article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
understanding the origin and evolution of disc galaxies is one of the major goals of modern astronomy .
the disc is a prominent feature of late type galaxies like the milky way .
as compared to distant galaxies , for which one can only measure the gross properties , the milky way offers the opportunity to study the disc in great detail . for the milky way
, we can determine 6-dimensional phase space information , combined with photometric and stellar parameters , for a huge sample of stars .
this has led to large observational programs to catalog the stars in the milky way in order to compare them with theoretical models .
the milky way stellar system is broadly composed of four distinct parts although in reality there is likely to be considerable overlap between them : the thin disc , the thick disc , the stellar halo and the bulge . in this paper ,
we mainly concentrate on understanding the disc components which are the dominant stellar populations . in the milky way ,
the thick disc was originally identified as the second exponential required to fit vertical star counts @xcite .
thick discs are also ubiquitous features of late type galaxies @xcite .
but whether the thick disc is a separate component with a distinct formation mechanism is highly debatable and a difficult question to answer . since the @xcite result
, various attempts have been made to characterize the thick disc .
some studies suggest that thick disc stars have distinct properties : they are old and metal poor @xcite and @xmath6 enhanced .
@xcite fit the sdss star counts using a two - component model and find that the thick disc has a larger scale - length than the thin disc .
in contrast , @xcite using a much smaller sample of sdss and segue stars find the opposite when they associate the thick disc with the @xmath6-enhanced component .
finally , the idea of a separate thick disc has recently been challenged .
@xcite argued that chemical evolutionary models with radial migration and mixing can replicate the properties of the thick disc ( see also * ? ? ? * who explore radial mixing using n - body simulations ) .
@xcite do not find the expected separation between metallicity and kinematics for f , g stars in the sdss survey , and @xcite argue that the thick disc is a smooth continuation of the thin disc
. opinions regarding the formation of a thick disc are equally divided .
various mechanisms have been proposed : accretion of stars from disrupted galaxies @xcite , heating of discs by minor mergers , radial migration of stars @xcite , a gas - rich merger at high redshift @xcite , and gravitationally unstable disc evolution @xcite , inter alia .
recently , @xcite have suggested that the thick disc can form without secular heating , mainly because stars forming at higher redshift had a higher velocity dispersion .
another possibility , proposed by @xcite , is misaligned angular momentum of in - falling gas .
how the angular momentum of halo gas becomes misaligned is described in @xcite .
however , @xcite and @xcite suggest that misaligned gas can destroy the discs . the obvious way to test the different thick disc theories is to compare the kinematic and chemical abundance distributions of the thick disc stars with those of different models . since , the thin and thick disc stars strongly overlap in both space and kinematics , it is difficult to separate them using just position and velocity . to really isolate and study the thick disc , one needs a tag that stays with a star throughout its life .
age is a possible tag but it is difficult to get reliable age estimates of stars .
chemical composition is another promising tag that can be used , but this requires high resolution spectroscopy of a large number of stars . in the near future , surveys such as galah using the hermes spectrograph @xcite and the gaia eso survey using the flames spectrograph @xcite should be able to fill this void . in our first analysis , we restrict ourselves to a differential kinematic study of the disc components .
we plan to treat the more difficult problem of chemo - dynamics in future .
the simplest way to describe the kinematics of the milky way stars of the solar neighborhood is by assuming gaussian velocity distributions with some pre - determined orientation of the principal axes of the velocity ellipsoid .
then if a single component disc is used , only three components of velocity dispersion and the mean azimuthal velocity @xmath7 need be known .
if a thick disc is included , one requires five additional parameters , one of them being the fraction of stars in the thick disc .
if stars are sampled from an extended volume and not just the solar neighborhood , then one needs to specify the radial dependence of the dispersions .
the velocity dispersion of a disc stellar population is known to increase with age , so one has to adopt an age velocity - dispersion relation .
discs heat because a cold , thin disc occupies a very small fraction of phase space , and fluctuations in the gravitational field cause stars to diffuse through phase space to regions of lower phase - space density .
the fluctuations arise from several sources , including giant molecular clouds , spiral arms , a rotating bar , and halo objects that come close to the disc .
one approach to computing the consequences of these processes is n - body simulation , but stellar discs are notoriously tricky to simulate accurately , with the consequence that reliable simulations are computationally costly .
in particular , they are too costly for it to be feasible to find a simulation that provides a good fit to a significant body of observational data . instead we characterize the properties of the milky way disc by fitting a suitable analytical formula .
the formula summarizes large amounts of data but its usefulness extends beyond this .
the formula is traditionally taken to be a power law in age .
the exponents @xmath8 , @xmath9 and @xmath10 of these power laws may not be the same for all three components .
the ratio @xmath11 and the values of @xmath8 , @xmath9 and @xmath10 are useful for understanding the physical processes responsible for heating the disc ( e.g. * ? ? ?
* ; * ? ? ?
the first generation of stellar population models characterized the density distribution of stars using photometric surveys .
@xcite assumed an exponential disc with magnitude - dependent scale heights .
an evolutionary model using population synthesis techniques was presented by . given a star formation rate ( sfr ) and an initial mass function ( imf ) , one calculates the resulting stellar populations using theoretical evolutionary tracks .
the important step forward was that the properties of the disc , like scale height , density laws and velocity dispersions , were assumed to be a function of age rather than being color - magnitude dependent terms . later introduced dynamical self - consistency to link disc scale and vertical velocity dispersions via the gravitational potential . further improved the constraints on sfr and imf of the disc .
the present state of the art is described in and is known as the besanon model . here , the disc is constructed from a set of isothermal populations that are assumed to be in equilibrium .
analytic functions for the density distribution , age / metallicity relation and imf are provided for each population .
a similar scheme is also used by the codes trilegal and _ galaxia _ @xcite .
there is a crucial distinction between kinematic and dynamical models . in a kinematic model ,
one specifies the stellar motions independently at each spatial location , and the gravitational field in which the stars move plays no role . in a dynamical model , the spatial density distribution of stars and their kinematics are self - consistently linked by the potential , under the assumption that the system is in steady state .
if one has expressions for three constants of stellar motion as functions of position and velocity , dynamical models are readily constructed via jeans theorem .
@xcite provides an algorithm for evaluating approximate action integrals , and has used these to fit dynamical models to the gcs data @xcite .
@xcite have confronted the predictions of the best of these models with rave data and shown that the model is remarkably , but not perfectly , successful .
our approach is different in two key respects : we fit kinematic rather than dynamical models , and we avoid adopting distances to , or using proper motions of , rave stars .
large photometric surveys such as denis , 2mass @xcite and sdss @xcite provide the underpinning for all galaxy modelling efforts .
the sdss survey has been used to provide an empirical model of the milky way stars @xcite .
the besanon model was fitted to the 2mass star counts , and its photometric parameters have been more thoroughly tested than its kinematic parameters because kinematic data for a large number of stars was not available when the model was constructed , the hipparcos satellite and the ucac2 catalog @xcite provided proper motions and parallaxes for @xmath12 stars in the solar neighborhood .
@xcite used the hipparcos data to study stellar kinematics as a function of color .
they also determined the solar motion with respect to the lsr and the axial ratios of the velocity ellipsoid .
@xcite also using hipparcos stars found the velocity dispersion to vary with function of age as @xmath13 . more recently ,
@xcite using data from a new reduction of the hipparcos mission estimated the solar motion and the avr for all three velocity components .
the avr is assumed to be a power law with exponents @xmath14 and @xmath10 for the three velocity components in the galactocentric cylindrical coordinate system .
they found @xmath15 .
they also investigated the star formation rate ( sfr ) and found it to be declining from past to present .
however , a degeneracy exists between the sfr and the slope of the imf , and constraining both of them together is challenging .
the gcs survey combined the hipparcos and tycho-2 proper motions with radial velocity measurements and strmgren photometry to create a kinematically unbiased sample of 16682 f and g stars in the solar neighborhood .
the data contains full 6d phase space information along with estimates of ages .
the temperature , metallicity and ages were further improved by and distances and kinematics were improved by using revised hipparcos parallaxes .
they investigated the avr and found @xmath16 which are at odds with @xcite . used the infrared flux method to improve the temperature , metallicity and age estimates for the gcs survey .
the uncertainty in estimated ages is an ongoing concern for studies that attempt to derive the avr directly from the gcs data . with the advent of large spectroscopic surveys
like rave @xcite and sdss / segue @xcite , we now have the radial velocity and stellar parameters for a large number of stars to beyond the solar neighborhood .
@xcite used sdss / segue to fit the spatial distributions of mono - abundance populations by double exponentials .
they showed that the vertical velocity dispersion declines exponentially with radius but varies little in @xmath17 .
finally , they argue that the thick disc is a continuation of the thin disc rather than a separate entity . the rave survey has also been used to study the stellar kinematics of the milky way disc . study the velocity dispersion and mean motion of the thin and thick disc stars in the @xmath18 plane .
they use the technique of singular value decomposition to compute the moments of the velocity distribution .
their analysis clearly shows that velocity dispersions fall as a function of distance from the galactic center .
@xcite explored the kinematics using red clump stars from rave and found complex structures in velocity space .
a detailed comparison with the prediction from the code _ galaxia _ was done , taking the selection function of rave into account .
the trend of dispersions in the @xmath18 plane showed a good match with the model .
however , the mean velocities showed significant differences .
studied the relation between kinematics and the chemical abundances of stars . by computing stellar orbits they deduced the maximum vertical distance @xmath19 and eccentricity @xmath20 of stars .
next they studied the chemical properties of stars by binning them in the @xmath21 plane .
they found that stars with @xmath22kpc and @xmath23 have two populations with distinct chemical properties , which hints at radial migration .
@xcite used full six - dimensional information for rave stars to fit a gaussian model to velocities in the @xmath24 plane .
they studied how the orientation and shape of the velocity ellipsoid varies with location in the galaxy , and provided analytic fits to the highly non - gaussian distributions of @xmath25 .
they also compared the observed kinematics of stars in different spatial bins with the predictions of a full dynamical model that had been fitted to the gcs data .
stellar kinematics allow us to measure the peculiar motion ( @xmath26 ) of the sun with respect to the local standard of rest ( lsr ) , and also the speed of the lsr ( in other words , the circular speed at the location of sun , @xmath27 ) .
there have been as many determinations of these as there have been new data , one of the earliest being ( @xmath28 km s@xmath29 by @xcite .
very precise measurements of these have been extracted from the hipparcos proper motions and the geneva copenhagen survey . @xcite and @xcite , using hipparcos proper motions , got ( @xmath30 km s@xmath29 .
a revision of @xmath4 was suggested by @xcite and @xcite .
later @xcite explained why the previous estimates , which used colors as a proxy for age , gave incorrect results . using a chemo - dynamical model calibrated on gcs data , they found ( @xmath31 km s@xmath29 .
@xcite described a model - independent method and suggests that @xmath32 could be as high as @xmath33 .
as further evidence of an unsettled situation , @xcite find from a sample of 3500 apogee stars @xmath34 , @xmath35 and @xmath36 km s@xmath29 and also suggest a revision of the lsr reference frame . [ cols="<,<,<,<,<",options="header " , ]
in this paper , we have constrained the kinematic parameters of the milky way disc using stars from the rave and the gcs surveys . to constrain kinematic parameters , we use analytic kinematic models based on the gaussian and shu distribution functions .
we use these distribution functions , padova stellar tracks and the selection functions of the surveys to predict the likelihood of each observed star . for gcs data , which has full phase - space information for the stars ,
we compute the likelihood in @xmath37 phase space . for rave data , we choose to fit the likelihood in @xmath38 space to avoid use of uncertain distances and proper motions .
we explored the full posterior distribution of model parameters using the markov chain monte carlo technique .
the parameters constrained include the solar peculiar motion ( @xmath26 ) , the circular speed at the sun @xmath3 , a parameter @xmath39 that controls the vertical gradient of @xmath40 , the age velocity dispersion relations ( via @xmath41 , @xmath42 and @xmath43 ) , and the scale lengths on which the dispersions vary , @xmath44 and @xmath45 . our results for both rave and gcs data are summarized in tables [ tab : tbgauss ] and [ tab : tbshu ] .
the final best fit model is given in .
in our kinematic modelling the main assumption we make is that we assume a sfr , imf and density laws that describe the spatial distribution of stars in accordance with the besanon model of , but with slight modifications as described in @xcite .
so the kinematic results that we present are conditional upon the above assumption .
moreover , kinematic models offer greater freedom than physics really allows .
so the accuracy of our kinematic results depends upon our ability to supply functional forms which are a good approximation to the actual velocity distribution . to overcome these concerns
, one should fit both the kinematics and the spatial distribution of stars together and they should be dynamically linked via the potential in which the stars move
. one could in principle constrain model parameters using the two surveys , rave and gcs , simultaneously . however , the two surveys probe different volumes , and it is not clear that a single value of a given parameter , for example the solar motion @xmath46 , is appropriate for both volumes : the immediate vicinity of the sun may be moving with respect to the wider disc , for example .
if such systematic differences exist , the simple models we are fitting can not provide an adequate account of the entire body of data , and parameter values obtained from a joint fit will be of doubtful physical significance .
hence , in this paper , we first analyzed the surveys separately and tried to understand the systematics .
then , having understood the extent to which each survey constrained each parameter , we fixed values of some parameters from the results of one survey while analysing the other .
we do this only for those parameters which we believe should take the same values for both surveys .
the gaussian model proves to be unsuitable for estimating disc parameters such as @xmath44 and @xmath4 because the fits prove to be strongly degenerate .
the gaussian model gives different values of @xmath44 for rave and gcs .
for rave it predicts negative values , implying that @xmath47 increases outwards .
this result is inconsistent with the disc s scale height and value of @xmath11 being constant .
negative values of @xmath44 also disagree with the findings of @xcite .
the shu model has three fewer parameters than the gaussian model and this helps it to break the degeneracy between @xmath44 and @xmath4 .
it gives positive and consistent values for @xmath44 for both rave and gcs .
the shu model also fits the rave data better than the gaussian model , especially with regard to stars values of @xmath25 .
the rave data allow us to constrain the solar peculiar motion and the local circular speed quite precisely .
our @xmath32 and @xmath48 are in good agreement with the results of @xcite , but our @xmath4 is lower by 5 km s@xmath29 .
the rave @xmath32 and @xmath48 are within 2-@xmath49 range of gcs values , but @xmath4 is lower by @xmath50 . using @xmath51 kpc and assuming @xmath52 we get @xmath53 km s@xmath29 . combining the estimates of @xmath3 and @xmath4 , we find the solar angular velocity with respect to the galactic center to be in good agreement with the measured proper motion of sgr a*. we find that if the fall of mean azimuthal velocity with height @xmath17 above the mid plane is neglected then this leads to an underestimation of @xmath3 .
although our random uncertainty regarding most parameters is quite small , due to large number of stars in the rave survey , significant sources of systematic uncertainty remain , especially regarding @xmath3 and @xmath4 .
our treatment of the vertical dependence of the kinematics is not fully self consistent .
this needs to be investigated with models that can handle the third integral of motion , e.g. , models based on action integrals . also we need to explore dynamical models that are self - consistent rather than pure kinematic models as studied here .
the values of @xmath3 and @xmath4 are also sensitive to the priors on age and distance distribution of stars .
so systematic errors of the order of the uncertainty in the priors are also expected .
when using the shu model , all parameters except @xmath4 and thick - disc parameters , show similar values for rave and gcs .
since there are very few thick disc stars in gcs , we deem the rave thick disc parameters to be more reliable . also , the uncertainty on @xmath44 and @xmath54 is substantially less for rave than for gcs .
the only parameter that is constrained better by gcs than rave is @xmath10 and this is partly due to the fact that we only use radial velocities in rave . in an attempt to build a concordance model , and to enable better comparison between the two data sets , we fix @xmath10 in rave to gcs values and then fix @xmath44 and @xmath45 in gcs to rave values . doing so we find that rave results are within 3@xmath49 of gcs results .
the most significant difference between the two is the value of @xmath4 , which is lower for rave by about @xmath55 .
the presence of prominent kinematic substructures in gcs could be responsible for this discrepancy
. however , inaccuracy in our vertical treatment of kinematics could also be responsible .
we find that the age - velocity dispersion relations in general satisfy @xmath56 , with @xmath9 is closer to @xmath8 than @xmath10 , contrary to the finding of @xcite .
this result is consistent with the physical principle that peculiar motions in the radial and azimuthal directions are strongly coupled by epicyclic dynamics , and large decoupled from vertical motions .
the fitted @xmath57 values depend on whether the thick disc is added separately or is left to be represented by the old tail of the thin disc , and they are naturally higher when it is not added separately .
the axial ratio @xmath11 of the thin disc velocity ellipsoid for the 10 gyr population is consistent with those predicted by @xcite for cloud scattering .
our values of @xmath8 and @xmath10 agree well with age - velocity profiles measured by for ages @xmath58 in simulations of disc galaxies . at ages larger than @xmath59 , a model that consists of power - law growth in the thin disc combined with a distinct thick - disc population , is too crude to represent the simulations adequately . in future
it may be appropriate to use more elaborate models inspired by simulations . in the shu model ,
the thick disc velocity dispersions for @xmath60 are very similar to those of the old thin disc . however , the radial scale length of the thick - disc velocity dispersions , @xmath45 , proved to be much smaller than that of the thin disc .
@xcite suggested a decrease of radial density scale length with age . in this regard , the role of our adopted priors on age and distance distribution of stars needs to be investigated further .
given the essential role that age plays in disc dynamics , it is unfortunate that the ages of stars are so hard to measure .
fortunately , big advances in this area are expected soon .
stellar astroseismology with missions like corot and kepler makes it possible to measure ages more accurately than before , and gaia will dramatically improve age estimates by geometrically determining distances to large numbers of stars .
meanwhile , chemical abundances , especially of the alpha elements , provides a fair proxy for age at a given metallicity .
hence , studying the relationship of kinematic properties with abundance will be crucial .
@xcite argued that each mono - abundance population has a distinct spatial distribution , and we expect cohorts of coeval stars to have spatial distributions that are characteristic of their ages .
ss is funded through arc dp grant 120104562 ( pi bland - hawthorn ) which supports the hermes project .
jbh is funded through a federation fellowship from the australian research council ( arc ) .
funding for rave has been provided by : the australian astronomical observatory ; the leibniz - institut fuer astrophysik potsdam ( aip ) ; the australian national university ; the australian research council ; the french national research agency ; the german research foundation ( spp 1177 and sfb 881 ) ; the european research council ( erc - stg 240271 galactica ) ; the istituto nazionale di astrofisica at padova ; the johns hopkins university ; the national science foundation of the usa ( ast-0908326 ) ; the w. m. keck foundation ; the macquarie university ; the netherlands research school for astronomy ; the natural sciences and engineering research council of canada ; the slovenian research agency ; the swiss national science foundation ; the science & technology facilities council of the uk ; opticon ; strasbourg observatory ; and the universities of groningen , heidelberg and sydney .
the rave web site is at http://www.rave-survey.org .
, a. e. , grenier , s. , udry , s. , et al .
1997 , in esa special publication , vol .
402 , hipparcos - venice 97 , ed .
r. m. bonnet , e. hg , p. l. bernacca , l. emiliani , a. blaauw , c. turon , j. kovalevsky , l. lindegren , h. hassan , m. bouffard , b. strim , d. heger , m. a. c. perryman , & l. woltjer , 621624 | we investigate the kinematic parameters of the milky way disc using the radial velocity ( rave ) and geneva - copenhagen ( gcs ) stellar surveys .
we do this by fitting a kinematic model to the data taking the selection function of the data into account . for stars in the gcs
we use all phase - space coordinates , but for rave stars we use only @xmath0 . using markov chain monte carlo ( mcmc ) technique ,
we investigate the full posterior distributions of the parameters given the data .
we investigate the ` age - velocity dispersion ' relation ( avr ) for the three kinematic components ( @xmath1 ) , the radial dependence of the velocity dispersions , the solar peculiar motion ( @xmath2 ) , the circular speed @xmath3 at the sun and the fall of mean azimuthal motion with height above the mid - plane .
we confirm that the besanon - style gaussian model accurately fits the gcs data , but fails to match the details of the more spatially extended rave survey .
in particular , the shu distribution function ( df ) handles non - circular orbits more accurately and provides a better fit to the kinematic data .
the gaussian distribution function not only fits the data poorly but systematically underestimates the fall of velocity dispersion with radius .
the radial scale length of the velocity dispersion profile of the thick disc was found to be smaller than that of the thin disc .
we find that correlations exist between a number of parameters , which highlights the importance of doing joint fits .
the large size of the rave survey , allows us to get precise values for most parameters .
however , large systematic uncertainties remain , especially in @xmath4 and @xmath3 .
we find that , for an extended sample of stars , @xmath3 is underestimated by as much as @xmath5 if the vertical dependence of the mean azimuthal motion is neglected .
using a simple model for vertical dependence of kinematics , we find that it is possible to match the sgr a * proper motion without any need for @xmath4 being larger than that estimated locally by surveys like gcs . |
since the seminal works on the small - world phenomenon by watts and strogatz @xcite and the scale - free property by barabsi and albert @xcite , the studies of complex networks have attracted a lot of interests within the physical community @xcite .
most of the previous works focus on the topological properties ( i.e. non - geographical properties ) of the networks . in this sense
, every edge is of length 1 , and the _ topological distance _ between two nodes is simply defined as the number of edges along the shortest path connecting them .
to ignore the geographical effects is reasonable for some networked systems ( e.g. food webs @xcite , citation networks @xcite , metabolic networks @xcite ) , where the euclidean coordinates of nodes and the lengths of edges have no physical meanings . yet , many real - life networks , such as transportation networks @xcite , the internet @xcite , and power grids @xcite , have well - defined node - positions and edge - lengths .
in addition to the topologically preferential attachment introduced by barabsi and albert @xcite , some recent works have demonstrated that the spatially preferential attachment mechanism also plays a major role in determining the network evolution @xcite .
very recently , some authors have investigated the spatial structures of the so - called _ optimal networks _ @xcite .
an optimal network has a given size and an optimal linking pattern , and is obtained by a certain global optimization algorithm ( e.g. simulated annealing ) with an objective function involving both geographical and topological measures .
their works provide some guidelines in network design . however , the majority of real networks are not fixed , but grow continuously . therefore , to study growing networks with an optimal policy is not only of theoretical interest , but also of practical significance . in this paper , we propose a growing network model , in which , at each time step , one node is added and connected to some existing nodes according to an optimal policy .
the degree distribution , edge - length distribution , and topological as well as geographical distances are analytically obtained subject to some reasonable approximations , which are well verified by simulations .
consider a square of size @xmath10 with open boundary condition , that is , a open set @xmath11 in euclidean space @xmath12 , where @xmath13 " signifies the cartesian product .
this model starts with @xmath14 fully connected nodes inside the square , all with randomly assigned coordinates . since there exists @xmath14 nodes initially , the discrete time steps in the evolution
are counted as @xmath15 .
then , at the @xmath16th time step ( @xmath17 ) , a new node with randomly assigned coordinates is added to the network .
rank each previously existing node according to the following measure : @xmath18 and the node having the smallest @xmath19 is arranged on the top . here
, each node is labelled by its entering time , @xmath20 represents the position of the @xmath16th node , @xmath21 is the degree of the @xmath1th node at time @xmath16 , and @xmath22 is a free parameter . the newly added node will connect to @xmath14 existing nodes that have the smallest @xmath19 ( i.e. on the top of the queue ) .
all the simulations and analyses shown in this paper are restricted to the specific case @xmath23 , since the analytical approach is only valid for the tree structure with @xmath23 .
however , we have checked that all the results will not change qualitatively if @xmath14 is not too large compared with the network size . [ 0.6 ] : ( a ) @xmath7 , ( b ) @xmath24 , ( c ) @xmath6 , ( d ) @xmath25 .
all the four networks are of size @xmath26 and @xmath23.,title="fig : " ] [ 0.8 ] and @xmath4 .
the solid line represents the assumption @xmath27 . in this simulation , we first generate a network of size @xmath28 by using the present optimal growing policy . to detect @xmath29 ,
some test nodes with randomly assigned coordinates are used , and for each test node , following eq .
( 1 ) , the existing nodes with minimal @xmath19 is awarded one point .
the test node will be removed from the network after testing .
the area @xmath29 for the @xmath1th node is approximately estimated as the ratio of the score node @xmath1 has eventually got after all the testings to the total number of test nodes .
the simulation shown here is obtained from 30000 test nodes , and the parameter @xmath30 is fixed.,title="fig : " ] in real geographical networks , short edges are always dominant since constructing long edges will cost more @xcite . on the other hand , connecting to the high - degree nodes will make the average topological distance from the new node to all the previous nodes shorter .
these two ingredients are described by the numerator and denominator of eq .
( 1 ) , respectively .
in addition , the weights of these two ingredients are usually not equal .
for example , the airline travellers worry more about the number of sequential connections @xcite , the railway travellers and car drivers consider more about geographical distances @xcite , and the bus riders often simultaneously think of both factors @xcite . in the present model , if @xmath7 , only the geographical ingredient is taken into account . at another extreme , if @xmath31 , the geographical effect vanishes .
1 shows some examples for different values of @xmath5 . when only the geographical ingredient is considered ( @xmath7 ) , most edges are very short and the degree distribution is very narrow . in the case of @xmath24 , the average geographical length of edges becomes longer , and the degree distribution becomes broader .
when @xmath6 , the scale - free structure emerges and a few hub nodes govern the whole network evolution . as @xmath5 becomes very large , the network becomes star - like .
at the @xmath16th time step , there are @xmath32 pre - existing nodes .
the square can be divided into @xmath32 regions such that if a new node is fallen inside the @xmath1th region @xmath33 , the quantity @xmath34 is minimized , thus the new node will attach an edge to the @xmath1th node . since the coordinate of the new node is randomly selected inside the square , the probability of connecting with the @xmath1th node is equal to the area of @xmath29 . if the positions of nodes are uniformly distributed , statistically , the area of @xmath29 is approximately proportional to @xmath35 with a time factor @xmath36 as @xmath37 .
2 shows the typical simulation results , which strongly support the valid of this assumption .
accordingly , by using the mean - field theory @xcite , an analytic solution of degree distribution can be obtained . however , when @xmath38 , most edges are connected to one single node ( see fig .
1d ) , so analytic solution is unavailable . here
, we only consider the case of @xmath39 .
assume @xmath40 where @xmath41 is a constant that can be determined self - consistently .
using the continuum approximation in time variable @xmath16 , the evolving of node @xmath1 s degree reads @xmath42 with the initial condition @xmath43 .
the solution is @xmath44 where @xmath45 accordingly , the degree distribution can be obtained as @xmath46 the constant @xmath47 is determined by the condition @xmath48 , where @xmath49 signifies the average degree .
the above solution is similar to the one obtained by using the approach of _ rate equation _ proposed by krapivsky _
in addition , one should note that if @xmath6 , the mean - field theory yields a solution @xmath50 , which is comparable to the exact analytic solution @xmath51 .
clearly , the degree distribution obeys a power - law form at @xmath6 , and an exponential form at @xmath7 .
when @xmath5 is in the interval @xmath8 , the networks display the so - called stretched exponential distribution @xcite : for small @xmath5 , the distribution is close to an exponential one , while for large @xmath5 , it is close to a power law .
this result is in accordance with the situation of transportation networks .
if only the geographical ingredient is taken into account ( e.g. road networks @xcite ) , then the degree distribution is very narrow . on the contrary , if the topological ingredient plays a major role ( e.g. airport networks @xcite ) , then the scale - free property emerges . when both the two ingredients are not neglectable ( e.g. bus networks @xcite ) , the degree distribution is intervenient between power - law and exponential ones .
[ 0.8 ] ( a ) and @xmath24 ( b ) .
the black squares and red curves represent simulated and analytic results , respectively .
all the data are averaged over @xmath52 independent runs , with network size @xmath28 ( i.e. @xmath53 ) fixed.,title="fig : " ] [ 0.8 ] ( a ) and @xmath24 ( b ) . the black squares and red curves represent simulated and analytic results , respectively . all the data are averaged over @xmath52 independent runs , with network size @xmath28 ( i.e. @xmath53 ) fixed.,title="fig : " ] fig .
3 shows the simulation results for @xmath6 and @xmath24 .
the degree distribution follows a power - law form when @xmath6 , which well agrees with the analytic solution . in the case of @xmath24 ,
the degree distribution is more exponential .
however , it is remarkably broader than that of the erds - rnyi model @xcite .
note that , the positions of all the nodes are not completely uniformly distributed , which will affect the degree distribution .
this effect becomes more prominent when the geographical ingredient plays a more important role ( i.e. smaller @xmath5 ) .
therefore , although the simulation result for @xmath24 is in accordance with the analysis qualitatively , the quantitative deviation can be clearly observed .
denote by @xmath54 the topological distance between the @xmath1th node and the first node . by using mathematical induction
, we can prove that there exists a positive constant @xmath55 , such that @xmath56 .
this proposition can be easily transferred to prove the inequality @xmath57 under the condition @xmath58 for @xmath59 . indeed ,
since the network has a tree structure , @xmath54 does not depend on time @xmath16 . under the framework of the mean field theory , the iteration equation for @xmath60 reads @xmath61 with the initial condition @xmath62 .
( 7 ) can be understood as follows : at the @xmath63th time step , the @xmath63th node has probability @xmath64 to connect with the @xmath1th node .
since the average topological distance between the @xmath1th node and the first node is @xmath54 , the topological distance of the @xmath63th node to the first one is @xmath65 if it is connected with the @xmath1th node .
according to the induction assumption , @xmath66 note that , statistically , @xmath67 if @xmath68 , therefore @xmath69 where @xmath70 denotes the average over all the nodes . substituting inequality ( 9 ) into ( 8) ,
we have @xmath71 rewriting the sum in continuous form , we obtain @xmath72 according to the mathematical induction principle , we have proved that the topological distance between the @xmath1th node and the first node , denoted by @xmath54 , could not exceed the order @xmath73 . for arbitrary nodes @xmath1 and @xmath74 ,
clearly , the topological distance between them could not exceed the sum @xmath75 , thus the average topological distance @xmath76 of the whole network could not exceed the order @xmath73 either .
this topological characteristic is referred to as the small - world effect in network science @xcite , and has been observed in a majority of real networks . actually , one is able to prove that the order of @xmath60 in the large @xmath16 limit is equal to @xmath77 ( see appendix a for details ) . furthermore , the iteration equation @xmath78 for general functions @xmath79 and @xmath80 , has the following solution : @xmath81 for the two special cases of @xmath7 ( @xmath82 , @xmath83 ) and @xmath6 ( @xmath82 , @xmath84 ) , the solutions are simply @xmath85 and @xmath86 , respectively . in fig .
4 , we report the simulation results about the average distance @xmath87 vs network size @xmath16 . in each case , the data points can be well fitted by a straight line in the semi - log plot , indicating the growth tendency @xmath88 , which agrees well with the analytical solution . [ 0.8 ] ( a ) and @xmath24 ( b ) .
as shown in each inset , the data points can be well fitted by a straight line in the semi - log plot , indicating the growth of the average distance approximately obeys the form @xmath89 .
all the data are averaged over @xmath90 independent runs , where the maximal network size is @xmath28 ( i.e. @xmath53).,title="fig : " ] [ 0.8 ] ( a ) and @xmath24 ( b ) .
as shown in each inset , the data points can be well fitted by a straight line in the semi - log plot , indicating the growth of the average distance approximately obeys the form @xmath89 .
all the data are averaged over @xmath90 independent runs , where the maximal network size is @xmath28 ( i.e. @xmath53).,title="fig : " ]
denote by @xmath91 the edge between nodes @xmath1 and @xmath74 , and the geographical length of edge @xmath91 is @xmath92 . when the @xmath93th node is added to the network
, the geographical length of its attached edge approximately obeys the distribution @xmath94 where @xmath95 in the large @xmath96 limit .
the derivation of this formula is described as follows .
the probability of the edge length being between @xmath97 and @xmath98 is given by the summation @xmath99 , where @xmath100 is the probability that @xmath101 falls between @xmath97 and @xmath98 , and the node @xmath1 minimizes the quantity @xmath102 among all the @xmath96 previously existing nodes .
this probability is approximately given by @xmath103 straightforwardly , the geographical length distribution of the newly added edge at the @xmath16th time step ( the @xmath16th edge for short ) is obtained as @xmath104 the lower boundary in the integral is replaced by 0 in the last step , which is valid only when @xmath105 .
the cumulative length distribution of the edges at time step @xmath106 is given by @xmath107,\end{aligned}\ ] ] where the argument of function @xmath108 is @xmath109 .
for @xmath110 , the approximate formula for @xmath111 reads @xmath112 and , when @xmath113 , @xmath114 if @xmath6 , the last step in eq .
( 16 ) is invalid but the analytic form for @xmath115 can be directly obtained as @xmath116\end{aligned}\ ] ] therefore , when @xmath117 , @xmath111 is approximately given by @xmath118 where @xmath119 is a numerical constant , and when @xmath113 , @xmath111 has the same form as that of eq .
( 19 ) .
5 plots the cumulative edge - length distributions . from this figure
, one can see a good agreement between the theoretical and the numerical results .
furthermore , one can calculate the expected value of the @xmath16th edge s geographical length , as @xmath120 which is valid only for sufficiently large @xmath16 and @xmath121 . according to eq .
( 22 ) , @xmath122 decreases as @xmath123 as @xmath16 increases , which is consistent with the intuition since all the @xmath16 nodes are embedded into a 2-dimensional euclidean space .
it may also be interesting to calculate the total length @xmath124 of all the edges at the time step @xmath16 , as @xmath125 @xmath124 is proportional to @xmath126 for @xmath127 .
when @xmath38 , a finite fraction of nodes will be connected with a single hub node and therefore we expect that @xmath128 in this case .
therefore , in the large @xmath16 limit , @xmath124 will increase quite abruptly when the parameter @xmath5 exceeds 1 .
this tendency is indeed observed in our numerical simulations , as shown in fig .
[ 0.8 ] ( a ) and @xmath24 ( b ) .
the black squares and red curves represent simulation and analytic results , respectively .
all the data are averaged over @xmath52 independent runs , with network size @xmath28 ( i.e. @xmath129 ) fixed.,title="fig : " ] [ 0.8 ] ( a ) and @xmath24 ( b ) . the black squares and red curves represent simulation and analytic results , respectively .
all the data are averaged over @xmath52 independent runs , with network size @xmath28 ( i.e. @xmath129 ) fixed.,title="fig : " ]
for an arbitrary path @xmath130 from node @xmath131 to @xmath132 , the corresponding geographical length is @xmath133 , where @xmath134 denotes the length of edge @xmath91 .
accordingly , the geographical distance between two nodes is defined as the minimal geographical length of all the paths connecting them .
now , we calculate the geographical distance @xmath135 between the @xmath1th node and the first node .
since our network is a tree graph , @xmath135 does not depend on time . by using the mean field theory
, we have @xmath136 or @xmath137 where , according to eq . ( 22 ) , @xmath138 it is not difficult to see that @xmath139 has an upper bound as @xmath16 approaches infinity .
one can use the trial solution @xmath140 to test this conclusion : @xmath141 where @xmath142 from eq .
( 27 ) , one obtains that @xmath143 .
similar to the solution of eq .
( 13 ) , @xmath144 as @xmath145 for @xmath7 and @xmath146 as @xmath145 for @xmath6 . however , it only reveals some qualitative property , and the exact numbers are not meaningful .
this is because the value of @xmath139 is obtained by the average over infinite configurations for infinite @xmath16 , while in one evolving process @xmath139 is mainly determined by the randomly assigned coordinates of the @xmath16th node .
[ 0.8 ] vs parameter @xmath5 with @xmath53 fixed . @xmath124
sharply increases when @xmath5 exceeds the critical value @xmath9 , which well agrees with the theoretical prediction.,title="fig : " ]
in many real - life transportation networks , the geographical effect can not be ignored .
some scientists proposed certain global optimal algorithms to account for the geographical effect on the structure of a static network @xcite . on the other hand , many real networks grow continuously .
therefore , we proposed a growing network model based on an optimal policy involving both topological and geographical measures .
we found that the degree distribution will be broader when the topological ingredient plays a more important role ( i.e. larger @xmath5 ) , and when @xmath5 exceeds a critical value @xmath9 , a finite fraction of nodes will be connected with a single hub node and the geographical effect will become insignificant .
this critical point can also be observed when detecting the total geographical edge - length @xmath124 in the large @xmath16 limit .
we obtained some analytical solutions for degree distribution , edge - length distribution , and topological as well as geographical distances , based on reasonable approximations , which are well verified by simulations .
.empirical degree distributions for geographical transportation networks .
[ cols="<,^,>",options="header " , ] although the present model is based on some ideal assumptions , it can , at least qualitatively , reproduce some key properties of the real transportation networks . in table 1 , we list some empirical degree distributions of transportation networks .
clearly , when building a new airport , we tend to firstly open some flights connected with previously central airports which are often of very large degrees .
even though the central airports may be far from the new one , to open a direct flight is relatively convenient since one does nt need to build a physical link .
therefore , the geographical effect is very small in the architecture of airport networks , which corresponds to the case of larger @xmath5 that leads to an approximately power - law degree distribution .
for other four cases shown in table 1 , a physical link , which costs much , is necessary if one wants to connect two nodes , thus the geographical effect plays a more important role , which corresponds to the case of smaller @xmath5 that leads to a relatively narrow distribution .
a specific measure of geographical network is its edge - length distribution .
a very recent empirical study @xcite shows that the edge - length distribution of the highly heterogenous networks ( e.g. airport networks , corresponding to the present model with larger @xmath5 ) displays a single - peak function with the maximal edge - length about five times longer than the peaked value ( see fig .
1c of ref .
@xcite ) , while in the extreme homogenous networks ( e.g. railway networks , corresponding to the present model with @xmath147 ) , only the very short edge can exist ( see fig .
1a of ref .
these empirical results agree well with the theoretical predictions of the present model .
firstly , when @xmath5 is obviously larger than zero , the edge - length distribution is single - peaked with its maximal edge - length about six times longer than the peak value ( see fig . 5 ) .
and , when @xmath5 is close to zero , eq .
( 17 ) degenerates to the form @xmath148 where @xmath96 denotes the network size . clearly , in the large @xmath96 limit , except a very few initially generated edges , only the edge of very small length @xmath97 can exist .
the analytical approach is only valid for the tree structure with @xmath23 .
however , we have checked that all the results will not change qualitatively if @xmath14 is not too large compared with the network size . some analytical methods proposed here are simple but useful , and
may be applied to some other related problems about the statistical properties of complex networks .
for example , a similar ( but much simpler ) approach , taken in section 4 , can also be used to estimate the average topological distance for some other geographical networks @xcite . finally , it is worthwhile to emphasize that , the geographical effects should also be taken into account when investigating the efficiency ( e.g. the traffic throughput @xcite ) of transportation networks .
very recently , some authors started to consider the geographical effects on dynamical processes , such as epidemic spreading @xcite and cascading @xcite , over scale - free networks .
we hope the present work can further enlighten the readers on this interesting subject .
the authors wish to thank dr .
hong - kun liu to provide us some very helpful data on chinese city - airport networks .
this work was partially supported by the national natural science foundation of china under grant nos .
10635040 , 70471033 , and 10472116 , the special research founds for theoretical physics frontier problems under grant no .
a0524701 , and specialized program under the presidential funds of the chinese academy of science .
substituting eq . ( 4 ) into eq . ( 7 ) , one obtains that @xmath149 then , define @xmath150 we next prove that @xmath151 in the large @xmath16 limit by using mathematical induction .
suppose for sufficiently large @xmath16 , all @xmath54 are less than @xmath152 for @xmath153 with @xmath119 being a constant greater than @xmath154 .
then , from eq .
( a1 ) , we have @xmath155 therefore , @xmath156 for all @xmath1 . similarly ,
suppose for sufficiently large @xmath16 , all @xmath54 are greater than @xmath157 for @xmath153 with @xmath158 being a constant less than @xmath154 .
then , from eq .
( a1 ) , we have @xmath159 therefore , @xmath160 for all @xmath1 . watts1998 d. j. watts , and s. h. strogatz , nature * 393 * , 440 ( 1998 ) .
barabsi , and r. albert , science * 286 * , 509 ( 1999 ) ; a. -l .
barabsi , r. albert , and h. jeong , physica a * 272 * , 173 ( 1999 ) .
r. albert , and a. -l .
barabsi , rev .
mod . phys . * 74 * , 47 ( 2002 ) .
s. n. dorogovtsev , and j. f. f. mendes , adv . phys . * 51 * , 1079 ( 2002 ) . m. e. j. newman , siam review * 45 * , 167 ( 2003 )
. s. boccaletti , v. latora , y. moreno , m. chavez , and d. -u .
hwang , phys .
rep . * 424 * , 175 ( 2006 ) .
d. garlaschelli , g. caldarelli , and l. pietronero , nature * 423 * , 165 ( 2003 ) .
k. brner , j. t. maru , and r. l. goldstone , proc .
natl . acad .
* 101 * , 5266 ( 2004 ) .
h. jeong , b. tombor , r. albert , z. n. oltvai , and a. -l .
barabsi , nature * 407 * , 651 ( 2000 ) .
p. sen , s. dasgupta , a. chatterjee , p. a. sreeram , g. mukherjee , and s. s. manna , phys .
e * 67 * , 036106 ( 2003 ) .
a. brrrat , m. barthlemy , r. pastor - satorras , and a. vespignani , proc .
* 101 * , 3747 ( 2004 ) ; r. guimer , s. mossa , a. turtschi , and l. a. n. amaral , proc .
102 * , 7794 ( 2005 ) .
m. faloutsos , p. faloutsos , and c. faloutsos , comput .
* 29 * , 251 ( 1999 ) .
r. pastor - satorras , a. vzquez , and a. vespignani , phys .
lett . * 87 * , 258701 ( 2001 ) .
r. crucitti , v. latora , and m. marchiori , physica a * 338 * , 92 ( 2004 ) .
r. albert , i. albert , and g. l. nakarado , phys .
e * 69 * , 025103 ( 2004 ) .
s. h. yook , h. jeong , and a. -l .
barabsi , proc .
* 99 * , 13382 ( 2001 ) .
m. barthlemy , europhys . lett . *
63 * , 915 ( 2003 ) ; c. herrmann , m. barthlemy , and p. provero , phys . rev .
e * 68 * , 026128 ( 2003 ) .
s. s. manna and p. sen , phys .
e * 66*,066114 ( 2002 ) ; p. sen , k. banerjee , and t. biswas , phys .
e * 66 * , 037102 ( 2002 ) ; p. sen and s. s. manna , phys.rev .
e * 68 * , 026104 ( 2003 ) ; s. s. manna , g. mukherjee , and p. sen , phys .
e * 69 * , 017102 ( 2004 ) ; g. mukberjee , and s. s. manna , phys .
e * 74 * , 036111 ( 2006 ) .
m. t. gastner , and m. e. j. newman , eur .
j. b * 49 * , 247 ( 2006 ) .
m. barthlemy , and a. flammini , e - print arxiv : physics/0601203 . b. m. waxman , ieee j. selected areas comm .
* 6 * , 1617 ( 1988 ) .
zhang , k. chen , y. he , t. zhou , b. -b .
su , y. -d .
jin , h. chang , y. -p .
zhou , l. -c .
sun , b. -h .
wang , d. -r .
he , physica a * 360 * , 599 ( 2006 ) .
krapivsky , s. redner , and f. leyvraz , phys .
. lett . * 85 * , 4629 ( 2000 ) ; p.l .
krapivsky , and s. redner , phys .
e * 63 * , 066123 ( 2001 ) .
j. laherrere , and d. sornette , eur .
j. b * 2 * , 525 ( 1998 ) .
p. erds , and a. rnyi , publ .
inst . hung .
sci . * 5 * , 17 ( 1960 ) .
l. a. n. amaral , a. scala , m. barthlemy , and h. e. stanley , proc .
97 * , 11149 ( 2000 ) ; s. h. strogatz , nature * 410 * , 268 ( 2001 )
. l. a. braunstein , s. v. buldyrev , r. cohen , s. havlin , and h. e. stanley , phys .
lett . * 91 * , 168701 ( 2003 ) .
v. latora , and m. marchiori , physica a * 314 * , 109 ( 2002 )
. k. h. chang , k. kim , h. oshima , and s. -m .
yoon , j. korean phys .
* 48 * , s143 ( 2006 ) .
m. kurant , and p. thiran , phys .
lett . * 96 * , 138701 ( 2006 ) ; phys .
e * 74 * , 036114 ( 2006 ) .
j. sienkiewicz , and j. a. hoyst , phys .
e * 72 * , 046127 ( 2005 ) .
liu , and t. zhou , acta physica sinica ( to be published ) .
g. bagler , e - print cond - mat/0409773 .
t. zhou , g. yan , and b. -h .
wang , phys .
e * 71 * , 046141 ( 2005 ) ; z. -m .
gu , t. zhou , b. -h .
wang , g. yan , c. -p .
zhu , and z. -q .
discret . impuls .
algorithms * 13 * , 505 ( 2006 ) .
zhang , l. -l .
rong , and f. comellas , physica a * 364 * , 610 ( 2006 ) ; z. -z .
zhang , l. -l .
rong , and f. comellas , j. phys .
a * 39 * , 3253 ( 2006 ) .
g. yan , t. zhou , b. hu , z. -q .
fu , and b. -h .
wang , phys .
e * 73 * , 046108 ( 2006 ) ; t. zhou , g. yan , b. -h .
wang , z. -q .
fu , b. hu , c. -p .
zhu , and w. -x .
wang , dyn .
discret . impuls .
algorithms * 13 * , 463 ( 2006 ) .
xu , z. -x .
wu , and g. chen , e - print arxiv : physics/0604187 ; x. -j .
xu , w. -x .
wang , t. zhou , and g. chen , e - print arxiv : physics/0606256 . l. huang , l. yang , and k. yang , phys . rev .
e * 73 * , 036102 ( 2006 ) . | in this article , we propose a growing network model based on an optimal policy involving both topological and geographical measures . in this model , at each time step , a new node , having randomly assigned coordinates in a @xmath0 square , is added and connected to a previously existing node @xmath1 , which minimizes the quantity @xmath2 , where @xmath3 is the geographical distance , @xmath4 the degree , and @xmath5 a free parameter .
the degree distribution obeys a power - law form when @xmath6 , and an exponential form when @xmath7 .
when @xmath5 is in the interval @xmath8 , the network exhibits a stretched exponential distribution .
we prove that the average topological distance increases in a logarithmic scale of the network size , indicating the existence of the small - world property .
furthermore , we obtain the geographical edge - length distribution , the total geographical length of all edges , and the average geographical distance of the whole network .
interestingly , we found that the total edge - length will sharply increase when @xmath5 exceeds the critical value @xmath9 , and the average geographical distance has an upper bound independent of the network size .
all the results are obtained analytically with some reasonable approximations , which are well verified by simulations . |
despite decades of deep and wide extragalactic surveys , we still do not understand the astrophysics behind the empirical relationship linking the star formation histories of galaxies and their morphologies .
observations show that galaxies with evolved stellar populations , so - called `` quiescent '' galaxies , have significantly smaller sizes and more concentrated light profiles than actively star - forming galaxies with a similar stellar mass and redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although we know that galaxies must shut down their star formation and migrate from the star - forming to quiescent population , there is much to be learned about the physical process(es ) that are primarily responsible for this structural evolution and the quenching of star formation .
one way to study the connection between this bimodal population of galaxies is through correlations between specific star formation rate ( ssfr@xmath6sfr / m@xmath7 ) and parameters describing various physical properties of galaxies , such as stellar mass ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , surface density ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , bulge mass ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , or environment ( e.g. * ? ? ? * ) .
the inverse of the ssfr defines a timescale for the formation of the stellar population of a galaxy , where lower ssfrs correspond to older stellar populations for a constant or single burst star formation history . in this sense , ssfr is a relatively straight forward diagnostic of quiescence that can be directly linked to other physical properties of galaxies . with a sample of galaxies selected from the sloan digital sky survey ( sdss ) ,
@xcite were the first to show that there is a turnover in the ssfr of galaxies at higher stellar surface mass densities ( see also * ? ? ? * ) .
the redshift evolution of this correlation was later presented in @xcite .
both works identified a threshold surface density at each redshift interval : below this threshold the ssfrs are high with little variation , and above the threshold density galaxies have low ssfrs .
@xcite reported that the density threshold increases with redshift , at least out to @xmath1=3 .
as stellar density and velocity dispersion are closely related ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , observations therefore indicate that galaxies are statistically more likely to be quiescent once they have surpassed a threshold in either density or velocity dispersion .
@xcite showed that these distant compact galaxies have similar densities to the central regions of local ellipticals by comparing their average stellar density profiles within a constant physical radius of 1 kpc ( see also * ? ? ?
* ; * ? ? ?
this study was the first to present a plausible link between these high redshift galaxies and where they end up in the local universe , but it is yet unclear what causes the quenching in the first place .
this work does however suggest that it may be more robust to define a quenching threshold in surface density within the _ central _ 1 kpc , as opposed to the half - light radius .
@xcite find that this central density threshold increases with stellar mass through a study of the correlation between galaxy structure and the quenching of star formation using a sample of sdss central galaxies .
@xcite extended the @xmath1=0 sdss study of @xcite out to @xmath1=3 . where the earlier work of @xcite found an evolving effective surface density threshold with redshift , @xcite do not find a strong redshift evolution in the central surface density threshold
however , as there still exist star - forming galaxies above this quenching threshold , results in the literature conclude that a dense bulge is a necessary but insufficient condition to fully quench galaxies ( see also * ? ? ?
while most studies have focused on the the stellar mass dependence of the central density alone , it is perhaps unsurprising that there is a tight correlation : the central density is a biproduct of the combined stellar mass and light profile
. the key comparison instead should be with ssfr , normalizing out the stellar mass dependence of sfr ( as also studied in * ? ? ?
* ) , where total ssfrs can be measured largely independent of the central density .
while the dynamic range in stellar mass enabled by the deep high resolution near - infrared ( nir ) imaging from the cosmic assembly near - ir deep extragalactic legacy survey ( candels ; * ? ? ?
* ; * ? ? ?
* ) improves dramatically from earlier multi - wavelength extragalactic surveys ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , the depth of the spitzer / mips 24@xmath8 m imaging used to derive the ir sfr indicator has remained unchanged . to therefore leverage the full range in stellar mass and galaxy structure probed by these _ hubble space telescope _
legacy programs , we must perform detailed stacking analyses of the 24@xmath8 m imaging to probe the sfr properties of the complete unbiased sample of galaxies using a single robust sfr indicator ( e.g. , * ? ? ? * ) . by combining the high resolution photometry from candels with the accurate spectroscopic information provided by the 3d - hst treasury program @xcite and a stacking analysis of the unobscured ( uv ) and obscured ( ir ) sfrs , we are in a unique position to perform a census across most of cosmic time of the simultaneous evolution of galaxy structure and star formation . while earlier results from this treasury dataset have shown that all quiescent galaxies have a dense stellar core , and
the formation of such cores is a requirement for quenching @xcite , there are several open questions that we aim to answer in this paper .
specifically , ( 1 ) how does star formation rate depend on galaxy size ?
, ( 2 ) is there a preferential galaxy size scale where star formation occurs ?
, ( 3 ) is there a physical parameter that will uniquely predict quiescence ? , and ( 4 )
does the quenching threshold in surface density and velocity evolve with redshift ?
there are a few differences that together separate the present analysis from earlier studies : the inclusion of accurate grism redshifts from 3d - hst improve the stellar population parameters , we derive the three dimensional deprojected central density and circular velocity instead of the surface density , and we stack the 24@xmath8 m imaging to robustly measure total sfrs for more extended , lower stellar mass , or low sfr galaxies . the paper is outlined as follows . in section [ sec : data ] , we introduce the data and sample selection , describing the details of the stellar masses , redshifts , rest - frame colors , structural parameters , total star formation rates , central densities and circular velocities used herein .
we present the correlations between galaxy size , stellar mass and ssfr for the overall population in section [ sec : size ] . in section
[ sec : sizescale ] , we determine at which galaxy size scale the most star formation occurs from @xmath1=0.5 to @xmath1=2.5 . and in section [ sec : sfalone ] , we proceed to analyze the residual offsets in sfr and size for star - forming galaxies alone once removing the well known correlations between log(sfr)log(m@xmath7 ) and log(@xmath9)log(m@xmath7 ) . in the second half of the paper ,
we explore the physical parameters that best predict quiescence .
first , we consider the role of galaxy size and srsic index in predicting quiescence in section [ sec : sersic ] .
next , in section [ sec : density ] , we study the dependence of ssfr on stellar mass density , parameterizing the redshift evolution of the density and velocity quenching thresholds in section [ sec : evolution ] . as this paper touches on a relatively wide range of topics ,
we integrate the discussion and implications of the results throughout the relevant sections , as well as further discussion in section [ sec : discussion ] . while we choose to place these empirical results in the context of current theoretical models
, we note that many of the correlations that we discuss can be interpreted in a different way ( e.g. , * ? ? ?
we caution that it is yet unclear if there is truly an evolutionary sequence causally linking galaxy structure with star formation .
we conclude the paper with a summary in section [ sec : summary ] of the results presented herein , in the context of current and future studies of galaxy formation and evolution . in this paper , we use a @xcite initial mass function ( imf ) and assume a @xmath10cdm cosmology with @xmath11 , @xmath12 , and @xmath13 km s@xmath14 mpc@xmath14 .
all magnitudes are given in the ab system .
we use the exquisite _ hst _ wide field camera 3 ( wfc3 ) multi - wavelength photometric and spectroscopic datasets of five well - studied extragalactic fields through the candels and 3d - hst surveys . using stellar masses , redshifts , and rest - frame colors from the 3d - hst 0.38@xmath8 m photometric catalogs
( see * ? ? ?
* for full details ) , we select samples in three redshift intervals of 0.5@xmath0@[email protected] , 1.0@xmath0@[email protected] , and 1.5@xmath0@[email protected] : 11266 , 9553 , and 7791 galaxies above the stellar mass limits for star - forming galaxies . when splitting the sample into sub - populations and accounting for stellar mass limits , our sample is reduced further to 9694 ( 1192 ) , 8643 ( 705 ) , and 6893 ( 766 ) star - forming ( quiescent ) galaxies greater than stellar mass limits of @xmath15 , 8.8 ( 9.4 ) , and 9.4 ( 10.0 ) , respectively .
the galaxies are defined to be either star - forming or quiescent based on their rest - frame @xmath16@xmath17 and @xmath17@xmath18 colors , following the definition of @xcite .
we have identified and removed luminous active galactic nuclei ( agns ) using the _ spitzer_/irac color selections presented in @xcite , as they may have significant contamination of their ir sfr ; only 2% of the sample were removed as agn candidates .
the total sample comprises 27,893 galaxies at 0.5@xmath0@[email protected] .
we only analyze data where we are mass - complete for star - forming galaxies .
the lower bounds of the stellar mass limits correspond to the mass - completeness limits down to which @xcite can determine structural parameters for star - forming and quiescent galaxies with good fidelity .
the values for star - forming galaxies are in agreement with the mass - completeness limits presented in @xcite , which are determined by comparing object detection in candels / deep with a re - combined subset of the exposures which reach the depth of candels / wide .
we indicate the stellar mass limits from @xcite to which we can trust the star - forming and quiescent structural measurements within figures [ fig : size ] and [ fig : sfuniverse ] .
we additionally correct the stellar masses of star - forming galaxies for contamination of the broadband fluxes from emission - lines using the values presented in appendix a of @xcite .
these corrections only begin to become significant at @xmath19 and @xmath20 . where available , we combine the spectral energy distributions ( seds ) with low - resolution hst / wfc3 g141 grism spectroscopy to derive grism redshifts with 0.3% accuracy @xcite .
@xcite present the full details of the 3d - hst grism data reduction and redshift analysis .
we select the `` best '' redshift to be the spectroscopic redshift , grism redshift or the photometric redshift , in this ranked order depending on the availability . photometric redshifts comprise 52% ( 57% ) of the @xmath21 ( @xmath22 ) sample , while 39% ( 40% ) have grism redshifts and 9% ( 2% ) spectroscopic redshifts .
size and srsic indices used herein are measured from deep hst / wfc3 @xmath23 and @xmath24 photometry , as presented in @xcite . the effective radius in each filter is defined to be the semi - major axis of the ellipse that contains half of the total flux of the best - fitting srsic model .
@xcite parameterize the effective radius as a simple function of redshift and stellar mass ( their equations 1 and 2 ) .
following equation 2 in @xcite , the rest - frame 5000 effective radius for galaxies with @[email protected] is measured from the @xmath23 effective radius , whereas @xmath24 is used at @[email protected] .
similarly , we adopt the srsic indices measured from the @xmath23 photometry at @xmath21 and @xmath24 at @xmath22 ( see * ? ? ?
* for details ) for the central density measurement in section [ sec : centraldensity ] .
the details of the error analysis on @xmath9 and @xmath5 are presented in @xcite .
total star formation rates are derived from stacking analyses of spitzer / mips 24@xmath8 m photometry , following the procedure detailed in @xcite .
the spitzer / mips 24@xmath8 m images in the aegis field are provided by the far - infrared deep extragalactic legacy ( fidel ) survey @xcite , cosmos from the s - cosmos survey @xcite , goods - n and goods - s from @xcite , and uds from the spitzer ukidss ultra deep survey(spuds ; pi : j. dunlop ) .
briefly , the analysis code uses a high - resolution @xmath23+@xmath25+@xmath24 detection image as a prior to model the contributions from neighboring blended sources in the lower resolution mips 24@xmath8 m image .
all galaxies are `` cleaned '' of the contaminating flux of the neighboring sources before stacking .
we refer the reader to section 3 of @xcite for the full details of the mips 24@xmath8 m stacking analyses .
the sfrs derived for quiescent galaxies herein are likely upper limits , as the 24@xmath8 m technique tends to overestimate the sfrs for galaxies with @xmath26@xmath0 - 10 yr@xmath14 @xcite .
we note that the uv+ir sfr technique is generally robust for star - forming galaxies .
we derive uncertainties in the average star formation rates from 50 monte carlo bootstrap simulations of the stacking analyses .
the error in the mean is therefore the width of the resulting distribution of star formation rates divided by the square root of the number of galaxies in each bin .
we choose to use the spitzer / mips 24@xmath8 m ir sfr because of the resolution and depth of the observations , and to mitigate systematic uncertainties when combining different sfr indicators .
however , we note that the observed 24@xmath8 m samples major spectral features arising from polycyclic aromatic hydrocarbons ( pahs ) . despite complications from these pah features , @xcite demonstrate that the luminosity - independent conversion from 24@xmath8 m to the bolometric ir luminosity used here yields estimates that are in good median agreement with measurements from herschel / pacs photometry ( see also * ? ? ? * and b. lee et al . in prep ) . when combining the 24@xmath8 m ir sfr indicator with the rest - frame uv , we can therefore successfully recover the average total amount of star formation in galaxies .
while considering the average correlations and including a bootstrap error analysis will reduce the potential noise in these measurements , we can not rule out that there exist biases due to the physical conditions of the dust and the star formation itself that are incredibly difficult to quantify .
given the observed projected two dimensional surface density , the surface brightness distribution can be deprojected to obtain a three dimensional light distribution if we assume spherical symmetry .
we derive this three - dimensional density profile from the best - fit structural parameters to the intensity profiles of the individual galaxies described in section [ sec : structure ] .
the projected luminosity within a projected radius is described in @xcite , assuming isotropic spherical galaxies with surface luminosity profiles following the srsic profile .
we adopt the asymptotic approximation for the term @xmath27 from @xcite .
this asymptotic expansion is truncated to the first four terms and is accurate to @xmath28 for exponential disks ( @xmath5=1 ) and @xmath29 for a de vaucouleur profile ( @xmath5=4 ) .
the approximation for @xmath27 presented in @xcite performs much better than previous formulae ( e.g. , * ? ? ? * and others ) .
however , despite the accuracy of the asymptotic expansion , we note that this methodology may lead to errors for galaxies that are far from spherical symmetry , in particular for flat disks .
we return to this issue in section [ sec : density ] .
following the equations summarized in section 2.2 of @xcite , we perform an abel transform to deproject the circularized , three - dimensional light profile . assuming mass follows the light and there are no strong color gradients , the total luminosity is converted to a total stellar mass using the stellar masses presented in @xcite .
these stellar masses ( labeled @xmath30 below ) are derived by modeling the spectral energy distributions and correspond to the final 3d - hst data release .
following @xcite , we apply a small correction to these stellar masses to take into account the difference between the total magnitude in the photometric catalog and the total magnitude implied by the srsic fit ( see * ? ? ?
on average , this correction is [email protected] .
the central density is therefore calculated by numerically integrating the following equation : @xmath31 where @xmath32 is the total , aperture - corrected luminosity of the galaxy from the 3d - hst catalogs in the filter corresponding to the srsic profile measurement ( e.g. , @xmath23 or @xmath24 ) .
@xmath33 is the total luminosity as measured from integrating the best - fit srsic profile . in 10% of the cases ,
the numerical integration does not converge and these unreliable measurements are removed from the subsequent analysis .
we find that these galaxies generally have gaussian profiles with @[email protected] , larger than average sizes with @[email protected] kpc , and low stellar masses within @xmath340.5 dex of the stellar mass limits ; these galaxies represent precisely the population one might expect to fail . while we are effectively using the same formalism as @xcite , who derive the `` core '' mass within the central 1 kpc , we instead parameterize the central density and circular velocity within 1 kpc to facilitate comparisons to earlier results by @xcite .
we note that these derived parameters are essentially equivalent , modulo constant factors , where m@xmath35@xmath36@xmath37@xmath36@xmath38 . to quantify the errors on the central density , we perform 50 bootstrap simulations of this numerical integration , perturbing @xmath9 and @xmath5 by pseudo random offsets drawn from a gaussian distribution with a standard deviation equal to the respective 1@xmath39 errors , as calculated by @xcite .
the width of the resulting distribution of central densities is taken as the error on each individual measurement .
the error on the average central density for each bin is then the square root of the sum of the errors in quadrature divided by the number of galaxies in the bin .
if we balance the gravitational force acting on the mass enclosed within the central 1 kpc with the centrifugal force , assuming spherical symmetry , the circular velocity of a test particle at radius @xmath40=1 kpc will be , @xmath41 where @xmath42 is the gravitational constant and equal to 4.302@xmath4310@xmath44 kpc m@xmath45 ( km / s)@xmath46 , and the stellar mass enclosed within the central 1 kpc sphere is determined from equation [ eq : rho1 ] .
the central circular velocity is a factor of @xmath47 greater than the velocity dispersion , which both @xcite and @xcite study .
we again caution that we must be careful to test that the results presented herein are not driven by flat galaxies , which deviate from the assumption of spherical symmetry .
we return to this issue in section [ sec : density ] .
in figure [ fig : size ] , we present the rest - frame 5000 size of galaxies as a function of their stellar mass in three redshift intervals , 0.5@xmath0@[email protected] , 1.0@xmath0@[email protected] , and 1.5@xmath0@[email protected] . the data are grouped into 0.2 dex bins of logarithmic stellar mass and size .
the size of the symbol represents the number of galaxies that populate that area in parameter space .
the symbols are color - coded by the ssfr derived from the uv+ir stacking analysis , with open symbols signifying upper limits .
whereas all galaxies are included in the left panels , the middle and right panels are separated into the star - forming and quiescent populations as determined from rest - frame u - v and v - j colors .
the middle and right panels of figure [ fig : size ] additionally show the size - mass contours in greyscale of the opposite population , quiescent and star - forming respectively . for the overall galaxy population ( figure [ fig : size ] , left ) , galaxies with smaller sizes at fixed stellar mass are forming stars at lower rates .
we confirm the earlier results of @xcite and numerous others , who show that the compact sizes of more massive quiescent galaxies are offset from the average size of star - forming galaxies by factors of approximately four at fixed stellar mass at least out to @[email protected] . here
we show that these compact galaxies indeed have low ssfrs ( see also * ? ? ?
* ) . as @xcite point out
, the true sfrs of quiescent galaxies may be even lower as the mid - ir flux density can originate from processes unrelated to ongoing star formation , such as cirrus dust heated by old stellar populations and circumstellar dust .
the uv+ir sfrs derived for quiescent galaxies herein are therefore likely upper limits ( right panels in figure [ fig : size ] ) , with the effect setting in for 24@xmath8m - derived sfrs with @xmath26@xmath0 - 10 yr@xmath14 ( e.g. , * ? ? ?
accounting for the overestimation of the quiescent sfrs by treating the sfrs measured for @xmath26@xmath0 - 10 yr@xmath14 as upper limits will only serve to accentuate the trends between galaxy size and sfr for the overall population . in the right most panels of figure [ fig
: size ] , we see a flattening of the galaxy size - mass relation for quiescent galaxies at low stellar masses ( @xmath010@xmath48 m@xmath49 ) , similar to @xcite and @xcite . this is most evident at 0.5@xmath0@[email protected] , where the stellar mass limits imposed by the structural measurements extend down to 10@xmath50 m@xmath49 .
these low - mass quiescent galaxies tend to have sizes similar to the bulk of the star - forming population at that epoch , as well as slightly higher ssfrs than more massive quenched galaxies .
these results hint at a more gradual quenching of star formation , perhaps due to a depleted gas supply .
these trends are likely not driven by environmental effects : results from @xcite find no significant environmental dependence of the sizes of central and satellite quiescent galaxies at fixed stellar mass at @xmath1@xmath40 . with the present observations , it is interesting to consider on what galaxy size scale most of the stars in the universe form . in figure [ fig : sfuniverse ] , the symbol size represents the product of the average ssfr for that bin with the number of galaxies .
a symbol can be small either because there are few galaxies that populate that parameter space and/or because those galaxies are not forming very many new stars on average .
the black line demarcates the 50th percentile , signifying the size scale at which half of the star formation for a given stellar mass occurs .
the grey shaded region then shows the 25th and 75th percentile range , encompassing half of the star formation in the universe .
we see that 50@xmath51 of stars are formed in galaxies with sizes within @xmath20.13 dex from the average star - forming size - mass relation , as shown as the solid blue from @xcite for comparison .
this width of 0.26 dex is of similar order to the @xmath52 intrinsic scatter of the size - mass relation ( see figure 6c in * ? ? ? * ) .
most of the stars in the universe are formed in star - forming galaxies with typical sizes .
while we find that amongst the overall galaxy population most of the star formation occurs within a narrow range of sizes , we can further explore if we see any dependence on the star formation rate _ within _ the star - forming population . in the following section
, we isolate these star - forming galaxies based on their rest - frame u - v and v - j colors ( see figure 26 in * ? ? ?
* ) . when solely selecting actively star - forming galaxies based on their rest - frame colors ( middle panels in figure [ fig : size ] ) , the dependence of the average ssfr of a galaxy on size is far less pronounced relative to the overall galaxy population ( left panels ) .
to first order , for any given stellar mass , larger star - forming galaxies have the same average ssfr as smaller star - forming galaxies within @xmath00.2 dex .
we start to see deviations from this trend at the highest stellar masses and lowest redshift interval .
this is likely related to the observed flattening in the slope of the star formation sequence towards later times for galaxies more massive than @xmath410@xmath53 m@xmath49 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and the correlation between this flattening and srsic index @xcite .
we note that the trends between the size - mass plane and ssfr for star - forming galaxies agree nicely with those at @xmath1=0 presented in @xcite .
ssfr , we find the same general trends as the uv+ir sfr amongst the overall galaxy population : large galaxies have higher ssfrs than smaller galaxies at fixed stellar mass .
similarly , we find the same lower envelope of compact star - forming galaxies with depressed ssfrs . ] in order to better quantify the dependence of the average star formation rate on galaxy size , we first must remove the well - known trends with stellar mass and redshift ( see the appendix for details ) . figure [ fig : size_sfr ] presents the residual relation between the average star formation rate and rest - frame 5000 galaxy size for star - forming galaxies only .
the grey - scale demarcates the typical observed 0.3 dex scatter in the star - formation sequence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , where the transparency is defined by a gaussian distribution .
if we split the data into quartiles in @xmath54log(sfr ) , taking into account the number of galaxies that go into each data point , we find no significant differences ; the sizes of galaxies in the highest quartile are [email protected] dex larger than galaxies in the lowest quartile . in other words , we see that the residual sfrs of the majority of galaxies show little to no dependence on galaxy size ( see also fang et al . in prep ) .
using spatially - resolved maps of h@xmath55 for star - forming galaxies at z@xmath41 in the 3d - hst survey , @xcite find that h@xmath55 is enhanced at all radii above the main ridgeline of the log(sfr)-log(m@xmath7 ) plane , and depressed at all radii below .
this suggests that the physical processes driving the rate of star formation acts throughout the entirety of galaxy disks . taking these results at face value with the present analysis ,
this supports the idea that galaxies follow `` parallel - tracks '' ( * ? ? ?
* see section [ sec : density ] for further discussion ) .
we note that this picture is complicated by the presence of large amounts of dust ( and associated cold molecular gas ) and apparent gradients thereof ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , as well as potential systematic uncertainties introduced by the stacking methods employed . while results from @xcite demonstrate that the stellar mass density profiles do not appear to depend strongly on any potential color gradients , other studies have found spatial variations in mass - to - light ratios to be important ( e.g. , * ? ? ?
* ; * ? ? ?
we discuss this issue further in section [ sec : density ] .
we have performed several tests in order to explore if the lack of trend between ( s)sfr and galaxy size is driven by either the relatively large redshift bins or the adopted sfr indicator .
if we assume that the ssfr of any given galaxy is independent of the size , we can use the same parent sample redshift distribution and the well observed average relation between log(sfr)log(m@xmath7 ) to randomly assign each galaxy a size from the @xcite size - mass relation .
we find that we can reproduce all observed trends , shy of the lower envelope of compact fading star - forming galaxies .
the large redshift bins therefore do not effect the conclusions of this paper .
the uv+ir total sfrs probe star formation timescales of order 100 myr .
when instead considering the h@xmath55 star formation rate indicator , which is sensitive to shorter timescales of order 10 myr , we find the same trends amongst star - forming galaxies in figure [ fig : halpha ] .
for this test , we use the h@xmath55 emission line fluxes from the 3d - hst survey in the two lowest redshift bins and correct for dust attenuation from the best - fit a@xmath56 from the spectral energy distribution . even though this dust correction is somewhat uncertain ( see * ? ? ? * ) , it is likely that we may be underestimating the dust attenuation which could make the trends between galaxy size and ssfr slightly more pronounced .
however , from this test we can conclude that the general independence of star formation rate and galaxy size is likely not sensitive to the timescale on which the star formation rate is probed .
the notable exception to the lack of dependence on ssfr are galaxies with intermediate stellar masses ( log(m@xmath7/m@xmath49)@xmath410.0 - 10.6 ) and compact sizes ( @xmath9@xmath02 kpc ) .
these massive compact galaxies have star formation rates significantly depressed relative to the bulk population of star - forming galaxies .
they may be in the process of fading to join the quiescent population ( see also * ? ? ?
larger statistical samples and/or deeper 24@xmath8 m observations are required to further improve the uncertainty in the star formation rates presented here .
nonetheless , we see evidence for an interesting trend between the star formation rates and sizes of intermediate mass , compact star - forming galaxies . both figure [ fig : sfuniverse ] and figure [ fig : size_sfr ] have implications for the formation and quenching of these compact galaxies .
formation models of distant compact galaxies must explain the cessation of star - formation ( `` quenching '' ) and the transformation to smaller , bulge - dominated galaxies , for which the progenitors appear to be larger rotating disks @xcite .
these compact galaxy populations are reproduced within cosmological simulations , where they either ( 1 ) have very early formation times when the universe was far denser @xcite , or ( 2 ) they are the result of a central starburst driven by violent disk instabilities @xcite or gas - rich mergers @xcite .
however , it may also be that galaxies do not undergo such `` compaction '' events , and that compact galaxies simply evolved from lower mass , slightly smaller galaxies @xcite . in cosmological simulations by @xcite , where a central starburst drives structural evolution , galaxies with high ssfrs ( above the average star formation sequence ) are predicted to be compact with short gas depletion timescales ( see also * ? ? ? * ) .
we do not see evidence that compact galaxies have higher than average ssfrs here .
if anything , we see the opposite trend , at least amongst intermediate stellar mass galaxies ( log(m@xmath7/m@xmath49)@xmath410.0 - 10.6 ) .
however , we note that a recent study by @xcite do indeed find extremely low ( co ) gas fractions in a pilot sample of compact star - forming galaxies , indeed suggesting short gas depletion timescales . as these compact star - forming galaxies exist in very small numbers
, they would need to quench rapidly ( @xmath00.5 gyr timescales ) in order to produce the required number of compact quiescent galaxies @xcite .
indeed , the early results on the gas depletion timescales suggest timescales of order 100 myr or less @xcite .
@xcite also show that galaxies undergoing mergers or showing signs of morphological disruptions have the shortest molecular gas depletion times .
these results hint that the timescale for galaxies to pass through this compact high ssfr phase is indeed short , and this is why the observational evidence is lacking when considering these average trends .
now that we have confirmed the dichotomy between the sizes of quiescent and star - forming galaxies ( section [ sec : size ] ) , with most of the star formation occurring within a narrow range of sizes ( section [ sec : sizescale ] ) , we additionally show that there is little to no dependence of ssfr on size within the star - forming galaxies alone .
taking these three points together , this suggests an abrupt change in star formation rate after a galaxy attains a particular structure . in the next section
, we will investigate how well various galaxy structure parameters can uniquely predict this decrease in ssfr .
amongst the several competing theories put forth to explain the simultaneous evolution of the structures of quiescent and star - forming galaxy populations across cosmic time , there are predictions in each case for how galaxies are expected to grow in stellar mass with respect to their structures .
@xcite postulate that the distribution of massive galaxies form a universal `` l - shaped track '' comprised of the two fundamental physical processes of compaction and quenching . here
, compaction is the process by which star - forming galaxies become structurally similar to quiescent galaxies through a rapid period of core - growth , where they grow in central density ( also increasing their core - to - total mass and srsic indices , and decreasing their size ) ( see * ? ? ?
* ; * ? ? ?
quenching occurs when these compact star - forming galaxies reach a central density threshold . on the other hand , the `` parallel tracks '' model by @xcite instead suggests that galaxies follow an inside - out growth track in the size - mass plane , where the stellar mass is gradually increased within a fixed physical radius and galaxies quench when they reach a stellar density or velocity dispersion threshold .
the relatively small observed scatter in the size - mass relation @xcite and the star formation sequence @xcite support this idea of a more gradual growth of galaxies . for all galaxies ( left ) and uvj - selected star - forming galaxies ( right )
is well correlated with the average ssfr for a given rest - frame u - v colors across all redshifts probed here .
as there is no universal trend , this implies that although @xmath5 is a good predictor of quiescence , it is not a sufficient condition to predict a low ssfr .
the implications of these results on the slope and scatter of the star formation sequence are presented in @xcite . ]
regardless for the timescale defining the build up of the central mass concentration and the transition to quiescence , the common denominator between these various physical mechanisms is the existence of a threshold in either stellar density or velocity dispersion . in section [ sec : sersic ]
, we will explore if galaxy size or srsic index alone could predict quiescence , followed by a detailed analysis of stellar mass surface density , central density and central circular velocity in sections [ sec : density ] and [ sec : evolution ] . from the present analysis ,
there are two interesting points regarding the ability to predict when a galaxy will quench : ( 1 ) compact galaxies have lower ssfrs on average ( section [ sec : size ] ) , and ( 2 ) most stars form in galaxies with typical ( larger ) sizes for their given total stellar mass ( section [ sec : sizescale ] ) . if a galaxy is observed to have a compact size for the given total stellar mass , does this uniquely predicted it will be quenched ?
we can recast the information in figure [ fig : size ] to instead plot the ssfr as a function of @xmath54log(@xmath9 ) , or the deviation in size from the average size - mass relation ( see appendix for more details ) . in figure
[ fig : size_ssfr ] , we see that galaxies with lower ssfrs are smaller than average with red rest - frame u - v colors . however , it is important to note that there also exist similarly compact galaxies with more typical ssfrs .
this emphasizes that galaxy size for a given stellar mass alone can not predict quiescence .
let us next consider the case of `` morphological quenching '' ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) where a galaxy will quench once it builds up a significantly massive bulge , as indicated observationally by a high srsic index @xmath5 ( e.g. , * ? ? ?
* ) . following an analogous analysis of the dependence of @xmath5 on the slope and scatter of the log(sfr)-log(m@xmath7 ) relation
, @xcite showed that the slope is of order unity for disk - like galaxies , equivalent to a constant ssfr . on the other hand , galaxies with @xmath5@xmath32 ( implying more dominant bulges and higher central densities ) have significantly lower ssfrs than the main ridgeline of the star formation sequence .
however , this earlier study did not explicitly present the dependence of ssfr on @xmath5 in the same manner as the present analysis . to that end
, we show the dependence of ssfr on @xmath5 , color - coded by the average rest - frame u - v color in figure [ fig : sersic_ssfr ] .
we see from figure [ fig : sersic_ssfr ] that while high @xmath5 generally indicates lower ssfr , @xmath5 is also not a unique predictor ; galaxies with high @xmath5 ( bulge - dominated ) exhibit a broad range of ssfrs that is strongly correlated with their rest - frame u - v colors ( see also discussion in * ? ? ? * ) .
the trends present in figure [ fig : sersic_ssfr ] are striking , with important implications for the derived slope and scatter of the star formation sequence ( as noted by * ? ? ?
@xcite present a schematic diagram of how the various physical mechanisms could move galaxies around the ssfr-@xmath5 plane , accounting for dry and wet mergers , disk instabilities , galaxy harassment , agn feedback and other slow gas depletion processes .
however , it is clear from the present data that both galaxy size and @xmath5 alone are insufficient to isolate quiescent galaxies with low ssfrs . , when selecting in a narrow range of stellar mass ( selected to be 10.45@xmath0log(m / m@xmath49)@xmath010.5 for demonstration purposes only ) .
the color - coding indicates the central density within a physical radius of 1 kpc for the individual galaxies : small galaxies with high @xmath5 have high central densities , whereas larger galaxies with low @xmath5 have lower central densities . ] the stellar mass density profile of a galaxy is defined by both the effective radius and the srsic index @xmath5 of a galaxy .
the combination of these two parameters may therefore be more powerful than either parameter alone .
figure [ fig : demo ] demonstrates the strong correlation between galaxy size and srsic index at fixed stellar mass . in this figure
, we have selected a small number of galaxies from the parent sample such that they inhabit a narrow range in stellar mass of 10.45@xmath0log(m / m@xmath49)@xmath010.50 at 1.0@xmath0@[email protected] .
the galaxies are color - coded by their deprojected central density within a fixed physical radius of 1 kpc , as derived in section [ sec : centraldensity ] .
we see in figure [ fig : demo ] that the stellar mass density within the central 1 kpc of these galaxies is strongly correlated with both size and srsic index . in the following sections
, we will next explore if this stellar mass density is the most robust predictor of quiescence , relative to stellar mass , @xmath9 or @xmath5 alone .
we perform a similar analysis as in figure [ fig : size ] , but instead derive stacked uv+ir ( s)sfrs in 0.2 dex bins of logarithmic stellar mass and stellar mass surface density ( log(@xmath57 ) ; figure [ fig : sigmae_ssfr ] ) and central stellar density ( log(@xmath37 ) ; figure [ fig : sigma1_ssfr ] ) in three redshift intervals , 0.5@xmath0@[email protected] , 1.0@xmath0@[email protected] , and 1.5@xmath0@[email protected] .
a similar bootstrap analysis on the stacked ir sfrs was performed to derive uncertainties in the average values , as described in section [ sec : centraldensity ] .
these measured values are then recast as ssfrs as a function of log(@xmath57 ) and log(@xmath37 ) in the subsequent analysis . by using ssfr , we are normalizing the sfrs by stellar mass and probing an inverse timescale sensitive to the ages of the stellar populations .
the individual measured values are shown in greyscale in figures [ fig : sigmae_ssfr ] and [ fig : sigma1_ssfr ] ; the plume rising towards higher ssfr at the lowest densities , offset from the average relations , reflects the limits of the data on a galaxy - by - galaxy basis .
similarly , we see a cut off at the lowest ssfrs and highest densities . in both of these extreme parameter regimes ,
many of the individual galaxies remain undetected in the 24@xmath8 m imaging , necessitating the stacking analysis presented herein .
figure [ fig : sigmae_ssfr ] shows a strong correlation between the average ssfrs of galaxies and their stellar mass density within the effective radius .
the result is strikingly different to that of figures [ fig : size_ssfr ] and [ fig : sersic_ssfr ] . for galaxies with low surface densities ,
the average ssfr is roughly constant at a given epoch . on the other hand ,
we see a strong drop in the ssfr for galaxies with high densities .
we observe these trends in all three redshift intervals , finding that the turnover in ssfr at higher densities is stronger for quiescent and star - forming galaxies together than for star - forming galaxies alone . as the majority of massive galaxies
are quiescent ( e.g. , * ? ? ?
* ) and have more centrally concentrated stellar mass profiles ( e.g. , * ? ? ?
* ) , they will preferentially pull down the average ssfr at the highest densities . if we instead consider the central density , which is derived from the deprojected stellar mass within a 1 kpc radius sphere , we see that the trend remains essentially the same shape but the scatter is significantly reduced in figure [ fig : sigma1_ssfr ] .
the offset in stellar density when comparing the solid and dashed black lines in figure [ fig : sigmae_ssfr ] result from the implicit difference between the stellar mass enclosed within the central 1 kpc as compared to the effective radius .
we have already accounted here for the constant factor of 0.12 dex introduced during the deprojection from a sphere to a circle . while the trend in @xmath54sfr-@xmath54@xmath58 from @xcite is coined a `` universal l - track '' , we find a more gradual evolution in ssfr with increasing central density when looking at the average values , that we will also demonstrate to evolve with redshift .
these results are in good quantitative agreement with b. lee et al .
( in prep ) , who perform a similar analysis but instead use sfr indicators from sed fitting using flexible star formation histories .
our results agree qualitatively with that of @xcite at @xmath1=0 , who show that sdss central galaxies with higher central surface densities have lower ssfr . it is somewhat challenging to directly compare to the @xmath1=0 results with the present analysis , as they separate their sample by stellar mass . in figure
[ fig : sigma1_mass ] , we isolate galaxies with log(m@xmath7/m@xmath49)@xmath5910.6 for both stellar surface mass density ( left panel ) and central surface density ( right panel ) .
to facilitate a direct comparison , we convert central density to surface density here by applying an offset of 0.12 dex corresponding to the equivalent difference in constants ( see equation [ eq : rho1 ] ) . while we only show the one redshift epoch , 0.5@xmath0@[email protected] , that has the largest dynamic range , we find similar trends at higher redshifts .
we see that our measurements for the most massive galaxies span over two decades in stellar mass central surface density and almost two decades in ssfr . for similar stellar masses in @xcite ,
the evolution in central surface density at @xmath1=0 is significantly less , of order @xmath00.5 dex .
when considering where the most massive galaxies reside in the log(ssfr)-log(@xmath57 ) plane in the left panel of figure [ fig : sigma1_mass ] , we see that they preferentially have lower ssfrs relative to the average relation , and thereby their less massive counterparts .
intriguingly , we no longer see this stellar mass dependence when instead considering central ( surface ) density ( right panel , figure [ fig : sigma1_mass ] ) .
the earlier results of @xcite showing that the bulge mass together with the bulge - to - total ratio ( or @xmath5 ) correlate most strongly with the degree of quiescence imply that the central mass concentration will be a key factor in quenching .
here we explicitly consider the density within the central 1 kpc out to @xmath1=2.5 and demonstrate that it is indeed a remarkably clean tracer of the average ssfr of galaxies .
these results confirm the low redshift study by @xcite , who use a novel technique to rank the relative importance of sdss central galaxy properties in the process of quenching star formation .
similar to our conclusions , @xcite find that central velocity dispersion and/or central stellar mass concentration are excellent predictors of the cessation of star formation . )
@xmath5910.6 ) lie systematically below the average log(ssfr)-log(@xmath57 ) relation ( left ) , we do not see any such offsets when instead considering the central surface density within 1 kpc ( where log(@xmath58 = log(@xmath37)-log(4@xmath60/3 ) ; right ) . ]
the present analysis rests on the assumption that light traces mass ( sections [ sec : structure ] and [ sec : centraldensity ] ) . when accounting for any potential variations in mass - to - light ratios from either the star formation histories , ages , or dust across the galaxies , @xcite and @xcite show that the stellar mass maps and profiles are generally smoother and more centrally concentrated than the profiles in rest - frame uv and optical light .
furthermore , the evidence is mounting for massive compact concentrations of gas and dust residing in a significant fraction of typical star - forming galaxies towards the high mass end from high - resolution studies of the rest - frame uv / optical light @xcite , as well as ( sub)mm studies @xcite .
the highest stellar mass star - forming galaxies may be effected most severely , where we could be underestimating their central mass concentration . preferentially underestimating the central densities and/or the sfrs ( e.g. , * ? ? ?
* ) of the most massive star - forming galaxies could make the turnover in ssfr above the density threshold ( falsely ) steeper .
figure [ fig : sigma1_mass ] demonstrates the dynamic range in central density that the most massive galaxies inhabit . despite the effects of variations in mass - to - light ratios being stronger towards higher stellar masses , the location of the break in the log(ssfr)-log(@xmath37 ) plane may not be affected severely as the vast majority of massive galaxies have central densities well above the turnover .
we fit the log(ssfr)log(@xmath57 ) relation in figure [ fig : sigmae_ssfr ] and log(ssfr)log(@xmath37 ) relation in figure [ fig : sigma1_ssfr ] with broken power laws to independently quantify the behavior of galaxies above and below the characteristic values of @xmath57 and @xmath37 .
the broken power law is parameterized as , @xmath61 where x equals either @xmath57 or @xmath37 .
the slope @xmath62 is roughly flat below the characteristic densities @xmath63 ( figure [ fig : bestfit_sigma_char ] ) , and steeply drops off above these characteristic values . as the central circular velocity is directly proportional to the central density ( see section [ sec : vcirc ] )
, we show the corresponding values of @xmath64 in the right axis of figure [ fig : bestfit_sigma_char ] .
the best - fit parameters and their associated uncertainties are listed in tables [ tab : sigmae_powlaw ] and [ tab : sigma1_powlaw ] .
we see that the best - fit power low is very similar between the log(ssfr)log(@xmath37 ) relation in figure [ fig : sigma1_ssfr ] , and that of log(ssfr)log(@xmath57 ) relation in figure [ fig : sigmae_ssfr ] .
the best - fit log(ssfr)log(@xmath58 ) relation is shown as a dashed line for reference in figure [ fig : sigmae_ssfr ] , where log(@xmath58)@xmath6log(@xmath37)-log(4/3 ) .
the main difference between these two measurements is that there is a significantly larger scatter amongst the average measured values of @xmath57 , whereas @xmath37 ( or @xmath58 ) shows remarkably little scatter .
lcccccccc & & + redshift range & @xmath65 & @xmath66 & @xmath63 & @xmath67 & @xmath65 & @xmath66 & @xmath63 & @xmath67 + @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & @xmath75 & @xmath76 + @xmath77 & @xmath78 & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath85 + @xmath22 & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 + _ notes .
_ broken power law coefficients parameterizing the evolution of the @xmath94(ssfr)-@xmath95 relation from the median stacking analysis ( equation [ eq : powlaw ] ) .
@xmath65 signifies the best - fit slope for galaxies below the characteristic stellar surface mass density @xmath57 , and @xmath66 is the slope above this limit . in order to test that the trends we observe in figures [ fig : sigma1_ssfr ] are not driven by flat galaxies , where the deprojected central density may not be robust given the assumptions , we repeat the analysis removing all galaxies with @xmath5@xmath02 .
we find that the best - fit power laws for the entire sample are unchanged , albeit we are missing the vast majority of data points falling below log(@xmath37)@xmath07.5
m@xmath49 kpc@xmath96 where most galaxies are disks with low @xmath5
. figure [ fig : bestfit_sigma_char ] shows that both the characteristic central density , @xmath97 , and the equivalent characteristic circular velocity , @xmath64 , evolve towards higher values at earlier times for both the analysis of all galaxies ( circles ) and uvj star - forming galaxies only ( stars ) . while this characteristic turnover marks a threshold above which galaxies become less efficient at forming new stars , it does not necessarily signal quiescence . with a broken power law , we find that the turnover in the broken power law , @xmath98 , evolves towards lower central densities at lower redshifts .
we note that there is not a large difference between @xmath1=2 and @xmath1=1.25 , for the measurements including all galaxies ( quiescent and star - forming ) .
we also show the equivalent central circular velocity on the right axis , where @xmath99 . ]
@xcite assert that galaxies quench once they reach a central surface density of @[email protected] m@xmath49 kpc@xmath101 ( or @[email protected] m@xmath49 kpc@xmath96 , when accounting for the deprojection of a sphere of radius 1 kpc ) .
similarly , @xcite quote a threshold in velocity dispersion of 225 km / s , based on an analysis of compact star - forming galaxies with a median size is @xmath9 = 1.8 kpc . when correcting to @xmath40 = 1 kpc following @xcite , this quenching threshold increases to 234 km / s .
this is equivalent to a central circular velocity of @xmath4331 km / s , assuming @xmath103=@xmath47@xmath39 .
we compare our results to these density and velocity thresholds in figure [ fig : bestfit_sigma_quenched ] .
however , we note that the interpretation of these results requires us to define what qualifies a galaxy as quenched first .
if we simply define a galaxy to have reached quiescence when log(ssfr)@xmath0 - 10 yr@xmath14 , we see in figure [ fig : bestfit_sigma_quenched ] that this results in an evolution in the critical central density for quenching , which we deem @xmath104 . alternatively , we could define a galaxy to quench once it deviates significantly below the average star formation sequence at that cosmic epoch .
this latter definition will result in an evolving limit in log(ssfr ) with cosmic time . to that end
, we estimate the average ssfr from results presented in @xcite . for the purpose of this exercise ,
we only consider lower mass galaxies where the ssfr is roughly constant , ignoring the increasing turnover in ssfr at the massive end towards later times .
if we assume the average log(ssfr ) equals -9.0 , -8.8 , and -8.6 yr@xmath14 at 0.5@xmath0@[email protected] , 1.0@xmath0@[email protected] , and 1.5@xmath0@[email protected] , respectively , an offset of 1 dex below the star formation sequence ( @xmath43@xmath39 ) reaches log(ssfr)=-9.6 yr@xmath14 at @xmath1=2 and lower values at later times ( log(ssfr)=-10 yr@xmath14 by @xmath1=0.75 ) .
@xcite adopt a 0.7 dex offset below the star formation sequence , however they normalize at the massive end where there is an evolving turnover towards increasingly lower ssfrs relative to the lower mass population as redshift decreases .
lcccccccc & & + redshift range & @xmath65 & @xmath66 & @xmath63 & @xmath67 & @xmath65 & @xmath66 & @xmath63 & @xmath67 + @xmath68 & @xmath105 & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 + @xmath77 & @xmath113 & @xmath114 & @xmath115 & @xmath81 & @xmath116 & @xmath117 & @xmath118 & @xmath119 + @xmath22 & @xmath109 & @xmath120 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 & @xmath93 + _ notes .
_ broken power law coefficients parameterizing the evolution of the @xmath94(ssfr)-@xmath102 relation from the median stacking analysis ( equation [ eq : powlaw ] ) .
@xmath65 signifies the best - fit slope for galaxies below the characteristic stellar mass density within 1 kpc , @xmath37 , and @xmath66 is the slope above this limit . accepting that defining quiescence can become complicated given the redshift evolution of the physical properties of galaxies ,
we see regardless that the average trends presented in this paper are indicating that the threshold for galaxies to quench is almost a decade higher at @xmath1@xmath42 compared to @[email protected] . in other words
, we find evidence that this threshold decreases to lower central densities and velocities at later times .
this implies that , on average , quiescent galaxies that quench at later times will have lower central densities and velocities . perhaps unsurprisingly ,
these trends are close to the expected cosmological evolution of density , as @xmath126@xmath36@xmath127 , the dashed line in figure [ fig : bestfit_sigma_quenched ] ( normalized to the lowest redshift measurement ) . similarly , as @xmath64@xmath36@xmath128 , the central circular velocity scales as @xmath129 .
the central densities of quiescent galaxies are shown in grayscale in figure [ fig : bestfit_sigma_quenched ] , with the running median shown in red .
the typical central density of quiescent galaxies at a given epoch is lower than our quenching threshold .
if we instead saw that these quiescent galaxies had higher central densities , this would suggest that the threshold we find is too low , as otherwise the star - forming galaxies would need to be able to continue to form more stars and further increase their central mass concentrations .
the results of @xcite show that the central densities of high redshift galaxies are slightly higher than low redshift ellipticals . as we observe galaxies to puff up over time via minor mergers and accretion ( e.g. , * ? ? ?
* ) , this suggests that although quiescent galaxies grow in size with time , their central densities will likely not continue to increase once quenched .
as these quiescent galaxies must have reached this threshold at earlier times , their distribution of central densities with redshift suggest that the evolution of the quenching threshold slows down above @xmath1@xmath42 .
it is unclear if this evolution in the quenching threshold is physical or simply an artifact of other processes triggering the shutdown of star formation .
despite the fact that is difficult to interpret the meaning behind this evolving quenching threshold , the observed central densities of quiescent galaxies with redshift hint that quiescence should not be defined as a non - evolving limit of log(ssfr)=-10 .
spatially resolved absorption line studies of the steller populations with extremely deep datasets such as the upcoming lega - c survey at @xmath1@xmath41 @xcite will potentially constrain the timescale for quenching as well as to better quantify the physical parameters predicting quiescence .
-10 yr@xmath14 , the corresponding central density measured from figure [ fig : sigma1_ssfr ] , @xmath130 , is significantly higher at earlier times . we find a similar trend if we instead define quenching to occur when galaxies have ssfrs 1 dex below the average observed ssfr ( open circles ; overlapping by definition at @xmath1=0.75 ) .
the greyscale shows the central densities measured for quiescent galaxies at analogous epochs , where the red line marks the running median of the quiescent population ; the median central density of the quiescent population lies just below the quenching threshold at @xmath1@xmath31 .
the dashed line indicates the expected cosmological evolution in density normalized at @xmath1=0.5 , where @xmath131 .
we also show the equivalent central circular velocity on the right axis , where @xmath99 . the long - dash and dot - dash lines are the assumed constant thresholds in surface density @xcite and velocity dispersion @xcite above which galaxies quench . ]
in this section , we will place our observational results in the context of various theoretical frameworks . when considering the trends we observe between galaxy size , stellar mass , and ssfr
, we find that they agree to first order with the size evolution predictions from cosmological simulations . if we compare the predictions from the eagle simulations at @xmath1=0 ( figure 2 in * ? ? ?
* ) to the observations presented in figure [ fig : size ] , we see the same lower envelope of compact star - forming galaxies with depressed ssfrs . the effect in the eagle simulations may be slightly more enhanced than the observations , with their ssfrs at fixed stellar mass ranging from 0.40.8 dex .
simulations also show this flattening in the slope of the size - mass relation at stellar masses below 10@xmath48 m@xmath49 amongst quiescent galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
interestingly , @xcite predict a trend for higher ssfr ( suggesting more recent assembly ) for larger quiescent galaxies at fixed stellar mass .
but we do not see any strong trends amongst these quiescent observations . as the 24@xmath8 m derived sfrs are likely upper limits for quiescent galaxies ,
it is possible that we are washing out a stronger intrinsic trend within the observations .
the key trend that the simulations do not reproduce amongst the star - forming population is the lack of variation in ssfr at lower stellar masses .
while the models show variations of order 0.30.5 dex , the variation in ssfr within the five extragalactic fields included in the 3d - hst dataset is @xmath00.2 dex .
unfortunately , @xcite do not present the higher redshift results , so a more direct comparison at the equivalent epochs is not possible . future comparisons between the observations and cosmological hydrodynamical simulations will prove illuminating .
both the remarkably small scatter and the evolution of the average relation between ssfr and central density is interesting in the context of recent arguments in the literature regarding the the nature of the most recently quenched galaxies and their role in the evolution of the size - mass relation of quiescent galaxies .
although it has been shown that quiescent galaxies will experience growth through minor mergers and accretion @xcite , the simplest explanation of their size growth is the continuous addition of ( larger ) recently quenched galaxies @xcite .
galaxies that quench at later times are expected to have larger sizes because the universe was less dense and therefore gas - rich , dissipative processes were less efficient @xcite .
however , observations at @[email protected] find that the most recently quenched galaxies are similar , if not more compact , than older quiescent galaxies at the same epoch @xcite .
there is a mixed bag of size measurements at @[email protected] @xcite , with recently quenched galaxies at @[email protected] exhibiting a range of sizes , whereas studies at @xmath1@xmath01 find that the most recently quenched galaxies are the largest @xcite .
a trend is emerging : recently quenched galaxies may be more compact at high redshift , whereas they are larger at later times . at @xmath1@xmath31 , we show here that the central density threshold for quenching is higher relative to the already quenched population in figure [ fig : bestfit_sigma_quenched ] ( shown as greyscale , with the running median in red ) . we therefore see evidence that higher redshift quiescent galaxies are predicted to have significantly higher central densities to quench , which may therefore alleviate some of the tension in the observations between low and higher redshift analyses .
when considering where massive galaxies populate the @xmath94(ssfr)@xmath94(@xmath37 ) plane ( e.g. , figure [ fig : sigma1_mass ] ) , we see that they tend to have the densest central concentrations of stellar mass .
it is important to note however , that the trend itself between central density and ssfr does not vary strongly with stellar mass at a given epoch . in other words , while different stellar mass regimes tend to populate the upper and lower end of this relation , it is the entire relation itself that evolves with redshift .
the stellar mass dependence of this quenching threshold in central density observed by @xcite at @xmath1=0 may be explained in part by more massive galaxies having formed earlier in the universe @xcite .
there is a thorough discussion of this downsizing in quenching in @xcite . as more massive dark
matter halos will tend to cross a threshold halo mass for viral shock heating earlier in the universe , halo quenching will occur preferentially in more massive galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
similarly , violent disk instabilities will also have a natural downsizing , which @xcite argue is the result of higher gas fractions in lower mass galaxies .
whether galaxies quenches fast through , e.g. , violent disk instabilities , or slow through halo quenching , more massive galaxies will tend to cross this quenching threshold earlier in time . in figures [ fig : bestfit_sigma_char ] and [ fig : bestfit_sigma_quenched ] , we show that the quenching threshold differs from the characteristic turnover value .
this difference may either imply that galaxies keep growing their centers after star formation starts actively diminishing . or , that there is significant scatter in this quenching threshold , where galaxies with central densities in between the characteristic turnover and the quenching threshold have an intermediate probability of being quenched .
although we observe that a high central density appears to predict quiescence on average , we caution that this does not necessarily imply causation .
@xcite propose instead that galaxies quench their star formation according to empirical probabilistic laws that depend solely on the total mass of the galaxy , not on the surface mass density or size .
with their simple model they can broadly reproduce all of the trends between galaxy structure and ssfr that we observe here , including the evolving quenching threshold .
@xcite argue that because galaxies form their stars inside - out and passive galaxies will form them at earlier epochs ( higher redshifts ) , passive galaxies will always have smaller sizes than their star - forming counterparts of the same stellar mass at any given redshift . determining the cause of quenching is beyond the scope of this paper .
but we note that even without causation , we show that the high central density holds a unique predictive power in identifying the population of galaxies that , on average , will be quiescent .
the aim of this paper is to connect rest - frame optical measurements of the size - mass and density - mass relations with robust measurements of the total specific sfrs from a purely empirical standpoint .
we can thereby connect galaxy structure and star formation to better understand the observed bimodal distribution of galaxies across cosmic time and the quenching of star formation .
this current study extends the original work by @xcite on a smaller field that included 1155 galaxies at 0.2@xmath0@[email protected] to now consider a mass - complete sample of 27,893 galaxies at 0.5@xmath0@[email protected] .
the sample is selected in five extragalactic fields from the 3d - hst photometric catalogs presented in @xcite , combining high spatial resolution _
nir imaging from the candels treasury program with total uv+ir sfrs derived from a stacking analysis of _ spitzer_/mips
24@xmath8 m imaging . 1 .
we find that 50% of new stars being formed amongst the overall population occurs in galaxies within @xmath20.13 dex of the average size - mass relation .
extremely compact or extended galaxies do not significantly contribute to the total stellar mass budget .
we show a flattening in the size - mass relation of quiescent galaxies at stellar masses below 10@xmath48 m@xmath49 at 0.5@xmath0@[email protected] .
these lower mass quiescent galaxies exhibit slightly higher ssfrs relative to more massive galaxies at the same epoch , suggesting more recent assembly
. however , the ssfrs of quiescent galaxies at fixed stellar mass do not show significant variations .
3 . after removing
the well known correlations between stellar mass and star formation rate and galaxy size , we show that the majority of star - forming galaxies show no dependence of their star formation rates on galaxy size .
there exists a small population of compact intermediate mass ( log(m@xmath7/m@xmath49)=10 - 10.6 ) star - forming galaxies who exhibit significant depressions of up to 0.5 dex in log(ssfr ) .
similar trends are found amongst massive galaxies in simulations ( e.g. , * ? ? ?
we find that the independence of star formation rate on galaxy size is not sensitive to the timescale on which the star formation rate is probed , with dust - corrected h@xmath55 ssfrs yielding similar trends .
we confirm earlier studies ( e.g. , * ? ? ?
* ) , showing that the central stellar density is a key parameter connecting galaxy morphology and star formation histories : galaxies with high central densities are red and have increasingly lower ssfrs , whereas galaxies with low central densities are blue and have a roughly constant ( higher ) ssfrs at a given redshift interval
we use a broken power - law to parameterize the correlation between log(ssfr ) and central density , log(@xmath37 ) , showing remarkably little scatter between the average measurements .
we find strong evolution in the central density threshold for quenching , as defined by both a constant and evolving threshold in ssfr , decreasing by @xmath30.5 dex from @xmath1@xmath42 to @[email protected] .
similarly , while the threshold in central circular velocity where most galaxies are considered quenched is @xmath3300 km / s at @xmath1@xmath42 , this decreases to @xmath4150 km / s by @[email protected] .
we show that neither a high @xmath5 nor a compact galaxy size will uniquely predict quiescence , whereas a threshold in central density ( or velocity ) may be a more robust and unique observable signature when considering the overall galaxy population . however , we emphasize that correlations between structure and star formation do not prove a causal effect .
for example , it remains to be seen whether small scale structure ( at the scale of the stars ) or large scale parameters ( the scale of the dark matter halo ) dominate the physical processes which quench galaxies . while we have presented the average global trends of the ssfr with structural parameters ( @xmath9 , @xmath5 , @xmath57 , @xmath37 , and @xmath64 ) amongst a mass - complete sample of galaxies using high resolution _
imaging and deep spitzer / mips 24@xmath8 m imaging , future studies with the james webb space telescope mid - ir spectroscopic and photometric capabilities will yield robust measurements of sfr for _ individual _ galaxies across the star - formation sequence .
such studies will allow us to resolve the detailed trends within the star - forming population as a function of structure .
the authors wish to acknowledge sandy faber for helpful discussions .
kew gratefully acknowledges support by nasa through hubble fellowship grant # hst - hf2 - 51368 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 .
the authors are grateful to the many colleagues who have provided public data and catalogs in the five deep 3d - hst fields ; high redshift galaxy science has thrived owing to this gracious mindset and the tacs and the observatory directors who have encouraged this .
this work is based on observations taken by the 3d - hst treasury program ( go 12177 and 12328 ) with the nasa / esa hst , which is operated by the associations of universities for resarch in astronomy , inc .
, under nasa contract nas5 - 26555 . in order to remove the
well known correlations between both star formation rate and galaxy size with stellar mass and redshift , we fit the best - fit coefficients describing the log(sfr)-log(m@xmath7 ) relations in @xcite and the log(@xmath9)-log(m@xmath7 ) relation in @xcite .
the methodology and data itself are identical between the present study and these earlier works .
the measured log(sfr)-log(m@xmath7 ) relation is defined in equation 2 of @xcite . in figure
[ fig : sfr_bestfit ] , we fit the redshift evolution of the observed relations with simple least - square linear fits for the three polynomial coefficients , weighted by their respective uncertainties .
we therefore parameterize the stellar mass and redshift dependence of the star formation sequence as follows , in order to remove the average redshift and stellar mass dependence on galaxy size , we also subtract the following stellar mass and redshift dependent log(@xmath9)-log(m@xmath7 ) relation from figure [ fig : size_bestfit ] : with the best - fit redshift evolution of the normalization to the size - mass relation presented at the top of the left panel in figure [ fig : size_bestfit ] .
the slope of the size - mass relation for star - forming galaxies only is assumed to be roughly constant , with @xmath55=0.22 . given the average redshift and stellar mass of each bin across the size - mass plane , we use the two equations above to subtract the average correlations . | in this paper , we investigate the relationship between star formation and structure , using a mass - complete sample of 27,893 galaxies at 0.5@xmath0@[email protected] selected from the 3d - hst survey .
we confirm that , at fixed stellar mass , star - forming galaxies are larger than quiescent galaxies at all redshifts .
however , within the star - forming population , there is no relation between star formation rate and size : when dividing star - forming galaxies into quartiles based on their residual offsets in star formation rate , we find that the sizes of galaxies in the highest quartile are [email protected] dex larger than galaxies in the lowest quartile , a difference that is not significant . we also show that 50% of all star formation in galaxies at fixed stellar mass takes place within a narrow range of sizes ( 0.26 dex ) . taken together
, these results suggest that there is an abrupt cessation of star formation after galaxies attain particular structural properties . confirming earlier results , we find that the central stellar density within a fixed physical radius of 1 kpc is the key parameter connecting galaxy morphology and star formation histories : galaxies with high central densities are red and have increasingly lower specific star formation rates , whereas galaxies with low central densities are blue and have a roughly constant ( higher ) specific star formation rate at a given redshift interval .
we find remarkably little scatter in the average trends and a strong evolution in the central density ( or equivalently central circular velocity ) threshold correlated with quiescence .
this quenching threshold decreases by @xmath30.5 dex from @xmath1@xmath42 to @[email protected] .
neither a compact galaxy size nor a high @xmath5 are sufficient to assess the likelihood of quiescence for the average galaxy ; rather , it is the combination of these two parameters together with stellar mass that results in a unique quenching threshold in central density or velocity . |
since the beginning of the last century , when some astronomers started to study the matter content of our neighborhood @xcite , many evidences have been collected leading one to believe that the universe in which we live is mainly constituted by non - luminous matter , whose existence is inferred through its gravitational effects on the remaining constituents of the universe . this missing matter " , was dubbed dunkle materie " ( dark matter ) , by @xcite .
nowadays , we know , according to plank s data mission fitted with the @xmath0cdm model , that the universe is composed of 26.8% non - baryonic dark matter ( dm ) and 68.3% dark energy ( represented by the @xmath0 term ) , and just 4.9% ordinary matter @xcite . in this context , the role of gravitational lensing ( gl ) can hardly be overestimated .
gl provides important evidences in favor of dm existence in galactic clusters @xcite .
the other application of gl , following the idea by @xcite , deals with the searches for compact objects in the galactic halo and inside the milky way @xcite .
gl effects over a wide range of lens masses can give us the possibility of analyzing the dm substructure characteristics ; especially it can give us a clue to solve the `` missing satellite '' problem @xcite .
here we analyze the possibilities to study some properties of dm using mainly photometric information induced by gravitational microlensing effects .
following the terminology adopted by the gl community , at least three kinds of gl phenomena , characterized by different lens masses and typical timescales , exist : 1 . macrolensing and the weak lensing by galaxies or groups of galaxies;[type_i ] 2 .
mesolensing : lenses are globular clusters , dwarf galaxies or dm clusters with mass in the range 10@xmath1 to 10@xmath2 m@xmath3 @xcite;[type_ii ] 3 . microlensing
: lenses are stellar - mass objects.[type_iii ] as distinct from [ type_i ] and [ type_ii ] dealing with almost static situations or very slow processes , the characteristic timescales of microlensing events in the milky way are of the order of weeks and in case of extragalactic microlensing events , they are of the order of months or years .
this makes image brightness variation effects due to [ type_iii ] quite observable .
based on the lens mass range / timescales , this terminology determines at the same time the possible observational appearance of these effects , i.e. multiple correlated static images with signal arrival time delays in the macrolensing case , slowly - varying distorted images due to mesolensing , and high amplification events on lightcurves when the source crosses the caustic of a microlens .
dark matter can manifest itself at the three mentioned levels , namely : 1 .
large haloes of dm with mass greater than @xmath4 can play the role of macrolenses ; the most prominent example is the `` bullet cluster''@xcite ; 2 .
dm subhaloes of intermediate masses can play the role of mesolenses , causing the anomalous flux ratios , image distortions @xcite and additional time delays @xcite in extragalactic gravitational lens systems .
several candidates to show anomalous flux ratios are known today , e.g. , b2045 + 265 @xcite , rx j11311231 , b1608 + 656 , wfi 2033 4723 @xcite , b1938 + 666 @xcite , and we would especially like to stress here the famous gls q2237 + 0305 for its sharp high - amplification microlensing events @xcite .
clumps of dm with stellar masses can play the role of microlenses , leading to both photometric ( high amplification events ) and astrometric ( jump - like shifts of brightness centroid of the images of the microlensed source ) appearances .
the main difference between dm microlensing and the `` usual '' one due to stars or black holes lies in the non - negligible size of the dm clumps ( non - dm microlenses are always considered as point masses ) . in this paper
, we pay attention to the last item , i.e. microlensing by extended clumps of stellar masses .
we note that continuous observations by eros , ogle and other groups @xcite neither revealed any sign of these extended clumps in the galactic halo , nor provided any proof of such structures inside the galaxy .
the overwhelming majority of the light curves observed for galactic microlensing events are well described by the gravitational field of stars ( and , sometimes , of planetary mass objects ) .
therefore , it seems that there is no room for the stellar mass extended clumps .
however , in this paper we show that the light curves due to extended clumps of stellar mass objects can mimic the light curves caused by ordinary compact objects .
therefore , more detailed investigation is needed in order to rule out ( or to confirm ) the existence of these dm clumps that may either not be very numerous ( so that we do not have these clumps within our solar system ) or not so dense to be observable .
photometric signatures of dm substructure via gravitational microlensing have been widely discussed ; see @xcite and references therein .
there are also some investigations of variability of spectral line profiles @xcite .
@xcite derived amplification distributions for static gravitational macro - lensing with a non - constant surface mass density , which includes the cases of a stochastic system of gaussian clumps and clumps in the form of truncated singular isothermal spheres .
@xcite , and @xcite have considered a model of microlensing due to a non - compact neutralino star . in this paper
we use several other models of dm clumps to analyze the observational appearance of dm microlensing within different cosmological models .
we focus our attention on item [ type_iii ] to compare the observational appearances of microlensed sources due to point - like and finite - size dm clump deflectors and determine how one can distinguish an extended dm clump microlensing event from a `` regular '' one in case of galactic and extragalactic structures . the main question we want to answer concerns the possibilities to detect signals from these putative extended mass structures and to estimate the accuracy needed to characterize such effects mainly from observed light curves . in section 2 ,
we briefly describe the existing dm models . in section 3
, we propose a `` toy model '' , which describes , for special choices of the parameters , microlensing by various types of extended structures .
the model is based on the deflection angle @xmath5 , where @xmath6 characterizes the effective size of a clump core ] .
for different choices of parameters , this model reproduces the effects either due to point microlenses , or cdm minihaloes with a cusped density profile , or sidm clumps with a cored density profile . in section 4
we discuss the superpositions of extended and point - mass deflecting objects . in the case of isolated microlenses
these models can be used to describe microlensing events within the galaxy .
we perform a numerical comparison of these events considering different microlens models .
namely , we calculate the `` amplification curves '' , that is the dependence of the total amplification of a microlensing system versus time , as the lens moves with respect to the line - of - sight to the source .
we also pay some attention to the trajectories of the brightness centroid of the microlensed images , because the fast progress in the positional accuracy measurements gives us a hope to detect the astrometric signatures of gravitational microlensing in the near future . in section 5
we consider the observational behavior of clumps in extragalactic glss .
namely , we study the statistical effects of gravitational microlensing due to a stochastic system composed of extended dm clumps and point masses .
there is a number of papers dealing with statistical subjects of microlensing systems ; see , e.g. , @xcite .
our technique is essentially the same as that used by @xcite and @xcite , followed by a number of authors ; cf . especially considerations by @xcite dealing with brightness variations in connection to the problem of anomalous brightness ratios .
unlike in the previous works we deal with the autocorrelation functions ( acfs ) of the amplification curves for our concrete model of a microlensing system .
we generate an ensemble of amplification curves for a fixed input of randomly distributed clumps and point mass microlenses in the total optical depth ; this enables us to derive acfs for these amplification curves as a function of different clump contribution , size and in presence of an external shear .
finally , in section 6 we discuss the results .
dm microstructure still remains the subject of debates ; there are many hypotheses about dm particles @xcite .
cold dark matter is supposed to contain heavy particles , weakly interacting with each other and with baryonic matter ( axions or wimps ) , and warm dark matter ( wdm ) consists of light , fast - moving particles ( sterile neutrinos , champs , neutralinos or gravitinos ) , which tend less to form small - scale compact structures @xcite .
self - interacting dark matter can be considered as a particular kind of cold one with a nonzero impact distance for the interaction between dm particles @xcite .
recent cosmological n - body simulations within cdm @xcite and wdm models @xcite have shown that dark matter is not distributed in space homogeneously , instead it forms more or less compact structures surrounded by less dense continuously distributed dm .
the most massive structures have been recognized for quite a long time already : these are galactic and galaxy cluster dm haloes . but numerical simulations demonstrate that these massive structures are also inhomogeneous : they contain smaller compact substructures over a wide range of masses .
such massive haloes containing a hierarchical substructure of dark matter are called host haloes .
these haloes also contain smaller substructures , etc .
however , one should note that only the existence of the more massive members of this hierarchy of dm structures has been proven from the results of observations .
the lower estimated limit on the substructure masses depends on the particular kind of considered hypothetical dm particles , varying over a very wide range from @xmath7 for cdm up to @xmath8 for wdm @xcite . to clear out the situation and the terminology used to characterize masses and scales of dm substructures , we have summarized here in table [ table0 ] the used nomenclature .
.dark matter hierarchy within the cdm / sidm cosmologies . [ cols="<,<,<",options="header " , ] [ table0 ] dm substructure depends very significantly on the cosmological model . within cdm models ,
the substructure formation process can be described as `` bottom - up '' : the smallest dm clumps were formed first ( therefore the nanohaloes with masses @xmath9 are often referred as primordial ) and do not contain any finer substructure @xcite . then
, in the merging process more and more massive structures up to galactic and cluster scales ( 10@xmath10 - 10@xmath11 m@xmath12 were gradually formed . due to various damping scales
the lower limits on substructure mass can vary over a wide range : from 10@xmath13 - 10@xmath14 m@xmath3 for various kinds of wimps ( including neutralinos ) , down to 10@xmath15 - 10@xmath16 m@xmath3 for axions . the mass density distribution in dm structures within the cdm model is often assumed to follow the cusped nfw profile of @xcite .
the best accepted alternative to the cdm cosmology is the velocity - dependent sidm ( self - interacting - dark - matter ) .
this model predicts identical behavior of the dm substructure over various scales and identical mass fractions , thus , the only difference lies in the density profile of a dm clump .
the sidm clump has a core and its density follows a burkert profile @xcite , or pseudo - isothermal sphere @xcite , contrarily to the cuspy coreless nfw profile typical for cdm clumps .
the set of warm dark matter models differs completely from the two mentioned above when we keep in mind gravitational microlensing processes .
the substructure formation in wdm models is a complex hybrid ; its leading role is played by the `` top - down '' formation process @xcite , i.e. bigger haloes and filaments were formed first , and smaller structures formed later , as a result of fragmentation processes in wdm filaments @xcite .
the lower limit on the subhalo mass is significantly higher than in cdm / sidm ones : e.g. , @xmath17 @xmath18 for gravitinos and even @xmath19 for axinos @xcite .
thus in the wdm cosmology , we may predict the non existence of dm gravitational microlensing .
even despite that it was recently shown @xcite that the wdm structure formation rather follows the hybrid scenario than the `` top - down '' one , the effects of fragmentation and collapse play a significant role only in massive haloes formation , and thus the situation with dark matter induced gravitational microlensing in wdm cosmology is not significantly altered .
the dm subhalo mass function ( shmf ) @xmath20 , such that @xmath21 represents the space number of substructures with a mass in the range @xmath22 in the host dm halo with a mass @xmath23 is strongly sensitive to the dm context ( i.e. warm , cold , collisionless , repulsive and so on ) .
the shmf appears to provide the essential source of crucial information both in cosmology and elementary particle physics .
the numerical simulations like `` millenium '' @xcite or `` via lactea '' @xcite allow us to determine it only at higher subhalo masses ( mainly @xmath24 than typical microlens masses ( i.e. not greater than @xmath25 .
however for lower masses of the substructure we can use here the extrapolated mass function obtained by @xcite : @xmath26 where @xmath27 and @xmath28 .
thus , the total mass in substructures within the cdm model ( with a lower limit of @xmath29 @xmath30 is 52% , following @xcite .
this value is in good agreement with the 50% of dark matter in the galactic halo in substructures obtained from the hydrodynamical simulations by @xcite . using this formula
, one can easily find that for the lower limit of @xmath31 ( which can be considered as a reasonable limit for microlens mass ) it is 43% , and for @xmath32 ( let us consider this value as an upper limit on microlens mass ) it is 30% .
thus within the desirable range @xmath33 of the substructures we have 13% of the total dm mass ( i.e. @xmath34 is in compact stellar - mass subhaloes ) . if we take into account the fact that the larger subhaloes ( with mass greater than @xmath35 ) also contain finer substructures , this value can be even larger .
however , estimating the probability of the dm subhalo microlensing we should take into account that up to 90% of the primordial subhaloes had to be fully or partially disrupted by tidal forces of stars , thus the percentage of the clumped matter in the areas containing stars can not exceed 10% @xcite .
thus we can expect to trace the dm - induced high amplification microlensing events in the galactic dm halo ( as well as in extragalactic systems ) rather than in its luminous parts .
the probability of the galaxy stars being microlensed by a dm microhalo is much lower then .
that is why we can hardly expect to find compact dm clumps in the vicinity of the solar system .
however , at the same time , in the luminous parts of a galaxy we can face the situation when a star ( or a star - mass black hole ) is surrounded by a dense cloud , formed by dark matter of a former clump , disrupted by star / black hole tidal force and trapped by its gravity . if such a clump is dense enough and thus has an optical depth large enough ( overcritical convergence ) , we can observe the ucmh @xcite , and one of the possible results of gravitational microlensing by such an object looks like an eclipse of the source image .
the role of optical depth / convergence in microlensing processes was analyzed by @xcite . in the simplest case
such an object can be modeled up by a schwarzschild or chang - refsdal lens with overcritical convergence .
speaking about the milky way systems , we confine ourselves to the simplest cases of circular symmetric mass distributions though investigations of double stars or planetary systems are also important .
one can get some insight into the latter case from considerations of clump models with an external shear @xcite . in this section ,
we use the lens equation in its normalized form ; the distances will be expressed in units of a typical length scale @xmath36 , where for the case of the point mass @xmath37 or cored clump microlens @xmath38 ( in case of a point mass this is the radius of the einstein ring ) and for the cusped clump @xmath39 $ ] ; here @xmath40 is the distance between the source and the observer , @xmath41 the distance between the deflector and the observer , and @xmath42 the distance between the source and the deflector .
furthermore @xmath43 is the normalized angular source position , @xmath44 is the normalized angular image position , @xmath45 .
we remind here that ( after normalization ) both these vectors are defined in the lens plane . the normalized lens equation is : @xmath46 where the deflection angle @xmath47 is supposed to be normalized to a corresponding length scale
, @xmath48 stands for a convergence ( optical depth ) that can be due to , e.g. , a background dm ; @xmath49 .
let us write @xmath50 , where for the cusped clump we assume @xmath51 to provide the convergence in the integral of eq.([phi ] ) , and for the point microlens @xmath52 .
for the cored clump we use a generalized expression @xmath53 that formally covers the previous cases if we put @xmath54 .
therefore , we consider the lens equation @xmath55 the determinant of the lens mapping ( [ eq1z ] ) is @xmath56 .
it factorizes as follows : @xmath57 where @xmath58 the image parity is defined by the sign of @xmath59 .
the relative amplification when the source is far from the lens . ] of the image at @xmath60 is @xmath61 . taking the absolute value of both sides of eq .
( [ eq1z ] ) yields @xmath62 where @xmath63 , \quad y=|\mathbf y|$ ] . evidently , if @xmath64 is a solution of eq.([eq1mod ] )
, then the solution of eq.([eq1z ] ) is either @xmath65 or @xmath66 , where @xmath67 .
our first goal will be to estimate the number of solutions of eq.([eq1z ] ) , i.e. the number of microlensed images for different source positions , and to find the caustics of the lens mapping ( [ eq1z ] ) where the lensed image of a point - source gets infinitely amplified .
the problem is reduced to investigate the monotonicity of the r.h.s . of eq .
( [ eq1z ] ) and the zeros of the determinant @xmath68 in eq .
( [ det ] ) .
note that any root of @xmath69 , if it exists , is smaller than the root of @xmath70 ( for @xmath71 ) because of the monotonicity of @xmath70 and due to @xmath72 .
simple calculations yield @xmath73,\ ] ] whence for @xmath74 we infer the existence of the only inflection point @xmath75 of the curve @xmath76 and there is no inflection for @xmath77 . ) for the subcritical case ( @xmath78 : solid ; a , b , c ) and the overcritical case ( @xmath79 : dashed ; d , e , f ) . here
we show the subcritical cases ( a ) @xmath80 , ( b ) @xmath81 , ( c ) @xmath82 ; in case of ( b ) and ( c ) there may be an inflection point that does not affect the number of images .
the overcritical cases : ( d ) @xmath83 ; ( e ) @xmath84 ; ( f ) @xmath85.,width=336,height=259 ] let us first consider the case @xmath87 and the subcritical values of @xmath86 : @xmath78 .
further we denote @xmath88 . for the case @xmath89 , the first factor @xmath70 of the r.h.s .
( [ det ] ) equals zero for @xmath90^{1/2}$ ] .
this corresponds to a circular critical curve with a radius @xmath91 around the origin of the image plane , which is mapped onto an isolated caustic point @xmath92 in the source plane .
thus , there is a solution for eq .
( [ eq1z ] ) with @xmath92 . for @xmath93
the function @xmath70 is always positive . for the second factor @xmath69 in eq .
( [ det ] ) we also have the condition ] @xmath94 ( i.e. @xmath89 ) for the existence of @xmath95 such that @xmath96 .
indeed , under this condition , @xmath97 and @xmath98 have different signs .
this is a necessary and sufficient condition for a unique value of @xmath99 to exist on @xmath100 .
the proof of the uniqueness is somewhat different for @xmath101 and @xmath102 . for @xmath101
the function @xmath69 is monotonically increasing . for @xmath102
one should use eq .
( [ dydr ] ) and take into account the existence of the unique inflection point for @xmath103 : we observe that @xmath69 is monotonically increasing for @xmath104 ( therefore there can be only one zero in this region provided that @xmath89 ) ; after passing through a maximum at @xmath105 it decreases and it is positive : @xmath106 .
it is easy to see that @xmath95 , when it exists , is the radius of a circular critical curve that is mapped onto a circular caustic with radius @xmath107 .
there exist two additional images of a point source for the case @xmath108 .
the problem can be easily treated using the graph of @xmath109 ( fig .
[ sideb ] , b , c ) and taking into account the fact that @xmath110 is either equal to @xmath111 or to @xmath112 . for @xmath82
the function @xmath109 starting at @xmath113 is monotonically increasing , therefore , for any @xmath43 there is a unique solution @xmath114 of eq .
( [ eq1mod ] ) .
the ( unique ) solution of eq .
( [ eq1z ] ) is @xmath115 having positive parity .
for @xmath89 we have @xmath116 .
the function @xmath109 has only one maximum @xmath107 .
for @xmath117 the r.h.s .
of ( eq . [ eq1mod ] ) is a monotonically increasing function .
therefore , for @xmath118 there are three solutions for eq .
( [ eq1mod ] ) : two solutions @xmath119 , which yield two solutions of the vectorial lens equation ( [ eq1z ] ) @xmath120 ( @xmath121 with positive parity , @xmath122 with negative parity ) ; and one solution @xmath123 yielding the solution @xmath124 with positive parity for eq .
( [ eq1z ] ) .
to sum up , for @xmath78 we have a unique image for the case @xmath82 ; in addition , for @xmath125 there is an image for @xmath126 exists , if @xmath127 , and if @xmath128 .
] at @xmath113 .
all images have a finite brightness . in the opposite case , for @xmath81
there is a circular caustic with radius @xmath129 and there are three images , if the source is inside this caustic @xmath118 , and one image , if @xmath130 .
two images acquire an infinite amplification when @xmath131 and then disappear after crossing @xmath132 .
in addition , there is a caustic point @xmath125 corresponding to a ring image like the einstein ring emerging in case of a point mass lens .
the caustics with source tracks and corresponding amplification curves are shown in fig.[two_lightcurves ] . .
the upper curve ( dashed ) : the point source moves along the straight line with the impact parameter @xmath108 yielding two caustic crossings . the lower curve ( solid ) : larger @xmath130 , no caustic crossings . in the right upper corner : the trajectories with respect to the circular caustic of the lens.,width=288,height=240 ] for @xmath79 ( @xmath87 ) , let us first consider the case @xmath101 .
in this case in equation ( [ eq1mod ] ) @xmath133 $ ] is a monotonically increasing function ( see fig .
[ sideb ] , d , dashed curve ) .
its derivative is @xmath134 .
the lens mapping has no caustics and critical curves ; for any @xmath43 there is always a unique solution @xmath64 for eq.([eq1mod ] ) and a unique solution @xmath135 ( positive parity ) for the vectorial lens equation ( [ eq1z ] ) . for @xmath79 ,
@xmath102 there is an inflection point of the function @xmath136 according to eq .
( [ dydr ] ) .
taking into account the sign of this inflection , we see that there exists a minimum for @xmath137 at @xmath105 , which is equal to @xmath138 then new roots of @xmath137 appear when @xmath139 if @xmath140 , the function @xmath109 is monotonically increasing ( not shown in fig . [ sideb ] ) .
let these new roots be @xmath141 and @xmath142 with @xmath143 , @xmath141 being a point of a local maximum of @xmath109 and @xmath142 being a point of a local minimum ( see fig .
[ sideb ] , e , dashed ) .
the roots correspond to the radii of the critical curves , and @xmath144 are the radii of the circular caustics . for @xmath145 eq.([eq1mod ] ) has three solutions @xmath146 correspondingly with a positive , negative and positive parity yielding three lensed images of a point source , and there is only one image with positive parity otherwise . the solution @xmath147 is extended for @xmath148 and the corresponding image has an infinite amplification when @xmath149 ; in this case @xmath147 and @xmath150 tend to @xmath151 and then disappear after @xmath152 crosses the value @xmath153 . the solution @xmath154 is extended to all values of @xmath155 ; it has analogous properties near the other caustic for @xmath156 . in this case @xmath150 and @xmath154 tend to @xmath142 and then disappear as @xmath152 decreases .
we have for the solutions of eq.([eq1z ] ) , @xmath157 . here
we have @xmath158 . for @xmath159 , the function @xmath160 for @xmath161 ; @xmath162 is the radius of a critical curve corresponding to the caustic point @xmath125 . for @xmath159 and @xmath101 ,
the modulus @xmath109 increases for @xmath163^{1/\a}$ ] and then decreases to zero for @xmath164 .
the value @xmath165^{1/\a-1}$ ] is the radius of a circular caustic .
the behavior of the graph of this function is roughly the same as that illustrated by the solid curve in fig .
[ sideb ] , b. there are three solutions for eq .
( [ eq1mod ] ) with @xmath118 : namely , two solutions @xmath166 for eq .
( [ eq1mod ] ) , @xmath167 corresponding to images at @xmath120 ( with positive ( @xmath121 ) and negative ( @xmath122 ) parity , and one solution @xmath168 corresponding to image at @xmath124 with positive parity .
the latter solution remains valid for @xmath130 . for @xmath159 , @xmath102 ,
the function @xmath169 is negative for @xmath170 , the modulus @xmath109 decreases along this interval and it increases for @xmath171 .
the behavior of @xmath109 is represented by the curve ( a ) in fig .
[ sideb ] . in this case
there are always two solutions of eq .
( [ eq1mod ] ) for @xmath172 corresponding to images at points @xmath173 , @xmath174 . for @xmath79 and @xmath101 the function @xmath109 increases for all values of @xmath175 starting from zero ( like the dashed curve ( d ) in fig .
[ sideb ] ) .
there is only one solution and one image for all @xmath152 values . for @xmath79 and @xmath102
the function @xmath109 shows a minimum at @xmath176^{1/\a}$ ] ; the minimum is @xmath177^{1 - 1/\a}$ ] ( the radius of a circular caustic ) .
there are two solutions for eq .
( [ eq1mod ] ) if @xmath130 and no solutions if @xmath108 .
this case is represented by the dashed curve ( f ) in fig .
[ sideb ] . in this subsection
, we list several situations whenever a simple analytical treatment is possible .
the point mass lens model ( schwarzschild lens ) is well known , it corresponds to @xmath52 , @xmath54 , @xmath178 .
the solutions for the image positions are @xmath179 and the total amplification of the two lensed images is @xmath180 . for a single sidm cored clump ,
let us assume @xmath52 , @xmath178 : critical curves for such a system exist only when @xmath181
. under this condition , two circular critical curves exist .
the first one has a radius @xmath182 corresponding to the caustic point @xmath183 ( the same as for the point - mass lens ) , and the second one has a radius @xmath99 such that @xmath184 ( this expression is positive for @xmath181 ) , corresponding to a caustic with a radius @xmath185 ) . in case of a cored clump microlens ,
the lens equation yields : @xmath186 the roots can be found using the cardano method .
for the lower plus , two roots must be taken in the interval @xmath187 for @xmath108 , and for the upper minus , one must take the root for @xmath117 . if @xmath188 ( low - density clump ) , no caustic exists and only one lensed image can be seen . for the ucmh model , it is interesting to consider an analytic solution for the overcritical case @xmath189 ( @xmath190 ) .
in this case there exists a critical curve @xmath191 and the radius of the corresponding caustic is described by @xmath192 .
two images for @xmath130 are produced at the positions : @xmath193/[{2y(\sigma-1 ) } ] .
$ ] the total amplification of these two lensed images is @xmath194/[y\sqrt{y^2 - 4 ( \sigma-1 ) } ] .$ ] for @xmath108 there are no images and we have an `` occultation '' . the lightcurves corresponding to this effect , both for point - like and continuous sources , were shown in our previous work @xcite . for the case of the cusped lens model
, the solutions can be written analytically for @xmath195 .
the lens equation @xmath196 has two solutions . the first solution must be written separately for @xmath197 and @xmath198 : @xmath199 ^ 2 } { \left(\xi+\sqrt{\xi^2 - 1}\right)^{2/3}},\,\ , \xi>1;\ ] ] and @xmath200 , \,\ , \xi<1.\ ] ] the second solution is @xmath201 ^ 2 } { \left(\xi+\sqrt{\xi^2 + 1}\right)^{2/ 3}}\ ] ] for any value of @xmath152 .
the magnification curves for these images can be found using the formula ( [ det ] ) : @xmath202 .
how can one distinguish between the different lens models discussed above on the basis of observations ?
of course , the most correct comparison must come after the fitting of concrete observational data .
however , let us first note the qualitative differences mainly arising from the topological properties of the corresponding lens mappings , i.e. the existence of the critical curves , caustics and number of images of a point source . from the above discussion
it is easy to see that , for different parameters of the source / lens motion , the caustic intersections can occur leading to appearances / disappearances of the lensed images .
these events could lead to such observable effects as high amplifications of the lensed image flux , `` occultations '' and sudden jumps of the average image positions ( image centroids ) .
there exist many observational data compiled by eros , ogle and other groups @xcite hunting for microlensing events caused by galactic objects .
some of these events show a complicated behavior like those characteristic of caustic crossings .
this is typically related to the existence of double star systems or planetary systems @xcite .
however , most typical haes @xcite can be interpreted using the isolated point mass lens model .
it is important to note that there exist some high amplification events with no detector identified . on the other hand , for most of the events
the deflector - star has been identified ; however this case does not rule out the situation ( discussed below ) when an extended clump surrounds the star .
correspondingly , at this stage we concentrate on the most simple models of extended microlenses ; and we choose the parameters of their motion without caustic crossings .
namely , we consider the clump models with two lensed images that will be compared with the `` fiducial '' model of the usual point mass microlens ( the schwarzschild lens ) . for galactic systems
we assume the background continuous matter density to be @xmath178 .
the critical curves of the two - image clump models appear to be very similar in case of the schwarzschild microlens and the cusped clump with @xmath203 and @xmath102 . in these cases
there is one circular critical curve of unit radius ; and also , one caustic point at @xmath125 exists .
the case of a cusped clump with @xmath77 is topologically different , however it will be also difficult to distinguish its light curve from that due to the schwarzschild lens . below we show amplification curves generated with different models corresponding to a straight line motion of the microlensing system with respect to the line - of - sight towards the remote point source .
it turns out that the shape of the amplification curves induced by two - image clumps are qualitatively similar to the ones induced by a schwarzschild microlens , and thus they can hardly be distinguished from one another . indeed , by an appropriate choice of the impact parameter and the velocity with respect to the line - of - sight
, the schwarzschild lens light curve can be well fitted by microlensing due to a cusped clump ( see fig . [ c - cusp ] ) . for a wide range of parameters that may be considered as typical for this problem ,
the difference between the amplification curves on a plot is sometimes not visible to the eye .
one can extract an additional information from the image centroid ( ic ) tracks in the reference frame of the source ( see fig .
[ c - cusp ] , lower panel ) .
this , however , requires an accuracy for the positioning measurements at the microarcsecond level ; this is typical for microlensing by galactic objects and is not accessible at present .
e.g. , the modern accuracy achieved with hst ( wfpc3 camera ) is around 20 - 40 microarcseconds @xcite . )
( @xmath204 corresponds to a pure clump , @xmath205 is the impact parameter of the clump center with respect to the line - of - sight , its transverse velocity @xmath206 ) .
the impact parameter and transverse velocity of the point mass microlens are derived from fitting the amplification curve for the clump model . the image centroid trajectory ( in the rest frame of the source ) in case of the schwarzshild lens is rescaled in order to clearly indicate the difference between the models.,width=288,height=192 ] ) ( @xmath204 corresponds to a pure clump , @xmath205 is the impact parameter of the clump center with respect to the line - of - sight , its transverse velocity @xmath206 ) .
the impact parameter and transverse velocity of the point mass microlens are derived from fitting the amplification curve for the clump model .
the image centroid trajectory ( in the rest frame of the source ) in case of the schwarzshild lens is rescaled in order to clearly indicate the difference between the models.,width=288,height=192 ] , @xmath207 ( upper panel ) @xmath208 , and @xmath209 ( lower panel ) .
the impact parameter of the microlens with respect to the line - of - sight is @xmath205 .
the curves are fitted by means of point mass microlens models ( small circles ) .
, height=192 ] , @xmath207 ( upper panel ) @xmath208 , and @xmath209 ( lower panel ) .
the impact parameter of the microlens with respect to the line - of - sight is @xmath205 .
the curves are fitted by means of point mass microlens models ( small circles ) .
, width=288,height=192 ] for cored clump models with @xmath210 , @xmath211 and @xmath212 .
the impact parameter of the microlens with respect to the line - of - sight is @xmath205.,height=192 ] for cored clump models with @xmath210 , @xmath211 and @xmath212 .
the impact parameter of the microlens with respect to the line - of - sight is @xmath205.,width=288,height=192 ] thus , we proceed in more detail with the differences between the amplification curves corresponding to different microlens models .
note that we performed a fitting of the amplification curves in figs .
[ c - cusp],[comp2],[comp3 ] so as to provide the best overlapping near the maximum of the curves ( where they can be well approximated by a parabola , and where one can expect a good measurement accuracy ) .
careful inspection of these figures shows that there is a slight deviation between the curves in the wings of the separate haes . in this connection , to compare the different microlens models , we proceed as follows . for a moving extended microlens characterized by an impact parameter @xmath213 with respect to the line - of - sight , and a transverse velocity ( assumed to be @xmath206 )
, we generate its amplification curve @xmath214 .
then we look for the best fit near the maximum of this curve using the schwarzschild lens model light curve ( @xmath215 ) ; see appendix [ a ] for details .
note that as distinct from dealing with the real observational data we have a simpler problem . in the real case , the fitting parameters , besides the transverse velocity and the impact parameter of the schwarzschild lens ( which are different from the assumed parameters @xmath216 of the clump+point mass complex ) , are : the time of maximum amplification and the maximum intensity of the image ( or the flux when the lens is far from the line - of - sight ) . in our case of the artificial amplification curve ,
the latter two parameters are fixed .
thus the difference between the two amplification curves is : @xmath217^{-1}|\mu(t)-\mu_{schw}(t)| \right\}.\ ] ] this is used to compare the amplification curves due to a somewhat more complicated model which corresponds to a combination of a point mass microlens and an extended clump .
it is natural to consider a situation where the cusped or cored dm clump is formed in the same region as where the point mass is situated ( and vise versa ) .
the case of a pure clump microlens and the schwarzschild lens represent two extreme cases of this situation . in this connection
we consider a microlens model that is represented by the following lens equation : @xmath218 where the coefficient @xmath219 describes the relative contribution of the clump , @xmath220 represents that of the point mass lens , and we omitted the contribution due to the background optical depth @xmath221 ; @xmath222 , @xmath223 .
the amplification is @xmath224 , where the determinant @xmath68 is also given by the product @xmath225 with @xmath226 @xmath227 furthermore , we assume such numerical values for the parameters of the microlens as to provide the qualitative behavior of the lens mapping ( [ p_cl ] ) to be like that of the schwarzschild lens with two lensed images , and so as to provide a considerable amplification ( up to 10 and higher ) , which can be the best to see the signals of the extended microlens structure
. we shall then estimate the accuracy of measurements needed to differentiate between the lightcurves of these models .
the dependence of this difference upon the contribution @xmath228 of the clump in the lensing complex `` cored clump + point mass '' is shown in fig .
[ diff_1 ] for some values of @xmath229 and @xmath6 . of the amplification curves for a different cored clump contributions @xmath228 in the microlensing complex `` point mass + clump '' .
upper panel : @xmath230 , impact parameter @xmath231 ( solid curves ) , @xmath232 ( dashed curves ) , the clump size @xmath211 ; the amplification is within the limits @xmath233 . for @xmath234
we have almost the same curves as for @xmath231 , but the amplification in this case varies from @xmath235 to @xmath236 .
lower panel : the same values of @xmath237 , @xmath231 , @xmath212 ( solid ) , @xmath238 ( dashed ) .
the curve with @xmath239 , @xmath212 is superimposed on the curve with @xmath210 , @xmath238 and it is omitted . amplification is within @xmath240.,width=267,height=228 ] of the amplification curves for a different cored clump contributions @xmath228 in the microlensing complex `` point mass + clump '' .
upper panel : @xmath230 , impact parameter @xmath231 ( solid curves ) , @xmath232 ( dashed curves ) , the clump size @xmath211 ; the amplification is within the limits @xmath233 .
for @xmath234 we have almost the same curves as for @xmath231 , but the amplification in this case varies from @xmath235 to @xmath236 .
lower panel : the same values of @xmath237 , @xmath231 , @xmath212 ( solid ) , @xmath238 ( dashed ) .
the curve with @xmath239 , @xmath212 is superimposed on the curve with @xmath210 , @xmath238 and it is omitted . amplification is within @xmath240.,width=267,height=228 ] in case of the configuration of a point mass and a cusped clump we assume @xmath241 , @xmath54 .
the relative differences between the amplification curve of this complex and that of the fitted schwarzschild lens are shown in fig .
[ diff_2 ] . of the amplification curves for different cusped clump contributions @xmath228 in the microlensing complex `` point mass + clump '' . here
for the impact parameter @xmath231 ( solid ) the amplification is within @xmath243 ; for @xmath205 ( dashed ) the amplification is @xmath244 .
for @xmath234 we have practically the same curves as for @xmath231 , but the amplification varies from @xmath235 to @xmath245.,width=267,height=228 ]
in a typical extragalactic gls we have several macro - images of one quasar .
comparison of the amplification curves of these images makes it possible to separate the proper brightness variations of the quasar and to derive variations due to the microlensing processes .
contrary to the case of the galaxy , because of the considerable microlensing optical depth in extragalactic systems , instead of an isolated point mass or putative dm clump , we must deal with an aggregate of masses in the lensing galaxy . in this case
we have a complicated caustic network generated by unknown masses with unknown positions .
therefore , the problem takes a statistical shade @xcite .
observationally , the main source of information in a concrete gls is at present an effective microlensed light curve of every image ( i.e. , separated from proper brightness variations that are intrinsic to the source ) , which arises because of the source motion .
as before , for a theoretical treatment we deal with the amplification curves , that is the dependence of the amplification coefficients upon the time .
our aim is to estimate the statistical effect of the extended masses ( clumps ) on the autocorrelation functions of the amplification curves . to compare the amplification curves in microlensing systems with a different content of these clumps , we consider a simple model of stochastic point and extended masses .
we confine ourselves to a special case of the cored clumps according to eq .
( [ eq1z ] ) with @xmath239 ( cf .
@xcite ) . an external shear owing to the average field of the lensing galaxy
will also be taken into account .
thus , we use the lens equation for n masses in one lens plane @xmath246 where @xmath247 is the two - dimensional external shear matrix ( the optical depth of the background continuous matter is taken to be zero ) , @xmath248 is a characteristic size of the @xmath249-th cored clump with mass @xmath250 , @xmath251 is the position of its center on the lens plane , and @xmath252 its einstein ring radius : @xmath253 . obviously , for @xmath254 we have ordinary point microlenses with mass @xmath250 . in our simulations ,
the microlens masses and positions were chosen in a random way .
the positions @xmath251 were distributed homogeneously inside a circle , which was big enough to minimize the boundary effects ; also , to check the result convergency we considered different sizes of the circle .
the mass distribution followed the salpeter s law @xcite with a power - law index @xmath255 within the mass range @xmath256 m_{\odot}$ ] . in every set of numerical experiments , the input parameters ( that were controlled to be the same in all the realizations of the set ) were as follows : the total optical depth of the microlensing field @xmath257 , the optical depths @xmath258 , @xmath259 of the point masses and the extended clumps ( @xmath260 ) , and the relative size of the clumps @xmath261 . to introduce the clumps into the microlensing field ,
we replaced some randomly chosen part of point masses by extended clumps without changing their positions @xmath262 , masses @xmath250 ; the size parameter was @xmath263 .
then , for each realization of the microlensing field we obtain an amplification map by means of the ray shooting method combined with our gpu - enabled microlensing c++ code based on the hierarchical tree algorythm @xcite .
convolution of the amplification map with the normalized brightness distribution over the initial source @xmath264 yields the amplification coefficient @xmath265 .
the mathematical representation of this procedure can be written as @xcite @xmath266 dx_1 dx_2,\ ] ] where @xmath267 is the position of the source center in the source plane projected in the lens plane and @xmath268 represents the lens mapping @xmath269 .
we used a gaussian model for @xmath270 ; the source half - brightness radius was @xmath271 , its motion has been assumed to be uniform . due to the motion of the source we have an amplification curve @xmath214 . having a large number of realizations of the microlensing field and the corresponding amplification curves @xmath214 , we have calculated the acfs @xmath272[\mu(t+\tau)-\mu_0]>,\ ] ] where @xmath273 ^ 2>},\ ] ] the brackets @xmath274 denote the averaging over all the realizations of the amplification curves for the fixed value of the optical depth of continuous masses @xmath259 and point masses @xmath258 . to consider the possible observational manifestations of the extended microlenses , their size @xmath275 has been chosen to be comparable with a typical einstein ring radius @xmath276 of the microlenses ; otherwise in the limiting situations we approach to either the well - known case of a continuous background matter for @xmath277 , or we have a standard microlensing of point masses . here
we present the results for @xmath278 , and @xmath279 .
the parameter @xmath259 has been varied from zero to @xmath280 .
the length unit corresponds to @xmath281^{1/2}=1 $ ] .
furthermore , @xmath282 and @xmath283 are in units of @xmath284 ( i.e. assuming a velocity @xmath206 ) ; the normalization to @xmath18 is just conventional .
, title="fig:",width=302,height=264 ] -3 mm -3 mm , here and below @xmath285 , and the size parameter is @xmath286.,title="fig:",width=302,height=264 ] -3 mm -3mm.,title="fig:",width=302,height=264 ] -3 mm -3 mm , @xmath279 , @xmath287 and for @xmath288.,title="fig:",width=302 ] -3 mm -3 mm , @xmath289 , @xmath290 and for @xmath288 .
, title="fig:",width=302 ] -3 mm the examples of the amplification maps in the source plane are shown in figs .
[ map_1 ] - [ map_3 ] .
the amplification distributions for various fractions of clumps ( and the same @xmath291 , @xmath287 ) noticeably differ one from another .
as can be expected , the presence of the clumps makes this distribution smoother .
the presence of non - zero @xmath292 ( fig .
[ map_3 ] ) stretches the maps in the direction of the shear . for every set of parameters @xmath293 we generated typically several hundred maps and the corresponding amplification curves . using the ensemble of these curves , the averaging procedure yields acf @xmath294 .
these functions are shown in figs .
[ autocor ] , [ auto - gamma ] for different fractions of point masses and clumps .
we note that the behavior of @xmath294 and @xmath295 as functions of @xmath259 appears to be rather complicated .
e.g. , for @xmath287 we see a non - monotonic variation of the curves as @xmath259 changes ( fig .
[ autocor ] ) .
this effect is not observed for non - zero values of the shear ( see the case @xmath290 in fig .
[ auto - gamma ] ) .
we have analyzed the observational signatures of gravitational microlensing in presence of extended masses ( clumps ) , which can represent , for example , hypothetical structures arising in cold dm models as a result of clustering . to produce significant effects due to the presence of the clumps , the masses of the clumps
were chosen of the same order as those of the point masses , and the sizes of the clumps were of the order of several typical einstein radii .
we compared the artificial amplification curves ( essentially , the light curves ) arising due to a relative motion of the microlens with respect to a remote source .
we have proposed simple models that describe circular symmetric cusped or cored clumps and their combinations with point masses .
we do not consider these idealized models as the only possible ones , rather , our aim was to look at some `` toy problems '' in order to see the main qualitative features accompanying the microlensing process in our galaxy and beyond in presence of small - scale dm clustering , and to estimate the order of the effects involved .
we believe that our main findings on this matter will be preserved at a qualitative level for more complicated models of the dm clumps , though an exact final answer will be the subject of a separate investigation . in case of galactic microlensing ,
the case of isolated circular symmetric microlenses of section [ extended_clump_models ] has led to a detailed analytical treatment ; here the regions of parameters of the lens mapping yielding the occurrence of one , two or three images were determined .
we note that any structures yielding changes in the number of images ( cf .
caustic crossings , occultations ) are possible only in case of a considerable local dark matter density , either near the clump centers or around the point masses .
the most vivid example is an eclipse - like effect with the total disappearance of the lensed images .
though , we do not believe this case to be realistic for the galaxy conditions , we can not exclude it at all .
it may be difficult to distinguish such complicated events from those related to double stars or planetary systems ; here a detailed investigation of the light curves of concrete high amplification events ( haes ) is required .
we considered relatively simple models and events .
we found that the light curve of a hae in presence of the extended microlens ( the isolated clump or its superposition with the point mass ) can be well mimicked by a light curve of the schwarzschild lens with appropriately chosen mass and parameters associated with its motion .
nevertheless , in case of such a fitting , there remain some residuals in the wings of the hae light curve that can be used to distinguish between the microlens models .
these residuals have been estimated through the fitting of the light curves of our clump model by those of the point mass lens ( sections [ estimations ] , [ combination ] ) ; these results determine the required level of photometric accuracy needed to detect the signals from the extended clumps .
this level can be estimated from figs .
[ diff_1 ] , [ diff_2 ] .
in particular , for cored clump + point mass system with @xmath296 ( that is , the core size and impact parameter are equal to the einstein ring radius ) and average values @xmath297 the light curve deviation from that of the ordinary point mass lens can be detected if the photometric accuracy is at the @xmath298 level .
so is also the case of the cusped clump , and the requirements to the accuracy in case of a pure cusped or cored clump ( @xmath204 ) are even lower .
we also note that the degeneracy in the choice of the model can be reduced by simultaneous astrometric observations of the source image centroid motion ( see fig .
[ c - cusp ] ) .
if this information is available , this may help to restrict the parameters of the microlensing system .
though this does not mean that all of them can be determined uniquely , not to mention the question of uniqueness of the microlensing model itself . in any case , there is a region of the clump parameters that induces a considerable difference between the light curves , which is within the reach of modern photometric measurements .
therefore , we see that at least some extended microlens models can be , in principle , either confirmed or rejected from the use of high - quality observations .
however , in real observations of microlensing by galactic objects , the situation will be aggravated by the possible presence of unobservable planets and star companions .
then in this case a more detailed treatment of concrete high amplification events is also necessary to compare the models that involve binaries and planets with those dealing with extended clumps . in extragalactic glss
the situation is much more complicated .
the view of amplification maps and the caustic network generated by simple point mass microlenses ( i1 ) seems to be rather different from those in presence of a considerable contribution of extended clumps ( i2 ) .
nevertheless , individual light curves in case ( i1 ) can also include high amplification and de - amplification effects which have an analogous look as in case ( i2 ) .
the other complication in extragalactic glss is that owing to the non - negligible source size since both the light curves and the centroid brightness trajectory will be smoothed out significantly .
to have an idea about microlensing in cases ( i1 ) and ( i2 ) , we considered statistical models of many - body microlensing systems that involve both point and extended microlenses .
the latter have been represented by a special case of the cored clumps .
considerations of section [ statistical ] have much in common with considerations by @xcite that , among other questions , deal with the issue of microlensing by extended structures .
in particular , @xcite pay attention to microlensing brightness variations in presence of a variable smooth matter depending on the parity of the close macrolens images .
these papers , however , are mainly aimed at investigating the problem of anomalous flux ratios and therefore they deal with other mass and spatial scales .
our simulations deal with extended microlenses ( e.g. , dm clumps ) of stellar mass , with sizes comparable to that of corresponding einstein rings in order to have a noticeable effect in the light curves .
a technical difference of our simulations owes to the simplified model of the lens mapping in presence of the clumps , which enables us to speed up the numerical calculations of acfs with a rather modest computer hardware . for every choice of parameters that characterize the content of the clumps , their sizes and masses
, we generated an ensemble of amplification curves assuming a uniform spatial distribution of microlenses in the lens plane . using these curves we built acfs of the amplification curves for a set of optical depths of extended clumps and point masses .
joint use of microlensed flux variations and acfs may provide a means for determining the clump fraction and the other parameters of the lensing system .
indeed , there is a notable difference between the acfs for different values of the clump contribution . in this view
, we note the remark by @xcite that the microlensing variations appear to be `` less dramatic for extended subhalos than for point microlenses '' .
this also has been mostly observed in our simulations , though with some exceptions .
one could expect that by and large the extended structures make acfs less steep and therefore , this effect shall increase the corresponding correlation length , and this typically is the case for @xmath299 and @xmath300 $ ] ( @xmath291 ) ) , but also with some exceptions showing a complicated dependence of acfs upon @xmath259 ( figs .
[ autocor ] , [ auto - gamma ] ) .
however , detection of these effects requires sufficiently long observations of the light curves in real glss .
the relevance of our results to extragalactic systems is connected also with the occurrence of a significant optical depth @xmath259 .
this depends on the surface mass density along the line - of - sight which is not well known .
for very rough estimations one can use the masses and sizes of the galactic haloes as compared to those of the stellar population .
the density of dm in the universe is five times higher than the baryonic one and the percentage of the luminous matter in galaxies is even lower by one order .
then , it may be quite possible to have the surface mass density of the extended clumps of some gls of the same order as that of the point masses ( stars ) .
moreover , despite the mass - luminosity ratio for typical galaxies is known to be at the level of @xmath301 , for some spherical dwarf galaxies it is @xmath302 @xcite .
to sum up , we studied microlensing of a remote source by point and extended masses , the latter representing hypothetical results of cold dm clustering . in case when the masses of the extended clumps are comparable to those of the point objects ( stars ) and their characteristic sizes are of the same order as the einstein radius of the stars , our findings are as follows : * we proposed a simple circular symmetric model of microlensing by isolated dark matter clumps that allowed us a detailed investigation of the corresponding lens mapping . * in case of a single point source motion , we built amplification curves of the lensed images ; this is done in case of microlensing by the isolated clumps and/or concentric clumps and point masses . for a wide range of model parameters
, we estimated the residuals after fitting these amplification curves with those due to the single point mass microlensing .
* the results allow one to estimate the photometric accuracy needed to differentiate the model with the extended clumps and without them .
we note that a number of model parameters can be ruled out under the level of photometric accuracy @xmath303 . * in case of extragalactic gravitational lens systems , we built autocorrelation functions of the amplification curves for different clump contributions .
we found a noticeable difference between acfs in case of a considerable value for the clump contribution and without the clumps .
we thank the anonymous referee for her / his helpful comments .
ef acknowledges the ago / aeos institute of astrophysics and geophysics of lige university for their warm hospitality and providing the technical facilities for this work .
ef and js acknowledge support from the esa prodex programme `` gaia - dpac '' , and from the belgian federal science policy office .
viz acknowledges partial support from the state fund for fundamental research of ukraine , project @xmath30464/42 - 2015 . this work has been supported in part by the scientific program
astronomy and space physics " , taras shevchenko national university of kyiv .
99 ade p. a. r. et al .
2013 a&a * 1303 * , 5062 .
alcock c. , akerlof c. w. , allsman r. a. , et al . , 1993 , nat ,
* 365 * , 621 .
alexandrov a.n .
, zhdanov v.i . , 2011 , mnras , * 417 * , 541 .
aubourg e. ; bareyre p. ; brhin s. e. et al . , 1993 ,
nat , * 365 * , 623 .
baryshev yu .
, bukhmastova yu .
1997 , astron.rep . , * 41 * , 436 . berezinsky v. , dokuchaev v. , eroshenko yu .
2008 , phys.rev.d , * 77 * , 083519 .
berezinsky v. , dokuchaev v. , eroshenko yu .
2013 , jcap , * 11 * , 059 .
brada m. , clowe d. , gonzalez a. h. et al .
2006 , apj , * 652 * , 937 .
chen j. et al .
2007 , apj , * 652 * , 52 .
chiba m. 2002 apj , * 565 * , 17 .
clowe d. , gonzalez a. , & markevitch m. 2004 , apj , * 604 * , 596 clowe d. , brada m. , gonzalez a. h. et al .
2006 , apjl , * 648 * , l109 .
congdon a. , keeton c. , nordgren c. 2010 , aph.j . , * 709 * , 552 .
contini e. , de lucia g. , borgani s. 2012 , mnras , * 420 * , 2978 .
dalal n. , kochanek c. s. 2002 , apj , * 572 * , 25 .
del popolo a. 2014 , int.journ .
d , * 23 * , n3 , 1430005 .
del popolo a. , lima j. a. s. , fabris j. c. , rodrigues d.c .
2014 , jcap , * 4 * , id.21 diemand j. , moore b. , stadel j. 2004 , mnras , * 352 * , is . 2 , 535 .
diemand j. , moore b. , stadel j. 2005 , nature , * 433 * , 389 - 391 .
dobler g. , keeton c.r . , wambsganss j. 2007 , mnras , * 377 * , 977 - 986 .
fedorova e. , del popolo a. , zhdanov v.i . ,
alexandrov a.n .
, sliusar v. 2014 , rencontres de moriond 49th .
cosmology , eds : e.auge , j.dumarchez and j. tran thanh van al .
la thuile 2014 proceedings , 407 .
jeans j. 1923 , mnras , * 84 * , 60 .
hisano j. , inoue k.t . , takahashi t. 2006 , phys.lett.b . , * 643 * , 141 . kains , n. ; street , r. a. ; choi , j .- y . ; han , c. ; udalski , a. et al .
2013 , a&a , * 552 * , id.a70 , 10 pp .
kapteyn j.c .
1922 , aph.j . *
55 * , 302 .
keeton c.r . , moustakas l.a .
2009 , apj , * 699 * , 1720 .
klypin a. , kravtsov a. , valenzuela o. , prada f. 1999 , aph.j .
, * 522 * , 82 .
knebe a. , arnold b. , power c. , gibson b.k .
2008 , mnras , * 386(2 ) * , 1029 .
knebe a. , devriendt j. , gibson b. , silk j. 2003 , mnras , * 345 * , 1286 .
kormendy j , freeman k.c .
2003 , iau symp.220 eds : s.d.ryder et al .
san francisco : astron .
, 377 ; astro - ph/0407321 . lee s.k .
ando a. , kamionkowski m. 2009 , jcap , * 07*. markevitch m. , gonzalez a. , clowe d. et al .
2003 , aph.j . ,
* 606 * , 819 .
macleod c. et al .
2013 , aphj .
, * 773 * , 35 .
mao s. , jing y. , ostriker j. p. , weller j. 2004 , apj , * 604 * , l5 .
mckean j. p. et al .
2007 , mnras , * 378(1 ) * , 109 .
metcalf r.b . , madau p. , 2001
, apj * 563 * , 9 .
metcalf r.b .
, moustakas l. a. bunker a. j. , parry i. r. 2003 , aph.j .
* 607(1 ) * , 43 .
moore b. et al . 1999 ,
aph.j . , * 524 * , l19 .
navarro j. , frenk c. , white s. 1996 , apj , * 462 * , 563 .
neindorf b. , 2003 , a&a , * 404 * , 83 .
oguri m. 2005 , mnras , * 361(1 ) * , l38 .
oort j.h . , 1932 , bull .
netherlands , * 6 * , 249 .
pik e. 1915 , bull . de la soc . astr .
de russie * 21 * , 150 .
paczyski b. 1986 , aph.j . * 301 * , 503 .
paczyski b. 1986 , aph.j . * 304 * , 1 .
paczyski b. , wambsganss j. 1989 , apj * 337 * , 581 .
paduroiu s. , revaz y. , pfenniger d. 2015 , arxiv 1506.03789 .
riess a.g . , casertano s. , anderson j. , mackenty j. , filippenko a. 2014 , aph.j . ,
* 785(2 ) * , 161 .
rocha m.e .
2012 , mnras , * 430(1 ) * , 81 .
salpeter e. , 1955 , apj , * 121 * , 161 .
schechter p. , wambssganss j. 2002 , aj , * 580 * , 685 .
schneider p. , ehlers j. , and falko e. e. , gravitational lenses , springer , new york , ny , usa , 1992 .
schneider a. , krauss l. , moore b. 2010 , phys.rev.d , * 82 * , 063525 .
schneider a. , krauss l. , moore b. 2011 , arxiv 1105.4106 .
schneider a. , smith r.e .
, macci a. , moore b. 2012 , mnras , * 424 * , 684 .
schneider p. , weiss a. , 1987 , mpa rep .
, * 311 * , 46 .
schmidt r. w. , wambsganss j. , 2010 , grg * 42 * , 2127 .
shaw l. d. , weller j. , ostriker j.p .
, bode p. 2007
, * 659 * , 1082 .
seitz c. , schneider p. 1994 , a&a * 288
* , 1 . shin i .-
; han c. ; choi j .- y .
2012 , aph.j .
, * 755 * , i d . 91 , 10 pp . shull j.m
2014 , aph.j . , * 784 * , is . 2 , 10 .
sliusar v.n . ,
zhdanov v.i . ,
alexandrov a.n . ,
fedorova e.v .
2015 , kinemat .
bodies * 31 * , is . 2 , 47 . springel v. et al .
2005 , nature , * 435 * , 629 .
springel v. , wang j. , vogelsberger m. et al .
2008 , mnras , * 391 * , 1685 . stadel j. , potter d. , moore b. , et al .
2009 , mnras , * 398 * , l21 .
swaters r.a . ,
sancisi r. , van albada t.s .
, van der hulst j.m .
2011 , aph.j . ,
* 729 * , doi:10.1088/0004 - 637x/729/2/11 .
tisserand p. , le guillou l. , afonso c. , et al .
2007 , a & a , * 469 * , 387 .
udalski a. , szymaski m. , kauny j. , et al .
1993 , acta astron . , * 43 * , 289 .
wambsganss , j. , paczyski , b. , katz , n. 1990 , aph.j * 352 * , 407 .
wambsganss j. , witt h.j . , schneider p. 1992 , a&a , * 258
* , 591 .
wyrzykowski . , kozowski s. , skowron j. , et al .
, 2009 , mnras , * 397 * , 1228 . vogelsberger m. et al .
2012 , mnras , * 423 * , 3740 .
zackrisson e. , riehm t. 2010 , advances in astronomy , i d .
478910 .
zakharov a. 2010 , gen.rel.grav . ,
* 42 * , 2301 .
zakharov a.f . ,
sazhin m.v . 1999 ,
a&at , * 18 * , 27 .
zhdanov v. , alexandrov a. , fedorova e. , sliusar v. 2012 , isrn a&a , * 2012 * , i d 906951 .
zhang d. 2011 , mnras , * 418(3 ) * , 1850 . zwicky f. 1933 , helvetica physica acta , * 6 * , 110 .
[ lastpage ]
in case of a point mass ( schwarzschild ) lens we have the mapping @xmath305 the amplification is well known : @xmath306 in case of a relative straight - line motion of the lens and the source we have @xmath307 , @xmath308 is the velocity , @xmath309 is the impact parameter .
therefore , the problem of fitting any given dependence @xmath214 ( e.g. , that has been obtained for a clumped or cusped lens model ) by means of ( [ schwarzschild ] ) is reduced to finding the coefficients @xmath310 of an approximate linear mapping @xmath311 : @xmath312 where @xmath313 .
obviously , @xmath309 and @xmath308 will differ from that of the initial clump model . in our case , to estimate the difference between models , the approximation is performed near the maximum of the amplification curve , where it has a parabolic form ( typically for @xmath314 . here
residuals of the approximation are of the order of @xmath315 ) . using the point mass model with the parameters obtained as the result of this approximation , we estimated @xmath316 for a larger interval .
note that to determine @xmath309 and @xmath308 we deal with the relative values @xmath317 irrespective of @xmath276 , which is not essential for fitting the amplification .
however , this is essential when considering the image centroid .
e.g. , in case of two images , the image centroid is @xmath318 where @xmath319 is the @xmath249-th image position and @xmath320 is its amplification ( @xmath321 ) . here
we have an additional parameter @xmath276 that can be used to rescale the trajectory @xmath322 ; this is done in the lower panel of fig .
[ c - cusp ] . | extended dark matter ( dm ) substructures may play the role of microlenses in the milky way and in extragalactic gravitational lens systems ( glss ) .
we compare microlensing effects caused by point masses ( schwarzschild lenses ) and extended clumps of matter using a simple model for the lens mapping .
a superposition of the point mass and the extended clump is also considered . for special choices of the parameters ,
this model may represent a cusped clump of cold dm , a cored clump of self - interacting dark matter ( sidm ) or an ultra compact minihalo of dm surrounding a massive point - like object .
we built the resulting micro - amplification curves for various parameters of one clump moving with respect to the source in order to estimate differences between the light curves caused by clumps and by point lenses .
the results show that it may be difficult to distinguish between these models .
however , some region of the clump parameters can be restricted by considering the high amplification events at the present level of photometric accuracy .
then we estimate the statistical properties of the amplification curves in extragalactic glss .
for this purpose , an ensemble of amplification curves is generated yielding the autocorrelation functions ( acfs ) of the curves for different choices of the system parameters .
we find that there can be a significant difference between these acfs if the clump size is comparable with typical einstein radii ; as a rule , the contribution of clumps makes the acfs less steep .
[ firstpage ] astrophysics , cosmology gravitational microlensing , dark matter . |
an _ independent _ set in a graph @xmath9 , is a subset of the vertices not containing any edges . for an integer @xmath10
we denote the collection of independent @xmath2-sets of @xmath0 by@xmath11 a subfamily @xmath12 of @xmath4 is said to be _ intersecting _ if @xmath13 implies @xmath14 . if @xmath15 then the collection of independent @xmath2-sets containing @xmath16 is @xmath17 such a family is called a _ star_. a graph @xmath0 is said to be _ @xmath2-ekr _ if no intersecting family @xmath18 is larger than the largest star in @xmath4 . if every intersecting family @xmath19 of maximum size is a star then @xmath0 is said to be _
strictly @xmath2-ekr_. the classical result in this area is the erds - ko - rado theorem which can be stated as follows .
if @xmath20 is the empty graph of order @xmath21 , then @xmath0 is @xmath2-ekr if @xmath22 and strictly @xmath2-ekr if @xmath23 .
there are several other recent results of this type .
[ blthm ] if @xmath24 , @xmath25 and @xmath0 is the disjoint union of @xmath21 copies of @xmath26 , then @xmath0 is @xmath2-ekr and strictly so unless @xmath27 and @xmath28 .
[ completethm]if @xmath0 is the disjoint union of @xmath29 complete graphs each of order at least two , then @xmath0 is @xmath2-ekr . in this paper
we consider the question of when a graph is @xmath2-ekr . in the next section
we give the first of our two main results : if a graph @xmath0 is @xmath2-ekr then its lexicographic product with any complete graph is also @xmath2-ekr . in section 3
we present some examples showing that graphs exhibit a variety of ekr properties .
these serve to motivate a conjecture we propose , giving a lower bound on the minimum @xmath2 such that a given graph @xmath0 can fail to be @xmath2-ekr .
this conjecture is known to be true for empty graphs , cycles , paths and disjoint unions of complete graphs . in the final section
we give our second main result : this conjecture is true for disjoint unions of two complete multipartite graphs . throughout @xmath0
is assumed to be a simple graph ( without loops or multiple edges ) and to have finite vertex set @xmath3 and edge set @xmath30 .
the independence number of a graph is denoted by @xmath31 and the _ minimax independence number _ ( the minumum size of a maximal independent vertex set ) by @xmath6 . an _ anomalous _ subfamily of @xmath4 is an intersecting subfamily that is not a subfamily of any star .
a vertex @xmath16 is an @xmath2-_centre _ of @xmath0 if @xmath32 for every intersecting subfamily @xmath12 of @xmath4 and is a _ strict _ @xmath2-_centre _ if @xmath33 for every anomalous subfamily @xmath12 of @xmath4 . where no confusion is caused
, we may omit the argument ` @xmath34 ' .
if @xmath35 is a family of sets then we define @xmath36 given two graphs @xmath0 and @xmath37 , the _ lexicographic product _
$ ] is constructed ( informally speaking ) by replacing each vertex of @xmath0 with a copy of @xmath37 .
more formally , @xmath39 ) = v(g)\times v(h)$ ] , where @xmath40 is adjacent in @xmath38 $ ] to @xmath41 if and only if _ either _ @xmath16 is adjacent to @xmath42 in @xmath0 _ or _ @xmath43 and @xmath44 is adjacent to @xmath45 in @xmath37 .
it is useful to develop a generalization of this concept : rather than insisting that each vertex of @xmath0 be replaced by a copy of a fixed graph , we may allow the replacement graphs to vary .
for example , if we begin with @xmath0 and replace each vertex @xmath46 with a copy of a graph @xmath37 and each vertex @xmath47 with a copy of a graph @xmath48 , then we denote the result by + @xmath49 $ ] .
we begin with a lemma concerning ekr properties of general set families , inspired by the elegant proof due to katona @xcite of the erds - ko - rado theorem and giving it a more general context .
a family of subsets of a set @xmath50 is a @xmath51-_covering _ of @xmath50 if each element of @xmath50 belongs to exactly @xmath51 sets of the family .
[ genkat ] let @xmath35 be a family of @xmath2-subsets of a finite set @xmath50 , let @xmath52 be a family of subfamilies of @xmath35 , let @xmath53 , and suppose suppose that , for some @xmath51 : * @xmath52 is a @xmath51-covering of @xmath35 ; * @xmath42 is an @xmath2-centre of each @xmath54 .
then @xmath42 is an @xmath2-centre of @xmath35 .
* let @xmath12 be any intersecting subfamily of @xmath35 .
since @xmath52 is a @xmath51-covering of @xmath35 , it is a @xmath51-covering of @xmath12 and so @xmath55 in particular , @xmath56 but for any intersecting subfamily @xmath12 of @xmath35 and any @xmath57 , the family @xmath58 is an intersecting subfamily of @xmath59 , and so ( since @xmath42 is an @xmath2-centre of each @xmath59 ) @xmath60 now , ( [ qf ] ) , ( [ qx ] ) and ( [ xg ] ) imply ( for any intersecting subfamily @xmath12 of @xmath35 ) : @xmath61 and so @xmath42 is an @xmath2-centre of @xmath35 .
* the ` strict ' extension of lemma [ genkat ] is false .
for example , let @xmath50 be the vertex set of an octahedron and let @xmath35 be the family of 3-subsets of @xmath50 corresponding to the faces .
let @xmath52 be the 1-covering ( i.e. partition ) of @xmath35 into pairs of opposite faces .
each @xmath54 is trivially ekr , and so each @xmath62 is a strict 3-centre of each such @xmath59 .
also , each @xmath53 is a 3-centre of @xmath35 with @xmath63 .
however , there exist anomalous subfamilies of @xmath35 of cardinality 4 , namely ( for each face @xmath64 ) the family of faces containing at least two of the vertices of @xmath64 .
thus the elements of @xmath50 are not strict 3-centres of @xmath35 .
[ lexiprod ] let @xmath16 be an @xmath2-centre of a graph @xmath0 and let @xmath65 ; then each vertex @xmath66 is an @xmath2-centre of the lexicographic product @xmath67 $ ] .
* when @xmath68 the statement is trivial , so assume @xmath69 . for the purposes of this proof , it is convenient to identify the vertices ( in some fixed way ) with the elements of the set @xmath70=1,2,\ldots , n$ ] , and to identify the vertices of @xmath71 with the elements of the cyclic group @xmath72 .
let @xmath35 be the family of functions @xmath73
\to z_m$ ] .
then , for each @xmath74 and each @xmath75 , we define @xmath76 we now define an equivalence relation @xmath77 on @xmath35 by @xmath78 that is , @xmath79).$ ] we denote by @xmath80 the family of equivalence classes , and for each @xmath81 we let @xmath82 denote the following subfamily of @xmath83)$ ] : @xmath84 each @xmath85)$ ] is of the form @xmath86 for exactly one @xmath87 and exactly @xmath88 functions @xmath89 ( each in a distinct equivalence class ) . that is , the family @xmath90 is a @xmath51-covering of @xmath83)$ ] where @xmath91 . by lemma [ genkat ] , it remains to show that each @xmath92 is an @xmath2-centre of @xmath82 for each @xmath81 .
let @xmath81 and let @xmath12 be an intersecting subfamily of @xmath82 .
let @xmath93 then @xmath94 is an intersecting subfamily of @xmath4 , and so @xmath95 . if @xmath87 and @xmath89 , @xmath96 are distinct elements of @xmath97 , then @xmath98 .
but @xmath12 is intersecting ; thus any two distinct elements of @xmath12 correspond to distinct elements of @xmath94 .
hence @xmath99 , and so @xmath100 let @xmath101 and consider the vertex @xmath102 of @xmath67 $ ] . for each @xmath81 and each @xmath103 , we have @xmath104 if and only if @xmath105 , in which case there is exactly one @xmath106 with this property .
thus @xmath107 , and it follows from ( [ lexic ] ) that @xmath102 is an @xmath2-centre of @xmath82 .
[ lexithm]if @xmath0 is @xmath2-ekr and @xmath108 then @xmath67 $ ] is @xmath2-ekr . *
* this follows directly from lemma [ lexiprod ] .
@xmath109 it is natural to ask whether theorem [ lexithm ] extends to lexicographic products that involve replacing the vertices of @xmath0 with complete graphs of variable rather than constant order .
we now show that this is not unconditionally true .
[ thirteen ] let @xmath0 be the graph with vertex set @xmath110 depicted below ( figure [ thirteenfig ] ) .
[ thirteenfig ] [ ] [ ] @xmath111 [ ] [ ] @xmath112 [ ] [ ] @xmath113 [ ] [ ] @xmath114 [ ] [ ] @xmath115 [ ] [ ] @xmath116 [ ] [ ] @xmath117 [ ] [ ] @xmath118 [ ] [ ] @xmath119 [ ] [ ] @xmath120 [ ] [ ] @xmath121 [ ] [ ] @xmath122 [ ] [ ] @xmath123 of example [ thirteen],title="fig : " ] it may straightforwardly be verified that @xmath0 is 3-ekr , the vertices @xmath124 being 3-centres , with @xmath125 the family of independent vertex 3-sets containing at least two of the vertices @xmath113 , @xmath120 , @xmath121 is of cardinality 16 and is one of two anomalous families of maximum cardinality .
now let @xmath65 and consider the graph @xmath126 $ ] .
then , @xmath127 ) \right\vert}}=15 + 2m\mbox { } ( v = v_1,\ldots , v_6)$ ] , the values for the remaining vertices being independent of @xmath128 . however , the anomalous family consisting of all independent 3-sets of @xmath126 $ ] containing at least two of the vertices @xmath113 , @xmath120 , @xmath121 is of cardinality @xmath129 .
thus , for @xmath130 , the vertices @xmath124 of @xmath126 $ ] are not 3-centres ( and @xmath126 $ ] is not 3-ekr ) .
trivially , any graph is 1-ekr .
the question of when a ( non - complete ) graph is 2-ekr is easy to deal with : [ 2-ekr ] let @xmath0 be any non - complete graph of order @xmath21 and with minimum degree @xmath131 . *
if @xmath132 , then @xmath0 is strictly 2-ekr . * if @xmath133 , then @xmath0 is 2-ekr if and only if @xmath134 and strictly so if and only if @xmath135 , the 2-centres being the vertices of minimum degree . * proof . *
let @xmath12 be an anomalous family of independent vertex 2-sets .
then @xmath136 , and @xmath12 must contain the three 2-subsets of some independent 3-set ; but then no other 2-set can intersect all three of these , and so @xmath12 must consist exactly of the three 2-subsets of an independent 3-set . thus : * if @xmath137 , then there is no anomalous family of independent vertex 2-sets , so @xmath0 is strictly 2-ekr ; * otherwise , the anomalous families of independent vertex 2-sets are all of cardinality 3 and the result follows from the fact that , for any vertex @xmath16 , @xmath138 for the remainder of the paper , then , we concentrate on on the question : for @xmath139 , when is @xmath0 @xmath2-ekr ?
all of the graphs studied in @xcite , including those arising from reinterpreting @xcite and @xcite , are @xmath140-ekr and also @xmath141-ekr , giving rise to the question : is this always true ?
the answer is no , as the following examples show .
[ dodeca ] let @xmath0 be the graph of the regular dodecahedron ( that is , the graph whose vertices and edges are those of the dodecahedron ) .
then @xmath142 , where @xmath143 consists of the vertex sets of the five inscribed cubes of the dodecahedron . any pair of these sets intersects on two ( opposite ) vertices , but any given vertex belongs to just two of them .
thus @xmath143 is an anomalous family and @xmath0 is not 8-ekr .
we note , without proof , that if @xmath0 is the graph of any of the platonic solids other than the dodecahedron , then @xmath0 @xmath140-ekr .
[ spiky ] let @xmath64 be the graph with vertices @xmath144 where @xmath145 are pairwise adjacent and @xmath146 is adjacent only to @xmath147 .
( see figure [ spikyfig ] . )
now let @xmath148 $ ] .
then the order of @xmath0 is 16 and @xmath149 .
moreover , the families @xmath150 are precisely the families of @xmath2-subsets of the unique independent 7-set .
thus , @xmath0 is 7-ekr in a trivial way and ( by the erds - ko - rado theorem ) is not 4- , 5- or 6-ekr .
more interestingly , @xmath0 fails to be 3-ekr , since no vertex belongs to more than 21 independent 3-sets but there is an anomalous family consisting of the 22 independent 3-sets containing at least two of @xmath119 , @xmath118 , @xmath112 .
thus it is possible for a graph to fail to be @xmath141-ekr and to fail to be @xmath151-ekr .
[ spikyfig ] [ ] [ ] @xmath152 [ ] [ ] @xmath113 [ ] [ ] @xmath114 [ ] [ ] @xmath120 [ ] [ ] @xmath118 [ ] [ ] @xmath121 [ ] [ ] @xmath153 of example [ spiky],title="fig : " ] in each graph studied so far , when @xmath0 is @xmath140-ekr , it is so in a trivial way ; but this is not so in general , as the next example shows .
[ icosa ] let @xmath0 be the graph of the regular icosahedron . then @xmath154 .
it is straightforward to check that @xmath155 for any vertex @xmath16 , and with a little care it is possible to construct an anomalous family of four independent 3-sets and to verify that no such family can be extended to a fifth member .
thus @xmath0 is ( strictly ) 3-ekr .
note that the antipodal pairs of vertices of @xmath0 are maximal independent sets , so that @xmath156 .
therefore , this example also shows that it is possible for a graph to be @xmath2-ekr for some @xmath157 .
it is easy to vary example [ spiky ] to produce a graph of arbitrarily large independence number that fails to be 3-ekr since , if we replace @xmath158 by @xmath159 and @xmath160 by @xmath161 in the generalized lexicographic construction of that example , then @xmath162 , the maximum value of @xmath163 is @xmath164 and there is an anomalous subfamily of @xmath165 of cardinality @xmath166 .
more generally it is possible , for any @xmath167 , to produce a graph of arbitrarily large independence number that fails to be @xmath2-ekr .
however , this does not seem to be true for the minimax independence number .
we make the following conjecture .
[ mu ] let @xmath0 be any graph and let @xmath168 ; then @xmath0 is @xmath2-ekr ( and is strictly so if @xmath169 ) .
each of the above bounds is sharp , as our final example shows .
[ bipexam ] let @xmath0 be the disjoint union of two copies of the complete bipartite graph @xmath170 .
then ( by theorem [ bip ] of section 4 ) @xmath171 and @xmath0 is non - strictly 3-ekr and strictly 2-ekr , but not 4-ekr .
it seems plausible that if any graphs fail to be @xmath2-ekr , for some @xmath172 , then the smallest examples should have @xmath173 ( that is , all maximal independent vertex sets should have the same cardinality ) .
this motivates the study of classes of graphs with this property .
conjecture [ mu ] is already known to hold for certain classes of graphs ; in particular it holds for empty graphs and disjoint unions of complete graphs ( both of which have @xmath174 ) .
we now show that the conjecture also holds for the class of unions of pairs of complete multipartite graphs ; moreover , if @xmath0 is such a graph and @xmath174 , then the bound is sharp in that @xmath0 fails to be @xmath2-ekr if @xmath175 .
let @xmath0 be the disjoint union of two complete bipartite graphs , @xmath183 $ ] and @xmath184 $ ] .
denote the partite sets of @xmath185 by @xmath186 where @xmath187 and those of @xmath188 by @xmath189 where @xmath190 . for @xmath191 ,
define @xmath192 as follows.@xmath193 similarly , for @xmath194 , define @xmath195 by @xmath196 with slight abuse of notation , if @xmath197 , we may write + @xmath198 and @xmath199 . note that + @xmath200 .
we now define the _ compressions _
@xmath201 on subfamilies of @xmath202 as follows .
let @xmath203 and let @xmath191 .
then @xmath204 more informally , for each @xmath205 that intersects @xmath206 , we replace @xmath207 by @xmath208 provided that @xmath208 is not already in @xmath12 ; otherwise , we leave @xmath207 alone .
we now note that , if @xmath12 is a non - empty intersecting subfamily of @xmath210 , then there is some partite set of @xmath185 or @xmath188 that intersects every set of @xmath12 ; for any @xmath205 is a subset of @xmath211 for some @xmath212 and now there can not be @xmath213 with @xmath214 and @xmath215 . by exchanging @xmath185 and @xmath188 if necessary
, we may assume that some fixed @xmath206 intersects each set of @xmath12 .
clearly , @xmath216 is an intersecting family with @xmath217 such that @xmath218 intersects each set of @xmath94 .
thus , in investigating the sizes of intersecting subfamilies @xmath12 of @xmath4 , we may assume that @xmath218 intersects each @xmath219 ; such a family is said to be _
standardized_. @xmath235 implies that @xmath236 .
also @xmath237 implies that there exists @xmath238 such that @xmath228 .
now if @xmath239 then @xmath240 , a contradiction , since @xmath229 .
so there exists @xmath241 .
but then @xmath242 as required
. hence @xmath221 is intersecting .
@xmath243 * proof .
* let @xmath248 . then , for any @xmath249 such that @xmath250 and @xmath251 , we have @xmath252 , and so @xmath94 is compressed . by lemma [ firstlem ]
, @xmath94 is intersecting and @xmath253 .
now let @xmath254 .
suppose @xmath255 where @xmath256 .
then @xmath257 , giving a contradiction since @xmath258 .
@xmath243 + * proof of theorem [ bip ] * * proof of ( i ) * let @xmath259 as above and let @xmath260 be an intersecting family . by using theorem [ 2-ekr ] or by direct consideration of small cases
, we may assume @xmath261 .
we shall show that @xmath262 partition @xmath12 as @xmath264 where @xmath265 and , for @xmath266 , @xmath267 correspondingly , let @xmath268 where @xmath269 , and partition @xmath270 as @xmath271 .
now , @xmath272 thus , by the erds - ko - rado theorem ( since @xmath273 ) , we have @xmath274 we now compare @xmath275 with @xmath276 . since each @xmath207 in @xmath277 intersects @xmath218 and @xmath278 , we have @xmath279 where @xmath280 . for @xmath281 ,
let @xmath282 and @xmath283 .
+ analogously , let @xmath284 and @xmath285 .
then , for @xmath194 : @xmath286 @xmath287 since @xmath12 is standardized and compressed , each @xmath288 is intersecting , by lemma [ downlem ] .
thus , by ( [ mu ] ) and the erds - ko - rado theorem , we have for @xmath289 : @xmath290 thus , if @xmath291 , then we may conclude from ( [ a0+a1])([bij ] ) that @xmath292 .
suppose now that @xmath293 .
for @xmath294 , we have @xmath295 moreover , by the intersecting property , no set in @xmath296 can be the complement of a set in @xmath297 , and hence @xmath298 two cases arise .
now , from @xmath302 , it is straightforward to deduce that @xmath303 together with inequality ( [ bs ] ) , this implies that ( [ a-1 ] ) still holds .
thus , + @xmath304 . with equation ( [ a0+a1 ] ) , this gives the result @xmath305 , as required .
first note that , for @xmath307 , equality must hold in each of the inequalities ( [ a0+a1 ] ) , ( [ ai ] ) , ( [ bij ] ) , ( [ bs ] ) ; moreover , for @xmath308 , the inequality ( [ dchooses ] ) is strict . thus , by the erds - ko - rado theorem , for @xmath307 , @xmath309 for some @xmath310 . by ( [ ai ] ) , ( [ ji ] ) , the erds - ko - rado theorem now implies , for @xmath194 : @xmath311 for some @xmath312 .
clearly , the @xmath313 must all be equal and the result follows .
b. bollobs and i. leader , an erds - ko - rado theorem for signed sets , _
computers math .
* 34 * ( 11 ) ( 1997 ) , 9 - 13 . c. godsil and g. royle , _ algebraic graph theory _ , graduate texts in mathematics * 207
* , springer - verlag , new york ( 2000 ) .
p. erds , c. ko and r. rado , intersection theorems for systems of finite sets , _ quart . j. math .
oxford ( 2 ) _ * 12 * ( 1961 ) , 313 - 320 .
holroyd and j. talbot , compression and erds - ko - rado graphs ( submitted )
katona , a simple proof of the erds - ko - rado theorem , _ j. combinatorial theory ( b ) _ * 13 * ( 1972 ) , 183 - 184 .
j. talbot , intersecting families of separated sets , _ j. london math .
( to appear ) | for a graph @xmath0 and integer @xmath1 we denote the family of independent @xmath2-sets of @xmath3 by @xmath4 .
a graph @xmath0 is said to be _ @xmath2-ekr _ if no intersecting subfamily of @xmath4 is larger than the largest such family all of whose members contain some fixed @xmath5 .
if this inequality is always strict , then @xmath0 is said to be _ strictly @xmath2-ekr_. we show that if a graph @xmath0 is @xmath2-ekr then its lexicographic product with any complete graph is @xmath2-ekr . for any graph @xmath0
, we define @xmath6 to be the minimum size of a maximal independent vertex set .
we conjecture that , if @xmath7 , then @xmath0 is @xmath2-ekr , and if @xmath8 , then @xmath0 is strictly @xmath2-ekr .
this is known to be true when @xmath0 is an empty graph , a cycle , a path or the disjoint union of complete graphs .
we show that it is also true when @xmath0 is the disjoint union of a pair of complete multipartite graphs . |
according to contemporary cosmology , the present energy content of the universe is composed of approximately @xmath22 ordinary matter , @xmath23 dark matter and @xmath24 dark energy @xcite .
the expansion of the universe began in a tremendous inflationary burst driven by the vacuum energy with the planck density @xmath25 . between @xmath26 and @xmath27 seconds after the beginning ( big bang ) , the universe expanded by a factor @xmath28 @xcite .
inflation does not offer any explanation for the time before that `` beginning '' .
the universe then entered in the radiation era and , when the temperature cooled down below approximately @xmath29 , in the matter era @xcite . at present
, it undergoes an accelerated expansion @xcite presumably due to the cosmological constant or to some form of dark energy with negative pressure violating the strong energy condition @xcite .
this corresponds to a second period of inflation , which is different from the first since it is driven by the cosmological energy density @xmath14 instead of the planck density . despite the success of this model , the nature of dark matter , dark energy , and of the very early universe ( pre - radiation era ) remains very mysterious and leads to many speculations .
the phase of inflation in the early universe is usually described by some hypothetical scalar field @xmath30 with its origin in quantum fluctuations of vacuum @xcite .
this leads to an equation of state @xmath31 , implying a constant energy density , called the vacuum energy .
this energy density is usually identified with the planck density @xmath32 . as a result of the vacuum energy ,
the universe expands exponentially rapidly on a timescale of the order of the planck time @xmath33 ( early inflation ) .
this phase of inflation is followed by the radiation era described by an equation of state @xmath34 . in our previous paper ( paper i )
, we showed that a unified description of the pre - radiation ( @xmath35 ) and radiation ( @xmath36 ) eras , connected by a phase of exponential inflation , could be obtained from a single equation of state of the form @xmath37 .
this provides a non - singular model of the early universe . the phase of acceleration in the present universe is usually ascribed to the cosmological constant @xmath18 which is equivalent to a constant energy density @xmath38 called dark energy
this can be modeled by an equation of state @xmath31 , implying a constant energy density identified with the cosmological density @xmath39 . as a result of the dark energy ,
the universe expands exponentially rapidly on a timescale of the order of the cosmological time @xmath15 ( de sitter solution ) .
this leads to a second phase of inflation ( late inflation ) .
inspired by the analogy with the early inflation , some authors have represented the dark energy by a scalar field called quintessence @xcite . as an alternative to the quintessence
, other authors @xcite have proposed to model the acceleration of the universe by an exotic fluid with an equation of state of the form @xmath40 called the chaplygin gas ( see @xcite for a complete list of references ) . at late times , this equation of state leads to a constant energy density implying an exponential growth of the scale factor that is similar to the effect of the cosmological constant . at earlier times
, this equation of state returns the results of the cold dark matter model .
therefore , it provides a unification of dark matter ( @xmath41 ) and dark energy ( @xmath42 ) in the late universe .
furthermore , it gives a real velocity of sound which is non - trivial for fluids with negative pressure .
some generalizations of this equation of state have been considered in the form @xmath43 with @xmath44 @xcite .
as mentioned in @xcite , the chaplygin gas has some connection with string theory and can be obtained from the nambu - goto action for @xmath45-branes moving in a @xmath46)-dimensional spacetime in the light - cone parametrization .
furthermore , it is the only fluid which , up to now , admits a supersymmetric generalization . from a theoretical point of view , it is desirable to study models of universe with a generalized equation of state @xmath47 having a standard linear component and a polytropic component .
the linear equation of state @xmath1 describes radiation ( @xmath2 ) , pressureless matter ( @xmath3 ) , stiff matter ( @xmath4 ) , and vacuum energy ( @xmath5 ) .
the polytropic equation of state @xmath48 with @xmath49 may be due to bose - einstein condensates with repulsive ( @xmath7 ) or attractive ( @xmath8 ) self - interaction @xcite , or have another origin .
when @xmath50 , the polytropic component dominates the linear component when the density is high .
these models , studied in paper i , describe the early universe .
conversely , when @xmath9 , the polytropic component dominates the linear component when the density is low .
these models , studied in the present paper , describe the late universe .
as we shall see , these two studies are strikingly symmetric .
interestingly , this symmetry seems to reflect the true evolution of the universe . in this paper ,
we propose an exhaustive study of the equation of state @xmath47 with @xmath9 .
when @xmath7 , the universe exhibits a future peculiarity : its density vanishes in infinite time ( @xmath51 ) , or periodically in time ( @xmath52 ) , while its radius tends to a constant value ( the universe is singular when @xmath53 since the pressure becomes infinite in finite time ) . for @xmath3 , @xmath10 and @xmath16 , we obtain a model of cyclic universe , equivalent to the anti-@xmath18cdm model , in which the universe `` disappears '' ( it becomes empty ) and `` reappears '' periodically . according to this model
, the universe would disappear in about @xmath17 billion years .
when @xmath8 , the universe exists for all times in the future and there is no singularity . for @xmath3 , @xmath10 and @xmath11 , we obtain a model of late universe , equivalent to the @xmath18cdm model , describing in a unified manner the transition from the matter era to the dark energy era .
coincidentally , we live close to the transition between these two periods , corresponding to @xmath12 and @xmath13 .
this universe exists eternally in the future and undergoes an inflationary expansion with the cosmological density @xmath14 on a timescale @xmath15 .
the paper is organized as follows . in sec .
[ sec_basic ] , we recall the basic equations of cosmology . in secs . [ sec_ges ] and [ sec_dark ] ,
we study the generalized equation of state @xmath54 for any value of the parameters @xmath55 , @xmath56 and @xmath9 . in sec .
[ sec_solid ] , we consider the case @xmath3 , @xmath10 , and @xmath57 , providing a model of non - singular inflationary universe describing the transition from the matter era to the dark energy era ( @xmath18cdm model ) . in sec .
[ sec_cyclic ] , we consider the case @xmath3 , @xmath10 , and @xmath58 , leading to a model of peculiar cyclic universe ( anti-@xmath18cdm model ) . in sec .
[ sec_simple ] , we discuss the connection between the early and the late universe , and propose a simple equation [ see eq .
( [ pr7 ] ) ] describing the whole evolution of the universe .
this leads to a model of universe that is eternal in past and future without singularity ( aioniotic universe ) .
this model exhibits a nice `` symmetry '' between an early and late phase of inflation , the cosmological constant in the late universe playing the same role as the planck constant in the early universe .
this model refines the standard @xmath18cdm model by removing the primordial singularity . in sec .
[ sec_scalar ] , we determine the potential of the effective scalar field ( quintessence , tachyon field ) corresponding to the generalized equation of state @xmath47 .
in a space with uniform curvature , the line element is given by the friedmann - lematre - roberston - walker ( flrw ) metric @xmath59 where @xmath60 represents the radius of curvature of the @xmath61-dimensional space , or the scale factor . by an abuse of language
, we shall call it the `` radius of the universe '' . on the other hand
, @xmath56 determines the curvature of space .
the universe can be closed ( @xmath7 ) , flat ( @xmath62 ) , or open ( @xmath8 ) .
if the universe is isotropic and homogeneous at all points in conformity with the line element ( [ b1 ] ) , and contains a uniform perfect fluid of energy density @xmath63 and isotropic pressure @xmath64 , the einstein equations can be written as @xmath65 @xmath66 @xmath67 where we have introduced the hubble parameter @xmath68 and accounted for a possibly non - zero cosmological constant @xmath18 .
these are the well - known friedmann equations describing a non - static universe @xcite . among these three equations ,
only two are independent .
the first equation can be viewed as an `` equation of continuity '' . for a given barotropic equation of state @xmath69
, it determines the relation between the density and the scale factor .
then , the temporal evolution of the scale factor is given by eq .
( [ b4 ] ) .
we will also need the thermodynamical equation @xmath70 which can be derived from the first principle of thermodynamics @xcite . for a given barotropic equation of state @xmath69 , this equation can be integrated to obtain the relation @xmath71 between the temperature and the density .
it can be shown @xcite that the friedmann equations conserve the entropy of the universe @xmath72 in this paper , we consider a flat universe ( @xmath62 ) in agreement with the observations of the cosmic microwave background ( cmb ) @xcite . on the other hand , we set @xmath73 because the effect of the cosmological constant will be taken into account in the equation of state .
the friedmann equations reduce to @xmath74 @xmath75 @xmath76 introducing the equation of state parameter @xmath77 , we see from eq .
( [ b8 ] ) that the universe is decelerating if @xmath78 ( strong energy condition ) and accelerating if @xmath79 . on the other hand , according to eq .
( [ b7 ] ) , the density decreases with the scale factor if @xmath80 ( null dominant energy condition ) and increases with the scale factor if @xmath81 .
we consider a generalized equation of state of the form @xmath82 this is the sum of a standard linear equation of state @xmath1 and a polytropic equation of state @xmath48 , where @xmath56 is the polytropic constant and @xmath49 is the polytropic index . concerning the linear equation of state , we assume @xmath83 ( the case @xmath5 is treated specifically in appendix [ sec_eosgm ] ) .
this equation of state describes radiation ( @xmath2 ) , pressureless matter ( @xmath3 ) and vacuum energy ( @xmath5 ) .
concerning the polytropic equation of state , we remain very general .
such an equation of state may correspond to self - gravitating becs with repulsive ( @xmath7 ) or attractive ( @xmath8 ) self - interaction , but it may have another origin . for an equation of state of the form ( [ b10 ] ) with positive index @xmath50 , the polytropic component dominates the linear component when the density is high .
these models , studied in paper i , describe the early universe .
conversely , when @xmath9 , the polytropic component dominates the linear component when the density is low .
these models , studied in the present paper , describe the late universe .
we shall see that these two situations are strikingly similar .
they reveal a form of `` symmetry '' between the early and the late universe . in this paper , as in paper i , we assume that @xmath84 .
this corresponds to the `` normal '' case where the density decreases with the radius ( @xmath80 ) .
the opposite case @xmath85 , leading to a `` phantom universe '' where the density increases with the radius ( @xmath81 ) , is treated in paper iii .
the purpose of our series of papers is to make an exhaustive study of all the possible cases @xmath86 , even if some of them seem to be unphysical or in conflict with the known properties of our universe . actually , by rejecting the cases leading to past or future singularities ( or peculiarities )
, we will be led naturally to the `` good '' model ( see sec .
[ sec_aioniotic ] ) . some of the equations established in paper i remain valid when @xmath9 . to avoid repetitions
, we shall directly refer to these formulae .
for example , eq .
( i-60 ) refers to eq .
( 60 ) of paper i. for the equation of state ( [ b7 ] ) , assuming @xmath80 , the friedmann equation ( [ b2 ] ) can be integrated into @xmath87 where @xmath88 and @xmath89 is a constant of integration .
the upper sign corresponds to @xmath7 and the lower sign corresponds to @xmath8 . and
we have taken @xmath90 and @xmath3 . ] for @xmath7 , the density is defined only for @xmath91 . when @xmath92 , @xmath93 corresponding to the linear equation of state .
when @xmath94 , @xmath95 for the behavior of the pressure , we must consider different cases .
when @xmath92 , @xmath96 for @xmath97 , @xmath98 for @xmath99 , @xmath96 for @xmath100 , @xmath101 for @xmath102 , and @xmath103 for @xmath104 .
when @xmath94 , @xmath105 for @xmath53 , @xmath106 tends to a finite value for @xmath10 , and @xmath101 for @xmath107 . for @xmath8
, the density is defined for all @xmath108 .
when @xmath92 , @xmath109 corresponding to the linear equation of state .
when @xmath110 , the density tends to a finite value @xmath111 . for the behavior of the pressure , we must consider different cases .
when @xmath92 , @xmath96 for @xmath97 , @xmath98 for @xmath99 , @xmath98 for @xmath100 , @xmath101 for @xmath102 , and @xmath103 for @xmath104 .
when @xmath110 , @xmath112 .
some curves giving the evolution of the density @xmath113 as a function of the scale factor @xmath108 are plotted in fig .
[ densitenneg ] for @xmath7 and @xmath8 . for the equation of state ( [ b10 ] ) ,
the evolution of the temperature is given by eqs .
( i-60 ) and ( i-61 ) .
let us consider some asymptotic limits .
when @xmath114 , eq .
( i-61 ) reduces to eq .
( i-62 ) returning the relation valid for a linear equation of state . when @xmath92 , @xmath115 for @xmath97 , @xmath116 for @xmath99 , and @xmath117 for @xmath3 .
when @xmath7 and @xmath94 , @xmath118 therefore , @xmath116 for @xmath107 , @xmath117 for @xmath10 and @xmath115 for @xmath53 .
when @xmath8 and @xmath110 , @xmath119 is given by eq .
( i-64 ) . therefore , @xmath116 for @xmath120 , @xmath117 for @xmath121 and @xmath115 for @xmath122 .
the extremum of temperature ( when it exists ) is located at the point defined by eqs .
( i-65)-(i-67 ) .
finally , the entropy ( [ t4 ] ) is given by eq .
( i-68 ) .
some curves giving the evolution of the temperature @xmath123 as a function of the scale factor @xmath108 are plotted in fig .
[ temp ] for @xmath7 and @xmath8 . and
we have taken @xmath90 and @xmath3 . ]
we can rewrite the equation of state ( [ b10 ] ) as @xmath125 where @xmath124 is given by eq .
the point @xmath126 where the pressure vanishes is defined by eqs .
( i-70 ) and ( i-71 ) . for @xmath7 , @xmath127 when @xmath92 , and @xmath128 when @xmath94 . for @xmath97
, the pressure is always positive ( @xmath129 ) . for @xmath99 ,
the pressure is negative when @xmath130 and positive when @xmath131 . for @xmath8 , @xmath127 when @xmath92 and @xmath132 when @xmath110 .
for @xmath99 , the pressure is always negative ( @xmath133 ) . for @xmath97 ,
the pressure is positive when @xmath130 and negative when @xmath134 . ,
@xmath135 and @xmath136 as a function of the scale factor @xmath108 for @xmath7 .
we have taken @xmath90 and @xmath3 . ] , @xmath135 and @xmath136 as a function of the scale factor @xmath108 for @xmath8 .
we have taken @xmath90 and @xmath3 . ]
some curves giving the evolution of @xmath137 as a function of the scale factor @xmath108 are plotted in figs .
[ wkposnneg ] and [ wknegnneg ] for @xmath7 and @xmath8 . for the equation of state ( [ b10 ] ) ,
the velocity of sound is given by eq .
the velocity of sound vanishes at the point @xmath138 defined by eqs .
( i-65)-(i-67 ) where the temperature is extremum . at that point
the pressure is extremum with value @xmath139 given by eq .
the case @xmath140 corresponds to an imaginary velocity of sound .
the point @xmath141 where the velocity of sound is equal to the speed of light is defined by eqs .
( i-74)-(i-76 ) .
different cases have to be considered .
we first assume @xmath7 . when @xmath92 , @xmath142 ; when @xmath94 , @xmath143 for @xmath107 and @xmath144 for @xmath53 . for @xmath107 and @xmath97 , @xmath145 is always positive . for @xmath107 and @xmath99 ,
@xmath145 is negative when @xmath146 and positive when @xmath147 .
the velocity of sound is smaller than the speed of light when @xmath148 and larger when @xmath149 . for @xmath53 and @xmath99 , @xmath145 in always negative .
for @xmath53 and @xmath97 , @xmath145 is positive when @xmath146 and negative when @xmath147 . for @xmath53 and @xmath150 ,
the velocity of sound is always smaller than the speed of light .
we now assume @xmath8 .
when @xmath92 , @xmath142 ; when @xmath110 , @xmath151 . for @xmath120 and @xmath97
, @xmath145 is always positive . for @xmath120 and @xmath99 ,
@xmath145 is negative when @xmath146 and positive when @xmath152 . for @xmath122 and @xmath99 , @xmath145 in always negative . for @xmath122 and @xmath97
, @xmath145 is positive when @xmath146 and negative when @xmath152 . for @xmath150 and @xmath153 ,
the velocity of sound is always smaller than the speed of light . for @xmath150 and @xmath154 ,
the velocity of sound is smaller than the speed of light when @xmath148 and larger when @xmath155 .
some curves giving the evolution of @xmath156 as a function of the scale factor @xmath108 are plotted in figs .
[ wkposnneg ] and [ wknegnneg ] for @xmath7 and @xmath8 .
the deceleration parameter is defined by eq .
( i-77 ) . in a flat universe ,
we obtain the relation of eq .
the universe is decelerating when @xmath157 and accelerating when @xmath158 .
for the equation of state ( [ b10 ] ) , we get eq .
the point @xmath159 defined by eqs .
( i-80 ) and ( i-81 ) corresponds to a possible inflexion point ( @xmath160 ) in the curve @xmath60 .
different cases have to be considered .
we first assume @xmath7 . when @xmath92 , @xmath161 ; when @xmath94 , @xmath162 . for @xmath163 ,
the universe is always decelerating ( @xmath157 ) .
for @xmath164 , the universe is accelerating when @xmath165 and decelerating when @xmath166 .
we now assume @xmath8 . when @xmath92 , @xmath167 ; when @xmath110 , @xmath168 . for @xmath164 , the universe is always accelerating ( @xmath158 ) . for @xmath163 ,
the universe is decelerating when @xmath165 and accelerating when @xmath169 .
some curves giving the evolution of @xmath135 as a function of the scale factor are plotted in figs .
[ wkposnneg ] and [ wknegnneg ] for @xmath7 and @xmath8 .
the temporal evolution of the scale factor @xmath60 is determined by the friedmann equation ( [ b4 ] ) . introducing the normalized radius @xmath170 ,
the density ( [ ges2 ] ) can be written as eq .
( i-82 ) . substituting this expression in eq .
( [ b4 ] ) , we obtain the differential equation @xmath171 where @xmath172 and @xmath173 . in general
, we shall select the sign @xmath174 corresponding to an expanding universe ( @xmath175 ) , except in the case of a cyclic universe where both signs of @xmath176 must be considered .
the solution of eq .
( [ dark6 ] ) can be written as in eq .
( i-84 ) or , after a change of variables @xmath177 , as in eq .
( i-85 ) . the solution of eq .
( [ dark6 ] ) can be expressed in terms of hypergeometric functions .
some simple analytical expressions can be obtained for specific values of @xmath178 .
actually , we can have a good idea of the behavior of the solution of eq .
( [ dark6 ] ) by considering asymptotic limits ( see below ) .
the complete solution is represented in the figures by solving eq .
( [ dark6 ] ) numerically . for @xmath7 ,
the universe starts at @xmath179 with a vanishing radius @xmath180 , an infinite density and an infinite pressure ( big bang singularity ) .
when @xmath181 , we recover the solution corresponding to a linear equation of state @xmath182 @xmath183 the final fate of the universe depends on the value of @xmath178 .
different cases must be considered . for @xmath184
, the radius tends to a finite value @xmath185 in infinite time , while the density and the pressure decrease to zero .
when @xmath186 , @xmath187 @xmath188 the density decreases algebraically rapidly . for @xmath189 , the radius tends to a finite value @xmath185 in infinite time , while the density and the pressure decrease to zero .
when @xmath186 , @xmath190 @xmath191 the density decreases exponentially rapidly . for @xmath52 , the evolution of the universe is cyclic .
starting from @xmath180 at @xmath179 ( big bang ) , the radius increases , reaches its maximum value @xmath185 at @xmath192 , then decreases and reaches zero at @xmath193 ( big crunch ) , increases again and so on ( since @xmath194 at @xmath193 , it may not be allowed to extend the solution after @xmath195 ) .
the density is infinite when @xmath180 and vanishes when @xmath185 . when @xmath196 , @xmath197 @xmath198 the universe displays a future peculiarity at @xmath199 since its density vanishes ( the universe `` disappears '' ) while its radius has a finite value @xmath185 . at @xmath192 , the pressure tends to zero for @xmath107 , is constant for @xmath10 , and is infinite for @xmath53 . therefore
, for @xmath53 there is a future singularity ( since @xmath200 at @xmath192 , it may not be allowed to extend the solution after @xmath201 in that case ) . as a function of time for @xmath7
we have represented the cases @xmath184 , @xmath189 , and @xmath52 ( specifically @xmath202 , @xmath189 , and @xmath90 ) .
we have taken @xmath3 .
the universe evolves towards a future peculiarity in which the density tends to zero while the radius tends to a finite value ( see text for details ) . for @xmath53 ,
the pressure is infinite at @xmath192 . for @xmath107 ,
the velocity of sound becomes larger than the speed of light when @xmath203 . ]
the evolution of the scale factor is represented in fig .
[ rayonkposnneg ] in the different cases described above .
some simple analytical results can be obtained in particular cases . for @xmath10 , using the identity @xmath204 we obtain @xmath205 @xmath206 this describes a peculiar cyclic universe ( see sec .
[ sec_cyclic ] ) .
the radius reaches its maximum value @xmath185 with a vanishing density @xmath207 for the first time at @xmath208 .
we can explicitly check that eq .
( [ dark14 ] ) has the asymptotic forms ( [ dark9b ] ) and ( [ dark10 ] ) .
for @xmath189 , using the identity @xmath209 we obtain @xmath210 @xmath211 this describes a universe reaching a peculiarity @xmath185 with @xmath207 in infinite time .
equation ( [ dark16 ] ) has the asymptotic forms ( [ dark9b ] ) and ( [ dark12 ] ) with @xmath212 . for @xmath90
, we have the identity @xmath213 which determines @xmath214 using eq .
( i-85 ) . for @xmath3
, this solution corresponds to the anti - chaplygin gas @xmath215 . for @xmath8 ,
the universe starts at @xmath179 with a vanishing radius @xmath180 , an infinite density , and an infinite pressure ( big bang singularity ) .
when @xmath181 , we recover the solution ( [ dark9b])-([toto1 ] ) corresponding to a linear equation of state .
when @xmath186 , the density tends to a constant value @xmath111 leading to an exponential growth of the radius @xmath216 this late inflationary phase can mimic the effect of the cosmological constant .
this was the original motivation of the chaplygin gas model corresponding to @xmath3 and @xmath90 @xcite .
the evolution of the scale factor is represented in fig .
[ rayonknegnneg ] .
some simple analytical results can be obtained in particular cases . as a function of time for @xmath8 .
we have taken @xmath90 and @xmath3 .
this corresponds to the original chaplygin gas @xcite . ] for @xmath10 , using the identity @xmath217 we obtain @xmath218 @xmath219 we can explicitly check that eq .
( [ dark20 ] ) has the asymptotic forms ( [ dark9b ] ) and ( [ dark18 ] ) with @xmath220 .
for @xmath189 , using the identity @xmath221 we obtain @xmath222 @xmath223 eq .
( [ dark22 ] ) has the asymptotic forms ( [ dark9b ] ) and ( [ dark18 ] ) with @xmath224 . for @xmath90
, we have the identity @xmath225 which determines @xmath214 using eq .
( i-85 ) . for @xmath3 ,
this returns the result of @xcite for the chaplygin gas @xmath40 .
the generalized equation of state ( [ b10 ] ) with @xmath9 and @xmath8 can be used to describe the transition between a phase where the universe has a linear equation of state ( @xmath1 ) and a phase of late inflation where its density is constant ( @xmath226 ) . therefore
, if we take @xmath3 , it describes the transition between the matter era and the dark energy era .
this provides a model of late inflationary universe . for simplicity
, we shall take @xmath10 . in this way
, we obtain a model of late universe that is `` symmetric '' to the model of early universe studied in paper i , corresponding to an index @xmath227 . for @xmath3 , @xmath10 , and @xmath8 , eqs .
( [ b10 ] ) and ( [ ges2 ] ) become @xmath228 @xmath229 where @xmath230 and we have written @xmath231 for @xmath89 . when @xmath232 , the density is approximately constant : @xmath233 .
this gives rise to a phase of late inflation .
the value @xmath111 defines a fundamental lower bound @xmath234 for the density .
we shall identify it with the dark energy density @xmath235 ( see appendix [ sec_pcscales ] ) .
hence , we take @xmath236 this fixes @xmath11 .
this is how the dark energy ( cosmological constant ) is taken into account in our model . when @xmath237 , eq .
( [ s2 ] ) reduces to @xmath238 corresponding to a pressureless universe ( @xmath239 ) .
this describes the matter era .
the conservation of @xmath240 implies @xmath241 .
writing @xmath242 and @xmath243 where @xmath244 is the present fraction of matter ( including baryons and dark matter ) and @xmath245 is the present fraction of dark energy , we obtain @xmath246 where we have used @xmath247 , @xmath248 , @xmath249 , and @xmath250 @xcite . introducing the cosmological length @xmath251 ( see appendix [ sec_pcscales ] ) , we obtain @xmath252 the value of the characteristic length @xmath231 is @xmath253 it corresponds to the typical radius of the universe at the transition between the matter era and the dark energy era .
regrouping these results , we obtain the basic equations of the model @xmath254 @xmath255 @xmath256 the temperature @xmath257 is constant . finally , we note that @xmath258 where we have introduced the cosmological time @xmath15 ( see appendix [ sec_pcscales ] ) .
as the universe expands from @xmath259 to @xmath260 , the density decreases from @xmath261 to @xmath39 , the equation of state parameter @xmath137 decreases from @xmath262 to @xmath263 , and the deceleration parameter decreases from @xmath264 to @xmath263 . _
remark : _ the equation of state ( [ s6 ] ) used to describe the late universe dominated by dark matter and dark energy corresponds to a constant negative pressure ( the velocity of sound @xmath265 ) .
this equation of state is related to , but different from , the linear equation of state @xmath31 .
this linear equation of state describes only the dark energy era while the constant equation of state ( [ s6 ] ) describes _ both _ the matter era and the dark energy era in a unified manner .
furthermore , it can be given a scalar field representation ( see sec .
[ sec_scalar ] ) .
it turns out that the equation of state ( [ s6 ] ) is equivalent to the standard @xmath18cdm model ( see sec .
[ sec_late ] ) , although this was not obvious _ a priori_. the chaplygin gas model @xcite is often presented as a unification of dark matter and dark energy .
we note that a constant ( not linear ) negative pressure ( [ s6 ] ) does the same job . in our opinion
, this gives a new light to the @xmath18cdm model . during the inflationary phase ( @xmath232 )
, the density has an approximately constant value @xmath266 .
the corresponding hubble parameter is @xmath267 numerically , @xmath268 and @xmath269 .
integrating eq .
( [ s9 ] ) , we find that the scale factor increases with time as @xmath270 this corresponds to de sitter s solution .
the radius of the universe increases exponentially rapidly on a timescale of the order of the cosmological time @xmath15 .
the universe exists at any time in the future and there is no singularity . before the late inflation ( @xmath237 ) , the universe is in the matter era . in that case , due to the smallness of @xmath39 , we can take @xmath239 and we recover the standard einstein - de sitter ( eds ) model .
the density is related to the scale factor by @xmath271 .
the friedmann equation ( [ b9 ] ) becomes @xmath272 yielding @xmath273 where we have determined the constant of integration such that @xmath259 at @xmath179 .
this is not quite correct because we should match the matter era to the radiation era ( see sec .
[ sec_simple ] ) , but this is a good approximation . using the approximate expression ( [ s12 ] ) of the scale factor in the matter era , we find that the time at which @xmath274 is @xmath275 numerically , @xmath276 . this marks the end of the matter era and the beginning of the dark energy era .
this numerical value will be revised in sec .
[ sec_gp ] using the exact solution ( [ s15 ] ) of the friedmann equations .
we also have @xmath277 during the matter era , the density decreases algebraically as the universe expands .
the equation of state ( [ s6 ] ) interpolates smoothly between the matter era described by a density @xmath41 and the dark energy era described by a constant density @xmath42 ( cosmological density ) .
it provides therefore a unified description of the late universe . for the equation of state ( [ s6 ] ) ,
the general solution of the friedmann equation ( [ b9 ] ) is @xmath278 the density evolves as @xmath279 the time at which @xmath274 is @xmath280 numerically @xmath281 .
the corresponding density is @xmath282 . for @xmath283 ,
the universe is in the matter era .
its radius increases algebraically according to eq .
( [ s12 ] ) while its density decreases algebraically according to eq .
( [ s14 ] ) .
for @xmath284 , the universe is in the dark energy era .
its radius increases exponentially according to eq .
( [ s10 ] ) while its density remain approximately constant @xmath285 . from eq .
( [ s15 ] ) we obtain the asymptotic behavior @xmath286 with @xmath287 numerically @xmath288 . for @xmath289 , the radius of the universe is larger than the cosmological length @xmath251 . with the time @xmath290 in logarithmic scales .
this figure clearly shows the transition between the matter era ( eds ) and the dark energy era ( de sitter ) . in the dark energy era , the radius increases exponentially rapidly on a timescale of the order of the cosmological time @xmath15 .
this corresponds to a phase of late inflation .
the universe is decelerating for @xmath165 and accelerating for @xmath169 with @xmath291 .
the transition between the matter era and the dark energy era takes place at @xmath292 . coincidentally
, the present universe turns out to be close to the transition point ( @xmath293 ) . ] with the time @xmath290 in linear scales . ] with the time @xmath290 in logarithmic scales . during the late inflation ,
the density remains approximately constant , with the cosmological value @xmath294 representing a lower bound . ] with the time @xmath290 in linear scales .
general relativity ( cosmological constant ) limits the decay of the density to the cosmological value @xmath235 . ] using the results of sec .
[ sec_dec ] , we find that the universe is decelerating for @xmath165 ( _ i.e. _ @xmath295 ) and accelerating for @xmath169 ( _ i.e. _ @xmath296 ) where @xmath297 the time @xmath298 at which the universe starts accelerating is given by @xmath299 this corresponds to the time at which the curve @xmath60 presents an inflexion point .
numerically , @xmath300 , @xmath301 , and @xmath302 . the evolution of the scale factor and density as a function of time are represented in figs .
[ tasansradiationloglog]-[trhosansradiationlinlin ] in logarithmic and linear scales . the hubble radius , the density and the hubble time of the present universe are @xmath303 , @xmath304 , and @xmath305 .
( [ s15 ] ) , the age of the universe is given by @xmath306 numerically , @xmath307 .
according to eq .
( [ s6b ] ) , the present values of the deceleration parameter and of the equation of state parameter are @xmath308 numerically , @xmath309 and @xmath310
. it is striking to note that the present size of the universe @xmath311 is of the order of the scale @xmath312 ( @xmath313 ) .
therefore , we live just at the transition between the matter era and the dark energy era ( see bullets in figs .
[ tasansradiationloglog]-[trhosansradiationlinlin ] ) .
another way to state this result is to say that the present ratio @xmath314 between dark energy and dark matter is of order unity .
this coincidence is intriguing and often referred to as the `` cosmic coincidence problem '' @xcite .
several theories have been proposed to explain why @xmath315 @xcite . however
, this may be just a pure coincidence without deeper reason .
life ( and researchers inquiring about cosmology ) may have emerged @xmath316 after the big bang , precisely at the epoch where @xmath317 .
we leave this `` problem '' open .
the equation of state ( [ b10 ] ) with @xmath9 and @xmath7 leads to a model of non - inflationary universe exhibiting a future peculiarity .
we take the same values of @xmath55 , @xmath178 , @xmath111 , and @xmath231 as in the previous section .
we also take @xmath16 .
the basic equations of the peculiar model are @xmath318 @xmath319 @xmath320 the temperature @xmath257 is constant .
as the universe expands from @xmath259 to @xmath231 , the density decreases from @xmath261 to @xmath262 , the parameter @xmath137 increases from @xmath262 to @xmath261 and the deceleration parameter increases from @xmath264 to @xmath261 . with the time @xmath290 .
the dotted line corresponds to the real evolution of the universe according to the model of sec .
[ sec_solid ] .
the bullet corresponds to the value of the scale factor at the time @xmath321 corresponding to the age of our universe . at this epoch ,
the cyclic universe has deviated relatively strongly from the real evolution . ] with the time @xmath290 .
the density vanishes periodically at the moment where the universe reaches its maximum size . on the other hand , it becomes infinite when the universe contracts to a point ( big bang or big crunch ) . ] for the equation of state ( [ s23 ] ) , the general solution of the friedmann equation ( [ b9 ] ) is @xmath322 the density evolves as @xmath323 this solution describes a peculiar cyclic universe .
the universe reaches its maximum radius @xmath324 for the first time at @xmath325 numerically , @xmath326 . for @xmath283
, we can make the approximation @xmath327 .
therefore , this universe evolves just like the einstein - de sitter ( eds ) universe describing the matter era ( see sec .
[ sec_rp ] ) .
its radius increases algebraically according to eq .
( [ s12 ] ) while its density decreases algebraically according to eq .
( [ s14 ] ) .
when @xmath328 , the radius approaches its maximum value according to @xmath329 correspondingly , the density decreases to zero as @xmath330 for @xmath331 , the radius decreases with time while the density increases ( this corresponds to the solution of eq .
( [ dark6 ] ) with @xmath332 ) .
the radius vanishes at @xmath333 ( big crunch ) , then increases again .
the density vanishes when @xmath274 and is infinite when @xmath259 .
this process continues periodically .
although there is no divergence at @xmath334 , this model is peculiar because , when the radius reaches its maximum value @xmath231 , the density vanishes ( @xmath207 ) , so the universe becomes empty . in a sense
, the universe `` disappears '' at @xmath334 , before re - appearing at @xmath331 , and this in a periodic manner . according to this model
, the universe would disappear for the first time at @xmath326 that is not very far from the present age of the universe @xmath321 .
therefore , if we were living in this cyclic universe , the end of the world ( in the sense of the whole universe ) would take place in about @xmath17 billion years .
we could fear to live in this universe .
however , this universe is always decelerating in the phase of expansion .
therefore , the fact that our universe is accelerating ( an observation that was made only recently @xcite ) shows that we do not live in this cyclic universe . in a sense
, this is reassuring .
the evolution of the scale factor and density as a function of time are represented in figs .
[ cycliqueradius ] and [ cycliquedensity ] . _
remark : _ the equation of state ( [ s6 ] ) corresponds to a constant positive pressure ( the velocity of sound @xmath265 ) .
this model turns out to be equivalent to the cold dark matter model with a negative cosmological constant ( anti-@xmath18cdm model ) , although this was not obvious _ a priori_. this is an example of cyclic universe corresponding to @xmath62 , @xmath335 and @xmath239 .
other types of cyclic universes ( corresponding to @xmath336 , @xmath335 and @xmath239 , or @xmath337 , @xmath73 and @xmath239 ) were studied by friedmann @xcite , einstein @xcite , and lematre @xcite , and were called `` phoenix - universes '' by lematre .
in paper i , we have described the evolution of the early universe , corresponding to the transition between the pre - radiation era and the radiation era , by a single equation of state of the form ( [ b10 ] ) with @xmath2 , @xmath227 and @xmath338 , namely @xmath339 this equation of state leads to a non - singular early inflationary universe .
the density is related to the scale factor by @xmath340 where @xmath341 .
this characteristic scale marks the transition between the pre - radiation era and the radiation era .
when @xmath342 , the density tends to a maximum value @xmath343 identified with the planck density @xmath344 .
this leads to a phase of early inflation @xmath345 whose timescale is the planck time @xmath33 .
when @xmath346 , we recover the equation @xmath347 corresponding to the pure radiation described by an equation of state @xmath34 .
the conservation of @xmath348 implies that @xmath349 .
writing @xmath350 , we can rewrite eq .
( [ early2 ] ) in the form @xmath351 where @xmath352 .
when @xmath346 , it reduces to @xmath353 which corresponds to the pure radiation . in the radiation era , the scale factor , the density , and the temperature evolve as @xmath354 , @xmath355 and @xmath356 . the friedmann equation ( [ b9 ] ) corresponding to eq .
( [ early5 ] ) can be written @xmath357 it has the analytical solution given by eq .
( i-128 ) .
the transition between the pre - radiation era and the radiation era takes place at a typical time @xmath358 corresponding to @xmath359 .
this time also corresponds to the inflexion point of the curve @xmath60 .
the universe is accelerating for @xmath360 and decelerating for @xmath361 .
_ remark : _ the equation of state ( [ early1 ] ) smoothly interpolates between a negative pressure vacuum energy regime ( @xmath31 , @xmath35 ) and a relativistic radiation phase ( @xmath34 , @xmath36 ) .
this amounts to summing the _ inverse _ of the densities of these two components taken independently . indeed , eq .
( [ early5 ] ) can be rewritten as @xmath362 in this paper , we have described the evolution of the late universe , corresponding to the transition between the matter era and the dark energy era , by a single equation of state of the form ( [ b10 ] ) with @xmath3 , @xmath10 and @xmath11 , namely @xmath363 this equation of state leads to a non - singular late inflationary universe .
the density is related to the scale factor by @xmath364 where @xmath365 .
this characteristic scale marks the transition between the matter era and the dark energy era . when @xmath232 , the density tends to a minimum value @xmath366 identified with the cosmological density @xmath367 .
this leads to a phase of late inflation @xmath368 whose timescale is the cosmological time @xmath369 .
when @xmath237 , we recover the equation @xmath370 corresponding to the pressureless matter described by an equation of state @xmath239 .
the conservation of @xmath371 implies that @xmath241 .
writing @xmath242 and @xmath372 , we can rewrite eq .
( [ late2 ] ) in the form @xmath373 when @xmath237 , it reduces to @xmath374 which corresponds to the pure matter . in the matter era , the scale factor and the density evolve as @xmath375 and @xmath355 .
the friedmann equation ( [ b9 ] ) corresponding to eq .
( [ late5 ] ) can be written as @xmath376 or , equivalently , as @xmath377 with @xmath378 .
it has the analytical solution given by eq .
( [ s15 ] ) . this can be rewritten as @xmath379 the evolution of the density is given by @xmath380 the transition between the matter era and the dark energy era takes place at a typical time @xmath281 at which @xmath274 .
the universe is decelerating for @xmath381 and accelerating for @xmath382 where @xmath383 . using the results of sec .
[ sec_solid ] and appendix [ sec_pcscales ] , we can write @xmath384 @xmath385 @xmath386 numerically , @xmath387 and @xmath388 . the age of the universe is given by eq .
( [ s21 ] ) which can be rewritten @xmath389 numerically , @xmath390 . _
remark : _ the equation of state ( [ late1 ] ) smoothly interpolates between a non - relativistic matter phase ( @xmath239 , @xmath41 ) and a negative pressure dark energy regime ( @xmath31 , @xmath42 ) .
this amounts to summing the density of these two components taken independently .
indeed , eq . ( [ late5 ] ) can be rewritten as @xmath391 therefore , the equation of state ( [ late1 ] ) is equivalent to the standard @xmath18cdm model @xcite .
the previous sections reveal a deep `` symmetry '' between the early and the late evolution of the universe . in the early universe ( @xmath227 )
, we have to add the inverse of the densities , and in the late universe ( @xmath10 ) we have to add the densities themselves .
this suggests that the universe must be described by _ two _ equations of state . in order to obtain a single equation describing the whole evolution of the universe
, we just have to add the contribution of radiation ( including pre - radiation ) @xmath392 baryonic matter @xmath393 dark matter @xmath394 and dark energy @xmath395 in the friedmann equation ( [ b9 ] ) .
the radiation dominates in the early universe , while baryonic matter , dark matter and dark energy dominate in the late universe .
now that the radiation phase has been regularized in order to avoid the initial singularity , it is necessary to state explicitly that matter is formed after radiation when @xmath92 while the matter density diverges as @xmath396 which is subdominant .
therefore , we can safely extrapolate the matter density to @xmath259 ( its contribution is negligible anyway ) . in the present model ,
the radiation density ( [ pr1 ] ) does not diverge anymore .
therefore , it becomes necessary to state explicitly that matter is formed after radiation in order to avoid spurious divergences of the matter density when @xmath92 .
] . to that purpose , it is sufficient to add a heaviside function @xmath397 in the matter density .
if we write @xmath398 and @xmath399 , the friedmann equation ( [ b9 ] ) can be written @xmath400 with @xmath401 . for @xmath402 , we recover the standard model of the universe @xcite .
it exhibits a singularity at @xmath179 ( big bang ) . for @xmath403
, we obtain a regularized model without primordial singularity .
the universe always existed in the past but , for @xmath404 , it has a very small radius , smaller than the planck length . at @xmath179 , it undergoes an inflationary expansion in a very short lapse of time of the order of the planck time and connects to the standard model .
a nice feature of this model is its simplicity since it incorporates an `` inflation phase '' in a very simple and very natural manner .
we just have to add a term @xmath405 in the standard equation of @xmath406 given in @xcite .
therefore , the modification implied by eq .
( [ pr7 ] ) to erase the initial singularity is very natural . on the other hand ,
the equation of state ( [ late1 ] ) in the late universe is equivalent to adding the contribution of matter and dark energy individually , as in the standard model @xcite .
this does not bring any modification to the usual equation , which is a virtue of this description since the standard model works well at late times .
due to quantum mechanics and a late inflation corresponding to the dark energy density @xmath39 due to the cosmological constant ( general relativity ) .
the evolution of the early and late universe is remarkably symmetric .
we have represented in dashed line the standard model leading to a primordial singularity ( big bang ) .
the dotted lines corresponds to the model of sec .
[ sec_late ] where the radiation is neglected .
this analytical model provides a good description of the late universe .
we have also represented the location of the present universe that is just at the transition between the matter era and the dark energy era . ]
determined by the planck constant ( quantum mechanics ) to a minimum value @xmath294 determined by the cosmological constant ( general relativity ) .
these two bounds are responsible for the early and late inflation of the universe . in between
, the density decreases as @xmath407 . ] . ]
the evolution of the scale factor with time is obtained by solving the first order differential equation ( [ pr7 ] ) .
this yields @xmath408 where @xmath409 in the standard model and @xmath410 in the regularized model .
the age of the universe is @xmath411 of course , for the determination of the age of the universe , we can neglect the pre - radiation era and take @xmath402 ( strictly speaking , the age of the universe is infinite since it has no origin ; however , we define the age of the universe from the time @xmath179 at which @xmath412 ) .
we obtain the standard result @xmath413 . where the dimensionless parameter @xmath414 is about @xmath415 of unity @xcite . for simplicity ,
we have taken @xmath416 in the numerical applications .
the current value is @xmath417 .
if we take @xmath418 , the age of the universe is @xmath419 . ] actually , we find the same result if we neglect radiation and use the analytical expression ( [ late15 ] ) instead of eq .
( [ pr9 ] ) .
it is rather fortunate that the age of the universe almost coincides with the hubble time @xmath420 . in figs .
[ taloglogcomplet]-[ttemploglogcomplet ] , we have represented the evolution of the radius , density and temperature of the universe as a function of time .
the universe exhibits two types of inflations : an early inflation due to the planck density @xmath344 and a late inflation due to the cosmological ( dark energy ) density @xmath367 .
the early inflation can be described by a polytropic equation of state with positive index @xmath227 .
the late inflation can be described by a polytropic equation of state with negative index @xmath10 .
there exists a striking `` symmetry '' between the early and the late evolution of the universe , the cosmological constant in the late universe playing the same role as the planck constant in the early universe .
in particular , fig . [ trhologlogcomplet ]
shows that the density varies between two bounds @xmath421 and @xmath294 that are fixed by fundamental constants ( see appendix [ sec_pcscales ] ) .
these values differ by a factor of the order @xmath422 .
the early universe is governed by quantum mechanics ( @xmath423 ) and the late universe by general relativity ( @xmath18 ) . .
the universe starts at @xmath179 from a singularity at which the radius is finite but the density and the temperature are infinite .
this is followed by the radiation era , the matter era , and the dark energy era . ] for the sake of completeness , we have also represented in fig .
[ taloglogcompletpos ] the evolution of the scale factor in the model with the new type of initial singularity considered in paper i. this model is obtained by replacing the sign @xmath424 by @xmath425 in eq .
( [ pr1 ] ) and in the following formulae ( we must then take @xmath426 in eq .
( [ pr8 ] ) ) . in that case
, the universe emerges `` from nothing '' at @xmath179 .
it has an infinite density and temperature but a finite radius @xmath341 .
this universe becomes physical only for @xmath427 when the velocity of sound becomes smaller than the speed of light .
_ remark : _ an additional refinement can be introduced if dark matter is made of becs instead of being pressureless . in that case , the term @xmath428 should be replaced by @xmath429 as proposed in @xcite .
it is shown in these papers that the presence of becs can substantially accelerate the formation of the large scale structures of the universe .
we may wonder whether the actual evolution of the universe can be predicted , without making any observation , from the principle of `` simplicity '' and `` aesthetics '' that was defended by great scientists such as einstein , eddington , and chandrasekhar @xcite .
we make the further assumption that the universe is non - singular and non - peculiar .
therefore , we try to construct the simplest non - singular model of universe .
the friedmann equations ( [ b2])-([b4 ] ) governing the evolution of the universe require the value of the curvature of space @xmath56 , the cosmological constant @xmath18 , and the equation of state @xmath69 .
concerning the curvature of space , the most natural guess is @xmath62 ( this leads to the simplest mathematical models ) . actually , this value is consistent with observations . even if an
apparently _ flat universe could be the result of the primordial inflation ( leading to @xmath430 ) @xcite , it may well be that @xmath62 exactly .
this is what we assume .
the cosmological constant can be eliminated from the basic equation since it can be taken into account in the equation of state . concerning the equation of state ,
the simplest guess is to take @xmath239 .
this choice was made by early cosmologists .
a pressureless universe describes the matter era ( eds model ) .
however , this equation of state neglects relativistic effects which become important in the past when the universe was hot .
therefore , cosmologists considered a linear equation of state of the form @xmath1 . for @xmath2 ,
this equation of state describes the ( relativistic ) radiation era .
however , this equation of state is not fully satisfactory since it leads to a primordial singularity ( big bang ) . the early inflation and the late inflation ( dark energy ) are usually described by a linear equation of state of the form @xmath31 leading to a constant density .
however , this equation of state does not contain in itself the value of the constant density ( a feature that we regard as a drawback ) . on the other hand , being linear
, it can not be `` added '' to another linear equation of state @xmath1 to describe the transition between the pre - radiation era ( @xmath5 ) and the radiation era ( @xmath2 ) in the early universe , or the transition between the matter era ( @xmath3 ) and the dark energy era ( @xmath5 ) in the late universe .
the next simplest choice is therefore to consider a mixed equation of state of the form @xmath0 with a linear component and a polytropic component .
it comes naturally from the polytropic model that the indices @xmath50 describe the early universe while the indices @xmath9 describe the late universe .
this suggests that the early and the late universe must be described separately by two different equations of state
. we must take @xmath2 ( radiation ) in the equation of state describing the early universe ( @xmath50 ) and @xmath3 ( matter ) in the equation of state describing the late universe ( @xmath9 ) . by analyzing all the possibilities ,
we have found that only the models with @xmath8 can produce a non - singular ( or non - peculiar ) universe .
the models with @xmath7 lead to past or future singularities / peculiarities .
therefore , the polytropic pressure must be _
negative_. then , it comes naturally from the polytropic model that the density has a maximum value @xmath431 in the early universe ( for @xmath50 ) and a minimum value @xmath234 in the late universe ( for @xmath9 ) .
it is natural to identify the upper bound with the planck density @xmath32 ( quantum mechanics ) and the lower bound with the cosmological density @xmath39 ( general relativity ) .
this determines the constant @xmath56 which is @xmath432 in the early universe and @xmath433 in the late universe . finally , according to the principle of simplicity , it is natural ( but not quite compulsory ) to select the index @xmath434 ( quadratic equation of state ) in the early universe and the index @xmath10 ( constant equation of state ) in the late universe .
this leads to the model represented in figs .
[ taloglogcomplet]-[ttemploglogcomplet ] , featuring two symmetric phases of inflation .
this evolution could have been predicted without making any observation of the universe , by only considering the simplest equation of state that does not yield any singularity or peculiarity .
interestingly , the observations of the real universe are remarkably consistent with that model . in conclusion ,
the most natural non - singular model of universe corresponds to a solution of the einstein equations that exists eternally in the past and in the future .
this has been called _ aioniotic _ universe in @xcite since it has no beginning nor end .
it is the only model of the family of equations of state ( [ b10 ] ) that has no singularity / peculiarity .
for that reason , it may be selected by nature .
furthermore , this universe is very `` symmetric '' in its early and late phases ( corresponding to a vanishing cosmological constant ) is also non - singular and simpler .
however , if we add a principle of `` symmetry '' ( @xmath435 aesthetics ) , in addition to the principle of `` simplicity '' , we are left with the model discussed previously . ] .
this symmetry can be seen in the values of the polytropic indices @xmath434 and @xmath10 that characterize the early and the late universe , respectively .
all the formulae obtained in paper i and in the present paper are symmetric with respect to each other ( compare secs .
[ sec_early ] and [ sec_late ] ) .
this symmetry is also apparent in the two types of inflation that describe the early and late universe : an early inflation due to the planck density @xmath32 ( quantum mechanics ) and a late inflation due to the cosmological density @xmath39 ( general relativity ) .
they correspond to the upper and lower bounds of the density shown in fig .
[ trhologlogcomplet ] , which is also strikingly symmetric .
therefore , this universe is the simplest and most symmetric non - singular cosmological solution of the einstein equations .
of course , some reservations should be made : the universe may not be `` simple '' . in particular , the very early universe ( before the planck time ) may not be described in terms of an equation of state @xmath436 , or even in terms of the einstein equations , as we have assumed .
the description of the very early universe may require the development of a theory of quantum gravity that does not exist for the moment .
an interesting description of the early inflation has been given by @xcite in terms of a quantized model based on a simplified wheeler - dewitt equation .
in that model , a quantum tunneling process explains the birth of the universe with a well defined size after tunneling .
therefore , other inflationary scenarios are possible in addition to the one given in paper i in terms of the generalized equation of state ( [ b10 ] ) . however , our aim was to explore all the consequences of this generalized equation of state , pushing it to its limits .
the cosmological constant @xmath18 is equivalent to a constant density @xmath38 called dark energy .
its value resulting from observations is @xmath367 ( see appendix [ sec_pcscales ] ) .
since the late universe is very dilute , it is oftentimes argued that the cosmological density @xmath39 should correspond to the vacuum energy density due to quantum fluctuations .
however , according to particle physics and quantum field theory , the vacuum energy density is of the order of the planck density @xmath344 which is @xmath422 times larger than the cosmological density .
this leads to the so - called cosmological constant problem @xcite . actually , as illustrated in fig .
[ trhologlogcomplet ] , the planck density and the cosmological density represent fundamental upper and lower density bounds acting in the early and late universe , respectively .
it is not surprising therefore that they are so different : @xmath437 . because of these bounds , the universe undergoes two phases of inflation .
the inflation in the early universe is due to quantum mechanics and is related to the planck density @xmath32 .
the inflation in the late universe is due to the cosmological constant and is related to the cosmological density @xmath39 .
quantum mechanics is negligible in the late universe .
the cosmological constant should be interpreted as a new fundamental constant of physics .
it applies to the very large universe ( cosmophysics ) exactly like the planck constant applies to the very small universe ( microphysics ) .
actually , there is a complete symmetry between the small and large universe where @xmath423 and @xmath18 play symmetric roles .
therefore , we interpret the cosmological constant as a fundamental constant of physics describing the cosmophysics ( late universe ) in the same sense that the planck constant describes the microphysics ( early universe ) . if this interpretation is correct , the origin of the dark energy density @xmath39 should not be sought in quantum mechanics , but in pure general relativity . in this sense ,
the cosmological constant `` problem '' may be a false problem .
if @xmath18 is a fundamental constant of physics , independent from the others , its value should not cause problem .
it is fixed by nature , just like the value of @xmath438 , @xmath439 , and @xmath423 .
one can just expect that @xmath32 is `` very large '' and @xmath39 is `` very small '' .
of course , the origin of the cosmological constant still needs to be understood by developing a theory of cosmophysics .
in addition , it is extremely important to understand why @xmath32 and @xmath39 represent upper and lower bounds , and if these bounds are as fundamental as , for example , the bound on the velocity fixed by the speed of light . _
remark : _ we may note that the appearance of maximum and minimum density bounds from the polytropic equation of state ( [ b10 ] ) is , in some sense , similar to the appearance of the maximum mass of relativistic white dwarf stars discovered by chandrasekhar @xcite also from a polytropic equation of state .
the phase of inflation in the very early universe is usually described by a scalar field @xcite . on the other hand , in alternative theories to the cosmological constant
, the present - day acceleration of the universe ( dark energy ) is also described by a scalar field called quintessence @xcite .
a tachyon field @xcite has also been considered due to its connection with string theory @xcite . actually , to any `` fluid '' described by a barotropic equation of state @xmath69 , it is possible to attach a scalar field with a potential @xmath440 . in this section
, we determine the potential of the scalar field ( quintessence and tachyon field ) corresponding to the equation of state ( [ b10 ] ) using the general methodology exposed in @xcite .
we treat polytropic indices @xmath178 and polytropic constants @xmath56 of arbitrary sign but we assume @xmath84 ( _ i.e. _ @xmath441 ) . the case @xmath85 ( _ i.e. _ @xmath442 ) leads to phantom ( ghost ) fields that are studied in paper iii .
quintessence @xcite is described by an ordinary scalar field @xmath30 minimally coupled to gravity .
the scalar field evolves according to the equation @xmath443 where @xmath440 is the potential of the scalar field .
the quintessence scalar field tends to run down the potential towards lower energies .
the density and the pressure of the universe are related to the scalar field by @xmath444 when the kinetic energy dominates the potential energy , we obtain the equation of state @xmath445 of stiff matter .
when the potential energy dominates the kinetic energy , we obtain the vacuum equation of state @xmath31 . from eq .
( [ quintessence2 ] ) , we get @xmath446 where we have written @xmath447 .
we assume @xmath441 in order to have @xmath448 . using @xmath449 , and the friedmann equation ( [ b9 ] ) valid for a flat universe
, we get @xmath450 for the equation of state ( [ b10 ] ) , using eqs .
( [ ges2 ] ) and ( i-69 ) , and setting @xmath170 , we can rewrite eq .
( [ quintessence4 ] ) in the form @xmath451 with the change of variables @xmath452 we find that @xmath453 for @xmath7 , we obtain @xmath454 and for @xmath8 , we get @xmath455 on the other hand , according to eq .
( [ quintessence2 ] ) , we have @xmath456 for the equation of state ( [ b10 ] ) , using eqs .
( [ ges2 ] ) and ( i-69 ) , we obtain @xmath457 for @xmath7 , since @xmath458 the scalar field potential is explicitly given by @xmath459 for @xmath8 , since @xmath460 the scalar field potential is explicitly given by @xmath461 in these models , @xmath462 .
let us consider particular cases .
\(i ) for @xmath227 , @xmath338 and @xmath2 ( non - singular early universe ) , we obtain @xmath463 and @xmath464 .
\(ii ) for @xmath227 , @xmath465 and @xmath2 ( singular early universe ) , we obtain @xmath466 and @xmath467 .
\(iii ) for @xmath10 , @xmath11 and @xmath3 ( late universe , @xmath18cdm model ) , we obtain @xmath468 and @xmath469 .
\(iv ) for @xmath10 , @xmath16 and @xmath3 ( late universe , anti-@xmath18cdm model ) , we obtain @xmath470 and @xmath471 .
\(v ) for @xmath90 , @xmath8 and @xmath3 ( chaplygin gas ) , we obtain @xmath472 and @xmath473 .
this returns the result of @xcite .
\(vi ) for @xmath90 , @xmath7 and @xmath3 ( anti - chaplygin gas ) , we obtain @xmath474 and @xmath475 .
the case @xmath5 and @xmath7 must be treated specifically ( see appendix a of paper i and appendix [ sec_eosgm ] of the present paper ) . repeating the preceding procedure , we find that the potential of the scalar field is @xmath476 @xmath477 finally , for a linear equation of state @xmath1 with @xmath478 , writing the relation between the density and the scale factor as @xmath479 , we obtain @xcite : @xmath480 @xmath481 where @xmath170 . since @xmath482 ,
the scalar field evolves with time as @xmath483 .
we note that @xmath484 for @xmath4 ( stiff matter ) . on the other hand
, there is no scalar field corresponding to the vacuum equation of state @xmath31 ( _ i.e. _ @xmath5 ) used to describe inflation .
by contrast , the equation of state ( [ b10 ] ) admits a scalar field representation in all cases , in particular for the models with @xmath8 exhibiting a phase of inflation .
this may be an advantage of the polytropic equation of state ( [ b10 ] ) with respect to the linear equation of state @xmath31 .
a tachyon field @xcite has an equation of state @xmath447 with @xmath485 .
this scalar field evolves according to the equation @xmath486 the density and the pressure are given by @xmath487 from these equations , we obtain @xmath488 where we have written @xmath447 . using @xmath449 , and the friedmann equation ( [ b9 ] ) , we get @xmath489 for the equation of state ( [ b10 ] ) , using eqs .
( [ ges2 ] ) and ( i-69 ) , we can rewrite eq .
( [ tachyon4 ] ) in the form @xmath490 with the change of variables @xmath491 we find that @xmath492 on the other hand , from eq .
( [ tachyon2 ] ) , we have @xmath493 for the equation of state ( [ b10 ] ) , using eqs .
( [ ges2 ] ) and ( i-69 ) , we obtain @xmath494 therefore , the scalar field potential @xmath495 is given in parametric form by eqs .
( [ tachyon7 ] ) and ( [ tachyon9 ] ) .
let us consider particular cases .
\(i ) for @xmath227 , we find that @xmath496 .
therefore , we obtain @xmath497 and @xmath498 . in the case
@xmath8 with @xmath499 , where @xmath137 is always between @xmath263 and @xmath262 , we obtain @xmath500 with @xmath462 .
we note that for @xmath227 , @xmath338 and @xmath2 ( early universe ) , the parameter @xmath137 becomes positive when @xmath501 so that the tachyon field is not defined for all times .
similarly , in the case @xmath7 , since @xmath128 when @xmath502 , the tachyon field is not defined for all times .
\(ii ) for @xmath10 and @xmath7 , we find that @xmath503 .
therefore , we obtain @xmath504 and @xmath505 , with @xmath506 . since @xmath128 when @xmath502 , the tachyon field is not defined for all times .
in particular , for @xmath10 , @xmath16 and @xmath3 ( anti-@xmath18cdm model ) , the parameter @xmath137 is always positive so that the tachyon field is not defined .
\(iii ) for @xmath10 and @xmath8 , we find that @xmath507 . therefore , we obtain @xmath508 and @xmath509 with @xmath510 . if @xmath499 , @xmath137 remains between @xmath263 and @xmath262 , so the tachyon field is well - defined . in particular , for @xmath10 , @xmath11 and @xmath3 ( late universe , @xmath18cdm model ) , we find @xmath511 and @xmath512 with @xmath510 .
\(iv ) for @xmath189 and @xmath7 , we find that @xmath513 .
therefore , we obtain @xmath514 and @xmath515 with @xmath516 . since @xmath128 when @xmath502 , the tachyon field is not defined for all times .
\(v ) for @xmath189 and @xmath8 , we find that @xmath517 .
therefore , we obtain @xmath518 and @xmath519 , with @xmath520 . if @xmath499 , @xmath137 is between @xmath263 and @xmath262 , so the tachyon field is well - defined . in particular , for @xmath3 , we have @xmath521 and @xmath522 with @xmath520 .
\(vi ) for @xmath90 , @xmath3 , and @xmath8 ( chaplygin gas ) , we find that @xmath523 is constant .
the tachyon field is not defined in the anti - chaplygin gas ( @xmath90 , @xmath3 , and @xmath7 ) .
the case @xmath5 and @xmath7 must be treated specifically ( see appendix a of paper i and appendix [ sec_eosgm ] of the present paper ) . repeating the preceding procedure , we find that the potential of the scalar field is @xmath524 @xmath525 for @xmath227 and @xmath526 , and for @xmath10 and @xmath527 , corresponding to the period over which the velocity of sound is smaller than the speed of light , the parameter @xmath137 is between @xmath263 and @xmath262 so that the tachyon field is perfectly well - defined . finally , for a linear equation of state @xmath1 with @xmath528 ,
writing the relation between the density and the scale factor as @xmath479 , we obtain @xcite : @xmath529 @xmath530 where @xmath170 .
since @xmath531 , the scalar field evolves with time as @xmath532 .
we note that @xmath484 for @xmath3 . on the other hand
, there is no tachyon field corresponding to the vacuum equation of state @xmath31 ( _ i.e. _ @xmath5 ) used to describe inflation .
by contrast , the equation of state ( [ b10 ] ) admits a tachyon field representation in the models with @xmath8 and @xmath499 exhibiting a phase of inflation .
this may be an advantage of the polytropic equation of state ( [ b10 ] ) with respect to the linear equation of state @xmath31 .
in this paper , and in the previous one , we have carried out an exhaustive study of the generalized equation of state ( [ b10 ] ) , considering all the possible cases .
we have obtained the following results ( we assume here @xmath533 ) : \(i ) for @xmath50 and @xmath8 , the universe undergoes an early inflation .
it starts from @xmath534 with a vanishing radius and a finite density @xmath431 .
its radius increases with time while its density decreases .
the universe exists at any time in the past and is non - singular .
\(ii ) for @xmath50 and @xmath7 , the universe undergoes a new type of primordial singularity .
it starts at @xmath179 with a finite radius and an infinite density . for @xmath535 ,
its radius increases while its density decreases .
\(iii ) for @xmath9 and @xmath8 , the universe undergoes a late inflation .
its radius increases to @xmath261 as @xmath186 while its density decreases to a constant value @xmath234 .
the universe exists at any time in the future and is non - singular .
\(iv ) for @xmath9 and @xmath7 , the universe undergoes a future peculiarity .
its radius increases to a maximum value while its density decreases to zero ( the universe `` disappears '' ) . for @xmath51 , this peculiarity
is reached in infinite time . for @xmath52 , this peculiarity
is reached in a finite time ( and there is a future singularity when @xmath53 due to the divergence of the pressure ) .
then , the radius decreases to zero while the density increases to @xmath261 .
these phases of expansion and contraction continue periodically ( cyclic universe ) .
the generalized equation of state ( [ b10 ] ) can be viewed as a `` mixture '' of a linear equation of state describing a classical universe filled with radiation ( @xmath2 ) or matter ( @xmath3 ) , and a polytropic equation of state whose origin remains to be understood ( in paper i , we have mentioned the connection with bose - einstein condensates , but other possibilities may be contemplated ) .
positive indices describe the early universe ( the polytropic component dominates the linear component when the density is high ) and negative indices describe the late universe ( the polytropic component dominates the linear component when the density is low ) . a positive polytropic pressure ( @xmath7 )
leads to singular or peculiar models , while a negative polytropic pressure ( @xmath8 ) leads to non - singular models that exist for all times .
for @xmath8 , the generalized equation of state ( [ b10 ] ) implies the existence of upper and lower bounds on the density .
it is natural to identify the upper bound with the planck density @xmath421 ( quantum mechanics ) and the lower bound with the cosmological density @xmath294 ( general relativity ) . by taking @xmath227 in the early universe and @xmath10 in the late universe ,
we have obtained a non - singular model that is consistent with the known properties of our universe .
this model improves the standard @xmath18cdm model by removing the primordial singularity .
an attractive feature of this model is its simplicity and the `` symmetry '' that it reveals between the past and the future .
furthermore , this model admits a scalar field representation based on a quintessence field ( in the early and late universe ) or a tachyon field ( in the late universe ) . by using only the principle of `` simplicity '' , and rejecting singular or peculiar solutions ,
we have obtained in a natural manner a simple model of universe .
it corresponds to an `` aioniotic '' universe that exists eternally in past and future and whose early and late evolutions present some striking symmetry .
the remarkable point is that this model , which is obtained in a purely theoretical manner independent of observations , turns out to be fully consistent with the known structure of the real universe .
a theoretical challenge would now be to _ justify _ a quadratic equation of state @xmath536 ( @xmath227 ) in the early universe and a constant pressure @xmath537 ( @xmath10 ) in the late universe .
it may also be interesting to study an equation of state of the form @xmath538 exhibiting two phases of inflation @xcite . for @xmath2
, it provides a unification of pre - radiation , radiation , and dark energy ( see appendix [ sec_unification ] ) .
our study suggests that the planck density @xmath32 and the cosmological density @xmath39 represent fundamental upper and lower bounds .
these bounds are responsible for a phase of early and late inflation .
it is oftentimes argued that the dark matter density @xmath39 should be identified with the vacuum energy .
since the vacuum energy is of the order of the planck density @xmath32 , which is @xmath422 times larger than the cosmological density , this leads to the so - called cosmological constant problem @xcite . actually ,
if the planck density and the cosmological density represent fundamental upper and lower bounds , it is not surprising that they differ by about @xmath19 orders of magnitudes .
therefore , the origin of the cosmological density should not be sought in quantum mechanics but in a new theory of cosmophysics based on general relativity .
we have also studied singular models of universe corresponding to a positive polytropic pressure ( @xmath8 ) .
although these models do not appear to be selected by nature , they are interesting on a mathematical point of view . for @xmath50
, we have obtained a model exhibiting a primordial singularity at @xmath179 at which the radius has a finite value while the density and the temperature are infinite . for @xmath9 ,
we have obtained models exhibiting future peculiarities in which the density of the universe vanishes while its radius reaches a maximum value .
this peculiarity may occur in infinite time ( @xmath51 ) , or periodically in time ( @xmath52 ) .
in particular , we have obtained a simple analytical solution of a cyclic universe ( anti-@xmath18cdm model ) in which the density disappears and re - appears periodically . since this solution is always decelerating ( in the phase of expansion ) ,
this peculiar model is not compatible with the known properties of our universe .
there remains a type of solutions that we have not described so far : the case where the density increases as the universe expands ( @xmath81 ) .
these _ a priori _ un - natural solutions which violate the null dominant energy condition are associated with phantom scalar fields @xcite .
they are studied in paper iii in which we construct models of `` phantom universes '' .
actually , there is a rich recent literature @xcite on these solutions since observations do not exclude the possibility that we live in a phantom universe @xcite .
these models usually predict a future singularity at which the radius and the density of the universe become infinite in a finite time .
this would lead to the death of the universe in a big rip or a cosmic doomsday @xcite .
the main feature of the generalized equation of state ( [ b10 ] ) is its polytropic component @xmath539 .
a polytropic equation of state occurs in many situations of astrophysical interest @xcite .
for example , a polytropic equation of state with index @xmath540 describes adiabatic processes in main sequence stars @xcite .
compact stars such as white dwarfs , neutron stars and bec stars are also described by a polytropic equation of state .
classical white dwarf stars correspond to a polytropic index @xmath541 and relativistic white dwarf stars correspond to a polytropic index @xmath542 @xcite .
newtonian and semi - relativistic bec stars with a self - interaction correspond to a polytropic index @xmath227 @xcite . from a mathematical point of view
, polytropic stars represent a particular class of steady states of the euler - poisson system .
similarly , stellar polytropes represent a particular class of steady states of the vlasov - poisson system @xcite .
polytropic distributions also appear in statistical physics , in relation with generalized thermodynamics @xcite and nonlinear fokker - planck equations @xcite . in previous works , we have studied polytropic equations of state in several situations of astrophysical @xcite , physical @xcite , and biological @xcite interest .
it was therefore natural to consider the application of polytropic equations of state in cosmology also .
in this appendix , we specifically study the equation of state ( [ b10 ] ) with @xmath5 , @xmath9 , and @xmath7 , namely @xmath544 it generalizes the equation of state @xmath31 of the vacuum energy .
for the equation of state ( [ eosgm0 ] ) , the continuity equation ( [ b7 ] ) can be integrated into @xmath545 where @xmath546 and @xmath89 is a constant of integration .
the density is defined for @xmath547 . when @xmath92 , @xmath548 and @xmath98 .
when @xmath94 , @xmath549 .
in the same limit , @xmath101 for @xmath107 , @xmath106 tends to a finite value for @xmath10 , and @xmath105 for @xmath53 .
the thermodynamical equation ( [ t2 ] ) can be integrated into @xmath550 where @xmath551 is a constant of integration . combined with eq .
( [ eosgm1 ] ) , we obtain @xmath552 when @xmath92 , @xmath116 . when @xmath94 , @xmath116 for @xmath107 and @xmath115 for @xmath53 . for @xmath107 ,
the temperature reaches its maximum at @xmath553 @xmath554 the evolution of the density and temperature as a function of the scale factor is plotted in fig .
[ annexedensity ] . .
] the equation of state can be written as @xmath447 with @xmath555 the pressure vanishes ( @xmath556 ) at @xmath557 when @xmath92 , @xmath132 ; when @xmath94 , @xmath128 .
the pressure is negative when @xmath130 and positive when @xmath131 .
the deceleration parameter is given by eqs .
( i-77 ) and ( i-78 ) . together with eq .
( [ eosgm5 ] ) , we obtain @xmath558 the curve @xmath60 presents an inflexion point ( @xmath559 ) at @xmath560 when @xmath92 , @xmath168 ; when @xmath94 , @xmath162 .
the universe is accelerating when @xmath165 and decelerating when @xmath166 .
finally , the velocity of sound is given by @xmath561 we have to distinguish two cases .
for @xmath53 , the velocity of sound is always imaginary .
when @xmath92 , @xmath562 ; when @xmath94 , @xmath144 . for @xmath107 , the velocity of sound vanishes at the point ( [ eosgm4a])-([eosgm4b ] ) where the temperature is maximum .
at that point , the pressure is maximum with value @xmath563 when @xmath92 , @xmath562 ; when @xmath94 , @xmath143 .
the velocity of sound is imaginary when @xmath146 and real when @xmath147 . on the other hand ,
the velocity of sound is equal to the speed of light at @xmath564 the velocity of sound is smaller than the speed of light when @xmath148 and larger when @xmath149 .
the evolution of @xmath137 , @xmath135 , and @xmath156 as a function of the scale factor @xmath108 is represented in fig .
[ annexew ] . , @xmath135 , and @xmath156 as a function of the scale factor @xmath108 .
we have taken @xmath90 . ] , @xmath189 , and @xmath52 ( specifically @xmath202 , @xmath189 and @xmath90 ) . ] setting @xmath170 , the friedmann equation ( [ b9 ] ) can be written @xmath565 where @xmath566 and @xmath173 as explained previously .
we must distinguish three cases . for @xmath184 , @xmath567 @xmath568 the universe starts at @xmath179 with a vanishing radius , an infinite density , an infinite pressure , and a vanishing temperature .
the universe expands and its density decreases .
it reaches its maximum radius @xmath185 in infinite time , while the density and the pressure tend to zero algebraically rapidly . for @xmath189 ,
@xmath569 @xmath570 the universe starts from @xmath534 with a vanishing radius , an infinite density , an infinite pressure , and a vanishing temperature .
the universe expands and its density decreases .
it reaches its maximum radius @xmath185 in infinite time , while its density and pressure tend to zero exponentially rapidly . for @xmath52 , @xmath571 @xmath572 the universe starts from @xmath534 with a vanishing radius , an infinite density , an infinite pressure , and a vanishing temperature .
the universe expands and its density decreases .
when @xmath192 , the universe reaches its maximum radius @xmath185 while its density vanishes ( future peculiarity ) .
the universe `` disappears '' . when @xmath573 , the universe contracts ( this corresponds to the solution of eq .
( [ eosgm12 ] ) with @xmath332 ) and its density increases .
when @xmath186 , the universe has a vanishing radius and an infinite density . at @xmath192 ,
the pressure vanishes for @xmath107 , is constant for @xmath10 , and is infinite for @xmath53 . in this last case , there is a future singularity . for @xmath107 , the velocity of sound becomes larger than the speed of light when @xmath574 corresponding to eq .
( [ eosgm10b ] ) .
this marks the end of the physical universe .
the solutions ( [ eosgm14])-([eosgm13marre ] ) may not have sense for @xmath575 .
the evolution of the scale factor with time is plotted in fig . [ annexekposnneg ] .
it is expected that quantum mechanics plays a fundamental role in the early universe where the scale factor is small . from the planck constant @xmath423 , the speed of light @xmath439 , and the constant of gravity @xmath438
, we can construct characteristic scales , known as the planck scales .
these are the planck density @xmath576 the planck time @xmath577 the planck length @xmath578 the planck mass @xmath579 and the planck temperature @xmath580 we have the obvious relations @xmath581 , @xmath582 , @xmath583 .
we have given in paper i an interpretation of these different scales .
in particular , the planck density is expected to represent a fundamental upper bound for the density leading to a phase of early inflation .
similarly , we expect that the cosmological constant @xmath18 plays a fundamental role in the late universe where the scale factor is large .
we shall regard the cosmological constant as a fundamental constant of nature that describes the cosmophysics ( hence the late universe ) exactly like the planck constant describes the microphysics ( hence the early universe ) .
if this interpretation is correct , we should not seek the meaning of the dark energy in the vacuum energy , or in quantum mechanics .
it would have another origin purely connected to general relativity . in this point of view , the cosmological constant problem is a false problem ( see sec . [ sec_problem ] ) .
however , the true nature of the cosmological constant remains to be found .
it would be necessary to develop the counterpart of quantum mechanics to cosmophysics . for the moment
, there is no such theory . however , using dimensional analysis , we can easily introduce cosmological scales that are the counterpart of the planck scales . from observations , we know the value of the dark energy density @xmath39 .
it is usually written as @xmath243 where @xmath250 is the present fraction of dark energy and @xmath584 is the present density of the universe determined by the hubble constant @xmath585 through the friedmann equation ( [ b9 ] ) .
this yields @xmath367 . from now on , @xmath39 will be called the cosmological density ( the word `` cosmological '' in the late universe replacing the name `` planck '' in the early universe ) . from the cosmological density @xmath39 , the speed of light @xmath439 , and the constant of gravity @xmath438 , we can construct a cosmological time @xmath586 , a cosmological length @xmath587 , and a cosmological mass @xmath588 .
we have @xmath243 , @xmath589 , and @xmath590 .
the dark energy density is related to the cosmological constant by @xmath38 . from now on
, we will consider that @xmath591 is a fundamental constant of physics .
therefore , the corresponding scales @xmath39 , @xmath592 , @xmath593 ... are fundamental scales determined from @xmath18 in the same sense that the planck scales are determined from @xmath423 .
these cosmological scales can be written as the planck scales by introducing the notation @xmath594 numerically , @xmath595 .
this constant might play a fundamental role in the theory of cosmophysics , just like the planck constant @xmath423 plays a fundamental role in the theory of microphysics .
this is , however , a pure speculation that is simply motivated by the apparent `` symmetry '' between the early universe and the late universe ( in any case , @xmath596 can be introduced as a convenient notation ) .
the theory of cosmophysics would then lead to characteristic scales that are the cosmological density @xmath597 the cosmological time @xmath598 the cosmological length @xmath599 the cosmological mass @xmath600 and the cosmological temperature @xmath601 we have given in this paper an interpretation of these different scales ( except for the temperature whose nature is unknown to us ) .
in particular , the cosmological density is expected to represent a fundamental lower bound for the density leading to a phase of late inflation .
actually , these scales were first introduced by einstein @xcite in his static model with a cosmological constant .
although they now have a different meaning ( since the universe is not static ) , the scaling is of course the same .
these scales , coming from the fundamental constant @xmath596 ( a redefinition of @xmath18 ) should ultimately emerge from a theory of cosmophysics that remains to be constructed .
the apparent symmetry between the early and the late universe , may be a source of inspiration in the construction of such a theory .
we have seen in paper i that the transition between the pre - radiation era and the radiation era can be described by an equation of state @xmath602 . on the other hand , the transition between a linear equation of state era and the dark energy era can be described by an equation of state @xmath603 ( see secs . [ sec_ges ] and [ sec_dark ] ) . in sec .
[ sec_solid ] , we have taken @xmath3 to describe the transition between the matter era and the dark energy era . alternatively ,
if we take @xmath2 , we describe the transition between the radiation era and the dark energy era .
finally , the quadratic equation of state @xmath604 describes , in a unified manner , the pre - radiation era , the radiation era , and the dark energy era .
a nice feature of this equation of state is that both the planck density ( early universe ) and the cosmological density ( late universe ) explicitly appear .
therefore , this equation of state reproduces both the early inflation and the late inflation .
this may suggest that the pre - radiation , radiation , and dark energy are the manifestation of a _
unique _ form of `` generalized radiation '' . when @xmath605 , we get @xmath606 ( vacuum energy ) ; when @xmath607 , we get @xmath608 ( dark energy ) ; when @xmath609 , we get @xmath610 ( radiation ) . for the equation of state ( [ unif1 ] ) , the continuity equation ( [ b7 ] ) can be integrated into @xmath611 where @xmath341 ( see paper i ) .
to obtain this expression , we have used the fact that @xmath612 so that @xmath613 . when @xmath92 , @xmath614 ( early inflation ) ; when @xmath110 , @xmath607 ( late inflation ) ; when @xmath609 , @xmath615 ( power - law evolution ) .
we note that eq .
( [ unif2 ] ) can be rewritten as @xmath616 it leads to eq .
( [ pr7 ] ) with @xmath617 .
it is therefore necessary to add `` by hand '' the density of matter ( [ pr4 ] ) and ( [ pr5 ] ) in the friedmann equation ( [ b9 ] ) .
this suggests that matter on the one hand , and pre - radiation @xmath424 radiation @xmath424 dark energy on the other hand should be treated as two different `` species '' .
this is at variance with usual models , including those considered in the main part of this paper , that try to unify matter and dark energy . for the equation of state ( [ unif1 ] ) , the thermodynamical equation ( [ t2 ] )
can be integrated into @xmath618 where we have determined the constant of integration in order to recover the stefan - bolzmann law in the radiation era @xmath619 ( see paper i ) .
combined with eq .
( [ unif2 ] ) , we get @xmath620 this is exactly the same expression as in the pre - radiation @xmath424 radiation era ( see paper i ) but it is also valid in the dark energy era .
this may be a confirmation that dark energy is of the same nature as pre - radiation and radiation . when @xmath342 , @xmath621 ( pre - radiation ) ; when @xmath346 , @xmath622 ( radiation and dark energy )
. simple analytical expressions of the friedmann equation ( [ b9 ] ) with the density - radius relation ( [ unif2 ] ) can be given in particular limits .
in the pre - radiation era , the scale factor increases exponentially as in eq .
( i-119 ) . in the pre - radiation @xmath424 radiation era , the evolution of the scale factor is given by eq .
( i-128 ) ; in the pure radiation era , the scale factor increases algebraically as in eq .
( i-124 ) ; in the radiation @xmath424 dark energy era , the evolution of the scale factor is given by eq .
( [ dark14 ] ) with @xmath623 , @xmath624 , and @xmath625 ; in the dark energy era , the scale factor increases exponentially as in eq .
( [ s10 ] ) .
actually , it is possible to solve the friedmann equation ( [ b9 ] ) with the density - radius relation ( [ unif2 ] ) exactly . introducing @xmath626 and @xmath627 , we obtain @xmath628 which can be integrated into @xmath629 where the constant @xmath630 is determined such that @xmath412 at @xmath179 ( see paper i ) . | we construct models of universe with a generalized equation of state @xmath0 having a linear component and a polytropic component .
the linear equation of state @xmath1 describes radiation ( @xmath2 ) , pressureless matter ( @xmath3 ) , stiff matter ( @xmath4 ) , and vacuum energy ( @xmath5 ) .
the polytropic equation of state @xmath6 may be due to bose - einstein condensates with repulsive ( @xmath7 ) or attractive ( @xmath8 ) self - interaction , or have another origin . in this paper
, we consider negative indices @xmath9 . in that case , the polytropic component dominates in the late universe where the density is low . for @xmath3 , @xmath10 and @xmath11 , we obtain a model of late universe describing the transition from the matter era to the dark energy era .
coincidentally , we live close to the transition between these two periods , corresponding to @xmath12 and @xmath13 .
the universe exists eternally in the future and undergoes an inflationary expansion with the cosmological density @xmath14 on a timescale @xmath15 . for @xmath3 , @xmath10 and @xmath16
, we obtain a model of cyclic universe appearing and disappearing periodically .
if we were living in this universe , it would disappear in about @xmath17 billion years .
we make the connection between the early and the late universe and propose a simple equation describing the whole evolution of the universe .
this leads to a model of universe that is eternal in past and future without singularity ( aioniotic universe ) .
it generalizes the @xmath18cdm model by removing the primordial singularity ( big bang ) .
this model exhibits a nice `` symmetry '' between an early and late phase of inflation , the cosmological constant in the late universe playing the same role as the planck constant in the early universe . we interpret the cosmological constant as a fundamental constant of nature describing the `` cosmophysics '' just like the planck constant describes the microphysics .
the planck density and the cosmological density represent fundamental upper and lower bounds differing by @xmath19 orders of magnitude .
the cosmological constant `` problem '' may be a false problem .
we determine the potential of the scalar field ( quintessence , tachyon field ) corresponding to the generalized equation of state @xmath20 .
we also propose a unification of pre - radiation , radiation and dark energy through the quadratic equation of state @xmath21 . |
during the past decade , much attention has been devoted to accommodate single high - dimensional sources of molecular data ( omics ) in the calibration of prediction models for health traits .
for example , microarray - based transcriptome profiling and mass spectometry proteomics have been established as promising omic predictors in oncology @xcite and , to lesser extent , in metabolic health @xcite .
nowadays , due to technical advances in the field and evolving biological knowledge , novel omic measures , such as nmr proteomics and metabolomics @xcite or nano - lcms and uplc glycomics @xcite are emerging as potentially powerful new biomolecular marker sets . as a result
, it is increasingly common for studies to collect a range of omic measurements in the same set of individuals , using different measurement platforms and covering different aspects of human biology .
this causes new statistical challenges , among which the evaluation of the ability of novel biomolecular markers to improve predictions based on previously established predictive omic sources , often referred as their added or incremental predictive ability @xcite .
an illustrative example of these new methodological challenges is given by our motivating problem .
we have access to data from 248 individuals sampled from the helsinki area , finland , within the dietary , lifestyle , and genetic determinants of obesity and metabolic syndrome ( dilgom ) study @xcite .
one - hundred - thirty - seven highly correlated nmr serum metabolites and 7380 beads from array - based transcriptional profiling from blood leukocytes were measured at baseline in 2007 , together with a large number of clinical and demographic factors which were also measured in 2014 , after seven years of follow - up .
our primary goal is the prediction of future body mass index ( bmi ) , using biomolecular markers measured at baseline .
more specifically , we would like to compare the predictive ability of the available metabolomics and transcriptomics , and to determine if both should be retained in order to improve single - omic source predictions of bmi at seven years of follow - up . in our setting
, it is necessary to both calibrate the predictive model based on each source of omic predictors and assess the incremental predictive ability of a secondary one relative to the first set , using the same set of observations .
hence , in order to avoid overoptimism and provide realistic estimates of performance , it is necessary to control for the re - use of the data , which has already been employed for model fitting within the same observations @xcite .
this is a very important issue in omic research , where external validation data are hard to obtain .
it is well known that biased estimation of model performance due to re - use of the data increases with large number of predictors @xcite and omic sets are typically high - dimensional ( @xmath5 , @xmath6 sample size and @xmath7 the number of predictors ) .
extra difficulties in our setting are the different dimensions ( number of features ) , scales and correlation structure of each omic source , and possible correlation between omic sources induced by partially common underlying biological information . evaluating
the added predictive ability of new biomarkers regarding classical , low dimensional , settings has been a topic of intense debate in the biostatistical literature in the last years ( see , for example , @xcite and references therein ) .
getting meaningful summary measures and valid statistical procedures for testing the added predictive value are difficult tasks , even when considering the addition of a single additional biomarker in the classical regression context .
in particular , widely used testing procedures for improvement in discrimination based on area under the roc curve ( auc ) differences @xcite and net reclassification index ( nri ) @xcite have shown unacceptable false positive rates in recent simulation studies @xcite .
overfitting is a big problem when comparing estimated predictions coming from nested regression models fitted in the same dataset .
moreover , the distributional assumptions of the proposed tests seem inappropriate , translating into poor performance of the aforementioned tests even when using independent validation sets @xcite . to date
, little attention has been given to the evaluation of the added predictive ability in high - dimensional settings , where the aforementioned problems are larger and new ones appear , such as the simultaneous inclusion in an unique prediction model of predictors sets of very different nature .
tibshirani and efron @xcite have shown that overfitting may dramatically inflate the estimated added predictive ability of omic sources with respect to a low - dimensional set of clinical parameters . to solve this issue
, they proposed to first create a univariate ` pre - validated ' omic predictor based on cross - validation techniques @xcite and incorporate it as a new covariate to the regression with low - dimensional clinical parameters . in a subsequent publication , hoefling and tibshirani @xcite have shown that standard tests in regression models are biased for pre - validated predictors . as a solution , the authors suggest a permutation test which seems to perform well under independence of clinical and omic sets .
boulesteix and hothorn @xcite have proposed an alternative method for the same setting of enriching clinical models with a high - dimensional set of predictors .
in contrast to @xcite , they first obtain a clinical prediction based on traditional regression techniques . in a second step ,
the clinical predictor is incorporated as an offset term in a boosting algorithm based on the omic source of predictors .
previous calibration of the clinical prediction is not addressed in the second step and the same permutation strategy than hoefling and tibshirani @xcite is used to derive p - values . in this paper
, we propose a two - step procedure for the assessment of additive predictive ability regarding two high - dimensional and correlated sources of omic predictors . to the best of our knowledge
, no previous work has addressed this problem before .
our approach combines double cross - validation sequential prediction based on regularized regression models and a permutation test to formally assess the added predictive value of a second omic set of predictors over a primary omic source .
let the observed data be given by @xmath8 , where @xmath9 is the continuous outcome measured in @xmath6 independent individuals and @xmath10 and @xmath11 are two matrices of dimension @xmath12 and @xmath13 , respectively , representing two omic predictor sources with @xmath7 and @xmath14 features .
we assume that we are in a high - dimensional setting ( @xmath15 ) and that the main goal is to evaluate the incremental or added value of @xmath16 beyond @xmath17 in order to predict @xmath18 in new observations .
our approach is based on comparing the performance of a primary model based only on @xmath10 with an extended model based on @xmath11 and adjusted by the primary fit based on @xmath10 .
we propose a two - step procedure based on the replacement of the original ( high - dimensional ) sources of predictors by their corresponding estimated values of @xmath18 based on a single - source - specific prediction model . in the first step ,
we build a prediction model for @xmath18 based on @xmath10 and a given model specification @xmath19 .
based on the fitted model , the fitted values @xmath20 are estimated . then , for each individual @xmath21 , we take the residual @xmath22 .
we consider @xmath23 as new response and construct a second prediction model based on @xmath16 as predictor source : @xmath24 this is equivalent to including @xmath25 as an offset term ( fixed ) in the model based on @xmath26 for the prediction of the initial outcome @xmath18 : @xmath27 several statistical methods are available to derive prediction models of continuous outcomes in high - dimensional settings . in this work ,
we focus on regularized linear regression models @xcite , where @xmath28 and the estimation of @xmath29 is conducted by solving @xmath30 , where @xmath31 .
the penalty parameter @xmath32 regularizes the @xmath33 coefficients , by shrinking large coefficients in order to control the bias - variance trade - off .
the pre - fixed parameter @xmath34 determines the type of imposed penalization .
we used two widely used penalization types : @xmath35 ( ridge , i.e. , @xmath36 type penalty @xcite ) and @xmath37 ( lasso , i.e. , @xmath38 penalty @xcite ) . note that other model building strategies for prediction of continuous outcomes could have been used in this framework , such as the elastic net penalization @xcite by setting @xmath39 , or boosting methods @xcite , among others . the use of a previously estimated quantity ( @xmath40 ) in the calibration of a prediction model based on @xmath16 ( expressions ( 1 ) and ( 2 ) ) requires , in absence of external validation data , the use of resampling techniques to avoid bias in the assessment of the role of @xmath40 and @xmath41 .
we use double cross - validation algorithms @xcite , consisting of two nested loops . in the inner loop a cross - validated grid - selection is used to determine the optimal prediction rule , i.e. , for model selection , while the outer loop is used to estimate the prediction performance by application of models developed in the inner loop part of the data ( training sets ) to the remaining unused data ( validation sets ) . in this manner , double cross - validation is capable of avoiding the bias in estimates of predictive ability which would result from use of a single - cross - validatory approach only . in our setting ,
the outer loop of a ` double ' cross - validatory calculation allows obtaining ` predictive'-deletion residuals which fully account for the inherent uncertainty of model fitting on the primary source ( @xmath10 ) , before assessing the added predictive ability of @xmath11 , given by @xmath41 .
the basic structure of the double cross - validation procedure to estimate unbiased versions of @xmath40 and @xmath41 is as follows : step 1 : : obtain double cross - validation predictions of @xmath18 , @xmath42 , based on @xmath43 : + - ; ; randomly split sample @xmath44 in @xmath45 mutually exclusive and exhaustive sub - samples of approximately equal size @xmath46 - ; ; for each @xmath47 , merge @xmath48 subsamples into @xmath49 + - : : randomly split sample @xmath50 in @xmath51 sub - samples @xmath52 - : : for each @xmath53 , merge @xmath54 subsubsamples into @xmath55 + * ; ; fit regression model @xmath56 for a grid of values of shrinkage parameters @xmath57 , @xmath58 to @xmath59 * ; ; evaluate @xmath60,@xmath58 in the @xmath61 held - out sub - sample @xmath62 by calculating @xmath63 - : : compute overall cross - validation error : @xmath64 , @xmath58 - : : choose @xmath65 and calculate predictions of @xmath18 in the @xmath66 held - out sub - sample @xmath67 , @xmath68 , @xmath69 - ; ; the vector of predictions of @xmath18 , @xmath42 is obtained by concatenating the @xmath45 @xmath70 , @xmath47 vectors , i.e. , @xmath71 step 2 : : repeat the process detailed in step 1 considering the double cross - validated residuals @xmath72 , as outcome and @xmath26 as set of predictors and obtain the double cross - validation predictions @xmath73 .
note that this is equivalent to obtaining the double cross - validation predictions of @xmath18 based on @xmath26 considering @xmath25 as offset variable in the @xmath45 fits of model ( 1 ) . in order to evaluate the performance of the sequential procedure introduced in subsection 2.1 .
, we propose three measures of predictive accuracy , denoted by @xmath74 , @xmath75 , and @xmath76 , based on sum of squares of the double cross - validated predictions @xmath40 and @xmath41 , obtained following the procedure described in subsection 2.2 .. these summary measures can be regarded as high - dimensional equivalents of calibration measurements for continuous outcomes in low - dimensional settings @xcite , and an extension of previously discussed proposals in the cross - validation literature @xcite . denote by @xmath77 the prediction sum of squares based on a vector of predictions @xmath78 , obtained according to some arbitrary model @xmath19 , @xmath79 and by @xmath80 the sum of squared differences between two cross - validated vectors of predictions , e.g. @xmath81 , @xmath82 .
let @xmath83 be the simplest cross - validated predictor of @xmath18 , based on the sample mean of @xmath18 only . to summarize the first step of the sequential procedure , we use double cross - validation to estimate the predictive ability of @xmath10 by @xmath84
intuitively , @xmath85 represents the proportion of the variation of the response @xmath18 that is expected to be explained by @xmath86 in new individuals , re - scaled by the total amount of prediction variation in the response @xmath18 . in the worst case scenario , when @xmath87 ( @xmath10 as predictive as a null model based on the mean of @xmath18 ) @xmath88 and @xmath89 if @xmath90 . since the computation of @xmath91 , @xmath92 for each of the @xmath47 random splits of the sample @xmath44
is based on the observations not belonging to @xmath67 , we proceed in an analogous way to compute the average predicted variation of @xmath18 .
hence , in order to get an appropriate re - scaling factor , for each subset @xmath67 , we compute @xmath93 , the mean value of the outcome variable @xmath18 calculated without the observations belonging to @xmath67 .
assume that @xmath94 , the contribution of the second omic source , @xmath11 , in the prediction of @xmath18 can be summarized by @xmath95 @xmath96 accounts for the predictive capacity of @xmath97 , after removing the part of variation in @xmath18 that can be attributed to the first source of predictors @xmath10 .
its computation relies on the squared difference between @xmath98 ( the double cross - validated predictions resulting from the second step of the proposed procedure in subsection 2.2 . ) and the corresponding residual from the step 1 ( @xmath99 ) based on @xmath10 , re - scaled by the remaining predicted variation on @xmath18 after the first step of the procedure . as a result
, @xmath96 can be regarded as the expected ability of @xmath11 to predict the part of @xmath18 , after adjusting for the predictive capacity of @xmath86 and accounting for all model fitting in the first stage of the assessment .
following the same arguments used in deriving expression ( 3 ) , for each subset @xmath67 , the computation of the average variation of @xmath100 is based on @xmath101 , i.e. , excluding the observations belonging to @xmath67 .
note that in step 1 of the sequential procedure @xmath45 models are fitted , each based on @xmath50 , providing residuals with expected zero mean ( given specification ( 1 ) ) , i.e. , @xmath102 , @xmath47 .
hence , @xmath103 and thus @xmath104 finally , we derive a third summarizing measurement of the overall sequential process , @xmath105 , defined as : @xmath106 @xmath107 represents the total predictive capacity of the overall sequential procedure based on @xmath10 and @xmath11 , i.e. , the combined predictive ability of @xmath10 and @xmath11 given by @xmath108 . note that @xmath107 is based on the same squared difference between @xmath98 and @xmath99 as @xmath96 , but the re - scaling factor refers to the total predictive variation of the original response @xmath18 .
the three introduced measures jointly summarize the performance of the two omic sources under study and their interplay in order to predict the outcome @xmath18 .
in all the cases higher values are indicative of higher predictive ability .
the three measurements vary between 0 ( null predictive ability ) and the maximal value of 1 .
the interpretation of @xmath74 is straightforward , as it simply captures the predictive capacity of the firstly evaluated omic source .
note that the difference between @xmath75 and @xmath76 relies on the denominator .
in general , if @xmath10 is informative , the denominator in expression ( 4 ) will be smaller than in expression ( 6 ) .
thus , the residual variation after step 1 will be smaller than the total initial variation .
the three summary measures are related by the following expression : @xmath109 consequently , we can rewrite @xmath110 as follows : @xmath111 note that in cases in which @xmath112 , we get that @xmath113 , and viceversa .
however , @xmath96 and @xmath114 differ when not zero .
specifically , from expression ( 8) , we obtain that @xmath115 . in short ,
@xmath96 may be regarded as the conditional contribution of @xmath11 for the prediction of @xmath18 with respect to what may be predicted using @xmath10 alone .
@xmath116 measures the absolute gain in predictive ability from adding @xmath26 to @xmath10 .
note that a given source @xmath11 may present a large conditional @xmath96 but a small absolute @xmath116 ( if , for example , @xmath10 presents high predictive ability itself ) .
moreover , due to the relation between @xmath25 and the resulting vector of predictions after combining @xmath10 and @xmath11 , @xmath108 , expression ( 8) implies that @xmath117 . this desirable property may not be fulfilled using alternative combination strategies . in practice ,
our sequential procedure relies on the realistic assumption of positive predictive ability of the first source of predictors , @xmath10 ( one would only be interested in assessing additional or incremental information on top of an informative source itself ) .
accordingly , we advise to conduct our sequential procedure using @xmath10 as primary source only if @xmath118 , which is , furthermore , required to derive expression ( 8) .
the summary measures may be used to introduce formal tests for assessing the added or augmented predictive value of @xmath11 over @xmath10 to predict @xmath18 .
we propose a permutation procedure to test the null hypothesis @xmath119 against the alternative hypothesis @xmath120 .
the test is based on permuting the residuals obtained after applying the first step of our two - stage procedure with the data at hand .
our goal is to remove the potential association between @xmath16 and @xmath18 while preserving the original association between @xmath18 and @xmath17 .
explicitly , we propose the following algorithm : step 1 : : calculate the residuals @xmath121 based on the predictions @xmath122 of @xmath18 based on @xmath10 , obtained in the first step of the procedure presented in section 2.1 . step 2 : : permute the values of @xmath100 , obtaining @xmath123 and generate values of the response @xmath18 under the null hypothesis : @xmath124 .
step 3 : : repeat the two - stage procedure from section 2.1 . for predicting @xmath125 and obtain the corresponding @xmath126 .
the procedure is repeated @xmath127 times and the resulting permutation p - values are obtained as follows : @xmath128 where @xmath127 is the number of permutations , and @xmath96 is the actual observed value with the data at hand .
note that in * step 2 * , we generate a ` null ' version of the original response @xmath18 and then we repeat the overall two - stage procedure , which implies that the ` null ' residuals used in * step 3 * are not fixed and are , in general , different from @xmath123 .
this is necessary in order to capture all the variability of the two - stage procedure and to correctly generate the null hypothesis of interest .
moreover , the cross - validation nature of the procedure protects against systematic bias of the residuals obtained in step 3 based on @xmath129 ( see chapter 7 of * ? ? ?
. given the aforementioned relations between @xmath130 , @xmath110 , and @xmath107 specified by expression ( 6 ) , note that @xmath119 is equivalent to @xmath131 .
this result immediately follows from expression ( 6 ) , given that @xmath132 if and only if @xmath133 ( assuming @xmath134 ) .
hence , both tests are equivalent provided that the distribution under the null hypothesis is generated by the aforementioned permutation procedure , i.e. , the p - values , resulting from using @xmath96 as test statistic or @xmath135 are approximately the same . due to the reliance of our method on resampling techniques ,
computational cost is a potential limitation of our approach .
however , our method is easy to split in @xmath127 independent realizations of the same computational procedure .
hence , we can use parallel computing to speed up the procedure @xcite .
we simulate two omic predictor sources @xmath17 ( dimension @xmath12 ) and @xmath16 ( dimension @xmath13 ) and a @xmath136 vector @xmath18 , the continuous outcome .
we use matrix singular value decomposition ( svd , @xcite ) of each of the two omic sources to generate common ` latent ' factors associated with @xmath17 , @xmath16 and @xmath18 .
common eigenvectors in the svd of @xmath10 and @xmath11 introduce correlation among the omic sources .
we consider different patterns in terms of the conditional association between @xmath11 and @xmath18 ( see figure 1 ) .
the details of the data generation procedure are as follows : ( x1 ) @xmath10 ; ( x2 ) @xmath11 ; ( lj ) @xmath137 ; ( l1 ) @xmath138 ; ( l2 ) @xmath139 ; ( y ) @xmath18 ; ( l1 ) ( x1 ) ; ( l2 ) ( x2 ) ; ( lj ) ( x2 ) ; ( lj ) ( x1 ) ; ( lj ) to [ out=120,in=120 , looseness=0.5 ] ( y ) ; ( x1 ) @xmath10 ; ( x2 ) @xmath11 ; ( lj ) @xmath137 ; ( l1 ) @xmath138 ; ( l2 ) @xmath139 ; ( y ) @xmath18 ; ( l1 ) ( x1 ) ; ( l2 ) ( x2 ) ; ( lj ) ( x2 ) ; ( lj ) ( x1 ) ; ( l1 ) to [ out=120,in=120 , looseness=0.5 ] ( y ) ; ( l2 ) to [ out=270,in=270 , looseness=0.5 ] ( y ) ; * step 1 * generate @xmath140 , a matrix of @xmath141 i.i.d .
latent factors of @xmath17 and @xmath16 .
* step 2 * define @xmath142 ( @xmath143 ) and @xmath144 ( @xmath145 ) , the correlation matrices of @xmath10 and @xmath11 , respectively , according to a predefined covariance structure of interest .
following the recent literature on pathway and network analysis of omics data @xcite , we generated @xmath146 , @xmath147 according to a hub observation model ( * ? ? ?
* see figure 2 ) .
* step 3 * draw @xmath148 , @xmath147 , and obtain the singular value decomposition for each of the independent matrices @xmath149 and @xmath150 : @xmath151 .
* step 4 * generate the final correlated @xmath10 and @xmath11 by manipulation of @xmath152 and @xmath153 , the left eigenvectors matrices from @xmath154 and @xmath155 , respectively .
specifically , for a certain number ( @xmath156 ) of predefined columns @xmath157 and @xmath158 , the original submatrices @xmath159 and @xmath160 ( independent ) are replaced by @xmath156 common independent latent factors @xmath161 , @xmath162 generated in * step 1*. in this manner , correlation between @xmath10 and @xmath11 is induced , while the within - omic source correlation structures @xmath163 and @xmath164 are preserved . *
step 5 * simulate the outcome @xmath165 , where @xmath166 and @xmath167 are vectors of regression coefficients of length @xmath7 and @xmath14 , respectively and @xmath168 . since @xmath169 , @xmath147 , we can rewrite @xmath170 and thus @xmath171 , where @xmath172 represents the association between @xmath173 and the outcome @xmath18 through the orthogonal directions given by @xmath174 . consequently , we first generate @xmath175 and we then transform it to the predictor space by using @xmath171 .
( @xmath176 ) , 4 groups of 250 features each .
right : correlation matrix of @xmath16 ( @xmath177 ) , 2 groups of 50 features each.,width=188,height=188 ] ( @xmath176 ) , 4 groups of 250 features each .
right : correlation matrix of @xmath16 ( @xmath177 ) , 2 groups of 50 features each.,width=188,height=188 ] * simulation 1 ( ` null ' scenarios ) : * the second omic @xmath11 source is non - informative , i.e. , @xmath178 , but is strongly correlated to @xmath10 , by imposing common first columns of @xmath179 and @xmath180 ( @xmath181 , the correlation between omic sources is driven through the maximal variance subspace ) .
we considered different assumptions regarding the regression dependence of @xmath18 on @xmath10 which has an impact on the ability to calibrate prediction rules based on @xmath17 for @xmath18 .
we consider two situations in which the association with @xmath18 is unifactorial , in the sense that only one latent factor ( one column of @xmath182 ) is associated with @xmath18 and two multi - factorial situations .
one of our objectives is to illustrate how changing the complexity of the calibration of a prediction rule based on @xmath10 ( by formulating the problem through regression on either larger or smaller variance latent factors ) may affect the results .
we consider the following ` null ' scenarios : scenario 1a : : @xmath183 , @xmath184 ; @xmath185 .
@xmath18 is associated to high - variance subspace of @xmath179 , corresponding to the largest eigenvalue of @xmath10 .
scenario 1b : : @xmath186 , @xmath187 , @xmath188 .
the association with @xmath18 relies on a low - variance subspace of @xmath179 .
hence , we expect lower values of @xmath74 , compared to scenario 1a . scenario 1c : : @xmath189 , @xmath190 otherwise . in this setting we consider a bifactorial regression , as association with @xmath18 is a combination of the effect of the two first eigenvectors of @xmath17 .
scenario 1d : : @xmath191 , @xmath190 otherwise . in this
setting we consider a multifactorial regression , as association with @xmath18 is a combination of the effect of the four first eigenvectors of @xmath17 .
* simulation 2 ( ` alternative ' scenarios ) : * @xmath11 is associated with @xmath18 through latent factors non - shared with @xmath10 .
the following ` alternative ' scenarios are investigated : scenario 2a : : @xmath192 , @xmath184 , @xmath193 , @xmath194 .
the eigenvector related to the largest eigenvalue of each source is associated to @xmath18 and the association between @xmath10 and @xmath11 is generated by sharing the second eigenvectors , i.e. , by setting @xmath195 .
scenario 2b : : @xmath196 , @xmath184 , @xmath197 , @xmath194 and @xmath198 , @xmath199 , @xmath200 , @xmath201 , and the association between @xmath10 and @xmath11 is generated by setting @xmath181 .
scenario 2c : : @xmath196 , @xmath187 , @xmath197 , @xmath194 and @xmath198 , @xmath199 , @xmath200 , @xmath201 , and the association between @xmath10 and @xmath11 is generated by setting @xmath181 .
figure 3 shows a monte carlo approximation based on a sample of @xmath202 observations of the regression coefficients @xmath166 and @xmath167 in the studied simulated scenarios . from panels 3(a ) to 3(d ) , we can observe that the different simulation settings differ in the level of imposed sparsity in the association between @xmath203 and @xmath10 . on the one hand ,
scenarios presented in panels 3(a ) ( scenarios 1a , 2a ad 2b ) and 3(b ) ( scenario 1b ) are relatively sparse , with most of the simulated coefficients close to zero . on the other hand , the @xmath166 of scenario 1c ( represented in panel 3(c ) ) and specially of scenario 1d ( represented in panel 3(d ) ) are less sparse , based on a large number of non - null regression coefficients in @xmath10 . with regard to @xmath167 , panel 3(e ) ( scenario 2a ) represents a sparser situation than panel 3(f ) ( scenarios 2b and 2b ) . , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] + , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] , y - axis ) corresponding to each of the @xmath7 predictors of @xmath10 ( x - axis ) .
( c)-(d ) : regression coefficients ( elements of @xmath167 , y - axis ) corresponding to each of the @xmath14 predictors of @xmath11 ( x - axis ) . the outcome variable is generated as @xmath165 , @xmath168.(e)-(f ) provide information about association between @xmath18 and @xmath17 and ( e ) and ( f ) corresponds to the independent association between @xmath18 and @xmath16 in the alternative scenarios ( for the null scenarios 1a-1d the independent association between @xmath18 and @xmath16 is null ) .
( a ) corresponds to scenarios 1a , 2a and 2b , ( b ) correspond to scenarios 1b and 2c respectively , while ( c ) and ( d ) correspond to scenario 1c and 1d .
( e ) shows the association ( @xmath167 ) between @xmath16 and @xmath18 in scenario 2a and ( e ) shows @xmath167 for scenarios 2b and 2c.,title="fig : " ] in our basic setting , we considered @xmath204 observations , @xmath176 features in @xmath10 and @xmath177 features in @xmath11 . for each scenario , we provide the mean values and standard deviations of @xmath85 , @xmath205 , and @xmath107 , based on 5-folds double cross - validation , jointly with the rejection proportions for testing @xmath206 along @xmath207 monte carlo trials .
we evaluated the permutation test introduced in subsection 2.3 . using @xmath208 permutations .
we complemented our empirical evaluations of the proposed sequential double cross - validation procedure by extending our basic simulation setting in two directions .
we checked the impact on modifying sample size ( @xmath209 ) and the complexity of the problem by varying the number of variables considered in the first stage ( @xmath210 ) .
additionally , we compared the performance of our procedure based on double - cross validation with two alternative strategies . on the one hand , we provide results based on a two - stage procedure using a single cross - validation loop ( cross - validation is used for model choice but predictions and therefore the residuals used as outcome in the second stage are directly computed on the complete sample ) . on the other hand
, we check the impact on the results of over - penalization . specifically , instead of taking @xmath211 as defined in the inner loop of the double cross - validation procedure presented in subsection 2.1 . , we choose a larger value for @xmath32 , namely @xmath212 .
both are usual strategies in penalized regression in single omic prediction frameworks , so it is of practical interest to quantify their impact from the added predictive value point of view .
the results of these alternative strategies are provided as supplemental material but discussed in the main text .
the results for the sequential double cross - validation procedure ( labeled as ` cv type= @xmath213 , @xmath211 ' ) are summarized in tables 1 and 2 .
the top part of each table contains the results concerning the ` null ' scenarios ( no added value of @xmath11 ) , while the bottom part shows the results of the ` alternative ' scenarios ( added value of @xmath11 ) .
table 1 contains results based on ridge regression , @xmath35 in expression ( 1 ) , while table 2 summarizes the results for the lasso penalty type ( @xmath37 ) . for
the four ` null ' scenarios 1a-1d , given that @xmath11 is not independently associated to @xmath18 , we expect @xmath112 and rejection proportions of @xmath214 about 0.05 .
the results of the sequential double cross - validation procedure based on ridge regression are satisfactory in this regard , with rejection proportions close to the nominal level in all the studied null scenarios and for different sample sizes ( @xmath209 , @xmath204 ) and levels of complexity of the first step ( @xmath176 , @xmath210 ) .
the top part of table 1 shows that the estimated @xmath215 for scenarios 1a , 1c and 1d are large and very similar ( @xmath216 ) .
as it was expected , the estimated predictive ability of @xmath10 is lower in scenario 1b and presents a larger variability , since the association between @xmath18 and @xmath10 relies on a small variance subspace . in general , for all 1a-1d scenarios the estimated @xmath96 is close to zero .
however , we observe that the sample size influences the estimated @xmath85 and hence , due to the correlation between @xmath10 and @xmath11 , also affects the estimation of @xmath96 .
we observe systematically lower values of @xmath217 for @xmath209 than for @xmath204 in all the studied ` null ' scenarios .
this feature translates in systematically larger values of @xmath96 for @xmath209 than for @xmath204 .
however , the permutation test is able to account for this issue and the level of the test is respected independently of the sample size .
analogously , increasing the number of features of the first source @xmath10 ( from @xmath176 to @xmath210 ) while keeping fixed the number of features of @xmath11 ( @xmath177 ) also affects the estimation of @xmath217 and @xmath96 . in this case , the values of @xmath217 are larger and hence , the values of @xmath96 tend to be closer to zero .
worth noting is that the level of the test is also well respected in this case .
the bottom part of table 1 shows the results for the alternative scenarios . as desirable
, the power increases with sample size for all the three studied alternative scenarios .
as it was the case for the ` null ' scenarios , increasing the sample size tends to lead to better predictive ability of @xmath10 .
this result matches intuition , since larger sample sizes provide more information for model building , and hence , under correct model specification , the resulting predicting models are expected to behave better in new data .
an exception to this is scenario 2c , where our double cross - validation procedure seems to overfit with @xmath209 .
this is due to the fact that scenario 2c , unlike scenarios 2a and 2b , is characterized as a ` difficult ' prediction problem when considering @xmath10 ( association with @xmath18 is driven by a low - variance subspace of @xmath10 ) . in line with this , the power of the test is different for the three different studied scenarios . the greatest power is reached in scenario 2a , in which the independent association between @xmath11 and @xmath18 is driven through the subspace of maximum variation and the first step of the procedure relies on a relatively ` easy ' prediction problem . even if scenarios 2b and 2c are based on the same independent association between @xmath16 and @xmath18 , the impact of the first source on the power of the test is large .
scenario 2b , in which @xmath218 for @xmath204 reaches a power of 71 % , while the rejection rate reduces to 19% in scenario 2c , corresponding to a more ` difficult ' prediction problem in the first stage , reflected in a low and unstable @xmath215 ( @xmath219 for @xmath209 and @xmath220 for @xmath204 ) .
[ table1 ] table 2 shows the results for double cross - validation procedure based on the lasso specification ( @xmath37 ) . with regard to the ` null ' scenarios ,
we observe a good performance for scenarios 1a and 1b , with rejection proportions close to the nominal level .
interestingly , the rejection proportion of the permutation test increases with sample size and the number of features in the first source in scenarios 1c and 1d , which indicates a bad performance of the procedure based on laso regression in these settings .
namely , the bad performance of the lasso specification for scenario 1d does not improve by increasing sample size ( 7% of rejections with @xmath209 , 9% of rejections with @xmath204 and 36% of rejections for @xmath210 ) .
the reason behind this difference with the ridge - based results is the mis - specification of the lasso with respect to the underlying data - generating mechanism .
lasso regression assumes that the true model is sparse , while , as mentioned , scenario 1c and specially 1d correspond to non - sparse solutions .
these findings illustrate how model mis - specification may result in an improvement of predictions by adding a second source of predictors , not because of independent association to the outcome , but just because of the correlation with the first source of predictors .
the bottom part of table 1 shows the results for the alternative scenarios . with respect to the alternative scenarios ( bottom part of table 2 ) , the conclusions are similar to those observed for ridge regression .
the power increases with the sample size , and the rejection proportions differ across the three scenarios .
however , we observe that ridge outperforms lasso in terms of power , specially for scenarios 2a and 2b . tables s3 and s4 summarize the results for the two aforementioned alternative strategies in the basic setting ( @xmath176 , @xmath177 ) and two sample sizes ( @xmath209 , @xmath204 ) : ` @xmath213 , @xmath221 ' corresponds to the strategy in which the sequential double cross - validation is over - shrunk ( by taking @xmath222 instead of @xmath211 in the inner cross - validation loop ) and ` @xmath223 , @xmath211 ' represents the sequential procedure based on one single cross - validation loop ( standard residuals as opposed to deletion - based residuals ) . in general ,
these two alternative strategies provide different estimates for the predictive ability of the two studied sources of predictors . taking the double - cross validation approach as gold - standard
, we observe that the over - shrinkage of the predictions in the first step of the ` @xmath213 , @xmath221 ' method provokes an under - estimation of @xmath215 , while the ` @xmath223 , @xmath211 ' provides an over - estimation , specially when the association between outcome and first source of predictors is driven through a low - variance space .
for example , in the scenario 1b , for @xmath204 , @xmath224 when based on a single cross - validation approach , notably larger than @xmath225 estimated by the double cross - validation approach . moreover , we observe that the effect of re - using the data is larger for small sample sizes , with systematically larger @xmath215 for @xmath209 than for @xmath204 . however , under the null hypothesis , the introduced bias on the first step for both alternatives does not translate in an inflated type i error . the method labeled as ` @xmath213 ,
@xmath221 ' , based on double cross - validation but based on under - fitting by over - penalization controls the false discovery rate under the null hypothesis in similar fashion than the procedure introduce in subsection 2.1 . with regard to the method based on single cross - validation ( ` @xmath223 , @xmath211 ' ) , its behavior is slightly conservative under the null hypothesis . for the alternative scenarios , as @xmath85 , @xmath96 and @xmath105 , are underestimated by ` @xmath213 , @xmath221 ' , while ` @xmath223 , @xmath211 ' overfits both .
even if power increases with sample size , both methods are systematically less powerful than our proposal,@xmath213 , @xmath211 , which makes it the preferable method from both an estimation and testing point of view .
to illustrate the performance of the proposed sequential double cross - validation procedure , and to compare it to the alternative strategies discussed in section 3 , we analyzed data from the dilgom study .
we are interested in the ability of serum nmr metabolites and microarray gene expression levels in blood to predict body mass index ( bmi ) at 7 years of follow - up .
the metabolomic predictor data consists of quantitative information on 137 metabolic measures , mainly composed of measures on different lipid subclasses , but also amino acids , and creatine .
the gene expression profiles were derived from illumina 610-quad snparrays ( illumina inc . ,
san diego , ca , usa ) .
initially , 35,419 expression probes were available after quality filtering .
in addition to the pre - processing steps described by @xcite , we conducted a prior filtering approach and removed from our analyses probes with extremely low variation ( see @xcite for details on the conducted pre - processing ) . as a result , we retained measures from 7380 beads for our analyses .
the analyzed sample contained @xmath226 individuals for which both types of omic measurements and the bmi after 7 years of follow - up ( mean=26 @xmath227 , sd=5 @xmath227 ) were available .
we carried out two distinct analyses using the added predictive value assessment approach described in this paper .
as a first analysis , we consider the metabolic profile as primary omic source for the prediction of the log - transformed bmi and we evaluated the added predictive value of blood transcriptomics profiles .
this approach is the most relevant in practice , because of both biological and economical reasons .
on the one hand , metabolome ( which contains , among other , cholesterol measures ) is presumably more predictive of bmi than gene expression in blood . on the other hand ,
nmr technology is typically more affordable @xcite than available technologies for transcriptomic profiling , so favoring the nmr source seems a sensible approach in our setting .
nevertheless , to illustrate the properties of our method , we also consider a second analysis in which we reversed the roles of the omic sources , first fitting a model based on gene expression and then evaluating the added predictive value of the metabolome . as in the simulation study , we considered ridge and lasso regression as prediction models , using the same alternative strategies to the sequential double cross - validation procedure presented in section 3 ( ` @xmath213 , @xmath211 ' ) : ` @xmath213 , @xmath221 ' , and ` @xmath223 , @xmath211 ' .
the main findings are summarized in table 3 . to check stability of the results
, we artificially reduced the sample size of the available dilgom data and checked the impact on the estimation of the added predictive ability with our sequential double cross - validation approach and its corresponding p - value .
we also compared our method with a naive approach , consisting in stacking both metabolites and transcriptomics , and hence ignoring their different origin .
the results of these two additional analyses are given as supplemental materials .
.application to dilgom data .
alternative cross - validation strategies .
p - values based on 1000 permutations [ cols="^,^,^,^,^,^",options="header " , ] [ tab : addlabel ]
simulation results based on two modification of the two - stage procedure presented in section 2 : @xmath223,@xmath211 relies on single cross - validation ( cross - validation is used for model choice but predictions and therefore the residuals used as outcome in the second stage are directly computed on the complete sample ) ; @xmath213,@xmath221 relies on over - penalization . specifically , instead of taking @xmath211 as defined in the inner loop of the double cross - validation procedure presented in subsection 2.1 .
, we choose a larger value for @xmath32 , namely @xmath212 . simulations for these two approaches are based on the same specifications detailed in subsection 3.1 . , considering @xmath176 and @xmath177 . | enriching existing predictive models with new biomolecular markers is an important task in the new multi - omic era .
clinical studies increasingly include new sets of omic measurements which may prove their added value in terms of predictive performance .
we introduce a two - step approach for the assessment of the added predictive ability of omic predictors , based on sequential double cross - validation and regularized regression models .
we propose several performance indices to summarize the two - stage prediction procedure and a permutation test to formally assess the added predictive value of a second omic set of predictors over a primary omic source .
the performance of the test is investigated through simulations .
we illustrate the new method through the systematic assessment and comparison of the performance of transcriptomics and metabolomics sources in the prediction of body mass index ( bmi ) using longitudinal data from the dietary , lifestyle , and genetic determinants of obesity and metabolic syndrome ( dilgom ) study , a population - based cohort from finland . * sequential double cross - validation for assessment of added predictive ability in high - dimensional omic applications * + mar rodrguez - girondo@xmath0 , perttu salo@xmath1 , tomasz burzykowski@xmath2 , markus perola@xmath1 , jeanine houwing - duistermaat@xmath3 and bart mertens@xmath0 @xmath0department of medical statistics and bioinformatics , leiden university medical center , leiden , the netherlands .
@xmath1national institute for health and welfare , helsinki , finland .
@xmath2interuniversity institute for biostatistics and statistical bioinformatics ( i - biostat ) , hasselt university , hasselt , belgium .
@xmath4department of statistics , leeds university , leeds , united kingdom .
* keywords : * _ added predictive ability ; double cross - validation ; regularized regression ; multiple omics sets _ |
let @xmath2 be a noetherian local ring that admits a surjection from an @xmath3-dimensional regular local ring @xmath4 containing its residue field @xmath5 , and @xmath6 be the kernel of the surjection . in @xcite
, lyubeznik introduced a new set of invariants @xmath7 as the @xmath8-th bass number of the local cohomology module @xmath9 , that is @xmath10 and they depend only on @xmath2 , @xmath11 and @xmath8 , but not on the choice of @xmath12 or the surjection @xmath13 . in the seminal works of huneke - sharp @xcite and lyubeznik @xcite it is proven that these bass numbers are all finite .
denote @xmath14 , lyubeznik numbers satisfy the following properties : * @xmath15 implies @xmath16 .
* @xmath17 . * _ euler characteristic : _
@xmath18 therefore , we can collect them in the so - called _ lyubeznik table _ : @xmath19 and we say that the lyubeznik table is _ trivial _ if @xmath20 and the rest of these invariants vanish . despite its algebraic nature
, lyubeznik numbers also provide some geometrical and topological information as it was already pointed out in @xcite .
for instance , in the case of isolated singularities , lyubeznik numbers can be described in terms of certain singular cohomology groups in characteristic zero ( see @xcite ) or tale cohomology groups in positive characteristic ( see @xcite , @xcite ) .
the highest lyubeznik number @xmath21 can be described using the so - called hochster - huneke graph as it has been proved in @xcite , @xcite .
however very little is known about the possible configurations of lyubeznik tables except for low dimension cases @xcite , @xcite or the just mentioned case of isolated singularities . in section 2
we will give some new constraints to the possible configurations of lyubeznik tables .
namely , the main result , theorem [ consecutiveness ] , establishes some consecutiveness of the non - vanishing superdiagonals of the lyubeznik tables using spectral sequence arguments . in section 3
we introduce a new set of invariants associated to the linear strands of a minimal free resolution of a @xmath0-graded ideal @xmath1 $ ] .
it turns out that these new invariants satisfy some analogous properties to those of lyubeznik numbers including the aforementioned consecutiveness property .
moreover , we provide a thom - sebastiani type formula for these invariants that is a refinement of the formula for betti numbers given by jacques and katzman in @xcite .
this section should be of independent interest and we hope it could be further developed in future work .
in the rest of the paper we treat the case where @xmath22 is a monomial ideal in a polynomial ring @xmath23 $ ] , and @xmath24 is the graded maximal ideal .
bass numbers are invariant with respect to completion so we consider @xmath25 where @xmath26\!]$ ] . in this sense ,
our study on the lyubeznik tables of monomial ideals is a ( very ) special case of that for local rings .
however , advanced techniques in combinatorial commutative algebra are very effective in this setting , and we can go much further than general case , so we hope that monomial ideals are good `` test cases '' for the study of lyubeznik tables .
since local cohomology modules satisfy @xmath27 we often will assume that a monomial ideal @xmath22 is _ squarefree _ , i.e. , @xmath28 .
in this case , @xmath22 coincides with the _ stanley - reisner ideal _
@xmath29 of a simplicial complex @xmath30 , more precisely , @xmath31 the stanley - reisner ring @xmath32 is one of the most fundamental tools in combinatorial commutative algebra , and it is known that @xmath32 reflects topological properties of the geometric realization @xmath33 of @xmath34 in several ways . in section 4
we get a deeper insight to the relation , given by the first author and a. vahidi in @xcite , between lyubeznik numbers of monomial ideals and the linear strands of their associated alexander dual ideals . in particular , we give a different approach to the fact proved in @xcite that if @xmath32 is sequentially cohen - macaulay , then its lyubeznik table is trivial .
we also provide a thom - sebastiani type formula for lyubeznik numbers .
one of the main results of this paper is left for section 5 .
namely , theorem [ topological ] states that lyubeznik numbers of stanley - reisner rings are not only an algebraic invariant but also a topological invariant , meaning that the lyubeznik numbers of @xmath32 depend on the homeomorphic class of the geometric realization @xmath33 of @xmath34 and the characteristic of the base field .
the proof of this result is quite technical and irrelevant to the other parts of the paper , so we decided to put it in the final section .
we also remark that this result holds in a wider setting .
more precisely , if @xmath12 is a normal simplicial semigroup ring which is gorenstein , and @xmath22 is a monomial ideal , then the corresponding result holds .
we will work in this general setting , since the proof is the same as in the polynomial ring case .
to give a full description of the possible configurations of lyubeznik tables of any local ring seems to be a very difficult task and only a few results can be found in the literature .
the aim of this section is to give some constraints to the possible configurations of lyubeznik tables , aside from the euler characteristic formula .
let @xmath4 be a regular local ring of dimension @xmath3 containing its residue field @xmath5 , and @xmath6 be any ideal with @xmath35 .
for each @xmath36 with @xmath37 , set @xmath38 for example , @xmath39 ( resp .
@xmath40 ) is the sum of the entries in the diagonal ( resp .
superdiagonal ) of the lyubeznik table @xmath41 .
clearly , @xmath42 .
we say @xmath43 is _ non - trivial _ , if @xmath44 clearly , @xmath41 is non - trivial if and only if @xmath43 is non - trivial for some @xmath45 .
it is easy to see that @xmath46 if @xmath47 and @xmath48 if @xmath49 , that is , @xmath50 is always trivial .
we also point out that @xmath51 when @xmath52 ( see @xcite ) . a key fact that we are going to use in this section
is that local cohomology modules have a natural structure over the ring of @xmath53-linear differential operators @xmath54 ( see @xcite , @xcite ) .
in fact they are @xmath54-modules of finite length ( see ( * ? ? ? * thm .
2.7.13 ) and ( * ? ?
* ex.2.2 ) for the case of characteristic zero and ( * ? ? ?
5.7 ) in positive characteristic ) . in particular , lyubeznik numbers are nothing but the length as a @xmath54-module of the local cohomology modules @xmath55 , i.e. @xmath56 the @xmath54-module length , that will be denoted simply as @xmath57 , is an additive function , i.e given a short exact sequence of holonomic @xmath54-modules @xmath58 we have @xmath59 the main result of this section is the following : [ consecutiveness ] let @xmath4 be a regular local ring of dimension @xmath3 containing its residue field @xmath5 , and @xmath6 be any ideal with @xmath35 .
then : * if @xmath43 is non - trivial for some @xmath45 with @xmath60 , then either @xmath61 or @xmath62 is non - trivial . *
if @xmath39 is non - trivial , then so is @xmath40 .
consider grothendieck s spectral sequence @xmath63 this is a spectral sequence of @xmath54-modules where @xmath64 .
notice also that the local cohomology modules @xmath65 vanish for all @xmath66 and in this case @xmath67 .
we will prove the assertion by contradiction .
so assume that @xmath43 is non - trivial for some @xmath60 , but both @xmath61 and @xmath62 are trivial ( the case @xmath68 can be proved by a similar argument ) .
we have some @xmath69 with @xmath70 such that @xmath71 ( equivalently , @xmath72 ) .
consider the maps on @xmath73-terms @xmath74 we will show first that @xmath75 .
since @xmath76 and @xmath77 concern @xmath62 and @xmath61 respectively , we have @xmath78 except the case when @xmath79 where we are assuming @xmath80 .
therefore @xmath75 is satisfied trivially except for @xmath81 . in this exceptional case
, we have @xmath82 .
hence , in the rest of this _ paragraph _ , we assume that @xmath39 is trivial but the map @xmath81 is non - zero .
then @xmath83 due to the fact that @xmath84 is a simple @xmath54-module .
it follows that @xmath85 .
it implies that @xmath86 on the other hand , since @xmath39 is trivial now , we have @xmath87 for all @xmath88 .
it contradicts the fact that , by the general theory of spectral sequences , there exists a filtration @xmath89 where @xmath90 .
anyway , we have shown that @xmath91 in all cases , and this implies that @xmath92 .
now we consider the maps on @xmath93-terms @xmath94 since @xmath95 and @xmath96 concern @xmath62 and @xmath61 respectively , we have @xmath97 by the same argument as above .
hence we have @xmath98 . repeating this argument , we have @xmath99 so we get a contradiction with the fact that @xmath100 ( recall that @xmath101 ) .
let @xmath102\!]$ ] be the ideal generated by the @xmath103 minors of a generic @xmath104 matrix , where @xmath5 is a field of characteristic zero .
its lyubeznik table is @xmath105 using similar spectral sequence arguments to those considered in theorem [ consecutiveness ] we can reprove and refine the result given by kawasaki @xcite and walther @xcite on the possible lyubeznik tables for rings up to dimension two .
namely , their result is : * if @xmath106 , then @xmath107 and the other lyubeznik numbers are 0 .
we also have : * if @xmath108 , then @xmath109 and @xmath110 for @xmath111 we can refine the last inequality , that is , @xmath112
throughout this section we will consider @xmath0-graded ideals @xmath22 in the polynomial ring @xmath23 $ ] , in particular @xmath22 is not necessarily a monomial ideal . for simplicity , we will assume that @xmath113 .
the minimal @xmath0-graded free resolution of @xmath22 is an exact sequence of free @xmath0-graded modules : @xmath114 & l_{n } \ar[r]^{d_{n } } & \cdots \ar[r ] & l_1 \ar[r]^{d_1 } & l_{0 } \ar[r ] & i \ar[r ] & 0},\ ] ] where the @xmath11-th term is of the form @xmath115 and the matrices of the morphisms @xmath116 do not contain invertible elements .
the _ betti numbers _ of @xmath22 are the invariants @xmath117 .
notice that @xmath118 as underlying @xmath12-modules where , for each @xmath11 , we set @xmath119 . hence , implies that @xmath120 given @xmath121 , we also consider the _ @xmath122-linear strand _ of @xmath123 : @xmath124 & l_{n}^{<r > } \ar[r]^{d_{n}^{<r > } } & \cdots \ar[r ] & l_1^{<r > } \ar[r]^{d_1^{<r > } } & l_{0}^{<r > } \ar[r ] & 0},\ ] ] where @xmath125 and the differential @xmath126 is the corresponding component of @xmath127 .
[ l^r and l^r-1 ] sometimes we will also consider the minimal @xmath0-graded free resolution @xmath128 of the quotient ring @xmath129 .
its truncation at the first term @xmath130 gives a minimal free resolution @xmath131 of @xmath22 .
for @xmath132 , @xmath133 is isomorphic to the @xmath134-linear strand @xmath135 up to translation .
however , this is not true for @xmath136 , since @xmath137 starts from the @xmath138-th term @xmath12 , which is irrelevant to @xmath139 .
to the minimal @xmath0-graded free resolution of @xmath22 we may associate a set of invariants that measure the acyclicity of the linear strands as follows : let @xmath140 denote the field of fractions @xmath141 of @xmath12 , and set @xmath142.\ ] ] since the complex @xmath143 is of the form @xmath144 & { \bbb k}^{\beta_{n , n+r}(i ) } \ar[r]^{\qquad \partial_{n}^{<r > } } & \cdots \ar[r ] & { \bbb k}^{\beta_{1,1+r}(i ) } \ar[r]^{\quad \partial_1^{<r > } } & { \bbb k}^{\beta_{0,r}(i ) } \ar[r ] & 0},\ ] ] we have @xmath145 for all @xmath146 ( if @xmath147 then @xmath148 ) , and @xmath149 for each @xmath122 . if we mimic the construction of the _ betti table _ , we may also consider the _
@xmath150-table _ of @xmath22 [ cols="^,>,>,>,>,>",options="header " , ] next we consider some basic properties of @xmath150-numbers .
it turns out that they satisfy analogous properties to those of lyubeznik numbers . for instance , these invariants satisfy the following _ euler characteristic _ formula .
[ eul char gamma ] for a @xmath0-graded ideal @xmath22 , we have @xmath151 the assertion follows from the computation below .
@xmath152 we can also single out a particular non - vanishing @xmath150-number . for each @xmath153 ,
let @xmath154 denote the ideal generated by the homogeneous component @xmath155 .
then we have : [ gamma_o , d ] if @xmath22 is a @xmath0-graded ideal with @xmath156 , then we have @xmath157 .
it is easy to see that there is a surjection @xmath158 . since @xmath159
, we have @xmath160 .
this fact allows us to consider the following notion : let @xmath22 be a @xmath0-graded ideal and set @xmath156 .
we say that @xmath22 has _
trivial @xmath150-table _ , if @xmath161 and the rest of these invariants vanish .
it might be an interesting problem to find necessary and/or sufficient conditions for a @xmath0-graded ideal to have a trivial @xmath150-table . in this direction
we have the following relation to the notion of _ componentwise linear _ ideals .
we say a @xmath0-graded ideal @xmath22 is _ componentwise linear _ if @xmath162 has a linear resolution for all @xmath121 , i.e. , @xmath163 unless @xmath164 .
r " omer ( @xcite ) and the second author ( ( * ? ? ? * theorem 4.1 ) ) independently showed that @xmath22 is componentwise linear if and only if @xmath165 for all @xmath122 and all @xmath166 .
[ comp linear = > trivial gamma ] a componentwise linear ideal @xmath22 has a trivial @xmath150-table .
since @xmath22 is componentwise linear , we have @xmath167 for all @xmath122 and all @xmath166 , and hence @xmath168 for all @xmath45 and all @xmath166 .
now the assertion follows from lemmas [ eul char gamma ] and [ gamma_o , d ] .
the converse of the above proposition is not true . for example , in corollary [ i_1 ne 0 ] below , we will show that if @xmath169 then it has trivial @xmath150-table . however , there is no relation between being componentwise linear and @xmath169 . for a @xmath0-graded ideal @xmath170 and @xmath153 , set @xmath171 if we denote @xmath172 , then @xmath173 . by lemma [ eul char gamma ]
, we have @xmath174 .
we say @xmath175 is _ non - trivial _ , if @xmath176 clearly , the @xmath150-table of @xmath22 is non - trivial if and only if @xmath175 is non - trivial for some @xmath11 . if @xmath177 , we have @xmath178 , and hence @xmath179 . in particular , @xmath180 is always trivial .
[ consecutive gamma ] let @xmath22 be a @xmath0-graded ideal of @xmath12 . then
; * if @xmath181 is non - trivial for @xmath182 , then either @xmath183 or @xmath184 is non - trivial . * if @xmath185 is non - trivial , then so is @xmath186 . in order to prove the theorem
, we will reconstruct @xmath187 using a spectral sequence .
let @xmath131 be the minimal free resolution of @xmath22 as before .
consider the @xmath188-adic filtration @xmath189 of @xmath131 with @xmath190 .
for any given @xmath12-module @xmath191 , we regard @xmath192 as an @xmath12-module via the isomorphism @xmath193 $ ] . since each @xmath194 is a free @xmath12-module , @xmath195 is isomorphic to @xmath194 ( if we identify @xmath196 with @xmath12 ) , while we have to forget the original @xmath0-grading of @xmath194 . since @xmath131 is a minimal free resolution , @xmath197 is the zero map for all @xmath198 , and hence @xmath199 .
it follows that @xmath200 is isomorphic to @xmath194 under the identification @xmath201 . collecting the maps @xmath202 for @xmath198 with @xmath203
, we have the @xmath12-morphism @xmath204 , and these morphisms make @xmath205 a chain complex of @xmath12-modules . under the isomorphism @xmath206
, @xmath207 is isomorphic to @xmath208 .
hence we have @xmath209 and @xmath210 . collecting the maps @xmath211
, we have the @xmath12-morphism @xmath212 moreover , we have the chain complex @xmath213 of @xmath12-modules whose @xmath45th homology is isomorphic to @xmath214 for all @xmath215 , @xmath216 satisfies the same property . by the construction of spectral sequences ,
if @xmath217 then the map @xmath218 is zero for all @xmath198 with @xmath203 , and hence @xmath219 is zero .
it implies that @xmath220 is isomorphic to @xmath221 for @xmath222 .
_ proof of theorem [ consecutive gamma ] .
_ we will prove the assertion by contradiction using the spectral sequence introduced above .
first , assume that @xmath181 is non - trivial for some @xmath223 , but both @xmath183 and @xmath184 are trivial ( the cases @xmath224 can be proved using similar arguments , and we will give a few remarks later )
. then we have @xmath225 and @xmath226 it follows that @xmath227 since it is the homology of @xmath228 similarly , we have @xmath229 repeating this argument , we have @xmath230 for all @xmath215 .
hence @xmath231 for some @xmath198 with @xmath203 .
however it contradicts the facts that @xmath232 and @xmath233 ( recall that @xmath234 now ) .
next we assume that @xmath186 is non - trivial , but @xmath185 and @xmath235 are trivial , that is , @xmath236 as we have seen above , we must have @xmath237 for @xmath222 .
so there is some @xmath238 such that @xmath239 is surjective .
hence @xmath240 for all @xmath241 , and @xmath242 .
however , since @xmath243 and @xmath244 , we have @xmath245 and hence @xmath246 .
this is a contradiction .
the case when @xmath185 is non - trivial can be proved in a similar way .
let @xmath247 be @xmath0-graded ideals in two disjoint sets of variables , say @xmath248 $ ] and @xmath249 $ ] .
the aim of this subsection is to describe the @xmath150-numbers of @xmath250 , where @xmath251 $ ] , in terms of those of @xmath22 and @xmath252 respectively .
when we just consider betti numbers we have the following results due to jacques - katzman @xcite .
[ jackat ] let @xmath128 and @xmath253 be minimal graded free resolutions of @xmath129 and @xmath254 respectively .
then , @xmath255 is a minimal graded free resolution of @xmath256 .
hence , betti numbers satisfy the following relation : the betti numbers of @xmath256 have the following form : @xmath257 we will extend their result to the case of @xmath150-numbers . to such purpose
it will be more convenient to consider separately the case of ideals with degree one elements .
thus , let @xmath170 be any @xmath0-graded ideal and assume for simplicity that @xmath252 is principally generated by an element of degree one , e.g. @xmath258 .
[ mapping cone ] let @xmath170 and @xmath259 be @xmath0-graded ideals . for @xmath132 , the @xmath122-linear strand @xmath260 is the mapping cone of the chain map @xmath261 it is easy to see that a minimal @xmath262-free resolution @xmath263 of @xmath256 is given by the mapping cone of the chain map @xmath264 \to l_\bullet(t / it)$ ] , where @xmath265 is a minimal @xmath262-free resolution of @xmath266 . since the operation of taking @xmath122-linear strand commutes with the operation of taking the mapping cone , we are done .
the general case is more involved .
assume now that @xmath170 and @xmath267 are @xmath0-graded ideal such that @xmath268 and @xmath269 .
[ tensor ] under the previous assumptions , let @xmath123 ( resp .
@xmath270 ) be a minimal graded @xmath12-(resp .
@xmath271-)free resolution of @xmath22 ( resp .
@xmath252 ) .
then @xmath272 \right).\ ] ] here , for a chain complex @xmath273 , @xmath274 $ ] means its translation , more precisely , it is the chain complex whose component of cohomological degree @xmath45 is @xmath275 .
consider the minimal @xmath0-graded free resolutions of @xmath129 and @xmath254 respectively @xmath276 & l_{m } \ar[r]^{d_{m } } & \cdots \ar[r ] & l_1 \ar[r]^{d_1 } & l_{0 } \ar[r ] & r / i \ar[r ] & 0},$ ] @xmath277 & { l'}_{n } \ar[r]^{d'_n } & \cdots \ar[r ] & { l'}_1 \ar[r]^{d'_1 } & { l'}_{0 } \ar[r ] & s / j \ar[r ] & 0 } , $ ] where @xmath278 and @xmath279 . according to proposition [ jackat ] , the minimal @xmath0-graded free resolution @xmath280 has the form simply as @xmath281 : @xmath282 & { \begin{array}{c } l_2 \otimes l'_0 \\ { { \oplus}}\\ l_1\otimes l'_1 \\ { { \oplus}}\\ l_0\otimes l'_2 \end{array } } \ar[r]^{\partial_2 } & { \begin{array}{c } l_1\otimes l'_0 \\ { { \oplus}}\\ l_0\otimes l'_1 \end{array } } \ar[r]^{\partial_1 } & l_0\otimes l'_0\ar[r ] & t / it+jt\ar[r ] & 0 } \ ] ] where , for any given @xmath283 , we have @xmath284 to describe the @xmath122-linear strand @xmath285 of the ideal @xmath250 we must consider the truncation at the first term of the above resolution and take a close look at the free modules and the components of the corresponding differentials . recall that @xmath286 corresponds to @xmath287 for all @xmath132 .
it is easy to see that both @xmath288 and @xmath289 are subcomplexes of @xmath285 .
moreover , @xmath290 and @xmath291 are direct summands of @xmath285 . in fact , since @xmath292 , the linear parts of the maps @xmath293 and @xmath294 vanish . in order to obtain the remaining components of @xmath285
we must consider the @xmath122-linear strand of @xmath282 & { \begin{array}{c } l_3 \otimes l'_1 \\ { { \oplus}}\\ l_2 \otimes l'_2 \\ { { \oplus}}\\ l_1\otimes l'_3 \end{array } } \ar[r]^{\partial_4 } & { \begin{array}{c } l_2 \otimes l'_1 \\ { { \oplus}}\\ l_1\otimes l'_2 \end{array } } \ar[r]^{\partial_3 } & l_1\otimes l'_1 \ar[r]^(.6){\partial_2 } & 0\ar[r]^{\partial_1 } & 0 } .\ ] ] this complex starts at the second term ( i.e. , the term of homological degree 1 ) , and the first term of the @xmath122-linear strand is @xmath295 if we take a close look at the free summands of these components and its differentials we obtain the following description @xmath296.\ ] ] so we are done .
the main result of this subsection is the following : [ join ] the @xmath150-numbers of @xmath297 have the following form : * if @xmath298 or @xmath299 then @xmath250 has trivial @xmath150-table . *
if @xmath300 and @xmath301 then we have : @xmath302 \i ) if @xmath299 we may assume , after a suitable coordinate change , that @xmath259 . for @xmath132 , the @xmath122-linear strand @xmath303 is given by the mapping cone of the chain map @xmath304 by lemma [ mapping cone ] .
hence @xmath305 is given by the mapping cone of the chain map @xmath306 clearly , this is the identity map , and hence its mapping cone is exact .
it means that @xmath307 for all @xmath132 and all @xmath11 . on the other hand
, @xmath308 is a complete intersection ideal generated by degree 1 elements , and hence we have @xmath309 . summing up
, we see that @xmath250 has trivial @xmath150-table .
\ii ) follows immediately from lemma [ tensor ] .
the following is just a rephrasing of part i ) of the previous result .
[ i_1 ne 0 ] let @xmath170 be a @xmath0-graded ideal with @xmath169 , then @xmath22 has trivial @xmath150-table .
in @xcite , the second author showed that , via alexander duality , the study of local cohomology modules with supports in monomial ideals can be `` translated '' into the study of the minimal free resolutions of squarefree monomial ideals .
this fact was later refined by a. vahidi and the first author in @xcite in order to study lyubeznik numbers of squarefree monomial ideals in terms of the linear strands of its alexander dual ideals . the aim of this section is to go further in this direction . in the sequel we will only consider monomial ideals in the polynomial ring @xmath23 $ ] . for a vector @xmath310 , set @xmath311 . for each @xmath312 ,
let @xmath313 be the @xmath11th standard vector .
the following notion was introduced by the second author , and serves a powerful tool for combinatorial commutative algebra .
[ sqf module for poly ring ] we say a finitely generated @xmath314-graded @xmath12-module @xmath315 is _ squarefree _ , if the multiplication maps @xmath316 is bijective for all @xmath317 and all @xmath318 .
the theory of squarefree modules is found in @xcite .
here we list some basic properties . * for a monomial ideal @xmath22 , it is a squarefree @xmath12-module if and only if @xmath319 ( equivalently , the stanley - reisner ideal @xmath29 for some @xmath34 ) . the free modules @xmath12 itself and the @xmath320-graded canonical module @xmath321 are squarefree . here
the stanley - reisner ring @xmath32 is also squarefree .
* let @xmath191 be a squarefree @xmath12-module , and @xmath323 its @xmath320-graded minimal free resolution .
then the free module @xmath324 and the syzygy module @xmath325 are squarefree for each @xmath11 .
moreover , @xmath326 is squarefree for all @xmath11 .
* let @xmath327 be the category of @xmath320-graded finitely generated @xmath12-modules , and @xmath328 its full subcategory consisting of squarefree modules .
then @xmath328 is an abelian subcategory of @xmath327 .
we have an exact contravariant functor @xmath329 from @xmath328 to itself .
the construction of @xmath329 is found in ( for example ) @xcite .
here we just remark that @xmath330 , where @xmath331 is the alexander dual simplicial complex of @xmath34 . in this framework
we have the following description of lyubeznik numbers .
[ yan01 , 3.10 ] let @xmath23 $ ] be a polynomial ring , and @xmath29 a squarefree monomial ideal . then we have @xmath332_0 < \infty.\ ] ] for a squarefree @xmath12-module @xmath191 , the second author defined the cochain complex @xmath333 of squarefree @xmath12-modules satisfying @xmath334 for all @xmath11 ( see @xcite ) . by (
* theorem 4.1 ) or ( * ? ? ? * theorem 3.8 ) , we have the isomorphism @xmath335\ ] ] of cochain complexes of @xmath320-graded @xmath12-modules and @xmath329 as individual operations , but treats the composition @xmath336 .
in fact , @xmath336 corresponds to the operation @xmath337 of @xcite up to translation . ] . here , for a cochain complex @xmath338 , @xmath339 $ ] means the @xmath340th translation of @xmath338 , more precisely , it is the cochain complex whose component of cohomological degree @xmath45 is @xmath341 , and we also regard a chain complex @xmath273 as the cochain complex whose component of cohomological degree @xmath45 is @xmath342 .
the following is a variant of a result given by the first author and a. vahidi .
[ lambda vs gamma ] let @xmath343 $ ] be a squarefree monomial ideal .
then we have @xmath344 by and the construction of @xmath329 , we have an isomorphism @xmath345_0)^ * \cong ( \mathbb{l}_\bullet^{<n - i>}(i_{\delta^\vee}))_{\mathbf 1}[-i]\ ] ] of cochain complexes of @xmath5-vector spaces . here
@xmath346 means the @xmath5-dual .
we also remark that , for a squarefree module @xmath191 , we have @xmath347 thus we have the following computation .
@xmath348_0\\ & = & \dim_{\bbbk}[h^{-p}({\mathbf{d}}({\mbox{\rm{ext } } } _ r^{n - i}(r / i , \omega_r ) ) ] _ 0\\ & = & \dim_{\bbbk}[h_{i - p}(\mathbb{l}_\bullet^{<n - i>}(i_{\delta^\vee}))]_{\mathbf 1}\\ & = & \dim_{\bbb k}h_{i - p}(\mathbb{l}_\bullet^{<n - i>}(i_{\delta^\vee } ) ) \otimes_r { \bbb k}\\ & = & \nu_{i - p , n - p}(i_{\delta^\vee}).\end{aligned}\ ] ] as mentioned in introduction , for a local ring @xmath2 containing a field , we have @xmath349 in the monomial ideal case , this equation is an immediate consequence of lemma [ eul char gamma ] and theorem [ lambda vs gamma ] . as a special case of theorem [ consecutiveness ] ,
the lyubeznik tables of monomial ideals in @xmath23 $ ] satisfy the consecutiveness property of nontrivial superdiagonals .
however , it also follows from the consecutiveness property of nontrivial columns of the @xmath150-tables ( theorem [ consecutive gamma ] ) via theorem [ lambda vs gamma ] . in this sense ,
both `` consecutiveness theorems '' are related .
let @xmath191 be a finitely generated graded module over the polynomial ring @xmath23 $ ] .
we say @xmath191 is _ sequentially cohen - macaulay _ if @xmath350 is either a cohen - macaulay module of dimension @xmath11 or the 0 module for all @xmath11 .
the original definition is given by the existence of a certain filtration ( see ( * ? ? ?
* iii , definition 2.9 ) ) , however it is equivalent to the above one by ( * ? ? ?
* iii , theorem 2.11 ) .
the sequentially cohen - macaulay property of a finitely generated module over a regular local ring is defined / characterized in the same way . in @xcite
, the first author showed that the sequentially cohen - macaulay property implies the triviality of lyubeznik tables in positive characteristic as well as in the case of squarefree monomial ideals . using proposition [ comp linear = > trivial gamma ] we can give a new proof / interpretation of this result for the case of monomial ideals .
[ seq cm ] let @xmath22 be a monomial ideal of the polynomial ring @xmath23 $ ] such that @xmath129 is sequentially cohen - macaulay .
then the lyubeznik table of @xmath129 is trivial . by (
? * theorem 2.6 ) , @xmath351 is sequentially cohen - macaulay again .
hence we may assume that @xmath22 is the stanley - reisner ideal @xmath29 of a simplicial complex @xmath34 .
herzog and hibi @xcite showed that @xmath32 is sequentially cohen - macaulay if and only if @xmath352 is componentwise linear .
now the assertion immediately follows from proposition [ comp linear = > trivial gamma ] and theorem [ lambda vs gamma ] .
the converse of proposition [ seq cm ] is not true , that is , even if @xmath129 has trivial lyubeznik table it need not be sequentially cohen - macaulay .
for example , if @xmath22 is the monomial ideal @xmath353 in @xmath354 $ ] , then @xmath129 has trivial lyubeznik table , but this ring is _ not _ sequentially cohen - macaulay . since all associated primes of @xmath22 have the same height , it is the same thing to say @xmath129 is not cohen - macaulay .
however , @xmath129 does not even satisfy serre s condition @xmath355 . in proposition [ dual_join ]
below , we will see that if a monomial ideal @xmath22 has height one ( i.e. , admits a height one associated prime ) , then the lyubeznik table of @xmath129 is trivial . of course
, @xmath129 need not be sequentially cohen - macaulay in this situation .
let @xmath248 $ ] and @xmath356 $ ] be squarefree monomial ideals in two disjoint sets of variables .
let @xmath357 and @xmath358 be the simplicial complexes associated to @xmath22 and @xmath252 by the stanley - reisner correspondence , i.e. @xmath359 and @xmath360 .
then , the sum @xmath361 corresponds to the simplicial _ join _ of both complexes .
let @xmath362 ( resp .
@xmath363 ) be the alexander dual of @xmath357 ( resp .
@xmath358 ) as a simplicial complex on @xmath364 ( resp .
@xmath365 ) . set @xmath366 and @xmath367 .
then it is easy to see that @xmath368 where @xmath329 denotes the alexander duality functor of @xmath369 .
[ dual_join ] the lyubeznik numbers of @xmath370 have the following form : * if either the height of @xmath22 or the height of @xmath252 is @xmath371 , then @xmath372 has trivial lyubeznik table . *
if both the height of @xmath22 and the height of @xmath252 are @xmath373 , then we have : @xmath374 the assertion easily follows from proposition [ join ] and theorem [ lambda vs gamma ] , but for completeness , we will give a few remarks .
\(i ) recall that , for a simplicial complex @xmath34 , the height of @xmath29 is 1 if and only if @xmath375_1 \ne 0 $ ] .
\(ii ) the last equality follows from the fact that @xmath376 which can be seen from theorem [ lambda vs gamma ] and the construction of linear strands .
it is well - know that local cohomology modules as well as free resolutions depend on the characteristic of the base field , the most recurrent example being the stanley - reisner ideal associated to a minimal triangulation of @xmath377 .
thus , lyubeznik numbers also depend on the characteristic .
namely , consider the ideal in @xmath378 $ ] : @xmath379 it has lyubeznik table in characteristic zero and two respectively : @xmath380 one can slightly modify this example and use proposition [ dual_join ] to obtain some interesting behavior of lyubeznik numbers : @xmath381 the ideal @xmath382 in @xmath383 $ ] has trivial lyubeznik table in any characteristic , so we obtain an example where the local cohomology modules depend on the characteristic but lyubeznik numbers do not .
@xmath381 the ideal @xmath384 in @xmath385 $ ] satisfies @xmath386 and both lyubeznik numbers are different from zero .
an interesting case where the lyubeznik table is completely described is that of ideals @xmath22 in a regular local ring @xmath387 , of dimension @xmath3 , such that the support of the local cohomology modules @xmath9 is contained in @xmath388 for all @xmath389 .
these rings @xmath129 are said to _ behave cohomologically like an isolated singularity _ or that they are _ cohomologically complete intersections in the punctured spectrum_. the following result was obtained by garca lpez - sabbah @xcite over the field of complex numbers and generalized to any characteristic by blickle - bondu @xcite ( see also @xcite ) .
namely , the lyubeznik numbers of @xmath129 satisfy @xmath390 if @xmath391 or @xmath392 .
moreover , @xmath393 and @xmath394 ( the latter follows from and the euler characteristic formula for the lyubeznik table ) . therefore , its lyubeznik table . ] is @xmath395 for a squarefree monomial ideal @xmath396 $ ] , the condition of being cohomologically like an isolated singularity is equivalent to @xmath397 having finite length for all @xmath88 , i.e. @xmath129 is buchsbaum .
we can easily prove the above result using theorem [ yan01 , 3.10 ] .
for example , the equation follows from ( * ? ? ? *
theorem 4.9 ) , which states that @xmath398 where @xmath346 means the graded @xmath5-dual ( recall that @xmath399 for a finitely generated graded @xmath12-module @xmath191 ) .
while the other sections treat the case where @xmath12 is a regular local ring or a polynomial ring , in this section we will work in a slightly different situation . here
the ring @xmath12 means a normal semigroup ring .
when @xmath12 is simplicial and gorenstein , the second author proved in @xcite that the local cohomology modules @xmath400 have finite bass numbers for radical monomial ideals @xmath401 .
in fact , without these conditions , bass numbers are out of control and can be infinite ( see @xcite for details ) .
let @xmath402 be an affine semigroup ( i.e. , @xmath403 is a finitely generated additive submonoid of @xmath320 ) , and @xmath404 \subset { \bbbk}[x_1^{\pm 1 } , \ldots , x_n^{\pm 1}]$ ] the semigroup ring of @xmath403 over @xmath5 . here
@xmath405 denotes the monomial @xmath406 for @xmath407 . regarding @xmath403 as a subset of @xmath408 ,
let @xmath409 be the polyhedral cone spanned by @xmath403 .
we always assume that @xmath410 , @xmath411 and @xmath412 .
thus @xmath12 is a normal cohen - macaulay integral domain of dimension @xmath3 with the graded maximal ideal @xmath413 .
we say @xmath12 is _ simplicial _ , if the cone @xmath414 is spanned by @xmath3 vectors in @xmath415 .
the polynomial ring @xmath416 $ ] is a typical example of a simplicial semigroup ring @xmath417 $ ] for @xmath418 .
clearly , @xmath419 is a @xmath320-graded ring .
we say that a @xmath320-graded ideal of @xmath12 is a _ monomial ideal _ and we will denote @xmath327 the category of finitely generated @xmath320-graded @xmath12-modules and degree preserving @xmath12-homomorphisms .
let @xmath420 be the set of non - empty faces of the polyhedral cone @xmath414 .
note that @xmath421 and @xmath414 itself belong to @xmath420 .
regarding @xmath420 as a partially ordered set by inclusion , @xmath12 is simplicial if and only if @xmath420 is isomorphic to the power set @xmath422 . for @xmath423
, @xmath424 is a prime ideal of @xmath12 .
conversely , any monomial prime ideal is of the form @xmath425 for some @xmath423 .
note that @xmath426 $ ] for @xmath423 .
for a point @xmath427 , we always have a unique face @xmath423 whose relative interior contains @xmath428 .
here we denote @xmath429 . for a monomial ideal @xmath22
, @xmath129 is a squarefree @xmath12-module if and only if @xmath22 is a radical ideal ( i.e. , @xmath436 ) .
we say that @xmath437 is an _ order ideal _ if @xmath438 implies @xmath439 .
if @xmath34 is an order ideal , then @xmath440 is a radical monomial ideal .
conversely , any radical monomial ideal is of the form @xmath29 for some @xmath34 .
clearly , @xmath441_{\mathbf{c}}\cong \begin{cases } { \bbbk } & \text{if $ { \mathbf{c}}\in c$ and $ s({\mathbf{c } } ) \in \delta$,}\\ 0 & \text{otherwise . }
\end{cases}\ ] ] if @xmath12 is simplicial , an order ideal @xmath34 is essentially a simplicial complex on the vertices @xmath442 . if @xmath12 is the polynomial ring @xmath416 $ ] , then @xmath32 is nothing but the stanley - reisner ring of the simplicial complex @xmath34 . for each @xmath423 , take some @xmath443 ( i.e. , @xmath444 ) . for a squarefree @xmath12-module @xmath191 and @xmath445 with @xmath446 , (
* theorem 3.3 ) gives a @xmath5-linear map @xmath447 these maps satisfy @xmath448 and @xmath449 for all @xmath450 .
we have @xmath451 for @xmath452 with @xmath453 . under these isomorphisms ,
the maps @xmath454 do not depend on the particular choice of @xmath455 s .
let @xmath328 be the full subcategory of @xmath327 consisting of squarefree modules . as shown in @xcite
, @xmath328 is an abelian category with enough injectives .
for an indecomposable squarefree module @xmath191 , it is injective in @xmath328 if and only if @xmath456 for some @xmath423 .
let @xmath457 be the @xmath320-graded canonical module of @xmath12 .
it is well - known that @xmath457 is isomorphic to the radical monomial ideal @xmath458 . as shown in (
* proposition 3.7 ) we have @xmath459 for @xmath460 .
notice that in this setting we have that whenever we have a multigraded isomorphism @xmath417/i_\delta \cong { \bbbk}[c']/i_{\delta'}$ ] between quotients of gorenstein normal simplicial semigroup rings by radical monomial ideals , then the corresponding lyubeznik numbers coincide .
this multigraded framework slightly differs from the original situation for regular local rings stated in @xcite .
however , as stated in ( * ? ? ?
* remark 5.14 ) , if @xmath463 as simplicial complexes , then @xmath32 and @xmath464 have the same lyubeznik numbers . in this sense , to study the lyubeznik numbers of a quotient @xmath32 of a gorenstein normal simplicial semigroup ring @xmath12 by a radical monomial ideal @xmath29
, we may assume that @xmath12 is a polynomial ring and @xmath32 is a stanley - reisner ring . in theorem
[ topological ] , we will prove a stronger result .
\(1 ) as in the polynomial ring case , we have the euler characteristic equation , @xmath465 since the original proof in @xcite also works in the present situation .
moreover , the statements corresponding to theorem [ consecutiveness ] ( the consecutiveness of nontrivial lines ) and theorem [ dual_join ] ( thom - sebastiani type formulae ) still hold . in fact , we may assume that @xmath22 is a radical ideal , and hence @xmath466 for some simplicial complex @xmath34 and then reduce to the case when @xmath12 is a polynomial ring as in ( * ? ? ?
* remark 5.14 ( b ) ) .
if we assume that @xmath28 , proposition [ seq cm ] also holds in the present situation . however , we can not drop this assumption , since we have no idea whether the condition of being sequentially cohen - macaulay is preserved after taking radicals .
what is known is that if @xmath129 is cohen - macaulay then so is @xmath351 ( see ( * ? ? ?
* theorem 6.1 ) ) .
hence if @xmath129 is cohen - macaulay then the lyubeznik table of @xmath129 is trivial .
\(2 ) set @xmath35 . for a radical monomial ideal @xmath29
, the highest lyubeznik number @xmath467_0\ ] ] has a simple topological ( or combinatorial ) meaning .
in fact , to study this number we may assume that @xmath12 is a polynomial ring , and we can use a combinatorial description of @xmath468 given in ( * ? ? ?
* p.96 ) . roughly speaking
, @xmath469 is the number of `` connected in codimension one components '' of @xmath33 .
( this result holds in a much wider context , see @xcite . ) in particular , if @xmath32 satisfies serre s condition @xmath355 then @xmath470 , while the converse is not true .
recall that if @xmath461 $ ] is simplicial then an order ideal @xmath34 of @xmath420 is essentially a simplicial complex , and hence it has the geometric realization @xmath33 .
it is natural to ask how lyubeznik numbers of @xmath471 depend on @xmath33 .
the next theorem shows that lyubeznik numbers are not only an algebraic invariant but also a topological invariant .
[ topological ] let @xmath461 $ ] be a simplicial normal semigroup ring which is gorenstein and @xmath472 a radical monomial ideal . then
, @xmath473 depends only on the homeomorphism class of @xmath33 and @xmath474 .
bearing in mind theorem [ yan01 - 2 , 5.12 ] , it suffices to show that @xmath475_0\ ] ] depends only on the topology of @xmath33 and @xmath474 . for this statement ,
the assumption that @xmath12 is simplicial and gorenstein is irrelevant ( if @xmath12 is not simplicial , then @xmath34 is essentially a cw complex ) . in (
* theorem 2.10 ) , r. okazaki and the second author showed that the invariant which is ( essentially ) equal to @xmath476 depends only on @xmath33 and @xmath474 for each @xmath11 .
our proof here uses similar arguments to the aforementioned result . to do so
, we have to recall some previous work of the second author in @xcite .
we have a hyperplane @xmath477 such that @xmath478 is an @xmath479-polytope ( an @xmath479-simplex , if @xmath12 is simplicial ) . for @xmath423 ,
set @xmath480 to be the relative interior of the face @xmath481 of @xmath482 .
we can regard an order ideal @xmath437 as a cw complex ( a simplicial complex , if @xmath12 is simplicial ) whose geometric realization is @xmath483 . for @xmath423 , @xmath484 is an open set of @xmath482 .
note that @xmath485 is an open covering of @xmath482 . in @xcite , from @xmath460
, we constructed a sheaf @xmath486 on @xmath482 .
more precisely , the assignment @xmath487 for each @xmath488 and the map @xmath489 for @xmath490 with @xmath491 ( equivalently , @xmath492 ) defines a sheaf .
note that @xmath493 is `` irrelevant '' to @xmath486 .
for example , @xmath494 , where @xmath495 is the constant sheaf on @xmath33 with coefficients in @xmath5 , and @xmath45 is the embedding map @xmath496 .
similarly , we have that @xmath497 , where @xmath498 is the constant sheaf on the relative interior @xmath499 of @xmath482 , and @xmath500 is the embedding map @xmath501 .
note that @xmath502 is the orientation sheaf of @xmath482 with coefficients in @xmath5 .
let @xmath437 be an order ideal , and set @xmath503 . for @xmath460
, @xmath191 is an @xmath32-modules ( i.e. , @xmath504 ) if and only if @xmath505 . in this case
, we have @xmath506 for all @xmath11 . here
@xmath507 is the restriction of the sheaf @xmath486 to the closed set @xmath508 .
combining this fact with ( * ? ? ?
* theorem 3.3 ) , we have the following . [ hoch ] with the above situation , we have @xmath509_0 \quad \text{for all $ i \geq 1$},\ ] ] and an exact sequence @xmath510_0 \longrightarrow m_0 \longrightarrow h^0 ( x ; m^+|_x ) \longrightarrow [ h_{{\mathfrak m}}^1(m)]_0 \longrightarrow 0.\ ] ] in particular , @xmath511_0 \cong \widetilde{h}^i(x ; { \bbbk})$ ] for all @xmath512 , where @xmath513 denotes the @xmath11th reduced cohomology of @xmath514 with coefficients in @xmath5 .
\(2 ) via the isomorphisms in ( 1 ) , for @xmath520 , the natural map @xmath522 coincides with the middle map @xmath523_0 \longrightarrow \gamma(x ; { \mbox{\rm{ext } } } ^{n - i}_r(m , \omega_r)^+|_x)\ ] ] of the sequence for @xmath524 . _ the proof of theorem [ topological ] . _
we show that the dimension of @xmath525_0 $ ] @xmath526_0)$ ] depends only on @xmath514 and @xmath474 . if @xmath527 , then we have @xmath528_0 \cong h^{p-1}(x ; { { \mathcal ext}}^{1-i}({\underline{{\bbbk}}}_x , { \mathcal{d}^\bullet}_x))\ ] ] by theorems [ hoch ] and [ verdier ] ( 1 )
. the right side of the equation clearly depends only on @xmath514 and @xmath474 for each @xmath69 .
next we consider the case @xmath529 . by theorem
[ hoch ] , @xmath530 and @xmath531 are the kernel and the cokernel of the map @xmath532_0 \longrightarrow \gamma(x ; { \mbox{\rm{ext } } } ^{n - i}_r(r / i_\delta , \omega_r)^+|_x)\ ] ] respectively . if @xmath520 , the above map is equivalent to the natural map @xmath533 by theorem [ verdier ] ( 2 ) , and the dimensions of its kernel and cokernel are invariants of @xmath514 .
it remains to show the case ( @xmath529 and ) @xmath534 . clearly , @xmath535 , if and only if @xmath536 , if and only if @xmath537 , if and only if @xmath538
. hence @xmath539 , if and only if @xmath540 , if and only if @xmath538 .
on the other hand , it is easy to check out that @xmath541 is always `` trivial '' , that is , @xmath542 ( the same is true for the local ring case using the spectral sequence argument as in the proof of theorem [ consecutiveness ] or adapting the techniques used in @xcite ) .
hence the remaining case is only @xmath543 , and it will be treated in lemma [ remaining case ] below .
we may assume that @xmath547 .
if @xmath548 , then @xmath32 is cohen - macaulay , and the assertion is clear .
so we may assume that @xmath549 .
first , we consider the case when @xmath29 does not have 1-dimensional associated primes , equivalently , @xmath33 does not admit isolated points ( i.e. , @xmath550 ) .
then we have @xmath551 since @xmath552 is a squarefree module , we have @xmath553_0.\ ] ] we also have @xmath554_0 \cong [ h_{{\mathfrak m}}^1(r / i_\delta)]_0 \cong \widetilde{h}^0(x ; { \bbbk } ) \cong { \bbbk}^{c-1},\ ] ] where the second isomorphism follows from the last statement of theorem [ hoch ] . hence @xmath555_0
= \dim_{\bbbk}[{\mbox{\rm{ext } } } _ r^n({\bbbk}^{c-1 } , \omega_r)]_0 = c-1,\ ] ] and we are done .
so we now consider the case where @xmath29 admits 1-dimensional associated primes
. set @xmath556 .
then there is a monomial ideal @xmath252 of @xmath12 with @xmath557 and @xmath558 .
note that @xmath559 .
the short exact sequence @xmath560 yields the exact sequence @xmath561 since lyubeznik numbers of type @xmath562 are always trivial , we have @xmath563_0 = [ { \mbox{\rm{ext } } } ^{n-1}_r({\mbox{\rm{ext } } } ^{n-1}_r(r / i , \omega_r ) , \omega_r)]_0 = 0\ ] ] and @xmath564_0 = { \bbbk}$ ] .
it is also clear that @xmath565 thus applying @xmath566 to , we obtain @xmath567_0 \ ( \cong { \bbbk } ) \longrightarrow [ { \mbox{\rm{ext } } } ^n_r({\bbbk } , \omega_r)]_0 \ ( \cong { \bbbk } ) \longrightarrow\\ & & [ { \mbox{\rm{ext } } } ^n_r({\mbox{\rm{ext } } } ^{n-1}_r(r / i_\delta , \omega_r ) , \omega_r)]_0 \longrightarrow [ { \mbox{\rm{ext } } } ^n_r({\mbox{\rm{ext } } } ^{n-1}_r(r / i , \omega_r ) , \omega_r)]_0 \longrightarrow 0 .\end{aligned}\ ] ] since @xmath568_0 \cong { \bbbk}^{c-1}$ ] as we have shown above , it follows that @xmath569_0 \cong { \bbbk}^{c-1},\ ] ] and we are done .
let @xmath570 $ ] be a polynomial ring , and consider the monomial ideal @xmath571 then @xmath33 consists of 1 isolated point and 3 segments , see fig 1 below .
so @xmath572 , which is @xmath573 , consists of 3 segments .
we have @xmath574 . | in this work we introduce a new set of invariants associated to the linear strands of a minimal free resolution of a @xmath0-graded ideal @xmath1 $ ] .
we also prove that these invariants satisfy some properties analogous to those of lyubeznik numbers of local rings .
in particular , they satisfy a consecutiveness property that we prove first for lyubeznik numbers . for the case of squarefree
monomial ideals we get more insight on the relation between lyubeznik numbers and the linear strands of their associated alexander dual ideals . finally , we prove that lyubeznik numbers of stanley - reisner rings are not only an algebraic invariant but also a topological invariant , meaning that they depend on the homeomorphic class of the geometric realization of the associated simplicial complex and the characteristic of the base field . |
the use of patches in image processing can be seen as an instance of the divide and conquer " principle : since it is admittedly very difficult to formulate a global prior / model for images , patch - based approaches use priors / models for patches ( rather than whole images ) , the combination of which yields the desired image prior / model . to keep the discussion and formulation at their essential and focus on the image modelling aspects
, we will concentrate on image denoising , arguably the quintessential image processing problem .
nevertheless , much of what will be presented below can be easily extended ( at least in principle ) to more general inverse problems .
there are basically two approaches to patch - based image denoising . in earlier methods
@xcite , @xcite , @xcite , patches are extracted from the noisy image , then processed / denoised independently ( or maybe even collaboratively , as in bm3d @xcite ) , and finally returned to their original locations .
since the patches overlap ( to avoid blocking artifacts ) , there are several estimates of each pixel , which are combined by some form of averaging ( unweighted or weighted ) .
this approach is also used in the nonlocal bayesian method @xcite , and in methods based on gaussian mixtures @xcite . for a comprehensive review of these and related methods ,
see @xcite .
arguably , a conceptual flaw of these methods is that they obtain patch estimates without explicitly taking into account that these will subsequently be combined additively . as a consequence , although some of these methods achieve state - of - the - art results , they do not explicitly provide a global image prior / model .
a more recent class of approaches does build global image models that are based on a function computed from image patches , but does not treat them as independent by explicitly taking into account that they are overlapping patches of the same image ; this approach was initiated with the _ expected patch log - likelihood _ ( epll ) @xcite , and is adopted by most of the recent work @xcite , @xcite .
these methods do not have the conceptual flaw pointed out in the previous paragraph and provide a coherent global image model .
the analysis vs synthesis dichotomy in global image models / priors ( _ e.g. _ , based on wavelet frames , or total variation ) has been first formalized in @xcite , and further studied in @xcite ; more recently , it has been ported to patch - wise models @xcite .
to the best of our knowledge , this dichotomy has not been pointed out before concerning the way in which patch - level models / priors are used to build a global image models ; that is precisely the central contribution of this paper . in the synthesis vs analysis dichotomy
, the epll - type class of patch - based models can be seen as an analysis method ( as explained below in detail ) .
this paper shows that there exists the synthesis counterpart of epll ; in other words , that the synthesis vs analysis dichotomy is also present in the way the whole image and the patches are related .
we also show that the two formulations are not , in general , equivalent .
the remaining sections of this paper are organized as follows . after reviewing the classical analysis / synthesis dichotomy in section [ sec : a_vs_s ] , we shown in section [ sec : patch_a ] that the classical patch - based methods follow an analysis formulation .
section [ sec : patch_s ] then introduces a synthesis patch - based formulation , and its relationship with the analysis counterpart is established in section [ sec : rel ] . in section [ sec : admm ] , we present an admm algorithm to efficiently perform image denoising under the proposed synthesis formulation .
finally , section [ sec : conclusion ] concludes the paper by referring to future work directions .
[ sec : a_vs_s ] before addressing patch - based models , we briefly review the analysis and synthesis global formulations of image denoising @xcite , where the goal is to estimate an unknown image @xmath0 ( @xmath1 is the total number of pixels in @xmath2 , which is a vectorized version of the corresponding @xmath3 image ) from a noisy version thereof @xmath4 where @xmath5 is a sample of a white gaussian noise field of zero mean and ( known ) variance @xmath6 , that is , @xmath7 . the classical approach to estimate @xmath2 from @xmath8 is to adopt ( or learn ) a prior @xmath9 for the unknown @xmath2 and seek a maximizer of the posterior density ( a _ maximum a posteriori_map estimate ) @xmath10 where @xmath11 , up to an irrelevant constant .
the analysis and synthesis formulations build priors for @xmath2 as follows .
analysis : : : in this formulation , the prior @xmath9 takes the form @xmath12 where @xmath13 is a constant and @xmath14 a linear ( analysis ) operator . in this case , the map estimate takes the form @xmath15 where map - a stands for _ map - analysis_. synthesis : : : here , the starting point is to assume that @xmath2 is linearly synthesized / represented according to @xmath16 , where @xmath17 is a vector of coefficients , and the prior is formulated on @xmath17 rather than directly on @xmath2 ; once an estimate @xmath18 is obtained , the corresponding estimate of @xmath2 is simply @xmath19 . in summary ,
@xmath20 and @xmath21 is a negative log - prior on @xmath17 .
one of the main distinguishing features of analysis and synthesis formulations is that in the former the object of the estimation procedure is the image itself , whereas in the latter , one estimates a representation from which the image estimate is synthesized .
a central tool in most patch - based approaches is a collection of operators @xmath22 that extracts @xmath23 patches from a given image of size @xmath3 ; each @xmath24 can be seen as a binary matrix with size @xmath25 , where @xmath26 is the total number of pixels in each patch ( assumed square , of size @xmath27 ) .
the standard way of formulating a patch - based prior is by writing @xmath28 where @xmath29 is a function expressing the patch - wise prior distribution and @xmath30 is a normalizing constant .
function @xmath31 may itself be a _ probability density function _
( pdf ) , _
e.g. _ , a gmm , in which case this prior is an instance of a so - called _ product of experts _ ( poe @xcite ) .
however , @xmath31 does not need to be a pdf , as long as it takes non - negative values ; in fact , can also be seen as a _ factor graph _ model , where each factor corresponds to a patch and the all factors have the same function @xmath31 @xcite .
moreover , this prior is also equivalent to the formulation known as epll ( _ expected patch log - likleihood _
@xcite ) , although epll was not originally interpreted as a prior . given a noisy image @xmath8 , a map estimate of @xmath2 is given by @xmath32 to tackle the large - scale optimization problem in , the so - called _ half quadratic splitting _
strategy replaces it with @xmath33 where @xmath34 , which obviously becomes equivalent to as @xmath35 @xcite .
the optimization problem in is tackled by alternating between minimizing with respect to @xmath36 and @xmath2 , while slowly increasing @xmath37 .
other strategies for setting @xmath37 have also been proposed @xcite . an obvious alternative ( as recently mentioned in @xcite )
is to reformulate as a constrained problem @xmath38 and tackle it with admm ( _ alternating direction method of multipliers _ ) @xcite .
of course , convergence of admm for this problem can only be guaranteed if the negative log factors @xmath39 are convex ; this is not the case if @xmath31 is a gmm , but it is true if @xmath39 is an @xmath40 norm or the @xmath41-th power thereof , _
e.g. _ , @xmath42 , with @xmath43 . examining ( with @xmath44 ) or reveals that this formulation seeks a consensus among the patches , in the sense that the several replicates of each pixels that exist in different patches are forced to agree on a common value for that pixel .
in other words , the clean patches are not modelled as additively generating a clean image , and are merely used to write a joint prior @xmath9 that factorizes across overlapping patches .
the estimation criterion in clearly falls in the analysis - type category @xcite , since it considers the image itself as the object to be estimated and as the argument of the prior .
however , as a generative model for clean images , its meaning is not very clear , since it is not a trivial task to obtain samples from this distribution . to obtain a more compact notation ,
let @xmath45 be the operator ( an @xmath46 matrix ) that extracts the set of @xmath23 patches , _
i.e. _ , @xmath47 the prior @xmath9 may then be written as @xmath48 where @xmath49 denotes a density defined in @xmath50 according to @xmath51 with this notation , the map denoising problem can be written as @xmath52 which clearly reveals the analysis nature of this formulation ( see ) .
we stress that the analysis / synthesis dichotomy addressed in this paper concerns the way in which an image relates to its patches , not the way the patches themselves are modelled .
in fact , the patch - analysis formulation just reviewed is compatible with a synthesis patch model , _
e.g. _ , one that models each patch @xmath53 as linear combination of elements of some dictionary @xmath54 , with coefficients @xmath55 equipped with some prior @xmath56 ( for example , a sparsity - inducing prior , as in @xcite ) .
using the half quadratic splitting approach , the formulation becomes @xmath57 where @xmath58 .
naturally , the patch - analysis formulation is also compatible with an analysis patch model @xcite , by using a patch prior of the form @xmath59 , _
i.e. _ , @xmath60 , for some function @xmath61 and matrix @xmath62 .
to identify the analysis / synthesis dichotomy in terms of how the patches are related to the underlying image , not in terms of how the patches are modelled , we refer to the formulation reviewed in this section as _ patch - analysis _ and to its synthesis counterpart ( to be introduced in the next section ) as _ patch - synthesis_.
we now present the patch - synthesis formulation , which can be summarized as follows : the clean image is generated by additively combining a collection of patches ; the patches themselves follow some probabilistic model , but are a priori mutually independent .
consider a collection of patches @xmath63 and let image @xmath2 be synthesized from these patches by combining them additively according to @xmath64 where matrices @xmath65 are such that they average the values in the several patches that contribute to a given pixel of @xmath2 .
a simple 1d example will help clarify the structure of the @xmath66 matrices .
consider that @xmath67^t$ ] is produced by combining all the consecutive 2-element patches ( with periodic boundary conditions ) , which correspond to the subsets of components @xmath68 , @xmath69 , @xmath70 and @xmath71 .
that is , @xmath72 and @xmath73 stacking all the patches in vector @xmath74 and considering a matrix @xmath75 \in \mathbb{r}^{n\times(mn)}$ ] , the synthesis expression in can be written compactly as @xmath76 with the patches modelled as independent and identically distributed samples of some patch - wise pdf @xmath77 , their joint log - prior is @xmath78 notice that obtaining samples from is as simple as obtaining samples from @xmath77 itself . consequently
, generating image samples under this synthesis model simply corresponds to generating samples from and then multiplying them by @xmath79 .
this is in contrast with the patch - analysis prior , where , even if @xmath31 is a valid pdf , it is not trivial to obtain samples from @xmath9 .
the resulting map denoising criterion can now be written as @xmath80 as the patch - analysis model , the patch - synthesis formulation that we have just presented is compatible with both analysis and synthesis priors for the patches , and of course with any valid pdf for vectors in @xmath81 .
an analysis formulation simply amounts to choosing a patch prior of the form @xmath82 , _
i.e. _ , @xmath83 , for some function @xmath84 and matrix @xmath85 . in a synthesis formulation
, each patch @xmath86 is synthesized using some dictionary @xmath54 as @xmath87 , and the @xmath55 follow some prior @xmath56 ; stacking all the patch coefficients in vector @xmath88 , we can write @xmath89 ( where @xmath90 is a block - diagonal matrix with @xmath23 replicas of @xmath54 ) , thus the map denoising criterion becomes @xmath91 where @xmath92
leaving aside for now the choice of the patch priors , let us focus on the relationship between formulations and .
the key observation underlying the relationship between these two formulations is that ( assuming the patch structure in both formulations is the same ) @xmath93 where @xmath94 denotes the identity matrix , but in general @xmath95 in other words , @xmath79 is a left pseudo - inverse of @xmath96 . to prove , simply notice that if a collection of patches is extracted from some image and then these patches are used to synthesize an image by averaging the overlapping pixels , an identical image is obtained .
of course , the converse is not true , in general : if an image is synthesized by averaging the overlapping pixels of a collection of patches , and then patches are extracted from the synthesized image , there is no guarantee that these patches are equal to the original ones , thus proving . in the trivial and uninteresting cases where the patches are singletons , or non - overlapping
, we would have @xmath97 . as shown in @xcite ,
given the analysis formulation , an equivalent synthesis formulation is @xmath98 where the constraint @xmath99 enforces @xmath100 to be in the subspace spanned by the columns of @xmath96 .
notice that this constraint corresponds to having a collection of patches extracted from some image , _
i.e. _ , it forces the patches to agree on the value of each shared pixel . since does not enforce this constraint , it is not , in general , equivalent to the patch - analysis formulation .
in this section , we derive an instance of admm to deal with . the first step is to rewrite it as a constrained problem , @xmath101 where @xmath102 is the negative log - prior , or regularizer .
admm for this problem takes the form @xmath103 the update equation for @xmath104 is a denoising step .
due to the separability of the squared @xmath105 norm and of @xmath106 ( see ) , this can be separately solved with respect to each patch : for @xmath107 , @xmath108 which is simply the proximity operator of @xmath109 @xcite , computed at @xmath110 . in a bayesian viewpoint ,
corresponds to obtaining the map estimate of @xmath111 from observations @xmath110 , assuming additive white gaussian noise of variance @xmath112 and a negative log - prior @xmath113 .
if the @xmath113 are convex , this map estimate is unique , due to the strict convexity of the quadratic term in .
computing @xmath114 corresponds to solving an unconstrained quadratic problem , the solution being @xmath115 the bottleneck in this update equation seems to be the matrix inversion , since @xmath116 is a huge @xmath117 matrix .
however , this inversion can be solved very efficiently by resorting to the sherman - morrison - woodbury matrix inversion formula .
in fact , @xmath118 where matrix @xmath119 is diagonal , thus its inversion is trivial . to prove that @xmath120 is diagonal , recall that @xmath121 , thus @xmath122 the element @xmath123 of matrix @xmath124 is the inner product between the @xmath125-th and the @xmath126-th rows of @xmath66 .
since each pixel in the @xmath127-th patch contributes to one and only one pixel in the synthesized image , the rows of @xmath66 have disjoint support , thus @xmath128 , that is , @xmath129 has no non - zero elements outside of its main diagonal , thus is a diagonal matrix . finally , since @xmath120 is a sum of diagonal matrices , it is a diagonal matrix .
it is also easy to obtain explicitly the elements of the diagonal of @xmath120 .
the diagonal elements @xmath130 are given by @xmath131 if the @xmath127-th patch contributes to pixel @xmath125 , the sum in contains exactly one non - zero term , thus it is equal to the square of the weight with which the @xmath127-th patch contributes to the synthesis of pixel @xmath125 . if the @xmath127-th patch does not contributes to pixel @xmath125 , then @xmath132 .
finally , since the weight with which each patch element contributes to each pixel equals the inverse of the number of patches that contributes to that pixel , @xmath133 is equal to the inverse of the number of patches that contribute to pixel @xmath125 . referring to the example in ,
we have simply @xmath134 , because each pixel is synthesized from two patches .
finally , letting @xmath135 , and denoting @xmath136 , we can write the update equation as @xmath137 where @xmath138 denotes element - wise product between two vectors . the leading cost of this update is that of the matrix - vector products involving @xmath79 and @xmath139 , which is @xmath140 ; all the other operations in have lower computational cost .
in this paper , we have revisited patch - based image priors under the light of the synthesis vs analysis dichotomy . after showing that the classical patch - based image models ( namely the epll ) corresponds to an analysis formulation
, we have proposed a patch - synthesis formulation , and analyzed its relationship with the analysis formulation , showing that they are , in general , not equivalent .
finally , we have shown how to address image denoising under the proposed formulation , via an admm algorithm .
we stress again that the purpose of this paper is not to introduce a new particular image prior , but a general patch - based synthesis formulation / framework , which ( to the best of our knowledge ) was missing from the literature on patch - based image processing , and which can be instantiated with many different patch models / priors . for this reason ,
we have abstained from presenting experimental results ; these would critically depend on the choice and estimation of a particular patch model , which is not the focus of this paper .
s. boyd1 , n. parikh , e. chu , b. peleato , j. eckstein , distributed optimization and statistical learning via the alternating direction method of multipliers " , _ foundations and trends in machine learning _ , vol . 3 , pp .
1-122 , 2011 . | in global models / priors ( for example , using wavelet frames ) , there is a well known analysis vs synthesis dichotomy in the way signal / image priors are formulated . in patch - based image models
/ priors , this dichotomy is also present in the choice of how each patch is modeled .
this paper shows that there is another analysis vs synthesis dichotomy , in terms of how the whole image is related to the patches , and that all existing patch - based formulations that provide a global image prior belong to the analysis category .
we then propose a synthesis formulation , where the image is explicitly modeled as being synthesized by additively combining a collection of independent patches .
we formally establish that these analysis and synthesis formulations are not equivalent in general and that both formulations are compatible with analysis and synthesis formulations at the patch level .
finally , we present an instance of the _ alternating direction method of multipliers _ ( admm ) that can be used to perform image denoising under the proposed synthesis formulation , showing its computational feasibility . rather than showing the superiority of the synthesis or analysis formulations , the contributions of this paper is to establish the existence of both alternatives , thus closing the corresponding gap in the field of patch - based image processing . |
for compact metric spaces @xmath5 and @xmath6 , _ the lipschitz distance _ @xmath7 is defined to be the infimum of @xmath8 such that an @xmath9-isometry @xmath10 exists . here a bi - lipschitz homeomorphism @xmath11 is called _
an @xmath9-isometry _ if @xmath12 where @xmath13 denotes the smallest lipschitz constant of @xmath14 , called _ the dilation of @xmath14 _ : @xmath15 let @xmath16 be the set of isometry classes of compact metric spaces .
it is well - known that @xmath17 is a complete metric space .
see , e.g. , ( * ? ? ?
* appendix a ) for the proof of the completeness and see , e.g. , @xcite for details of the lipschitz distance .
then the following question arises : ( q ) : : : is the metric space @xmath17 separable ?
the answer is _ no _ , which can be seen easily by the following facts : ( a ) : : if @xmath18 , the hausdorff dimensions of @xmath0 and @xmath2 must coincide ; ( b ) : : for any non - negative real number @xmath19 , there is a compact metric space @xmath0 whose hausdorff dimension is equal to @xmath19 .
see , e.g. , ( * ? ? ?
* proposition 1.7.19 ) for ( a ) and @xcite for ( b ) .
the fact ( b ) indicates that @xmath17 is too big to be separable .
then we change the question ( q ) to the following more reasonable one ( q ) : for a compact metric space @xmath0 , let @xmath1 be the set of isometry classes of compact metric spaces @xmath2 such that @xmath18 .
any elements of @xmath1 have a common hausdorff dimension by ( a ) .
then the following question arises : ( q ) : : : is the metric space @xmath4 separable ? the main results of this paper give the negative answer for this question for several @xmath0 .
to be more precise , we give two examples for @xmath0 such that @xmath4 is not separable : ( i ) : : infinite unions of shrinking closed intervals with zero @xmath20;\ ] ] ( ii ) : : closed interval @xmath21 $ ] .
we would like to stress that @xmath4 becomes non - separable even when @xmath0 are the above elementary cases .
we note that the non - separability of the first example follows from the non - separability of the second example .
the first example , however , is easier to show the non - separability than the second example .
the present paper is organized as follows : in the first section , we show that the set of isometry classes of the infinite unions of shrinking closed intervals with zero is not separable . in the second section , we show that the set of isometry classes of the closed interval is not separable .
let @xmath22 denote the set of positive integers .
for @xmath23 , let @xmath24 be an interval in @xmath25 defined as follows : @xmath26.\ ] ] for each @xmath27 , we define the following subset in @xmath25 : @xmath28 we equip @xmath29 with the usual euclidean metric in @xmath25 : @xmath30 then it is easy to check that @xmath31 is a compact metric space .
let @xmath32 denote the element in @xmath33 such that all components are equal to one .
let @xmath34 be the set defined in ( [ eq : xu ] ) for the element @xmath35 .
let @xmath36 denote the set of isometry classes of compact metric spaces @xmath0 whose lipschitz distances from @xmath34 are finite , that is , @xmath37 .
then we have the following result : [ thm : xu ] @xmath38 is not separable .
it is enough to find a certain discrete subset @xmath39 with the continuous cardinality .
we introduce a subset @xmath40 , which is the set of isometry classes of all @xmath41 for @xmath42 : @xmath43 it is clear that the cardinality of @xmath44 is continuum .
we show that @xmath45 and @xmath46 is discrete ( i.e. , every point in @xmath44 is isolated ) .
we first show that @xmath45 . for @xmath47 and @xmath48 , let @xmath49 be a function from @xmath29 to @xmath50 defined by @xmath51
then @xmath49 is a bi - lipschitz continuous function from @xmath29 to @xmath50 and for @xmath52 , @xmath53 therefore the lipschitz distance between @xmath41 and @xmath54 is bounded by @xmath55 thus we have that @xmath45 .
second we show that @xmath46 is discrete : [ ce ] let @xmath56 . if @xmath57 , then @xmath58 .
let @xmath47 and @xmath48 .
we show that @xmath59 for all @xmath60 . by the assumption @xmath57
, there exists a bi - lipschitz function @xmath61 such that latexmath:[\[\label{ex1-ass } @xmath14 is homeomorphic , any intervals must be mapped to intervals by @xmath14 .
that is , there exists a bijection @xmath63 as @xmath64 such that @xmath65 to show @xmath59 for all @xmath60 , we have two steps : ( i ) : : @xmath66 ; ( ii ) : : @xmath67 . we start to show ( i ) by contradiction .
assume there exists @xmath68 such that @xmath69 .
since @xmath70 is homeomorphic , the endpoints of @xmath71 must be mapped to the endpoints of @xmath72 by @xmath70 .
therefore @xmath73 thus the dilation of @xmath14 is at least bigger than @xmath74 by the assumption of @xmath69 , we have that @xmath75 or @xmath76 .
this implies @xmath77 this contradicts the inequality ( [ ex1-ass ] ) .
hence we have @xmath66 for all @xmath78 .
we start to show ( ii ) by contradiction .
assume there exists @xmath68 such that @xmath79 .
let us define @xmath80 then @xmath81 by definition .
since we know that @xmath82 by the first step ( i ) , and that @xmath83 and @xmath84 are in @xmath85 , thus the possibility of values of @xmath86 is that @xmath87 , @xmath88 or @xmath89 .
this implies that @xmath90 , @xmath91 and @xmath92 or @xmath93 .
since the endpoints of intervals must be mapped to the endpoints of intervals by @xmath14 , the possibility of values of @xmath94 and @xmath95 is @xmath96 and @xmath97 thus , by noting @xmath98 , we have the following estimate : @xmath99 this shows @xmath100 and contradicts the inequality ( [ ex1-ass ] ) .
hence we have @xmath67 for all @xmath101 . by the above two steps
, we have that @xmath59 for all @xmath60 , and we have completed the proof of lemma [ ce ] .
we resume the proof of theorem [ thm : xu ] .
_ proof of theorem [ thm : xu ] .
_ by using lemma [ ce ] , we know that @xmath102 is discrete . since the cardinality of @xmath44 is continuum and @xmath39 , we have that @xmath103 is not separable .
we have completed the proof .
in this section , we show the non - separability of @xmath104}$ ] : [ thm : int ] the metric space @xmath105},d_l)$ ] is not separable . _ proof .
_ it is enough to find a certain discrete subset @xmath106}$ ] with the continuous cardinality .
define two subsets , _
flat parts _ @xmath107 , and _ pulse parts _ @xmath108 in @xmath109 : * flat part : for @xmath60 , @xmath110\times \{0\},\end{aligned}\ ] ] * pulse part : for @xmath60 , @xmath111\times \{0\}\\ & \cup { \left}\ { { \left}(x,\frac{3}{2^{n+1}}-x { \right } ) : \frac{5}{2^{n+2 } } \leq x \leq \frac{3}{2^{n+1 } } { \right}\}\\ & \cup { \left}\{{\left}(x , x-\frac{1}{2^n}{\right } ) : \frac{1}{2^n } \leq x \leq \frac{5}{2^{n+2}}{\right}\}.\end{aligned}\ ] ] see the figures below : and the right is @xmath108 . ] for each @xmath112 , let @xmath113 be a subset in @xmath109 as an infinite union of flat parts and pulse parts with the origin : @xmath114 see the figure below : . ]
we equip @xmath113 with the usual euclidean distance in @xmath109 : @xmath115 it is easy to check that @xmath116 is a compact metric space .
let @xmath117 be the set of isometry classes of @xmath113 for all @xmath118 : @xmath119 now we show that @xmath106}$ ] .
for @xmath120 , let @xmath121 be the projection from @xmath113 to @xmath21 $ ] such that @xmath122 .
then it is easy to see that @xmath121 is bi - lipschitz continuous and , for @xmath123 , @xmath124 therefore the lipschitz distance between @xmath21 $ ] and @xmath113 is bounded by @xmath125,y_{u})\leq \frac{1}{2}\log 2 \quad \text{$\forall u \in \{0,1\}^{{{{\mathbb z}}_{>0}}}$}.\ ] ] thus we have @xmath106}$ ] .
now we show that @xmath117 is discrete : [ ce2 ] let @xmath126 if @xmath127 then @xmath58 .
let @xmath47 and @xmath48 .
we show that @xmath59 for all @xmath60 . by the assumption , there exists a bi - lipschitz function @xmath14 from @xmath128 to @xmath129 such that @xmath130 let us define a subset in @xmath131 as follows : @xmath132 without loss of generality , we may assume that @xmath133 is not empty .
that is , @xmath113 has at least one pulse .
the pulse part @xmath134 of @xmath113 for @xmath135 is called _ @xmath88-pulse of @xmath113_.
we note that , by the definition of the pulse parts , the peak of the @xmath88-pulse is attained at @xmath136 in @xmath137-axis .
it is enough for the desired result to show that @xmath138 .
we show that there is a bijection @xmath139 such that @xmath140 . to show this , we have the following three steps : ( i ) : : the first step : for @xmath135 , @xmath141 ( ii ) : : the second step : for @xmath135 , @xmath142 ( iii ) : : the third step : for @xmath135 , @xmath143 in fact , if we show the above three statements , each maximizers @xmath136 of @xmath88-pulses of @xmath113 are mapped to the maximizers @xmath136 of @xmath88-pulses of @xmath144 by @xmath145 .
this correspondence of @xmath88-pulses defines the map @xmath139 such that @xmath140 .
the proof of the all three steps ( i)-(iii ) are governed by the same scheme : ( a ) : : assume that the statements do not hold ( proof by contradiction ) ; ( b ) : : estimate lower bounds of the dilations of @xmath14 and @xmath146 ; ( c ) : : the lower bounds obtained in ( b ) contradict the inequality ( [ dil-2 ] ) .
we start to show the first step ( i ) .
since @xmath14 is homeomorphic , the maximizer @xmath136 of the pulse can not be mapped to the endpoints of @xmath21 $ ] by @xmath147 .
assume that , for some @xmath135 , @xmath148 and prove ( i ) by contradiction . by the continuity of @xmath14 , there exists @xmath149 such that , for any @xmath150 $ ] , we have @xmath151 therefore we have @xmath152 here we use the fact that the three points @xmath153 , @xmath154 and @xmath155 are on the same line . by using the inequality ( [ ineq : lip2 ] ) ,
the dilation of @xmath14 is estimated as follows : @xmath156 in the above last line , we just calculated the distance following the euclidean distance ( [ equal : euc ] ) in the @xmath88-pulse @xmath108 .
this implies that @xmath157 , or @xmath158 .
thus we have @xmath159 this contradicts the inequality ( [ dil-2 ] ) .
therefore we have , for any @xmath135 , @xmath160 we start to show the second step ( ii ) by contradiction .
assume that , for some @xmath135 , @xmath161 then there exists @xmath162 such that @xmath163 by the same argument as the first step ( i ) , we can obtain @xmath164 , that is , the @xmath165-pulse exists in @xmath144 . by the continuity of @xmath14 ,
there exists @xmath166 such that @xmath167 and @xmath168 { \right } ) \subset & { \left}(\frac{5}{2^{n+2}}-\frac{1}{2^{n+3}},\frac{5}{2^{n+2 } } { \right } ) \\ & \cup { \left}(\frac{5}{2^{n+2}},\frac{5}{2^{n+2}}+\frac{1}{2^{n+3 } } { \right } ) .
\end{aligned}\ ] ] noting the definition of @xmath169 , for @xmath170 and @xmath171 $ ] , we have @xmath172 since @xmath173 , we have the following inequality : @xmath174 on the other hand , there exist @xmath175 and @xmath176 $ ] such that @xmath177 thus the triangle determined by the three vertices @xmath178 , @xmath179 and @xmath180 is an isosceles right triangle , and we can calculate @xmath181 by ( [ eq : step2 ] ) , ( [ ineq : triangle1 ] ) and ( [ eq : triangle ] ) , we have a bound for the dilation of @xmath14 : @xmath182 here we used the equality ( [ eq : triangle ] ) in the second line , the equality ( [ eq : step2 ] ) and the definition of the dilation in the third line , and the inequality ( [ ineq : triangle1 ] ) in the last line .
the inequality ( [ ineq : dilation ] ) implies that @xmath183 or @xmath184 .
thus we have @xmath185 this contradicts the inequality ( [ dil-2 ] ) .
therefore we have , for any @xmath135 , @xmath186 by the same argument as above , we have , for any @xmath135 , @xmath187 now we start to show the third step ( iii ) . by the above two steps ( i ) and
( ii ) , we have that , for any @xmath135 , there exists @xmath188 such that @xmath189 by the same argument as the first step ( i ) , we can check that @xmath190 , that is , @xmath191 .
also for the inverse function @xmath146 , we have that , for any @xmath192 , there exists @xmath193 such that @xmath194 since @xmath14 is a bijection , the map @xmath195 is a bijection from @xmath133 to @xmath196 and @xmath197 .
now it suffices to show that @xmath198 for all @xmath135 .
we assume that there exists @xmath199 such that @xmath200 . without loss of generality
, we may assume @xmath201 .
we first show that @xmath202 to show this , it suffices to show that @xmath203 where the above inequality means that the point @xmath204 belongs to one of two edges in the @xmath205-pulse crossing at the right angle .
by @xmath206 , we have @xmath207 since we have @xmath208 ( by the inequality ( [ dil-2 ] ) ) and @xmath209 , it holds @xmath210 thus we have shown ( [ incl2 - 0 ] ) . by the continuity of @xmath14 and ( [ incl2 - 0 ] ) , there exists @xmath211 such that @xmath212 and @xmath213{\right } ) \\ & \subset { \left}(\frac{1}{2^{l}},\frac{5}{2^{l+2}}{\right } ) \cup { \left}(\frac{5}{2^{l+2}},\frac{3}{2^{l+1}}{\right } ) .
\end{aligned}\ ] ] since the three points @xmath214 and @xmath215 are on the same line , we have @xmath216 thus the inclusion ( [ incl2 - 0 ] ) and the equality ( [ eq-2 - 1 ] ) imply the following bound of the dilation of @xmath14 : @xmath217 here we used the following equality in the second line : @xmath218 thus @xmath219 or @xmath220 .
this contradicts the inequality ( [ dil-2 ] ) .
therefore we have @xmath198 for any @xmath135 .
we have completed all of the three steps .
setting @xmath221 , we have that the map @xmath139 is a bijection such that @xmath140 and this implies @xmath138 .
we have completed the proof .
we resume the proof of theorem [ thm : int ] .
_ proof of theorem [ thm : int ] .
_ by using lemma [ ce2 ] , we know that @xmath222 is discrete . since the cardinality of @xmath117 is continuum and @xmath106}$ ] , we have that @xmath105 } , d_l)$ ] is not separable .
we have completed the proof .
[ rem : ref ] theorem [ thm : int ] states that @xmath104}=\{x \in \mathcal m : d([0,1],x)<\infty\}$ ] is not separable . by the proof of theorem [ thm : int ] , moreover we know the following stronger result , that is , the non - separability holds locally : _ let @xmath223 , \delta)$ ] denote the ball in @xmath104}$ ] centered at @xmath21 $ ] with radius @xmath211 with respect to the lipschitz distance @xmath224 , that is , @xmath225 , \delta)=\{x \in \mathcal m_{[0,1 ] } : d_l([0,1 ] , x)<\delta\}.\ ] ] then , for any @xmath211 , @xmath226 , \delta)$ ] is not separable .
_ in fact , let @xmath227 \times \ { 0 \ } \\ & \cup \ { ( x , \epsilon ( 3/2^{n+1}-x ) : 5/2^{n+2}\le x \le 3/2^{n+1 } \ } \\ & \cup \ { ( x , \epsilon ( x-1/2^n ) ) : 1/2^n \le x \le 5/2^{n+2 } \ } , \\ j^\epsilon(n,0)=&j(n,0 ) , \\ y^{\epsilon}_{u}=&\{(0 , 0)\ } \cup \bigcup_{n=1}^{\infty}j^{\epsilon}(n , u_n ) , \quad u=(u_n)_{n \in { { \mathbb z}}_{>0 } } \in \{0,1\}^{{{\mathbb z}}_{>0}}.\end{aligned}\ ] ] then , by the similar proof to that of theorem [ thm : int ] , we obtain ( i ) : : for every @xmath228 , the set + @xmath229 is discrete with cardinality of the continuum . (
ii ) : : for every @xmath230 , there exists @xmath228 such that @xmath231 , \delta)$ ] .
the statement ( ii ) implies that @xmath223 , \delta)$ ] is not separable for any @xmath211 .
we would like to thank an anonymous referee for careful reading of our manuscript and pointing out remark [ rem : ref ] .
the first author was supported by grant - in - aid for jsps fellows number 261798 and daad pajako number 57059240 .
burago , d. , burago , y. , and ivanov , s. : _ a course in metric geometry _ , graduate studies in mathematics vol . *
33 * , american mathematical society , providence , ri , ( 2001 ) .
errata list available online at link : www.pdmi.ras.ru / staff/ burago.html.[www.pdmi.ras.ru/staff/ burago.html . ]
gromov , m. : _ metric structures for riemannian and non - riemannian spaces _ , birkhuser boston inc , boston , ma , 1999 , based on the 1981 french original [ mr 85e:53051 ] , with appendices by m. katz , p. pansu , and s. semmes , translated from the french by sean michael bates . | let @xmath0 be a compact metric space and @xmath1 be the set of isometry classes of compact metric spaces @xmath2 such that the lipschitz distance @xmath3 is finite .
we show that @xmath4 is not separable when @xmath0 is a closed interval , or an infinite union of shrinking closed intervals . |
measurements of time - dependent cp asymmetries in @xmath2 decays including @xmath3 , carried out in the past few years at the two @xmath4 @xmath5 factories at slac and kek @xcite , are interpreted in the standard model as @xmath6 , where @xmath7 .
these measurements were proposed in @xcite to provide a first test of the kobayashi - maskawa mechanism of cp violation @xcite in the @xmath5 meson system .
this test is theoretically clean because a single weak phase dominates @xmath8 within a fraction of a percent @xcite .
these measurements have not only passed successfully the standard model test ; they also improve our knowledge of the cabibbo - kobayashi - maskawa ( ckm ) matrix parametrized in terms of the unitarity triangle .
( see fig . 1 @xcite . )
a recent time and angular analysis of @xmath9 @xcite seems to resolve the plotted ambiguity , @xmath10 .
cp asymmetries in @xmath5 decays are often related to the three angles of the unitarity triangle , for which currently allowed ranges are , at 95@xmath11 confidence level ( cl ) @xcite : [ albega ] 78 ^ 122^,21 ^ 27^ , 38 ^ 80^ the essence of measuring cp asymmetries in a variety of @xmath5 and @xmath12 decays is first to see whether the measured asymmetries are consistent with ( [ albega ] ) .
then one would like to restrict these ranges further , hoping to eventually observe deviations from standard model expectations .
a class of processes , susceptible to a possibe early observation of new physics , mimicking loop effects , consists of @xmath13 penguin - dominated @xmath14 decays into cp - eigenstates @xcite , where cp asymmetries may deviate from @xmath15 .
while the phase @xmath16 characterizes cp violation in the interference between @xmath17 mixing and a @xmath2 decay amplitude or a @xmath13 penguin - dominated amplitude , the phase @xmath18 is responsible for direct cp violation .
the phase @xmath19 occurs when @xmath17 mixing interferes with @xmath20 , where an additional @xmath21 penguin amplitude implies also direct cp violation .
since a direct cp asymmetry involves a ratio of two hadronic amplitudes and their relative strong phase , both of which are not reliably calculable , one faces hadronic uncertainties whenever one is relating cp asymmetries to @xmath22 or @xmath23 .
a model - independent way of resolving or at least reducing these uncertainties is by flavor symmetries , isospin or broken su(3 ) , relating certain processes to others .
su(3 ) breaking corrections are introduced using qcd factorization results proven in a framework of a soft collinear effective theory @xcite .
an impressive progress is being reported at this conference in measurements of cp asymmetries @xcite .
the purpose of my talk is to study theoretical implications of these measurements . in section 2
i discuss tree dominated decays , including @xmath24 .
section 3 studies pure tree decays @xmath25 , while section 4 focuses on penguin - dominated processes , @xmath26 .
section 5 concludes .
the amplitude for @xmath29 contains two terms @xcite , conventionally denoted tree " and penguin " amplitudes , @xmath30 and @xmath31 , involving a weak phase @xmath22 and a strong phase @xmath32 : a(^+ ^- ) = @xmath14 or a @xmath33 , are given by @xcite [ sc ] & & ( b^0(t)/(t)^+^- ) + & & e^-t where & = & = 2_eff , + & = & , + & & e^-2i .
recent measurements by the babar @xcite and belle @xcite collaborations , which belle used to announce evidence for cp violation , are @xmath34 these values imply averages , = -0.46 0.13 , = -0.74 0.16 , which also determine a range for @xmath35 . _
2.1.1 isospin symmetry in @xmath36 _ since the two measurables , @xmath37 and @xmath38 , depend on @xmath39 ( and @xmath40 which may be assumed to be given ) , one needs further information to study the weak phase .
this information is provided by isospin symmetry @xcite , which distinguishes between tree and penguin amplitudes , also relating a tiny electroweak penguin term to the tree amplitude @xcite .
one forms an isospin triangle for @xmath5 decays and a similar one for @xmath41 decays , [ pipi ] a(^+^- ) + a(^0 ^ 0 ) = a(^+^0 ) .
a mismatch angle between the two triangles determines @xmath42 , which then fixes @xmath23 from @xmath35 up to a discrete ambiguity . as long as separate @xmath14 and @xmath33 decays into @xmath43
have not yet been measured , one may try to use the three measured charge averaged decay rates for bounds on @xmath44 @xcite .
current branching ratios of the three processes in ( [ pipi ] ) are , in units of @xmath45 @xcite , @xmath46 and @xmath47 , respectively .
these values imply @xmath48 at 90@xmath11 cl , which is not very useful because of the sizable branching ratio into @xmath43 . _
2.1.2 isospin symmetry in @xmath49 _ the two @xmath50 mesons in @xmath51 are described by three possible polarization states , longitudinal or transverse to the momentum direction , parallel or orthogonal to each other in the case of transversely polarized states .
angular distributions of the outgoing pions measured by babar @xcite show that the @xmath50 mesons are dominantly longitudinally polarized , @xmath52 , describing cp - even states .
that is , the case of @xmath53 is almost identical to the case of @xmath54 , albeit a possible small correction from a cp - odd state , and a slight violation of the @xmath49 isospin relation when the two @xmath50 mesons are observed at different invariant masses @xcite .
time - dependent asymmetries in @xmath53 , analogous to @xmath38 and @xmath37 in ( [ sc ] ) , were measured by babar for longitudinally polarized @xmath50 s @xcite , [ csrhorho ] c _ & = & -0.17 0.270.14 , + s _ & = & -0.42 0.42 0.14 .
this implies two possible central values @xmath55 with large experimental errors .
a small branching ratio of @xmath56 , @xmath57 @xcite , much smaller than @xmath58 @xcite and @xmath59 @xcite , implies a @xmath60 cl upper bound @xmath61 @xcite .
this bound and ( [ csrhorho ] ) exclude the range @xmath62 at 90@xmath11 cl @xcite consistent with ( [ albega ] ) . _
2.1.3 broken su(3 ) symmetry for @xmath63 _ as long as separate @xmath14 and @xmath33 decays into @xmath43 have not yet been measured , one may replace isospin symmetry by the less precise flavor su(3 ) symmetry relating @xmath64 to @xmath65 @xcite .
this imposes constraints on the ratio of penguin - to - tree amplitudes in @xmath36 , @xmath66 . to improve the quality of the analysis , one introduces su(3 ) breaking in tree amplitudes in terms of a ratio of @xmath67 and @xmath68 decay constants , as given by factorization @xcite , neglecting a very small variation of the @xmath5 to @xmath68 form factor when @xmath69 varies from @xmath70 to @xmath71
this assumption and the neglect of small annihilation amplitudes may be checked experimentally . here
we follow briefly a study presented recently in @xcite , which contains further references .
writing [ @xmath72 [ akpi+ ] a(b^+ k^0^+ ) & = & -^-1pe^i , + [ akpi- ] a(b^0 k^+^- ) & = & - te^i + ^-1pe^i , one defines two ratios of charged averaged rates , [ r ] r_+ & & = , + r_0 & & = where @xmath73 and , z ( + ) .
current experimental values @xcite , [ r+,0 ] r_+ = 0.235 0.026 , r_0 = 0.209 0.020 , imply a large penguin amplitude 0.51 r 0.85 ( assuming || < /2 ) .
this range is consistent with a recent global su(3 ) fit to charmless @xmath5 decays into two pseudoscalar mesons , @xmath74 @xcite .
however , it is somewhat in conflict with most qcd - based calculations which typically obtain smaller values for @xmath75 @xcite . a study of @xmath23 using the cp asymmetries @xmath38 and @xmath37 proceeds as follows .
one expresses the two asymmetries in terms of @xmath75 , @xmath32 and @xmath23 , [ c ] & = & , + [ s ] & = & . vs. @xmath76.,width=264,height=238 ]
we now impose in addition the su(3 ) constraints ( [ r ] ) and ( [ r+,0 ] ) .
the resulting two plots of @xmath38 vs. @xmath37 @xcite , shown in fig .
2(a ) using @xmath77 and in fig .
2(b ) using @xmath78 , are rather similar , supporting the su(3 ) assumption in these processes . the plotted experimental average ( 7 )
implies an allowed range , @xmath79 , which overlaps nicely with the range of @xmath23 in ( [ albega ] ) , favoring large values in this range .
su(3 ) breaking corrections in penguin amplitudes modify the allowed range of @xmath23 .
the bounds become stronger ( weaker ) if su(3 ) breaking enhances ( suppresses ) the penguin amplitude in @xmath65 relative to @xmath36 .
a @xmath80 su(3 ) correction , implied for instance by a factor @xmath81 ( or by its inverse ) , modifies the bounds by @xmath82 .
very recently broken flavor su(3 ) has also been applied to study @xmath23 in @xmath84 @xcite .
the number of hadronic parameters and the number of measurable quantities are twice as large as in @xmath29 which complicates somewhat the analysis . here
we discuss only a bound on @xmath23 , while a complete determination of @xmath23 will be presented by j. zupan at this conference @xcite .
time - dependent decay rates for initially @xmath14 decaying into @xmath85 are given by @xcite , [ gammat ] ( b^0(t ) ^^ ) 1 + ( c c)mt + - ( s s)mt , while for initially @xmath33 decays the @xmath86 and @xmath87 terms have opposite signs . as in @xmath63 ,
one defines a measurable phase , @xmath88 , which equals @xmath23 in the limit of vanishing penguin amplitudes , 4_eff + + .
[ alphaeff ] the difference @xmath89 , which is governed by penguin contributions , can be shown to be bounded by ratios of strangeness changing decay rates of @xmath90 and @xmath91 ( dominated by penguin amplitudes ) and decay rates of @xmath92 and @xmath93 ( dominated by tree amplitudes ) .
in particular , using the following two ratios of rates , ^+_+ & ^2(b^+k^*0^+)/ ( b^0 ^+^- ) , + r^0_- & ^2(b^0 ^ - k^+ ) ( b^0 ^ -^+ ) , one may show that 2|_eff - | + . using current values @xcite [ expr ] r^+_+ = 0.032 0.007 , r^0_- = 0.047 0.015
, one obtains an upper bound at 90@xmath11 cl , [ delta - alpha ] |_eff - | 12^. this bound , which assumes exact su(3 ) for penguin amplitudes , becomes stronger ( weaker ) if @xmath94 penguin amplitudes are enhanced ( suppressed ) relative to @xmath95 penguin amplitudes . in any event
, one expects this bound to change by no more than 30@xmath11 , implying @xmath96 in the presence of su(3 ) breaking corrections .
both babar @xcite and belle @xcite measured time - dependence in @xmath97 : [ csdelta ] c=\ { c0.35 0.14 0.250.17 .
c=\ { c0.20 0.14 0.380.18 .
s=\ { c-0.13 0.18 -0.280.24 .
s=\ { c0.33 0.18 -0.300.26 . where the first values are babar s and the second are belle s .
these values imply single solutions , [ aleff ] _ eff = \ { c(93 7)^ babar(10211)^ belle , . when making a mild and experimentally testable assumption @xcite that the two arcsin s in ( [ alphaeff ] ) differ by much less than @xmath98 . combining ( [ aleff ] ) with the above upper bound on @xmath99 one obtains = \
{ c(93 7 15)^= ( 93 17)^ babar(1021115)^= ( 102 19)^ . where experimental and theoretical errors are added in quadrature .
these values are in good agreement with the range of @xmath23 in ( [ albega ] ) and largely overlap with this range .
one may show that the relative contributions of penguin amplitudes in @xmath97 are much smaller than in @xmath29 , involving ratios of penguin and tree amplitudes , @xmath100 @xcite .
consequently , effects of su(3 ) breaking in penguin amplitude lead to an intrinsic uncertainty of only a few degrees in the determination of @xmath23 in @xmath84 @xcite .
the process @xmath102 and its charge conjugate provide a way of determining @xmath22 in a manner which is pure in principle , avoiding uncertainties in penguin amplitudes . here
we wish to study a scheme proposed in @xcite , which uses both @xmath103 cp - eigenstates and @xmath103 flavor states
. an extensive list of variants of this method is given in @xcite . in all variants
one makes use of an interference between tree amplitudes in decays of the type @xmath104 , from @xmath105 and @xmath106 , for which the weak phase difference is @xmath22 .
we choose to also discuss briefly one variant @xcite , in which the dalitz plot of @xmath107 is being analyzed in @xmath101 .
we denote by @xmath75 the magnitude of the ratio of two amplitudes , @xmath108 from @xmath109 and @xmath110 from @xmath111 , and we denote by @xmath32 the strong phase of this ratio .
one then obtains the following expressions for two ratios of rates , for even and odd @xmath103-cp states , and for two corresponding cp asymmetries : r _ & = & + & = & 1 + r^2 2r , + a _ & = & + & = & 2r /r_. in principle , these three independent observables determine @xmath112 and @xmath22 .
however , this is difficult in practice since one must be sensitive to an @xmath113 term , where the current 90@xmath11 cl upper limit on @xmath75 is @xmath114 @xcite .
a crude estimate is @xmath115 @xcite , since this ratio involves a ckm factor , @xmath116 , and probably a comparable color - suppression factor .
taking averages of belle and babar measurements when both are available , one finds @xcite r_+&=&1.09 0.16 ( belle & babar ) , + r_-&=&1.30 0.25 ( belle ) , + a_+&=&0.07 0.13 ( belle & babar ) , + a_-&=&-0.19 0.18 ( belle ) .
this implies r = 0.44^+0.14_-0.22 , |a_|_ave = 0.11 0.11 . in order to obtain constraints on @xmath22
we eliminate @xmath32 , plotting in fig
. 3 @xmath117 versus @xmath22 for allowed @xmath118 .
we are using @xmath119 bounds on @xmath120 and @xmath121 .
the ratios @xmath122 and @xmath123 are described by the lower and upper branches , respectively , corresponding to @xmath124 .
this implies a very strong @xmath119 lower bound , @xmath125 , and requires @xmath126 for allowed values of @xmath22
. more precise measurements of @xmath117 are needed for constraints beyond @xmath119 . as functions of @xmath22 for @xmath127 and @xmath128 ( solid curve ) or @xmath129 ( dashed curve ) .
horizontal dashed lines denote @xmath119 experimental lower limits of @xmath123 and @xmath122.,width=264,height=226 ] a recent study by belle of @xmath130 @xcite , analyzed the dalitz plot for @xmath131 in terms of a sum of a non - resonant term and a set of resonances given by breit - wigner forms .
this study , involving some model - dependence , confirmed @xmath126 and obtained @xmath132 bounds , @xmath133 , considerably wider than and including the range of @xmath22 in ( [ albega ] ) .
in a class of penguin - dominated @xmath14 decays into cp - eigenstates , including the final states @xmath134 and @xmath135 , decay amplitudes contain two terms : a penguin amplitude , @xmath136 , involving a dominant ckm factor @xmath137 , and a color - suppressed tree amplitude , @xmath138 , with a smaller ckm factor @xmath139 .
the first amplitude by itself would imply a cp asymmetry of magnitude @xmath140 .
the second amplitude modifies the coefficient of this term , and introduces a @xmath86 term in the asymmetry @xcite .
the coefficients of @xmath87 and @xmath86 for a final state @xmath141 are denoted by @xmath142 and @xmath143 , respectively , as given in eq .
( [ sc ] ) for @xmath63 .
the observables , @xmath144 ( where the sign depends on the final state cp ) and @xmath145 , increase with @xmath146 , but are functions of unknown strong interaction phases .
a search for new physics effects in these processes requires a careful theoretical analysis within the standard model of @xmath147 and @xmath145 and not only of @xmath146 .
model - independent studies of the ratios @xmath146 , providing estimates for @xmath147 for the above final states , were performed in @xcite . here
we will follow ref .
@xcite in order to obtain correlated bounds directly on @xmath142 and @xmath145 in the two cases of @xmath148 and @xmath149 . for simplicity of expressions
we will expand the two asymmetries up to terms linear in @xmath146 .
we will not study theoretical bounds on asymmetries in @xmath150 , where one is awaiting a greater consistency between babar and belle measurements @xcite .
writing a(b^0 ^ 0 k^0 ) = |p|e^i - |c|e^i , and denoting @xmath152 , one obtains @xcite s_k & & 2- 2r2 , + c_k & & - 2r. [ sckpi ] the allowed region in the @xmath153 plane is confined to an ellipse centered at @xmath154 , with semi - principal axes @xmath155_{\rm max } \cos 2\beta$ ] and @xmath156_{\rm max}$ ] .
an approximate ellipse providing these bounds is obtained by relating @xmath157 within flavor su(3 ) to @xmath158 and @xmath159 . for simplicity
, we will first neglect the second process given by an exchange - type amplitude which is expected to be negligible .
it then follows from su(3 ) that a(b^0 ^ 0 ^ 0 ) = -|p|e^i - ^-1|c|e^i . defining a ratio of rates , r_/k , with @xmath160 @xcite
, one has [ rpik ] r_/k = . a scatter plot of @xmath161 vs. @xmath162 , using ( [ rpik ] ) and exact expressions instead of ( [ sckpi ] ) ,
is shown in fig
. 4 @xcite .
it describes the allowed region .
also shown is the experimental point given by a babar measurement @xcite .
the inner ellipse , which is excluded when neglecting @xmath163 , is still allowed when using the current upper bound on this amplitude . in any event
, current errors in the asymmetries are seen to be too large to provide a sensitive test for new physics , and must be improved for such a test .
plane satisfying ( [ rpik]).,width=264,height=170 ] in order to apply su(3 ) to asymmetries in @xmath149 , where measurements are @xcite , s_k=0.270.21 , c_k=0.040.13 , one may optimize bounds over a whole continuum of combinations of corresponding strangeness conserving decays .
these include @xmath14 decays involving @xmath165 and @xmath166 in the final state .
considerable improvements in some of these branching ratios were achieved recently by babar @xcite .
the resulting bounds are shown in fig . 5 @xcite .
regions enclosed by the solid ( dashed ) curve give current bounds including ( neglecting ) annihilation - type amplitudes , while earlier bounds are given by the dot - dashed curve . also shown is a point labeled x denoting the central value predicted in a global su(3 ) fit to @xmath74 @xcite . with recent improvement in bounds on @xmath95 rates ,
only a mild improvement in the asymmetry measurements is required for achieving sensitivity to new physics . )
* nothing is as pure and easy as @xmath8 . *
the evidence for cp violation in @xmath63 is consistent with the km mechanism , and is beginning to exclude low values of @xmath23 permitted by other ckm bounds . *
cp asymmetries in @xmath167 exclude values @xmath168 , in agreement with ckm constraints .
smaller errors in asymmetry measurements may restrict the range of @xmath23 . * bounds on @xmath23 from @xmath83 overlap nicely with the range obtained from other ckm constraints . the intrinsic uncertainty from su(3 )
breaking is merely a few degrees .
* current @xmath102 measurements are consistent with ckm constraints on @xmath22 and require more statistics for stronger bounds . *
all this indicates that the km phase governs cp violation in tree - dominated decays .
* some improvement in asymmetry measurements in @xmath169 , and convergence of babar and belle in @xmath150 , are required for sensitivity to new physics .
# 1#2#3am .
j. phys .
* # 1 * ( # 3 ) # 2 # 1#2#3ann .
( n.y . ) * # 1 * , # 2 ( # 3 ) # 1#2#3acta phys .
polonica * # 1 * , # 2 ( # 3 ) # 1#2#3ann .
nucl . part .
sci . * # 1 * , # 2 ( # 3 ) # 1#2#3comments on nucl . part .
phys . * # 1 * , # 2 ( # 3 ) # 1#2#3electronic conference proceedings * # 1 * , # 2 ( # 3 ) # 1#2#3eur .
j. c * # 1 * ( # 3 ) # 2 # 1#2#3 * # 1 * , # 2 ( # 3 ) # 1#2#3int . j. mod .
a * # 1 * ( # 3 ) # 2 # 1#2#3jhep * # 1 * ( # 3 ) # 2 # 1#2#3j .
b * # 1 * , # 2 ( # 3 ) # 1#2#3mod .
lett . a * # 1 * ( # 3 ) # 2 # 1#2#3nature * # 1 * , # 2 ( # 3 ) # 1#2#3nuovo cim .
* # 1 * , # 2 ( # 3 ) # 1#2#3nucl . instr .
meth . a * # 1 * , # 2 ( # 3 ) # 1#2#3nucl .
b * # 1 * ( # 3 ) # 2 # 1#2#3nucl . phys .
suppl . *
# 1 * ( # 3 ) # 2 # 1#2#3#4pisma zh .
fiz . * # 1 * , # 2 ( # 3 ) [ jetp lett . *
# 1 * , # 4 ( # 3 ) ] # 1#2#3phys .
lett . * # 1 * , # 2 ( # 3 ) # 1#2#3phys .
lett . a * # 1 * , # 2 ( # 3 ) # 1#2#3phys .
b * # 1 * ( # 3 ) # 2 # 1#2#3phys .
# 1 * ( # 3 ) # 2 # 1#2#3phys .
d * # 1 * ( # 3 ) # 2 # 1#2#3phys
. rep .
* # 1 * ( # 3 ) # 2 # 1#2#3prog .
phys . * # 1 * ( # 3 ) # 2 # 1#2#3rev .
* # 1 * ( # 3 ) # 2 # 1 rp # 1 # 1#2#3#4yad .
# 1 * , # 2 ( # 3 ) [ sov . j. nucl .
# 4 ( # 3 ) ] # 1#2#3#4#5#6zh .
# 1 * , # 2 ( # 3 ) [ sov.phys
. - jetp * # 4 * , # 5 ( # 6 ) ] # 1#2#3zeit . phys .
* # 1 * ( # 3 ) # 2 # 1#2#3zeit .
c * # 1 * ( # 3 ) # 2 # 1#2#3zeit .
d * # 1 * , # 2 ( # 3 ) | recent cp asymmetry measurements in tree - dominated processes , @xmath0 , and in penguin - dominated decays , @xmath1 , are interpreted in the framework of the kobayashi - maskawa ( km ) mechanism of cp violation .
the km phase emerges as the dominant source of cp violation in tree - dominated decays , which are beginning to constrain the unitarity triangle beyond other constraints .
improving precision of cp asymmetry measurements in penguin - dominated decays may indicate the need for new physics . |
helium , the second most abundant chemical element in the universe , is expected to be singly ionised at the same time as hydrogen due to its similar ionisation potential ( @xmath4 ) , larger photo - ionisation cross - section and lower abundance relative to hydrogen .
even for ionising sources with relatively soft , stellar like spectra , the reionisation of and in the igm is expected to be concurrent ( _ e.g. _ @xcite ) .
in contrast , the much larger ionisation threshold ( @xmath5 ) and smaller photo - ionisation cross - section of singly ionised helium requires that reionisation proceeds only when sources with sufficiently hard spectra become numerous . a putative population of massive , metal free stars ( @xcite ) , high redshift x - rays
( @xcite ) or thermal emission from shock heated gas ( @xcite ) may provide the requisite ionising photons .
the stacked spectra of lyman break galaxies at @xmath6 ( @xcite ) also exhibit an recombination line strong enough to suggest that galaxies may contribute significantly to reionisation ( @xcite ) .
however , quasars are considered to be the most likely candidates for completing reionisation ( @xcite ) , although bright quasars are rare at high redshift and unlikely to contribute substantially to either the or ionising photon budget at @xmath7 ( @xcite ) . consequently , unless reionisation was completed by an as yet unidentified population of sources with hard spectra at @xmath7 , its latter stages are expected to coincide with the observed peak in quasar activity around @xmath8
. there are several pieces of observational evidence which may support this picture .
observations of the gunn - peterson trough and patchy absorption in the forest at @xmath9 indicate a situation analogous to that implied by the opacity data at @xmath10 ( @xcite ) . however , like the measurements at higher redshift ( _ e.g. _ @xcite ) , the interpretation of the data is hampered by its insensitivity to the fraction once the absorption becomes saturated .
other indirect evidence includes an apparent boost to the igm temperature at @xmath11 ( @xcite ) , measurements of metal line ratios which indicate the ultraviolet ( uv ) background is hardening around the same redshifts ( @xcite ) , and the detection of a sharp dip in the forest effective optical depth at @xmath12 ( @xcite ) .
the latter result may be associated with the observed boost in the igm temperature at the same redshifts ( @xcite ) . on the other hand
, conflicting evidence exists for all of these cases ( _ e.g. _ @xcite ) .
thus , the observational evidence for reionisation remains controversial .
nevertheless , from a theoretical standpoint it is still expected reionisation will have a significant impact on the thermal state of the igm ( @xcite ) .
firstly , energy injected by the hard sources which reionise will raise the igm temperature .
this is consistent with some of the observational constraints derived from forest line widths which imply a boost of @xmath13@xmath14 around @xmath6 ( @xcite ) .
the magnitude of the increase will depend on the spectrum of the intrinsic ionising radiation , and thus also on the amount it is subsequently filtered by the intervening igm ( @xcite ) .
secondly , the reionisation of will alter the temperature - density relation of the igm . in the optically thin limit
, the igm will follow a tight polytropic temperature - density relation , @xmath15 , expected to arise for normalised densities @xmath16 when photo - heating by a spatially uniform uv background is balanced by cooling due to adiabatic expansion ( @xcite ) .
however , the assumption of a spatially uniform uv background ionising an optically thin igm will break down during reionisation . instead
, radiative transfer effects and/or the inhomogeneous distribution of the ionising sources will produce a complex , multi - valued and perhaps even inverted ( @xmath17 ) temperature - density relation ( @xcite ) . despite all of these studies , very little work has been done to establish a physical argument for exactly how large the temperature boost in the igm during reionisation will be .
furthermore , it is unclear over what timescale any global temperature boost should occur .
this depends on understanding how photo - heating in an optically thick igm proceeds , and thus on the details of radiative transfer through the inhomogeneous igm .
the observational evidence provides valuable hints , but the data are still too poorly constrained to distinguish between different models in detail . a wide variety of values for the temperature boost during reionisation , @xmath18 , have therefore been adopted in the literature , with the timescales for the temperature boost varying from instantaneous to periods spanning several redshift units ( _ e.g. _ @xcite ) .
the main reason for this is that photo - heating in an optically thick igm during reionisation is almost always modelled in an approximate way . only a handful of studies
have explicitly quantified the expected igm temperature boost during reionisation using radiative transfer calculations .
@xcite were the first to point out the importance of radiative transfer through an optically thick igm .
using one dimensional radiative transfer simulations , they found the filtering of ionising radiation through the igm could increase photo - heating rates by a factor of @xmath19@xmath20 above the optically thin values , leading to @xmath21 .
more recently , @xcite demonstrated that the temperature boost depends sensitively on the typical spectra of the sources which drive reionisation .
finally , @xcite have presented a detailed , three dimensional radiative transfer simulation of reionisation .
they found that radiative transfer effects could boost the igm temperature by around a factor of two .
however , none of these studies explicitly discuss over what timescale this heat input may reasonably occur , or how the temperature boost depends in detail on the density distribution of the intervening igm .
a clear understanding of photo - heating during reionisation is thus vital for interpreting the thermal history of the igm . in this paper
we critically re - examine the photo - heating of the igm by quasars during reionisation .
in particular , using a combination of analytical arguments and line - of - sight radiative transfer calculations , we shall address what the likely temperature boost in the igm during reionisation is , and how rapidly this temperature boost should occur .
furthermore , we consider what the fate of the hard photons responsible for heating the igm is likely to be : where do they end up in the igm , and how important is the filtering of ionising radiation through the igm in setting the temperature during reionisation ? we begin in 2 , where we briefly outline the relationship between photo - heating and the photo - ionisation timescale .
this argument is key to establishing the amount of heat which may be injected into the igm within a given time . in
3 we present a model for estimating the maximum temperature achievable during reionisation in an optically thick igm .
a more detailed treatment of the filtering of ionising photons through a clumpy medium is presented in 4 , and in 5 we consider the impact of photo - heating in recombining fossil regions .
we then compare our analytical results with detailed line - of - sight radiative transfer simulations in 6 , and use these results to discuss the impact of reionisation on the igm temperature density relation in 7 .
we summarise our results , compare them to current observational constraints and conclude in 8 . throughout this paper
we shall assume @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 and an igm of primordial composition with a helium mass fraction of @xmath28 ( _ e.g. _ @xcite ) .
the cosmological parameters are consistent with the most recent wmap results ( @xcite ) aside from a slightly larger normalisation for the power spectrum . unless otherwise stated , all distances are expressed as comoving quantities .
we begin by considering an ideal gas parcel of density @xmath29 exposed to a spatially uniform ionising radiation field with specific intensity @xmath30 , where @xmath31 and @xmath32 are the specific intensity and frequency at the ionisation threshold , respectively , and @xmath33 is a power - law spectral index .
the thermal evolution of the gas parcel in a cosmologically expanding medium is then ( _ e.g. _ @xcite ) @xmath34 + \frac{2t}{3(1+\delta)}\frac{d\delta}{dt } + \frac{t}{\mu}\frac{d\mu}{dt } - 2ht , \label{eq : temp}\ ] ] where @xmath35^{-1}$ ] is the mean molecular weight of the gas , @xmath36 is the total photo - heating rate per unit volume summed over the species @xmath37[h @xmath38 , he @xmath39 , he @xmath40 with number density @xmath41 , @xmath42 is the cooling rate per unit volume , @xmath43 is the density contrast , @xmath44 is the hubble parameter and all other symbols have their usual meaning . the first term in eq .
( [ eq : temp ] ) is the dominant heating term ; during reionisation @xmath45 at the low densities and temperatures typical of most of the igm ( @xcite ) .
the photo - heating rate for species
@xmath37[h @xmath38 , he @xmath38 , he @xmath46 is @xmath47 where @xmath48 is the optical depth to the ionising photons summed over the species @xmath49[h @xmath50 , he @xmath38 , he @xmath40 and @xmath51 is the frequency at the relevant ionisation threshold .
note here that @xmath52 is an _ intrinsic _ spectrum and does not include any reprocessing by the igm .
the second term in eq .
( [ eq : temp ] ) describes the adiabatic heating or cooling due to structure formation , and is small for the modest density contrasts associated with most of the igm . the third term , which results from changes in the mean molecular weight of the gas parcel , is also small compared to the photo - heating rate during reionisation .
the final term corresponds to the adiabatic cooling driven by the expansion of the universe , and is the dominant cooling term at the moderate to low densities ( @xmath53 ) still expanding with the hubble flow .
the relevant cooling timescale for the low density , photo - ionised gas in the igm is therefore always the hubble time , @xmath54 ( @xcite ) .
let us now assume the and in the gas parcel has already been ionised at some earlier time , and that the reionisation of is now completed by the radiation field @xmath52 .
we shall further assume that the gas parcel is exposed to this radiation field for a time @xmath55 .
the temperature evolution of the gas parcel at @xmath9 may then be approximated by @xmath56 ignoring the collisional ionisation of he @xmath57 , which is a reasonable approximation in the presence of large photo - ionisation rates , the rate of change of @xmath58 is @xmath59 where @xmath60 are the recombination rates for species @xmath61 and in general the photo - ionisation rate for species @xmath37[h @xmath50 , he @xmath38 , he @xmath40 is @xmath62 finally , assuming the case b recombination timescale @xmath63^{-3 } \gg t_{\rm s}$ ] , eq .
( [ eq : rates ] ) reduces to @xmath64 , and upon substitution in eq .
( [ eq : tempapprox ] ) yields @xmath65 where @xmath66 is the mean excess energy per photo - ionisation . in the optically thin limit ( @xmath67 ) and assuming the photo - ionisation cross - sections are proportional to @xmath68 , this reduces to @xmath69 ( @xcite ) . eq .
( [ eq : tempapprox2 ] ) thus elucidates the key parameters which influence the temperature of an optically thin gas parcel during reionisation ; @xmath70 , which depends _ only _ on the spectral shape and not the specific intensity of the ionising radiation , and the rate of change of @xmath58 , which _ does _ implicitly depend on the intensity .
expressed differently , @xmath71 , where @xmath72 is the photo - ionisation timescale .
if @xmath73 , no significant heating will occur in the gas parcel irrespective of the average excess energy per photo - ionisation . )
ionisation equilibrium is achieved and recombinations become important . then in general for a species @xmath61 , @xmath74 ( _ e.g. _ @xcite ) .
the heating rate is now proportional to the igm density and is independent of the intensity of the ionising radiation if the ionised fraction is close to unity . ]
although this argument is applied here to a single optically thin gas parcel for simplicity , it also extends more generally to the igm as a whole . a substantial temperature boost in the igm
can only occur globally within a timescale @xmath75 if the ionising background is sufficiently hard _ and _ intense .
if the condition @xmath76 is not met everywhere , the ionisation rate is too low to significantly photo - heat the entire igm .
instead , temperature boosts are possible only in the vicinity of quasars where the local photo - ionisation rate is large .
alternatively , a large , global igm temperature boost over longer timescales is still possible if the ionising background is weak , yet hard and persistent .
crucially , these scenarios depend on exactly how many hard photons are produced by quasars over time and the impact of the subsequent transfer of radiation though the igm on the intrinsic quasar emission . bearing these points in mind , we now proceed to discuss the expected temperature of the igm during reionisation in more detail .
the temperature boost in the igm during reionisation is approximately ( @xcite ) @xmath77 where @xmath78 is the fraction in the igm when photo - heating commences .
this assumes the excess energy per photo - ionisation is shared equally among all the baryons in the igm through coulomb interactions . in the optically thin limit ,
@xmath79 for @xmath80 , giving @xmath81 .
radiative transfer effects during reionisation can substantially increase this temperature boost .
an optically thick igm will alter the spectral shape of any incident ionising radiation ( @xcite ) . in a uniform medium
the ionising photon mean free path @xmath82 ; high energy ionising photons have a strongly suppressed photo - ionisation cross - section , where @xmath83 . in an optically thick igm
the average excess energy per photo - ionisation is therefore progressively larger the further from the ionising source ( @xcite ) .
this hardened , filtered ionising radiation can then impart a significant temperature boost to the igm compared to the optically thin case . in the optically thick limit , where every ionising photon is eventually absorbed and one may ignore the @xmath68 weighting from the photo - ionisation cross - section , @xmath84 for @xmath80 , leading to @xmath85 .
thus , a conventional interpretation is that any large , global boost to the igm temperature at @xmath9 ( inferred either directly or indirectly from observations ) may be due to photo - heating rates which are amplified by radiative transfer effects during reionisation ( _ e.g. _ @xcite ) .
recall , however , that irrespective of how large @xmath86 is , no significant photo - heating will occur in a patch of the igm over timescales shorter than the local photo - ionisation timescale , @xmath72 .
ionising radiation is significantly weakened as well as hardened at progressively larger distances from an ionising source .
this clearly has implications for the @xmath87 attainable over a fixed timescale .
the hardest , most highly filtered radiation may be too weak to significantly heat a patch of the igm over the short timescales associated with quasar activity .
quantifying the effect of this filtering and weakening of the ionising radiation field on the igm temperature evolution is non - trivial ; the temperature will depend on the density distribution of the intervening igm , the intrinsic spectra and luminosities of the ionising sources , their spatial distribution and their lifetimes .
detailed numerical simulations of radiative transfer are therefore required to fully address this issue .
we may nevertheless develop some intuition by first using a simple toy model for the effect of radiative transfer on the igm temperature . consider a quasar which emits @xmath88 ionising photons per second above the ionisation threshold where @xmath89 and which has a broken power law spectrum given by ( _ e.g. _ @xcite ) @xmath90 there are a wide range of values for the extreme uv ( euv ) spectral index , @xmath33 , observed in the quasar population at @xmath91 ( @xcite ) .
we adopt a fiducial value of @xmath80 , consistent with the mean @xcite obtain for radio - quiet quasars .
however , the euv spectral index is a significant uncertainty .
for example , @xcite find a somewhat softer mean euv index of @xmath92 may be appropriate , while on the other hand @xcite find @xmath93 at @xmath94 .
we therefore also consider softer and harder quasar spectra with @xmath95 and @xmath96 , broadly consistent with the dispersion in @xmath33 measured by @xcite .
the rate at which ionising photons are produced by such a quasar with b - band luminosity @xmath97 ( where @xmath98 evaluated at @xmath99 ) is then @xmath100 an @xmath101 quasar at @xmath102 yields @xmath103 for @xmath104 .
note the quasar spectrum which is harder / softer than the fiducial @xmath105 model also emits many more / less ionising photons below the ionisation threshold for fixed @xmath97 .
quasars with @xmath106 are expected to dominate the ionising emissivity at @xmath107 ( @xcite ) .
we adopt @xmath108 as the fiducial luminosity throughout this paper .
let us now invoke a cut - off at some arbitrary frequency in the quasar spectrum above the ionisation threshold , @xmath109 .
one may think of this as a condition which requires @xmath110 for @xmath111 and @xmath112 for @xmath113 , so the igm is optically thin only to photons with frequencies above the cut - off .
this is obviously a simplification of the full radiative transfer problem ; in reality the optical depth is a continuous function of frequency and the amount of attenuation will depend on the density distribution of the intervening igm .
this approach is nevertheless useful because it crudely mimics the effect of the filtering and weakening of the ionising radiation emitted by the quasar .
the photo - ionisation rate at a comoving distance @xmath114 from the quasar is @xmath115 it is then straightforward to integrate this expression to obtain @xmath116 @xmath117 the photo - ionisation rate is thus strongly dependent on the frequency of the assumed cut - off in the ionising spectrum , where @xmath118 . on the other hand ,
the mean excess energy per ionisation increases with the cut - off frequency , such that @xmath119.\ ] ] assuming @xmath120 , substituting into eq .
( [ eq : igmtemp ] ) yields @xmath121 .
\label{eq : cutofftemp}\ ] ] hence , for a quasar with a fixed luminosity and intrinsic spectral shape , we may estimate the _ maximum _ temperature boost which can be achieved through photo - heating of if the intrinsic spectrum has been filtered by some arbitrary but constant amount .
note that this simple argument does not include any information regarding the likelihood ionising radiation is filtered by the assumed amount , nor the timescale for which it remains filtered ; to do so requires a detailed knowledge of the intervening igm density distribution .
rather , this model assumes _ a priori _ that the quasar radiation is already filtered by the igm as parameterised by the frequency cut - off @xmath122 , and remains so throughout the lifetime of the ionising source .
we will return to the issue of how likely ionising radiation is filtered by a certain amount in
[ peng_section ] .
finally , we must also assume a fiducial lifetime for the quasar .
recent numerical studies indicate optically bright quasars have lifetimes of @xmath123 ( @xcite ) , while a wide variety of observational data suggest quasar lifetimes vary from @xmath124 ( @xcite ) .
we adopt the upper limit of @xmath125 yr which conservatively maximises the effect of photo - heating as the fiducial lifetime , although we will discuss the effect of shorter quasar lifetimes later . this toy model
is illustrated in fig .
[ fig : mfp ] .
the photo - ionisation timescale @xmath126 , computed from eq .
( [ eq : tion ] ) , is displayed as a function of @xmath114 and @xmath122 for three different quasar models at @xmath127 . in all panels
the black lines correspond to the ionisation timescale assuming @xmath128 , where @xmath129 , and the horizontal dotted line denotes our fiducial model quasar lifetime of @xmath130 .
the underlying colour scale shows the logarithm of the expected temperature increase calculated using eq .
( [ eq : cutofftemp ] ) .
the unshaded regions in the lower right of each panel lie below the minimum ionisation timescale corresponding to the unfiltered quasar radiation field , where @xmath131 .
the left - most panel fig .
[ fig : mfp ] displays the photo - ionisation timescales for our fiducial quasar model ( @xmath132 , @xmath80 ) .
recalling that we require @xmath76 for significant heating to occur , we see a temperature increase @xmath85 ( @xmath133 ) , expected in the optically thick limit where every ionising photon is absorbed so that @xmath134 , is only possible very close to the quasar if @xmath135 ( or equivalently @xmath136 at @xmath127 ) . ionising radiation which has been hardened sufficiently to produce such a large temperature boost is too weak to photo - ionise the at @xmath137 .
a hard , intense ionising radiation field is in any case unlikely to be maintained in such close proximity to the quasar for long periods . as noted by @xcite ( and see also 6 ) , the igm will instead quickly become optically thin to all ionising photons , thereby limiting the temperature boost within this region to a substantially smaller value . over a quasar lifetime of @xmath130 ,
the temperature boost is restricted to @xmath138 at @xmath139 from the quasar , and even less at larger distances .
note again , however , that the exact amount of filtering will depend sensitively on the intervening density distribution of the igm .
in addition , we stress these temperatures represent the _ maximum _ possible values within a fixed volume around a quasar , since all photons with @xmath140 are neglected in eq .
( [ eq : cutofftemp ] ) .
although these low energy photons do increase the total ionisation rate , they _ decrease _ the average heating .
these estimates also ignore the effect of any cooling on the resulting igm temperature .
greater temperatures boosts can be achieved out to larger distances around a quasar for either a harder intrinsic spectrum , a brighter luminosity , or both .
the former increases the average excess energy per photo - ionisation , while the latter decreases the photo - ionisation timescale . the central panel in fig .
[ fig : mfp ] shows the results for the model with @xmath108 and @xmath141 .
in addition to having a harder spectrum , this quasar model produces around @xmath142 times the number of ionising photons in our fiducial model . on the other hand , a softer , fainter quasar spectrum with @xmath108 and @xmath143 , as shown in the right - most panel of fig .
[ fig : mfp ] , yields much smaller temperatures for a given amount of filtering .
this simple model clearly illustrates the difficulty of attaining substantial temperature boosts over large volumes in the igm within short timescales .
for typical quasar euv spectral indices and luminosities , except for regions in close proximity to the quasar , there are simply not enough hard photons to raise the igm temperature substantially above the value expected in the optically thin limit .
furthermore , in this example we have assumed @xmath144 and @xmath145 , which is around the upper limit for plausible quasar lifetimes ( @xcite ) .
if the lifetimes of optically bright quasars are significantly shorter than @xmath130 ( @xcite ) , or if a patch of the igm has @xmath146 as might be expected if fossil regions are common prior to the completion of reionisation ( @xcite ) , then the temperature boost around a single quasar will be accordingly smaller .
thus , for our fiducial quasar model a boost to the igm temperature on timescales @xmath147 ( @xmath148 at @xmath9 ) in excess of @xmath149 appears unlikely over scales of @xmath150 .
given that the average separation between @xmath151 quasars is @xmath152 ( @xcite ) , this implies that a rapid _ global _ igm temperature boost in excess of @xmath153 is difficult to achieve . one important caveat , however , is that we have applied the above argument to the photo - heating around individual quasars only .
over longer timescales , the integrated effect of photons emitted by distant quasars with highly filtered spectra could produce a more gradual yet larger igm temperature boost .
however , we shall see it is quite difficult to reionise the igm with _ only _ hard photons , so the actual temperature boost will likely be even smaller than these estimates .
we address this issue in more detail in
[ peng_section ] .
thus far we have have focused on the expected temperature boost in the igm without explicitly quantifying the _ amount _ of filtering which occurs through an optically thick , inhomogeneous igm .
furthermore , we have not considered _ where _ in the igm these hard photons are subsequently absorbed . in this section ,
we consider some simple analytical arguments for the amount of filtering expected , and the fate of hard photons . while less rigorous , these serve to motivate the detailed numerical calculations that follow later , and promote physical intuition .
specifically , we shall address two key arguments for substantial heating during reionisation , which we briefly restate here .
firstly , we know from the lifting of the gunn - peterson trough at @xmath154 along multiple lines - of - sight that most of the in the universe has been reionised by this time ( _ e.g. _ @xcite ) .
thus , all that matters is the average heat input per photo - ionisation , which as discussed earlier is simply @xmath155 for an optically thick igm .
this has to be true from a simple energy conservation standpoint , and implies significant heating .
secondly , the spacing between quasars is much larger than the spacing between lyman limit systems , which dominate the opacity ( @xcite ) .
thus , most of the igm is exposed to filtered , hardened radiation . while such radiation is undoubtedly weakened , even if the photo - ionisation rate is low over long periods of time it may again still be possible to achieve substantial heating from a weak but persistent and hard ionising background .
thus , while we have so far demonstrated that it is difficult to achieve a large temperature boost globally in the igm over short timescales , it still appears to be possible over longer periods of time . while the latter scenario is certainly feasible , there are a number of uncertainties and caveats , which have not been emphasised in the literature . in particular : ( i ) while the average energy per ionisation in an optically thick igm is indeed @xmath156 , dense regions preferentially absorb hard photons .
the energy per photo - ionisation in the low density regions may therefore be systematically less than the average , since the average heat input in the observable igm depends on the relative collective opacity of optically thin and thick systems .
note that virtually all statistical measures of temperature in the forest are sensitive to the optically thin , underdense to mildly overdense regions . as an example , the low column density forest ( where doppler line widths are directly measurable ) corresponds to regions of fairly low overdensity @xcite .
( ii ) there are many generations of quasars . while at any given time a patch of the igm may be exposed only to filtered radiation , over many generations a nearby quasar will eventually be able to directly ionise that patch with only modestly filtered radiation . reionising the igm exclusively via soft photons does not require an unrealistic photon budget , even with a substantial population of lyman limit systems .
why should hard photons be preferentially absorbed in dense systems ?
we can gain some physical insight from a simple toy model .
let us consider a uniform igm filled with a single population of absorbers of fixed mass and some overdensity @xmath157 .
for this population , the radial size of the absorbers @xmath158 , and hence the optical depth @xmath159 . the effective optical depth due to the absorbers has two interesting limits .
when @xmath157 is low so that the absorbers are optically thin , @xmath160 , where @xmath161 is the mean number of absorbers encountered along a line - of - sight , @xmath162 is the mean optical depth of a single absorber , and @xmath163 is the absorber cross - section . in this limit , the effective optical depth is independent of the overdensity @xmath157 of absorbers : @xmath164 , but for increasing overdensity their cross - section falls , and thus fewer absorbers , @xmath165 , are encountered along a line - of - sight . on the other hand , once the absorbers become optically thick , their contribution to the opacity saturates , and @xmath166 , where @xmath167 is the total mass fraction of in the absorbers .
this implies that @xmath168 thus , on average fewer photons are absorbed in optically thick absorbers , since only the photosphere ( @xmath169 ) contributes to the opacity . for hard photons
the absorbers are optically thin , while for soft photons the absorbers are optically thick . thus , a greater fraction of hard photons are absorbed by dense absorbers , rather than the igm , compared to soft photons .
similarly , in the interstellar medium of galaxies , continuum photons can have a higher probability of being absorbed by dusty atomic / molecular clouds ( which are optically thin at these frequencies ) , compared to photons , for which the clouds are optically thick @xcite .
photons is always larger in a clumpy igm compared to a uniform igm , so that the transmission is always dominated by underdense voids ( e.g. @xcite ) .
all this stems from the triangle inequality , @xmath170
\rangle \leq { \rm exp } \left [ - \langle \tau \rangle \right]$ ] . ] from the above , it is clear that the relative absorption of hard and soft photons depends on the abundance of optically thick systems , with column densities which depend on frequency as @xmath171 . in particular , the fate of hard photons depends strongly on the abundance of high column density systems , which are optically thick to photons just above the lyman edge , but optically thin to hard photons .
unfortunately , unlike the hydrogen forest , it is not possible to directly observe the column density distribution @xmath172 of the forest in the uv .
one is therefore forced to theoretical models of @xmath173 @xcite .
there are a number of subtle modifications which need to be made to such models while reionisation is in progress , and we present detailed calculations elsewhere ( oh et al .
2009 , in prep . ) .
for now , we merely remark that @xmath172 can be significantly model dependent , and point out the implications of these uncertainties .
the effective optical depth experienced by a photon which has frequency @xmath174 at redshift @xmath175 between redshifts @xmath176 due to discrete , poisson distributed clouds is @xcite @xmath177 , \label{eq : tau_eff}\ ] ] where @xmath178 , the frequency @xmath179 , and @xmath180 .
note it is generally assumed that the opacity is negligible @xcite , although there may be regimes where this is not justified . while the column density distribution @xmath181 is observationally determined as a function of redshift
, @xmath182 is not .
thus , above 4 ry , @xmath183 depends strongly on the unknown function @xmath184 . given a model for @xmath185 which usually depends on @xmath186 and @xmath187 and an assumed ionising source population of quasars and galaxies
, @xmath188 and @xmath189 can be self - consistently computed as a function of redshift .
the coupled equations have to be solved iteratively @xcite , since the opacity depends on the ionising background and vice - versa .
while @xmath188 can be compared to observations of the ly@xmath190 forest , @xmath189 can not above @xmath6 , when absorption in the forest saturates .
this is unfortunate : the mean free path of ionising photons and hence @xmath189 evolves strongly during reionisation , a fact which can cause considerable uncertainty in the redshift evolution of @xmath182 .
let us consider the qualitative behaviour of @xmath185 @xcite . in the optically thin limit , @xmath191 , where @xmath192 is the ionisation fraction and
@xmath193 is the ionising background for a given ion . in this case
, @xmath182 is a constant independent of @xmath194 .
however , this is no longer true at higher column densities . at some point ,
becomes self - shielding and recombines , causing @xmath182 to rise above the optically thin value . at even higher column densities , recombines and
@xmath195 plummets drastically .
this modulation of @xmath182 with @xmath196 has been modelled by radiative transfer calculations of slabs illuminated by ionising radiation on both sides @xcite .
the results depend on assumptions about @xmath188 and @xmath189 , but also crucially the gas density @xmath187 , which then requires an uncertain and model - dependent correspondence to be made between @xmath187 and @xmath197 .
for instance , in the widely used models of @xcite , which assume separate fixed densities for systems above and below the self - shielding limit @xmath198 , this turnover takes place at @xmath199 . on the other hand ,
@xcite , following the _ ansatz _ of @xcite , assume that @xmath200 , and instead find that the turn - over takes place at @xmath201 , depending on the assumed value of @xmath189 .
this model - dependence has little impact on self - consistent calculations of @xmath189 ( at least once reionisation is complete ) .
this is because ionising photons are strongly weighted toward the lyman edge , and the turn - over in @xmath182 only takes place in systems which are already optically thick to photons at @xmath202 ry .
for such photons , it does not really matter whether the optical depth of a system is @xmath203 or @xmath204 ; the high absorption probability changes only slightly . for this reason , previous studies
correctly did not regard this model - dependence as a liability . on the other hand , for higher energy photons for which such systems are still optically thin the exact column density _ is _ important , since @xmath205 in this regime .
this in turn translates into considerable uncertainty about the frequency dependence of the opacity , and hence the spectral shape of ionising background and the photo - heating of the igm .
for instance , let us consider the mean excess energy per photo - ionisation @xmath206 @xmath207 here @xmath208 is the ionising emissivity @xmath209 $ ] , @xmath210 is the mean free path for ionising photons and @xmath211 , where @xmath212 is the photo - ionisation cross - section at the ionisation threshold . for a uniform optically thick igm , @xmath82 and @xmath213 , while for the optically thin case , @xmath214 is constant and comparable to the size of the system under consideration , yielding @xmath215 . of course ,
the full inhomogeneous density structure of the igm must be taken into account , using eq .
( [ eq : tau_eff ] ) . if we ignore the effect of continuum recombination radiation ( which can significantly soften the spectrum ) , the widely used model of @xcite yields @xmath216 , and hence @xmath217 ( f. haardt , private communication ) .
this level of heating is intermediate between the optically thick and optically thin case .
the result is easy to understand : the turnover in @xmath182 occurs at high values of @xmath194 , and the assumption of constant @xmath182 is a reasonable first approximation .
doing so , and assuming that @xmath218 , we obtain : @xmath219 \left ( \frac{\lambda}{\lambda_{\rm ll } } \right ) \left ( \frac{\nu}{\nu_{\rm heii } } \right)^{-3(\beta-1 ) } , \label{eq : tau_lls}\ ] ] where @xmath220 is the comoving distance travelled by the photon , @xmath221 is the mean free path at the lyman limit and the term in the brackets apply when @xmath222 $ ] .
note here that @xmath223 ( with no subscript ) represents the gamma function . for @xmath224 ,
as assumed by @xcite in the relevant regime @xmath225 , we therefore obtain @xmath226 , or @xmath227 . on the other hand ,
the assumption of constant @xmath182 is not appropriate for models where @xmath182 turns over at much lower column densities @xmath228 .
for instance , in the modified version of the model presented by @xcite in which @xmath229 , we obtain @xmath230 , assuming @xmath231 .
accounting for the thermalisation between all species , this corresponds to a temperature jump of @xmath232k : less than the optically thick number of @xmath233k , but also significantly greater than the @xmath234 k associated with the @xcite model for @xmath182 .
these uncertainties ultimately have to be observationally resolved . to summarise : _
if _ high @xmath235 column density systems are abundant , then most high energy photons will be deposited there . however , the abundance of such systems particularly during the course of reionisation is far from clear , leading to considerable uncertainty in heating rates .
we will consider the effect of dense clumps on photo - heating further in [ sec : sims ] using our radiative transfer simulations .
the second key argument for the importance of filtered radiation is the fact that the spacing between quasars is much less than the spacing between lyman limit systems @xcite .
let us consider an updated version of this argument . at @xmath236
, quasars with @xmath237 dominate the photon budget ( @xcite ) .
these have a space density of @xmath238 , or a mean separation of @xmath239 mpc .
let us now reasonably suppose that these quasars can only directly ionise the igm out to distances where they will on average encounter a self - shielded lyman limit system . assuming the typical size of an absorber with overdensity @xmath157 is the local jeans length , then in ionisation equilibrium its column density is approximately given by ( @xcite ) @xmath240 where @xmath241
. the clump will then become self - shielded once @xmath242 .
this corresponds to a characteristic overdensity ( @xcite ) @xmath243 @xcite provide a prescription for estimating the separation between clumps of overdensity @xmath244 calibrated to numerical simulations at @xmath245@xmath246 . in their model ,
the separation between clumps of density @xmath247 is given by @xmath248^{-2/3},\ ] ] @xmath249 where @xmath250 is the fraction of volume with @xmath251 , and @xmath252 is a good fit to their simulations .
the second equality uses the fact that the @xcite density distribution rapidly approaches an isothermal profile at high densities , @xmath253 , and provides an excellent approximation to their expression at @xmath127 ( see the appendix of @xcite ) . setting @xmath254 and using eqs .
( [ eq : tion ] ) , ( [ eq : delta_i ] ) and ( [ eq : mfp ] ) , we thus find that a photon at the lyman edge can travel a comoving distance @xmath255 before it is absorbed by a lyman limit system .
we have ignored the weak dependence ( @xmath256 ) on temperature in this expression . since @xmath257
, it would appear that most of the igm is inevitably exposed to filtered , hardened radiation @xcite .
while such filtered radiation is undoubtedly weak and results in a low photo - ionisation rate ( see [ sec : analytic_temp ] ) , over long periods of time , it may still seem possible to achieve substantial heating . however , there are some caveats to this argument .
the mean separation between @xmath151 quasars is not the correct metric in this instance .
the region over which a quasar ionises is generally substantially less than @xmath258 . instead , must become reionised over many generations of quasar formation , since the lifetime of quasars , @xmath259 ( @xcite ) , is much less than the hubble time .
again for @xmath260 , a quasar can ionise a region of radius @xmath261 where @xmath262 is the fraction of photons absorbed within the region .
we have again taken the _ upper limit _ for the quasar lifetime to be conservative with regard to the maximum temperature boost . since @xmath263 , most of the photons that ionised the igm need not have undergone significant filtering , at least until @xmath264 .
note that this requirement becomes even stricter if the quasars have shorter lifetimes than @xmath130 . and @xmath265
; hereafter we shall ignore this difference .
] instead , most photons photo - ionise and heat the region directly around their source quasar .
significant photo - heating by strongly filtered radiation is thus more likely to occur towards the end of reionisation , once most of the is reionised .
even then , as discussed previously , much of this heating could potentially occur in the lyman limit systems themselves .
again , it appears difficult to reionise the entire igm _ exclusively _ with hard photons and hence achieve a large , global temperature boost .
, is shown in the case of no photon sinks ( dashed curve ) , recombinations as in eq .
( [ eq : q2 ] ) with @xmath266 ( dot - dashed curves ) , lyman limit systems as in eq .
( [ eq : q1 ] ) ( thick solid curve ) and both photon sink terms with @xmath267 ( thin solid curve ) . the fact that lyman limit systems only allow quasars to ionise a fixed volume , regardless of their previous ionisation history ( as represented by eq .
[ eq : q2 ] ) , does not incur a significant photon cost , and reionisation can still continue to completion.,scaledwidth=45.0% ] one worry with this argument might be that the photon budget for successive generations of quasars to reionise the igm is prohibitive .
if every quasar only ionises a fixed , local region of the igm , as implied above , then are there enough photons available to complete reionisation ? to be conservative
we will assume @xmath268 .
then , if the filling fraction of ionised regions is @xmath269 , on average a quasar will only be able to freshly reionise a new volume @xmath270 around it .
, for the relatively luminous quasars observed with the sloan digital sky survey at @xmath271 ( @xcite ) is comparable to @xmath272 , we expect this to be an order unity effect . ]
the rest of the photons are consumed by lyman limit systems in the existing ionised regions . by taking this photon
sink due to lyman limit systems into account ( and ignoring all other sinks ) , the filling factor evolves as @xmath273 where @xmath274 is the rate at which ionising photons are produced per helium atom .
note that this differs from the customary relation @xmath275 @xcite , where @xmath276 is the average recombination rate per atom in ionised regions which also includes the effective clumping factor , @xmath277 , of ionised gas .
instead , eq . ( [ eq : q1 ] ) ignores the effect of recombinations in the igm and only considers the lyman limit systems as a photon sink .
it mimics the effect that @xmath277 must increase rapidly as @xmath278 when dense pockets of gas must begin to be ionised @xcite ; quasar regions are sufficiently large that this could potentially happen at all stages of the reionisation process .
note also that since the recombination time in underdense regions of the igm is long , fossil cavities which were ionised by extinct quasars far outnumber active regions during reionisation ( @xcite , see also [ sec : fossil ] ) .
the ionising emissivity required to _ keep _ the igm ionised should thus be significantly less than that required to initially reionise it , and can be maintained by the active generation of quasars . in fig .
[ fig : filling_factor ] we show the evolution of @xmath279 according to these two equations , where we have computed the ionising emissivity using the @xcite luminosity function .
the result for no photon sinks is shown by the dashed curve , the dot - dashed curves include recombinations as in eq .
( [ eq : q2 ] ) with @xmath266 , the thick solid curve includes the sink due to lyman limit systems as in eq .
( [ eq : q1 ] ) ( but not recombinations ) , and the result including both photon sink terms ( recombinations and lyman limit systems ) with @xmath267 is indicated by the thin solid curve .
note that since @xmath280 is simply the number of ionising photons produced minus the photon sink terms , it can exceed unity once reionisation is complete , at least in the case of eq .
( [ eq : q2 ] ) .
the most important point to take away is that allowing quasars to only ionise a fixed volume , regardless of the previous ionisation history of their surroundings , does not incur a large cost in the photon budget and reionisation can still continue to completion .
most of the igm can therefore be reionised by a direct source of soft photons ; there are few isolated patches which have only been exposed to hard photons ( and hence which are heated significantly ) .
it is also worth noting that photo - ionisation timescales are then likely to be short . unlike hydrogen reionisation , the size of regions
are already controlled by lyman limit systems before overlap , since @xmath281 .
hence , the mean free path and thus the ionising radiation field can only increase mildly at best by a factor of 2 or 3 during percolation .
the strength of the radiation field after overlap is constrained by transmission along all lines - of - sight in the gunn peterson trough .
the results of @xcite suggest a fit to the effective optical depth of @xmath282^{3.5}$ ] for @xmath283 , or @xmath284 .
this ionising background implies ionisation timescales of @xmath285 .
this directly contradicts the possibility that most of the igm is pervaded by weak , filtered radiation ; instead , most or all of the igm must lie within the proximity zone of bright source(s ) which gives rise to a strong ionising radiation field .
such sources at a slightly earlier epoch would cause rapid ionisation .
thus far we have considered photo - heating in regions of the igm where is reionised only once .
however , @xcite have recently pointed out that fossil regions will be common during the epoch of reionisation ; less than 50 percent of the created by dead quasars will have enough time to recombine before another quasar appears .
the time lag between successive ionisations of a patch of the igm is @xmath286 at @xmath127 ( @xcite ) , based on the expected number density of @xmath287 quasars from @xcite .
could a second round of reionisation in a partially recombined fossil region then boost the igm temperature even further , leading to temperatures in excess of those otherwise expected from our previous arguments ?
the case - b recombination timescale , @xmath288 , can be written as @xmath289 whereas the time to replenish a given fraction , @xmath290 , is @xmath291 .
the fraction replenished in @xmath292 is thus @xmath293 for @xmath294 and @xmath295 .
note further from eq .
( [ eq : igmtemp ] ) that the temperature increase due to photo - heating is proportional to @xmath296 .
thus , since the recombination time in voids filled with is so long compared to @xmath297 , they will only experience a small fraction of the heating achieved during the first round of reionisation .
in contrast , clumps with @xmath298 have sufficient time to recombine nearly completely before undergoing a second phase of reionisation , potentially allowing the to be fully reheated for a second time .
however , we must also consider the cooling timescale to establish whether any _ additional _ temperature increase is actually achieved following a second reionisation . if @xmath299 , even if a clump in the fossil region has had sufficient time to fully recombine such that @xmath300 , the igm will have cooled more quickly and any subsequent photo - heating by a quasar can , at most , only restore the previous temperature in the region .
[ fig : fossil ] displays the recombination and cooling timescales as a function of temperature for gas of primordial composition at @xmath127 in photo - ionisation equilibrium with a uv background specified by @xmath301 where @xmath302 and @xmath303 .
the amplitude is consistent with measurements of the photo - ionisation rate at @xmath304 based on the observed forest opacity ( @xcite ) .
the solid curves show the radiative cooling timescale @xmath305 at three different igm densities .
the cooling rate is computed using the rates listed in @xcite , with the exception of the case - b recombination cooling rates which use the functions given by @xcite .
the dashed curves display the corresponding case - b recombination timescale .
the minimum in the radiative cooling timescale at @xmath306 results from excitation cooling , while the high temperature tail is due to the inverse compton scattering of free electrons off cosmic microwave background photons .
the horizontal dotted line shows the adiabatic cooling timescale for the low density gas which is still expanding with the hubble flow ( @xmath53 ) . for temperatures @xmath307 at @xmath294 , the recombination time exceeds the adiabatic cooling timescale .
hence , the igm temperature in the voids following a second reheating will in general be smaller than the temperature immediately following the initial reionisation , irrespective of the time lag between successive ionisations . in photo - ionisation equilibrium with the uv background given by eq .
( [ eq : uvb ] ) .
the solid curves show the radiative cooling timescale at three different densities , from top to bottom @xmath308 , while the dashed curves display the corresponding case - b recombination timescale as labelled on the figure .
the horizontal dotted line shows the adiabatic cooling timescale for low density gas still expanding with the hubble flow ( @xmath53).,scaledwidth=45.0% ] at higher densities , adiabatic expansion will no longer dominate the cooling timescale for clumps which have already separated from the hubble flow . in this instance , radiative cooling becomes dominant , and the cooling timescale falls below the recombination timescale for @xmath309 .
furthermore , the cooling timescale becomes comparable to the time lag between successive ionisations at @xmath127 for @xmath310 .
if @xmath311 , even high density clumps in the fossil region which can recombine efficiently are unlikely to receive a substantial _ additional _ temperature boost following a second reheating . note further that the radiative cooling timescale in overdense regions of the igm is likely to be even shorter than we have assumed here .
although the assumption of primordial composition is a reasonable approximation for voids in the igm , metals are ubiquitous in higher density regions ( _ e.g. _ @xcite ) and substantially increase the radiative cooling rate , depending on their abundance ( @xcite ) .
this would make any net temperature increase even more difficult to achieve . finally , at higher redshifts the time lag between successive ionisations for a given patch of the igm increases as the quasar number density drops ; @xcite estimate @xmath312 at @xmath313 .
on the other hand , the recombination and collisional excitation cooling timescales at fixed density decrease as @xmath314 .
hence , at higher redshifts the fossil region is even more likely to have sufficient time to recombine and cool before another quasar can reheat the igm .
we conclude that partially or fully recombined fossil regions which are reheated a second time are therefore unlikely to exhibit temperatures in excess of those expected following the initial phase of reionisation .
note however , heating in these regions is still likely to contribute significantly to the _ complexity _ of the relationship between temperature and density in the igm during reionisation ( see [ sec : trho ] ) .
in the previous sections we have employed analytical arguments to model the photo - heating of the igm during reionisation . however , detailed radiative transfer simulations ( ideally , with a full three - dimensional , multi - frequency approach ) are required to fully address this problem .
unfortunately , such simulations are computationally challenging , requiring very high spatial resolution within large simulation volumes in order to accurately track the topology of reionisation , as well as high frequency resolution to correctly follow the thermal state of the igm .
existing calculations are thus either one dimensional ( @xcite ) , neglect the modelling of multiple sources ( @xcite ) , or resort to approximation schemes which do not recover the reprocessing of the ionising radiation field precisely ( @xcite ) .
the latter is crucial for correctly modelling the igm thermal evolution .
high spatial resolution is also needed to resolve the overdense regions in the igm which play an important role in filtering the ionising radiation . furthermore ,
large , computationally expensive simulations allow only a limited parameter space to be explored .
we therefore explore the results discussed previously using an updated version of the one - dimensional , multi - frequency photon conserving algorithm described and tested in @xcite and @xcite .
this approach is based on the monochromatic photon - conserving algorithm originally developed by @xcite .
note that our simulations do not model reionisation using large volumes ( _ e.g. _ @xcite ) and will therefore not correctly capture any three dimensional effects such as long range photo - heating by multiple quasars .
instead , our models employ very high spatial and mass resolution and thus resolve the dense regions in the igm which are important for filtering the ionising radiation .
this high resolution is computationally prohibitive in larger , three dimensional simulations . the frequency integration in the updated code is now performed over the range @xmath315 , allowing the simulations to follow the radiative transfer of soft x - ray and uv photons .
photons with energies in excess of @xmath316 are relatively rare and do not contribute significantly to photo - heating in the igm over a typical quasar lifetime .
secondary ionisations are included following the prescription of @xcite , and the photo - ionisation cross - sections from @xcite are now used instead of the fits from @xcite .
our one - dimensional radiative transfer code does not follow the diffuse emission from recombination radiation , and so an option to include the on - the - spot approximation using the case - b recombination and cooling rates of @xcite has also been added . in this work
we shall always use the case - b values , although case - a would be the more appropriate choice in underdense , highly ionised regions of the igm .
however , in this work we only consider the reionisation of an igm with an initial fraction @xmath144 , and so case - b is the more appropriate choice . in order to simulate the propagation of ionising radiation through the inhomogeneous igm we use density distributions drawn from a high resolution hydrodynamical simulation run using the parallel tree - sph code gadget-2 ( @xcite ) .
the simulation volume is a periodic cube @xmath317 mpc in length containing @xmath318 gas and dark matter particles .
each gas particle has a mass of @xmath319 .
this mass resolution adequately resolves the forest at @xmath127 ( @xcite ) .
a friends - of - friends halo finding algorithm was used to identify the ten most massive haloes in the simulation volume at @xmath127 .
the haloes span a mass range of @xmath320 and lie above the minimum mass expected for quasars with @xmath321 at @xmath245@xmath322 ( @xcite ) .
density distributions were then extracted around these haloes and spliced with other lines - of - sight drawn randomly from the simulation volume at the same redshift following the procedure described in @xcite .
this resulted in @xmath323 different line - of - sight igm density distributions , all of which start at the location of one of the most massive haloes in the simulation volume .
each complete line - of - sight is @xmath324 mpc in length and is divided into @xmath325 equally spaced pixels in the radiative transfer calculation , giving a cell size of @xmath326 . at mean density at @xmath127 for @xmath144 , this corresponds to an optical depth per cell of @xmath327 at the ionisation edge .
these density distributions are used to run a total of @xmath328 separate line - of - sight radiative transfer simulations , divided into nine groups of @xmath323 . in all these simulations
, we shall assume the and in the igm is initially in photo - ionisation equilibrium with the uvb given by eq .
( [ eq : uvb ] ) .
the first group of simulations is our fiducial quasar model with @xmath108 , @xmath329 and @xmath330 .
five further groups are run by varying one of the quasar parameters in the model : specifically @xmath96 , @xmath95 , @xmath331 , @xmath332 and @xmath333 . in all of these models
we assume an isothermal igm with @xmath295 prior to reionisation , consistent with the measurements of @xcite .
the final three groups again use the fiducial quasar model , plus the two models with @xmath334 and @xmath20 .
however , instead of adopting an isothermal igm with @xmath295 prior to the quasar turning on , a power - law temperature - density relation is adopted ( _ e.g. _ @xcite ) . specifically , we set @xmath15 for @xmath335 , and @xmath336 for @xmath337 , with @xmath338 and @xmath339 , again consistent with the @xcite data .
this initial temperature distribution resembles the results from optically thin hydrodynamical simulations of the igm ( @xcite ) .
the results for six of the simulated lines - of - sight are shown in detail in figs .
[ fig : los1 ] and [ fig : los2 ] .
the temperature of the igm is displayed in the upper left panels , with the corresponding and fractions in the upper centre panels .
the igm density distributions along the lines - of - sight are displayed in the upper right panels .
the lower panels display , from left to right , the integrated optical depth along the lines - of - sight for photons with energies @xmath340 and @xmath341 , the photo - ionisation rates and the average excess energy per photo - ionisation .
the thin curves in the latter two panels correspond to the relevant quantities computed in the optically thin limit , while the filled circles correspond to the distance from the quasar where the optical depth at the ionisation threshold first exceeds unity .
the dotted horizontal line in the lower central panel indicates the inverse of the adopted quasar lifetime , @xmath75 .
a selection of numerical results from the six lines - of - sight are summarised in table [ tab : losdata ] .
c|c|c|c|c|c|c|c model & @xmath342 [ k ] & @xmath343 [ ev ] & @xmath114 [ mpc ] & @xmath157 & @xmath344 [ k ] & @xmath345 & @xmath346 + & ( @xmath347 ) & ( @xmath347 ) & ( @xmath348 ) & ( @xmath348 ) & ( @xmath348 ) & ( @xmath348 ) & ( @xmath348 ) + fiducial , fig .
[ fig : los1 ] & 6200 & 22.9 & 24.5 & 5.1 & 7900 & 0.66 & 1.21 + @xmath96 , fig .
[ fig : los1 ] & 10700 & 39.5 & 37.7 & 46.0 & 24200 & 0.25 & 2.14 + @xmath95 , fig .
[ fig : los1 ] & 4600 & 17.0 & 0.8 & 32.0 & 10900 & 0.94 & 1.03 + fiducial , fig .
[ fig : los2 ] & 5400 & 19.9 & 33.5 & 0.2 & 5200 & 0.65 & 1.22 + @xmath332 , fig .
[ fig : los2 ] & 6300 & 23.3 & 71.4 & 4.7 & 7400 & 0.47 & 1.42 + @xmath349 , fig .
[ fig : los2 ] & 21100 & 77.9 & 0.8 & 222.4 & 43100 & 0.09 & 3.16 + [ tab : losdata ] in each panel of fig .
[ fig : los1 ] the solid , dashed and dash - dotted curves correspond to the fiducial , soft and hard quasar models , respectively .
the temperature for all three models increases somewhat within @xmath350 of the quasar .
the large gas overdensities encountered approaching the centre of the haloes ( typically @xmath351 ) strongly filter the ionising radiation , exposing the nearby igm to very hard , intense ionising radiation for a short period of time .
the analytical estimates presented in fig . [ fig : mfp ] are consistent with such temperature boosts occurring close to a quasar .
however , once these regions have been fully ionised , this filtering disappears , so it is not important for the distant igm . outside of this region ,
the igm temperatures for the fiducial and soft quasar models are fairly modest , with maximum temperatures of @xmath352 and @xmath353 occurring at @xmath354 and @xmath355 , respectively .
the temperature in the soft model corresponds to the dense clump located around @xmath356 from the quasar .
this is consistent with many of the hardest photons being deposited in lyman limit systems , as argued in detail in
[ peng_section ] . in the hard quasar model
a dramatic boost in the igm temperature is apparent .
the maximum temperature is @xmath357 at @xmath358 , again corresponding to the overdense structure with @xmath359 shown in the upper right panel .
the void which appears immediately down - stream from this clump is irradiated by hard , filtered yet intense ionising radiation which photo - heats the gas to a temperature substantially higher than the mildly overdense regions closer to the quasar .
increased temperatures in the underdense regions in the igm may be required to reconcile the observed flux distribution of the forest with simulations ( @xcite ) .
note also that the presence of a second highly overdense clump at @xmath360 , which has yet to be fully ionised by the quasar , further attenuates and hardens the radiation .
however , photo - heating by this very hard but weak radiation is now less effective beyond this clump , where @xmath361 . in general the temperatures within the ionised zone
all increase with distance from the quasar ; regions further from the quasar are more likely to be ionised by hard photons with long mean free paths ( @xcite ) .
the temperature then falls again beyond the ionisation front , where the ionisation rate declines .
the average temperatures of the reionised igm ( @xmath347 ) are @xmath362 for @xmath363 . for an initial temperature of @xmath295
, this implies @xmath364 , again consistent with the simple analytical estimates for @xmath87 in fig .
[ fig : mfp ] . from eq .
( [ eq : igmtemp ] ) , these temperature boosts imply @xmath365 , corresponding to @xmath366 times @xmath343 evaluated in the optically thin limit . excluding the biased regions within @xmath367 of the quasar
yields similar values ; @xmath368 .
thus on average , most of the photo - heating over the quasar lifetime is achieved by the moderately filtered radiation close to the leading edge of the ionisation front . for comparison to other estimates ,
when neglecting re - emission from the igm the updated calculations of @xcite indicate that @xmath369 for @xmath370 within a quasar dominated uv background model ( f. haardt , private communication ) .
this is in good agreement with our results .
note , however , that including recombination emission lowers the average energy per photo - ionisation , such that @xmath371 .
this suggests that significant recombination emission may lower the expected igm temperature boost even further by increasing the number of soft ionising photons . in fig .
[ fig : los2 ] , the results for the fiducial quasar model along a different line - of - sight are compared to two further models with @xmath349 ( dashed curves ) and @xmath372 ( dot - dashed curves ) .
the igm temperature again increases within @xmath350 of the quasar in all models .
however , in the case of the @xmath349 quasar , there is a significant temperature boost over a region @xmath373 . at first this
may seem counter - intuitive ; from the argument presented in
[ sec : toymodel ] , one might naively expect a faint ionising source to produce less heating due to longer photo - ionisation timescales .
however , in this instance , the faint quasar has not had enough time to overcome the recombinations in its host halo , and as a result the radiation has been persistently filtered and hardened @xmath374 .
this effect is relatively common in our simulated lines - of - sight for @xmath331 .
note , however , this effect will be less prominent or even absent if all the within @xmath350 of the quasar host halo has been previously photo - ionised or is instead shock - heated and hence collisionally ionised . in comparison to the line - of - sight in fig .
[ fig : los1 ] , the igm density distribution in fig .
[ fig : los2 ] is also less dense on average .
this results in slightly less spectral hardening close to the quasar . as a consequence , the fiducial model in fig .
[ fig : los2 ] exhibits lower temperatures , with a maximum of @xmath375 at @xmath376 , while the @xmath332 model has a maximum of @xmath352 at @xmath377 .
the average temperatures in regions with @xmath347 are @xmath378 for @xmath379 . as before
, this implies that the average excess energy per photo - ionisation is @xmath380 , corresponding to @xmath381 times @xmath343 evaluated in the optically thin limit .
again , the first two values are consistent with most of the photo - heating being due to only modestly hardened radiation , whereas the latter model results in very large temperature boosts close to the quasar . excluding the biased regions with @xmath367 , the average temperatures are instead @xmath382 for @xmath383
; there are no regions with @xmath347 at @xmath384 in the @xmath349 line - of - sight .
we note briefly that observations of the and forest are another useful diagnostic of reionisation .
large fluctuations are observed in the measured column density ratio , @xmath385 ( @xcite ) .
the average value of @xmath182 also increases towards higher redshifts .
direct measurements of the softness parameter , @xmath386 , give @xmath387 at @xmath388 ( @xcite ) .
the large error bars are due to the very uncertain effective optical depth measured from the forest data .
our simulations are not able to follow @xmath182-fluctuations during reionisation in detail , since this requires a detailed treatment of the spatial distribution of ionising sources as well as radiative transfer ( see _ e.g. _ @xcite ) .
however , we may still estimate the average softness parameter in our simulations along the line - of - sight from a single quasar . in regions with @xmath347 , this yields @xmath389 for the lines - of - sight with @xmath390 in fig .
[ fig : los1 ] and @xmath391 for @xmath392 in fig .
[ fig : los2 ] .
these values are broadly consistent with the observational constraints , aside from the model with @xmath334 which produces a spectrum which is too hard , indicating that this model may be somewhat extreme .
we have only shown the details for six of our line - of - sight radiative transfer calculations here , although these are quite typical examples from our large sample .
the results from all the simulations will in any case be summarised in the next section .
it is nevertheless clear from these examples that variations in quasar luminosities , lifetimes , and intrinsic spectra , as well as the intervening igm density distribution , will result in a great deal of variation in the thermal and ionisation state of the igm on small scales .
such detailed behaviour can not be adequately captured by analytical or semi - analytical arguments .
they do , however , confirm that our general expectations hold .
for typical quasar luminosities and spectra at @xmath6 , temperature boosts during reionisation in excess of @xmath149 are unlikely to occur rapidly ( @xmath393 ) over the entire igm .
although large temperatures are indeed possible , these are most likely to occur either around very hard sources , or in highly overdense regions of the igm .
furthermore , the hardest photons tend to be preferentially deposited in the densest clumps in the simulations , as is evidenced by the larger temperatures associated with these regions .
most of the low density igm is reionised by the more numerous soft photons , leading to temperature boosts in the range of @xmath394 for @xmath329 .
note that in all instances , there is no strong photo - heating wherever @xmath395 .
there is some recent evidence to suggest that the igm temperature - density relation may be substantially more complex than is usually assumed around @xmath107 .
@xcite and @xcite have independently noted that the flux distribution of the forest is well reproduced by a model where the voids in the igm are significantly hotter than expected , perhaps due to radiative transfer effects during reionisation which invert ( @xmath17 ) the igm temperature - density relation .
complementary to this , @xcite pointed out that the inhomogeneous distribution of ionising sources will also complicate the igm temperature - density relation ; voids are generally photo - ionised and heated last and are thus hotter towards the end stages of reionisation .
@xcite also demonstrated that the inhomogeneous nature of reionisation will blur the igm temperature - density relation .
we can not address the three dimensional effects discussed in these latter studies with our radiative transfer simulations .
however , we can quantify the impact filtering by the igm will have on the igm temperature - density relation .
contour plots of the temperature - density planes for all nine sets of our radiative transfer simulations are displayed in fig .
[ fig : trho ] .
each panel shows the results for @xmath323 different lines - of - sight . in all panels ,
the annotated arrows indicate the temperature of the igm for all gas at mean density with @xmath396 ( upper arrows ) and @xmath397 ( lower arrows ) .
the dotted lines represent the initial temperature - density relation adopted in the models .
regions with @xmath398 are associated with which has not yet been significantly photo - ionised and heated . the upper row in fig .
[ fig : trho ] displays , from left to right , the temperature - density plane for the fiducial quasar model , the hard quasar model ( @xmath96 ) and the soft model ( @xmath95 ) .
again note the quasar models are normalised to have the same rest frame b - band luminosity , such that @xmath399 and @xmath400 .
the largest temperatures are associated with the hard quasar model , where additionally the most has been photo - ionised and heated .
the middle row , from left to right , displays the models with @xmath372 , @xmath349 and @xmath333 .
note that the @xmath349 temperature - density plane exhibits the highest temperatures around mean density .
this is due to photo - heating of the close to the quasar by intense but highly filtered radiation , as discussed in the previous section .
lastly , the lower row displays the same quasar models shown in the upper row , but now the igm initially follows a tight power - law temperature density relation , @xmath15 for @xmath335 and @xmath336 for @xmath337 , where @xmath338 and @xmath339 .
the temperature - density relation in an optically thin igm is expected to slowly asymptote to a power - law with @xmath401 following reionisation ( @xcite ) .
this changes the _ absolute _ temperature of the underdense material in the igm following reionisation , although the temperature boost at all densities remains similar to the isothermal case .
thus , the igm temperature - density relation prior to reionisation clearly impacts its post - reionisation form , although again note that the heat input into the igm remains consistent with the analytical arguments presented earlier . c|c|c|c|c|c model & @xmath342 [ k ] & @xmath342 [ k ] & @xmath342 [ k ] & @xmath342 [ k ] & @xmath402 [ k ] + & ( @xmath403 ) & ( @xmath404 ) & ( @xmath405 ) & ( @xmath406 ) & + fiducial & 5300 & 5900 & 9400 & 21300 & 4200 + @xmath96 & 10400 & 10600 & 14900 & 30100 & 5900 + @xmath95 & 3500 & 3700 & 4600 & 16100 & 3300 + @xmath332 & 6600 & 6800 & 9500 & 21400 & 4200 + @xmath349 & 5300 & 7500 & 15700 & 26700 & 4200 + @xmath407 & 5300 & 6200 & 9100 & 14400 & 4200 + fiducial ( non - isothermal ) & 5900 & 5900 & 7200 & 15600 & 4200 + @xmath96 ( non - isothermal ) & 11000 & 10600 & 12800 & 24800 & 5900 + @xmath95 ( non - isothermal ) & 4000 & 3600 & 2000 & 9400 & 3300 + [ tab : trhodata ]
the average temperatures measured at four different densities , @xmath408 , in each of the models are summarised in table [ tab : trhodata ] .
the largest temperature boosts occur in the densest regions in the simulations and are substantially higher than those expected in the optically thin limit as discussed in [ peng_section ] .
this suggests the hardest photons are preferentially deposited in these regions over the lifetime of the quasars .
however , at mean density and below , the temperature boosts are much closer to the optically thin values , and only modest temperature boosts are achieved . in all panels there is also considerable scatter in the temperature - density plane .
this is very different from the tight , power - law temperature - density relation expected for an optically thin igm ( _ e.g. _ @xcite ) . a similar result was noted by @xcite following reionisation , who found that the temperature - density relation may even be inverted ( @xmath17 ) .
however , this was for a single line - of - sight only and thus not necessarily representative of the igm as a whole .
@xcite also found that radiative transfer effects can alter the temperature - density relation of the igm , and will furthermore depend on the types of sources responsible for reionisation . on the other hand , using three dimensional radiative transfer simulations of reionisation , @xcite found the igm temperature - density relation is not inverted , although the igm temperature is still increased by around a factor of two at mean density .
however , the necessarily approximate frequency integration scheme they use in order to run such a detailed simulation may not model the reprocessing of the ionising radiation field precisely .
it is nevertheless clear from fig .
[ fig : trho ] that although voids in the igm are significantly heated during reionisation , a well - defined , inverted temperature - density relation appears difficult to achieve with line - of - sight radiative transfer alone ( see also @xcite ) .
empirical interpretations of the forest data seem to indicate such an inversion , but ( as noted by @xcite ) the mild inversion apparently required in such studies may simply mimic a more complex relationship between temperature and density in the igm .
following reionisation , the temperature scatter in each model increases considerably , and over the entire igm it may be larger still .
differences in quasar properties and the local density distribution of the igm , as well as the thermal state of the igm immediately prior to reionisation , could enhance the overall scatter .
the simulations also suggest the extent to which reionisation has progressed will have an important effect on the igm temperature - density relation .
prior to the completion of reionisation , a two - phase medium with hot regions and cooler regions of should develop , leading to large fluctuations in the igm temperature .
note that a correct treatment of the timing of reionisation at different densities is also required to model this properly ( @xcite ) , and the reionisation of partially recombined fossil regions will complicate the temperature - density relation yet further .
on the other hand , studies using wavelets ( @xcite ) have failed to detect spatial fluctuations in the igm temperature on small scales at @xmath9 . however , these studies are implicitly predicated on fairly large ( around a factor of two ) temperature boosts ; more modest boosts over small scales may be hard to extract from the data .
in this work we have performed a detailed analysis of the photo - heating of the igm during reionisation using a combination of analytical arguments and numerical radiative transfer models .
our aim was to critically examine the physical processes involved in modelling the fate of hard , ionising photons in the igm and their impact on the thermal evolution of the igm .
the main results of this work are as follows . 1
. filtering through the igm weakens as well as hardens ionising radiation ( @xcite ) .
however , significant photo - heating of the igm during reionisation can only occur over timescales comparable to or shorter than the local photo - ionisation timescale , @xmath409 , irrespective of the mean excess energy per photo - ionisation , @xmath343 .
consequently , this limits the temperature boost , @xmath87 , attainable over a given timescale within a fixed volume of the igm . for an @xmath151 quasar at @xmath127 with a typical euv spectra index of @xmath80 and a lifetime @xmath410
, @xmath87 will not exceed @xmath411 at distances @xmath150 from the quasar , and will furthermore only do so at @xmath412 if the softer photons near the ionisation edge do not contribute towards photo - ionisation .
for an average separation between @xmath151 quasars of @xmath152 , this implies a global temperature boost of @xmath413 over @xmath414 is difficult to achieve unless quasars with much harder euv spectra are common .
hard photons are preferentially absorbed in the densest regions of the igm where the largest temperature boosts will thus occur .
however , current constraints on the igm temperature based on observations of the low column density forest ( @xcite ) are mainly sensitive to densities close to the cosmic mean or below ( @xcite ) .
the fate of hard photons and expected heating rates thus depends on the relative abundance of high column density @xmath235 absorbers : if most of the heat is deposited in dense systems , we have no way of inferring this from the forest .
unfortunately , the abundance of such systems is rather model - dependent , and varies with the unknown radiation field @xmath189 during reionisation and the correspondence between physical density @xmath187 and observed column densities @xmath415 .
this translates into considerable uncertainties in the photo - heating rates and the resulting igm temperature boost .
a plausible range of values , given model uncertainties , is @xmath416k , somewhat intermediate between the optical thin and thick values of @xmath417k and @xmath418k respectively , for a @xmath419 unprocessed source spectrum .
even though the mean separation between lyman limit systems is comparable to the expected size of regions around @xmath151 quasars , most of the igm could potentially still be ionised primarily by soft photons , at least if quasars are randomly distributed .
if this were the case , the average amount of heating over the entire igm would be relatively modest .
in general , only unusual regions that are far from the ionising sources will be exposed to a highly - filtered spectrum for a long enough period to heat substantially .
this can only happen toward the end of reionisation , because the photo - ionisation timescale is so long for such heavily filtered spectra .
the reheating of partially recombined fossil regions is unlikely to substantially increase the igm temperature beyond that achieved during the first round of reionisation . in general ,
recombination timescales in these fossil regions typically exceed cooling timescales and the average time lag between ionisations by successive generations of quasars .
the net heating is therefore not increased .
we run a detailed set of @xmath328 line - of - sight radiative transfer simulations using a variety of initial conditions to confirm these analytic predictions .
we further demonstrate that the filtering of ionising radiation through the igm will produce a rich thermal structure .
this is likely to be further complicated by the inhomogeneous distribution of the ionising sources and the typically short lifetimes of quasars ( @xcite ) .
this complex , multi - valued temperature - density relation may help to resolve discrepancies between optically thin hydrodynamical simulations of the forest and recent detailed observations ( @xcite ) how do these conclusions compare to the current observational constraints on the igm thermal state at @xmath127 ?
in the case of the magnitude of the igm temperature boost , measurements of the doppler widths of absorption lines in the forest currently offer the best constraints .
the data of @xcite are consistent with a relatively sharp ( @xmath420 ) , large ( @xmath421 ) temperature jump occurring around @xmath422 at densities around the cosmic mean .
however , the error bars on these measurements are still large , and a more gradual , moderate temperature boost , as suggested by this work , is also consistent with the data .
similarly , the results of @xcite are also consistent with a significant and sudden temperature boost at @xmath6 , although large statistical errors again leave a more gentle thermal evolution possible . on the other hand
, @xcite find no evidence for a rapid boost in the igm temperature at @xmath6 .
their constraints , corrected to correspond to measurements at mean density , at most allow only for a modest evolution in the igm temperature ( @xmath423 between @xmath304 ) , and are also consistent with no temperature evolution at all .
a detailed analysis of a large set of high resolution forest spectra , perhaps using alternative temperature sensitive statistics , is necessary to obtain improved constraints on the thermal state of the igm at these redshifts .
another intriguing result which may indirectly favour a rapid heat input scenario , somewhat at odds with the results presented here , is the evidence for a sharp dip ( @xmath424 ) of around @xmath425 per cent in the effective optical depth of the forest at @xmath426 ( @xcite )
. @xcite used hydrodynamical simulations to interpret this feature as being due to a sudden increase in the igm temperature at @xmath427 , such that @xmath428 . however
, our results suggest that it is unlikely this transition could occur globally over such a short period .
alternatively , this feature may instead be due to a rapid change in the hydrogen photo - ionisation rate at @xmath426 , although again there is no current evidence to suggest this may be the case . further investigation into this feature using detailed hydrodynamical simulations of the igm is thus desirable . our results therefore suggest a general picture for the thermal evolution of the igm during reionisation by quasars where : ( i ) the temperature boost during reionisation is significantly uncertain , and could be considerably less than has been commonly assumed in the optically thick heating limit , ( ii ) any such temperature boost must be achieved over a timescale longer than @xmath429 and ( iii ) the resulting temperature - density relation in the igm will be much more complex than the tight , power - law temperature density relation expected in an optically thin igm . finally , we note that these results will not apply if the in the igm is predominantly reionised by sources other than quasars .
if the in the igm is reionised early on by stellar sources , high redshift x - rays , thermal emission from shocked gas or some other , as yet unidentified source , then the thermal evolution of the igm during reionisation may be rather different .
we thank george becker , adam lidz , martin haehnelt and matteo viel for helpful discussions during the course of this work , and the anonymous referee for a helpful report which improved this manuscript .
we also thank francesco haardt for sharing his numerical results with us , and we are especially grateful to matthew mcquinn for drawing our attention to several errors in the original manuscript . the hydrodynamical simulation used in this work
was run using the sgi altix 4700 supercomputer cosmos at the department of applied mathematics and theoretical physics in cambridge .
cosmos is a uk - ccc facility which is sponsored by sgi , intel , hefce and stfc .
this research was also supported in part by the national science foundation under grant no .
phy05 - 51164 ( jsb , through the mpa / kitp postdoctoral exchange programme ) , ast-0407084 ( spo ) and ast-0829737 ( srf ) , and nasa grant nng06gh95 g ( spo ) .
jsb thanks the staff at the kavli institute for theoretical physics , santa barbara , for their hospitality during the early stages of this work . | we use a combination of analytic and numerical arguments to consider the impact of quasar photo - heating during reionisation on the thermal evolution of the intergalactic medium ( igm ) .
we demonstrate that rapid ( @xmath0@xmath1 ) , strong ( @xmath2 ) photo - heating is difficult to achieve across the entire igm unless quasar spectra are significantly harder than implied by current observational constraints .
although filtering of intrinsic quasar radiation through dense regions in the igm does increase the mean excess energy per photo - ionisation , it also weakens the radiation intensity and lowers the photo - ionisation rate , preventing rapid heating over time intervals shorter than the local photo - ionisation timescale .
moreover , the hard photons responsible for the strongest heating are more likely to deposit their energy inside dense clumps , which cool rapidly and are furthermore invisible to most observational probes of the igm temperature .
the abundance of such clumps is , however , uncertain and model - dependent , leading to a fairly large uncertainty in the photo - heating rates .
nevertheless , although some of the igm may be exposed to a hardened and weakened ionising background for long periods , most of the igm must instead be reionised by the more abundant , softer photons and with accordingly modest heating rates ( @xmath3 ) , although localised patches of much higher temperature are still possible .
the repeated ionisation of fossil quasar regions does not increase the net heating because the recombination times in these regions typically exceed the igm cooling times and the average time lag between successive rounds of quasar activity .
detailed line - of - sight radiative transfer simulations confirm these expectations and predict a rich thermal structure in the igm during reionisation .
the resulting complex relationship between temperature and density may help resolve discrepancies between optically thin simulations of the forest and recent observations .
[ firstpage ] radiative transfer - methods : numerical - intergalactic medium - quasars : absorption lines - cosmology : theory . |
exactly solvable one - dimensional quantum models are widely used as proving grounds for key ideas in condensed matter physics and the theory of critical phenomena , since their conceptual simplicity often makes it possible to derive exact analytical expressions for the relevant physical quantities .
historically , most of the work in this field has been focused on systems with short - range interactions , like the well - known heisenberg and ( quantum ) ising chains . in the last few years , however , it has become feasible to realize in the laboratory quantum spin chains featuring various types of long - range interactions through different experiments involving , e.g. , optical lattices of ultracold rydberg atoms and trapped ions , or neutral atoms in optical cavities @xcite .
in particular , with the help of hyperfine `` clock '' states of trapped @xmath5 ions it is now possible to simulate quantum spin chains in which the coupling @xmath6 between the @xmath7-th and @xmath8-th sites is inversely proportional to a power @xmath9 of their distance @xcite .
an important model of this type is the integrable haldane
shastry ( hs ) chain @xcite , whose sites are the equispaced points @xmath10 ( @xmath11 ) on the unit circle with a coupling proportional to @xmath12 . in fact , this chain is a limiting case of a more general model due to inozemtsev , in which the coupling @xmath6 is an elliptic function of the difference @xmath13 with real period @xmath14 @xcite .
although the particles in the original hs chain carried spin @xmath15 , the model was shortly generalized to @xmath16 spin without losing its remarkable integrability properties @xcite . as a matter of fact
, the @xmath17 supersymmetric version of the hs chain , originally introduced by haldane @xcite , has also been studied in the literature @xcite .
of particular interest is the @xmath18 hs chain ( with @xmath19 ) , since it is essentially equivalent to an @xmath16 supersymmetric @xmath20-@xmath21 model @xcite with exchange and transfer energies proportional to @xmath12 .
this chain , first introduced by kuramoto and yokoyama in the @xmath22 case @xcite , is an exactly solvable model which provides one of the simplest realizations of spin - charge separation . in this work
we introduce a wide class of @xmath0 supersymmetric spin chains with general translation - invariant couplings @xmath23 and a chemical potential term . for zero chemical potential ,
these models include in particular the supersymmetric elliptic chain studied in ref .
@xcite and its two limiting cases , namely the @xmath0 hs chain and the xx model .
the class of models under study are technically simpler than their @xmath18 counterparts , essentially due to the fact that they can be transformed into a system of free spinless fermions in a straightforward way .
however , they still exhibit a sufficiently rich structure which makes it possible to examine a number of key properties in the theory of quantum critical systems in an analytic fashion .
more precisely , our main objective is to study whether the models under consideration are quantum critical for suitable values of the chemical potential , and to determine their corresponding central charge . as is well known ,
a characteristic feature of @xmath1-dimensional cfts is the fact that at low temperature @xmath24 their free energy per unit length is approximately given ( in appropriate units ) by @xmath25 , where @xmath26 is a constant and @xmath27 is the effective speed of `` sound '' @xcite . since the low temperature behavior of @xmath28
is determined by the low - lying states of the theory , this should also be the case for any one - dimensional quantum system whose low energy spectrum is described by a @xmath1-dimensional cft . in particular , the determination of the low temperature behavior of the free energy of a one - dimensional critical model provides an efficient way of determining the central charge of its underlying cft . in this way we have been able to show that if the dispersion relation @xmath4 is monotonic in the range @xmath29 $ ] the models under study are critical when the chemical potential @xmath30 belongs to the open interval @xmath3 , with central charge @xmath2 . as further confirmation of this result
, we have studied the ground state entanglement entropy , i.e. , the entropy of the reduced density matrix of a block of @xmath31 consecutive spins when the whole chain is in its ground state .
indeed , it is well known that in a @xmath1-dimensional cft the rnyi and von neumann entanglement entropies scale as @xmath32 and @xmath33 , respectively , where @xmath34 is the central charge and @xmath35 is the rnyi parameter @xcite .
thus the entanglement entropy of a quantum critical one - dimensional system should be proportional to @xmath36 for @xmath37 , where the proportionality constant fixes the central charge of the underlying cft .
again , we have verified that when the chemical potential belongs to the open critical interval @xmath3 the entanglement entropy of the models under consideration scales as that of a @xmath1-dimensional cft with @xmath2 .
we have also examined the behavior of the entanglement entropy and the zero - temperature fermion density as @xmath30 approaches the endpoints of the critical interval , showing that it is consistent with a quantum phase transition at both ends . for the @xmath0 chain with elliptic interactions
we have studied numerically the fermion density at finite temperature , finding that its behavior is far more complex when the chemical potential lies in the critical interval .
more precisely , for suitable values of @xmath30 inside this interval the fermion density is not a monotonic function of the temperature , but can rather present up to two extrema .
the paper is organized as follows . in section [ sec.model ]
we introduce the class of supersymmetric spin chains with which this work is concerned , and recall how these models can be fermionized using the algebraic properties of the @xmath0 permutation operator . in section [ sec.tdc ] we analyze the thermodynamics of the general @xmath0 chain when the dispersion relation is monotonic in the range @xmath38 , showing that at low temperature it behaves as a @xmath1-dimensional cft with @xmath2 . in section [ sec.gsee ]
we outline the computation of the von neumann and rnyi ground state entanglement entropies of the latter models in terms of the eigenvalues of the ground state correlation matrix .
section [ sec.asymp ] is devoted to deriving asymptotic formulas for these entropies , both when the size of the block of spins tends to infinity and when the chemical potential approaches the endpoints of the critical interval . in section [ sec.fd ] we perform a numerical study of the fermion density of the elliptic @xmath0 chain at finite temperature , and determine analytically its low temperature behavior for arbitrary interactions .
finally , in section [ sec.sumout ] we summarize our conclusions and discuss some future developments suggested by the present work .
consider a translation - invariant ( closed ) spin chain whose @xmath14 sites are occupied by either a boson or a ( spinless ) fermion .
if we denote by @xmath39 and @xmath40 the operators that respectively create a boson or a fermion at the @xmath7-th site , the hilbert space of the model is the @xmath41-dimensional subspace of the infinite - dimensional fock space determined by the constraints @xmath42 we shall take as the model s hamiltonian the operator to @xmath14 unless otherwise stated . ]
@xmath43 where @xmath44 , @xmath45 is the total fermion number operator , @xmath46 is a nonnegative smooth function and @xmath47 is the @xmath0 spin permutation operator @xcite defined by @xmath48 if we denote by @xmath49 and @xmath50 respectively the states occupied by a boson or a fermion , the action of the operator @xmath47 on the canonical spin basis with elements @xmath51 , with @xmath52 , is given by @xmath53 where @xmath54 if @xmath55 while for @xmath56 @xmath57 equals the number of fermions at the sites @xmath58 . note that @xmath47 is invariant under the supersymmetry transformation @xmath59 , so that the term @xmath60 in @xmath61 is @xmath0 supersymmetric , while the last term @xmath62 , i.e , the chemical potential of the fermions , transforms into @xmath63 due to the constraints .
furthermore , we shall exclusively be concerned in this paper with _ closed _
( i.e. , periodic ) chains , for which @xmath64 .
it is customary to extend the function @xmath46 to the whole real line as an @xmath14-periodic function , so that @xmath65 it was shown in ref
. @xcite that any chain of the form can be recast into a model of spinless hopping fermions by identifying the boson state @xmath49 with the fermion vacuum .
more precisely , we define a new set of fermion creation operators @xmath66 , @xmath67 , which indeed satisfy the canonical anticommutation relations ( car ) on account of . for instance
, we have @xmath68 the chain sites can now be either empty ( i.e. , in the state @xmath49 ) or occupied by a fermion ( in the state @xmath50 ) , and thus the hilbert space is the whole @xmath41-dimensional fock space built acting on the vacuum @xmath69 with the operators @xmath70 . as first shown by haldane @xcite , from eqs . and the constraints it follows that the @xmath0 exchange operator @xmath47 admits the following simple expression in terms of the new fermion operators @xmath71 : @xmath72 likewise , @xmath73 ( since @xmath74 is idempotent ) , so that @xmath75 is simply the chemical potential for the new fermions .
taking into account the latter identities , the hamiltonian can be rewritten as @xmath76 where we have set @xmath77 ( see ref .
@xcite for more details ) .
this hamiltonian describes a system of @xmath14 hopping ( spinless ) free fermions on a circle , with hopping amplitude between the @xmath7-th and @xmath8-th sites given by @xmath78 and chemical potential @xmath30 .
the translation invariance of this model ( encoded in the periodicity of the function @xmath79 ) suggests introducing the fourier - transformed operators @xmath80 it can be readily shown that these operators satisfy the car , and can therefore be considered as a new set of fermionic operators ; in fact , as we shall see below , @xmath81 creates a fermion with momentum @xmath82 ( @xmath83 ) .
it is shown in ref .
@xcite that @xmath61 is diagonal when written in terms of the new operators @xmath84 and their adjoints . in fact , we have @xmath85 where @xmath86h_n(j)\,.\ ] ] likewise , the system s total momentum operator @xmath87 is given by @xmath88 which shows that the operator @xmath81 creates a fermion with momentum @xmath89 ( @xmath83 ) .
in this work we shall be concerned with systems for which @xmath90 depends on @xmath91 and @xmath14 only through the corresponding momentum @xmath89 , i.e. , @xmath92 where the _ dispersion relation _ @xmath93 is a smooth function defined in the interval @xmath94 $ ] .
it easily follows from eq . that if such a function @xmath93 exists it is necessarily unique , and that @xmath95 .
an important type of interaction @xmath96 satisfying the above requirement is given by the elliptic function @xmath97 where @xmath98 is a real parameter , @xmath99 and @xmath100 , @xmath101 and @xmath102 denoting respectively the weierstrass elliptic and zeta functions with half - periods @xmath103 and @xmath104 @xcite .
it can be shown @xcite that the function satisfies the three conditions in eq . .
moreover , since @xmath105 the model with interaction strength smoothly interpolates between the heisenberg ( for @xmath106 ) and haldane shastry ( for @xmath107 ) @xmath0 chains ( with a chemical potential term added ) .
in fact , the former of these models can be transformed into the spin @xmath15 ( closed ) xx heisenberg hamiltonian @xmath108 where @xmath109 is the @xmath110-th pauli matrix acting on the @xmath111-th site and @xmath112 , with the help of the standard wigner
jordan transformation @xcite @xmath113 the dispersion relation @xmath4 for the elliptic interaction was computed in closed form in ref .
more precisely , from eq .
( 2.21b ) in the latter reference and the homogeneity properties of the weierstrass functions we have @xmath114,\ ] ] where @xmath115 in particular , we see that in this case the dispersion relation is a pure @xmath116-periodic that @xmath4 is not an elliptic function .
] function , independent of the number of particles @xmath14 . taking the @xmath117 and @xmath118 limits in the above equation for @xmath4 one recovers the well - known dispersion relations of the xx model @xcite and the @xmath0 haldane shastry chain , namely @xmath119
in this section we shall exploit the equivalence of the @xmath0 supersymmetric chain to the free fermion model to analyze the critical behavior of this chain as a function of the chemical potential @xmath30 . to this end
, we first need to determine the ground state of the model , which is straightforward from eq . .
indeed , it is obvious from the latter equation that the modes excited in the ground state are precisely those whose momenta @xmath82 satisfy the condition @xmath120 , so that the ground state is nondegenerate . strictly speaking ,
this is only true if we assume that @xmath121 for @xmath122 . indeed , if @xmath123 the mode with momentum @xmath89 ( and @xmath124 , if @xmath125 and @xmath126 ) can be either present or absent in the ground state , which is therefore degenerate .
since we shall be exclusively concerned with the thermodynamic limit @xmath127 , from now on we shall implicitly assume without loss of generality that @xmath121 for @xmath128 .
we shall also assume in what follows that the dispersion relation has a positive derivative in the interval @xmath129 , so that it is monotonically increasing in the latter interval and reaches its maximum at @xmath130 .
this is `` generically ''
true , and it certainly holds for the dispersion relation of the elliptic interaction and , in particular , for the xx model and the @xmath0 haldane shastry chains . if this is the case , it is straightforward to show that _ the model is gapless for @xmath131$]_. as a function of the mode number @xmath122 ( the range of modes excited in the ground state for the given @xmath132 has been represented by a thick red line ) . ] indeed , first of all it is clear that the system is gapped for @xmath133 or @xmath134 .
for instance , for @xmath133 the gap between the first excited state @xmath135 and the ground state is @xmath136 , which remains positive as @xmath127 .
similarly , when @xmath134 the gap is approximately equal to @xmath137 independently of @xmath14 .
suppose , on the other hand , that @xmath138 , and let @xmath139 be the root of the equation @xmath140 in the interval @xmath141 $ ] , which exists and is unique on account of the monotonicity of @xmath93 in the interval @xmath29 $ ] .
the modes excited in the ground state are now those with @xmath142 and @xmath143 , where @xmath144 denotes the integer part of @xmath139 ( cf .
[ fig.disp ] ) . thus if @xmath138 the gap between the first excited state and the ground state , given by @xmath145 is @xmath146 , since @xmath147 .
thus @xmath148 tends to zero as @xmath127 and the system is gapless , as claimed .
( in fact , when @xmath139 is an integer the modes with @xmath149 or @xmath150 may or may not be present in the ground state , but this does not affect the ground state energy and therefore the foregoing argument . ) we shall next show that when the chemical potential @xmath30 belongs to the open interval @xmath3 the @xmath0 chain is indeed critical , or , more precisely , that at low energies its spectrum is that of a @xmath1-dimensional cft with one free boson . to begin with , we note that when @xmath151 the low energy excitations of the chain are linear in the excitation momentum .
indeed , let @xmath152 denote the fermi momentum , where @xmath153 is the inverse function of the restriction of the dispersion relation to the interval @xmath29 $ ] . adding a fermion with momentum @xmath154 ( or , equivalently , @xmath155 ) , with @xmath156 , to the ground state increases the energy by @xmath157 .
the same excess energy is approximately obtained when removing from the ground state a fermion with momentum @xmath158 ( or @xmath159 ) .
thus for low excitation momenta we have @xmath160 , as in a @xmath1-dimensional cft with speed of `` sound '' @xmath161 .
the simple argument outlined above , based on linearizing the dispersion relation near the fermi momentum @xmath162 ( or @xmath163 ) the only region in momentum space relevant at low excitation energies does not provide any information on the central charge of the underlying cft . a more precise way of establishing the equivalence at low energies of the @xmath0 spin chain with a @xmath1-dimensional cft , and in particular of determining its central charge , is based on the analysis of the chain s free energy . indeed , as mentioned in the introduction , at low temperatures the free energy per unit length of a @xmath1-dimensional cft is given ( in natural units @xmath164 ) by @xmath165 where @xmath166 is a constant , @xmath34 is the central charge and @xmath27 is the effective speed of sound . on the other hand , by eq . the free energy of the spin chain is simply given by @xmath167 where @xmath168 ( with @xmath169 ) is the partition function of the @xmath91-th normal mode .
substituting in the previous equation and using the relation @xmath95 we obtain the closed formula @xmath170{\mathrm{d}}p.\ ] ] in order to determine the low temperature behavior of @xmath171 , we note that @xmath172 is negative for @xmath173 and positive for @xmath174 , so that @xmath175 , where @xmath176\,{\mathrm{d}}p = f(0)\ ] ] is constant and @xmath177{\mathrm{d}}p\,,\\ \label{f2def } f_2(t)&=-\frac t\pi\int_{p_0}^\pi\log\big[1+{\mathrm e}^{-{\beta}({{\mathcal e}}(p)-{\lambda})}\big]{\mathrm{d}}p\end{aligned}\ ] ] vanish at @xmath178 . the low temperature behavior of @xmath179 can be determined by performing the change of variable @xmath180 , which yields @xmath181 the condition @xmath182 implies that @xmath183 and hence @xmath184 , where @xmath161 is the fermi velocity .
we thus have @xmath185 so that for @xmath186 we obtain for @xmath187 , so that the error incurred in extending the upper integration limit to @xmath188 is bounded by @xmath189 for all @xmath190 , where @xmath191 is a positive constant . ]
@xmath192 the last term @xmath193 can be similarly dealt with through the change of variable @xmath194 , with the same result .
hence at low temperatures we have @xmath195 which coincides with eq . with @xmath2 .
this shows that the spin chain is indeed critical for @xmath151 , with central charge @xmath2 .
the critical behavior of the @xmath0 chain at the endpoints @xmath196 can be similarly investigated . indeed , suppose to begin with that @xmath197 .
in this case @xmath198 , where @xmath199 is as in eq . with @xmath200
, so that performing the change of variable @xmath201 we obtain @xmath202 the dispersion relation can be expanded around @xmath203 as @xmath204 , where @xmath205 denotes the order of the lowest nonvanishing derivative of @xmath93 at the origin ( generically , therefore , @xmath206 ) and @xmath207^{1/{\kappa}}\,.\ ] ] from the latter expansion we have @xmath208 $ ] , and therefore @xmath209 substituting into the previous equation for @xmath171 we thus obtain @xmath210 with @xmath211 the integral @xmath212 can actually be evaluated using the technique of ref .
@xcite , namely : @xmath213 where @xmath214 is riemann s zeta function , @xmath215 is dirichlet s eta function , and we have used the identity @xmath216 . substituting into eq .
we finally obtain @xmath217 with @xmath218 we thus see that for @xmath197 the chain can not be critical unless @xmath206 , i.e. , @xmath219 .
moreover , for @xmath206 we have @xmath220 , and therefore @xmath221 this shows that when @xmath197 and @xmath219 the chain is still critical but has central charge @xmath222 , and its low energy behavior is therefore described by a cft with one free _
fermion_. for instance , for the elliptic interaction @xmath223 for @xmath224 , while @xmath206 for @xmath107 . in particular , for @xmath224 eqs . - with @xmath223 reproduce the result in ref .
@xcite . on the other hand
, it is well known that the @xmath0 haldane shastry chain ( i.e. , the @xmath107 case ) can indeed be described at low energies by a @xmath1-dimensional cft with one free fermion @xcite .
the analysis is totally analogous at the other endpoint @xmath225 . indeed , since now @xmath226 for @xmath227 we have @xmath228 , where @xmath26 and @xmath229 are respectively given by eqs . and with @xmath225 . performing the usual change of variable @xmath230 $ ] we thus arrive at eq . .
near @xmath130 we have @xmath231^\nu+{\operatorname{o}}[(\pi - p)^{\nu+1}]$ ] , where @xmath232 denotes the lowest nonvanishing derivative of the dispersion relation at @xmath130 and @xmath233^{1/\nu}\,.\ ] ] note that , by the symmetry @xmath95 , @xmath232 is necessarily _ even _ and @xmath234 . proceeding as before we readily obtain eqs . - with @xmath110 and @xmath235 respectively replaced by @xmath236 and @xmath232 .
in particular , since in this case @xmath237 we see that at the endpoint @xmath225 the model is not critical . in summary
, our analysis indicates that the latter model is critical for @xmath151 , and for @xmath197 when @xmath219 .
we shall study in this section the von neumann entanglement entropy @xmath238 of the ground state of the @xmath0 supersymmetric model , defined as the von neumann entropy of the reduced density matrix @xmath239 of a block of @xmath31 consecutive sites when the system is in its ground state . in other words ,
if we denote by @xmath240 the ground state of the chain then @xmath241 , where @xmath242 denotes the trace over the hilbert space of the remaining @xmath243 sites , and the von neumann entanglement entropy is given by @xmath244 more generally , we shall also consider the rnyi entropy @xmath245 where @xmath246 is a real parameter , which reduces to that of von neumann in the @xmath247 limit . as pointed out in the introduction , the von neumann and rnyi ground - state entanglement entropies of a @xmath1-dimensional cft scale as @xmath248
when @xmath249 , where the coefficient @xmath250 is related to the holomorphic and anti - holomorphic central charges @xmath34 and @xmath251 by @xmath252 ( with @xmath253 for the von neumann entropy ) .
since the @xmath0 supersymmetric chain is critical for @xmath151 , with central charge @xmath254 , it is to be expected that for this model @xmath255 in the limit @xmath249 .
in fact , we shall rigorously establish this asymptotic formula in the next section ( cf .
eq . ) .
before addressing the actual computation of the entanglement entropy of the @xmath0 chain , we note that the result is the same for its `` antiferromagnetic '' version @xmath256 . this is most easily proved by considering the equivalent hamiltonian , whose ground state entanglement entropy is obviously unchanged if we reverse the roles of the occupied and empty sites . in other words ,
the entanglement entropy is the same for the hamiltonian as for its image under the replacement @xmath257 . using the car and the even character of the interaction @xmath79 ,
it is immediate to show that the latter transformation maps @xmath61 into @xmath258 , which establishes our claim .
first of all , it is clear that the ground state is not entangled for @xmath30 outside the interval @xmath259 $ ] .
indeed , if ( for instance ) @xmath133 the ground state is obviously the vacuum @xmath69 ( i.e. , the state with all sites occupied by bosons for the original hamiltonian ) , since in this case all the modes have positive energy @xmath260 .
in particular , the ground state is a product state ( @xmath261 ) , and is therefore not entangled .
the situation is completely analogous for @xmath134 , since in this case @xmath262 for all @xmath91 and therefore all the modes are excited in the ground state .
thus @xmath263 for all @xmath122 , and therefore @xmath264 hence @xmath265 ( i.e. , the state with all sites occupied by fermions ) , which is again a product state and therefore not entangled .
( this is also true when @xmath225 if @xmath14 is _
odd_. ) from the previous considerations it follows that the ground state entanglement entropy of the model vanishes for @xmath30 outside the interval @xmath259 $ ] ( and when @xmath225 , if @xmath14 is odd ) , since in these cases the ground state is a product state . for this reason , in the rest of this section we shall suppose that @xmath30 belongs to the open critical interval @xmath3 . we shall next find a closed form expression for the entanglement entropy of the @xmath0 chain by applying the method of ref .
@xcite to the equivalent fermionic hamiltonian .
the first step in our computation is the evaluation of the ground - state correlation matrix @xmath266 of the latter model , with matrix elements @xmath267 this matrix can be easily determined ( in the thermodynamic limit ) from the relations @xmath268 which in turn are a straightforward consequence of the car and the conditions @xmath269 characterizing the ground state .
indeed , from the inverse fourier transform formula @xmath270 and the previous relations it immediately follows that , it is obvious that @xmath271 coincides with the @xmath272 limit of the latter equation . ]
@xmath273 let us now consider the analogous correlation matrix @xmath274 for a block of @xmath31 consecutive sites , which by translation invariance we can take as the first @xmath31 ones . by the defining property of the reduced density matrix @xmath239 @xcite , for @xmath275
we have @xmath276 where @xmath277 denotes the trace over the hilbert space of the first @xmath31 sites .
thus @xmath274 is just the submatrix of @xmath266 consisting of its first @xmath31 rows and columns .
following ref .
@xcite , we now consider an alternative basis of fermionic operators whose correlation matrix is diagonal .
more precisely , let @xmath278 be a unitary matrix diagonalizing the hermitian matrix @xmath274 , i.e. , satisfying @xmath279 where @xmath280 $ ] are the eigenvalues of @xmath274 .
we then define the operators @xmath281 ( @xmath282 ) by @xmath283 ; note that @xmath281 , though certainly nonlocal , acts on the hilbert space of the first @xmath31 sites .
the operators @xmath281 and their adjoints satisfy the car by the unitarity of the matrix @xmath284 , and their correlation matrix is given by @xmath285 on account of eq . .
as shown in ref .
@xcite , the latter equation and wick s theorem for gaussian states imply that the correlation matrix factorizes as @xmath286 , with @xmath287 the hilbert space of the system is the tensor product of the two - dimensional spaces spanned by the vectors @xmath288 ( @xmath11 ) , where @xmath289 .
moreover , from the car it easily follows that @xmath290 is diagonal in the basis @xmath291 , with respective eigenvalues @xmath292 and @xmath293 .
thus the von neumann and rnyi entropies of @xmath290 are respectively equal to @xmath294 and @xmath295 , where @xmath296\ , . \end{aligned } \right.\
] ] by the additivity property of both of these entropies we then have @xmath297 equations - , which are _ exact _ for any @xmath31 , make it possible to evaluate numerically the ground state entanglement entropy of _ any _ supersymmetric @xmath0 chain of the form in _ polynomial _ time , since they are based on the diagonalization of the @xmath298 matrix with elements . from the latter equations it also follows that the entropy of all of these models is a _
universal _ function of the fermi momentum @xmath162 , the difference between two models being manifested only in the dependence of @xmath162 on the parameter @xmath30 through eq . .
equations - can be used to obtain approximate expressions for the entanglement entropy of the general @xmath0 supersymmetric chain in several interesting regimes . to begin with
, we shall investigate the behavior of the entropy as @xmath30 approaches its extreme critical values @xmath299 and @xmath300 .
suppose , in the first place , that @xmath30 tends to zero for fixed @xmath31 , so that the fermi momentum @xmath162 is much smaller than @xmath301 . in this case
all the matrix elements of the correlation matrix @xmath274 in eq .
are approximately equal to @xmath302 , so that @xmath303 , where @xmath304 is the @xmath298 matrix with all matrix elements equal to @xmath305 . since the eigenvalues of @xmath304 are @xmath299 ( with multiplicity @xmath306 ) and @xmath31 , when @xmath307 the rnyi entanglement entropy is approximately given by @xmath308 \dfrac{q}{q-1}\dfrac{lp_0}{\pi}\,,\quad & q>1\ , .
\end{cases}\ ] ] for the same reason , when @xmath307 the von neumann entropy can be approximated by @xmath309 in particular , we see that both @xmath310 and @xmath238 are continuous exactly equal to @xmath299 the previous calculation only shows that the entropy has the same limit as @xmath311 and @xmath312 . ] at @xmath197 .
similarly , suppose now that @xmath162 is close to its upper critical value @xmath300 , so that @xmath313 with @xmath314 .
in this case we have @xmath315 while @xmath316 .
thus @xmath317 , where @xmath318 is the @xmath298 matrix with matrix elements @xmath319 .
it is easy to check that the eigenvalues of @xmath318 are again @xmath299 ( with multiplicity @xmath306 ) and @xmath31 , so that the previous asymptotic expressions for @xmath310 and @xmath238 still hold with @xmath162 replaced by @xmath320 . in particular , this shows that the von neumann and rnyi entanglement entropies are both continuous also at the upper critical value @xmath225 . on the other hand ,
it is clear that these entropies have a discontinuous first derivative ( with respect to the chemical potential @xmath30 ) at both endpoints @xmath197 and @xmath225 .
for instance , for @xmath321 we have @xmath322 where @xmath235 is the order of the first non - vanishing derivative of @xmath93 at @xmath203 and @xmath110 is defined in eq . .
thus @xmath323 diverges as @xmath324 when @xmath312 .
similarly , for @xmath325 the derivative of the rnyi entropy diverges as @xmath326 in this limit , while for @xmath327 @xmath328 diverges as @xmath329 for @xmath330 and tends to a non - zero finite limit when @xmath206 .
the situation is similar at the other endpoint @xmath225 , i.e. , @xmath331 with @xmath236 defined by eq . , except that now @xmath232 ( the order of the lowest nonvanishing derivative of @xmath93 at @xmath130 ) is necessarily even and thus greater than or equal to @xmath332 .
hence in all cases the derivatives of @xmath238 and @xmath310 diverge as @xmath333 .
the above analysis strongly suggests that there is a quantum phase transition at @xmath197 and @xmath225 between an ordered ( non - entangled ) and a disordered ( entangled ) ground state , with the entanglement entropy as the order parameter .
this conclusion is confirmed by the behavior of the zero - temperature fermion density @xmath334 , which by translation invariance is simply given by @xmath335 in the critical interval @xmath151 .
indeed , by eqs . - , near the two critical points @xmath196 the fermion density respectively behaves as @xmath336 and @xmath337 . since @xmath338 for @xmath133 and @xmath339 for @xmath134 ,
this behavior is typical of a quantum phase transition with exact exponents @xmath340 and @xmath341 at the critical points @xmath197 and @xmath225 .
for instance , for the elliptic interaction it is known @xcite that @xmath342 and @xmath223 for @xmath224 , while @xmath206 for @xmath107 ( i.e. , for the @xmath0 hs chain ) .
the parameters @xmath110 and @xmath236 can also be exactly computed in this case , namely @xmath343^{-1/2},\end{aligned}\ ] ] where @xmath344 and @xmath345 is the second invariant of the weierstrass function with half - periods @xmath346 @xcite .
chain with elliptic interactions for several values of the parameter @xmath347 in the range @xmath348 $ ] .
the red ( top ) and blue ( bottom ) dashed curves correspond respectively to the xx model ( @xmath106 ) and the @xmath0 haldane shastry chain ( @xmath107 ) . ] for the general elliptic @xmath0 model with interactions ( with @xmath349 ) and dispersion relation , it is of course unfeasible to explicitly invert @xmath93 to obtain a closed - form expression for the fermi momentum @xmath350 .
note , however , that the graph of the fermion density @xmath334 admits the simple parametrization @xmath351 , with @xmath352 . in this way we have generated the plot in fig .
[ fig.nf ] , where @xmath334 is represented as a function of the normalized parameter @xmath353 , where @xmath354 $ ] , for several values of @xmath347 in the range @xmath355 $ ] and for @xmath107 .
the fermion density can be easily computed in closed form for the limiting cases @xmath106 and @xmath107 , i.e. , for the xx model and the @xmath0 haldane shastry chain , due to the simple form of their dispersion relations .
indeed , from eq .
we immediately obtain @xmath356 respectively for @xmath357 and @xmath358 .
as expected , the first of these formulas agrees with the result in ref .
@xcite , taking into account that our parameter @xmath30 is related to the parameter @xmath79 in the latter reference by @xmath359 . on the other hand ,
the formula for the @xmath0 hs chain is , to the best of our knowledge , new .
it is also of interest to determine the asymptotic behavior of the von neumann and rnyi entropies for @xmath151 fixed and @xmath37 . to this end
, we note that eq .
implies that @xmath360 is a function of @xmath361 only , and hence the correlation matrix @xmath274 is a toeplitz matrix .
this fact can be exploited to find a simple asymptotic formula for the von neumann and rnyi entanglement entropies in the @xmath249 limit , as shown in ref .
@xcite for the xx model .
the formula in the latter reference , which is based on a particular case of the general fisher hartwig conjecture @xcite proved by e. basor in 1979 @xcite , is also valid for a general model of the form provided only that we express the result in terms of the fermi momentum @xmath162 .
indeed , this formula relies only on eq . for the correlation matrix which , as we have just seen , holds for the model with @xmath350 . in this way one obtains the following asymptotic formula for the rnyi entropy in the limit @xmath362 : @xmath363 while the corresponding formula for the von neumann entropy is obtained from the above by setting @xmath253 . here
@xmath364 denotes a function of @xmath31 and @xmath162 which tends to @xmath299 as @xmath365 , and @xmath366 is a constant ( independent of @xmath31 and @xmath162 ) whose precise value , which can be found in ref . @xcite , will not be needed in what follows . equation can be easily applied in the case of the xx and @xmath0 haldane shastry chains . indeed , for the former of these models we have @xmath367 , so that agrees with the result in ref .
@xcite . on the other hand , for the @xmath0 hs chain @xmath368 , and hence eq .
yields the following asymptotic formulas for the von neumann and rnyi ground state entanglement entropies : @xmath369 + { \gamma}_1^{(q)}+{\operatorname{o}}(1)\ , .
\label{shs}\ ] ] these formulas are valid for @xmath30 belonging to the critical interval @xmath370 , in the asymptotic regime @xmath371 . for the general elliptic @xmath0 model with @xmath349
no such closed formulas are available . however , as for the fermion density , the graph of @xmath310 admits the simple parametrization @xmath372 where for simplicity we have dropped the @xmath364 terms . ) of the elliptic @xmath0 chain - for @xmath373 and several values of the parameter @xmath347 between @xmath332 and @xmath374 .
the red and blue dashed curves correspond respectively to the xx heisenberg model ( @xmath106 ) and the @xmath0 haldane shastry chain ( @xmath107 ) . ] for instance , in fig .
[ fig.splot ] we present a plot of the approximation to the von neumann entropy of the elliptic @xmath0 chain - for @xmath373 and several values of the parameter @xmath347 , including the limiting cases @xmath106 ( the heisenberg xx model ) and @xmath107 ( the @xmath0 haldane shastry chain ) .
it is apparent that all of these plots are qualitatively similar , although only in the case of the xx model ( @xmath106 ) the graph of @xmath238 is symmetric about the midpoint @xmath375 .
more precisely , the maximum of @xmath238 at @xmath376 is increasingly displaced towards the right as @xmath347 tends to infinity , with @xmath377 varying continuously from @xmath15 to @xmath378 as @xmath347 ranges from @xmath299 to @xmath379 .
in the previous sections we have seen that the @xmath0 chain is critical for @xmath151 , with central charge @xmath2 . this is confirmed by the asymptotic behavior of the ground state entanglement entropy when the size of the block of spins considered tends to infinity . in this section
we shall show that the fermion density at finite temperature , given by @xmath380 also exhibits a qualitatively richer behavior when @xmath30 lies in the critical interval @xmath3 . chain - with @xmath381 for @xmath382 and @xmath383 . ] as a concrete example , we shall first focus on the @xmath0 elliptic chain - . in fig .
[ fig.cplot ] we present a contour plot of @xmath334 for this model with @xmath381 for @xmath384 $ ] and @xmath385 $ ] , obtained by numerically evaluating the integral in eq . .
for @xmath30 outside the critical interval @xmath3 , it is clear that @xmath334 is a monotonic function of @xmath24 ( increasing for @xmath386 , decreasing for @xmath387 ) , since @xmath388}\,{\mathrm{d}}p\,.\ ] ] on the other hand , it is apparent from fig .
[ fig.cplot ] that there is a range of values of @xmath30 in the interval @xmath3 for which the fermion density is not a monotonic function of the temperature . remarkably , the @xmath0 elliptic chain exhibits this interesting behavior for all values of the parameter @xmath347 , including the limiting cases @xmath106 and @xmath107 .
more precisely , for each @xmath347 there are three critical values @xmath389 ( @xmath390 ) of the chemical potential @xmath30 such that : i ) for @xmath391 , the fermion density reaches an absolute minimum at some positive temperature and then increases monotonically towards its limiting value @xmath15 ; ii ) for @xmath392 , @xmath334 first reaches a maximum at some @xmath393 and then a minimum , after which it tends monotonically to @xmath15 ; iii ) for @xmath394 , @xmath334 is monotonically increasing , and iv ) for @xmath395 , the fermion density attains an absolute maximum at some @xmath393 and then decreases monotonically towards @xmath15 .
this is also true for the limiting values @xmath106 ( xx model ) and @xmath107 ( @xmath0 hs chain ) , for which @xmath396 and @xmath397 , @xmath398 , respectively .
this behavior is qualitatively apparent from fig .
[ fig.dnfplot ] , where we have represented the implicit curve @xmath399 vs. @xmath30 and @xmath24 for @xmath400 , and is also confirmed by the plots of @xmath334 vs. @xmath24 for these values of @xmath347 and @xmath401 presented in fig .
[ fig.nfplots ] . for the elliptic chain - and several values of the parameter @xmath402 .
the red and blue dashed curves correspond respectively to the xx model ( @xmath106 ) and the @xmath0 haldane shastry chain ( @xmath107 ) . ] for several values of the parameter @xmath347 and @xmath403 ( inset : blowup of the range @xmath404 ) .
bottom : analogous plot for @xmath405 . in both plots ,
the red and blue dashed lines correspond to the limiting cases @xmath106 and @xmath107.,title="fig : " ] + for several values of the parameter @xmath347 and @xmath403 ( inset : blowup of the range @xmath404 ) .
bottom : analogous plot for @xmath405 . in both plots
, the red and blue dashed lines correspond to the limiting cases @xmath106 and @xmath107.,title="fig : " ] although in general the integral in eq .
can not be computed in closed form , its low temperature behavior can be readily determined , as we shall now explain . to begin with ,
when @xmath133 the exponent @xmath406 is positive throughout the whole integration range , so that @xmath407 where @xmath201 .
using eq . and
extending the integration range to @xmath188 we obtain @xmath408 proceeding in a similar way we obtain an analogous formula when @xmath134 : @xmath409 we thus see that for @xmath410 $ ] the fermion density at low temperature is monotonic , approaching exponentially its zero temperature values @xmath299 ( for @xmath133 ) and @xmath305 ( for @xmath134 ) . for @xmath197 , the change of variable @xmath201 and eq .
easily yield @xmath411 and therefore @xmath412 \frac{a}{\pi}\,(1 - 2^{1 - 1/{\kappa}}){\gamma}(1+{\kappa}^{-1}){\zeta}_{\mathrm r}({\kappa}^{-1})\,t^{1/{\kappa}}\,,&{\kappa}>1\ , .
\end{cases}\ ] ] likewise , at the other endpoint @xmath225 we have @xmath413 since now @xmath232 is even and hence greater than @xmath305 .
suppose next that @xmath30 lies in the critical interval @xmath3 .
we start by writing the fermion density as the sum @xmath414 where the first term is the value of @xmath334 at @xmath178 ( cf .
. in this case the leading ( @xmath415 ) contributions to the two integrals @xmath416 cancel each other , so that we need to evaluate the @xmath417 corrections . for the first
integral , after performing the change of variable @xmath418 we have @xmath419 expanding @xmath420 to first order in @xmath421 we obtain @xmath422\\ & = v-\frac{{{\mathcal e}}''(p_0)}v\,tx+{\operatorname{o}}[(tx)^2]\,,\end{aligned}\ ] ] where we have used the expansion @xmath423 $ ] .
we thus have @xmath424\right).\ ] ] substituting in the definition of @xmath425 and using the estimate @xmath426 ( with @xmath427 ) we obtain @xmath428 the term @xmath429 can be similarly dealt with through the analogous change of variable @xmath194 , so that @xmath430 $ ] and hence @xmath431\right).\ ] ] from the definition of @xmath429 we immediately obtain @xmath432 and combining this result with eq .
we finally have @xmath433 in particular , for the xx and @xmath0 hs chains the low temperature expansion reads @xmath434 the absence of a term linear in @xmath24 in eq .
is in agreement with the low temperature behavior of @xmath334 apparent from fig .
[ fig.nfplots ] .
it is also interesting to observe that the sign of the leading correction to the @xmath178 value of @xmath334 is opposite to that of the second derivative of @xmath93 at the fermi momentum @xmath162 .
this behavior can be understood by noting that the energy difference between adding a fermion with momentum @xmath154 ( or @xmath155 ) and removing a fermion with momentum @xmath158 ( or @xmath159 ) , with @xmath435 , is given by @xmath436 .
thus when @xmath437 the addition of a fermion is energetically more favorable than its removal for momenta close to the fermi momentum @xmath162 ( or to @xmath163 ) , and consequently the fermion density should increase at sufficiently low temperatures . for instance , for the elliptic interaction with
@xmath438 finite @xmath439 is positive for @xmath440 less than a critical momentum ( which depends on @xmath347 ) and negative for larger momenta , while for the @xmath0 hs chain ( i.e. , for @xmath107 ) @xmath441 is always negative . again , these facts are consistent with the behavior of @xmath334 observed in fig .
[ fig.nfplots ] .
in this paper we introduce a general class of @xmath0 supersymmetric spin chains with long - range interactions generalizing the @xmath0 haldane shastry and inozemtsev ( elliptic ) chains , which can be fermionized using the algebraic properties of the @xmath0 permutation operator .
we exploit this fact to study the critical behavior of this class of models ( with nonzero chemical potential @xmath30 ) in terms of their dispersion relation @xmath4 .
more precisely , we show that they are gapless when the chemical potential lies on the interval @xmath259 $ ] , and that their ground state is a product state unless @xmath30 belongs to this interval .
we prove that the models under study are actually critical when @xmath151 by verifying that their low energy excitations are linear in the excitation momentum , and that their free energy at low temperature exhibits the characteristic quadratic behavior found in a @xmath1-dimensional cft with @xmath2 @xcite . as further confirmation of this critical behavior , we find an exact asymptotic formula for the von neumann and rnyi entanglement entropies for the ground state , showing that when @xmath30 belongs to the open interval @xmath3 they both scale as @xmath36 when the size @xmath31 of the block of spins considered tends to infinity .
moreover , in both cases the constant multiplying @xmath36 is the same as for a @xmath1-dimensional cft with central charge @xmath2 @xcite .
likewise , we show that the asymptotic behaviors of the entanglement entropy and the zero - temperature fermion density as @xmath30 approaches the endpoints of the critical interval @xmath3 are consistent with a quantum ( continuous ) phase transition
. we also analyze the fermion density at finite temperature for a particular class of models with elliptic interactions , finding that its behavior is nontrivial ( for instance , it is not always a monotonic function of the temperature , and it can in fact present up to two extrema at finite temperature ) when @xmath30 belongs to the critical interval .
the results of this paper suggest several lines for future research .
for one thing , they might prove relevant for the @xmath22 analogs of the models discussed in this paper , and most notably the spin @xmath15 inozemtsev and hs chains in the presence of a magnetic field .
indeed , it has been analytically shown that the @xmath0 hs chain with zero chemical potential @xmath30 is equivalent in the thermodynamic limit to its @xmath22 counterpart with zero magnetic field @xcite .
more recently , a numerical computation of the free energy of the spin @xmath15 elliptic chain with no magnetic field suggests that this model and its @xmath0 version with @xmath197 studied in this paper are also equivalent in the thermodynamic limit @xcite .
if this equivalence could be extended to the case of non - zero chemical potential ( or magnetic field strength , for the @xmath22 models ) , the results of this work could be used , for instance , to evaluate the ground state entanglement entropy of the spin @xmath15 elliptic chain and its asymptotic limit when @xmath31 tends to infinity .
another line of research suggested by the present work is the study of the entanglement entropy of the low - lying states of the @xmath0 supersymmetric model when the chemical potential lies in the critical interval @xmath3 .
indeed , it has been recently shown @xcite that in a @xmath1-dimensional cft the quotient between the entanglement entropy of an excited state created by acting on the vacuum with a primary field and that of the ground state is a universal quantity , essentially determined by the conformal weights of the field .
thus the computation of the entanglement entropy of the lowest excited states of the model when @xmath442 , which can be constructed from the equivalent fermionic model , could shed some light on its underlying cft .
this work was partially supported by spain s mineco under grant no . fis2015 - 63966 , and by the universidad complutense de madrid and banco santander under grant no .
gr3/14 - 910556 .
pt has been partly supported by the icmat severo ochoa project sev-2015 - 0554 ( mineco ) .
jac would also like to thank the madrid township and the `` residencia de estudiantes '' for their financial support .
40ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _
_ , , vol . ,
( ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) , , , and , eds .
, @noop _ _ ( , ) @noop _ _ , ed .
( , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , ed .
( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) , @noop * * , ( ) @noop * * , ( ) | we introduce a general class of @xmath0 supersymmetric spin chains with long - range interactions which includes as particular cases the @xmath0 inozemtsev ( elliptic ) and haldane shastry chains , as well as the xx model .
we show that this class of models can be fermionized with the help of the algebraic properties of the @xmath0 permutation operator , and take advantage of this fact to analyze their quantum criticality when a chemical potential term is present in the hamiltonian .
we first study the low energy excitations and the low temperature behavior of the free energy , which coincides with that of a @xmath1-dimensional conformal field theory ( cft ) with central charge @xmath2 when the chemical potential lies in the critical interval @xmath3 , @xmath4 being the dispersion relation .
we also analyze the von neumann and rnyi ground state entanglement entropies , showing that they exhibit the logarithmic scaling with the size of the block of spins characteristic of a one - boson @xmath1-dimensional cft .
our results thus show that the models under study are quantum critical when the chemical potential belongs to the critical interval , with central charge @xmath2 . from the analysis of the fermion density at zero temperature
, we also conclude that there is a quantum phase transition at both ends of the critical interval .
this is further confirmed by the behavior of the fermion density at finite temperature , which is studied analytically ( at low temperature ) , as well as numerically for the @xmath0 elliptic chain . |
the substitutional insertion of a point defect impurity into an otherwise perfect host crystal will typically modify the spectrum of the @xmath0 normal modes of the crystal,@xcite leading in certain circumstances ( such as the insertion of impurities which are lighter in mass than the host atoms which they replace or which are more strongly coupled to the host crystal atoms than the ones they replace ) to the generation of modes with frequencies which lie beyond the band maximum @xmath1 of the crystal .
such modes will not be plane waves which propagate throughout the crystal , but will instead fall off exponentially fast away from the defect site and thus be localized to it .
moreover , with the rest of the crystal atoms not participating appreciably in such localized modes , the intensity of the defect in such modes will be @xmath2 times larger than the intensity it would otherwise have had in a crystal plane wave mode , to thus give the localized mode enough intensity to render it observable .
while such modes could be of relevance for phenomena such as the @xmath3 effect associated with the insertion of @xmath3 active defects into host crystals , historical recoil - free fraction @xmath3 studies only involved an averaging over all the lattice modes of the system , to thereby only allow one to infer the possible presence of localized modes indirectly .
however , with the advent of dedicated synchrotron rings it became possible to monitor @xmath3 active systems mode by mode directly ; and via nuclear resonant inelastic x - ray scattering studies , modes lying beyond the host crystal phonon band maximum have now explicitly been seen in the @xmath4 and the @xmath5 systems.@xcite consequently , knowing whether the insertion of a defect into pure crystal hosts might generate localized modes can be of great value for such studies . in this paper
we use a very straightforward trace technique to enable us to identify some very general conditions under which such localized modes can be produced .
we shall restrict our study to the most straightforward case of single point defects which are substitutionally inserted in the three primitive simple , body - centered and face - centered cubic crystals ( these specific cases being amongst the most commonly experimentally studied ones ) , though with our approach being quite generic , it could in principle be adapted to encompass other crystal structures as well if desired .
we shall provide results for various combinations of mass and force - constant changes , some which are specific to nearest - neighbor force - constants and some of which are general to all - neighbor force - constants . in sec .
ii we derive the trace condition on which all of our results are based , with the provided relation ( eq . ( [ 15 ] ) below ) being an exact , all - neighbor relation which permits an arbitrary mass change ( @xmath6 replaced by @xmath7 ) at the defect site and arbitrary force - constant changes between the defect atom and any other atom in the entire crystal , despite which the resulting relation only involves the self - force constant @xmath8 at the defect site and the change @xmath9 in it . in sec .
iii we apply our trace condition to the case of an isotopic substitution where the only change is that in the mass at the defect site , to show that while having a lighter impurity ( @xmath10 ) is necessary for the generation of a localized mode , it is the condition @xmath11 which is the sufficient one . in secs .
iv and v we apply our trace condition to some straightforward nearest - neighbor force - constant change cases for which there are extremely simple exact analytic solutions to the impurity problem ( some typical examples of which being crystals with force - constants which are central or which are isotropic ) , to find that for all of them the sufficient condition for localized mode production by force - constant changes alone is given as the requirement that @xmath12 be greater than @xmath13 . finally in sec .
v we examine more complicated force - constant change cases and go beyond nearest - neighbor force - constants , and while these situations do not admit of as straightforward a treatment as the cases considered in sec .
iv , nonetheless for them we are still able to extract a general sufficiency condition for the generation of localized modes in an arbitrary crystal by force - constant changes alone , namely that @xmath12 be greater than two .
since the very introduction of a defect breaks the translation invariance of the lattice , to derive a condition such as one involving a trace , we will need to work entirely in the coordinate space crystal site representation . to actually derive the trace condition we recall that in the harmonic approximation the equations of motion for the displacements from equilibrium @xmath14 of the atoms of a pure 3n - dimensional cubic crystal lattice are given by @xmath15u_{\beta}(\ell^{\prime})=0~~ , \label{1}\ ] ] where @xmath16 ranges from @xmath17 to @xmath18 , @xmath19
, @xmath20 is the mass of the atom at site @xmath16 , and @xmath21 are the second order force - constants . since eq .
( [ 1 ] ) is an eigenvalue equation for the frequencies @xmath22 , on defining a matrix @xmath23 , we immediately see that the sum of the eigenfrequencies of the crystal is given by @xmath24 for a pure crystal in which all masses are equal to a common @xmath25 and all self force - constants are equal to a common @xmath26 , eq .
( [ 2 ] ) reduces to @xmath27 with the left - hand side of eq .
( [ 3 ] ) being recognized as @xmath28 , the second moment of the density of states , eq .
( [ 3 ] ) thus recovers the well - known relation for pure crystals @xcite @xmath29 for the system with a substitutional point impurity of mass @xmath7 located at the origin of coordinates and changed force - constants @xmath30 , the displacements from equilibrium are now given as @xmath31 , with eq .
( [ 1 ] ) then being replaced by @xmath32u_{\beta}(\ell^{\prime } ) = \sum_{\beta,\ell^{\prime}}v_{\alpha \beta}(\ell , \ell^{\prime})u_{\beta}(\ell^{\prime})~~ , \label{5}\ ] ] where the changes from the pure crystal case are described by the perturbation @xmath33 on now defining a matrix @xmath34 , the sum of the eigenfrequencies of the perturbed crystal is then given by @xmath35^{-1}~~ , \label{7}\ ] ] with the change in the sum of the eigenfrequencies thus being given by @xmath36={\rm tr } a^{\prime}[b^{\prime}]^{-1}-{\rm tr } ab^{-1}~~. \label{8}\ ] ] the utility of eq .
( [ 8 ] ) is that its left - hand side measures whether modes are shifted to higher or lower frequency , while on its right - hand side it is only the sites which are explicitly involved in @xmath37 which do not drop out of the difference between the two 3n - dimensional traces .
our task is thus to first evaluate the right - hand side of eq .
( [ 8 ] ) and then to seek constraints on its left - hand side . in the simplest case where the only change is a mass change at the defect site , it is only the defect site contribution itself which is not cancelled in the trace difference , with the full change in frequency immediately being given by @xmath36= \frac{3a_{xx}(0,0)}{m^{\prime}}- \frac{3a_{xx}(0,0)}{m}~~. \label{9}\ ] ] when there is also a change in force - constant at the defect site , we recall that by newton s third law of motion there must also be changes in the force - constants at other sites too .
moreover , this same newtonian law entails that the force - constants of a perturbed system with a defect have to obey @xmath38 for all @xmath16 in exactly the same way as the pure crystal force - constants have to obey @xmath39 with the @xmath40 , @xmath41 and @xmath42 directions being equivalent in cubic crystals , in the general case which allows for arbitrary force - constant changes , eq .
( [ 8 ] ) takes the form @xmath43= & & \frac{a^{\prime}_{xx}(0,0)}{m^{\prime}}-\frac{a_{xx}(0,0)}{m } \nonumber \\ & & + \frac{a^{\prime}_{xx}(1,1)}{m}-\frac{a_{xx}(1,1)}{m } + \frac{a^{\prime}_{xx}(2,2)}{m}-\frac{a_{xx}(2,2)}{m}+ .... ~~. \label{12}\end{aligned}\ ] ] as regards the pure and the perturbed force - constants , we recall that in terms of the two - body interatomic potential @xmath44 , the force - constants between an atom vibrating around site @xmath45 and one vibrating around the origin are defined as @xmath46\bigg{|}_0= -\left[\frac{\phi^{\prime\prime}(r)}{r^2 } -\frac{\phi^{\prime}(r)}{r^3}\right]r_{\alpha}(\ell)r_{\beta}(\ell ) -\frac{\phi^{\prime}(r)}{r}\delta_{\alpha\beta}~~ , \label{13}\ ] ] as calculated at the equilibrium separation @xmath47 .
then , since the effect of the introduction of the defect is to alter the two - body potential between the defect and the host atoms while not affecting the potentials between any two host atoms themselves , the only non - self force - constants which will change will then be the @xmath48 with @xmath49 , with eq .
( [ 10 ] ) then obliging the self force - constants at the defect site and those at the @xmath49 sites to respectively change as @xmath50 given these relations and eqs .
( [ 10 ] ) and ( [ 11 ] ) , eq .
( [ 12 ] ) can thus be simplified to @xmath43&&= \frac{a^{\prime}_{xx}(0,0)}{m^{\prime}}-\frac{a_{xx}(0,0)}{m } -\frac{1}{m}\sum_{\ell \neq 0}\left[a^{\prime}_{xx}(\ell,0 ) -a_{xx}(\ell,0)\right ] \nonumber \\ & & = \frac{a^{\prime}_{xx}(0,0)}{m^{\prime}}+\frac{a^{\prime}_{xx}(0,0)}{m } -\frac{2a_{xx}(0,0)}{m}~~. \label{15}\end{aligned}\ ] ] equation ( [ 15 ] ) is our main result , and is derived here with no restriction at all on the number of neighbors of the defect for which the force - constants might change . nor
does it presuppose any relation between @xmath51 and @xmath52 . despite the fact that eq .
( [ 15 ] ) conveniently only involves the self force - constant at the defect site , nonetheless it is an all - neighbor result , one which additionally holds for any combination of central and non - central force - constants.@xcite
in the treatment of crystal impurity problem it is conventional to solve eq .
( [ 5 ] ) by the lattice green s function technique , and since we will use some of its aspects to constrain the left - hand side of eq . ( [ 15 ] ) , we briefly recall the procedure .
one first introduces the dynamical matrix of the pure crystal @xmath53 as expressed in terms of the phonon modes @xmath54 of the translational invariant pure crystal , and then defines its eigenvectors and eigenvalues according to @xmath55 and then uses these eigenvectors and eigenvalues to construct the pure crystal lattice green s functions according to @xmath56}}{[\omega_{j}^2(\vec{\bf k})-\omega^2 ] } \label{18}\ ] ] as summed over the polarizations @xmath57 and momenta @xmath54 of all the modes in the brillouin zone .
as constructed these green s functions obey @xmath58 and thus immediately allow us to solve eq .
( [ 5 ] ) in the form @xmath59 with the eigenmodes being given as the solutions to the ( @xmath0-dimensional ) determinantal condition @xmath60 as written in an obvious notation . for the simplest case of just a change in mass at the defect site , eq . ( [ 20 ] ) requires the three components of the defect displacement vector to obey @xmath61 where the cubic symmetry of the host lattice requires that @xmath62 , @xmath63 and @xmath64 all be equal to each other , and thus that each one of them be given as @xmath65 } = \frac{1}{3}\left[g_{xx}(\omega^{\prime};0,0 ) + g_{yy}(\omega^{\prime};0,0 ) + g_{zz}(\omega^{\prime};0,0)\right ] \nonumber \\ & = & \frac{1}{3nm}\sum_{\bar{k},j } \frac{[\sigma_x^{*j}(\vec{\bf k})\sigma_x^{j}(\vec{\bf k } ) + \sigma_y^{*j}(\vec{\bf k})\sigma_y^{j}(\vec{\bf k } ) + \sigma_z^{*j}(\vec{\bf k})\sigma_z^{j}(\vec{\bf k } ) ] } { [ \omega_j^2(\vec{\bf k})-\omega^{\prime 2 } ] } \nonumber \\ & = & \frac{1}{3nm}\sum_{\bar{k},j } \frac{1 } { [ \omega_j^2(\vec{\bf k})-\omega^{\prime 2 } ] } \nonumber \\ & = & \frac{1}{nm}\sum_{i=1}^n \frac{1}{(\omega_i^2-\omega^{\prime 2 } ) } = \frac{1}{m}\int_0^{\omega^2_{\rm max}}d\omega^{2}\frac{\nu(\omega^{2})}{(\omega^2-\omega^{\prime 2 } ) } \label{22}\end{aligned}\ ] ] as now summed over @xmath2 threefold degenerate pure crystal eigenmodes @xmath66.@xcite and with the determinantal condition then being given by @xmath67 ^ 3(1)^{3n-3}=0 $ ] , in the mass defect case the perturbed simple , body - centered and face - centered cubic crystal modes are thus given as the solutions to the familiar @xcite @xmath68 where @xmath69 is given by @xmath70 and @xmath71 is given by @xmath72 as summed over the @xmath2 eigenfrequencies @xmath66 of the pure crystal . for our purposes here
we note that the function @xmath71 is divergent at every pure crystal frequency , to thus contain the set of @xmath2 asymptotes indicated in fig .
@xmath71 starts out equal to minus one at @xmath73 , and decreases to minus infinity at the first asymptote , with @xmath71 thus having no zero below the lowest pure crystal band mode . since the derivative of @xmath71 , viz
@xmath74 is negative definite , the function @xmath71 falls monotonically between asymptotes , to thus have a zero somewhere on the @xmath75 axis between any two adjacent pure crystal modes , while beyond the last pure crystal frequency eigenmode at @xmath76 the function @xmath71 is positive , falling to zero at @xmath77 .
as introduced , the function @xmath78 thus has @xmath18 zeroes inside the pure crystal phonon band and one zero outside . in consequence of this structure for @xmath71 , we see that the curve @xmath79 will intercept the @xmath71 curve @xmath2 times no matter what the value of @xmath80,@xcite with eq .
( [ 21 ] ) thus always having the same number of mode solutions as the pure crystal ( viz .
@xmath2 solutions for each of the three @xmath81 , @xmath82 and @xmath83 sectors of eq .
( [ 21 ] ) ) , with eq .
( [ 23 ] ) and the threefold degeneracy of eq .
( [ 21 ] ) thus accounting for the complete spectrum of eigenfrequencies of the perturbed crystal .
when @xmath7 is less than @xmath6 ( typical intercept @xmath84 ) the @xmath2 modes which satisfy eq .
( [ 23 ] ) are all shifted to higher frequency compared to their pure crystal counterparts , with both sides of eq .
( [ 9 ] ) thus consistently being positive . similarly , when @xmath7 is greater than @xmath6 ( typical intercept @xmath85 ) the @xmath2 modes which satisfy eq .
( [ 23 ] ) are all shifted to lower frequency compared to their pure crystal counterparts , with both sides of eq .
( [ 9 ] ) then consistently being negative . finally , once the eigenfrequencies have been found , a return to eq .
( [ 5 ] ) will then allow a determination of all of the @xmath0 displacements @xmath86 in every mode , to then completely specify the crystal displacements of the perturbed crystal . since the zeroes of @xmath71 other than the final one at @xmath77 lie between adjacent asymptotes , an initial glance at fig .
( 1 ) would suggest that when shifted upwards , each of @xmath87 modes other than the mode at the band maximum itself could be shifted upward as far as the immediate next asymptote , to suggest that the quantity @xmath88 $ ] could be as large as @xmath89 ( the net shift between the highest and lowest in - band frequency modes ) .
however , we will now show that in fact the net shift must be much less than this , an outcome that will sharply constrain the implications of eq .
( [ 15 ] ) . to this end
we need to obtain a bound on the sum of the in - band zeroes , @xmath90 , of @xmath71 , and note that since the quantity @xmath91 is an @xmath18 dimensional polynomial in @xmath92 , the sum of its zeroes can immediately be given as @xmath93 where @xmath94 on defining the positive semi - definite quantity @xmath95 via @xmath96 ^ 2\rangle = \alpha~~ , \label{29}\ ] ] through use of eq .
( [ 3 ] ) we then obtain @xmath97 = \omega_{\rm max}^2-\mu_2 -\frac{\alpha}{\mu_2 } = \omega_{\rm max}^2-\frac{a_{xx}(0,0)}{m}-\frac{\alpha}{\mu_2}~~. \label{30}\ ] ] now in general it can be shown that the quantity @xmath98 is related to @xmath76 according to@xcite @xmath99 where the quantity @xmath100 is given by @xmath101 as summed over the even neighbors of the @xmath102 site alone . with this sum only beginning with the second nearest neighbors , to good approximation we can neglect the contribution of @xmath100 , and can thus rewrite eq .
( [ 30 ] ) as @xmath97 = \frac{\omega_{\rm max}^2}{2}-\frac{\alpha}{\mu_2}~~. \label{33}\ ] ] with @xmath95 being positive , we thus see that the quantity @xmath103 $ ] can not be larger than @xmath104 , with the zeroes of @xmath71 on average being no more than midway between adjacent asymptotes rather than close to the immediately adjacent higher ones .
further support for this result can be obtained by considering eq .
( [ 23 ] ) in the infinitely heavy defect limit in which @xmath105 , @xmath106 , corresponding to the line @xmath85 in fig .
( 1 ) . in this case
( [ 9 ] ) reduces to @xmath107= -\frac{a_{xx}(0,0)}{m}=-\frac{\omega_{\rm max}^2}{2}~~ , \label{34}\ ] ] with the structure of fig .
( 1 ) then yielding @xmath108-\omega_{\rm max}^2~~ , \label{35}\ ] ] from which the bound @xmath109<\frac{\omega_{\rm max}^2}{2 } \label{36}\ ] ] then follows . having now established the bound on the zeroes of @xmath71 which is given in eqs .
( [ 33 ] ) and ( [ 36 ] ) , we now note that for the pure mass defect case eq .
( [ 9 ] ) requires that @xmath110+\omega^{\prime 2}_{\rm max } -\omega^ { 2}_{\rm max}=\frac{a_{xx}(0,0)}{m^{\prime}}- \frac{a_{xx}(0,0)}{m}~~ , \label{37}\ ] ] where @xmath111 is the largest shifted frequency . since in the @xmath10 case
the in - band modes have to lie to the left of the @xmath112 zeroes of @xmath71 , the in - band @xmath113 $ ] can then never be any larger than @xmath114 .
consequently , no matter what the value of @xmath95 of eq . (
[ 29 ] ) ( a quantity which varies from one crystal to the next ) , and no matter which particular crystal one might choose , if @xmath7 is less than @xmath115 , the largest perturbed eigenfrequency @xmath111 would than have to be larger than the pure crystal maximum @xmath116 , and not only would it then lie beyond the pure crystal phonon band , once @xmath7 has been reduced enough to move the in - band modes as far to higher frequency as they are able to go ( viz . to the zeroes of @xmath71 ) , further reduction in @xmath7 would then cause @xmath111 to move further and further away from the band maximum .
hence while the condition @xmath10 is a necessary condition for producing a mode beyond the band maximum in simple , body - centered and face - centered cubic crystals , it is @xmath11 which is the sufficient one , a result previously obtained@xcite by entirely different means.@xcite
because of newton s third law of motion , in the presence of force - constant changes the matrix @xmath117 of eq .
( [ 6 ] ) will necessarily involve more atoms than just the one at the defect site and the problem is essentially intractable other than numerically if @xmath117 extends beyond the nearest neighbors of the defect . moreover , even in the event that one restricts the force - constant changes to the defect and its nearest neighbors alone , for the simple cubic , body - centered cubic and face - centered cubic crystals the defect and its nearest neighbors respectively consist of a cluster of seven , nine and thirteen atoms , making @xmath117 21- , 27- and 39-dimensional in those respective cases .
fortunately , because of the cubic crystal @xmath118 symmetry at the defect site the @xmath117 matrix can be block diagonalized in the irreducible representations of the octahedral group , with typical decompositions@xcite @xmath119 since the displacement of the defect atom itself transforms as a 3-dimensional vector , the defect displacements must be located entirely in the @xmath120 modes , with all of the other relevant @xmath118 irreducible representations being built solely out of appropriate @xcite linear combinations of the displacements of the nearest neighbors of the defect .
as such the defect displacement appears in 3-dimensional , 3-dimensional and 4-dimensional representations in the three respective cubic crystal cases , with the defect sector thus leading to respective 3-dimensional , 3-dimensional and 4-dimensional blocks in eq .
( [ 20 ] ) each one of which ( just as in eq .
( [ 21 ] ) ) is threefold degenerate
. while these blocks are still quite complicated,@xcite in the case of central force - constants alone ( as well as in some other specific cases such as isotropic force - constant crystals which we discuss in sec .
v ) the determinantal condition @xmath60 associated with the @xmath120 block can be treated completely analytically , with both the body - centered cubic and face - centered cubic crystal eigenmodes being found @xcite to obey precisely the same eq .
( [ 23 ] ) as before , viz .
@xmath121 save only that this time @xmath69 is given by @xmath122~~. \label{40}\ ] ] for our purposes here we note that since the function @xmath78 is the same one as discussed previously , it still has the asymptote structure shown in fig .
( 1 ) . to determine the number of times the function @xmath123 will intersect @xmath78 requires considering each possible combination of mass and force - constant changes in eq .
( [ 40 ] ) separately . for @xmath124 and @xmath125
the quantity @xmath123 is everywhere positive yielding the typical curve @xmath126 in fig .
( 2 ) with its @xmath2 intercepts ( the point @xmath127 is at @xmath128 when @xmath129 ) , with each crystal eigenmode having been shifted upwards to a higher frequency . for @xmath130 and @xmath131
the quantity @xmath132 is everywhere less than or equal to @xmath133 yielding the typical curve @xmath134 in fig .
( 2 ) with its @xmath2 intercepts ( the point @xmath135 is at @xmath136 when @xmath129 ) , with each crystal eigenmode having been shifted downwards . for @xmath10 but @xmath131 the quantity @xmath132 has to diverge somewhere , yielding the typical curve @xmath137 in fig . ( 2 ) with its @xmath2 intercepts if the divergence in @xmath132 falls inside the band,@xcite and analogously also yielding @xmath2 intercepts if the divergence is outside the band ( not shown).@xcite finally , for @xmath138 and @xmath125 there will again be @xmath2 intercepts ( typical curve @xmath139).@xcite as we thus see , no matter what particular values @xmath80 and @xmath140 might actually take , in all cases there are precisely @xmath2 intercepts , and thus precisely @xmath0 eigenmodes in the @xmath120 sector .
however , since this exhausts the number of degrees of freedom for the problem , eq .
( [ 20 ] ) can not generate any further eigenmode solutions .
consequently , none of the determinantal conditions associated with any of the other irreducible octahedral representations in the cluster can yield eigenmode solutions , though just as with all the rest of the atoms in the crystal , the @xmath86 displacements associated with these other irreducible representations will , via eq .
( [ 5 ] ) , still participate in the @xmath120 mode oscillations . with the eigenmodes associated with the @xmath120 sector thus exhausting the eigenspectrum
, it will be these modes alone which will contribute to the trace condition of eq .
( [ 15 ] ) . and with the zeroes of @xmath71 still obeying eq .
( [ 33 ] ) , for a defect with @xmath129 in a crystal with central force - constants alone and eigenmodes which then obey @xmath141=\sum_{i=1}^{n-1}[\omega^{\prime 2}_i -\omega^{2}_{i}]+\omega^{\prime 2}_{\rm max } -\omega^ { 2}_{\rm max}=\frac{2a^{\prime}_{xx}(0,0 ) } { m}-\frac{2a_{xx}(0,0)}{m}~~ , \label{41}\ ] ] and in - band modes which can never be shifted up beyond the immediate next zeroes of @xmath71 ( typical curve @xmath126 in fig .
( 2 ) ) , we can conclude that no matter which central force - constant crystal we may choose , there will definitely be a localized mode if @xmath140 is greater than @xmath13 , with this condition thus being sufficient to guarantee localized modes in nearest - neighbor central force - constant cubic crystals when there is no change in mass.@xcite
to extend these results to other cases , we need to consider both non - central force - constants and go beyond nearest neighbors . as regards the issue of non - central force - constants , we note that within the nearest - neighbor approximation a complete and exact general relation for locating the perturbed crystal @xmath120 modes has recently actually been obtained.@xcite specifically , it was found for the body - centered cubic crystal ( and thus by extension for the face - centered cubic crystal as well since it is the reciprocal lattice of the body - centered cubic and we work in the harmonic approximation where momentum and position are treated equivalently)@xcite that the general determinantal condition @xmath60 in the perturbed @xmath120 mode sector can be written as @xmath142 where the nearest - neighbor , arbitrary force - constant @xmath143 is given as @xmath144\left[1-\hat{r}\right ] -\mu\hat{r}\left[s(\omega^{\prime 2})\left(1-\frac{m}{m^{\prime}}\right ) + 1\right]\right\}~~ , \label{42}\ ] ] expressed here in terms of the two quantities @xmath69 and @xmath71 which were given previously and two additional quantities @xmath145 and @xmath146 which are needed now . to define these two additional quantities
one needs to introduce the pure crystal lattice green s function combination @xmath147 } \left[e^{i\vec{\bf k}\cdot\vec{\bf r}(222)}+ e^{i\vec{\bf k}\cdot\vec{\bf r}(220)}\right]~~ , \label{43}\end{aligned}\ ] ] and in terms of the interatomic potential @xmath44 between two pure crystal nearest neighbors and the interatomic potential @xmath148 between the defect and a host crystal nearest neighbor , define pure and impure lattice force - constants via @xmath149 where @xmath150 ( as evaluated at the pure crystal nearest - neighbor equilibrium separation @xmath47 ) , with @xmath146 , @xmath145 and @xmath151 then being given by @xmath152 equation ( [ 42 ] ) is not only a very compact relation ( it only requires knowledge of two pure crystal green s functions combinations , viz . @xmath153 and @xmath154),@xcite but it additionally reduces to the previous condition given in eq .
( [ 39 ] ) whenever @xmath151 is zero .
now while the quantity @xmath151 would vanish when @xmath155 , viz . the previously discussed central force - constant case , it would also vanish when @xmath156 , viz . the isotropic force - constant case , and also whenever @xmath157 and @xmath158 are equal , viz . in the mixed case in which the fractional changes in the central and isotropic components of the force - constants are equal to each other , with @xmath159 being equal to @xmath160 in all such cases .
for all of these cases then , the condition @xmath161 is sufficient to guarantee localized modes in nearest - neighbor crystals when there is no change in mass . to treat cases where the quantity @xmath145 is not zero is not as straightforward , since unlike the @xmath162 green s function
, the @xmath154 combination can not be reduced to a sum solely over pure crystal eigenfrequencies as the polarization vectors can not readily be eliminated from eq .
( [ 43 ] ) , with eq .
( [ 42 ] ) not obviously being reducible to an expression which only involves the pure crystal density of states @xmath163 . however , despite this , it is still possible to extract a sufficiency condition for localized modes . specifically , even though @xmath154 does involve the pure crystal polarization vectors , as can be seen from eq .
( [ 43 ] ) , its poles , and thus its asymptotes , are nonetheless precisely the same as those possessed by @xmath153 , with the asymptotes of @xmath143 then being none other than the asymptotes of @xmath78 , i.e. none other than the ones exhibited in fig .
( 2 ) . to determine what happens
when we take @xmath145 to be non - zero then , we need to monitor how the intercept structure displayed in fig .
( 2 ) gets modified as we slowly switch @xmath145 on . to explicitly see what happens when we switch @xmath145 on , it is instructive to first return to the central force - constant case , and consider the situation in which we start with some arbitrary @xmath9 which is bigger than @xmath8 and some initial @xmath7 which is equal to @xmath6 , and them slowly start to increase @xmath7 .
we thus start with a @xmath69 which is equal to @xmath164 $ ] , and thus with a @xmath123 which is everywhere positive and infinite at @xmath73 ( viz .
curve @xmath126 of fig .
when we now allow @xmath7 to be just a little bit bigger than @xmath6 , @xmath69 will now take small negative values at the smallest @xmath92 , and then quickly revert back to being positive again as @xmath92 is increased . consequently , @xmath123 will now take large negative values at the smallest @xmath92 , while also quickly reverting back to being positive again as @xmath92 is increased ( viz .
curve @xmath139 of fig .
as can therefore be seen from fig .
( 3 ) , the net effect of letting @xmath7 be just a little bit bigger than @xmath6 is that rather than increasing , the frequency of the lowest mode is instead decreased .
now since the trace condition of eq .
( [ 15 ] ) holds for both of these two cases , we can write a trace condition for the difference between the two cases , viz . @xmath165= \frac{a^{\prime}_{xx}(0,0)}{m^{\prime } } -\frac{a^{\prime}_{xx}(0,0)}{m}~~ , \label{47}\end{aligned}\ ] ] with the net effect of a small difference between @xmath7 and @xmath6 on the right - hand side of eq .
( [ 47 ] ) entailing a small net total shift on the left - hand side , and thus a net shift in each individual eigenmode of order @xmath166 of the shift on the right - hand side . moreover , with the change in the right - hand side of eq .
( [ 47 ] ) being continuous , the change on the left - hand side must be continuous as well .
consequently , with the transition from curve @xmath126 to curve @xmath139 needing to also be continuous , the lowest impure crystal eigenfrequency must then have continuously traversed the lowest lying of the pure crystal asymptotes on its way , with the lowest lying pure crystal eigenfrequency then being an impure crystal eigenfrequency at the point at which the asymptote is reached .
since @xmath71 is infinite at any pure crystal eigenfrequency , for such a pure crystal eigenfrequency to also be a mode of the impure crystal , @xmath123 must equally be infinite at the pure crystal eigenfrequency , and it indeed is in this particular case since an increase in mass and an increase in force - constant act oppositely in @xmath69 , to thereby allow @xmath69 to indeed vanish at the lowest pure crystal eigenfrequency . an alternate way to modify the curve @xmath126 is not to change the mass at the defect site at all , but to keep @xmath7 fixed at @xmath6 and to instead start to allow an @xmath9 which is already bigger than @xmath8 to get even bigger .
such an increase will cause all of the impure crystal eigenfrequencies of fig .
( [ 3 ] ) to increase , but will never permit any of them to ever reach or traverse the next immediate pure crystal asymptote since a @xmath167 which only involves force - constant changes can never vanish at any non - zero @xmath92 .
we thus recognize two possible options as we start to vary parameters , namely that an impure crystal eigenmode can cross a pure crystal asymptote if @xmath69 has a zero there , or can not do so if @xmath69 has no zero .
if we thus picture the @xmath2 asymptotes in fig .
( 3 ) as dividing the @xmath168 region into @xmath169 compartments ( @xmath18 of which lie between pure crystal eigenfrequencies , with the other two lying below the lowest pure crystal eigenfrequency and above the largest one ) , we see that small changes in parameters can cause the locations of eigenmodes to either move slightly within any given compartment or to cross into an adjacent one , doing so in either case without radically altering the value of the total in - band difference @xmath88 $ ] contained on the left - hand side of eq .
( [ 15 ] ) . to see how this analysis pertains to non - central force - constant crystals with non - zero @xmath145
, we need to determine whether or not switching on @xmath145 can cause modes to change compartments .
to make such a determination , we note that according to eq .
( [ 42 ] ) , in the event of no mass change at the defect site , the eigenfrequencies associated with a general force - constant change in a nearest - neighbor crystal are given as the solutions to @xmath170 } \right ] \nonumber \\ & = & \frac{\omega_{\rm max}^2a_{xx}(0,0)}{2\omega^{\prime 2}[a^{\prime}_{xx}(0,0)-a_{xx}(0,0 ) ] } \left[\frac{a^{\prime}_{xx}(0,0)}{a_{xx}(0,0)}-\frac { \mu\hat{r}}{[1-\hat{r}]}\right]~~. \label{48}\end{aligned}\ ] ] since the pure crystal green s function @xmath146 has the same set of asymptotes as @xmath71 , whenever @xmath71 diverges , @xmath146 will do so also . however since @xmath171 is finite at frequencies at which @xmath146 is infinite , we see from eq .
( [ 48 ] ) that in the force - constant change case no matter what value we allow for @xmath145 , the perturbed eigenmodes are unable to ever leave their respective frequency compartments . despite this
though , since @xmath146 does have asymptotes , it can be expected that @xmath146 would change sign on its way between adjacent asymptotes ( though it could drop to some minimum and then go back to the next asymptote without ever changing sign ) , and thus we can anticipate that there will be some frequencies at which @xmath146 can take an assigned value equal to one . at such frequencies the right - hand side of eq .
( [ 48 ] ) would then diverge , and since @xmath146 ( and thus @xmath71 ) is not itself diverging at those points , such points would not be solutions to eq .
( [ 48 ] ) , but would instead be points between pure crystal eigenfrequencies at which the right - hand side of eq .
( [ 48 ] ) would undergo a discontinuity .
now while such a discontinuity can not take the eigenmode out of its frequency compartment , if the discontinuity is to have any effect at all , its only possible one would be to move the eigenmode out of the @xmath120 mode sector determinant altogether . specifically ,
unlike the case of a perturbation of a central force - constant crystal where the @xmath120 sector determinant accounts for all of the perturbed crystal eigenmodes , once we introduce non - central force - constants we have to consider the possibility that eigenmodes could move into the determinantal blocks of @xmath172 which are associated with the other irreducible octahedral representations of the cluster as given in eq .
( [ 38 ] ) .
however , recalling that all such eigenmodes are also controlled by eq .
( [ 20 ] ) , and recalling that all the pure lattice green s functions which appear in eq .
( [ 20 ] ) have the common asymptote structure associated with eq .
( [ 18 ] ) and thus a compartment structure identical to that of @xmath71 itself , we see that even if the eigenmodes do migrate into different irreducible @xmath118 sectors , since the left - hand side of the trace condition of eq .
( [ 15 ] ) is a sum over all of the modes of the crystal and not just over those associated with the @xmath120 sector alone ( a sum that thus includes the entire cluster of eq .
( [ 38 ] ) anyway ) , and since we are only making a small change on the right - hand side of eq .
( [ 15 ] ) , when the eigenmodes do move to other irreducible representations , they can still only move to equivalent compartments with the same frequency ranges as the ones they started in . as such then , independent of which specific sector the modes actually find themselves in after crossing any @xmath173 discontinuity , they could only cross into the same frequency compartments they were already in . however , once the eigenmodes have moved out of the @xmath120 sector , they would then be in some other irreducible @xmath118 representation where the @xmath120 mode eq .
( [ 48 ] ) would not apply . hence in these other representations additional changes in @xmath145 ( beyond the ones needed to first bring the force - constant dependent @xmath146 of eq .
( [ 46 ] ) to one ) could perhaps then move the eigenmodes into adjacent compartments .
however , for that to happen the eigenmodes would have to cross the pure crystal asymptotes , with those particular eigenmodes then needing to be common eigenmodes of both the pure and impure crystals at the requisite value of @xmath145 .
however , in order for an eigenmode to be an eigenmode at all , the eigenmode would need to be at a zero of the full @xmath172 , i.e. it would not only need to be at a zero of the determinantal block of some particular irreducible sector , but also to not give an infinite value to any of the other blocks which multiply the given determinantal block of interest in the full @xmath172 of the full crystal .
however , inspection of the @xmath120 mode @xmath143 of eq .
( [ 42 ] ) shows that it is actually divergent at a pure crystal eigenmode ( where it diverges as @xmath174 ) , and thus once there is no asymptote crossing in the @xmath120 mode sector , there can not be any in any other irreducible representation either . thus regardless of whether or not the @xmath173 discontinuities take the impure crystal eigenmodes out of the @xmath120 sector , the modes can not leave the frequency compartments they were already in when @xmath145 was zero .
hence if a mode already was in the compartment which lies beyond the pure crystal maximum eigenfrequency when the parameter @xmath145 was equal to zero , the mode will remain beyond the band no matter how @xmath145 might then vary .
moreover the in - band modes will remain in which ever compartments they had been in when @xmath145 was zero .
we thus recognize two specific consequences of switching @xmath145 on , namely that modes have to stay within their respective @xmath175 compartments , but that they are no longer prevented from traversing the zeroes of @xmath71 .
thus suppose we start off with a central force - constant crystal with @xmath175 and @xmath129 , and with @xmath176 conveniently set right at the threshold value of @xmath13 , and then slowly switch @xmath145 on .
the in - band modes could now move up or down , moving to no lower than the immediately previous pure crystal eigenmodes ( net shift of the in - band mode @xmath88 $ ] of down to @xmath177 ) , or moving upwards possibly as far as the next immediate pure crystal asymptotes ( net shift of up to @xmath178).@xcite then regardless of what value @xmath145 may actually take and regardless of which particular irreducible representation any specific eigenmode may actually lie in , eq .
( [ 41 ] ) will nonetheless still hold when there is no mass change since the trace is taken over the entire set of irreducible @xmath118 representations and not just over the @xmath120 mode .
on recalling that @xmath179 for a nearest - neighbor crystal , we see that the necessary condition that @xmath180 be positive is given by having the in - band modes move down as far as they possibly can , to thereby require @xmath9 to be positive , while the sufficient condition is given by having the in - band modes move up by as far as they possibly can , viz . @xmath181 with the condition that @xmath140 be greater than two thus being sufficient to guarantee the generation of localized modes by force - constant changes alone in an arbitrary nearest - neighbor cubic crystal .
to proceed beyond nearest neighbors is also not straightforward as no exact analog of eqs .
( [ 39 ] ) and ( [ 42 ] ) is currently known .
( for the case of second - nearest neighbors , the second - neighbor generalization of the irreducible decomposition of eq .
( [ 38 ] ) has been used to bring the problem to a reasonably manageable form.@xcite ) however , in the event of non - nearest neighbor force - constants , even though the cluster required for the matrix @xmath182 in eq .
( [ 20 ] ) would be much larger , and even though more pure lattice green s functions would be required , nonetheless every single one of these green s functions would still have exactly the same asymptote and compartment structure as @xmath71 .
thus if we start at the eigenmodes of a general nearest - neighbor crystal with non - zero @xmath145 and @xmath129 and slowly start to increase the strength of the non - nearest neighbor force - constants ( central or otherwise ) , we would again be constrained by the same set of pure crystal compartments in exactly the same way as before ( save that there would be yet more irreducible representations of @xmath118 to migrate to with all of them also contributing to the left - hand side of the trace condition ) , to again allow us to infer that even in the most general possible case imaginable ( arbitrary central and non - central force - constants and arbitrary number of participating neighbors ) , in the absence of any change in mass , the condition @xmath183 would still be sufficient to guarantee localized modes.@xcite the force - constants yielded by potentials for which @xmath185 is zero when the atoms are in their equilibrium positions are referred to as being central , while those associated with potentials which obey @xmath186 at equilibrium are referred to as being isotropic . even though @xmath153 involves a @xmath0-dimensional sum on both @xmath54 and @xmath187 , the @xmath188 term in the expression for @xmath153 only permits the @xmath187 sum to contribute once , since in the basis in which @xmath189 is diagonalized we can choose the polarization vectors @xmath190 so that only one polarization eigenvector has a non - zero @xmath40 component .
the function @xmath153 thus only possesses @xmath2 poles , poles then also possessed by @xmath191 and @xmath192 as well , with the full @xmath0 dimensionality of the eigenfrequencies then being recovered via the existence of eigenmode equations for all three of @xmath193 , @xmath194 and @xmath195 in eq .
( [ 21 ] ) . to reflect this our notation here
is that any eigenfrequency which appears once in an eigenmode summation from @xmath196 to @xmath197 will appear three times in an eigenmode summation from @xmath196 to @xmath198 . for any choice of the @xmath80 ratio ,
the curve @xmath79 is parallel to the @xmath75 axis in fig .
( 1 ) , lying above the line @xmath75 when @xmath199 and below the line @xmath85 when @xmath138 , being at @xmath200 when @xmath201 .
an initial glance at fig .
( 1 ) would appear to imply that no matter how close @xmath7 is to @xmath6 , as long as @xmath7 is less than @xmath6 there would be a solution to eq .
( [ 23 ] ) beyond the band maximum , with the @xmath202 condition then being both necessary and sufficient . however if this solution is as close to @xmath1 as @xmath1 itself is to the in - band mode just below it ( viz . within @xmath203 ) , the effect of @xmath7 would then be to just slightly displace the band maximum , with the wave function of the mode then being a plane wave which is slightly distorted at the defect site but not at all localized to it ; with it only being the condition that @xmath7 be less than @xmath115 which guarantees that the mode outside the band necessarily move away from the band edge ( i.e. all of the decrease in @xmath80 in going from one to @xmath204 could be used up entirely in moving the in - band modes to their allowed highest frequency maxima at the zeroes of @xmath71 . ) in passing we note that in order to determine whether there is a bona fide mode outside the band which is localized to the defect site , instead of using the trace condition we can also simply solve eq .
( [ 23 ] ) directly .
however , in doing so we need to allow for the fact that beyond the band edge @xmath71 falls precipitously from its asymptote at @xmath205 .
specifically , since @xmath71 is completely well - behaved outside the band and nowhere near infinite in value there , we can define its value at the band maximum as a continuation of its values outside the band .
this is equivalent to replacing @xmath71 by its principal value @xmath206 ( essentially one half of the sum of the values of @xmath71 on the two sides of the asymptote ) , to then yield an @xmath207 which is completely finite . with @xmath71 decreasing monotonically beyond the band maximum , localized modes
will then appear outside the band when @xmath208 .
since it can be shown on general grounds @xcite that for any cubic crystal the quantity @xmath209 defined this way has to lie in the range @xmath210 ( with the particular value that @xmath209 takes in any given crystal being dependent on that crystal ) , the quantity @xmath211 must lie in the range @xmath212 , with @xmath213 thus being a sufficient condition@xcite for producing a localized mode via a mass defect in any cubic crystal . for the face - centered cubic crystal for instance the normalized four - dimensional basis of displacements for the @xmath214 mode @xmath215 is given ( in face - centered cubic crystal site notation ) by @xmath216 , @xmath217 $ ] , @xmath218 $ ] , @xmath219 $ ] .
degenerate with this basis are two others , one based on @xmath220 and the other on @xmath221 . on the @xmath137
curve the loss of an intercept outside the band is compensated for by having an additional intercept in the region where the @xmath123 divergence is located , with all modes below the line @xmath222 being shifted upwards , and all modes above the line @xmath222 being shifted downwards .
after first intercepting the @xmath71 curve @xmath2 times in the positive region before reaching the point where it diverges , on reaching this point the function @xmath123 then crosses into the negative region with no further intercepts then being encountered . on the @xmath139 curve the absence of any intercept at all in the region where the @xmath123 divergence is located is compensated for by having an additional intercept outside the band , with all modes below the line @xmath223 being shifted downwards , and all modes above the line @xmath223 being shifted upwards . while the curve @xmath126 in fig .
( 2 ) shows that it is necessary to have @xmath12 be greater than one in order to shift frequencies upwards , this requirement is only necessary for the production of localized modes by central force - constant changes alone , with it being the requirement that @xmath12 be greater than @xmath13 which is then the sufficient one .
( all of the increase in @xmath12 in going from one to @xmath13 could be used up in moving the in - band modes to their allowed highest frequency maxima at the zeroes of @xmath71 . ) in passing we note that as such , this sufficiency condition initially appears to be at variance with eq .
( [ 39 ] ) , since eq .
( [ 39 ] ) would produce a mode right at the band maximum if @xmath224 and would thus suggest that @xmath225 is needed to produce a mode outside the band since @xmath209 can be small as one .
however , while we had indicated earlier that in general @xmath209 could be as small as one , for pure central force - constant cubic crystals it turns out that @xmath209 is actually altogether larger,@xcite making the @xmath226 condition the more powerful . for pure central force - constant crystals @xmath209 is formally found to take the value @xmath227 , a value which however very quickly comes down to order one when there is a small non - central force - constant admixture for rare gas solids for instance @xmath209 is given @xcite by @xmath228 even though there is only a two percent non - central component .
the utility of eq .
( [ 42 ] ) is that while the sum over all of the @xmath120 mode basis neighbor sites would initially generate a rather large number of pure lattice green s functions in eq .
( [ 20 ] ) , through repeated use of the green s function relations of eq .
( [ 19 ] ) one is , in the nearest - neighbor approximation , able@xcite to reduce the number of needed green s functions to a primitive set of just @xmath162 and @xmath154 .
the presence of any analogs of the zeroes of @xmath71 in the other irreducible @xmath118 sectors could prevent the eigenmodes from actually being able to go all the way to the next adjacent asymptotes . to be slightly more precise , in relating @xmath229 to @xmath230 in the non - nearest neighbor case
, one should also take into account the factor @xmath100 as given in eq .
( [ 31 ] ) , though we neglect this effect here . | using general properties of the crystal site representation normal mode matrix , we provide some very simple bounds on localized modes in simple , body - centered and face - centered cubic crystals with substitutional point defects .
we derive a trace condition constraint on the net change in crystal eigenfrequencies caused by the introduction of a defect , with the condition being a completely general one which holds for any combination of central and non - central crystal force - constants and for all - neighbor interactions . using this condition
we show that the sufficient condition for producing localized modes in an arbitrary cubic crystal by a mass change at the defect site is that the defect mass be less than one half of that of the host atom mass which it replaces , and that the sufficient condition for producing localized modes in an arbitrary cubic crystal by force - constant changes alone is that the defect site self force - constant be greater than twice that of the pure crystal self force - constant of the host atom which it replaces . |
ever since the realisation that cepheids are stars undergoing radial pulsation @xcite , the oscillation of stars has been an important research area . with much improved sensitivity , observations over the last few years have established a plethora of pulsating stars .
the best case by far is provided by the sun for which it is known @xcite that many high order pressure p - modes , and perhaps also the gravity g - modes and the coriolis restored r - modes , are excited . by combining observational data with theoretical models
researchers have been able to infer details of the sun s internal structure , e.g. the sound speed at different depths .
these impressive results of so - called `` helioseismology '' provide inspiration for further research into stellar oscillations and the hope that `` asteroseismology '' @xcite will help further unveil details of the structure of distant stars .
the purpose of this paper is to advance our modelling of non - radial oscillations of non - rotating , old and cold neutron stars that contain superfluid components . within the framework of general relativity
oscillating compact stars provide an interesting potential source of gravitational radiation .
there are several scenarios in which one would expect a neutron star to pulsate wildly , e.g. following its formation in a gravitational collapse , and it is interesting to ask whether the associated gravitational waves could be detected on earth .
should this be the case , one can hope to use the gravitational - wave data to probe the star s interior and possibly put constraints on the supranuclear equation of state @xcite . in the last few years
much attention has been focussed on gravitational - wave driven mode - instabilities in rapidly rotating stars .
ever since the seminal work of chandrasekhar , friedman and schutz @xcite in the 1970s it has been known that gravitational waves can drive various modes of oscillation unstable , and recently it has been shown that the r - modes are particularly susceptible to the instability @xcite .
most astrophysical r - mode scenarios regard hot young neutron stars , but it has also been proposed @xcite that the instability may operate in mature accreting neutron stars , e.g. in low - mass x - ray binaries .
these stars are expected to have core temperatures well below the superfluid transition temperature , and hence one would need to account for superfluidity in any detailed r - mode model .
this is particularly important since mutual friction may provide a strong dissipative mechanism on any mode in a superfluid star @xcite .
in addition to this , it has been argued that the presence of a viscous boundary layer at the base of the crust of a mature neutron star will lead to a very strong damping on the r - modes @xcite . while the issue of mutual friction has been discussed by lindblom and mendell @xcite , there are as yet no studies of dissipation due to a `` realistic '' core - crust interface , that is likely to involve the transition from a regime where superfluid neutrons co - exist with a lattice of crust nuclei to a region composed of superfluid neutrons and ( possibly superconducting ) protons
. further motivation for our current work is provided by results from attempts to model the equation of state at supranuclear densities .
many modern equation of state calculations ( within relativistic mean field theory ) predict a sizeable hyperon fraction at high densities . however , the hyperons act as an effective `` refrigerant '' which means that the star would cool very fast .
in fact , it has been argued that the presence of hyperons would lead to core temperatures well below those indicated by observations .
the favoured resolution to this problem corresponds to the hyperons being superfluid .
in addition , neutron stars may have cores composed of deconfined quarks which may also pair into exotic superfluid states .
in other words , if we want to understand the dynamics of the core of a realistic mature neutron star we have to allow for one ( or more ) partially decoupled superfluid components .
the need for an improved understanding of these problems was recently emphasized by the suggestion that the presence of hyperons in the deep core would lead to a strong bulk viscosity wich could potentially completely suppress the unstable r - modes @xcite .
this suggestion brings may difficult issues to the top of the agenda . for example , a detailed study of unstable modes must necessarily account for the fact that a superfluid component(s ) may move relative to the normal fluid component .
one might expect this to have a significant effect on the estimates of time scales for an unstable mode .
furthermore , as has been pointed out by andersson and comer @xcite , one would expect new classes of r - modes ( and indeed other inertial modes ) to exist in a superfluid star .
this means that the superfluid problem is likely to be richer than the ordinary , single fluid , one .
given that there have not yet been any detailed investigations into these issues it may be premature to draw any definite conclusions regarding the effect that superfluidity may have on mode - instabilities .
it is clear that a considerable amount of work on stars with one , or several , superfluid components remains to be done before we can claim to have an understanding of possible instabilities in mature neutron stars .
these problems provide strong motivation for our current work .
studies of oscillating superfluid stars were pioneered by epstein @xcite , who was the first to suggest that there ought to exist modes of oscillation that are unique to the superfluid case .
these modes have since been calculated both in newtonian theory @xcite and general relativity @xcite .
it is now well established that a simple two - fluid model of a superfluid neutron star core has two families of fluid pulsation modes .
the first of these is essentially the standard pressure p - modes , for which the two fluids tend to move together .
the superfluid modes , on the other hand , are distinguished by the fact that the two fluids are counter - moving @xcite .
another distinguishing characteristic of the superfluid modes is that their mode frequencies have a fundamental dependence on entrainment .
andersson and comer @xcite have used a local analysis of the newtonian superfluid equations to show that the modes are essentially acoustic in nature , with frequencies that depend on the stellar parameters as ^2_s c^2 _ , [ sfreq ] where @xmath0 is ( roughly ) the speed of sound in the proton fluid , @xmath1 and @xmath2 are the bare and effective proton masses , respectively , @xmath3 is the radial coordinate , and @xmath4 is the index of the relevant spherical harmonic @xmath5 .
these results have recently been confirmed by detailed mode - calculations @xcite .
the strong dependence of the superfluid modes on parameters that are difficult to constrain otherwise suggest a strategy that may be used in conjunction with future observations to narrow down the theoretical uncertainties @xcite .
specifically , an observational determination of the mode frequencies of an oscillating neutron star could be used to constrain the proton effective mass in dense nuclear matter .
this would have immediate implications for bcs energy gap calculations since the effective mass is part of the required information ( see , for instance , khodel , khodel , and clark @xcite ) . in this paper
we concentrate on the equations that describe polar perturbations ( even parity ) of a general relativistic superfluid neutron star , using as our starting point the formalism developed by comer , langlois , and lin @xcite .
the neutron star is assumed to consist of two distinct regions : a core consisting of matter in superfluid states , and an envelope of ordinary fluid matter .
we report progress in three important directions .
firstly , we have obtained the first ever results for gravitational - wave damping rates of the superfluid oscillation modes .
this is important as it allows us to assess the relevance of these modes for gravitational - wave astronomy .
secondly , we have developed a framework in which a superfluid region can be matched to an ordinary fluid region .
( lindblom & mendell @xcite have employed a similar model in the newtonian regime . )
the introduction of an ordinary fluid envelope is the first step towards allowing for the presence of superfluids that are confined to distinct regions in the star .
however , it is important to point out that since we do not incorporate elasticity in our model the envelope does not play the same role as the crust of a mature neutron star .
in essence , our model corresponds to a star in which the fluid degrees of freedom contain a conglomerate `` proton '' fluid ( consisting of nuclei and electrons in the envelope , and electrons and superconducting protons in the core ) that extends from the surface to the center of the star and superfluid neutrons that exist only in the core .
we work out the relevant junction / boundary conditions at the core / envelope interface and determine the quasinormal modes for such a neutron star model .
thirdly , we have included a simple model for the so - called entrainment effect , based on an explicit equation of state and an approximation that assumes that the fluid velocities are small compared to the speed of light .
this allows us to provide the first detailed results describing the effect that entrainment has on the oscillation modes of a superfluid star .
as we will show , this leads to the presence of so - called `` avoided crossings '' between neighbouring modes as the strength of entrainment is varied .
the main body of the paper begins with a discussion , in sec .
[ eqns ] , of the basic formalism used to describe non - radial , linearized oscillations of general relativistic superfluid neutron stars .
this contains a discussion of the background , spherically symmetric and static equilibrium configurations , and the introduction of the variables that are used to model the non - radial oscillations .
we also describe the key details of the numerical methods used to solve the linearized einstein / superfluid field equations .
the equation of state and inclusion of the entrainment effect are described in sec .
[ aem ] , and the results of our numerical analysis are presented in sec .
[ numres ] . in sec .
[ nonrad ] we prove that all pulsation modes of a superfluid star must radiate gravitational waves , unless the equation of state belongs to a very special class . finally , we consider in sec . [ gws ]
the question of whether or not superfluid modes , excited , for instance , during a pulsar glitch , can be reliably extracted from gravitational - wave data .
some final remarks are offered in the concluding sec .
for clarity of presentation and emphasis of the main physical results , we have relegated some of the formal , mathematical details to appendices .
appendix i is devoted to a derivation of the junction conditions used to tie together the core with the envelope . in appendix
ii we present an analytical equation of state that can be used to incorporate the lowest - order effects of entrainment .
finally , in appendix iii a new method is presented which can be used to accurately determine long - lived quasinormal modes .
we will be using geometrized units ( except in appendix ii where the speed of light is restored ) and `` mtw '' @xcite conventions throughout the paper .
in this section we summarise the equations that govern the equilibrium configurations of the superfluid core and the normal fluid envelope . what must be determined in the core are the neutron and proton number densities , and two metric coefficients . in the envelope
only the proton number density remains but there are still two metric coefficients to specify .
we also describe the complicated set of coupled linear perturbation equations that need to be integrated , and outline the computational strategy that we have adopted . in what follows
, the center of the star is at radial coordinate @xmath6 , the core / envelope interface will be at @xmath7 , and the surface at @xmath8 .
the formalism to be used here , and motivation for it , has been described in great detail elsewhere @xcite .
hence , we will only mention the key features here .
the central quantity of the superfluid formalism is the so - called master function @xmath9 .
it depends on the three scalars @xmath10 , @xmath11 and @xmath12 that can be formed from the conserved neutron ( @xmath13 ) and proton ( @xmath14 ) number density currents .
the master function is such that , when the two fluids are co - moving , @xmath15 corresponds to the total thermodynamic energy density . once the master function is provided ( see sec .
[ numres ] for a simple analytic equation of state ) the stress - energy tensor is given by t^_= ^_+ p^_+ n^ _ , where = - n^_- p^_[press ] is the generalised pressure , and @xmath16 are the chemical potential covectors .
we have also introduced = - x^2 , = -2 n^2 , = -2 p^2 .
the momentum covectors @xmath17 and @xmath18 are dynamically , and thermodynamically , conjugate to @xmath19 and @xmath20 .
the two covectors also make manifest the so - called entrainment effect , that affects the dynamics of a superfluid neutron star in a crucial way .
it is easy to see that the momentum of one constituent ( @xmath17 , say ) carries along some of the mass current of the other constituent if @xmath21 ( since @xmath17 is a linear combination of @xmath22 and @xmath23 ) . on the other hand , if @xmath24 , i.e. if the master function does not depend on @xmath25 , then there is no entrainment .
while the standard one - fluid problem can be expressed solely in terms of the einstein equations , with the fluid equations of motion being automatically satisfied `` by virtue of the bianchi identities '' , the two - fluid problem is different . because of the additional dynamic degrees of freedom associated with the second fluid we need to use also ( a subset of ) the fluid equations of motion .
the equations that need to be solved , in addition to the einstein field equations , are two `` continuity '' equations _
n^= _ p^= 0
. [ conteq]these equations represent conservation of the superfluid neutrons and the protons separately .
that is , we ignore `` transfusion '' from one component to the other due to , for example , weak interactions @xcite .
this is likely to be a reasonable approximation for the timescales and amplitudes that are relevant for nonradial pulsation .
we also have two euler - type equations n^ _ [ _ ] = p^ _ [ _ ] = 0 . [ euleq ] in order to determine the background fluid configuration we need to evaluate the associated metric .
we take our equilibrium configurations to be spherically symmetric and static , so the metric can be written in the schwarzschild form ds^2 = - e^ dt^2 + e^ dr^2 + r^2 ( d^2 + sin^2 d^2 ) .
[ bgmet ] the two metric coefficients are determined from two components of the einstein equations , which can be written in the form ^ = 1 - e^ r - 8 r e^ , ^ = - 1 - e^ r + 8 r e^ , [ bckgrnd ] where a primes represents a radial derivative , and it is to be understood that @xmath26 and @xmath27 in the interval @xmath28 and @xmath29 and @xmath30 in the interval @xmath31 . the equations that determine the radial profiles of @xmath32 and @xmath33 in the core have been derived in comer , langlois , and lin @xcite . they follow from ( [ euleq ] ) , and can be written ^0_0 ^ + ^0_0 ^ + 1 2 ( + a ) ^ = 0 , ^0_0 ^ + ^0_0 ^ + 1 2 ( + ) ^ = 0 , [ bgndfl_c ] where @xmath34 in the coefficients above , one sets @xmath35 after the partial derivatives are taken for the equilibrium configuration .
for the particular model equation of state we use in our numerical calculations , the above equations require that the two fluids have a common surface ( i.e. @xmath36 as @xmath37 ) if one assumes chemical equilibrium .
the oscillations of such models were studied in @xcite . here
we want to consider models such that the two - fluid regime is enclosed within a single fluid envelope .
it is easy to build such a model by relaxing the assumption of chemical equilibrium in the core .
the resultant models are likely not too unrealistic given the relatively long timescales required for nuclear interactions to reinstate equilibrium .
our models are such that the superfluid neutron number density is zero in the envelope , and thus the only matter equation is ^0_0 ^ + 1 2 c ^ = 0 , [ bgndfl_e ] where now = - 2 ^2 , 00 = + 2 ^2 ^2 .
there are three sets of `` boundary '' conditions that must be dealt with : a set at the center , one at the interface , and the remaining one at the surface of the star . in view of eq .
( [ bckgrnd ] ) , requiring a non - singular behavior at the center of the star will impose that @xmath38 , and consequently @xmath39 and @xmath40 must also vanish .
this in turn implies , in view of eq .
( [ bgndfl_c ] ) , that @xmath41 and @xmath42 have to vanish as well .
although our analysis of the junction conditions at the interface allows for other possibilities , our numerical calculations assume that the phase transition that leads to the formation of superfluid neutrons is second order .
in other words , we consider the energy density to be continuous at @xmath43 . from the analysis presented in appendix i
, we see that the two metric coefficients and the pressure must then also be continuous .
we will also assume that the proton number density and the proton chemical potential are continuous at the interface .
at the surface , we will only consider configurations that satisfy @xmath44 . a smooth joining of the interior spacetime to a schwarzschild vacuum exterior at the surface of the star implies that the total mass @xmath45 of the system is given by m = - 4 ^r_0 r^2 ( r ) dr and that @xmath46 . it is well - known that all non - trivial pulsation modes of a nonrotating fluid star correspond to polar pertubations ( often referred to as `` even parity '' ) .
in the so - called regge - wheeler gauge @xcite , the corresponding metric components are g _ = - e^i t p_l ( ) .
[ lemet ] where @xmath47 are the legendre polynomials .
this decomposition will be applied to both the core and the envelope . writing the neutron and
proton four - currents as @xmath48 and @xmath49 , where @xmath50 and @xmath51 , one can show that in the core the independent components are u^i = e^- /2 t ^i _ , v^i = e^- /2 t ^i _ , where _
^r = e^-/2r^l - 1 ( r ) p_l e^i t , _ ^ = - r^l - 2 ( r ) p_l e^i t , and _ ^r = e^-/2 r^l - 1 ( r ) p_l e^i t , _ ^ = - r^l - 2 ( r ) p_l e^i t .
the lagrangian variations in the neutron and proton number densities can be written as , respectively , @xmath52 and the conservation equations ( [ conteq ] ) for the particle number currents yield @xmath53 + { l ( l + 1 ) \over r^2 } v_\n - { 1 \over 2 } h_0 - k\right ) \ , \\ & & \cr { \delta p \over p } & = & - r^l \left(e^{- \lambda/2 } \left[{l + 1 \over r^2 } w_\p + { 1 \over r } w^{\prime}_\p \right ] +
{ l ( l + 1 ) \over r^2 } v_\p - { 1 \over 2 } h_0 - k\right ) \ . \label{lang}\end{aligned}\ ] ] thus , all matter variables can be expressed in terms of the velocity variables @xmath54 and @xmath55 . in the envelope , we have as the only independent matter variables @xmath56 and @xmath57 .
the set of perturbation equations that we solve for in the superfluid core have already been listed in @xcite , but since our core / envelope problem requires a slightly different method of solution we repeat the relevant equations here .
we also need these equations for the analysis in section v. first we define ( in analogy with lindblom and detweiler s @xcite approach to the one - fluid problem ) two new variables as - e^(- ) /2 ^ r ( 00 + 00 ) , [ xn ] and - e^(- ) /2 ^ r ( 00 + 00 ) . [ xs ] then we find that the einstein and superfluid field equations yield an algebraic constraint equation @xmath58 \ha + \left[{2 \omega^2 \over e^{\nu}}-{l(l+1)\over 2}e^\lambda\left({1-e^{- \lambda } \over r^2 } + 8 \pi \psi\right)\right ] \h1 \cr & & + \left[-2 e^{\lambda - \nu } \omega^2 + e^\lambda{l^2 + l - 2 \over r^2 } + e^{2 \lambda } \left({1 - e^{- \lambda } \over r^2 } + 8 \pi \psi \right)\left(1 - { 3 \over 2 } \left(1 - e^{- \lambda}\right ) - 4 \pi r^2 \psi\right)\right ] k \cr & & + 16 \pi e^{\lambda - \nu/2 } \left(\xn + \xs\right ) = 0 \ , \label{cnstrt}\end{aligned}\ ] ] and a system of coupled ordinary differential equations ( where we use the definition @xmath59 ) : @xmath60 \h1 + { e^\lambda \over r } k - 16 \pi { e^\lambda \over r } \left(\mu \n \vn + \chi \p \vs\right ) \ , \label{h1prime } \\ & & \cr k^{\prime } & = & { \ha \over r } + { l ( l + 1 ) \over 2 r } \h1 + \left[{\nu^{\prime } \over 2 } - { l + 1 \over r}\right ] k - 8 \pi { e^{\lambda/2 } \over r}\left[\mu \n \wn + \chi \p \ws\right ] \ , \\ & & \cr \wn^{\prime } & = & { e^{\lambda/2}r\over 2 } \ha + e^{\lambda/2 } r k - e^{\lambda/2 } { l ( l + 1)\over r } \vn - \left({l + 1 \over r } + { n^{\prime } \over \n}\right ) \wn + \cr & & + { \c00 \over \n^2 \d00 } \left[e^{(\lambda - \nu)/2 } r \xn + n^{\prime } \left(\b00 \n \wn + \a00 \p \ws\right)\right ] - \cr & & { \a00 \over \n \p \d00 } \left[e^{(\lambda - \nu)/2}r \xs + \p^{\prime } \left(\a00 \n \wn + \c00 \p \ws\right)\right ] \ , \label{wnprim } \\ & & \cr \ws^{\prime } & = & { e^{\lambda/2 } r \over 2 } \ha + e^{\lambda/2 } r k - e^{\lambda/2 } { l ( l + 1 ) \over r } \vs - \left({l + 1 \over r } + { \p^{\prime } \over \p}\right ) \ws \cr & & + { \b00
\over \p^2 \d00 } \left[e^{(\lambda - \nu)/2 } r \xs + \p^{\prime } \left(\c00 \p \ws + \a00 \n \wn\right)\right ] - \cr & & { \a00\over \n \p \d00 } \left[e^{(\lambda - \nu)/2}r \xn + \n^{\prime } \left(\a00 \p \ws + \b00 \n \wn\right)\right ] \ , \\ & & \cr \xn^{\prime } & = & - { l \over r } \xn + { e^{\nu/2 } \over 2 } \left[\n \mu \left({1 \over r } - \nu^{\prime}\right ) - \n^{\prime } \left(\b00 \n + \a00 \p\right)\right ] \ha \cr & & + \mu \n \left[{e^{\nu/2 } \over 4}{l ( l + 1 ) \over r } + { \omega^2 \over 2 } r e^{-\nu/2}\right ] \h1 \cr & & + e^{\nu/2}\left[\mu \n \left({\nu^{\prime } \over 4 } - { 1 \over 2 r } \right ) -
\left(\b00 \n + \a00 \p\right ) \n^{\prime}\right ] k + { l ( l + 1 ) \over r^2 } e^{\nu/2 } \n^{\prime } \left(\b00 \n \vn + \a00 \p \vs\right ) \cr & & - e^{(\lambda - \nu)/2 } { \omega^2 \over r } \n \left(\b \n \wn + \a \p \ws\right ) - 4 \pi e^{(\lambda+\nu)/2 } { \mu \n \over r } \left(\mu \n \wn + \chi \p \ws\right ) \cr & & + e^{-(\lambda - \nu)/2 } \left[- { \n^{\prime } \over r } \left ( \b00^{\prime } \n \wn + \a00^{\prime } \p \ws\right ) + \left({2 \n^{\prime } \over r^2 } + { \lambda^{\prime } - \nu^{\prime } \over 2 r } \n^{\prime } - { \n^{\prime \prime } \over r}\right)\right . \cr & & \left.\left(\b00 \n \wn + \a00 \p \ws\right)\right ] \ , \\ & & \cr \xs^{\prime } & = & - { l \over r } \xs + { e^{\nu/2 } \over 2 } \left[\p \chi \left({1 \over r } - \nu^{\prime}\right ) - \p^{\prime } \left(\c00 \p + \a00 \n\right)\right ] \ha \cr & & + \chi \p \left[{e^{\nu/2 } \over 4 } { l ( l + 1 ) \over r } + { \omega^2 \over 2 } r e^{-\nu/2}\right ] \h1 \cr & & + e^{\nu/2 } \left[\chi \p \left({\nu^{\prime } \over 4 } - { 1 \over 2 r } \right ) -
\left(\c00 \p + \a00 \n\right ) \p^{\prime}\right ] k + { l ( l + 1 ) \over r^2 } e^{\nu/2 } \p^{\prime } \left(\c00 \p \vs + \a00 \n \vn\right ) \cr & & - e^{(\lambda - \nu)/2 } { \omega^2 \over r } \p \left(\c
\p \ws + \a \n \wn\right ) - 4 \pi e^{(\lambda + \nu)/2 } { \chi \p \over r } \left(\chi \p \ws + \mu \n \wn\right ) \cr & & + e^{- ( \lambda - \nu)/2 } \left[- { \p^{\prime } \over r } \left ( \c00^{\prime } \p \ws + \a00^{\prime } \n
\wn\right ) + \left({2 \p^{\prime } \over r^2 } + { \lambda^{\prime } - \nu^{\prime } \over 2 r } \p^{\prime } - { \p^{\prime \prime } \over r}\right)\right.\cr & & \left.\left(\c00 \p \ws + \a00 \n \wn\right)\right ] \ . \label{xpprime}\end{aligned}\ ] ] the equations for the ordinary fluid envelope are obtained by taking the @xmath61 limit of the field equations .
it is , however , not quite as straightforward as letting @xmath36 in the above set of two - fluid perturbation equations .
the reason for this is that we have in places divided through by @xmath62 which vanishes in the one - fluid limit . in the one - fluid case ,
the constraint equation becomes @xmath63 \ha + \left[{2 \omega^2 \over e^{\nu } } - { l ( l + 1)\over 2 } e^\lambda \left({1 - e^{- \lambda } \over r^2 } + 8 \pi \psi\right)\right ] \h1 \cr & & \cr & & + \left[- 2 e^{\lambda - \nu } \omega^2 + e^\lambda{l^2 +
l - 2 \over r^2 } + e^{2\lambda } \left({1 - e^{- \lambda } \over r^2 } + 8 \pi \psi \right ) \left(1 - { 3 \over 2 } \left(1 - e^{- \lambda}\right ) - 4 \pi r^2 \psi\right)\right]k \cr & & \cr & & + 16 \pi e^{\lambda - \nu/2 } \xs = 0 \ . \label{envcon}\end{aligned}\ ] ] the other two equations for the metric are @xmath64 \h1 + { e^\lambda \over r } k - 16 \pi { e^\lambda \over r } \chi \p \vs \ , \label{h1env } \cr & & \cr k^{\prime } & = & { \ha \over r } + { l ( l+1 ) \over 2 r } \h1 + \left [ { \nu^{\prime } \over 2 } - { l + 1 \over r}\right ] k - 8 \pi { e^{\lambda/2 } \over r } \chi \p \ws \ . \label{kenv}\end{aligned}\ ] ] the final two equations are for the fluid and they take the form @xmath65 \ha + \chi \p \left [ { e^{\nu/2 } \over 4}{l ( l + 1 ) \over r } + { \omega^2 \over 2 } r e^{-\nu/2}\right ] \h1 \cr & & \cr & & + e^{\nu/2 } \left[\chi \p \left({\nu^{\prime } \over 4 } - { 1 \over 2 r}\right ) - \c00 \p \p^{\prime}\right ] k + { l ( l + 1 ) \over r^2 } e^{\nu/2 } \p^{\prime } \c00 \p \vs \cr & & \cr & & - e^{(\lambda-\nu)/2 } { \omega^2 \over r } \c \p^2 \ws - 4 \pi e^{(\lambda+\nu)/2 } { ( \chi \p)^2 \over r } \ws \cr & & \cr & & + e^{-(\lambda-\nu)/2 } \left[-{\p^{\prime } \over r } \c00^{\prime } \p \ws + \left({2 \p^{\prime } \over r^2 } + { \lambda^{\prime}-\nu^{\prime } \over 2 r } \p^{\prime } - { \p^{\prime \prime } \over r}\right ) \c00 \p \ws\right ] \ , \label{xpenv}\end{aligned}\ ] ] where = - e^(-)/2 ^ r ( 00 ) .
this set of equations is identical to that previously used @xcite to study the oscillations of normal fluid neutron stars for a range of supranuclear equations of state . at the center of the star ,
the conditions are those given in appendix a of comer , langlois , and lin @xcite , i.e. all functions are regular . at the surface ,
the conditions are the one - fluid conditions used by detweiler and lindblom @xcite .
the main difference here concerns the interface .
the detailed treatment of the interface is discussed in appendix i. it is found that the relativistic junction conditions imply that the three metric perturbations @xmath66 and @xmath67 must be continuous at the interface .
we assume that @xmath68 is free to vary at the interface , i.e. , the value it takes at the interface is determined by the general solution produced for the core .
we also assume that @xmath69 , which will be shown below to be consistent with the chosen equation of state . from the results presented in appendix i
, it is found that two conditions remain .
one of these implies that @xmath70 must be continuous at the interface .
recalling that we assume the proton number density and proton chemical potential to be continuous at the interface , i.e. that we have a second order phase - transition , the last condition is that @xmath56 must also be continuous at the interface .
even though our strategy for integrating the perturbation equations is similar to that used by comer , langlois , and lin @xcite there are some subtleties associated with the presence of the one - fluid envelope .
hence , it is worthwhile outlining the approach we have taken to the problem . in the core the perturbation equations can be written as the matrix equation = * q * , [ me ] where = \{h_1,k , w_,w_,x_,x _ } is an abstract six - dimensional vector field .
the @xmath71 matrix @xmath72 depends on @xmath4 , @xmath73 , the background fields , and the various coefficients @xmath74 , @xmath75 , etc . as was shown by comer , langlois , and lin one need
only specify the set of values @xmath76 at the center of the star .
the remaining variables , @xmath77 , @xmath78 , and @xmath79 , then follow from the @xmath80 limit of the perturbation equations .
all of the second derivatives , @xmath81 , @xmath82 , etc , are likewise determined by @xmath76 .
this means that , in order to provide information that enables us to construct the general solution , an integration starting from the centre must generate three linearly independent solutions @xmath83 , @xmath84 , and @xmath85 .
the corresponding general solution can thus be written ( r ) = ^3_i = 1 c_i * y*_i(r ) , where the @xmath86 ( @xmath87 ) are constants to be determined . in the envelope ,
the problem is equivalent to the standard one for a single fluid and our strategy is identical to that of lindblom and detweiler @xcite .
we write the perturbation equations as = * * , where = \{h_1,k , w_,x _ } and the matrix @xmath88 can be deduced from eqns .
( [ envcon])([xpenv ] ) . at the surface of the star
our solution must satisfy the single condition @xmath89 ( cf .
this means that we must generate three linearly independent solutions in the envelope and the general solution can therefore be written ( r ) = ^6_i = 4 c_i * * _ i(r ) . at the core / envelope interface
we must enforce the additional condition that @xmath90 .
we also know , from the analysis in appendix i , that the variables @xmath91 , @xmath67 , @xmath92 and @xmath93 should all be continuous across the interface .
this means that only the function value @xmath94 remains unspecified .
this is , however , as it should be since the interface represents a `` free boundary '' for the superfluid neutrons .
( an analogy can be made with the water / air interface of the oceans , where the water oscillation is free at the interface . ) our strategy for solving the problem is based on two steps .
first we continue the three solutions from the envelope into the core assuming in addition that @xmath95 .
then we determine a fourth solution ( needed to generate the general solution in the core ) by assuming that @xmath96 , but that all the other variables vanish at the interface .
this means that we have determined a general solution of form ( r ) = ^7_i = 4 c_i * y*_i(r ) , where @xmath97 for @xmath98 .
the remaining step is to match the various solutions at some point @xmath99 in the core @xmath100 . at @xmath101
we must have ^3_i = 1 c_i * y*_i(r_m ) = ^7_i = 4 c_i * y*_i(r_m ) . [ eq ]
once we have provided the overall normalisation ( by specifying one of the @xmath102 coefficients ) , this problem can readily be solved for the remaining coefficients .
this completes the solution of the interior problem . to determine the global solution we must also solve for the exterior metric perturbations
this problem reduces to that of integrating the zerilli equation , cf .
@xcite . finally , in order to find a quasinormal mode ( qnm ) of the system we need to identify solutions that correspond to purely outgoing waves at spatial infinity .
our method for determining long - lived qnms is described in appendix iii .
in order to extend the previous work by comer , langlois , and lin @xcite further we want to construct a sensible equation of state that incorporates the entrainment effect .
this effect arises whenever there is a coupling between two interpenetrating fluids , and has the net result that the momentum of one fluid is not simply proportional to that fluid s velocity .
rather , it is a linear combination of the velocities of both fluids .
hence , when one constituent starts to flow it will necessarily induce a momentum in the other
. however , the entrainment effect is poorly understood , and there are as yet no completed relativistic models that we can base our discussion on .
( the problem is currently being considered by comer and joynt , and we are hopeful that a fully relativistic formalism will soon be available . ) the involved microphysics is , in fact , so uncertain that the best strategy corresponds to introducing some convenient parameterisation and then studying whether a variation in the chosen parameters affects the overall properties of the star and/or it s modes of pulsation .
we noted earlier , in sec .
[ eqns ] , that the two chemical potential covectors @xmath103 and @xmath104 make manifest the entrainment effect by being linear combinations of the conserved four - currents , @xmath105 and @xmath106 .
ultimately , this traces back to the master function @xmath9 depending explicitly on the `` entrainment variable '' @xmath25 .
its presence in the background and perturbation equations listed in sec .
[ eqns ] is most apparent through the values of the coefficients @xmath107 and @xmath108 . in the following
we will use an expansion in terms of @xmath25 as outlined in appendix ii . at the heart of this expansion are the dimensionless ratios of the neutron and proton three - velocities with respect to the speed of light .
this expansion makes sense since one would expect these ratios to be extremely small under most circumstances .
then the expansion becomes particularly accurate .
the equation of state so constructed should have applications to the studies of quasinormal modes of neutron stars , with both static spherically symmetric and slowly rotating backgrounds .
we thus expand the master function as ( n^2,p^2,x^2 ) = _ i = 0^ _ i(n^2,p^2 ) ( x^2 - ) ^i , [ expand ] where @xmath109 is expected to be small with respect to @xmath110 .
it should be noted that there are combinations other than @xmath109 that could be used in the expansion , the most obvious being those that have no dimensions , say , @xmath111 .
these combinations are , however , not convenient in that the @xmath74 , @xmath112 , etc coefficients require that partial derivatives with respect to @xmath113 and @xmath114 be taken .
the effect of this is that the expansions for @xmath74 , @xmath112 , etc would then not take the same form as eq .
( [ expand ] ) above , since every derivative would bring in extra factors of @xmath25 outside the terms @xmath115^i$ ] . a quick glance at appendix ii will show that if we use an expansion based on @xmath109 the basic form of the expansion is preserved for all coefficients that enter the field equations . in order to make contact with previous newtonian studies of oscillating superfluid stars we will adopt the particular entrainment model used by lindblom and mendell @xcite . to do this , a point of connection
must be made between the coefficients used in the general relativistic superfluid equations and the corresponding newtonian equations .
this connection can be made by taking the newtonian limit of the general relativistic superfluid formalism and then comparing the dynamical variables ( as in @xcite ) . in this way the mass matrix components @xmath116 , @xmath117 , and @xmath118 of the newtonian formalism ( cf .
@xcite ) can be directly connected to the analytical entrainment coefficient @xmath119 of the general relativistic formalism ( cf .
appendix ii ) and the particle number densities @xmath120 and @xmath121 . following this strategy
we find that @xmath119 can be written as _ 1 = - c^2 m _
m _ _ ^2 - _ _ _ , where m _ = _ + _ , m
_ = _ + _ . where @xmath122 is the neutron ( proton ) mass . the particular model of lindblom and mendell @xcite sets _ = - m _ , where @xmath123 is a constant .
in the following we will refer to @xmath123 as the `` entrainment parameter . '' from the analysis of prix , comer and andersson @xcite one can infer that = m_m _ ( m_m^ * _ - 1 ) , where @xmath2 is the proton effective mass .
given that @xmath124 ( see , for instance , sjberg @xcite ) `` typical '' values for a neutron star core may lie in the range @xmath125 . in the following
we take this range as being `` physically reasonable '' . at this point
it is relevant to note that , after all the numerical work described in this paper had been completed , prix et al @xcite developed an alternative description of entrainment .
while we could , in principle , have adopted our formulas here to this new description we have decided not to do this .
such a change would not have affected our discussion or the implications of our results in any way .
we will now apply our formalism to a simple model equation of state .
we consider the case where each core fluid ( at the lowest order in the expansion presented in appendix ii ) behaves as a relativistic polytrope , i.e. we take _ 0(^2,^2 ) = - m_- _ ^ _ - m_- _ ^ _ .
this master function is clearly separable in @xmath126 and @xmath127 . using the formalism developed in appendix ii , the relevant coefficients are ( for the equilibrium configurations where @xmath35 ) = , 00 = 0 , = m _ + _ _ ^_- 2 - , 00 = _ _ ( _ - 1 ) ^_- 2 , and = m _ + _ _ ^_- 2 - , 00 = _ _ ( _ - 1)^_- 2 .
the pressure and chemical potentials of the core are @xmath128 furthermore , one can verify that these coefficients and thermodynamic variables are such that the variable @xmath129 must vanish at the core / envelope interface , cf .
( [ xn ] ) . of course
, we must also consider what happens in the envelope .
this is much simplified because there entrainment is not relevant .
we get ( ^2 ) = - m_- _ ^ _ .
the relevant coefficients are = m _ + _ _ ^_- 2 , 00 = _ _ ( _ - 1)^_- 2 , and the pressure and chemical potential are given by @xmath130 we have considered two stellar models . the first model is identical to model 2 of comer , langlois , and lin @xcite , and has no envelope .
the second model has a significant envelope , and a more realistic ( slightly smaller ) proton fraction in the core .
the relevant parameters that determine the two models are listed in table [ modtab ] . in particular
, the features of the core / envelope model are illustrated in fig .
[ backfig ] .
the figure shows the radial profiles of the background neutron and proton particle number densities , @xmath32 and @xmath33 , respectively .
the model has been constructed to have the following features that are reckoned to be characteristic of real neutron stars : ( i ) a mass of about @xmath131 , ( ii ) a total radius of about @xmath132 , ( iii ) an envelope of roughly one kilometer thickness , and ( iv ) a central proton fraction of about @xmath133 .
.parameters describing our stellar models i and ii .
model i is identical to model 2 of @xcite , and has no envelope .
model ii , on the other hand , has an envelope of roughly 1 km and could be seen as a slightly more realistic neutron star model . [
cols="^,^,^",options="header " , ] this does not , however , mean that we should give up on the main idea behind this analysis .
we simply have to acknowledge that we are likely to require significant improvements in technology if we are to be able to make this into a viable approach .
but it seems inevitable that the available technology will improve over the next decades .
in fact , various groups are already discussing possible improvements in detector sensitivity that may be achievable in the future . in order to illustrate the levels that are being discussed , we consider the so - called euro detector , for which the sensitivity has been estimated by sathyaprakash and schutz ( for further details see @xcite ) .
we consider two possible configurations : in the first , the sensitivity at high frequencies is limited by the photon shot - noise , while the second configuration reaches beyond this limit by running several narrowbanded ( cryogenic ) interferometers as a `` xylophone '' .
the corresponding noise - curves are illustrated , and compared to the current generation of interferometers , in figure [ figure1 ] .
it is immediately clear from figure [ figure1 ] that a euro detector would be a superb instrument for studying pulsating neutron stars .
this means that previously suggested strategies @xcite for unveiling the supranuclear equation of state may eventually be put to the test .
in fact , as is clear from the estimates in table [ ftab ] , where we list the signal - to - noise ratio estimated from ( [ sign ] ) , one should also be able to detect the superfluid oscillation modes .
the various modes would be marginally detectable given this level of excitation and a third generation detector limited by the photon shotnoise .
if this limit can be surpassed by configuring several narrowbanded interferometers as a xylophone , the achievable signal - to - noise ratio will be excellent .
hence , it seems plausible that one could infer the parameters of neutron star superfluidity . as an obvious `` by product '' one might hope to shed light on the mechanism for pulsar glitches .
we can also confirm that , provided that the modes can be detected , the oscillation frequencies can be extracted with good accuracy from the data . by combining eqns .
( 11 ) and ( 12 ) from kokkotas et al @xcite with ( [ sign ] ) one can obtain a formula for the signal - to - noise required to determine the mode frequency with a relative error @xmath134 .
we get ( since @xmath135 for the modes under consideration ) a relation ( s n ) 0.33 ( 1 t_d ) ( f ) ( _ f f ) ^-1 from this relation we can deduce that a detection with signal - to - noise of ( say ) 10 would enable one to infer the fluid mode frequencies with an accuracy of the order of a percent or so . with this level of precision
one should be able to distinguish clearly between the `` normal fluid '' f and p - modes and the superfluid s - modes .
in other words , we would have the information required not only to infer the mass and radius of the star @xcite , we could also hope to constrain the parameters of neutron star superfluidity .
the main motivation behind the work presented in this paper is the fact that neutron star physics is not adequately modelled within newtonian gravity .
it is well known that newtonian results differ greatly from the correct relativistic ones already at the level of determining the mass and radius of a star with a prescribed central density from a given `` realistic '' equation of state . as far as the qnms of the star
are concerned , newtonian studies are extremely useful since they help us understand the physics of different classes of modes .
but at the same time it is clear that if we require a detailed model of the oscillation spectrum of an astrophysical neutron star we must approach the problem from the relativistic point of view .
this is particularly crucial if we are interested in the gravitational - wave damping rates .
furthermore , since mature neutron stars are likely to contain several superfluid components , it is important that we develop a framework for modelling multi - fluid systems in full general relativity .
the present paper represents significant progress towards this goal .
we have considered a core - envelope model , with superfluid neutrons being present only in the core of the star while the outer region is composed of an ordinary fluid .
our model includes a simple , yet reasonable , model for entrainment and we have shown how neighbouring qnms undergo avoided crossings as the entrainment parameter is varied . finally , we have ruled out the possibility that a two - fluid star could have non - radiative pulsation modes , and discussed the possible detection of gravitational - wave signals from oscillating superfluid neutron stars . in the near future , our aim is to turn our attention to the oscillations of rotating superfluid stars .
this is an exciting problem area , since various modes of oscillation may be driven unstable by the emission of gravitational waves .
of particular recent interest are the so - called r - modes , and the damping due to superfluid mutual friction . in a recent work , lindblom and
mendell @xcite showed that mutual friction was effective at damping unstable r - modes only for very particular values of the entrainment parameter .
andersson and comer @xcite have speculated that this peculiar behavior might be due to the existence of avoided crossings between two classes of r - modes ( analogous to the two classes of acoustic fluid modes discussed in the present paper ) .
the results of the present paper lend support to this idea .
yet it is clear that our understanding of the various involved issues is far from complete , and that a considerable amount of work remains to be done before we can model the dynamics of rotating superfluid neutron stars in a `` realistic '' way .
we thank brandon carter and john clark for their excellent answers to several questions that arose as this work was being pursued .
we are also grateful to reinhard prix for useful discussions and a careful reading of a draft version of this paper .
na gratefully acknowledges support from the leverhulme trust via a prize fellowship , pparc grant ppa / g/1998/00606 as well as the eu programme improving the human research potential and the socio - economic knowledge base ( research training network contract hprn - ct-2000 - 00137 ) .
glc gratefully acknowledges partial support from a saint louis university slu2000 faculty research leave award , and epsrc visitors grant gr / r52169/01 in the uk .
in this appendix will be presented a geometrical approach to deriving the junction conditions that must be used to smoothly join together an inner superfluid core with an exterior normal fluid envelope .
the junction conditions will be obtained via an analysis of the first and second fundamental forms associated with the ( in general , timelike ) hypersurfaces given by the level surfaces of the generalized pressure @xmath136 ( cf .
( [ press ] ) ) .. however , we believe that the pressure is the most natural choice from both mathematical and physical points of view . ] if there are no `` delta - function - like '' discontinuities in the pressure , then the first and second fundamental forms will be continuous throughout the star @xcite . turning the problem around , by demanding the continuity of the first and second fundamental forms a smooth joining at the core / envelope interface can be achieved .
consider that the level surfaces of @xmath136 are timelike , i.e. the normal to these surfaces is given by ^ = g^ _ , where @xmath137 .
the so - called first fundamental form @xmath138 ( i.e. the induced three - metric ) of these level surfaces is _
= ^ _ ^ _ g _ , where the `` perp '' operator @xmath139 is given by ^ _
= ^ _ - n^ n _ . the second fundamental form @xmath140 ( i.e. the extrinsic curvature ) of the level surfaces is defined as k _ = - ^ _ ^ _ _ ( n _ ) , where the parentheses imply symmetrization of the indices .
let us consider the pressure to be of form ( t , r , ) = _ 0(r ) + ( t , r , ) .
the components of the unit normal @xmath141 are thus found to be @xmath142 the non - zero components of the first fundamental form are @xmath143 and we also find @xmath144 for the non - trivial components of the second fundamental form . given that a smooth background can be constructed independently of the oscillations , we can assume that the background and linearized pieces of @xmath138 and @xmath140 are separately continuous at the core - envelope interface .
we will consider the background pieces first .
for clarity of presentation the matter and metric variables in the envelope will be distinguished by a tilde .
the continuity of the first fundamental form thus yields for the background ( r_c ) = ( r_c ) , and continuity of the second fundamental form implies ^(r_c ) = ^(r_c ) , e^(r_c ) = e^(r_c ) . using the background einstein equations , and defining e^- = 1 - 2 m(r)/r , e^- = 1 - 2 ( r)/r , where m(r ) = - 4 _ 0^r r^2 _
0(r ) dr , ( r ) = - 4 _ r_c^r r^2 _ 0(r ) dr + c , we find that the background junction conditions imply @xmath145 and @xmath146 , where @xmath147 and @xmath148 are the pressure and energy density of the envelope , respectively .
it is also useful to note that the tolman - oppenheimer - volkov equations for the core and envelope imply = _
0(r_c ) - _ 0(r_c ) _ 0(r_c ) - _ 0(r_c ) . [ pderivcont ] before dealing with the linear perturbations , it will be convenient to write out the linearized pressure as a function of the fundamental matter and metric variables , for both the core and the envelope .
we have used the field equations to help simplify the formulas .
the final forms are ( for the radial dependence ) = r^l and = r^l where we have again put a tilde over all the linearized variables associated with the envelope .
now , it can be seen that the junction conditions imply that the metric perturbations are continuous at the core / envelope interface , i.e. @xmath149 the matter variables , on the other hand , must satisfy two conditions , which are ( r_c ) ( r_c ) _ ( r_c )
= ( r_c ) n(r_c ) w_(r_c ) + ( r_c ) ( r_c ) w_(r_c ) and @xmath150 it is important to notice that the junction conditions do _ not _ imply that @xmath151 at the core / envelope interface , but rather that @xmath152 .
this fact is the clearest demonstration why a geometrical approach is crucial for obtaining the correct junction conditions , since this particular result is a direct consequence of the fact that schwarzschild - like coordinates cause derivative discontinuities whenever the energy density is not continuous . if it is the case that @xmath153 , then we can see from eq .
( [ pderivcont ] ) that the junction conditions _ will _ imply @xmath151 .
having a continuous energy density also implies that @xmath154 .
in this appendix we develop the analytical equation of state used in the main part of the text .
the essential strategy is to introduce an expansion based on the assumption that the fluid velocities are small compared to the speed of light .
this is a reasonable assumption for neutron star pulsations .
recall that the entrainment variable @xmath25 is given by @xmath155 .
it is convenient to write each conserved four - current as in the main text : ^= u^ , ^= v^ , except that now we take @xmath156 and @xmath157 where @xmath158 is the speed of light .
if @xmath159 and @xmath160 denote the proper times of the neutron and proton fluid elements , respectively , then the worldlines of each fluid element are obtained from the functions x_^ ( _ ) = ( t(_),x_^i ( _ ) ) , x_^ ( _ ) = ( t(_),x_^i ( _ ) ) .
the `` unit '' four - velocities are thus given by u^ = dx_^d _ , v^ = dx_^d _ .
consider , for the moment , a region within the fluid that is small enough that the gravitational field does not change appreciably across the region . in this case
, a locally minkowski coordinate system can be set up and the metric can be approximated by the flat metric : ds^2 = - d(c t)^2 + _ ij dx^i dx^j . letting @xmath161 and @xmath162 , as well as @xmath163 and @xmath164 , one can show that the four - velocity components can be written as u^0 = c , u^i = u_^i , and v^0 = c , v^i = v_^i . with this decomposition
one finally obtains for the entrainment variable x^2 = ( 1 - _ i j ( u_^i / c ) ( v_^j / c ) ) .
if it is the case that the individual three - velocities are small with respect to the speed of light , i.e. that < < 1 , < < 1 , then it will be true that @xmath165 to leading order in the ratios @xmath166 and @xmath167 .
this basic fact will be at the heart of the expansion considered below .
given what was just discussed , it makes sense to consider equations of state that can be expanded like ( n^2,p^2,x^2 ) = _ i = 0^ _ i(n^2,p^2 ) ( x^2 - ) ^i , since @xmath109 is small with respect to @xmath110 . in terms of this expansion , one can show the following for the @xmath74 , @xmath75 , etc .
coefficients that appear in the field equations : @xmath168 it is important to note that @xmath75 vanishes if the master function is such that @xmath169 are separable in @xmath126 and @xmath127 .
the utility of this expansion is especially apparent for the quasinormal mode calculations , because when any of the coefficients are evaluated on the background , then one sets @xmath35 , and thus only the first few @xmath169 are needed .
in fact , in our analysis we only retain @xmath170 and @xmath119 , where the latter contains the information concerning the entrainment effect .
the problem of calculating quasinormal modes of relativistic systems , such as black holes and neutron stars , is in many ways far from trivial . in general
, a strategy for finding qnms has to involve a prescription for imposing a pure outgoing wave boundary condition at infinity for a linear second - order differential equation . in the context of the present paper
the relevant equation is the one derived by zerilli @xcite : z = 0 .
[ zerilli ] here @xmath171 is the standard tortoise coordinate and we have assumed that the perturbed quantities have a harmonic dependence on time , i.e. behave as @xmath172 .
the effective potential @xmath173 is rather complicated but here we need only know that it is such that the behaviour of a general solution at spatial infinity ( as @xmath174 ) is z ~a_out ( ) e^-ir _ + a_in ( ) e^ir _ . a qnm of the system is a solution that combines some physical constraints ( no waves coming out of the event horizon in the case of a black hole or a regular solution to the equations for the interior of a star ) with purely outgoing waves , @xmath175 , at infinity .
two typical difficulties arise in the determination of such mode solutions .
both are due to the fact that the qnms are damped by gravitational radiation emission , and thus the qnm frequency @xmath176 must be complex ( with a positive imaginary part unless the mode is unstable ) .
this means that an outgoing - wave solution to eq .
( [ zerilli ] ) will be exponentially growing as @xmath174 and one would , in principle , need exponentially high numerical precision to discard the ingoing solution .
this problem is particularly challenging for rapidly damped modes , like those of black holes and the neutron star w - modes .
a second difficulty that arises is relevant for the p - modes of a neutron stars .
the high order p - modes are damped very slowly by gravitational radiation .
thus , the characteristic frequencies are such that the imaginary part is many orders of magnitude smaller than the real part .
although not as conceptually challenging as the first qnm problem , the difficulty associated with extremely small imaginary parts typically prohibits the determination of any but the first few of the neutron star p - modes . in this appendix
, we describe a new method for dealing with this problem .
the various mode results presented in the main body of the paper were obtained within this scheme .
let us assume that our relativistic system has a qnm with complex frequency @xmath176 .
then the solution to the zerilli equation is such that @xmath177 as @xmath178 whereas @xmath179 asymptotically .
from the fact that it is the frequency squared that appears in all the relevant perturbation equations ( see sec .
[ eqns ] ) we can draw two general conclusions : i ) another outgoing - wave mode is characterized by @xmath180 ( where the bar represents complex conjugation ) , and ii ) time - reversal of an outgoing - wave mode leads to a solution that corresponds to purely ingoing waves at infinity .
that is , a solution such that , at infinity , @xmath181 while @xmath182 .
moreover , it follows that a second such solution corresponds to @xmath183 . as we will demonstrate below
, this information can be very useful when trying to identify normal - mode frequencies that have small imaginary parts .
most schemes for determining qnms are based on numerical constructions of the ratio @xmath184 . assuming that a_in ( ) ~(-_n ) close to a qnm one can try to iterate for the zero of @xmath185 using a standard scheme such as mller s method .
this strategy works fine for rapidly damped modes ( and typically also for the fluid f - mode ) , but does usually not provide a reliable estimate for an imaginary part that is several orders of magnitude smaller than the real part .
an alternative method is inspired by resonant scattering problems in quantum theory .
letting @xmath186 ( with @xmath187 and @xmath188 both positive ) we can easily identify the real part @xmath187 from the position of the minimum of the standard `` breit - wigner resonance '' in a graph of @xmath189 , cf . the example shown in fig .
[ ainfig ] .
it is not all that simple to extract the imaginary part , however . to do
this one must approximate the half - width of the peak , eg . by curve fitting .
to achieve satisfactory precision in this process is not trivial . vs the real part of the frequency .
in the left panel we see a standard `` breit - wigner resonance . ''
the right panel illustrates the properties of the phase of the ration of the asymptotic amplitudes .
we use the location of the two poles and zeros to deduce the frequency and damping rate of a long - lived pulsation mode.,height=264 ] to devise a better scheme for determining the damping rate of a very long - lived qnm , we employ the two general properties we deduced earlier . from these
it is clear that , if we have a zero of @xmath185 close to the real frequency axis , there must also be a zero of @xmath190 on the opposite side of the axis .
how does that alter the above results ?
close to a zero of @xmath190 we will have a_out ( ) ~(-|_n ) .
then we immediately see that the absolute value of the ratio between the asymptotic amplitudes , @xmath191 , is a smooth , slowly varying function of the frequency .
but if we consider its phase , we find some interesting and useful features .
we get ( for real @xmath73 ) where @xmath192 is a complex `` constant , '' and therefore . from this
we see that , instead of having a singularity at @xmath193 we now have a function with two zeros and two poles on the real frequency axis , cf .
[ ainfig ] . provided that the calculation of the asymptotic amplitudes can be done with sufficiently high precision , the location of these zeros ( @xmath194 ) and poles ( @xmath195 ) can readily be deduced .
given this information one can show that @xmath196^{1/2 } \\ { \gamma_r \over \gamma_i } & \approx & { 2 \beta_n \over p_1 + p_1 - 2 \alpha_n } = { 1 \over 2 \beta_n } [ 2 \alpha_n - z_1 - z_2 ] \
.\end{aligned}\ ] ] it should be pointed out that one gets four equations for the real and imaginary parts of the qnm frequency , as well as the ratio @xmath197 .
the last equality above can therefore be used as a `` sanity check '' on the calculation .
in essence , it provides information about the accuracy of the obtained quantitites .
k. d. kokkotas , t. apostolatos and n. andersson , mnras * 302 * , 307 ( 2001 ) .
s. chandrasekhar , phys . rev
. lett .
* 24 * , 611 ( 1970 ) .
j. l. friedman , commun .
* 62 * , 247 ( 1978 ) .
j. l. friedman and b. f. schutz , astrophys .
j. * 222 * , 281 ( 1978 ) .
n. andersson , ap . j. * 502 * , 708 ( 1998 ) .
j. l. friedman and s. m. morsink , ap . j. * 502 * , 714 ( 1998 ) . n. andersson , k. d. kokkotas , and n. stergioulas , ap . j. * 516 * , 307 ( 1999 ) .
g. mendell , ap .
j. * 380 * 530 ( 1991 ) l. bildsten and g. ushomirsky astrophys .
* 529 * 33 ( 2000 ) .
l. lindblom and g. mendell , phys .
d * 61 * , 104003 ( 2000 ) .
jones , phys .
lett . * 86 * 1384 ( 2001 ) .
l. lindblom and b.j .
owen , _ effect of hyperon bulk viscosity on neutron - star r - modes _ preprint astro - ph/0110558 n. andersson and g. l. comer , mon . not .
r. astron .
soc . * 328 * , 1129 ( 2001 ) .
r. i. epstein , ap .
j. * 333 * , 880 ( 1988 ) .
l. lindblom and g. mendell , ap .
j. * 421 * , 689 ( 1994 ) .
u. lee , astron .
astrophys . * 303 * , 515 ( 1995 ) .
r. prix , in preparation ( 2002 ) .
g. l. comer , d. langlois , and l. m. lin , phys . rev .
d * 60 * , 104025 ( 1999 ) .
n. andersson and g. l. comer , phys .
* 24 * 241101 ( 2001 ) .
v. v. khodel , v. a. khodel , and j. w. clark , nuc .
a * 679 * , 827 ( 2001 ) . c. w. misner , k. s. thorne , and j. a. wheeler , _ gravitation _ ( freeman press , san francisco , 1973 ) .
b. carter , `` relativistic fluid dynamics '' , a. anile and m. choquet - bruhat , eds . , springer - verlag ( 1989 )
. g. l. comer and d. langlois , class . and
grav . * 11 * , 709 ( 1994 ) . b. carter and d. langlois , phys .
d * 51 * , 5855 ( 1995 ) .
b. carter and d. langlois , nucl .
b * 454 * , 402 ( 1998 ) .
b. carter and d. langlois , nucl . phys .
b * 531 * , 478 ( 1998 ) .
d. langlois , a. sedrakian , and b. carter , mon . not .
r. astron .
soc . * 297 * , 1189 ( 1998 ) .
r. prix , phys .
d * 62 * , 103005 ( 2000 ) n. andersson and g. l. comer , class . and quant .
grav . * 18 * , 969 ( 2001 ) .
t. regge and j. a. wheeler , phys .
rev . * 108 * , 1063 ( 1957 ) .
l. lindblom and s. l. detweiler , ap .
j. supplement series * 53 * , 73 ( 1983 ) .
s. l. detweiler and l. lindblom , ap .
j. * 292 * , 12 ( 1985 ) .
a. f. andreev and e. p. bashkin , sov .
jetp * 42 * ( * 1 * ) , 164 ( 1975 ) .
r. prix , g. l. comer , and n. andersson , astron .
astrophys . , * 381 * 178 ( 2002 ) .
o. sjberg , nucl .
a * 265 * , 511 ( 1976 ) .
a. akmal , v.r .
panharipande and d.g .
ravenhall , phys rev c * 58 * , 1804 ( 1998 ) k. d. kokkotas and b. f. schutz , mon . not .
r. astron .
268 * , 1230 ( 1992 ) .
a. reisenegger and p. goldreich , ap .
j. * 395 * , 240 ( 1992 ) .
a. sedrakian and i. wasserman , phys . rev .
d * 63 * , 024016 ( 2000 ) .
w. unno , y. osaki , h. ando and h. shibahashi _ nonradial oscillations of stars _ ( university of tokyo press , 1989 ) l. m. franco , b. link and r. i. epstein , ap . j. * 543 * 987 ( 2000 ) . | we develop a general formalism to treat , in general relativity , the nonradial oscillations of a superfluid neutron star about static ( non - rotating ) configurations .
the matter content of these stars can , as a first approximation , be described by a two - fluid model : one fluid is the neutron superfluid , which is believed to exist in the core and inner crust of mature neutron stars ; the other fluid is a conglomerate of all charged constituents ( crust nuclei , protons , electrons , etcetera ) .
we use a system of equations that governs the perturbations both of the metric and of the matter variables , whatever the equation of state for the two fluids . the entrainment effect is explicitly included .
we also take the first step towards allowing for the superfluid to be confined to a part of the star by allowing for an outer envelope composed of ordinary fluid .
we derive and implement the junction conditions for the metric and matter variables at the core / envelope interface , and briefly discuss the nature of the involved phase - transition .
we then determine the frequencies and gravitational - wave damping times for a simple model equation of state , incorporating entrainment through an approximation scheme which extends present newtonian results to the general relativistic regime .
we investigate how the quasinormal modes of a superfluid star are affected by changes in the entrainment parameter , and unveil a series of avoided crossings between the various modes .
we provide a proof that , unless the equation of state is very special , all modes of a two - fluid star must radiate gravitationally .
we also discuss the future detectability of pulsations in a superfluid star and argue that it may be possible ( given advances in the relevant technology ) to use gravitational - wave data to constrain the parameters of superfluid neutron stars .
0.5 true in |
controlling the dynamics of quantum systems is a problem of great recent interest with many applications from quantum chemistry to quantum computing . since the dynamics of a quantum system
is essentially determined by its hamiltonian , a central theme in quantum control is hamiltonian engineering .
this is typically accomplished by the application of external fields , whose interaction with the system modifies its effective hamiltonian , thereby changing its dynamical evolution .
the type of control fields that are available tends to vary significantly depending on the type of system and the actuators available . in traditional applications such as nuclear spin engineering and nuclear magnetic resonance ( nmr ) , the fields are often radiofrequency pulses , while in applications involving transitions between electronic states in atoms or vibrational modes of molecules , the control fields are usually laser pulses . in more recent applications such as various solid - state architectures based on semi - conductor quantum dots or josephson junctions , the controls are often electric fields produced by external electrodes . despite the disparate physical realizations , all of these applications share important similarities , in that the external fields modify the hamiltonian in a certain way , and
the objective generally is to find a particular control that modifies the dynamics in a certain way , and achieves a certain objective , e.g. , of steering the system to a particular state or realizing a desired quantum process .
there are many differences in the specifics , however , which can be very important for the design of effective and experimentally feasible controls . in nmr applications , for example , we can often approximate the system well by assuming that the control fields induce rapid local rotations of individual spins , which interact through a fixed , weak coupling between them that is not affected by the control fields .
furthermore , we can often assume that we can selectively address individual spins by frequency - selective control pulses , and that we can perform local rotations about orthogonal axes .
the same approximations , however , are not necessarily valid for other systems such as solid - state architectures . in particular for systems controlled by local voltage gates , it is often difficult to perform rotations about orthogonal axes , and a voltage applied to a local electrode may have non - negligible effects on the dynamics of nearby quantum dots , and in the presence of multiple control voltages there can be signficant cross - talk effects , all of which can signficantly complicate effective control design . in this paper
we attempt to incorporate some of these effects to derive more realistic control system models for two types of systems of practical interest , and apply optimal control theory to find control solutions to implement a set of elementary gates for the resulting more complex control system models .
the starting point for many optimal control problems in quantum computing such as optimal implementation of quantum gates has often been a control system model inspired by liquid - state nuclear magnetic resonance ( nmr ) . using the common shorthand @xmath1 to denote an @xmath2-factor tensor product in which the @xmath3th factor is the pauli matrix @xmath4 for @xmath5 and all other factors are the identity @xmath6 , the hamiltonian for a system consisting of @xmath2 spin-@xmath7 nuclei ( qubits ) in this model is usually given by @xmath8 = \hbar\sum_{n < n ' } j_{nn ' } \sz^{(n ) } \sz^{(n ' ) } + \hbar\sum_{n=1}^n \left(u_{2n-1}(t)\sx^{(n ) } + u_{2n}(t)\sy^{(n)}\right),\ ] ] where the first term is a fixed inter - qubit coupling term , in this case of ising type , and we have @xmath9 independent control fields @xmath10 and @xmath11 capable of performing local @xmath12 and @xmath13 rotations on the @xmath3th spin , respectively .
if the interqubit couplings @xmath14 are controllable , or at least can be switched off , then it is very easy to see that we can implement any local unitary operation @xmath15 on the @xmath3th qubit , e.g. , by performing a sequence of three local rotations about the orthogonal axes @xmath16 and @xmath17 , e.g. , using the standard euler decomposition @xmath18 with @xmath19 , etc , for suitable values of the rotation angles @xmath20 , and two - qubit gates can be implemented using the cartan decomposition @xcite , @xmath21\ , [ \op{u}_x^{-1}\ , \op{z}(\alpha_2 ) \ , \op{u}_x]\ , \op{z}(\alpha_1 ) \ , \op{u}_2,\ ] ] where @xmath22 corresponds to free evolution under the ising - coupling hamiltonian for a suitable time , @xmath23 and @xmath24 correspond to local operations on both qubits , which can be implemented using the euler decomposition , and @xmath25 correspond to local @xmath12-rotations by @xmath26 on both qubits , and similarly for @xmath27 .
if the inter - qubit coupling terms @xmath14 are fixed and not controllable then this gives rise to a drift term , which complicates the situation .
if the coupling is weak and we can apply `` hard pulses '' , e.g. , if we can ensure @xmath28 then the drift term is negligible ( as far as the implementation of local gates is concerned ) , and the geometric control pulse sequences derived from the euler and cartan decomposition are often a good approximation .
however , if the coupling is stronger or hard pulses are not available , neglecting the drift term will result in inaccurate gates . moreover , in realistic applications , hard pulses may lead to other problems such as off - resonant excitation .
although we can in principle try to find geometric decompositions that take the drift term into account @xcite , geometric control pulse sequences derived from lie group decompositions are usually not optimal even when the drift is negligible , and optimal control has been shown to be able to produce more effective or efficient control pulses with regard to various perfomance indices such as gate operation timesor average fidelity even for cases where geometric decompositions are valid , at least theoretically [ see e.g. @xcite ] .
the success of optimal control theory in finding more effective control pulse sequences for various model problems has fueled a surge of interest in this area .
however , the control system models on which many of the optimal control calculations so far have been based are generally too simple and often unrealisitc for the systems they are supposed to model . the model described in the previous section , for example , is based on many approximations that may be questionable even in nmr settings , and certainly for other applications .
for instance , in the ideal case the control fields @xmath10 and @xmath11 induce only an @xmath12-rotation and @xmath13-rotation , respectively , on the @xmath3th qubit , but the desire to speed up gate implementation times may produce strong control fields likely to induce off - resonant excitations not accounted for the in model .
off - resonant excitation means that the fields are not only felt by the target qubit but also by other qubits . in the simplest case we can try to model off - resonant excitation by replacing the individual interaction hamiltonians @xmath16 and @xmath17 for the independent control fields @xmath10 and @xmath11 by linear combinations of the coupling terms [ eq : lin_xtalk ] @xmath29 the new interaction hamiltonians @xmath30 for @xmath31 no longer correspond to local rotations about mutually orthogonal axes , which makes finding geometric decompositions for this case very challenging in general .
however , even arbitrary ( linear ) cross - talk need not be an obstacle for optimal control as shown in @xcite for a model system of this type .
provided the cross - talk hamiltonians are known , based on a combination of theoretical models and experimental characterization , it still appears to easy to find effective control pulses using optimal control algorithms .
the main drawback is that this simple model does not accurately model real cross - talk .
the intrisic system hamiltonian for a system of @xmath2 spins with resonance frequencies @xmath32 with static ising coupling @xmath14 in a fixed laboratory frame is @xmath33 if we apply a global electromagntic field @xmath34 with components @xmath35 and @xmath36 , respectively , the resulting control hamiltonian is @xmath37\ ] ] transforming to a multiply rotating frame defined by @xmath38 , the interaction picture rotating wave approximation hamiltonian is @xcite @xmath39\ ] ] where @xmath40 and @xmath41 .
if the field is exactly resonant with the @xmath3th spin , @xmath42 , then the term corresponding to the @xmath3th spin simplifies to @xmath43.\ ] ] if we can further argue that all other spins have frequencies @xmath32 far detuned from @xmath44 , @xmath45 is slowly varying and not too large , @xmath46 for all @xmath47 , then for sufficiently large @xmath48 the rapidly oscillating terms will `` average to zero '' , i.e. , we can assume @xmath49 in this case the effective control hamiltonian associated with the field @xmath34 simplifies to @xmath50.\ ] ] thus , if we are able to simultaneously apply @xmath2 independent fields @xmath51 with frequencies @xmath32 and @xmath2 independent fields @xmath52 with frequencies @xmath32 , for instance , then assuming all the previous assumptions and approximations still hold , we obtain @xmath53,\ ] ] exactly the hamiltonian ( [ eq : sys0 ] ) with @xmath54 and @xmath55 . in practice ,
it is questionable , however , whether these approximations are still valid in general , especially if we simultaneously apply @xmath9 control fields @xmath56 and @xmath57 , all of which may be strongly modulated in time as a result of an optimization procedure , and thus not slowly varying .
furthermore , the interaction picture rwa hamiltonian ( [ eq : sys1c ] ) is not a very good starting point for a general optimization either because each of the independent control hamiltonians @xmath58\ ] ] with @xmath59 for a field @xmath60 with @xmath61 and @xmath62 , in the total hamiltonian @xmath63 is time - dependent , containing rapidly oscillating sine and cosine terms .
therefore , we might as well let the component functions @xmath64 and @xmath65 of the field be completely arbitrary and work with the general hamiltonian @xmath66 with @xmath67 as in eq .
( [ eq : sys1a ] ) , instead . in some applications such as nmr the only type of control
we have is globally applied control fields . however ,
for many systems , especially solid - state systems such as the nuclear spins of donor atoms embedded in a substrate @xcite , or electron spins in quantum dots @xcite , for example , some local control in the form of control electrodes is available , and in a typical setting , control is achieved by combining globally applied electromagnetic fields with local fields provides by control electrodes @xcite . in this case
we often have a global electromagnetic field @xmath34 with components @xmath68 , @xmath69 , in addition to a static magnetic field in the @xmath70 direction , and a hamiltonian of the form @xmath71 , \end{split}\ ] ] as before , but in addition we now have some degree of control over the resonance frequency @xmath32 of the @xmath3th qubit via local control electrodes .
note that unlike in the previous model we have assumed isotropic heisenberg coupling , i.e. , @xmath72 , instead of @xmath73 ising coupling as this seems to be more appropriate for solid - state systems , although the assumption of isotropy may not be appropriate for some systems .
one control approach is to simultaneously vary the voltages applied to the local control electrodes as well as the globally applied field .
however , it practice it is much simpler to apply a global field @xmath34 with a fixed amplitude @xmath74 and fixed frequency @xmath44 and only vary the voltages applied to the local control electrodes . in this case
we can introduce a rotating frame , this time rotating at the field frequency @xmath44 , @xmath75 , to obtain the slightly simpler interaction picture hamiltonian @xmath76 where we set @xmath77 and take @xmath78 to be the coupling constants @xmath79 in units of some coupling strength @xmath80 , so that for a homogeneous system consisting of a chain of spins that couple identically to the field we have @xmath81 for @xmath82 . in this model the constants @xmath14 and @xmath79 are usually fixed and determined by the physics of the system .
the frequency @xmath44 and rabi frequency @xmath83 of the field can be chosen initially but are then fixed .
thus , all the dynamic control is achieved by dynamically changing the control voltages and thus @xmath84 .
this type of control therefore leads to a different partitioning of the total hamiltonian into a fixed drift and variable control part @xmath85 in general , due to cross - talk effects @xcite the energy level shift of the @xmath3th qubit may be a complicated function of all the voltages applied , not only to the electrode directly above it , but to neighbouring control electrodes as well , i.e. , @xmath32 may be a function of all the control voltages @xmath86 applied , i.e. , @xmath87 .
one way to solve the problem of finding controls @xmath88 that implement a desired set of quantum gates , especially for control systems that are not amenable to simple geometric decomposition schemes , such as the ones discussed in the previous section , is to formulate the task as an optimization problem .
this typically involves ( a ) choosing a target function to be optimized , ( b ) finding a suitable parametrization for the set of admissible ( and experimentally feasible ) controls , and ( c ) finding a solution to the resulting , possibly constrained , optimization problem based on application of pontryagin s maximum principle @xcite . for the task of implementation of quantum gates or processes , a natural choice for the objective function
is the gate fidelity @xcite @xmath89 the reason for the factor of @xmath90 , where @xmath2 is the system dimension , is to ensure that @xmath91 varies between @xmath92 ( for @xmath93 ) and @xmath94 ( for @xmath95 ) independent of the system dimension . in general , maximizing the gate fidelity is equivalent to minmizing the gate error @xmath96 this is very to see if the target operator @xmath97 and the actual operator @xmath98 are both unitary ( and thus @xmath99 ) as @xmath100 \\ & = \tr[\op{u}^\dagger\op{u } + \op{u}_t^\dagger\op{u}_t -\op{u}^\dagger\op{u}_t -\op{u}_t^\dagger \op{u } ] \\ & = 2n - 2\re\tr(\op{u}_t^\dagger\op{u } ) .
\end{split}\ ] ] noting that the implemented operator @xmath101 is a function of the control @xmath102 , the control objective is to find an admissible control @xmath103 such that @xmath104 and of course , @xmath105 has to satisfy the schrodinger equation @xmath106 \op{u}_{\vec{u}}(t),\ ] ] where @xmath107 $ ] is the control - dependent hamiltonian of the system .
once the hamiltonian @xmath107 $ ] , the target time @xmath108 and the target operator @xmath97 have been chosen , the fidelity depends only on the choice of the control @xmath88 , i.e. , @xmath109 now let @xmath102 and @xmath110 be two controls . noting that @xmath111 and that for a control - linear hamiltonian @xmath107=\op{h}_0+\sum_m u_m(t ) \op{h}_m$ ] we have @xmath112 we obtain immediately @xmath113\\ & = \frac{2}{n\hbar } \im \tr \left[\op{u}_t^\dagger\int_{t_0}^{t_f } \op{u}_{\vec{u}}(t_f , t ) \sum_m \delta u_m(t ) \op{h}_m \op{u}_{\vec{u}+\delta\vec{u}}(t)\ , dt \right ] \\ & = \frac{2}{n\hbar } \sum_m \int_{t_0}^{t_f } \delta u_m(t ) \im \tr \left[\op{u}_t^\dagger \op{u}_{\vec{u}}(t_f , t ) \op{h}_m \op{u}_{\vec{u}+\delta\vec{u}}(t)\right ] \ , dt \end{aligned}\ ] ] which shows that setting @xmath114\ ] ] with @xmath115 for all @xmath116 will increase the fidelity .
thus , starting with any initial guess @xmath117 , iteratively solving the schrodinger equation while updating the control in each iteration according to the rule @xmath118 , with @xmath119 chosen as in eq .
( [ eq : deltau ] ) , will monotonically increase the fidelity in each iteration , and as the fidelity is of course bounded , the iteration must converge to some limiting fidelity @xmath120 , although we can not guarantee that the limiting value will be equal to the global maximum of the fidelity .
in fact , many optimal control algorithms are based precisely on this iterative update scheme . to solve the schrodinger equation numerically
, we still require some discretization .
the simplest approach is to assume piecewise constant fields @xmath121 for @xmath122 , in which case the exact solution of the schrodinger equation is explicitly @xmath123 where @xmath124 , @xmath125 , and @xmath126\ ] ] is a simple matrix exponential and the update rule becomes @xmath127.\ ] ] with this type of discretization there are two ways of implementing the update rule . in the global update approach all the matrix exponentials for all times
@xmath128 are evaluated to solve the schrodinger equation , and the fields are updated for all times @xmath128 simultaneously at the end of each iteration step .
this is essentially the well - known grape algorithm as described in @xcite .
an alternative is to update the fields @xmath129 at each time step @xmath128 .
our analysis above shows that both approaches will monotonically increase the fidelity provided @xmath130 for all @xmath128 , although they will tend to converge to different solutions for the fields . in our limited numerical experiments for the systems considered here
, both approaches seemed equally effective in finding satisfactory control fields but the local update approach tended to converge faster in terms of the computational time required to find a solution .
the main reason for this appears to be that in the local update approach only a single @xmath131 is incremented at a time , and hence we can optimize the multiplier @xmath132 with very little computational overhead to speed up the convergence , as any local change of @xmath132 requires only re - evaluation of a single matrix exponential ( and two matrix multiplications and a trace operation ) to assess its effect on the fidelity , while in the global update any change in @xmath133 requires the re - evaluation of all the @xmath134 matrix exponentials and @xmath135 matrix multiplications to assess the fidelity of the new control .
an alternative to this simple step - size control is to use higher derivatives to predict the optimal step - size but this also involves potentially significant computational overhead to find the best step size , in addition to the need to calculate analytical expressions for these derivates , thus it is not clear if this will yield a computational advantage or not , although it is worth investigating .
a general shortcoming of this simple optimization procedure is that it does not take constraints into account or include penalities for less desirable fields .
simple constraints such as upper bounds on the field amplitudes can easily be incorporated into the algorithm .
more complex constraints such as limiting the frequency bandwidth of pulses or minimizing the overall pulse energies etc require augmentation of the objective functional by constraint or penalty terms and modifications to the update rules . for some types of penalty terms iterative algorithms very similar to the algorithm above exist [ see e.g. @xcite ] , but a detailed discussion of appropriate penalty functions , resulting iterative update rules , and comparison of the resulting optimal fields is beyond the scope of this paper .
the specific task we wish to accomplish here is finding controls @xmath88 that implement the following complete set of universal gates ( i.e. , target operators ) @xmath136 for a two - qubit system , where we employ the common appreviations @xmath137 , @xmath138 and @xmath139 .
the phase factor for the @xmath140 is necessary to ensure that the target operators are in @xmath141 since our model hamiltonians have zero trace and hence can only generate operators in @xmath141 . as a second ,
more challenging task we consider the implementation of the toffoli - like three - qubit gate @xmath142 for the simulations we assume uniform nearest - neighbour coupling , i.e. , @xmath143 .
since the coupling constants @xmath144 in our model have units of frequency , by choosing time in units of @xmath145 , we can set the coupling constants to unity . all other frequencies , unless otherwise states , are also chosen in units of @xmath144 .
we can furthermore choose the units of energy such that @xmath146 , which has units of energy @xmath147 time , equals @xmath148 , and can thus be omitted .
for the two - qubit gates we set the target gate operation time to @xmath149 , while for the three - qubit gate we choose @xmath150 as a toffoli gate normally requires five two - qubit gates .
we choose these values as reasonable target times although we were able to find solutions for smaller @xmath108 for many gates , suggesting that the gate operation times could in fact be improved . for our first model system with hamiltonian ( [ eq : sys1 ] ) , assuming @xmath151 for all @xmath3 , and setting @xmath152 and @xmath153 , the hamiltonian becomes @xmath154 for a two - qubit system , and for a three - qubit system we have @xmath155 fig .
[ fig1 ] shows the actual control field solutions for a system with @xmath156 and @xmath157 , i.e. , a system with ising coupling frequency about 10% of the smaller qubit frequency , i.e. , strong enough to be non - negligible for single qubit operations , and a qubit frequency difference large enough to allow some selectivity but still small enough so that off - resonant excitation is a concern for pulses of the magnitude and duration required for to achieve the desired gate operation times . for this model system
we choose a relatively large number of time steps , @xmath158 , motivated by the assumption that the @xmath12 and @xmath13 components of the globally applied electromagnetic field could be varied more or less arbitarily , although simulations suggest that we can find piecewise constant solutions with much fewer steps , if necessary .
the solutions are not unique and in fact different initial fields or update rules generally produce different solutions . again , despite the increased complexity of the model ( off - resonant excitation , no rwa ) and the reduced number of control fields ( two vs four in the simpler model ( [ eq : sys0 ] ) ) , it appeared easy to find solutions that achieved the target fidelity of @xmath159 , or gate errors @xmath160% . however , as the figure shows , the magnitudes of the fields for at least some of the gates are relatively large .
further analysis also shows that the spectral range of the optimal controls produced by this algorithm tends to be rather broad , which may be a concern in practice when there are bandwidth limitations as filtering of high or low frequency components will reduce the gate fidelity .
some numerical simulations for related problems suggest that solutions with much lower spectral bandwidth often exist but it appears that a different parametrization of the fields and optimization approach are required to find such solutions .
finding controls to implement the three - qubit gate proved more of a challenge for this system , in part because convergence was extremely slow .
[ fig](left ) shows a solution with fidelity @xmath161% .
notice that the fields have high amplitudes and are very noisy .
[ fig4](left ) also shows that the fields have very large spectral bandwidth as expected , which would make faithful implementation in the laboratory challenging . for our second model system with hamiltonian ( [ eq : sys2 ] ) , assuming again @xmath151 for all @xmath3 , and setting @xmath162 for @xmath163 , the hamiltonian becomes @xmath164 for a two - qubit system , and for a three - qubit system we have @xmath165 notice that we take @xmath166 to be independent controls here to circumvent the complex and highly architecture specific problem of mapping the actual control voltages @xmath167 onto the frequency shifts @xmath84 , although it must be stressed that optimal control solutions for @xmath88 will only be useful for a specific system if the @xmath88 can be actually realized through proper choice of the voltages , which is likely to impose additional , non - trivial constraints .
again , we chose target gate operation times of @xmath149 and @xmath150 for the two- and three - qubit gates , respectively , but in this case we tried to minimize the number of steps @xmath134 , assuming that frequent switching of the control voltages would be experimentally more challenging and increase errors . as fig .
[ fig2 ] shows for @xmath168 ( in units of @xmath144 ) we were able to find satisfactory solutions for all six two - qubit gates with only @xmath169 switches .
we note that @xmath170 appeared to be optimal in the sense that we had to increase @xmath134 significantly for larger or smaller values of @xmath83 . for @xmath171 ,
for example , we had to increase the number of steps to @xmath172 and double the target time to be able to implement the first five two - qubit gates ( corresponding to local operations ) with fidelity @xmath173 , and for the final cnot gate the fidelity we achieved was still only @xmath174 , i.e. , below threshold .
for @xmath175 we were able to find solutions above threshold fidelity for all target gates for @xmath176 ; for @xmath169 the fidelity of the cnot gate remained below threshold ( @xmath177% ) . similarly , for @xmath178 we were able to find solutions above the threshold fidelity for all gates for @xmath176 but not for @xmath169 .
for even larger @xmath83 , such as @xmath179 or @xmath180 , @xmath134 had to be increased to @xmath179 , and for @xmath83 in the 50 - 60 range @xmath181 steps were required .
these results can be partially explained considering that if @xmath83 is on the order of the fixed coupling strengths then the weight of the drift term makes it harder to implement local operations hence the increase in the time required . for large @xmath83 local operations
become easier to implement but entangling gates such as the cnot , which rely on generation of entanglment via the fixed ising coupling term , become more challenging , requiring more complex pulse shapes , hence necessitating a larger number of steps @xmath134 .
it is interesting to note that for this system we had no trouble finding solutions to implement the three - qubit gate with fidelity @xmath161% .
a possible solution is shown in fig .
[ fig3](right ) .
although the pulse profiles appear more complicated , their magnitudes are actually lower than for some of the two - qubit gates , and although @xmath181 , the number of switches per time unit is still only @xmath182 . unlike for the previous model fig .
[ fig4](right ) shows that the spectral profile and bandwidth of the solutions are much more desirable .
starting with the most basic control system model commonly used in applications such as ( liquid - state ) nmr , we have derived more realistic control system models for two types of control of special interest for many applications : ( a ) control of qubits with fixed inter - qubit coupling using globally applied electromagnetic fields in the regime where off - resonant excitation by the control fields is non - negligible , and ( b ) control of qubits with fixed coupling using a combination of a fixed globally applied control field and local control electrodes to shift the resonance frequencies of the qubits . despite the substantially increased complexity of the resulting models , reduced number of independent controls , and non - trivial constraints on the allowed frequency of voltage changes in the latter model , optimal control techniques even in the rather basic form implemented , overall still proved very effective in finding solutions that achieved the target fidelity of @xmath183% for almost all cases , although there appears to be scope for further improving the controls with regard to amplitude and bandwidth considerations , and for improving the overall convergence of the algorithm , which even with some of the modifications discussed , appears to be rather slow especially for the three - qubit gates .
a somewhat unexpected conclusion of the calculations is that our second model system , which is controlled mainly by local voltage gates that can only alter the detuning of the qubits from the fixed ( unmodulated ) global field appears to be easier to control , even when constraints on the number of voltage changes are imposed , than the first model system , which is controlled by global fields with arbitrary pulse envelopes .
sgs acknowledges uk research council funding from an advanced research fellowship and additional support from the and hitachi .
she is currently also a marie curie fellow under the european union knowledge transfer programme mtdk - ct-2004 - 509223 . | starting with the basic control system model often employed in nmr pulse design , we derive more realistic control system models taking into account effects such as off - resonant excitation for systems with fixed inter - qubit coupling controlled by globally applied electromagnetic fields , as well as for systems controlled by a combination of a global fields and local control electrodes . for both models optimal control
is used to find controls that implement a set of two- and three - qubit gates with fidelity @xmath0 . |
quantum systems subjected to external time - periodic fields which are not small have been studied in various settings . in constant amplitude small enough oscillating fields ,
perturbation theory typically applies and ionization is generic ( the probability of finding the particle in any bounded region vanishes as time becomes large ) . for larger time - periodic fields ,
a number of rigorous results have been recently obtained , see @xcite and references therein , showing generic ionization . however , outside perturbation theory , the systems show a very complex , and often nonintuitive behavior .
the ionization fraction at a given time is not always monotonic with the field @xcite .
there even exist exceptional potentials of the form @xmath10 with @xmath11 periodic and of zero average , for which ionization occurs for all small @xmath12 , while at larger fields the particle becomes confined once again @xcite . furthermore
, if @xmath13 is replaced with smooth potentials @xmath14 such that @xmath15 in distributions , then ionization occurs for all @xmath12 if @xmath16 is kept fixed .
numerical approaches are very delicate since one deals with the schrdinger equation in @xmath17 , as @xmath18 and artefacts such as reflections from the walls of a large box approximating the infinite domain are not easily suppressed .
the mathematical study of systems in various limits is delicate and important . in physical experiments one deals with forcing of finite effective duration , often with exponential damping
this is the setting we study in the present paper , in a simple model , a delta function in one dimension , interacting with a damped time - harmonic external forcing .
the equation is @xmath19 where @xmath20 is the amplitude of the oscillation ; we take @xmath21 ( the analysis for other values of @xmath22 is very similar . ) the quantity of interest is the large @xmath23 behavior of @xmath24 , and in particular the survival probability @xmath25 where @xmath26 is a bounded subset of @xmath27 . _ perturbation theory , fermi golden rule_. if @xmath22 is small enough , @xmath28 decreases exponentially on an intermediate time scale , long enough so that by the time the behavior is not exponential anymore , the survival probability is too low to be of physical interest . for all practical purposes ,
if @xmath22 is small enough , the decay is exponential , following the fermi golden rule , the derivation of which can be found in most quantum mechanics textbooks ; the quantities of interest can be obtained by perturbation expansions in @xmath22 .
this setting is well understood ; we mainly focus on the case where @xmath22 is not too small , a toy - model of an atom interacting with a field comparable to the binding potential .
_ no damping . _ the case @xmath29 is well understood for the model ( [ eq : eqa ] ) in all ranges of @xmath22 , see @xcite . in that case , @xmath30 as @xmath2 . however , since the limit @xmath31 is singular , little information can be drawn from the @xmath29 case .
for instance , if @xmath5 , the limiting value of @xmath28 is of order @xmath32 , while with an abrupt cutoff , @xmath33 , the limiting @xmath28 is @xmath7 ( as usual , @xmath34 is the characteristic function of the set @xmath35 ) .
thus , at least for fields which are not very small , the shape of the pulse cut - off is important .
even the simple system ( [ eq : eqa ] ) exhibits a highly complex behavior .
we obtain a rapidly convergent expansion of the wave function and the ionization probability for any frequency and amplitude ; this can be conveniently used to calculate the wave function with rigorous bounds on errors , when the exponential decay rate is not extremely large or small , and the amplitude is not very large .
for some relevant values of the parameters we plot the ionization fraction as a function of time . we also show that for @xmath5 the equation is solvable in closed form , one of the few nontrivial integrable examples of the time - dependent schrdinger equation .
[ thm:1 ] let @xmath36 be the solution of ( [ eq : eqa ] ) with initial condition @xmath37 .
let @xmath38 then as @xmath39 we have @xmath40 where @xmath41 , @xmath42 , @xmath43 , and where @xmath44_{\sigma=1}\ ] ] where @xmath45 solves @xmath46 there is a unique solution of ( [ eq : inhomoina1 ] ) satisfying @xmath47 where @xmath48 is a constant .
it is this solution that enters ( [ eq : rina ] ) .
there is a rapidly convergent representation of @xmath49 , see [ subsec : sumrepofamn ] . clearly , @xmath50 is the probability of survival , the projection onto the limiting bound state .
[ thm:3 ] ( i ) for @xmath5 we have @xmath51 where @xmath52 .
\(ii ) we look at the case when @xmath53 , the bound state of the limiting time - independent system . assuming the series of @xmath54 is borel summable in @xmath3 for @xmath55 $ ] ( summability follows from ( [ eq : inhomogeneousmainsys2 ] ) , but the proof is cumbersome and we omit it ) , as @xmath56 we have @xmath57 note : the behavior ( [ eq : inhomogeneousmainsy1 ] ) is confirmed numerically with high accuracy , constants included , see [ subsec : numericalevidence ] .
we also discuss results in two limiting cases : the short pulse setting ( see [ sec : shortpulse ] ) and the special case @xmath29 ( see [ sec : anotherspecialcase ] ) .
existence of a strongly continuous unitary propagator for ( [ eq : eqa ] ) ( see @xcite v.2 , theorem x.71 ) implies that for @xmath58 , the laplace transform @xmath59 exists for @xmath60 and the map @xmath61 is @xmath62 valued analytic in the right half plane @xmath63 the laplace transform of ( [ eq : eqa ] ) is @xmath64 let @xmath65 and @xmath66 where @xmath67 .
[ rmk : changeofnotion ] _ since the @xmath68 plane equation only links values of @xmath68 differing by @xmath69 , @xmath70 , it is useful to think of functions of @xmath68 as vectors with components @xmath71 and @xmath16 , parameterized by @xmath72 .
_ thus we rewrite ( [ eq : eqa ] ) as @xmath73 when @xmath74 , the resolvent of the operator @xmath75 has the integral representation @xmath76 with @xmath77 where the choice of branch is so that if @xmath78 , then @xmath79 is in the fourth quadrant , and where the green s function is given by @xmath80 [ rmk : propofg ] if @xmath81 , using integration by parts we have , as @xmath82 @xmath83 where we regard @xmath84 as an operator with @xmath68 as a parameter ; see also remark [ rmk : changeofnotion ] .
furthermore , ( [ eq : defofg ] ) implies @xmath85 . define the operator @xmath86 by @xmath87\ ] ] then eq .
( [ eq : inhomo2 ] ) can be written in the equivalent integral form @xmath88 where @xmath84 is defined in ( [ eq : defofg ] ) .
[ rmk : propofc ] because of the factor @xmath89 in ( [ eq : defofg ] ) , we have , with the identification in remark [ rmk : changeofnotion ] , @xmath90 as @xmath91 , for any function @xmath92 . in this section
we assume @xmath37 . as in remark
[ rmk : propofg ] , we obtain @xmath93 for some @xmath94 .
let @xmath95}(t ) \right)\ ] ] for large @xmath68 we have @xmath96 [ rmk : invlaplaceofh1 ] _ as a function of @xmath97 , @xmath98 is clearly in @xmath99 and @xmath100}(t)\ ] ] thus for @xmath101 we have @xmath102 _ substituting @xmath103 in ( [ eq : vector21 ] ) we have @xmath104 let @xmath105 , then remark [ rmk : propofc ] implies that for large @xmath68 @xmath106 and by construction @xmath107 as a function of @xmath97 .
we analyze ( [ eq : innewvariable ] ) in the space @xmath108 , @xmath48 , where @xmath109 we denote by @xmath110 the transformed wave function corresponding to @xmath111 . writing @xmath112 instead of @xmath111 , we obtain from ( [ eq : innewvariable ] ) , @xmath113 [ remark : c ] @xmath86 is a compact operator on @xmath114 , and analytic in @xmath115 .
compactness is clear since @xmath86 is a limit of bounded finite rank operators .
analyticity is manifest in the expression of @xmath86 ( see ( [ eq : defofg ] ) and ( [ eq : definitionofc ] ) ) .
[ prop : analyticityofy ] equation ( [ eq : vector2 ] ) has a unique solution iff the associated homogeneous equation @xmath116 has no nontrivial solution . in the latter case ,
the solution is analytic in @xmath117 .
this follows from lemma [ remark : c ] and the fredholm alternative . when @xmath118 , @xmath119 , and @xmath120 , @xmath86 is singular , but the solution is not . indeed , by adding @xmath121}$ ] , @xmath122 , to both sides of ( [ eq :
inhomo2 ] ) we get the equivalent equation @xmath123 } \right)y_{m , n}\\ = \
i\psi_0 + \left(\mathbf{1}_{[-a , a]}- 2\delta(x)\right)y_{m , n } + \delta(x)\left(y_{m+1,n+1}+y_{m+1,n-1}\right)\end{gathered}\ ] ] arguments similar to those when @xmath121}$ ] is absent show that the operator @xmath86 associated to ( [ eq : inhomo2mod ] ) is analytic in @xmath117 , thus @xmath124 is analytic in @xmath117 .
componentwise ( [ eq : vector21 ] ) reads @xmath126 \end{split}\ ] ] with @xmath127 , we have @xmath128 where @xmath129 is defined in ( [ eq : defofgmn ] ) . [ prop : yisdeterminedbya ] the solution to ( [ eq : detailed ] ) is determined by the @xmath130 through @xmath131 it thus suffices to study ( [ eq : inhomoina ] ) .
taking @xmath132 in ( [ eq : detailed ] ) we obtain ( [ eq : inhomoina ] ) ; using now ( [ eq : inhomoina ] ) in ( [ eq : detailed ] ) we have @xmath133\\ = \int_{\rr } \frac{1}{2}\kappa_{m , n}^{-1}e^{-\kappa_{m , n}|x - x'| } \psi_0(x ' ) dx'+ e^{-\kappa_{m , n } |x| } a_{m , n } - \frac{1}{\kappa_{m , n } } e^{-\kappa_{m , n } |x| } g_{m , n}\end{gathered}\ ] ] [ rmk : spaceofa ] if @xmath134 , then @xmath127 satisfies ( [ eq : norm_in_a_inhomo ] ) .
let @xmath135 where @xmath136 is a solution to ( [ eq : vector3 ] ) .
the solution of ( [ eq : vector3 ] ) has the freedom of a multiplicative constant ; we choose it by imposing @xmath137 it is clear @xmath138 satisfies the homogeneous equation associated to ( [ eq : inhomoina ] ) @xmath139 define @xmath140 to simplify notation we take @xmath141 in which case @xmath142 .
the general case is very similar
. denote @xmath143 [ prop : a ] the system ( [ eq : homoina ] ) has nontrivial solutions in @xmath114 iff @xmath144(=1 as discussed above ) .
if @xmath145 , then the solution is a constant multiple of the vector @xmath146 given by @xmath147 and obtained inductively from ( [ eq : homoina ] ) for all other @xmath148 .
( note that @xmath145 is used crucially here since ( [ eq : homoina ] ) allows for the nonzero value of @xmath149 . )
let @xmath145 . by construction ,
@xmath150 defined in proposition [ prop : a ] satisfies the recurrence and we only need to check ( [ eq : norm_in_a_inhomo ] ) . since @xmath151 and @xmath152 , we have @xmath153 proving the claim .
now , for any @xmath72 , if there exists a nontrivial solution , then for some @xmath154 we have @xmath155 . by ( [ eq : homoina ] )
, we have either @xmath156 or @xmath157 it is easy to see that if @xmath158 or @xmath159 , the above inequalities lead to @xmath160 for large @xmath161 ( note that in these cases @xmath162 ) , contradicting ( [ eq : norm_in_a_inhomo ] ) . finally ,
if @xmath145 , then @xmath163 is determined by @xmath149 via the recurrence relation ( [ eq : homoina ] ) ( note that @xmath164 ) .
this proves uniqueness ( up to a constant multiple ) of the solution . combining proposition [ prop : analyticityofy ] and proposition [ prop : a ] we obtain the following result .
the solution @xmath165 to equation ( [ eq : inhomo1 ] ) is analytic with respect to @xmath115 , except for poles in @xmath166 .
proposition [ prop : a ] shows that ( [ eq : homoina ] ) has a solution @xmath150 for @xmath167 ; by proposition [ prop : analyticityofy ] , @xmath125 has singularities in @xmath166 , and the conclusion follows from proposition [ prop : yisdeterminedbya ] .
so far we showed that the solution has possible singularities in @xmath166 . to show that indeed @xmath168 has poles for generic initial conditions , we need the following result : [ lemma : betterfredholm]let @xmath169 be a hilbert space .
let @xmath170 be compact , analytic in @xmath72 and invertible in @xmath171 for some @xmath172 .
let @xmath173 be analytic in @xmath72 .
if @xmath174 solves the equation @xmath175 , then @xmath176 is analytic in @xmath72 in @xmath171 but singular at @xmath120 . by the fredholm alternative ,
@xmath176 is analytic when @xmath177 .
if @xmath176 is analytic at @xmath120 then @xmath178 is analytic and @xmath179 which is a contradiction .
the operator @xmath86 is compact by remark [ remark : c ] .
the inhomogeneity @xmath180 in equation ( [ eq : vector3 ] ) is analytic in @xmath117 .
furthermore , at @xmath145 , @xmath181 is of codimension 1 ( proposition [ prop : a ] ) . combining with lemma [ lemma : betterfredholm ] we have for a generic inhomogeneity @xmath182
, @xmath183 is singular at @xmath145 .
equivalently , @xmath165 has a pole at @xmath184 .
it can be shown that @xmath165 has a pole at @xmath184 for generic @xmath185 .
we prefer to show the following result which has a shorter proof .
[ prop : psihasapole ] the residue @xmath186 of the pole for @xmath187 at @xmath184 is given by @xmath188_{\sigma=1}\ ] ] in particular , @xmath189 for large @xmath3 and generic initial condition @xmath190 .
when @xmath118 and @xmath119 ( [ eq : inhomoina ] ) gives @xmath191 clearly @xmath192 is singular as @xmath193 , which implies that @xmath187 has a pole at @xmath184 with residue given in ( [ eq : r00 ] ) .
thus @xmath186 is not zero if the quantity @xmath194_{\sigma=1}$ ] is not zero .
first , @xmath195 is not zero by definition : @xmath196 next , taking @xmath197 , @xmath198 , and @xmath145 in ( [ eq : inhomoina ] ) we obtain @xmath199_{\sigma=1}\ ] ] thus for any @xmath200 when @xmath3 is large enough we have @xmath201_{\sigma=1}\ ] ] estimating similarly @xmath202 and @xmath203 and so on , we see that @xmath204 .
when @xmath205 is large enough we have @xmath206 .
analogous bounds hold for @xmath207 showing that @xmath194_{\sigma=1}$ ] is not zero .
[ cor : psihaspolesinb ] for generic initial condition @xmath187 has simple poles in @xmath166 , and their residues are given by @xmath208 .
take a small loop around @xmath120 and integrate equation ( [ eq : detailed ] ) along it .
this gives a relation among @xmath209 which is identical to ( [ eq : homoina ] ) : @xmath210 proposition [ prop : psihasapole ] and ( [ eq : choiceofa000 ] ) implies that @xmath211 .
the rest of the proof is follows from proposition [ prop : a ] .
it is easy to see that there exist initial conditions for which the solution has no poles . indeed ,
if the solutions @xmath212 and @xmath213 have a simple pole at @xmath184 with residue @xmath214 and @xmath215 respectively , then for initial condition @xmath216 , the corresponding solution @xmath190 has no pole at @xmath184 .
taking @xmath145 in ( [ eq : inhomoina ] ) we get @xmath217 for @xmath218 , we define @xmath219 .
( note that @xmath220 ) .
we denote @xmath221 and @xmath222 .
let @xmath223 and for some @xmath224 define @xmath225 equation ( [ eq : inhomoasigmais1 ] ) implies @xmath226 as @xmath227 we have @xmath228 and @xmath130 goes to zero as @xmath229 , and thus we have @xmath230 in the limit @xmath227 we obtain @xmath231 [ remark : errorcontrol ] truncating the infinite expansion to @xmath232 , the error is bounded by @xmath233
in [ subsec : positionsandstrengthofthepoles ] it was shown that for a generic initial condition @xmath234 , the solution @xmath235 has simple poles in @xmath166 , with residues @xmath236 . since @xmath134 ,
the inverse laplace transform can be expressed using bromwich contour formula .
recall that @xmath112 differs from the original vector form of @xmath187 by ( [ eq : changeofvariable ] ) , we have @xmath237 the fact that @xmath134 also implies that @xmath238 fast enough as @xmath239 .
thus the contour of integration in the inverse laplace transform can be pushed into the left half @xmath68-plane , after collecting the residues . as a result , for some small @xmath240 the contour becomes one coming from @xmath241 , joining @xmath242 , @xmath243 , and @xmath244 ( for arbitrarily small @xmath245 ) in this order , then going towards @xmath246 .
thus we have @xmath247 by corollary [ cor : psihaspolesinb ] we have @xmath248 the third term in ( [ eq : psidecomposition ] ) decays exponentially for large @xmath23 ( since the integral is bounded ) , while the last two terms yield an asymptotic power series in @xmath249 , as easily seen from watson s lemma .
combining these results and the fact @xmath250 ( remark [ rmk : invlaplaceofh1 ] ) concludes the first part of theorem [ thm:1 ] , with @xmath251 .
the rest follows from proposition [ prop : psihasapole ] .
when @xmath5 , the equation @xmath252 becomes @xmath253 rewriting @xmath130 and @xmath129 as @xmath254 and @xmath255 , ( [ eq : inhomoina ] ) becomes @xmath256 since @xmath5 , ( [ eq : sumrep ] ) simplifies to @xmath257 when @xmath198 ( [ eq : inhomogeneousmainn ] ) becomes @xmath258 with the notation @xmath259 ( [ eq : inhomogeneousmain1 ] ) becomes @xmath260 let @xmath261 we have @xmath262 differentiating in @xmath3 we obtain @xmath263 let @xmath264 be so that @xmath265 then , @xmath266 by taking the inverse laplace transform of ( [ eq : diff_in_u ] ) in @xmath267 we get ( we use @xmath68 as the transformed variable here ) @xmath268 integrating ( [ eq : add1 ] ) with respect to @xmath3 gives @xmath269 thus @xmath270 finally we obtain @xmath271 here we assume that expansion of @xmath272 as @xmath31 is invariant under a @xmath273 rotation ; that is , there are no stokes lines in the fourth quadrant ; this would be ensured by borel summability of the expansion in @xmath3 .
let @xmath274 with @xmath275 , and for simplicity let @xmath276 , then ( [ eq : inhomogeneousmain1 ] ) implies @xmath277 the euler - maclaurin summation formula gives @xmath278 where @xmath279 therefore @xmath280 since @xmath281 applying the euler - maclaurin summation formula again gives @xmath282 figure [ fig:2 ] shows @xmath283 as a function of @xmath284 , very nearly a straight line with slope @xmath285 ( corresponding to the @xmath286 behavior ) , with good accuracy good even until @xmath3 becomes as large as 1 . as a function of @xmath3 for @xmath5 .
@xmath186 is the residue of the pole of @xmath287 at @xmath184 , see corollary [ cor : psihaspolesinb ] . ]
we now consider a short pulse , with fixed total energy and fixed total number of oscillations .
the corresponding schrdinger equation is @xmath288 where @xmath3 is now a large real parameter ( note the @xmath3 in front of the exponential ) .
we are interested in the ionization rate as @xmath289 . by similar arguments as in [ subsec : sumrepofamn ] we have the convergent representation @xmath290 figs . [ fig:5 ] and [ fig:6 ] give @xmath291 ( see corollary [ cor : psihaspolesinb ] ) and @xmath3 under different @xmath292 ratios . on small scales on the @xmath3 axis , @xmath291 exhibits rapid oscillations , easier seen if @xmath292 is smaller . as a function of @xmath3 , with fixed ratio @xmath293 .
@xmath186 is the residue of the pole of @xmath287 at @xmath184 , see corollary [ cor : psihaspolesinb ] . ] as a function of @xmath3 when @xmath294 .
@xmath186 is the residue of the pole of @xmath287 at @xmath184 , see corollary [ cor : psihaspolesinb ] . ]
we briefly go over the case @xmath29 , where ionization is complete ; the full analysis is done in @xcite . in this case @xmath187
does not have poles on the imaginary line ; we give a summary of the argument in @xcite .
the homogeneous equation now reads @xmath296 thus we have @xmath297 the first sum and the second sum on the right hand side are conjugate to each other , and each term in the third sum is real .
so the right hand side is real , thus the left hand side is also real . for @xmath298
, @xmath299 has same the sign as @xmath300 .
therefore the sum can not be real and the equation has no nontrivial solution .
when @xmath301 , for @xmath302 , all @xmath303 have the same sign and for @xmath304 , @xmath305 is real .
since the final sum is purely real , this means @xmath306 for @xmath302 .
but then , recursively , all @xmath307 should be 0 .
zero is thus the only solution to ( [ eq : a01 ] ) . by the fredholm alternative the solution @xmath125 is analytic in @xmath117 and
thus the associated @xmath112 is analytic in @xmath117 .
this entails complete ionization .
, at @xmath308 , as a function of @xmath4 .
@xmath186 is the residue of the pole of @xmath287 at @xmath184 , see corollary [ cor : psihaspolesinb ] . ]
we expect that the behavior of the system at @xmath29 is a limit of the one for small @xmath3 . however , this limit is very singular , as the density of the poles in the left half plane goes to infinity as @xmath31 , only to become finite for @xmath29 .
nonetheless , given a @xmath3 , small but not extremely small , formula ( [ eq : sumrep ] ) allows us to calculate the residue .
figure [ fig:7 ] shows the behavior of the residue versus @xmath4 , for @xmath308 .
we show for comparison the corresponding result when @xmath29 , in figure [ fig:8 ] :
this work was supported in part by the national science foundation dms-0601226 and dms-0600369 .
o costin , r d costin and j lebowitz , `` transition to the continuum of a particle in time - periodic potentials '' in advances in differential equations and mathematical physics , ams contemporary mathematics series ed .
karpeshina , stolz , weikard , and zeng ( 2003 ) .
o. costin , r. d. costin and j. l. lebowitz , _ time asymptotics of the schrdinger wave function in time - periodic potentials , dedicated to elliott lieb on the occasion of his 70th birthday _ , j. stat
14 * 283 - 310 ( 2004).@xmath309 a galtbayar , a jensen and k yajima , local time - decay of solutions to schrdinger equations with time - periodic potentials ( j. stat .
phys . , to appear ) .
l hrmander , _ linear partial differential operators _ , springer ( 1963 ) .
j s howland , stationary scattering theory for time dependent hamiltonians . math .
ann . * 207 * , 315335 ( 1974 ) .
h r jauslin and j l lebowitz , spectral and stability aspects of quantum chaos , chaos * 1 * , 114121 ( 1991 ) .
t kato , perturbation theory for linear operators , springer verlag ( 1995 ) .
p d miller , a soffer and m i weinstein , _ metastability of breather modes of time dependent potentials _ , nonlinearity volume 13 ( 2000 ) 507 - 568 .
c miranda , _ partial differential equations of elliptic type _ , springer - verlag ( 1970 ) .
m reed and b simon , _ methods of modern mathematical physics _
( academic press , new york , 1972 ) . b simon , schrdinger operators in the twentieth century ,
phys . * 41 * , 3523 ( 2000 ) . a soffer and m i weinstein , nonautonomous hamiltonians , jour .
. phys . * 93 * , 359391 ( 1998 ) .
f treves , _ basic linear partial differential equations _ , academic press ( 1975 ) .
w wasow , _
asymptotic expansions for ordinary differential equations _ , interscience publishers ( 1968 ) . | we study the asymptotic behavior of the wave function in a simple one dimensional model of ionization by pulses , in which the time - dependent potential is of the form @xmath0 , where @xmath1 is the dirac distribution .
we find the ionization probability in the limit @xmath2 for all @xmath3 and @xmath4 .
the long pulse limit is very singular , and , for @xmath5 , the survival probability is @xmath6 , much larger than @xmath7 , the one in the abrupt transition counterpart , @xmath8 where @xmath9 is the heaviside function . |
neutrino energy transfer to the matter adjacent to the nascent neutron star is supposed to trigger the explosion of a massive star ( @xmath1 ) as a type ii supernova .
since the energy released in neutrinos by the collapsed stellar iron core is more than 100 times larger than the kinetic energy of the explosion , only a small fraction of the neutrino energy is sufficient to expel the mantle and envelope of the progenitor star .
numerical simulations have demonstrated the viability of this neutrino - driven explosion mechanism ( wilson @xcite ; bethe & wilson @xcite ; wilson et al .
@xcite ) but the explosions turned out to be sensitive to the size of the neutrino luminosities and the neutrino spectra ( janka @xcite ; janka & mller @xcite ; burrows & goshy @xcite ) both of which determine the power of the neutrino energy transfer to the matter outside the average neutrinosphere .
the rate of energy deposition per nucleon via the dominant processes of electron neutrino absorption on neutrons and electron antineutrino absorption on protons is given by : @xmath2 here @xmath3 and @xmath4 are the number fractions of free neutrons and protons , respectively ; the normalization with the baryon density @xmath5 indicates that the rate per baryon is calculated in eq .
( [ eq-1 ] ) .
@xmath6 denotes the luminosity of either @xmath7 or @xmath8 in units of @xmath9 and @xmath10 is the radial position in @xmath11 .
the average of the squared neutrino energy , @xmath12 , is measured in units of @xmath13 and enters through the energy dependence of the neutrino and antineutrino absorption cross sections .
@xmath14 is the angular dilution factor of the neutrino radiation field ( the `` flux factor '' , which is equal to the mean value of the cosine of the angle of neutrino propagation relative to the radial direction ) which varies between values much less than unity deep inside the protoneutron star atmosphere , about 0.25 around the neutrinosphere , and 1 for radially streaming neutrinos very far out .
the factor @xmath14 determines the local neutrino energy density according to the relation @xmath15 and thus enters the heating rate of eq .
( [ eq-1 ] ) . only far away from the neutrino
emitting star , @xmath16 , and @xmath17 dilutes like @xmath18 .
although it was found in two - dimensional simulations that convective instabilities in the neutrino - heating region can help the explosion ( herant et al .
@xcite ; janka & mller @xcite , @xcite ; burrows et al .
@xcite ; miller et al .
@xcite ; shimizu et al .
@xcite ) by the exchange of hot gas from the heating layer with cold gas from the postshock region , the strength of this convective overturn and its importance for the explosion is still a matter of debate ( janka & mller @xcite , mezzacappa et al .
@xcite , lichtenstadt et al .
in addition , it turns out that the development of an explosion remains sensitive to the neutrino luminosities and the mean spectral energies even if convective overturn lowers the required threshold values .
this is the case because convective instabilities can develop sufficiently quickly only when the heating is fast and an unstable stratification builds up more quickly than the heated matter is advected from the postshock region through the gain radius ( which is the radius separating neutrino cooling inside from neutrino heating outside ) down onto the neutron star surface ( janka & mller @xcite , janka & keil @xcite ) .
`` robust '' neutrino - driven explosions might therefore require larger accretion luminosities ( to be precise : a larger value of the product @xmath19 in eq .
( [ eq-1 ] ) ) during the early post - bounce phase , or might call for enhanced neutrino emission from the core .
the latter could be caused , for example , by convective neutrino transport within the nascent neutron star ( burrows @xcite ; mayle & wilson @xcite ; wilson & mayle @xcite , @xcite ; keil et al .
@xcite ) or , alternatively , by a suppression of the neutrino opacities at nuclear densities through nucleon correlations ( sawyer @xcite ; horowitz & wehrberger @xcite ; raffelt & seckel @xcite ; raffelt et al .
@xcite ; keil et al .
@xcite ; janka et al .
@xcite ; burrows & sawyer @xcite,@xcite ; reddy et al .
@xcite ) , nucleon recoil and blocking ( schinder @xcite ) and/or nuclear interaction effects in the neutrino - nucleon interactions ( prakash et al . @xcite ; reddy et al .
@xcite , @xcite ) , all of which have to date not been taken into account fully self - consistently in supernova simulations .
the diffusive propagation of neutrinos out from the very opaque inner core is determined by the value of the diffusion constant and thus sensitive to these effects . most of the current numerical treatments of neutrino transport , however , are deficient not only concerning their description of the extremely complex neutrino interactions in the dense nuclear plasma but also concerning their handling of the transition from diffusion to free streaming .
while the core flux is fixed in the diffusive regime , the accretion luminosity as well as the spectra of the emitted neutrinos depend on the transport in the semitransparent layers around the sphere of last scattering .
since neutrino - matter interactions are strongly dependent on the neutrino energy , neutrinos with different energies interact with largely different rates and decouple in layers with different densities and temperatures .
the spectral shape of the emergent neutrino flux is therefore different from the thermal spectrum at any particular point in the atmosphere .
even more , through the factor @xmath14 in the denominator of eq .
( [ eq-1 ] ) the energy deposition rate depends on the angular distribution of the neutrinos in the heating region .
a quantitatively reliable description of these aspects requires the use of sophisticated transport algorithms which solve the boltzmann equation instead of approximate methods like flux - limited diffusion techniques ( janka @xcite , @xcite ; mezzacappa & bruenn @xcite,@xcite,@xcite ; messer et al .
the detection of electron antineutrinos from sn 1987a in the kamiokande ii ( hirata et al .
@xcite ) and imb laboratories ( bionta et al .
@xcite ) and the construction of new , even larger neutrino experiments for future supernova neutrino measurements have raised additional interest in accurate predictions of the detectable neutrino signals from type ii supernovae .
neutrino transport in core collapse supernovae is a very complex problem and difficult to treat accurately even in the spherically symmetric case .
some of the major difficulties arise from the strong energy dependence of the neutrino interactions , the non - conservative and anisotropic nature of the scattering processes such as neutrino - electron scattering , the non - linearity of the reaction kernels through neutrino fermi blocking , and the need to couple neutrino and antineutrino transport for the neutrino - pair reactions .
therefore various simplifications and approximations have been employed in numerical simulations of supernova explosions and neutron star formation .
the so far most widely used approximation with a high degree of sophistication is the ( multi - energy - group ) flux - limited diffusion ( mgfld ) ( bowers & wilson @xcite , bruenn @xcite , myra et al .
@xcite , suzuki @xcite , lichtenstadt et al .
@xcite ) where a flux - limiting parameter is employed in the formulation of the neutrino flux to ensure a smooth interpolation between the diffusion regime ( where the neutrinos are essentially isotropic ) and the free streaming regime ( where the neutrinos move radially outward ) .
although the limits are accurately reproduced , there is no guarantee that the intermediate regime is properly treated . since in a quasi - stationary situation ( e.g. , for the cooling protoneutron star )
the flux and the mean energy of the emitted neutrinos are determined in the diffusion regime , little change of these is found when the flux - limiter is varied ( suzuki @xcite ) or the transport equation is directly solved , e.g. , by monte carlo calculations ( janka @xcite ) .
this , however , is not true when the spectral form is considered , because the spectra are shaped in the semitransparent surface - near layers .
moreover , significant differences are also expected for problems where the local angular distribution is important in the region between the diffusion and free streaming limits . due to the factor
@xmath14 appearing in eq .
( [ eq-1 ] ) the hot - bubble heating is such a problem , neutrino - antineutrino pair annihilation is another problem of this kind .
in fact , monte carlo simulations ( janka & hillebrandt @xcite,@xcite ; janka @xcite ; janka et al .
@xcite ) have shown that all flux - limiters overestimate the anisotropy of the radiation field outside the neutrinosphere , i.e. , @xmath20 is enforced too rapidly ( see also cernohorsky & bludman @xcite ) .
this leads to an underestimation of the neutrino heating in the hot - bubble region between neutrinosphere and supernova shock
( [ eq-1 ] ) ) , and sensitivity of the supernova dynamics to the employed flux - limiting scheme must be expected ( messer et al .
@xcite , lichtenstadt et al .
@xcite ) .
modifications of flux - limited diffusion have been suggested ( janka @xcite , @xcite ; dgani & janka @xcite , cernohorsky & bludman @xcite ) by which considerable improvement can be achieved for spherically symmetric , static and time - independent backgrounds ( smit et al .
@xcite ) , but satisfactory performance for the general time - dependent and non - stationary case has not been demonstrated yet .
therefore the interest turns towards direct solutions of the boltzmann equation for neutrino transport , also because the need to check the applicability of any approximation with more elaborate methods remains .
moreover , the rapid increase of the computer power and the wish to become independent of ad hoc constraints on generality or accuracy yield a motivation for the efforts of several groups ( in particular mezzacappa & bruenn @xcite,@xcite,@xcite and messer et al .
@xcite ; more recently also burrows @xcite ) to employ such boltzmann solvers in neutrino - hydrodynamics calculations of supernova explosions .
there are different possibilities to solve the boltzmann equation numerically , one of which is by straightforward discretization of spatial , angular , energy , and time variables and conversion of the differential equation into a finite difference equation which can then be solved for the values of the neutrino phase space distribution function at the discrete mesh points .
dependent on the number @xmath21 of angular mesh points , this procedure is called s@xmath22 method . since solving
the equation is computationally very expensive , there are limitations to the resolution in angle and energy space .
therefore tests need to be done whether a chosen ( and affordable ) number of energy and angle grid points is sufficient to describe the spectra well and , in particular , to reproduce the highly anisotropic neutrino distribution outside the neutrinosphere .
another , completely different approach to solve the boltzmann equation is the monte carlo ( mc ) method by which the probabilistic history of a large number of sample neutrinos is followed to simulate the neutrino transport statistically ( tubbs @xcite ; janka @xcite , @xcite ; janka & hillebrandt @xcite ) . in principle
, the accuracy of the results is only limited by the statistical fluctuations associated with the finite number of sample particles .
since the mc transport essentially does not require the use of angle and energy grids , it allows one to cope with highly anisotropic angular distributions and to treat with high accuracy neutrino reactions with an arbitrary degree of energy exchange between neutrinos and matter .
however , the mc method is also computationally very time consuming , in particular if high accuracy on a fine spatial grid or at high optical depths is needed .
therefore it is not the transport scheme of one s choice for coupling it with a hydrodynamics code . in the present work ,
we make use of the advantages of the mc method in order to test the accuracy and reliability of a newly developed neutrino transport code that follows the lines of the s@xmath0 scheme described by mezzacappa & bruenn ( @xcite,@xcite,@xcite ) . in particular
, we shall test the influence of the number of radial , energy , and angular mesh points on predicting the spectra and the radial evolution of the neutrino flux in `` realistic '' protoneutron star atmospheres as found in hydrodynamic simulations of supernova explosions ( wilson @xcite ) . since our investigations
are restricted to static and time - independent backgrounds , we concentrate on generic properties of the transport description which should also hold for more general situations .
the radial evolution of the angular distribution of the neutrinos is such a property , because it is primarily dependent on the profile of the opacity and the geometry of the neutrino - decoupling region , but is not very sensitive to the details of the temperature and composition in the neutron star atmosphere .
finally , good overall agreement of the mc and s@xmath0 results would strengthen the credibility of the mc transport with its limited ability to yield high spatial resolution .
the paper is organized as follows .
the details of the boltzmann solver and essential information for the mc method are given in sect .
2 . in sect . 3
we describe the background models .
section 4 presents the results of our comparative calculations , i.e. , neutrino spectra , luminosities , and eddington factors .
some of our calculations are also compared against results obtained with a mgfld code developed by suzuki ( @xcite ) .
the dependence of the results from the s@xmath0 scheme on the energy , angular , and radial grid resolution is discussed , too .
it is shown that an angular mesh that varies with the position in the star can improve the angular resolution and the representation of the beamed neutrino distributions without increasing the number of angular mesh points .
finally , a summary of our results and a discussion of their implications can be found in sect .
our s@xmath0 code is based on a finite difference form of the general relativistic boltzmann equation for neutrinos .
we assume spherical symmetry of the star throughout this paper .
for the misner - sharp metric ( misner & sharp @xcite ) : @xmath23 the boltzmann equation in the lagrangian frame takes the following form : @xmath24 \left ( \frac{f_{\nu}}{\rho_{\rm b } } \right)\right\ } \nonumber \\ & + & \left [ \mu ^{2 } \frac{1}{\rm c } { \rm e}^{- \phi } \frac{\partial \ln ( \rho_{\rm b } r^{3})}{\partial \ , t } \right . \nonumber \\ & - & \frac{1}{\rm c } { \rm e}^{- \phi } \frac{\partial \ln r}{\partial \ , t } \nonumber \\ & - & \left .
\mu 4 \pi r^{2 } \rho_{\rm b } \frac{\partial \phi}{\partial m } \right ] \frac{\partial \ \ } { \partial \frac{1}{3 } \varepsilon _ { \nu } ^{3 } } \left\ { \varepsilon _ { \nu } ^{3 } \left ( { \frac{f_{\nu}}{\rho_{\rm b } } } \right ) \right\ } \nonumber \\ & = & \frac{1}{\rho_{\rm b } } \frac{1}{\rm c } { \rm e}^{- \phi } \left ( \frac{\delta f_{\nu}}{\delta \ , t } \right)_{\rm coll } \quad .
\end{aligned}\ ] ] in the above formula @xmath25 and @xmath26 are the velocity of light and the gravitational constant , respectively , @xmath27 is the coordinate time and @xmath28 is the baryonic mass coordinate which is related to the circumference radius @xmath29 through the conservation law of the baryonic mass . in view of combination with a lagrangian hydrodynamics code ( yamada @xcite ) ,
the baryonic mass is chosen to be the independent variable instead of the radius .
@xmath30 , @xmath31 and @xmath29 are the metric components which are determined by the einstein equations . in this paper , however , these quantities are given from the background models and set to be constant with time during the neutrino transport calculations .
@xmath32 is the neutrino phase space distribution function . under the assumption of spherical symmetry
, @xmath32 is a function of @xmath27 , @xmath28 , @xmath33 and @xmath34 , where @xmath33 is the cosine of the angle of the neutrino momentum with respect to the outgoing radial direction and @xmath34 is the neutrino energy .
@xmath35 is the baryonic mass density . the right hand side of eq .
( [ eqn : be ] ) is the so - called collision term , which actually includes absorption , emission , scattering and pair creation and annihilation of neutrinos , details of which are described below .
the left hand side , on the other hand , looks a little bit different from the form used , for example , in mezzacappa et al .
( @xcite,@xcite,@xcite ) .
this is not only because it is fully general relativistic but also because all the velocity dependent terms are expressed as time derivatives so that it can easily be coupled to the implicit general relativistic lagrangian hydrodynamics code ( yamada @xcite ) . in this way
a fully coupled , implicit system of the radiation - hydrodynamics equations is formed , in which the time derivatives can be treated easily because they are just off - diagonal components of the matrix set up from the linearized equations .
it should be noted , however , that since we assume that the matter background is static in this paper , all these time derivatives are automatically set to be zero , although these terms have been already implemented in the code .
the conserved neutrino number @xmath36 in the absence of source terms is represented in terms of the chosen independent variables as @xmath37 this suggests to cast eq .
( [ eqn : be ] ) in a conservation form with respect to the neutrino number .
it is also evident that the combination of @xmath38 is more convenient to be used than @xmath39 itself . in the following , therefore
, we define the specific neutrino distribution function as in mezzacappa et al .
( @xcite,@xcite,@xcite ) by @xmath40 and use this quantity as the dependent variable to be solved for . as mentioned above the specific neutrino distribution function @xmath41 is a function of @xmath27 , @xmath28 , @xmath33 and @xmath42 .
this four - dimensional phase space is discretized and eq .
( [ eqn : be ] ) is written as a finite - difference equation . in the time direction
we adopt a fully implicit differencing .
the discretized specific distribution function @xmath43 is defined at the mesh centers of the spatial , angular and energy grids . here
the subscripts @xmath44 refer to the spatial , angular and energy grid points , respectively .
the superscript @xmath45 corresponds to the time step .
the value at each cell interface is evaluated by interpolation of the distribution at two adjacent mesh centers .
our finite difference method is essentially the same as that of mezzacappa et al .
( @xcite,@xcite,@xcite ) with some modifications .
for the spatial advection , the upwind difference and the centered difference are linearly averaged with the weights determined by the ratio of the mean free path to the distance to the stellar surface unlike mezzacappa et al .
( @xcite,@xcite,@xcite ) who used the ratio of the mean free path to the local mesh width .
in fact , in the latter case we found that the upwind distribution was given too large a weight in the optically thick region and thus the flux was overestimated .
this issue will be addressed later .
the angular mesh is determined such that each mesh center and cell width correspond to the abscissas and weights of the gauss - legendre quadrature , respectively . in angular direction , the neutrino distribution at each interface is simply taken as the upwind value .
the advection in the energy space is also approximated by an upwind scheme following mezzacappa & matzner ( @xcite ) , although this does not allow to conserve both lepton number and energy in non - static situations ( which are not considered here ) unless a large number of energy zones is used ( mezzacappa , private communication ) . in typical calculations ,
105 , 6 and 12 mesh points are used for the spatial , angular and energy discretizations , respectively .
the dependence of the results on the numbers of mesh points will be discussed below .
the finite - differenced boltzmann equation forms a nonlinear coupled system of equations for all radial grid points which is linearized and solved iteratively by using a newton - raphson scheme .
the linearized equations adopt a block tridiagonal matrix form , which can be efficiently solved by the feautrier method .
different from the finite difference method , the monte carlo method constructs the statistical ensemble average by following the destinies of individual test particles and performing the average when all particles have been transported . due to the fact that neutrinos are fermions
it is impossible to propagate them independently .
instead , the full time - dependent problem has to be simulated by following a large number ( typically @xmath46 ) of sample particles along their trajectories simultaneously in order to be able to construct the local phase space occupation functions and to include anisotropies as well as phase space blocking effects self - consistently into the calculation of the reaction rates and source terms .
the modeling of the phase space distribution function from the local particle sample must guarantee the correct approach to chemical equilibrium .
also the pauli exclusion principle has to be satisfied by the statistical average .
we refer readers to janka ( @xcite ) and references therein for details .
the stellar background is divided into 15 equispaced spherical shells of homogeneous composition and uniform thermodynamical conditions , the number of which was determined both from physical requirements for spatial resolution and from the requirement to have acceptably small statistical errors in the local neutrino phase space distributions constructed from the chosen number of sample particles . although the monte carlo method is essentially mesh free , about 60 energy bins and approximately 35 angular bins are used only for representing the phase space distribution functions and for calculating the reaction kernels .
neutrinos are injected into the computational volume at the inner boundary in the way described in the next section , while particles passing inwardly through the inner boundary are simply forgotten .
the outer boundary is treated as a free boundary , where particles escape unhindered and no neutrino is assumed to come in from outside . in the presented calculations we are mainly interested in the neutrino transport in the region where the neutrinos decouple from the stellar background and the emitted spectra form .
therefore we calculate the neutrino transport only in the vicinity of the `` neutrinosphere '' . for this reason we have to set an inner boundary condition as well as an outer boundary condition for each model . at the outer boundary
we impose the condition that no neutrinos enter the computational volume from outside . in the boltzmann solver , this is realized by setting @xmath47 where @xmath48 is the radius of the outermost mesh center , and @xmath49 is the radius of the outer surface which is dislocated outward from the outermost mesh center by half a radial cell width . on the other hand
, we have to specify the distribution of the neutrinos coming into the computational volume at the inner boundary @xmath50 which is dislocated inward from the innermost mesh center @xmath51 by half a radial cell width . for this purpose
we adopt fermi - dirac distribution functions , in which the temperature , chemical potential and a normalization factor are determined such that the neutrino number density at @xmath50 ( where the neutrinos are essentially isotropically distributed because the inner boundary is chosen to be located at high optical depth ) , the average energy @xmath52 and the width of the energy spectrum , measured by the parameter @xmath53 ( see janka & hillebrandt 1989@xcite ) , are reproduced as given by wilson s ( @xcite ) models .
thus the inner boundary condition is set as : @xmath54 where @xmath55 , @xmath56 and @xmath57 are the fitting parameters , the values of which are summarized in table [ tab2 ] for all considered models and neutrino species . concerning the distribution of neutrinos that leave the computational volume
, we impose in eq .
( [ eqn : ibc ] ) in the boltzmann solver the condition that it is the same as the distribution at the innermost mesh center . from the physical point of view , however , and treated correctly in the monte carlo simulations , it should be determined by the fraction of neutrinos which is emitted or backscattered towards the inner boundary .
this is in general different from what the phase space distribution at the innermost mesh center yields because the mean free path of the neutrinos near the inner boundary is shorter than the mesh width . as a result
the imposed condition for @xmath58 in eq .
( [ eqn : ibc ] ) leads to a minor discrepancy of the treatment of the inner boundary condition in the monte carlo and boltzmann computations and sometimes causes a small oscillation of the neutrino distribution near the inner boundary in the latter computations .
this issue will be revisited later .
ccrrr models & neutrinos & @xmath57 [ mev ] & @xmath56 [ mev ] & + w1 & @xmath59 & 9.56932 & 7.85190 & 1.221490 + & @xmath60 & 9.38259 & 7.13610 & 3.874140 + & @xmath61 & 10.32539 & 30.87195 & 17.911456 + w2 & @xmath60 & 10.52454 & 5.75932 & 3.672680 + w3 & @xmath60 & 9.87875 & 11.27370 & 7.108076 + the neutrino opacities of dense neutron star matter are still one of the major uncertainties of supernova simulations .
theoretical and numerical complications arise from the description and treatment of nucleon thermal motion and recoil ( schinder @xcite ) , nuclear force effects and nucleon blocking ( prakash et al .
@xcite ; reddy et al .
@xcite ) , and nucleon correlations , spatial ( sawyer @xcite ; burrows & sawyer @xcite,@xcite ; reddy et al .
@xcite ) as well as temporal ( raffelt & seckel @xcite ; raffelt et al .
although in particular nucleon recoil and auto - correlations might play an important role even in the sub - nuclear outer layers of the protoneutron star down to densities below @xmath62 ( janka et al .
@xcite ; hannestad & raffelt @xcite ) we do not concentrate on this problem here but rather employ the standard description of the neutrino opacities , according to which neutrinos interact with isolated nucleons ( see , e.g. , tubbs & schramm @xcite ; bruenn @xcite ) .
also , as in most other simulations , bremsstrahlung production of neutrino - antineutrino pairs is neglected here , although it may be important as pointed out by suzuki ( @xcite ) and more recently by burrows ( @xcite ) and hannestad & raffelt ( @xcite ) . doing so , we intend to enable comparison with other ( already published ) work and want to avoid the mixing of effects from a different numerical treatment of the transport with those from a non - standard description of neutrino - nucleon interactions or from the inclusion of processes typically not considered in the past .
the following neutrino reactions have been implemented in our codes .
+ [ 1 ] @xmath63 + ( electron - type neutrino absorption on neutrons ) , + [ 2 ] @xmath64 + ( electron - type anti - neutrino absorption on protons ) , + [ 3 ] @xmath65 + ( neutrino scattering on nucleons ) , + [ 4 ] @xmath66 + ( neutrino scattering on electrons ) .
+ in addition to the above reactions , the following processes have also been implemented in both codes , although these reactions are not used in the present paper .
+ [ 5 ] @xmath67 + ( electron - type neutrino absorption on nuclei ) , + [ 6 ] @xmath68 + ( neutrino coherent scattering on nuclei ) , + [ 7 ] @xmath69 + ( electron - positron pair annihilation and creation ) , + [ 8 ] @xmath70 + ( plasmon decay and creation ) . + since the pair processes are not taken into account in this paper , we can treat each species of neutrinos separately .
a test showed that in the considered protoneutron star atmospheres neutrino pair creation and annihilation as well as processes involving nuclei are unimportant to determine the fluxes and spectra .
in this paper we use a simplified equation of state , in which only nucleons , electrons , alpha particles and photons are included .
they are all treated as ideal gases .
for given density , temperature and electron fraction we derive the mass fractions and the chemical potentials of nucleons and the electron chemical potential from this equation of state .
the disregard of nuclei is well justified for the densities and temperatures we are considering , where most of the nuclei are dissociated into free nucleons .
this is actually confirmed by comparing our eos with the more realistic eos of wolff that is based on the skyrme - hartree - fock method ( hillebrandt & wolff @xcite ) .
only very small differences of the nucleon chemical potentials are found for the innermost region where small amounts of nuclei appear and for the outermost region where some contribution from alpha particles is mixed into the stellar medium .
we also repeated some of the calculations making use of the wolff eos with the nuclei - related reactions [ 5 ] , [ 6 ] ( neutrino emission , absorption and scattering on nuclei ) turned on and found qualitatively and quantitatively the same results .
hence the nuclei - related reactions are switched off in the calculations described below .
the time - dependent transport calculations presented here were performed for background profiles which are representative of protoneutron star atmospheres during the quasi - static neutrino cooling phase ( wilson @xcite ) . at this stage ,
several seconds after core bounce , the typical evolution timescale of density , temperature , and electron fraction is much longer than the timescale for neutrinos to reach a stationary state .
therefore our assumption of a static and time - independent background is justified . in addition , our interest is focused on the radial evolution of the eddington factors and on a test of the influence of the energy and angle resolution used in the s@xmath0 boltzmann solver .
both aims do not require a fully self - consistent approach which takes into account the evolution of the stellar background ( in particular of the temperature and composition ) .
in fact , the eddington factors are normalized angular integrals of the radiation intensity and as such reflect very general characteristics of the geometrical structure of the atmosphere where neutrinos and matter decouple . profiles from wilson s ( @xcite ) protoneutron star model were taken for three different times , 3.32 , 5.77 , and 7.81 s after core bounce . with the chosen fundamental variables density @xmath71 , temperature @xmath72 , and electron fraction @xmath73
, the thermodynamical state is defined for the plasma consisting of non - relativistic free nucleons , arbitrarily relativistic and degenerate electrons and positrons , and photons in thermal equilibrium .
figures [ fig_1][fig_3 ] show the input used for the three models . in fig .
[ fig_3 ] also the general relativistic metric coefficients @xmath74 and @xmath75 are given as provided by wilson s data and used for a comparative general relativistic calculation of the neutrino transport in model w3 .
= 8.8 cm = 8.8 cm = 8.8 cm all models computed with the boltzmann code are summarized in table [ tab3 ] .
models st are the standard models , in which 105 uniform spatial , 6 angular and 12 energy mesh points were used .
the energy mesh is logarithmically uniform and covers 0.9 110 mev .
the numbers of angular grid points and energy grid points were increased in models fa and fe , respectively . in model
cs we used the same radial grid as in the monte carlo simulations where 15 radial zones were chosen .
105 spatial mesh points were again used in model ni with no interpolations of density , temperature and electron fraction in the radial grid of the monte carlo simulations .
model gr took into account the general relativistic effects .
we used a non - uniform spatial mesh in model nu .
a different interpolation of up - wind differencing and centered differencing was tried for the radial advection term in model di .
we assumed the nucleon scattering to be isotropic in model is .
as is understood from table [ tab3 ] , most of the comparisons were done for the electron - type anti - neutrinos , since they are most important from the observational point of view .
clccl model & & background model & @xmath76 species & + st1 & @xmath77 & w1 & @xmath78 & interpolation in spatial grid of mc + st2 & @xmath77 & w1 & @xmath60 & interpolation in spatial grid of mc + st3 & @xmath77 & w2 & @xmath60 & interpolation in spatial grid of mc + st4 & @xmath77 & w3 & @xmath60 & interpolation in spatial grid of mc + st5 & @xmath77 & w1 & @xmath61 & interpolation in spatial grid of mc + fa & @xmath79 & w1 & @xmath60 & interpolation in spatial grid of mc + fe & @xmath80 & w1 & @xmath60 & interpolation in spatial grid of mc + cs & @xmath81 & w1 & @xmath60 & same spatial grid as in mc + ni & @xmath77 & w1 & @xmath60 & no interpolation in spatial grid of mc + gr & @xmath77 & w3 & @xmath60 & general relativity included + nu & @xmath77 & w1 & @xmath60 & non - uniform spatial mesh + di & @xmath77 & w1 & @xmath82 & different interpolation for conservative radial advection + is & @xmath77 & w1 & @xmath60 & isotropic @xmath76-@xmath21 scattering +
in the following , we consider the neutrino transport results after the time - dependent simulations have reached steady states .
first we compare observable quantities such as the luminosity , average energy and average squared energy of the neutrino flux for models st and gr .
these quantities are calculated at the outermost spatial zone as @xmath83 where @xmath49 is again the surface radius , and the redshift corrections from the surface up to infinity are not taken into account .
the results are summarized in table [ tab4 ] .
cccccccc & & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & @xmath89 + @xmath90 & b & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 & @xmath96 + & mc & @xmath97 & @xmath92 & @xmath98 & @xmath94 & @xmath95 & @xmath96 + @xmath99 & b & 12.8 & 16.3 & 15.9 & 16.2 & 24.3 & 15.5 + & mc & 12.7 & 16.1 & 15.8 & 16.3 & 24.2 & 15.6 + @xmath100 & b & 198.5 & 329.0 & 309.9 & 324.5 & 727.8 & 300.3 + & mc & 198.6 & 322.6 & 308.9 & 327.1 & 724.3 & 300.6 + the electron - type neutrino has the lowest energy while the muon - type neutrino has the highest .
the reason for this is that the electron - type neutrino has the shortest mean free path due to absorptions on the abundant neutrons , and decouples in the surface - near layers where the temperature is lower .
in contrast , the muon and tau neutrinos do not interact with particles of the stellar medium by charged currents and therefore their thermal decoupling occurs at a higher temperature .
the luminosity for the electron - type anti - neutrino gets smaller as time passes , which is due to the cooling and shrinking of the protoneutron star .
as can be seen in the table , the agreement of all quantities between the two methods is very good , which confirms the statistical convergence of the monte carlo simulations .
the average energy is in general determined accurately because the energy spectrum is shaped in the region where the neutrino angular distribution is not very anisotropic .
moreover , possible effects due to the rather coarse angular resolution of the boltzmann code essentially cancel out by taking the ratios of eqs .
( [ eqn : emin ] ) and ( [ eqn : e2min ] ) . on the other hand ,
the luminosity and the number flux of the monte carlo computations are also well reproduced by the boltzmann results .
this is due to the fact that these quantities are also determined deep inside the star and are nearly conserved farther out .
in fact , when integrating eq .
( [ eqn : be ] ) over angle and energy multiplying with unity and @xmath34 , respectively , ignoring all time derivatives and general relativistic effects , one gets @xmath101 where @xmath102 and @xmath103 are the number and energy fluxes of neutrinos at radius @xmath29 , respectively , and defined as , @xmath104 as expected intuitively , the scattering kernels drop out of the right hand side of eq .
( [ eqn : nexch ] ) while only the isoenergetic scattering does not contribute on the right hand side of eq .
( [ eqn : eexch ] ) . in the given boltzmann code eqs .
( [ eqn : nexch ] ) and ( [ eqn : eexch ] ) are discretized in a conservative form for the radial advection .
therefore it is clear that the boltzmann code can calculate the number and energy fluxes accurately if also the number and energy exchange by the reactions are calculated accurately in the source terms on the right hand sides of the equations .
moreover , the radial evolutions of @xmath105 and @xmath106 are entirely determined by the number and energy exchange through reactions of which the net effect is small in an atmospheric layer which is in a stationary state and reemits as much energy and lepton number as it absorbs .
changes of the number fluxes and luminosities in the considered protoneutron star atmospheres occur only in regions where the neutrino distribution is still essentially isotropic and possible effects due to an insufficient angular resolution in the boltzmann s@xmath0 scheme do not cause problems .
for all these reasons it is not surprising that the same quality of agreement is also found for the radial evolutions of the luminosity , average energy and average squared energy and that this is also true for the general relativistic case . in the previous section we discussed only the energy and angle integrated quantities .
however , we also provide information about the energy spectra of each neutrino species , because they yield more evidence about the quality of the agreement between the calculations with the different codes . figs .
[ fig_4 ] and [ fig_5 ] show energy flux spectra defined as @xmath107 at the protoneutron star surface for different cases . for the reasons discussed in section [ la ] ,
the spectra computed with the boltzmann code ( symbols ) and the monte carlo code ( lines ) show excellent agreement .
the number and the distribution of the energy bins in the boltzmann code seem to be adequate to reproduce the highly resolved monte carlo spectra .
= 8.8 cm = 8.8 cm so far we discussed only angle integrated quantities since they are observable .
however we are also interested in the angular distributions of the neutrinos , because information about the angular distributions is important to determine the neutrino heating rate in the hot - bubble region ( see eq .
( [ eq-1 ] ) ) .
although it is an advantage of the boltzmann solver over mgfld that one does not have to assume an ad hoc closure relation between the angular moments of the distribution function , one should remember that the usable number of angular mesh points is severely limited . in the feautrier method the computation time increases in proportion to the third power of the dimension of the blocks in the tridiagonal block matrix
which has to be inverted when one chooses the radius as the outermost variable of the do - loops .
the dimension of one block , in turn , is linearly proportional to the number of angular mesh points .
the same dependence holds for the number of energy mesh points and the number of neutrino species . in the standard calculations we use 6 angular mesh points and 12 energy grid points , and we treat a single neutrino species at a time , which corresponds to a block matrix size of 72 . on the other hand ,
the number of spatial grid points is about 100 .
the cpu time is a few seconds per inversion of the whole matrix on a single vector processor of a fujitsu vpp500 .
hence , use of more than 10 angular mesh points is almost prohibitive for calculations with three neutrino species even with a highly parallelized matrix inversion routine ( sumiyoshi & ebisuzaki @xcite ) .
it is , therefore , important to clarify the sensitivity of the accuracy to the angular resolution .
for this reason we consider the flux factor @xmath108 and the eddington factor @xmath109 which are defined as : @xmath110 here the subscript `` @xmath111 '' means that the averages are defined with the weight of energy . in mgfld
these factors are related with each other by a closure condition which can be derived from the employed flux - limiter ( janka @xcite , @xcite ) . for simplicity
we discuss here only energy integrated quantities as defined above .
the fundamental features are similar for the individual energy groups . in figs .
[ fig_6][fig_8 ] , we show the radial evolutions of the flux factors and the eddington factors for all neutrino species in case of background model w1 .
the upper panels show the flux factors and the lower panels the corresponding eddington factors .
the solid lines are the results of the boltzmann simulations ( having the finer radial resolution ) and the filled triangles are those of the monte carlo simulations .
as can be seen , near the inner boundary the flux factors are almost zero while the eddington factors are @xmath112 , which implies that the neutrino angular distribution is nearly isotropic , a consequence of the fact that the neutrinos are in equilibrium with the surrounding matter .
as we move outward , both factors begin to deviate from these values , reflecting the increase of the mean free path and a more rapid diffusion .
farther out , the angular moments increase monotonically towards unity , the value in the free streaming limit , as the angular distribution gets more and more forward peaked with increasing distance from the source . as can be seen clearly in figs .
[ fig_6][fig_8 ] , the boltzmann solver tends to _ underestimate _ both angular moments in the outermost region , where the neutrino angular distribution is most forward peaked .
this can be explained by the insufficient angular resolution , or , to be more specific , by the fact that in case of the employed gauss - legendre quadrature the maximum angle cosine @xmath113 of the angular grid is significantly less than unity , if not a large number of angular grid points is used .
this is directly confirmed by using a larger number of angular mesh points or , in particular , a variable angular mesh ( see sect .
[ vam ] ) .
the same trend is also present in the general relativistic case , model gr , as shown in fig .
[ fig_9 ] .
it turned out that the ray bending effect which tends to isotropize the angular distribution of the neutrinos is not very important and that differences between the monte carlo results and the boltzmann results are significantly larger . in fig .
[ fig_10 ] we show the flux factor and the eddington factor for model fa where we employed 10 angular mesh points instead of 6 .
the long dashed lines are for model fa and the short dashed lines depict the result of model st2 , the corresponding standard model , for comparison .
the discrepancy between the boltzmann simulation and the monte carlo simulation is reduced with the increase of the number of angular mesh points .
this supports our interpretation that the deviation stems entirely from the insufficient angular resolution and/or the unfavorable location of the angular grid points for the gauss - legendre quadrature used in the boltzmann simulations . indeed , the degree of improvement is consistent with the fact that our finite difference scheme is of first order for the angular advection , since we always take the upwind differencing as explained above .
even with the finer 10-zone angular mesh the deviation of both factors from the exact values given by the monte carlo result is significant .
the maximum deviations , @xmath114 for the flux factor and @xmath115 for the eddington factor , are approached as one goes farther out into the optically thin regime .
this is visible in fig .
[ fig_10 m ] which displays the ratios of the boltzmann to the monte carlo results for the flux factors and the eddington factors in case of 6 and 10 angular bins and the variable angular mesh .
( the relatively large discrepancies of the flux factors for smaller radii are explained by slight differences of the treatment of the inner boundary condition , see sect .
[ sec_boundary ] , and by the fact that the flux factor adopts very small values in the optically thick region . )
it should be mentioned that the flux factors calculated in the recent paper by messer et al .
( @xcite ) do not converge to unity but saturate at a nearly constant lower level ( around 0.9 ) even far outside of the neutrinosphere .
this reflects the use of @xmath116 for the largest @xmath33-bin of the angular grid in the 6-point quadrature of the s@xmath117 method .
it is interesting to see that this tendency of the boltzmann solver is completely opposite to that of mgfld .
janka ( @xcite , @xcite ) pointed out that all flux limiters used so far tend to overestimate the flux factor and the eddington factor in the optically thin region , which implies that the neutrino angular distribution approaches the free streaming limit much too rapidly ( see also messer et al .
@xcite ) . in order to confirm this statement , transport calculations with mgfld
were done for the same models with three different flux limiters , which are bruenn s ( br ) , levermore & pomraning s ( lp ) and mayle & wilson s ( mw ) .
we refer readers to janka ( @xcite ) and suzuki ( @xcite ) and references therein for details on the flux limiters .
we show in figs .
[ fig_11 ] and [ fig_12 ] the flux factors and local number densities of the electron - type neutrino and electron - type anti - neutrino for model w1 , respectively .
it is clear that all flux limiters overestimate the forward peaking of the angular distributions of the neutrinos in the optically thin region , a trend that holds for all neutrino species and is not dependent on the background model .
the typical deviation of mgfld results from the monte carlo results is much larger than that between the boltzmann solver and the monte carlo method .
= 8.8 cm = 8.8 cm = 8.8 cm = 8.8 cm = 8.8 cm = 8.8 cm = 8.8 cm = 8.8 cm from the lower panels of figs .
[ fig_11 ] and [ fig_12 ] we learn that the local neutrino number density , which is given by @xmath118 is overestimated in case of the boltzmann solver ( by about 10% ) and underestimated for mgfld ( by approximately 30% ) in the optically thin region .
this is understood from the fact that the number flux , which is defined as @xmath119 is related to the local neutrino number density by @xmath120 @xmath121 denotes the average angle cosine for the neutrino number flux and is calculated from eq .
( [ eqn : ffac ] ) with a factor @xmath122 omitted under the integrals in the numerator and denominator . since the number flux is determined deep inside the protoneutron star atmosphere where the neutrinos are still nearly isotropic , and conserved farther out , it is not affected by problems with a coarse angular resolution occurring in the optically thin regime . for this reason , the number and energy fluxes agree well between the monte carlo method and the boltzmann solver irrespective of the angular resolution as long as the boltzmann solver is based on conservative finite differencing in the radial direction .
the good agreement of the fluxes is confirmed by fig .
[ fig_13 ] , which depicts the radial behavior of the number flux in case of models st1 , st2 and st5 .
it is now clear from eq .
( [ eqn : nfac ] ) that an underestimation ( overestimation ) of the flux factor leads to an overestimation ( underestimation ) of the number density , if the flux is the same .
it should be noted that even in the outermost zone of our computational region in the atmosphere of a protoneutron star , the neutrino angular distribution is not so strongly forward peaked as in the hot - bubble region farther out .
hence it must be expected that the errors by an over- or underestimation of the neutrino number density might be even larger in the hot - bubble region where neutrino heating takes place .
since the neutrino heating rate is proportional to the local neutrino number density ( actually : energy density)this is why the inverse of the flux factor appears in eq .
( [ eq-1])the application of the boltzmann solver with a relatively small number of angular mesh points may lead to an overestimation of the neutrino energy deposition in the hot bubble for disadvantageous situations , in particular when most of the heating occurs in those regions where the deviation of the flux factor from the correct value is significant .
in contrast , all flux limiters used in mgfld underestimate the heating rate significantly .
therefore , even for the boltzmann solver , improvement of the angular resolution or a redistribution of the angular grid points is desirable in order to a priori avoid inaccurate evaluation of the heating rate .
as already mentioned , an increase of the number of angular mesh points is not feasible .
choosing a variable angular mesh which adjusts mesh point locations in dependence of time and spatial position might be a solution .
this issue will be discussed in the next subsection .
= 8.8 cm here we attempt to improve the angular resolution of the boltzmann solver by redistributing the angular grid points in dependence of time and position so that their density is enhanced in the forward direction in the optically thin region where the neutrino angular distribution becomes strongly forward peaked and the boltzmann solver underestimates the flux factor and the eddington factor .
this requires adding extra angular advection terms in the numerical scheme which compensate for the motions of the angular mesh points .
we assume that the position of each interface of the angular grid is a function of time , baryonic mass and neutrino energy , i.e. , @xmath123 .
integrating eq.([eqn : be ] ) over angular bins then leads to the following additional angular advection fluxes at each angular mesh interface i : @xmath124 @xmath125 @xmath126 \left ( \frac{\partial \ \mu _ { \rm i}}{\partial \frac{1}{3 } \varepsilon _ { \nu } ^{3 } } \right ) \left ( \frac{f_{\nu}}{\rho_{\rm b } } \right ) .\end{aligned}\ ] ] it is easy to understand that eqs .
( [ eq : vat])([eq : vae ] ) originate from the variability of the angular mesh points because of the differentials of the @xmath127 with respect to time , mass and neutrino energy
. since the neutrino reaction rates are strongly energy dependent and so is the neutrino angular distribution , it would be desirable to implement the energy dependent angular mesh according to eq .
( [ eq : vae ] ) . in the current preliminary attempt ,
however , we installed only eqs .
( [ eq : vat ] ) and ( [ eq : vam ] ) for simplicity . incidentally , since eq .
( [ eq : vae ] ) is proportional to @xmath128 in static background calculations , it is anyway negligible for the models considered here .
we note that the motion of mesh points is not calculated implicitly , that is , the angular mesh points for the next time step are determined from the neutrino angular distribution at the current time step and are kept fixed during the implicit calculation of the transport for the next step . in fig .
[ fig_10 ] , we show both the flux factor and the eddington factor obtained from the computation with the variable angular mesh .
the improvement is clear from a comparison with the result of model fa which employed 10 angular mesh points and is also shown in the figure .
it should be emphasized that increasing the number of angular mesh points from 6 to 10 leads to an increase of cpu time by a factor of @xmath1294.5 , while the additional operations for the variable angular mesh imply negligible computational load .
we repeated all boltzmann calculations with the variable angular mesh method and found that the same improvement could be achieved for all cases .
our scheme is stable at least for the static background models , although the stability for dynamical background models remains to be tested .
thus we think this method is promising in applying the boltzmann solver to the study of neutrino heating in the hot - bubble region of supernovae , although there is room for improvement concerning the prescription of the motion of the mesh points and the implementation of the energy - dependent angular mesh . in this section
we discuss how the numerical results change in dependence of the number of spatial and energy grid points , the boundary condition and the treatment of the radial advection in the boltzmann solver . in fig .
[ fig_14 ] we show the energy spectrum of electron - type anti - neutrinos for model fe , in which we used 18 energy mesh points , compared with the spectrum for the corresponding standard model st2 which has 12 energy zones .
no qualitative or quantitative difference is found between the two cases .
this is also true for the luminosity and the angular distribution .
thus we think that about 15 energy mesh points are sufficient for the calculation of the energy spectrum .
these results are in agreement with previous findings by mezzacappa & bruenn ( @xcite,@xcite,@xcite ) .
= 8.8 cm in models cs and ni we reduced the spatial resolution , because the monte carlo simulations were done with only 15 radial mesh points which were used to represent the stellar background on which the reaction kernels were evaluated .
another motivation for testing the sensitivity to the radial resolution is that it is hardly possible to describe the protoneutron star atmosphere with about 100 radial grid points in the context of a full supernova simulation . in model
cs we used the same 15 spatial grid points as in the monte carlo simulations .
on the other hand , in model ni we used 105 spatial mesh points but the density , temperature and electron fraction were not interpolated between the grid points of the monte carlo simulations . in fig .
[ fig_15 ] we show the radial evolution of the average energy as defined in eq .
( [ eqn : emin ] ) and that of the number flux given by eq .
( [ eqn : nflx ] ) for model cs , to be compared with the corresponding result for model st2 in fig .
[ fig_13 ] .
it is clear that the agreement between the monte carlo and the boltzmann results for both quantities is good .
we note also that the angular distribution as well as the energy spectrum are hardly affected by this change of the spatial resolution .
model ni agrees with the monte carlo data nearly perfectly ( except for the problems with the angular distribution discussed in sect .
[ fefac ] ) after averaging over spatial mesh zones in accordance to the way the monte carlo data represent the transport result .
since the boltzmann results do not change with the number of radial grid points , we conclude that the quality of the numerical solutions is not degraded very much for simulations with a decreased spatial resolution .
= 8.8 cm minor oscillations of the number flux near the inner boundary can be seen in fig .
[ fig_13 ] .
this problem results from the fact that one can not consistently specify the distribution of neutrinos which leave the computational volume at the inner boundary .
while this distribution should be determined from the transport result just above the inner boundary , the boltzmann solver , however , requires an ad hoc specification in order to calculate the flux at the inner boundary .
this leads inevitably to an inconsistency of the flux in the innermost zone and thus to the observed oscillations .
in fact , when an inhomogeneous spatial mesh was used in model nu , in which the innermost grid zone was five times smaller than in the standard models , the oscillations were diminished as well . finally , we illustrate possible errors which are associated with the treatment of the finite differencing of the spatial advection term in the boltzmann solver . in the radial advection term a linear average of the centered difference and of the upwind difference
is used with a weight factor that changes according to the ratio of the neutrino mean free path to some chosen length scale .
mezzacappa et al .
( @xcite,@xcite,@xcite ) took the ratio of the mean free path to the local mesh width in order to construct the weighting .
however , we found that this does not work well if the mesh width becomes of the same order as the mean free path but is much smaller than the scale height of the background .
this is indeed the case in the inner optically thick region of our standard models with 105 radial zones . in a more recent version of his code ,
mezzacappa ( private communciation and @xcite ) defines the weighting factors by refering them to the neutrinospheric radius . in fig .
[ fig_16 ] the dashed line shows the number flux of muon - type neutrinos for model di which used the prescription suggested by mezzacappa et al .
( @xcite,@xcite,@xcite ) .
the flux is slowly increasing with radius because the upwind differencing is given too large a contribution in the optically thick region where the centered differencing should actually be chosen . as demonstrated by the solid line in fig .
[ fig_16 ] , the constancy of the flux , however , is recovered when the ratio of the mean free path to the distance up to the surface is chosen instead of the ratio of the mean free path to the local mesh width .
yet , this issue is probably not very important for realistic calculations of the entire neutron star , since the mesh width is usually not much smaller than the typical scale height of the matter distribution .
= 8.8 cm to finish , we comment briefly on a last model in which we assumed that the nucleon scattering is taken isotropic to see to what extent the result changes .
no significant effect was found by modifying the angular distribution of the dominant scattering reaction .
in this paper an extensive comparison was made between a newly developed boltzmann neutrino transport code based on the discrete ordinate ( s@xmath0 ) method as described by mezzacappa & bruenn ( @xcite,@xcite,@xcite ) , and a monte carlo transport treatment ( janka @xcite , @xcite ) by performing time - dependent calculations of neutrino transport through realistic , static profiles of protoneutron star atmospheres under the assumption of spherical symmetry .
in particular , the sensitivity of the results of the boltzmann solver to the employed numbers of radial , angular and energy grid points and to the treatment of the radial advection terms was investigated . the flux factor and eddington factor , which contain information about the angular distribution of the neutrinos in the neutrino - decoupling region ,
were also compared with the approximate treatment of this regime by a multi - energy - group flux - limited diffusion code ( mgfld ) .
the boltzmann and monte carlo results showed excellent agreement for observables such as the luminosity and the flux spectra which are determined in those regions of the star where the neutrino - matter interactions are still very frequent and thus the neutrino distributions are still nearly isotropic . since the luminosity and the spectra
are essentially conserved farther out , the spatial evolutions as well as the surface values exhibit this agreement , as long as the finite differencing of the boltzmann solver is done in a conservative way . some problems , however ,
were observed concerning the description of the angular distribution of the neutrinos by the boltzmann results in the semitransparent and transparent regimes . due to severe limitations of the number of angular grid points which can be used
typically only 610 angular bins between zero and 180 degrees are compatible with the steep increase of the requirements of computer time for better resolution the boltzmann code is not able to describe strong forward peaking of the neutrino distributions very well , if a quadrature set is employed for the angular integration where the maximum angle cosine @xmath113 is significantly less than unity . in this case
the boltzmann results _ underestimate _ the anisotropy in the optically thin region outside the average `` neutrinosphere '' , and the exact limits for the flux factor , @xmath130 , and for the eddington factor , @xmath131 , at large distances away from the neutrino source can not be satisfactorily reproduced .
this is in agreement with the trends also seen in recent results of messer et al .
( @xcite ) and is exactly opposite to the deficiencies of mgfld which tends to _ overestimate _ the radial beaming of the radiation because flux limiters enforce the free - streaming limit when the optical depth of the stellar background becomes very low ( janka @xcite , @xcite ) . since the dominant energy deposition rate by absorption of electron neutrinos and antineutrinos in the hot - bubble region of the supernova core is inversely proportional to the flux factor ( see eq .
( [ eq-1 ] ) ) , which means that the energy transfer from neutrinos to the stellar plasma scales with the neutrino energy density ( or number density ) in the heating region , one can not exclude that the boltzmann solver may lead to an overestimation of the neutrino heating in disadvantageous situations , whereas mgfld yields rates which are definitely too low .
for a set of typical post - core bounce situations , messer et al .
( @xcite ) , however , claim on grounds of numerical tests that the net neutrino heating is converged with s@xmath117 and that the differences between s@xmath117 and s@xmath132 are minor .
the problems may be more serious for neutrino reactions like neutrino - antineutrino annihilation which are sensitive to both the flux factor @xmath133 and the eddington factor @xmath134 of the neutrinos ( see janka @xcite ) .
the description of the angular neutrino distribution by the boltzmann solver and thus the agreement with the highly accurate monte carlo data can be significantly improved by employing a variable angular mesh , even without increasing the total number of angular mesh points .
the positions of the angular grid points must be moved at each time step and in every spatial zone such that they are clustering in the forward direction in the optically thin regime .
we found that the energy spectra can be well calculated with about 15 energy mesh points .
the fact that a reduction of the number of spatial grid points from more than 100 to only 15 in the neutron star atmosphere did not change the quality of the transport results means that the boltzmann code can be reliably applied to realistic simulations which involve the whole supernova core .
moreover , it was demonstrated that the details of the interpolation between centered differencing and upwind differencing in the spatial advection term can affect the accuracy of the transport results .
the excellent overall agreement of the results obtained with the boltzmann code and the monte carlo method confirms the reliability of both of them .
good performance of the s@xmath0 method for solving the boltzmann equation of neutrino transport has been found before by mezzacappa & bruenn ( @xcite,@xcite,@xcite ) and messer et al .
( @xcite ) even for realistic dynamic situations .
we hope that the work described here also helps to reveal possible deficiencies and weaknesses and thus will serve for further improvements of the numerical treatment of neutrino transport in supernovae and protoneutron stars .
encouraging discussions and helpful suggestions by e. mller are acknowledged .
we are grateful to a. mezzacappa for critical comments on a first version of the paper and for updating us with the most recent improvements of his code .
this work was partially supported by the japanese society for the promotion of science ( jsps ) , postdoctoral fellowships for research abroad , and by the supercomputer projects ( no.97 - 22 and no.98 - 35 ) of the high energy accelerator research organization ( kek ) .
t.j . was supported , in part , by the deutsche forschungsgemeinschaft under grant
the numerical calculations were mainly done on the supercomputers of kek .
bethe , h. a. and wilson .
j. r. 1985 , apj , 295 , 14 bionta , r. m. et al .
1987 , phys .
lett . , 58 , 1494 bowers , r. l. and wilson , j. r. 1982 , apjs , 50 , 115 bruenn , s. w. 1985 , apjs , 58 , 771 burrows , a. 1987 , apj , 318 , 57 burrows , a. 1997 , to be published in the proceedings of the 18th texas symposium on relativistic astrophysics , held in chicago , 15 - 20 december 1996 , edited by olinto , a. , frieman , j. and schramm , d. , ( world scientific press , singapore , 1997 ) burrows , a. and goshy , j. 1993 , apj , 416 , 75 burrows , a. , hayes , j. and fryxell , b. a. 1995 , apj , 450 , 830 burrows , a. , sawyer , r. f. 1998a , phys .
, c58 , 554 burrows , a. , sawyer , r. f. 1998b , submitted to phys .
cernohorsky , j. and bludman , s. a. 1994 , apj , 433 , 250 dgani , r. and janka , h .- th .
1992 , a&a , 256 , 428 hannestad , s. and raffelt , g. 1997 , to appear in apj herant , m. , benz , w. , hix , j. , freyer , c. and colgate , s. a. 1994 , apj , 435 , 339 hillebrandt , w. and wolff , r. g. 1985 , nucleosynthesis : challenges and new developments , edited by arnett , w. d. and truran , j. w. , ( university of chicago press , chicago , 1985 ) , p131 hirata , k. et al .
1987 , phys .
lett . , 58 , 1490 horowitz , c. j. and wehrberger , k. 1991 , phys .
b , 266 , 236 janka , h .- th .
1987 , nuclear astrophysics ; proceedings of the workshop , tegernsee , germany , apr . 21.24 . , 1987 , ( springer - verlag , berlin and new york , 1987 ) , p319 janka , h .- th .
1991a , ph.d .
thesis , technische universitt mnchen janka , h .- th .
1991b , a&a , 244 , 378 janka , h .- th .
1992 , a&a , 256 , 452 janka , h .- th .
1993 , frontier objects in astrophysics and particle physics , proc . of the vulcano workshop 1992 , conf .
40 , edited by giovannelli , f. and mannocchi , g. , ( sif , bologna , 1993 ) , p345 janka , h .- th . ,
dgani , r. and van den horn , l. j. 1992 , a&a , 265 , 345 janka , h .- th . and hillebrandt , w. 1989a , a&as , 78 , 375 janka , h .- th . and hillebrandt , w. 1989b , a&a , 224 , 49 janka , h .- th . and keil , w. 1998 , supernovae and cosmology , proc . of a colloquium in honor of prof .
g. tammann on the occasion of his 65th birthday , augst , switzerland , jun . 13 . , 1997 , edited by labhardt , l. , binggeli , b. and buser r. , ( astronomisches
institut der universitt basel , basel , 1998 ) p7 janka , h .- th . , keil , w. , raffelt , g. and seckel , d. 1996 , phys .
lett . , 76 , 2621 janka , h .- th . and mller , e. 1993 , frontiers of neutrino astrophysics , proc . of the international symposium on neutrino astrophysics , takayama / kamioka , japan , oct .
19.22 . , 1992 , edited by suzuki , y. and nakamura , k. , ( universal academy press , tokyo , 1993 ) p203 janka ,
h .- th . and mller , e. 1996 , a&a , 306 , 167 keil , w. , janka , h .- th . and mller , e. 1996 , apj lett .
, 473 , l111 keil , w. , janka , h .- th . and raffelt , g. 1995 , phys . rev . ,
d51 , 6635 lichtenstadt , i. , khokhlov , a. m. and wheeler , j. c. 1998 , submitted to apj mayle , r. and wilson , j. r. 1988 , apj , 334 , 909 messer , o. e. b. , mezzacappa , a. , bruenn , s. w. and guidry , m. w. 1998 , submitted to apj , astro - ph 9805276 mezzacappa , a. 1998 , j. computational and applied mathematics , in press mezzacappa , a. and matzner , r. a. 1989 , apj , 343 , 853 mezzacappa , a. and bruenn , s. w. 1993a , apj , 405 , 637 mezzacappa , a. and bruenn , s. w. 1993b , apj , 405 , 669 mezzacappa , a. and bruenn , s. w. 1993c , apj , 410 , 740 mezzacappa , a. et al . 1998 , apj , 495 , 911 miller , d. s. , wilson , j. r. and mayle , r. w. 1993 , apj , 415 , 278 misner , c. w. and sharp , d. h. 1964 , phys .
rev , 136 , b571 myra , e. s. et al .
1987 , apj , 318 , 744 prakash , m. et al .
1997 , phys .
, 280 , 1 raffelt , g. g. and seckel , d. 1995 , phys .
d52 , 1780 raffelt , g. g. , seckel , d. and sigl , g. 1996 , phys . rev . ,
d54 , 2784 reddy , s. , prakash , m. and lattimer , j. m. 1997 , apj .
, 478 , 689 reddy , s. , prakash , m. and lattimer , j. m. 1998a , to appear in proc .
second oak ridge symposium on nuclear and atomic and nuclear astrophysics reddy , s. , prakash , m. and lattimer , j. m. 1998b , phys .
, d58 , 1309 sawyer , r. f. 1989 , phys .
, c40 , 865 schinder , p. j. 1990 , apjs , 74 , 249 shimizu , t. , yamada , s. and sato , k. 1994 , apj lett . , 432 , l119 smit , j. m. , cernohorsky , j. and dullemond , c. p. 1997 , a&a , 325 , 203 sumiyoshi , k. and ebisuzaki , t. 1998 , parallel computing , 24 , 287 suzuki , h. 1990 , ph.d .
thesis , university of tokyo suzuki , h. 1993 , frontiers of neutrino astrophysics , proc . of the international symposium on neutrino astrophysics , takayama / kamioka , japan , oct .
19.22 . , 1992 , edited by suzuki , y. and nakamura , k. , ( universal academy press , tokyo , 1993 ) p219 suzuki , h. 1994 , physics and astrophysics of neutrinos , edited by fukugita , m. and suzuki , a. , ( springer - verlag , tokyo , 1994 ) , p763 tubbs , d. l. 1978 , apjs , 37 , 287 tubbs , d. l. and schramm , d. n. 1975 , apj , 201 , 467 wilson , j. r. 1982 , proc .
illinois meeting on numerical astrophysics wilson , j. r. 1988 , private communication wilson , j. r. and mayle , r. w. 1988 , phys .
163 , 63 wilson , j. r. and mayle , r. w. 1993 , phys .
227 , 97 wilson , j. r. , mayle , r. w. , woosley , s. e. and weaver , t. 1986 , ann .
n. y. acad .
, 470 , 267 yamada , s. 1997 , apj , 475 , 720 | we have coded a boltzmann solver based on a finite difference scheme ( s@xmath0 method ) aiming at calculations of neutrino transport in type ii supernovae .
close comparison between the boltzmann solver and a monte carlo transport code has been made for realistic atmospheres of post bounce core models under the assumption of a static background .
we have also investigated in detail the dependence of the results on the numbers of radial , angular , and energy grid points and the way to discretize the spatial advection term which is used in the boltzmann solver .
a general relativistic calculation has been done for one of the models .
we find overall good agreement between the two methods .
this gives credibility to both methods which are based on completely different formulations .
in particular , the number and energy fluxes and the mean energies of the neutrinos show remarkably good agreement , because these quantities are determined in a region where the angular distribution of the neutrinos is nearly isotropic and they are essentially frozen in later on . on the other hand , because of a relatively small number of angular grid points ( which is inevitable due to limitations of the computation time ) the boltzmann solver tends to underestimate the flux factor and the eddington factor outside the ( mean ) `` neutrinosphere '' where the angular distribution of the neutrinos becomes highly anisotropic . as a result , the neutrino number density is overestimated in this region .
this fact suggests that one has to be cautious in applying the boltzmann solver to a calculation of the neutrino heating in the hot - bubble region because it might tend to overestimate the local energy deposition rate .
a comparison shows that this trend is opposite to the results obtained with a multi - group flux - limited diffusion approximation of neutrino transport , employing three different flux limiters , all of which lead to an underestimation of the hot - bubble heating .
the accuracy of the boltzmann solver can be considerably improved by using a variable angular mesh to increase the angular resolution in the semi - transparent regime .
+ 3.0 cm |
one of the most important issues of the current high energy physics is to find the higgs boson and to reveal its nature .
the lhc will most likely find a higgs candidate and the ilc is expected to follow it to complete the mission @xcite@xcite . to study the higgs properties in detail we need a high performance central tracker .
a tpc is a natural candidate for the ilc central tracker because of its very good performance in the past collider experiments @xcite . at the ilc , however , we need the highest possible tracking efficiency in a jetty environment and a momentum resolution one order - of - magnitude better than those in the past .
this requires very high 3-dimensional granularity ( a small voxel size ) and a space point resolution in the @xmath4 plane better than 100 @xmath5 m throughout the sensitive volume . in order to realize a tpc with such unprecedented performance , intense r&d programs are now on going in an international framework called the lc - tpc collaboration .
three technologies have been considered as the gas amplification device for the ilc - tpc : mwpc , micromegas @xcite and gem @xcite . in order to compare the different readout technologies
, we prepared a small prototype tpc with a replaceable readout plane consisting of a gas amplification device and a pad plane .
the results of the beam test at kek for a micromegas readout using a gas mixture of ar - isobutane(5% ) have already been published @xcite . in this paper our results of cosmic ray tests for an mwpc readout using a tdr gas ( ar - ch@xmath3(5%)-co@xmath6(2% ) ) @xcite and for a triple gem readout operated in a t2k gas ( ar - cf@xmath3(3%)-isobutane(2% ) ) @xcite are presented .
the gem - based prototype is described in ref .
@xcite , where the preliminary results of the beam test for the gem readout using a tdr gas or a p5 gas ( ar - ch@xmath3(5% ) ) are presented .
the sensitive volume of the prototype was @xmath7 260 mm in length and 152 mm in diameter .
the size of the pad plane was 100 mm @xmath1 100 mm .
the induction gap of the triple gem stack was increased later from 1 mm to 1.5 mm for the cosmic ray tests using a t2k gas , in order to increase the charge spread after gas amplification , while the transfer gaps ( 1.5 mm ) were maintained . a picture of the prototype is shown in fig .
[ fig1 ] .
photograph of the prototype just before installation into the gas vessel . ] for the mwpc readout the gem foils were replaced by two wire planes @xcite .
the sense - wire plane consisted of 20 @xmath5 m thick wires spaced with 2 mm pitch and was placed 1 mm above the readout pad plane , on which pad rows were arranged in parallel to the wires .
the drift region was delimited by a grid wire plane .
the grid wire plane consisted of 127 @xmath5 m thick wires strung 1 mm above the sense - wire plane with 2 mm spacing .
it should be noted that the sense - wire pitch and the wires - to - pads distance were kept small as compared to those in conventional tpcs in order to obtain the best spatial resolution and the highest two - track resolving power achievable with a practical mwpc readout ) was about 1.3 mm . ] .
the pad pitch was increased to 2.3 mm , from 1.27 mm in the case of the gem readout , corresponding to the increase in the width of the pad response function while the pad - row pitch ( 6.3 mm ) was retained .
the mwpc - tpc was operated inside a 5-t solenoid at desy and the gem - tpc data was taken inside a 1-t solenoid at the kek cryogenic center . a pair of long scintillation counters was used for cosmic ray triggers .
the readout electronics system for the aleph tpc @xcite was used for the chamber signal processing and the traditional multifit ( doublefit ) program @xcite was used for the data analysis of the cosmic ray events .
the hit coordinate along a pad row was determined by a simple charge centroid ( barycenter ) method using the pad signals within a charge cluster in the row . among the tracks recorded in the tpc , those which were nearly perpendicular to the pad rows ( sense wires ) and parallel to the readout plane were selected for later analysis .
the spatial resolution as a function of the drift distance ( @xmath8 ) obtained under a magnetic field ( @xmath2 ) of 4 t is shown in fig .
the sense - wire potential was set to 1200 v while the grid wires were grounded .
the drift field strength ( @xmath9 ) was 220 v / cm , at which the transverse diffusion constant ( @xmath10 ) of drift electrons is about 60 @xmath5m/@xmath11 . also shown in the figure is the result of a monte carlo simulation .
the simulation took into account the primary ionization statistics of incident tracks , the diffusion of drift electrons , the lateral displacement of the drift electrons ( charge spread ) near the wire planes due to the @xmath12 effect , and finally the avalanche fluctuation in gas amplification . spatial resolution measured with the mwpc readout in a magnetic field of 4 t.
the solid line shows the result of the simulation . ]
the simulation program garfield / magboltz @xcite was used to estimate the displacement of the drift electrons along the wire direction ( @xmath13 ) due to the @xmath12 effect as a function of the initial position along the track ( @xmath14 ) ( see fig .
[ fig3 ] and [ fig4 ] ) .
the simulated resolution clearly shows the large contribution of the @xmath12 effect at short drift distances and its reduction with increasing @xmath8 due to the declustering effect caused by the diffusion of the drift electrons @xcite .
simulated drift electron paths without diffusion in tdr gas for a track perpendicular to the wires ( @xmath2 = 4 t ) . ]
displacement of the arrival position of drift electrons on the sense wire as a function of the initial position along the track for @xmath2 = 1 , 2 , 3 and 4 t. the coordinate @xmath14 is measured from the point on the track right above the sense wire .
the increase of displacement ( @xmath13 ) , especially for large @xmath14 , degrades the spatial resolution significantly . ]
[ fig5 ] shows the simulated spatial resolution of the ilc - tpc equipped with the mwpc readout for @xmath2 = 0 , 1 and 4 t. the resolution for @xmath2 = 4 t is dominated by the @xmath12 effect throughout the drift region in spite of its decrease at long drift distances where the contribution of the electron diffusion increases .
the situation is worse for t2k gas because of the larger charge spread near the wire planes caused by the @xmath12 effect and the smaller declustering effect due to its small diffusion constant ( @xmath10 @xmath7 30 @xmath5m/@xmath11 at @xmath9 = 250 v / cm and @xmath2 = 4 t ) .
therefore t2k gas would not be a good choice for mwpc readout even if the mwpc itself operated properly in this gas mixture .
simulated spatial resolution of the ilc - tpc equipped with the mwpc readout operated in tdr gas ( left ) and in t2k gas ( right ) for tracks perpendicular to the wires and the pad rows . ]
the spatial resolutions measured under magnetic fields of 0 and 1 t are shown in fig .
the high voltage applied across each gem of the triple stack was 250 v and the electric fields in the transfer and induction gaps were set to 1.6 kv / cm .
the solid curves in the figure are the results of analytic calculation @xcite .
the resolution deterioration at short drift distances prominent for the t2k gas at 1 t is not caused by an @xmath12 effect but by the finite pitch of the readout pads ( 1.27 mm ) , which is not small enough compared to the charge spread after gas amplification in the gem stack ( @xmath7 270 @xmath5 m ) @xcite . spatial resolution as a function of the drift distance obtained with a t2k gas ( @xmath9 = 250 v / cm ) for @xmath2 = 0 and 1 t. the resolutions obtained in the beam test with a p5 gas ( @xmath9 = 100 v / cm ) are also shown for comparison . ] the spatial resolution ( @xmath15 ) at long drift distances is given by @xmath16 where @xmath17 is the intrinsic resolution and @xmath18 is the effective number of electrons , which is determined by the primary ionization statistics and the avalanche fluctuation for a single drift electron @xcite . since the diffusion constant
is determined from the @xmath8-dependence of the pad response width @xcite@xcite ( or given by magboltz ) one can evaluate @xmath18 from the @xmath8-dependence of the spatial resolution at long drift distances .
the value of @xmath18 thus obtained was about 21 , which is in good agreement with a simple estimate of 22.2 @xcite , indicating that possible dissociative attachment of drift electrons by cf@xmath3 molecules @xcite@xcite@xcite at the entrance to the gem stack is not significant if any .
we also measured the diffusion constant and the drift velocity in a t2k gas as functions of the drift field , covering from the diffusion minimum ( @xmath9 = 100 v / cm ) to the drift velocity near - maximum ( @xmath9 = 250 v / cm ) .
the drift velocity was estimated from the full width of the drift time distribution and the maximum drift length of the prototype .
the results are shown in fig .
[ fig7 ] along with the corresponding magboltz predictions , which closely reproduce the data points .
the reliability of magboltz for this gas mixture was thus confirmed for the practical range of the electric field for tpc operation .
transverse diffusion constants ( @xmath19 ) and drift velocities ( @xmath20 ) measured in t2k gas .
the curves represent the magboltz predictions . .
] fig .
[ fig8 ] shows the calculated resolution of a gem - based ilc - tpc operated in a t2k gas , assuming the diffusion constant given by magboltz for @xmath9 = 250 v / cm .
expected resolution of a gem - based tpc operated in t2k gas for tracks perpendicular to the pad rows . ]
mwpc readout turned out to be inappropriate for a tpc operated in a strong magnetic field because of a large @xmath12 effect on its spatial resolution .
therefore it can not be a fall - back option for the ilc - tpc .
a gem - based tpc using a t2k gas is a good candidate for the central tracker of the ilc experiment .
it operates stably and is expected to give the required spatial resolution of better than 100 @xmath5 m throughout the sensitive volume under a 4 t magnetic field .
we are grateful to the kek cryogenic center and the ipns cryogenic group for the operation of the superconducting magnet . this study is supported in part by the creative scientific research grant no .
18gs0202 of the japan society for promotion of science .
ronan , time - projection chambers , review of particle physics , phys .
b592 , 1 ( 2004 ) 261 .
y. giomataris , et al .
instr . and meth . a 376 ( 1996 ) 29 .
f. sauli , nucl .
instr . and meth . a 386 ( 1997 ) 531 .
matthias e. janssen ( on behalf of the flc - tpc group at desy ) , r&d ongoing at desy for a gem based tpc at the ilc : resolution studies ; techniques and results , ieee nuclear science symposium conference record ( 2006 ) 38 , and references cited therein . | a time projection chamber ( tpc ) is a strong candidate for the central tracker of the international linear collider ( ilc ) experiment and we have been conducting a series of cosmic ray experiments under a magnetic field up to 4 t , using a small prototype tpc with a replaceable readout device : multi - wire proportional chamber ( mwpc ) or gas electron multiplier ( gem ) .
we first confirmed that the mwpc readout could not be a fall - back option of the ilc - tpc under a strong axial magnetic field of 4 t since its spatial resolution suffered severely from the so called @xmath0 @xmath1 @xmath2 effect in the vicinity of the wire planes . the gem readout , on the other hand ,
was found to be virtually free from the @xmath0 @xmath1 @xmath2 effect as had been expected and gave the resolution determined by the transverse diffusion of the drift electrons ( diffusion limited ) .
furthermore , gems allow a wider choice of gas mixtures than mwpcs
. among the gases we tried so far a mixture of ar - cf@xmath3-isobutane , in which mwpcs could be prone to discharges , seems promising as the operating gas of the ilc - tpc because of its small diffusion constant especially under a strong magnetic field .
we report the measured drift properties of this mixture including the diffusion constant as a function of the electric field and compare them with the predictions of magboltz .
also presented is the spatial resolution of a gem - based ilc - tpc estimated from the measurement with the prototype .
ilc , tpc , mwpc , gem , cf@xmath3 , spatial resolution 29.40.cs , 29.40.gx |
dileptons are crucial probes for studying the properties of the strongly interacting , hot and dense matter which is created in ultrarelativistic heavy - ion collisions at the relativistic heavy - ion collider ( rhic ) @xcite .
they are produced during the whole evolution of the created matter , and are not subject to strong interactions with the medium .
dielectron pairs are sensitive probes of the medium properties throughout the spacetime evolution of the medium @xcite because they are produced through a variety of mechanisms and in several different kinematic regimes . in the low invariant mass region , @xmath7 gev/@xmath4 ( lmr ) ,
the dilepton production is dominated by in - medium decay of vector mesons ( @xmath2 , @xmath8 and @xmath9 ) in the hadronic gas phase . in - medium
modifications to the mass and width of the vector mesons are considered a link to chiral symmetry restoration @xcite . in the vacuum , chiral symmetry
is spontaneously broken , which results in mass differences between chiral partners [ e.g. @xmath2 and @xmath10 . in the hot , dense medium , chiral symmetry
is expected to restore and the mass distributions of @xmath2 and @xmath11 are expected to change and degenerate . since it is extremely challenging to measure a spectral function for the @xmath11 meson , one can not directly observe the disappearance of the mass splitting between the @xmath2 and @xmath11 experimentally .
instead , efforts are devoted to studying the modification of vector meson spectral function .
two schematic scenarios are used to describe the in - medium @xmath2 spectrum function : a broadened and a dropping - mass @xmath2 .
the broadened @xmath2 scenario incorporates finite temperature effects into self - energy corrections through medium interactions and @xmath12 annihilations @xcite .
the dropping mass scenario uses the quark mean field from a high temperature / density regime wherein constituent quarks are the relevant degrees of freedom , and then extrapolates down to a low temperature / density regime wherein hadrons are appropriate degrees of freedom @xcite .
the ceres experiment at the cern - sps reported an excess dielectron yield with respect to the known hadronic sources in the lmr in pb+au collisions at @xmath1 = 17.2 gev , which indicates that the vector mesons are modified in medium @xcite .
more recently , na60 published a precise measurement of the dimuon invariant mass spectra in in+in collisions at @xmath1 = 17.3 gev @xcite .
the results show a significant excess in the lmr above the hadronic sources . in both cases ,
the excess is consistent with a broadened @xmath2 spectral function @xcite , but not with a @xmath2 dropping - mass scenario @xcite , where both models have been evaluated for the same fireball evolution . in the model calculation , the coupling to the baryons in the medium plays a dominant role in the broadening of the @xmath2 spectral function @xcite . at rhic ,
a significant enhancement in the dielectron continuum , compared with the known hadronic sources , has been observed in the lmr by both the phenix and star collaborations in au+au collisions at @xmath13 gev @xcite .
results from the star collaboration show that the excess dielectron yield in the mass region 0.3 - 0.76 gev/@xmath6 follows an @xmath14 dependence , where @xmath15 is the number of participant nucleons in a collision @xcite . however , the phenix collaboration reported significant higher excess dielectron yields in central collisions @xcite .
theoretical calculations @xcite , which describe the sps dilepton data , fail to consistently describe the low - mass enhancement at low transverse momentum ( @xmath16 ) observed by phenix in both 0 - 10% and 10 - 20% central au+au collisions @xcite . the same calculations
, however , correctly describe the star measurement of the low-@xmath16 and low - mass enhancement from peripheral to central au+au collisions @xcite . while the discrepancy between star and phenix in central au+au collisions at @xmath13 gev is still under investigation , it is important to have dilepton measurements at rhic at lower beam energies with the same large acceptance as for the 200 gev data . since the total baryon density does not change significantly from @xmath17 17.3 gev to @xmath17 200 gev @xcite , it is essential to confirm that the broadened @xmath2 spectral function , which describes the results at 17.3 gev and the 200 gev star data , is consistent with the 19.6 gev results . in the intermediate mass region , @xmath18 gev/@xmath4 ( imr ) , dilepton production
is expected to be directly related to thermal radiation of the partonic phase , which is considered to be the prime signature of deconfinement @xcite .
an enhanced yield in this region was first observed by helios/3 @xcite and na38/na50 @xcite .
more recently , the na60 collaboration reported an enhancement in the imr which can not be connected to decays of _ d _ mesons , but may be the result of thermal radiation @xcite .
however , it is experimentally challenging to extract the signal in the presence of significant background sources from open heavy - flavor semi - leptonic decays , such as @xmath19 or @xmath20 . in this letter , we report the first dielectron measurements at mid - rapidity in minimum - bias au+au collisions at @xmath1 = 19.6 gev with the star detector @xcite .
furthermore , we present the first acceptance - corrected dielectron excess mass spectra in au+au collisions at @xmath17 19.6 and 200 gev which are compared with measurements from na60 and theoretical model calculations .
the invariant excess dielectron spectra at different centralities and energies allow for a first systematic study of the lifetime of the hot , dense medium using electromagnetic probes at rhic .
it was pointed out that the excess dielectron yield at low mass is proportional to the total lifetime of the hot , dense medium at @xmath1 = 6 - 200 gev @xcite .
in this analysis , 33 million minimum - bias ( mb ) au+au ( 0 - 80% ) events at @xmath17 19.6 gev , recorded by the star experiment in the year 2011 , were used .
the results at @xmath17 200 gev are derived from the same data analysis reported in ref .
@xcite . the mb trigger at @xmath17 19.6 gev was defined as a coincidence of the two beam beam counters covering the pseudorapidity range @xmath21 @xcite .
charged tracks were reconstructed by the time projection chamber ( tpc ) @xcite , which has full azimuthal coverage at @xmath22 . the absolute distance between collision vertices and the tpc center along the beam direction
was required to be less than 70 cm .
the transverse momentum resolution is measured to be @xmath23 gev/@xmath24 $ ] for @xmath25 5 gev/@xmath26 .
the time - of - flight ( tof ) @xcite detector , which covers the pseudorapidity range @xmath27 , provides the arrival time of charged tracks from the collision vertex .
slow hadrons can be rejected by a velocity cut @xmath28 in the range of @xmath29 gev/@xmath30 , where @xmath31 is the measured velocity and @xmath32 is the expected velocity calculated using the track length and momentum with the assumption of the electron mass .
after the velocity cut , electron identification is achieved by cutting on the normalized ionization energy loss ( @xmath33 ) measured by the tpc , where @xmath34 is the energy loss , @xmath35 is the expected @xmath34 for an electron and @xmath36 is the @xmath34 resolution of an electron , which is better than 8% @xcite .
the @xmath37 cut is momentum dependent and results in a high electron purity of @xmath38 93% and an efficiency of @xmath38 65% on average @xcite . ) .
all panels are presented as a function of dielectron invariant mass in au+au collisions at @xmath1 = 19.6 gev.,scaledwidth=48.0% ] the electron and positron candidates are paired by opposite and same sign charges , called unlike - sign and like - sign pairs , respectively .
the like - sign pairs are used to statistically reproduce the combinatorial and correlated pair backgrounds .
the combinatorial background comes from two random tracks without correlation .
the correlated background is the result of two electrons , each of which comes from a different but correlated process of a particle decay or a jet fragmentation .
for example , consider a @xmath39 dalitz decay where the gamma may convert on some material to form an additional @xmath40 pair .
the @xmath41 from the @xmath42 paired with a @xmath43 from the @xmath44 can produce a correlated background pair .
this correlated background can be reproduced by like - sign pairs .
the unlike - sign and like - sign pairs have different acceptances due to dead areas of the detector and the different bending curvatures of positively and negatively charged particles in the magnetic field .
the dead area fraction is 13% along the azimuthal distribution at @xmath45 .
a mixed - event technique @xcite is applied to estimate the acceptance differences between the unlike - sign and like - sign distributions .
figure [ fig : eemass ] ( a ) shows the ratio between mixed - event unlike - sign pairs and mixed - event like - sign pairs as a function of dielectron mass . a zoom - in version is shown in fig .
[ fig : eemass ] ( b ) .
the background subtraction is based on the measured like - sign spectra with the assumption that the shape and magnitude of the correlated background are the same in the unlike- and like - sign spectra .
we subtract the like - sign background ( corrected for the acceptance difference using the mixed event technique mentioned above ) from the unlike - sign distributions to obtain the raw dielectron signals .
the mixed - event background is not used for background subtraction , since the correlated background contribution is difficult to address with limited statistics at @xmath46 gev/@xmath4 for @xmath1 = 19.6 gev .
figure [ fig : eemass ] ( c ) shows the invariant mass distributions of unlike - sign pairs , like - sign pairs and background - subtracted signals .
the signal to background ratio is shown in fig .
[ fig : eemass ] ( d ) .
dielectron pairs from photon conversions in the detector materials are suppressed by selecting tracks with a distance of closest approach to the collision vertex that is less than 1 cm , and a minimum opening angle cut between the two electron candidates @xcite .
the minimum opening angle is 0.84 rad at @xmath47 gev/@xmath6 and decreases as a function of @xmath48 according to a function form of @xmath49 $ ] , in which a , b , and c are input parameters . for @xmath50 gev/@xmath6 ,
the minimum opening angle is zero .
spectra of pions , kaons and protons in pb+pb at @xmath1 = 17.3 gev @xcite .
the data points o f @xmath51 completely overlap with that of @xmath52 on the figure .
other meson @xmath53 spectra are predicted by the tbw function . for @xmath54 ,
the @xmath16 shape is determined by an independent tbw function fit to the @xmath54 @xmath16 spectra measured by na50 @xcite . more details are in the text . ]
gev . ] , @xmath55 gev/@xmath26 , @xmath56 ) after efficiency correction , compared with the hadronic cocktail consisting of the decays of light hadrons and correlated decays of charm in au+au collisions at @xmath57 gev . the data to cocktail ratio
is shown in the bottom panel .
theoretical calculations @xcite of a broadened @xmath2 spectral function are shown up to 1.5 gev/@xmath4 for comparison .
systematic uncertainties for the data points are shown as green boxes , and the grey band represents the uncertainties for the cocktail simulation . ]
the raw dielectron signal is corrected for the electron reconstruction efficiency .
the single electron reconstruction efficiency includes tpc tracking , electron identification and tof matching efficiencies .
the tpc tracking efficiency is determined by embedding monte carlo ( mc ) tracks into real raw data events , processing the track reconstruction with a geant model of the star detector @xcite , and determining the fraction of those embedded mc tracks which are reconstructed as good tracks .
the efficiency correction includes the effect of dead areas in the detector .
the tof matching and electron identification efficiencies are reproduced from real data .
detailed procedures to obtain the tpc and tof efficiencies are explained in ref .
the energy loss and bremsstrahlung radiation effects for electrons are reproduced by the geant simulation .
the single electron efficiency is convoluted into the pair efficiency with the decay kinematics in the simulation .
the hadronic sources of dielectron pairs include : dalitz decays @xmath39 , @xmath58 and @xmath59 ; vector meson decays : @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 and @xmath65 ; heavy - flavor hadron semi - leptonic decays : @xmath66 ; drell - yan .
the @xmath2 meson contribution is not evaluated in the simulation , but included in the model calculation ( as described in sec .
the @xmath67 process is not included as it has negligible contribution to the cocktail in au+au collisions at @xmath1 = 19.6 gev .
the input hadron spectra to the cocktail are derived from a tsallis blast wave ( tbw ) function fit @xcite to the na49 @xmath16 spectra of pions , kaons and protons in pb+pb at @xmath1 = 17.3 gev @xcite , as shown in fig .
[ fig : tsallisfit ] . other meson @xmath53 spectra
are predicted by the tbw function using the same freeze - out parameters from @xmath53 fit of pions , kaons and protons . the extra uncertainty caused by the input @xmath53 spectra is found to be less than 10% and has been propagated to the final cocktail uncertainty .
for @xmath54 , the @xmath16 shape is determined by an independent tbw function fit to the @xmath54 @xmath16 spectra measured by na50 @xcite .
the @xmath42 contribution is obtained by matching the dielectron mass distribution from simulated @xmath39 and @xmath68 decays to the efficiency - corrected dielectron mass spectrum for @xmath69 gev/@xmath6 .
we also match the @xmath70 distribution from simulation to the measured dielectron production in the corresponding mass region .
the meson yields of other mesons are derived by the meson - to - pion ratios @xcite and the pion yields .
table [ tab : yield ] lists the integrated yields used in the simulation at mid - rapidity for au+au collisions at @xmath57 gev .
the branching ratios of mesons to dielectrons and their uncertainties are from ref .
@xcite .
.the meson yields , @xmath71 , at mid - rapidity used in the hadronic cocktail for 0 - 80% au+au collisions at @xmath1 = 19.6 gev .
the uncertainty includes contributions from the tbw fit and the meson - to - pion ratio .
[ tab : yield ] [ cols="^,^,^",options="header " , ] , in au+au collisions at @xmath57 ( solid circles ) and 200 gev ( diamonds ) .
the @xmath72 values in au+au collisions at @xmath73 19.6 and 200 gev are from refs .
@xcite and @xcite , respectively .
comparison to the na60 data @xcite for in+in collisions at @xmath17 17.3 gev ( open circles ) is also shown .
bars are statistical uncertainties , and systematic uncertainties are shown as grey boxes . a model calculation ( solid curve )
@xcite with a broadened @xmath2 spectral function in hadron gas ( hg ) and qgp thermal radiation is compared with the excess in au+au collisions at @xmath1 = 19.6 gev .
the normalization uncertainty from the star measured @xmath71 is about 10% , which is not shown in the figure .
, scaledwidth=50.0% ] to quantify the yield , the known hadronic cocktail , @xmath74 and drell - yan contributions were subtracted from the dielectron mass spectrum at @xmath57 gev . at @xmath75 gev ,
the known hadronic sources , @xmath66 , @xmath67 , and drell - yan contributions were subtracted .
the excess dielectron mass spectra , corrected for detector acceptance , are shown in fig .
[ fig : excesscomp ] for au+au mb collisions at @xmath57 and 200 gev .
the spectra are normalized to mid - rapidity @xmath76 in absolute terms to cancel out the volume effect , and compared with the excess dimuon yields from the na60 measurements in in+in collisions at @xmath77 gev .
the model calculation @xcite including a broadened @xmath2 spectral function and qgp thermal radiation is consistent with the acceptance - corrected excess in au+au collisions at @xmath17 19.6 gev .
the excess at @xmath17 200 gev is higher than that at @xmath17 17.3 gev in the lmr and imr , but within 2@xmath78 uncertainty .
further measurements with better precision are needed to obtain the average temperature of the hot , dense medium created .
gev/@xmath4 as a function of @xmath76 .
the solid circle and diamond represent the results in 0 - 80% au+au collisions at @xmath17 19.6 and 200 gev , respectively .
the squares are the results for 40 - 80% , 10 - 40% , and 0 - 10% au+au at @xmath17 200 gev .
the open circle represents the dimuon result from the na60 measurement with @xmath79 .
bars are statistical uncertainties , and systematic uncertainties are shown as grey boxes . the theoretical lifetimes for @xmath17 200 gev au+au as a function of @xmath76 in the model calculations @xcite
are shown as a dashed curve .
the lifetimes for @xmath17 17.3 gev in+in and @xmath17 19.6 gev au+au in the same model calculations @xcite are shown as the two horizontal bars .
the @xmath76 values for the horizontal bars are shifted for clarity .
, scaledwidth=50.0% ] figure [ fig : excesscomp ] shows that the excess dielectron yield in the lmr at @xmath17 19.6 gev has a magnitude similar to the excess dimuon yield at @xmath17 17.3 gev .
to quantitatively compare the excess in the lmr , the integrated excess yields of dielectrons in the mass region @xmath80 gev/@xmath4 are shown in fig .
[ fig : excesslmrcomp ] for 0 - 80% au+au collisions at @xmath17 19.6 and 200 gev . the results in finer centralities 0 - 10% , 10 - 40% , and 40 - 80%
are also shown for @xmath17 200 gev collisions .
the excess yield has a centrality dependence and increases from peripheral to central collisions at @xmath17 200 gev .
comparing to the results from in+in collisions at @xmath17 17.3 gev , the excess yield at @xmath17 19.6 gev is consistent within the uncertainties while the excess at @xmath17 200 gev is higher in central collisions , but within 2@xmath78 uncertainty .
this might indicate that the lifetime of the medium created in central collisions at @xmath17 200 gev is longer than those in peripheral collisions and at @xmath17 17.3 gev , which enhances the dilepton production from thermal radiation .
the same model calculations @xcite that consistently describe the dilepton excesses in the @xmath17 17.3 , 19.6 , and 200 gev a+a data give lifetimes of @xmath81 fm/_c _ , @xmath82 fm/_c _ , and @xmath83 fm/_c _ for the 17.3 gev in+in , 19.6 gev au+au , and 200 gev au+au data as shown in fig .
[ fig : excesslmrcomp ] @xcite .
in addition , the lifetime has a strong centrality dependence in @xmath17 200 gev au+au collisions in the calculations , as indicated by the dashed curve in fig .
[ fig : excesslmrcomp ] . with the total baryon density nearly a constant and the dilepton emission rate dominant in the critical temperature region at @xmath17 17.3 - 200 gev ,
the normalized excess dilepton yields in the low mass region from the measurements are proportional to the calculated lifetimes of the medium @xcite .
we note that the lifetime might be model dependent .
it is important to have the calculated lifetimes from other models to verify this proportionality .
in summary , the dielectron mass spectrum is measured in au+au collisions at @xmath17 19.6 gev by the star experiment at rhic .
compared with known hadronic sources , a significant excess is observed , which can be consistently described in all beam energies by a model calculation in which a broadened @xmath2 spectral function scenario at low temperature and chiral symmetry restoration are included .
furthermore , the excess dielectron mass spectra , corrected for the star detector acceptance , are reported for the first time in au+au collisions at @xmath17 19.6 and 200 gev . in the lmr ,
the excess yield at @xmath17 19.6 gev , normalized to the charged particle multiplicity @xmath76 , is comparable to that in in+in collisions at @xmath17 17.3 gev .
for @xmath17 200 gev , the normalized excess yield is higher in central collisions than that at @xmath17 17.3 gev and increases from peripheral to central collisions .
these measurements indicate that the hot , dense medium created in central au+au collisions at top rhic energy has a longer lifetime than those in peripheral collisions and at @xmath17 17.3 gev .
we thank the rhic operations group and rcf at bnl , the nersc center at lbnl , the kisti center in korea , and the open science grid consortium for providing resources and support .
this work was supported in part by the offices of np and hep within the u.s .
doe office of science , the u.s .
nsf , cnrs / in2p3 , fapesp cnpq of brazil , the ministry of education and science of the russian federation , nnsfc , cas , most and moe of china , the korean research foundation , ga and msmt of the czech republic , fias of germany , dae , dst , and csir of india , the national science centre of poland , national research foundation ( nrf-2012004024 ) , the ministry of science , education and sports of the republic of croatia , and rosatom of russia .
d. adamova _ et al .
_ , ; g. agakichiev _
et al . _ , ; d. adamova _ et al . _ , .
r. arnaldi _ et al .
_ , ; r. arnaldi _
et al . _ , ; r. arnaldi _
et al . _ , .
et al . _ , .
l. adamczyk _
et al . _ , .
erratum - ibid .
* 113 * , 049903 ( 2014 ) .
r. rapp , phys .
c * 63 * , 054907 ( 2001 ) ; h. van hees and r. rapp , phys . rev .
lett . * 97 * , 102301 ( 2006 ) .
h. van hees and r. rapp , nucl .
a * 806 * , 339 ( 2008 ) ; r. rapp , adv . high energy phys .
* 2013 * ( 2013 ) 148253 .
o. linnyk _
et al . _ , ; o. linnyk _ et al . _ , .
et al . _ , .
star beam energy scan ii white paper : https://drupal.star.bnl.gov/star/starnotes/public/sn0598 .
a. l. s. angelis _ et al . _ , .
m. c. abreu _ et al .
_ , ; m. c. abreu _ et al .
. k. h. ackermann _
instr . meth .
a * 499 * , 624 ( 2003 ) .
r. rapp and h. van hees , arxiv : 1411.4612 .
j. adams _ et al .
lett . * 92 * , 171801 ( 2004 ) .
m. anderson _
et al . _ ,
instr . meth .
a * 499 * , 659 ( 2003 ) .
b. bonner _
a * 508 * , 181 ( 2003 ) ; m. shao _
et al . _ ,
instr . meth .
a * 492 * , 344 ( 2002 ) ; j. wu _ et al .
a * 538 * , 243 ( 2005 ) .
h. bichsel , .
l. adamczyk _
et al . _ , .
geant 3.21 , cern program library http://wwwasdoc.web.cern.ch/wwwasdoc/geant_html3/geantall.html
. z. tang _
et al . _ ,
c * 79 * , 051901 ( 2009 ) .
et al . _ , j. phys .
g * 37 * , 085104 ( 2010 ) . c. alt _ et al .
et al . _ , .
j. beringer _
et al . _ ,
d * 86 * , 010001 ( 2012 ) .
t. sjstrand _
et al . _ , .
r. rapp , private communications .
g. a. alves _ et al .
. lett . * 77 * , 2388 ( 1996 ) .
erratum - ibid * 81 * , 1537 ( 1998 ) ; s. p.
k. tavernier , rep .
. phys . * 50 * , 1439 ( 1987 ) , and references therein ; l. adamczyk _
d * 86 * , 072013 ( 2012 ) ; a. adare _
* 97 * , 252002 ( 2006 ) .
r. nelson _
et al . _ , .
s. damjanovic _
et al . _ ,
a * 783 * , 327 ( 2007 ) .
l. adamczyk _
et al . _ ,
arxiv:1504.01317 , submitted to phys .
rev . c. p.m. hohler and r. rapp , .
et al . _ , .
b. i. abelev _
et al . _ , . h. specht _
et al . _ ,
proc . * 1322 * , 1 ( 2010 ) . | the acceptance - corrected dielectron excess mass spectra , where the known hadronic sources have been subtracted from the inclusive dielectron mass spectra , are reported for the first time at mid - rapidity @xmath0 in minimum - bias au+au collisions at @xmath1 = 19.6 and 200 gev .
the excess mass spectra are consistently described by a model calculation with a broadened @xmath2 spectral function for @xmath3 gev/@xmath4 .
the integrated dielectron excess yield at @xmath1 = 19.6 gev for @xmath5 gev/@xmath6 , normalized to the charged particle multiplicity at mid - rapidity , has a value similar to that in in+in collisions at @xmath1 = 17.3 gev . for @xmath1 = 200 gev ,
the normalized excess yield in central collisions is higher than that at @xmath1 = 17.3 gev and increases from peripheral to central collisions .
these measurements indicate that the lifetime of the hot , dense medium created in central au+au collisions at @xmath1 = 200 gev is longer than those in peripheral collisions and at lower energies .
# 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) |
supermassive black holes ( smbhs ) found in the centers of all massive galaxies grow by accreting matter , and are dubbed active galactic nuclei ( agn ) during this phase .
x - rays are the most effective way to find moderate to high - luminosity agn because they emit a significant fraction of their bolometric luminosity at these energies compared with inactive galaxies and nearly all x - ray point sources with l@xmath7 erg / s are agn ( e.g. * ? ? ?
( this dividing line misses low luminosity accreting black holes and can potentially include a few very energetic starburst galaxies that might attain these x - ray luminosities . ) optical studies of the quasar luminosity function ( qlfs ) from large areas covered by the sloan digital sky survey ( sdss ) have indicated that black hole growth evolves as a function of luminosity ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , although these unobscured agn likely represent only @xmath830% of the total black hole growth ( e.g. , * ? ? ?
the deepest x - ray surveys to date have been successful in elucidating the low to moderate luminosity agn population .
the _ chandra _
deep fields north and south ( cdf - n and cdf - s ) have peered to depths of 2 ms and 4 ms , reaching to 0.5 - 2 kev ( 2.0 - 8.0 kev ) flux limits of 2.5@xmath9 ( 1.4@xmath10 ) erg s@xmath3 @xmath4 and 9.1@xmath11 ( 5.5@xmath9 ) erg s@xmath3 @xmath4 respectively , in cdf - n and cdf - s @xcite .
the extended _ chandra _ deep field south ( ecdf - s ) surrounded the original cdf - s survey , covering @xmath80.3 deg@xmath0 with sensitivity limits of @xmath12 and @xmath13 erg @xmath4 s@xmath3 in the 0.5 - 2 kev and 2.0 - 8.0 kev bands respectively @xcite .
these observations are integral components of the larger multi - wavelength surveys of great observatories origins deep survey ( goods ; * ? ? ?
* ; * ? ? ?
* ) and multi - wavelength survey by yale - chile ( musyc ; * ? ? ?
_ xmm - newton _ and _ chandra _ have done deep surveys of the cosmos field , reaching 0.5 - 2 kev fluxes down to 7@xmath10 erg @xmath4 s@xmath3 on the whole field @xcite and 1.9@xmath10 erg @xmath4 s@xmath3 in the inner area @xcite , respectively . in the next two years ,
_ chandra _ will survey the whole cosmos field to the depth now available only in the central area ( cosmos legacy , pi : f. civano ) .
however , since these surveys cover relatively small areas , high luminosity and/or high redshift agn remain under - sampled compared with the population that would be detected in much wider areas . as half of the black hole growth occurs in high luminosity agn @xcite , the existing census of black hole growth over cosmic time is incomplete .
indeed , x - ray qlfs at high redshift ( @xmath14 ) are only constrained at lower luminosities , necessitating optical studies to fill in the high luminosity parameter space to allow the qlf to be fit over several decades of luminosity ( e.g. , * ? ? ?
. such a method assumes optically selected quasars represent the same population as x - ray selected agn , which may be incorrect .
indeed , this assumption is incorrect at lower luminosities ( e.g. , * ? ? ?
larger volume x - ray surveys are necessary to sample a representative population of growing black holes since high redshift and high luminosity sources are relatively rare .
so far , only xbotes ( 9 deg@xmath0 , * ? ? ?
* ; * ? ? ?
* ) , champ , a serendipitous survey based on archival _ chandra _ data ( 10 deg@xmath0 , * ? ? ?
* ; * ? ? ?
* ) , extended champ / sdss ( @xmath832 deg@xmath0 , * ? ? ?
* ) and the _ xmm - newton _ large scale structure ( xmm - lss , @xmath811 deg@xmath15 , with a full survey coverage of 50 deg@xmath0 planned , * ? ? ? * ; * ? ? ?
* ) surveys come close to exploring this part of parameter space .
ancillary multi - wavelength data are then needed to provide additional critical information about these sources , such as optical identifications , with redshifts , to accurately calculate agn and host galaxy properties like accretion luminosity , galaxy morphology , stellar content , star formation activity , galaxy mass , and so on .
follow - up spectroscopy over such large areas have been obtained only very slowly .
about half of the x - ray sources in xbotes have mmt hectospec spectra as part of the agn and galaxy evolution survey ( ages , * ? ? ?
* ) , but this campaign took over 4 years . the optical follow - up for champ also required a similar time investment @xcite . though @xcite produced the log@xmath5-log@xmath6 distribution for the full _ xmm - newton _ serendipitous source catalog , covering @xmath8130 deg@xmath0 , multi - wavelength follow - up observations over this area to identify these x - ray sources and to produce an accurate quasar luminosity function is prohibitively time - consuming . to efficiently probe the high luminosity and high redshift agn population ,
we have begun a large area x - ray survey in sdss stripe 82 , a @xmath8300 deg@xmath0 area in a region of space that harbors substantial observational investments from an assortment of multiwavelength facilities , including first and evla ( radio , * ? ? ?
* ; * ? ? ?
* ) ; atacama cosmology telescope ( millimeter , * ? ? ?
* ) ; ukidss and vista hemisphere survey ( near - infrared , * ? ? ?
* ; * ? ? ?
* ) and galex ( ultraviolet , * ? ? ?
stripe 82 extends from -60@xmath16ra@xmath1760@xmath18 and -1.25@[email protected] , and is thus accessible from both hemispheres .
an advantage of stripe 82 over the full sdss catalog , and by extension the champ / sdss survey , is the increased depth in the optical : the imaging in stripe 82 reaches two magnitudes deeper than any individual sdss scan since each individual area was imaged @xmath880 times .
the optical spectral density of stripe 82 is also exceptionally high thanks to observations with sdss , 2df and wigglez @xcite , and continues to increase with spectroscopic campaigns from sdss - iii boss @xcite and aus ( croom et al .
, in prep . )
: at present , @xmath20400 optical objects per square degree have spectra . combining archival _ xmm - newton _ and _ chandra _ observations as well as newly obtained _ xmm - newton _ observations ( which we will report in a future publication due to the complicated data reduction procedures required for these _ xmm - newton _ data ) , the total x - ray coverage of stripe 82 ( stripe 82 x ) reaches @xmath817 deg@xmath0 so far . taking advantage of the high level of optical spectroscopic completeness
, this survey will reveal more rare high luminosity ( l @xmath21 erg s@xmath3 ) agn at all redshifts and in the distant universe ( @xmath22 ) than previous x - ray surveys . in this pilot study , we present x - ray source counts from a new large - area archival chandra survey in stripe 82 ( hereafter , `` stripe 82 acx '' ) , covering a largely non - contiguous area of @xmath86.2 deg@xmath0 , and make comparisons for the first time to the other three largest chandra surveys .
we also describe the general characteristics of agn found in stripe 82 acx . though the full 17 deg@xmath0 data will be required to generate the high end of the x - ray quasar luminosity function , the present number counts - and in particular , the area - flux curve - are essential for deriving the qlf .
we demonstrate that these number counts do indeed sample the high luminosity agn population under - represented in smaller area surveys .
furthermore , our results suggest that current population synthesis models are incorrect at high luminosity and/or high redshift . throughout the paper ,
we use cosmology h@xmath23 = 70 km / s / mpc , @xmath24 = 0.27 and @xmath25=0.73 .
we utilized the _
chandra _ source catalog @xcite to identify x - ray sources and estimate the survey sensitivity of stripe 82 acx .
chandra _ source catalog currently contains point and compact ( @xmath26 ) x - ray sources from the first eight years of public _
imaging observations , with associated data products , such as events files , exposure maps , etc . , reduced homogeneously .
full details of the data reprocessing are provided in @xcite . in brief , all observations for the _ chandra _ source catalog were recalibrated with a calibration database ( caldb ) version created specifically for the _ chandra _ source catalog .
acis observations had the time - dependent gain calibration applied and were corrected for cti ( ccd charge transfer inefficiency ) .
the ciao task _ acis_run_hotpix _ was run to remove pixel afterglow events , and bad pixels and hot pixels were flagged for removal .
the background was screened to reduce the non x - ray background , removing intervals with strong background flaring . for a source to be included in the _ chandra _ source catalog ,
a conservative threshold was chosen to mitigate the inclusion of spurious detections as real sources .
source detection was performed by the ciao task _ wavdetect _ , with a limiting significance level of 2.5@xmath27 , corresponding to @xmath81 false detection per pointing .
an additional cut on flux significance was then applied .
chandra _ source catalog estimates this significance by @xmath28 , where @xmath29 and the fwhm is of the posterior probability density ( @xmath30 ) for the flux ( @xmath6 ) assuming that the total counts ( @xmath31 ) and background counts ( @xmath32 ) have a poisson distribution .
the flux had to be significant at @xmath333-@xmath1 level in at least one flux band , generally corresponding to @xmath810 source photons on - axis , increasing to @xmath820 - 30 source counts off - axis since the _ chandra _ psf degrades with off - axis distance .
this threshold is more conservative than those used in previous _
surveys ( see * ? ? ?
* ; * ? ? ?
* for a comparison ) and removes @xmath81/3 of the sources detected by _
wavdetect_. though this high signal - to - noise cut makes stripe 82 acx incomplete at low fluxes , this does not adversely affect the science goals of this project , which are to probe the high x - ray luminosity agn population and to accurately constrain their number density and evolution .
surveys such as cdf - s @xcite are optimized to uncover x - ray sources at the faintest levels , whereas the wide survey area of stripe 82 x complements this parameter space by better exploring the high x - ray flux regime .
figure [ pointings ] shows the positions of the archival _ chandra _ observations in stripe 82 .
we note that fields targeting galaxy clusters in stripe 82 were _ a priori _ removed from our analysis .
we note that including galaxy cluster fields would make no noticeable differences on our derived log@xmath5-log@xmath6 relationships .
the two dense pointing regions at ra = @xmath8352@xmath18 and ra = @xmath837@xmath18 are observations from the xdeep2 survey , fields 3 and 4 , respectively @xcite .
we estimated the stripe 82 acx survey sensitivity at the 4.5@xmath1 flux significance threshold , corresponding to @xmath81 spurious source per 100 deg@xmath151100 unique sources were detected at all significance levels by the _
chandra _ source catalog in the 6.2 deg@xmath0 covered here and then extrapolating to a 100 deg@xmath0 area .
] , by creating sensitivity maps for each observation in the soft ( 0.5 - 2 kev ) , hard ( 2 - 7 kev ) and full ( 0.5 - 7 kev ) bands .
these sensitivity maps give the limiting flux at the cited significance threshold for each pixel in the detector .
we note that the _ chandra _ source catalog runs the source detection algorithm and provides products and fluxes in the more narrow `` s '' ( 0.5 - 1.2 kev ) and `` m '' ( 1.2 - 2 kev ) bands . for consistency with previous x - ray surveys ,
however , we explore the _ chandra _ number counts in the traditional `` soft '' band , 0.5 - 2 kev .
the hard and full bands defined here are consistent with those from the _ chandra _ source catalog , i.e. , only the `` soft '' band is redefined .
first , exposure maps were created , where we used the level 3 event files from the _ chandra _ source catalog and downloaded the aspect solution file , mask file and parameter block file from the _ chandra _ archive ; these level 3 event files are similar to the level 2 event files included in the _ chandra _ data archive ( i.e. , filtered files suitable for data analysis ) , but with different good time interval ( gti ) filters applied to remedy background flaring .
we created aspect histograms for each active chip in an observation with ciao tool _ asphist _ , using the aspect solution and event files as input .
we then created an instrument map for each chip , which provides information about the effective area as a function of detector position , using the ciao tool _ mkinstmap_. as the effective area is a function of energy , we provided an input file of spectral weights derived from the ciao routine _
make_instmap_weights _ using a power law model with @xmath34 = 1.4 , to be consistent with previous _ chandra
_ surveys to which we compare our number counts ( i.e. , ecdf - s , _ chandra_-cosmos , xdeep2 and champ ) . _
mkinstmap _ was also given the mask file and parameter block file , which defines the clocking parameters for each pixel ( i.e. , how long the pixel is exposed before read - out ) . using these instrument maps and aspect histograms , an exposure map for each chip
is then created using ciao tool _
mkexpmap_. finally ,
these individual chip by chip exposure maps are combined into one image with ciao routine _ reproject_image _ , producing an exposure map for each observation in each band . to create the sensitivity maps
, we used the background images in the hard and full bands provided by the _
chandra _ source catalog , with the narrow `` s '' and `` m '' bands added to produce a soft band background image . using ciao tool _
mkpsfmap _ , new psf images
were then generated at the spectral weighted mean energies of each band , i.e. , 1.02 kev , 3.79 kev and 1.98 kev for the soft , hard and full bands , respectively , again assuming a powerlaw with @xmath34=1.4 .
finally , sensitivity maps for each obsid were created from these exposure maps and psf images , with ciao tool _
lim_sens _ , using a 4.5@xmath1 flux threshold .
these sensitivity maps were then converted from photon flux units to energy units using an absorbed power law model with @xmath34 = 1.4 and n@xmath35 = 3@xmath36 @xmath4 , the approximate absorption through our galaxy in the direction of stripe 82 ( see table [ info ] for conversion factors ) .
we effectively mask out the chip edges where source detection becomes improbable by `` turning off '' pixels where the exposure map drops below 15% of the maximum value .
we note that no sources fall within these masked out pixels . to gauge full sensitivity across the survey ,
the individual sensitivity maps were overlaid on a grid of pixels spanning the stripe 82 area . in overlapping regions , the most sensitive pixel ( i.e. , the lowest limiting flux ) was chosen . from this stripe 82 region sensitivity map , we then calculated the cumulative histogram of survey area as a function of limiting flux , producing the area - flux curves in figure [ area - flux ] .
stripe 82 acx covers @xmath86.2 deg@xmath0 of non - overlapping area , with cluster fields removed , which would add an additional 1.3 deg@xmath0 of coverage .
as noted in table [ info ] , the approximate flux limits of this survey are 1.2@xmath2 , 5.4@xmath2 and 2.9@xmath2 erg @xmath4 s@xmath3 , with half the survey area visible to approximate depths of 1.5@xmath37 , 4.8@xmath37 and 2.9@xmath37
erg @xmath4 s@xmath3 in the soft , hard and full bands , respectively . from the source observation table in the _ chandra _ source catalog
, we isolated point sources lying in stripe 82 that are not saturated , on the acis read - out streak , or suffering from pile - up .
, @xmath38 , @xmath39 , and where the pileup fraction was less than 10% .
a source is considered extended ( i.e. , @xmath40 is set to 1 ) if the intrinsic source extent , found by deconvolving the local psf from the observed source extent ( parameterized as a rotated elliptical gaussian ) , is inconsistent with a point source at the 90% confidence level .
only one source suffered from significant pileup ( i.e. , pileup fraction 25% ) and was thus discarded from our source list . ] for the hard and full bands , the aperture corrected flux in photon units was converted to energy units using the conversion factors listed in table [ info ] . to derive the soft flux
, we had to ensure the `` s '' and `` m '' fluxes were combined with appropriate weights to match that of the sensitivity map .
the aperture corrected net counts in the `` s '' ( @xmath41 ) and `` m '' ( @xmath42 ) bands from the _ chandra _ source catalog were added , giving soft band counts ( @xmath43 ) : @xmath44 these counts were then divided by the pixel from the soft band exposure map corresponding to the location where the source was detected ( @xmath45 ) , where the exposure map was created as described in section [ sens ] : @xmath46 again , the soft flux in photon units were converted to energy units via the conversion factor listed in table [ info ] . in practice , a source could be detected with _ wavdetect _
run in the full soft band , but not in the more narrow `` s '' and `` m '' bands .
however , the _ chandra _ source catalog includes a flux measurement for each band as long as the source was detected at the 3@xmath1 level in any given band and we therefore have enough information to calculate the soft band flux and then test its significance . as we further impose a 4.5@xmath1 threshold for inclusion of a source in the stripe 82 acx catalog
, we expect that missed objects , sources that would have been detected at the 4.5@xmath1 level in the 0.5 - 2 kev band but not at the 3@xmath1 level in any individual band , are rare . to determine whether a detection is significant at the 4.5@xmath1 level , and for consistency with the area - flux curves from which we generate the log@xmath5 - log@xmath6 relationships , we compared the x - ray source list with the sensitivity maps : the source flux had to meet or exceed the limiting flux from the sensitivity map at the pixel where the source was identified in order for us to consider the source as significantly detected in that specific energy band .
there were 112 sources detected in multiple observations ( identified by the _
flag in the master source observation table ) . for these objects we chose the flux corresponding to the most sensitive observation . to avoid skewing our statistics and introducing a bias into our catalog
, we also removed targeted sources from our catalog , identified as the source within 5@xmath47 of the user supplied ra and dec for the target of the observation ; we note that only 19 pointings from the 73 observations in this analysis had sources that met this criterion and were subsequently removed .
as mentioned previously , 29 pointings are from the xdeep2 survey , and thus do not have targeted sources . for the remaining 25 observations ,
either the targeted object was not detected at the 4.5@xmath1 level or the observer supplied x - ray coordinates of the target were not well determined .
though we _ a priori _ removed observations that targeted galaxy clusters and groups , the fields that targeted specific objects could potentially be biased by clustering . to test
if such an effect factors into our sample , we compared the spectroscopic redshifts for sources detected serendipitously in each field with the redshift of the targeted source ( see section [ opt_info ] for details on matching x - ray sources with optical counterparts and finding associated spectroscopic redshifts ) . in 14 of the 19 fields where the targeted source was detected at a significant level and thus removed ,
the spectroscopic redshifts of the target and more than one x - ray field source were available .
we find that in all cases , these serendipitous sources are not associated with the target : the dispersion between the target redshift and those in the field range from 0.32 to 3.13 .
bias due to source clustering around targeted objects therefore does not affect this survey . in total
, we detect 709 unique x - ray sources in stripe 82 acx , with 705 detected in the full band , 480 in the soft band and 137 in the hard band at the 4.5@xmath1 level .
of these , 4 objects were detected in only the soft band and 214 were detected solely in the full band ; none were detected in just the hard band .
we present the number density of sources as a function of flux , i.e. , the log@xmath5 - log@xmath6 relation .
the binned differential number counts are given by : @xmath48 where n is the number of sources in bin @xmath49 , @xmath50 is the limiting sky coverage associated with the @xmath51th source and @xmath52 is the width of the flux bin . here , we follow the prescription of @xcite to calculate @xmath53 , a weighted flux to represent the bin centroid : @xmath54 the error is given by poissonian statistics : @xmath55 we have binned by 20 , 10 and 30 sources in the soft , hard and full bands , respectively , so that each bin , except for the highest flux bin , has an equal number of sources rather than equal flux widths . in integral form
, the cumulative source distribution is represented by : @xmath56 where
n(@xmath20s ) is the number of sources with a flux greater than @xmath6 and @xmath57 is defined as above .
the associated error is the variance : @xmath58 we have imposed a lower sky coverage limit of 0.02 deg@xmath0 , restricting the log@xmath5 - log@xmath6 analysis to sources whose fluxes exceed the limiting flux at this sky coverage .
the normalized representations of the stripe 82 acx log@xmath5 - log@xmath6 relationships ( i.e. , dn / ds @xmath59 s@xmath60 and n(@xmath20s ) @xmath59 s@xmath61 , where s@xmath62 is flux in units of 10@xmath63 erg @xmath4 s@xmath3 ) are shown in figure [ logn - logs ] as black circles , with the binned differential counts on the left and unbinned cumulative integral counts on the right . at the bright flux end ,
the stripe 82 acx number counts are euclidean , as shown by the horizontal shape in this normalized representation .
we compare our number counts with previous x - ray surveys that span the range from deep , small area ( ecdf - s , at 0.3 deg@xmath15 with a soft band flux limit of @xmath81.1@xmath5910@xmath64 erg @xmath4 s@xmath3 , * ? ? ?
* ) , to slightly larger area , though still rather deep ( _ chandra_-cosmos at 0.9 deg@xmath0 with a 0.5 - 2kev flux limit of @xmath81.9@xmath5910@xmath64erg @xmath4 s@xmath3 , * ? ? ?
* ) , to moderate area and moderate depth ( xdeep2 at 3.2 deg@xmath0 with a 0.5 - 2kev flux limit of @xmath65erg @xmath4 s@xmath3 , * ? ? ?
* ) and finally to wide area and moderate depth ( champ , at 9.6 deg@xmath0 and a soft band flux limit of @xmath82.5@xmath5910@xmath64 erg @xmath4 s@xmath3 , xbootes at 9 deg@xmath0 with a 0.5 - 7 kev flux limit of @xmath66 erg @xmath4 s@xmath3 ; * ? ? ?
* ; * ? ? ?
* ) ; only champ ( xbootes ) has differential number counts ( fits ) available for comparison . a comparison with the agn number counts from the population synthesis models from @xcite is also presented for the soft and hard bands .
we note that although the extended champ / sdss study has a greater survey area , it represents an optically selected qso sample incorporating x - ray detections and flux limits , thus no x - ray log@xmath5-log@xmath6 is published in that work .
stripe 82 acx , however , considers all x - ray detected sources , regardless of whether or not an optical counterpart has yet been identified .
@xcite also produces log@xmath5-log@xmath6 for several _ chanda _ surveys simultaneously , including xbootes and ecdf - s , however as we discuss below , in this analysis it is advantageous for us to consider the surveys independently rather than in aggregate .
ecdf - s , c - cosmos and champ adopt the same spectral model ( absorbed power law with @xmath34=1.4 ) to estimate source flux while the xdeep2 survey uses a spectral index of 1.7 . only xbootes ( xdeep2 ) defines the hard ( and full ) energy bands in the same way as stripe 82 acx ( 2 - 7 kev and 0.5 - 7 kev , respectively ) .
we have adjusted the values from the remaining comparison surveys using the assumed spectral model of a power law with @xmath34=1.4 , to shift the hard ( 2 - 10 kev for c - cosmos and the @xcite model predictions , 2 - 8 kev for e - cdfs and champ ) and full ( 0.5 - 10 kev for c - cosmos and 0.5 - 8 kev for champ ) bands to be within our defined bandpasses ( i.e. , we multiplied the comparison fluxes by factors of 0.68 , 0.86 , 0.76 , and 0.91 , respectively ) .
general agreement exists among the comparison surveys presented here , though champ is systematically higher in all bands and xbootes ( xdeep2 ) is somewhat lower in the hard ( and full ) band(s ) . in the soft and full bands , stripe
82 acx agrees with ecdf - s , c - cosmos , xdeep2 within the quoted error bars , and is slightly lower than champ ( though consistent within the error bars ) , which has similar ( or slightly better ) sensitivity at high fluxes .
agreement also exists in the soft band between stripe 82 acx and the @xcite model .
the consistency in the soft band among stripe 82 acx and previous _
surveys and the @xcite population synthesis model suggests that our method of calculating the soft band flux from the individual `` s '' and `` m '' bands is robust : we do not see a great number of `` missing '' soft sources that would have been identified in a 0.5 - 2 kev band detection . we note that stripe 82 acx is incomplete at low fluxes due to the shallow to moderate depth covered by the majority of archival observations , but the deeper surveys like cdf - s @xcite cover this regime better . the characteristic signature of eddington bias ( promotion of spurious sources into the source list due to statistical fluctuations at the detection limit of each observation ) , which manifests as an `` up - turn '' in the number counts at low fluxes , is absent , likely due to the stringent significance detection threshold we have imposed .
this , combined with the consistency between the number counts of stripe 82 acx with other surveys , suggests that eddington bias is negligible in our survey .
as @xcite only publish the soft band xdeep2 log@xmath5-log@xmath6 and @xcite do not compare the xbootes differential number counts with other x - ray surveys , the present work is the first to demonstrate a disagreement among _ chandra _ surveys in the hard band log@xmath5-log@xmath6 .
intriguingly , the stripe 82 acx log@xmath5-log@xmath6 is consistent with xdeep2 and xbootes in the hard band , yet all three are somewhat lower than other _
chandra _ surveys . between 10@xmath63 erg @xmath4 s@xmath3 and @xmath67
erg @xmath4 s@xmath3 , the 1@xmath1 error bars between stripe 82 acx and champ do not overlap , suggesting that the difference in this range is significant . though the cumulative log@xmath5-log@xmath6 data points are not independent , the agreement among stripe 82 acx , xbootes and xdeep2 indicates that the systmatically lower relation we find is not due to several abnormally low data points
. what could be the cause of the discrepant normalizations ?
all three surveys are wide area and are therefore minimally affected by cosmic variance , unlike the smaller area surveys to which we compare our results .
however , champ also covers a wide area and is systematically higher , rather than lower , than other _
chandra _ surveys . to further rule out cosmic variance , and to test whether our results are driven by the inclusion of the xdeep2 fields
, we removed these pointings from the stripe 82 area , to obtain a completely independent measurement of the hard band log@xmath5-log@xmath6 .
the remaining 44 stripe 82 pointings , covering @xmath84.3 deg@xmath0 , are non - contiguous and spread over a wide area , so like champ , these fields provide a more random sampling of the general x - ray population than a contiguous survey .
as shown in figure [ hard_test ] left , removal of the xdeep2 fields does not change the normalization of the hard band number counts .
another possibility for the agreement among xbootes , xdeep2 and stripe 82 acx , and their discrepancy with other _
chandra _ surveys , is the short average exposure time of the former studies : the xbootes pointings are 5 ks in duration , the xdeep2 survey is a series of 10 ks observations and a majority of the stripe 82 fields have exposure times under 10 ks . in figure [ hard_test ] right , we show the hard band log@xmath5-log@xmath6 for the @xmath82 deg@xmath0 of observations with exposure time exceeding 10 ks ( where the survey sensitivity to account for the subset of pointings has been updated accordingly ) .
the normalization at high fluxes is clearly higher , though a siginficant disagreement still exists betweeen 2@xmath37 erg @xmath4 s@xmath3 and 7@xmath37 erg @xmath4 s@xmath3 .
these results indicate that exposure times can play a significant role in determining the source number counts at the high flux end in the hard , but not soft and full bands in _ chandra _ surveys .
however , we note that though stripe 82 acx disagrees significantly over a limited flux range with champ in the hard band , it is consistent within the error bars compared to the other 4 surveys we consider .
the full 17 deg@xmath0 survey data are needed to generate the high luminosity end of the x - ray qlf , but these stripe 82 acx log@xmath5-log@xmath6 relationships are an important first step towards that goal .
an additional area up to 100 deg@xmath0 would be ideal for more precise qlf constraints , as well as the ability to study its evolution . for now , as we show below , the current number counts do preferentially trace the high luminosity agn population and we find many more of these sources in this pilot study than were detected in smaller area x - ray surveys .
here we investigate the agn population found in stripe 82 acx prior to a time - intensive follow - up spectroscopic campaign . by combining the x - ray catalog with the sloan digital sky survey ( sdss )
, we find that over half of the x - ray sources have optical counterparts and half of those have spectroscopic redshifts from existing optical catalogs , enabling identification of these x - ray sources .
below , we comment on this agn population , comparing the parameter space in which these sources live with those found in smaller area surveys that have extensive spectroscopic follow - up .
we also compare the agn we immediately identify with the population expected from theoretical predictions of agn population synthesis models .
spectroscopic redshifts , or well constrained photometric redshifts , are necessary to robustly calculate x - ray luminosities .
chandra _ source catalog provides a matched catalog of x - ray sources with sdss data release 7 @xcite .
details of the matching algorithm are described in @xcite , who use the bayesian probabilistic algorithm presented in @xcite . in brief , for each possible association , a bayes factor is calculated using a positional uncertainty of 0.1@xmath47 for the sdss sources and a varying error for _ chandra _ sources based on the 95% accuracy limit . a uniform prior
is then assumed , which is a function of the number of x - ray and optical sources as well as the number of true pairs .
all sources in the cross - matched catalog are those where the probability of a match exceeds 50% .
four hundred thirty _ chandra _ sources have optical counterparts in sdss , 11 of which have two possible optical counterparts .
we have verified these matches by visual inspection and have removed optical counterparts that are saturated , contaminated by proximity to a bright non - point source , coincident with diffraction spikes , or are faint photometric sources from dr7 that do not have photometric detections in data release 8 ( dr8 ) . for the 11 x - ray objects with two possible optical counterparts , we chose the source with a higher probability of a match ( generally the brighter , closer source ) . this vetting yields 409 matched x - ray and optical sources , with redshifts from the following optical catalogs : sdss dr8 @xcite , sdss - iii @xcite , 2slaq @xcite , wigglez @xcite and the 4th release of the deep2 catalog @xcite .
we find spectroscopic redshifts for 204 x - ray sources ( 186 from sdss dr8 and dr9 , 9 from 2slaq , 3 from wigglez and 6 from xdeep2 ) , of which 197 have l@xmath68 erg s@xmath3 ( 0.5 - 7 kev ) , and are therefore likely agn ( e.g. , * ? ? ?
here l@xmath69 represents the observed full band luminosity .
of the 300 x - ray sources not detected in sdss dr7 , 107 were detected by wise ( lamassa et al .
, 2013 , in prep . ) .
these objects are candidates for elusive obscured high luminosity quasars and will be discussed in a follow - up paper . in figure
[ frac_opt ] , we plot the total number of x - ray sources and the subset with optical identifications and spectroscopic redshifts as a funcion of full band flux
. interestingly , optical counterparts and spectroscopic identifications are found at all flux levels
. the fraction of x - ray sources with optical associations and those with redshifts do increase with full band flux , but a low x - ray flux threshold where optical identifications become improbable is not apparent . consequently , follow - up spectroscopy will target sources at all x - ray fluxes . in figure [ lum_z ] , we compare stripe 82 acx with the musyc survey of cdf - s + e - cdfs @xcite and c - cosmos @xcite to illustrate the value of larger areas in expanding the parameter space explored by pencil beam and moderate area surveys . here we focus only on x - ray - identified agn with spectroscopic redshifts in all surveys .
the spectroscopic completeness between cdf - s + e - cdfs and stripe 82 acx is comparable , @xmath828% and @xmath829% , respectively ( 314 of 1134 x - ray sources have reliable spectroscopic redshifts , see cardamone et al .
2010 for details ) ; c - cosmos has a higher level of spectroscopic completeness ( 48% ) .
we note that the x - ray luminosities for stripe 82 acx and for cdf - s + e - cdfs are observed luminosities , while the c - cosmos luminosities have been k - corrected to the rest frame .
we also include the predicted agn luminosity distribution for stripe 82 acx using the @xcite population synthesis model where the observed full band 0.5 - 7 kev area - flux curve was given as the model input .
it is immediately apparent that the agn identified in stripe 82 acx are preferentially at high luminosities ( i.e. , l@xmath70 3@xmath71 erg s@xmath3 ) at all redshifts , including in the distant universe at @xmath22 .
stripe 82 acx uncovers half again as many systems as c - cosmos beyond log(l@xmath69 ) of 44.5 dex ( 83 vs. 57 ) , although c - cosmos has a higher level of spectroscopic completeness due to a long - term dedicated follow - up campaign .
though over 100 @xmath14 agn have been identified by _
chandra_-cosmos and @xmath14 quasar luminosity functions have been generated using x - ray data from _
chandra_-cosmos @xcite , _ xmm - newton_-cosmos @xcite , and the _ chandra _ deep field south @xcite , all these sources have faint to moderate luminosities .
wide - area surveys like stripe 82 acx are essential to locate rare , l@xmath72 erg s@xmath3 quasars , which is necessary to populate the high end of the x - ray qlf .
the @xcite population synthesis model indicates that 580 agn should be located within the 6.2 deg@xmath0 survey area , 137 of which are at high luminosities ( l@xmath70 3@xmath71 erg s@xmath3 ) and 75 beyond a redshift of 2 .
figure [ lum_z ] indicates a majority of the yet - to - be classified agn are at moderate luminosities .
we have found more than half of the expected high luminosity agn ( 83 out of 137 ) and a third of those at @xmath22 ( 25 of 75 ) . according to the model predictions , 56 of the @xmath22 agn are luminous sources ( l@xmath73 dex ) ; we have immediately identified half ( 23 ) of these .
our observational results further indicate that population synthesis models can be better refined with our survey data .
only 580 agn are predicted in stripe 82 acx though @xmath8700 x - ray sources lie within this region . though stars , normal galaxies and x - ray binaries contribute to the total number of x - ray sources , such non - agn are not numerous enough to explain the discrepancy between the predictions and observations .
additionally , as shown in figure [ lum_z ] , more extremely luminous agn ( i.e. , l@xmath74 erg s@xmath3 ) are found than accounted for in this model . indeed , the paucity of known agn at high x - ray luminosities is the reason stripe 82 x is needed .
we have presented the first 6.2 deg@xmath0 of the current @xmath817 deg@xmath0 x - ray survey covering sdss stripe 82 .
this pilot work concerns the analysis of archival _ chandra _ observations in the region ( stripe 82 acx ) .
we utilized source lists and products generated and provided by the _
chandra _ source catalog @xcite , and the extensive multiwavelength data available in stripe 82 , which greatly streamlined the analysis compared with previous x - ray surveys .
our main results are summarized as follows : * stripe 82 acx reaches approximate flux limits of 1.2@xmath2 , 5.4@xmath2 and 2.9@xmath2 erg s@xmath3 @xmath4 , with half area survey coverage at fluxes of 1.5@xmath37 , 4.8@xmath37 and 2.9@xmath37 erg s@xmath3 @xmath4 , in the soft ( 0.5 - 2 kev ) , hard ( 2 - 7 kev ) and full ( 0.5 - 7 kev ) bands , respectively .
we detect a total of 709 unique x - ray sources , with 480 , 137 and 705 sources at the 4.5@xmath1 level in the soft , hard and full bands .
of these , 214 were detected solely in the full band and 4 sources detected only in the soft band .
* the number counts for stripe 82 acx are euclidean at high fluxes .
our soft and full band log@xmath5-log@xmath6 relations are consistent with the predictions from @xcite and with previous _
chandra _ x - ray surveys : ecdf - s @xcite , c - cosmos @xcite , xdeep2 @xcite , xbootes @xcite and champ @xcite .
we agree with the xdeep2 and xbootes hard band number counts , but all three are somewhat systematically lower than other surveys .
we show that this discrepancy may be due to exposure times , indicating that the length of observations may have a significant impact on the high flux end of the hard band log@xmath5-log@xmath6 from _ chandra _ surveys . *
409 _ chandra _ sources are matched to optical counterparts in the sdss .
we obtained spectroscopic redshifts for 204 of these x - ray objects from sdss dr8 @xcite , sdss - iii @xcite , 2slaq @xcite , wigglez @xcite and deep2 @xcite .
one hundred ninety - seven of these objects have x - ray luminosities consistent with agn ( i.e. l@xmath75 erg s@xmath3 ) ; 83 of these are at high luminosity ( l@xmath76 erg s@xmath3 ) and 25 are at @xmath22 . * we compared the agn luminosity distributions from stripe 82 acx with cdf - s + e - cdfs @xcite and c - cosmos @xcite ; e - cdfs + cdf - s has similar level of spectroscopic completeness with stripe 82 acx while c - cosmos has a somewhat higher level of spectroscopic completeness .
this comparison shows that the agn we immediately identify represent the high luminosity ( l@xmath70 3@xmath71 erg s@xmath3 ) population , at all redshifts , including beyond @xmath22 , while smaller surveys preferentially locate more moderate luminosity agn . * using population synthesis models from @xcite and the full - band area - flux curve in stripe 82 acx , we have demonstrated that more than half of the high luminosity agn expected to lie in the survey area have been identified prior to any follow - up spectroscopic campaign .
we find a third of the high redshift ( @xmath22 ) agn and half of the luminous agn at high redshift that are predicted to lie within stripe 82 acx . slight disagreements between model predictions and our observational results indicate that population synthesis models need to be refined to better accommodate the high luminosity agn population .
this result affirms the importance of a large volume x - ray survey like stripe 82 x , which will provide important constraints for these models .
we thank the referee for insightful critiques and comments that improved this manuscript . this research has made use of data obtained from the chandra source catalog , provided by the chandra x - ray center ( cxc ) as part of the chandra data archive .
we thank frank primini and nina bonaventura for answering many csc related questions .
we thank bret lehmer and andy goulding for thoughtful discussions .
we thank bret lehmer , andy goulding and minsun kim for access to their logn - logs data .
00344 & 40.670 & -0.0132 & 47.4 + 02101 & 7.898 & 0.5724 & 6.7 + 02115 & 358.223 & -0.4809 & 5.8 + 02178 & 9.267 & -1.1523 & 27.5 + 02252 & 5.639 & 0.3486 & 71.2 + 03039 & 334.183 & 0.2301 & 7.4 + 03576 & 43.217 & -1.2749 & 39.7 + 04011 & 40.439 & 0.4431 & 5.0 + 04100 & 29.209 & 0.8857 & 5.6 + 04105 & 58.676 & 0.6173 & 9.8 + 04686 & 40.910 & 1.4036 & 5.7 + 04825 & 12.526 & -0.8886 & 13.0 + 04827 & 359.326 & 0.7306 & 12.7 + 04829 & 3.275 & 0.0755 & 6.7 + 04830 & 13.480 & -0.0526 & 7.1 + 04832 & 18.115 & -1.2061 & 5.9 + 04861 & 0.628 & 0.8331 & 5.7 + 04963 & 19.723 & -1.0019 & 39.3 + 05244 & 38.691 & -0.8400 & 13.3 + 05617 & 1.468 & -0.1155 & 16.9 + 05650 & 55.807 & 1.0603 & 7.9 + 05654 & 50.956 & -0.4971 & 8.3 + 05694 & 12.541 & -0.6502 & 7.9 + 05695 & 49.960 & -0.9807 & 11.3 + 05699 & 359.630 & -0.3740 & 6.3 + 06128 & 359.225 & -1.0006 & 17.9 + 06129 & 359.225 & -1.0006 & 19.3 + 06802 & 20.923 & 0.7433 & 9.8 + 06890 & 331.189 & 0.5283 & 9.4 + 07241 & 45.300 & -0.5539 & 49.5 + 07746 & 14.090 & 0.5433 & 9.9 + 07747 & 20.134 & -0.9172 & 10.1 + 07748 & 23.568 & 0.2371 & 9.9 + 07749 & 27.385 & -0.8010 & 10.0 + 07750 & 29.320 & -0.8847 & 9.6 + 07867 & 352.998 & 0.2874 & 21.9 + 07868 & 352.998 & 0.2874 & 29.8 + 08173 & 343.778 & 0.9778 & 15.9 + 08259 & 15.255 & -0.4123 & 16.8 + 08601 & 353.253 & 0.2457 & 9.1 + 08602 & 352.662 & 0.2818 & 8.9 + 08603 & 353.468 & 0.2078 & 8.8 + 08604 & 351.640 & 0.2577 & 8.8 + 08605 & 353.372 & 0.0153 & 8.8 + 08606 & 352.973 & 0.2194 & 8.8 + 08607 & 351.899 & 0.2501 & 8.8 + 08608 & 351.723 & 0.0178 & 8.8 + 08609 & 353.090 & -0.0110 & 8.8 + 08610 & 352.159 & 0.3047 & 8.8 + 08611 & 352.012 & 0.0234 & 8.8 + 08612 & 351.474 & 0.0274 & 8.8 + 08613 & 352.254 & 0.0648 & 8.8 + 08614 & 352.803 & 0.0608 & 8.8 + 08615 & 351.481 & 0.2616 & 8.8 + 08616 & 352.543 & 0.0553 & 8.8 + 08617 & 352.422 & 0.2914 & 8.8 + 08619 & 36.640 & 0.7024 & 9.0 + 08620 & 36.730 & 0.4813 & 8.7 + 08621 & 36.887 & 0.7025 & 9.1 + 08622 & 37.144 & 0.7045 & 8.7 + 08623 & 37.392 & 0.7045 & 8.7 + 08624 & 37.639 & 0.7027 & 8.7 + 08625 & 37.890 & 0.7613 & 8.7 + 08626 & 36.979 & 0.4814 & 8.9 + 08627 & 37.227 & 0.4739 & 8.5 + 08628 & 37.483 & 0.4721 & 8.5 + 08629 & 37.727 & 0.4740 & 8.5 + 08630 & 37.890 & 0.5974 & 8.5 + 08960 & 323.372 & -0.8231 & 11.5 + 09594 & 343.778 & 0.9778 & 15.0 + 09719 & 352.998 & 0.2874 &
8.1 + 10388 & 318.970 & 0.0210 & 9.5 + 11351 & 36.618 & 1.1605 & 7.5 + | we present the first results of a wide area x - ray survey within the sloan digital sky survey ( sdss ) stripe 82 , a 300 deg@xmath0 region of the sky with a substantial investment in multi - wavelength coverage .
we analyzed archival _
chandra _
observations that cover 6.2 deg@xmath0 within stripe 82 ( `` stripe 82 acx '' ) , reaching 4.5@xmath1 flux limits of 1.2@xmath2 , 5.4@xmath2 and 2.9@xmath2 erg s@xmath3 @xmath4 in the soft ( 0.5 - 2 kev ) , hard ( 2 - 7 kev ) and full ( 0.5 - 7 kev ) bands , to find 480 , 137 and 705 x - ray sources , respectively .
two hundred fourteen sources are detected only in the full band and 4 sources are detected solely in the soft band . utilizing data products from the _ chandra _ source catalog , we construct independent log@xmath5-log@xmath6 relationships , detailing the number density of x - ray sources as a function of flux .
the soft and full bands show show general agreement with previous _
chandra _
surveys ; the hard band number counts agree among stripe 82 acx , xbootes and xdeep2 , but all 3 are somewhat systematically lower than the counts derived from champ .
we compare the luminosity distribution of stripe 82 acx with the smaller , deeper cdf - s , e - cdfs and _ chandra_-cosmos surveys , to illustrate the benefit of wide - area surveys in locating high luminosity and/or high redshift agn .
finally , we compare the identified agn with predictions from population synthesis models , noting that prior to any spectroscopic follow - up campaign , we have already located roughly half the high luminosity quasars at high redshift expected to lie within the survey area . however , our data also suggests that refinements to population synthesis models will be required . |
new polycrystalline samples of oxygen - substituted ( @xmath0o/@xmath1o ) la@xmath3ba@xmath4cuo@xmath5 with @xmath6 = 1/8 were prepared in the same way as described in @xcite .
the oxygen isotope enrichment of the samples was determined in situ during isotope exchange using mass spectrometry .
the @xmath1o enriched samples contain @xmath19 82(5 ) @xmath62 @xmath1o and @xmath19 18(5 ) @xmath62 @xmath0o .
magnetization experiments were performed with a squid magnetometer ( @xmath63 @xmath64 mpms - xl ) in a field of 0.5 mt and at temperatures between 2 k and 50 k. the @xmath30sr experiments were carried out at the @xmath65m3 beam line at the paul scherrer institut ( psi ) in switzerland using the general purpose instrument ( gps ) with a standard veto setup providing a low - background @xmath30sr signal .
the @xmath30sr time spectra were analyzed using the free software package musrfit @xcite .
for the @xmath0o and @xmath1o samples of lbco-1/8 .
the arrows denote the superconducting transition temperatures @xmath27 .
( b ) the normalized transverse - field @xmath30sr asymmetry @xmath66/@xmath67 plotted as a function of temperature for the @xmath0o and @xmath1o samples of lbco-1/8 .
the crosses mark the spin - stripe order temperatures @xmath0@xmath9 and @xmath1@xmath9 for the @xmath0o and @xmath1o sample , respectively . ]
o ( a ) and @xmath1o ( b ) samples of lbco-1/8 at 2 k. the observed , calculated , and difference plots are shown by open circles , black solid lines , and green solid lines , respectively .
tick marks below each panel represent the positions of allowed bragg reflections compatible with space group @xmath154@xmath16/@xmath17 .
panels ( c ) and ( d ) show the ( 400 ) and ( 004 ) bragg reflections at various temperatures for the @xmath0o and @xmath1o samples , respectively , illustrating the suppression of the orthorhombic splitting below @xmath19 55 k. ] the temperature dependence of the zero field cooled ( zfc ) diamagnetic susceptibility @xmath68 of the @xmath0o and @xmath1o samples of lbco-1/8 reveals an analogous behavior as observed in our previous work @xcite , namely a two - step superconducting transition @xcite where at @xmath27 @xmath19 30 k ( the linearly extrapolated susceptibilities intersect the zero line ) 2d superconductivity sets in in the cuo@xmath16 planes , and at @xmath69 @xmath19 5 k a transition to 3d superconductivity with enhanced diamagnetic response takes place . as discussed previously it is possible to quantify the oie only for the so called high temperature sc transition @xmath27 .
accordingly we show in fig .
6a @xmath70 close to @xmath27 . upon replacing @xmath0o by @xmath1o ,
@xmath27 decreases by @xmath19 1.4 k , corresponding to an oie exponent @xmath71 = 0.46(6 ) in perfect agreement with our previous results @xcite .
the transverse field @xmath30sr asymmetry @xmath66 ( normalized to its maximum value @xmath67 ) has been measured in an applied field of @xmath72@xmath73 = 3 mt following the procedure given in ref .
@xcite ( see fig .
above 40 k , @xmath66 saturates consistent with the fact that the whole sample is paramagnetic .
below 40 k , @xmath66 systematically decreases signaling the appearance of magnetic order .
as already reported previously , @xmath74 is shifted to higher temperatures for the @xmath1o sample as compared to the @xmath0o sample , with an oie exponent @xmath8 = -0.56(9 ) in agreement with ref .
@xcite . during the neutron diffraction experiments ,
the @xmath0o and @xmath1o samples of lbco-1/8 were each placed in 8-mm - diameter vanadium containers and measured in the temperature range from 2 k to 300 k. the experiments were done using a sample changer inside the cryostat to ensure identical temperature conditions for the @xmath0o and @xmath1o samples .
neutron powder diffraction patterns recorded at @xmath75 = 2 k and the corresponding structural refinements for the @xmath0o and @xmath1o samples of lbco-1/8 are shown in fig .
7a and b. the average structure was assessed through rietveld refinements to the raw diffraction data using the program fullprof @xcite employing @xmath124/@xmath13 ( htt ) , @xmath14 ( lto ) , and @xmath154@xmath16/@xmath17 ( ltt ) models from the literature @xcite .
figure 7c and d shows that the orthorhombic splitting between the ( 400 ) and ( 004 ) bragg reflections is clearly resolved .
below @xmath19 55 k the orthorhombic splitting is suppressed , signaling the transition to the ltt phase . | a novel negative oxygen - isotope ( @xmath0o/@xmath1o ) effect ( oie ) on the low - temperature tetragonal phase transition temperature @xmath2 was observed in la@xmath3ba@xmath4cuo@xmath5 ( @xmath6 = 1/8 ) using high - resolution neutron powder diffraction .
the corresponding oie exponent @xmath7 = - 0.36(5 ) has the same sign as @xmath8 = -0.57(6 ) found for the spin - stripe order temperature @xmath9 .
the fact that the ltt transition is accompanied by charge ordering ( co ) implies the presence of an oie also for the co temperature @xmath10 .
furthermore , a temperature dependent shortening of the @xmath11-axis with the heavier isotope is observed .
these results combined with model calculations demonstrate that anharmonic electron - lattice interactions are essential for all transitions observed in the stripe phase of cuprates . since the discovery of high temperature superconductivity in la@xmath3ba@xmath4cuo@xmath5 ( lbco ) @xcite numerous intensive studies have revealed a complex interplay between charge , spin , orbital , and lattice degrees of freedom @xcite .
while the phase diagram of this cuprate family displays a generic complexity , a specific peculiarity is associated with it .
in addition to the phase transition from the high temperature tetragonal ( htt ) phase with space group @xmath124/@xmath13 to the low temperature orthorhombic ( lto ) phase ( @xmath14 ) , a further transition takes place in lbco from the lto to a low temperature tetragonal ( ltt ) phase ( @xmath154@xmath16/@xmath17 ) which strongly depends on the ba content @xmath6 @xcite . the phase transition from htt to lto occurs at @xmath18 @xmath19 235 k @xcite and
is accompanied by the tilting of the oxygen cuo@xmath20 octahedra around the [ 110 ] direction , resulting in buckling distortions of all cu - o - cu bonds in the cuo@xmath16 planes . at @xmath2 @xmath19 55 k alternating octahedra tilt along [ 100 ] and [ 010 ] @xcite .
this restores the macroscopic fourfold symmetry , but breaks the one of the individual planes where half of the cu - o - cu bonds are buckled . the structural transition from lto to ltt
is accompanied by a prominent suppression of superconductivity which is largest for @xmath6 = 1/8 @xcite .
this specific doping level is associated with static stripe order and linked to charge - spin - ordering @xcite . the static charge order ( co )
appears at the same temperature @xmath10 @xmath19 @xmath2 @xmath19 55 k , and spin order ( so ) sets in at @xmath9 @xmath19 35 k @xcite.these results provide clear evidence for a subtle interplay between lattice , spin , and charge degrees of freedom .
the experimental data indicate that uni - directional stripe - like ordering is common to cuprates @xcite . despite of various attempts @xcite , no consensus on the microscopic mechanism of stripe formation and
the relevance of stripe correlations for high - temperature superconductivity in cuprates has been achieved .
the great interest in structural aspects of cuprates is related to the original idea leading to their discovery , namely , jahn - teller polarons @xcite .
substantial experimental evidences for a strong electron - lattice interaction has been reported for cuprates ( see , e.g.,@xcite ) , including the unconventional oxygen isotope ( @xmath0o/@xmath1o ) effects ( oie s ) on various quantities @xcite . however , remarkable electronic properties of cuprates stimulated alternative proposals in terms of purely electronic pairing mechanisms ( see , e.g. @xcite ) . to explore the role of the lattice dynamics on the various ordering temperatures in the stripe phase of cuprates , oie studies @xcite are essential .
so far , a large oie on the superconducting ( sc ) transition temperature @xmath21 was reported for la@xmath22nd@xmath23sr@xmath4cuo@xmath5 @xcite and la@xmath24eu@xmath25sr@xmath4cuo@xmath5 @xcite showing stripe order at @xmath6 = 1/8 @xcite .
recently , we observed substantial oie s on magnetic quantities characterizing the static spin - stripe order in la@xmath3ba@xmath4cuo@xmath5 with @xmath6 = 1/8 ( lbco-1/8 ) @xcite .
however , no oie investigations on the ltt phase transition and charge order in 1/8 doped cuprates have been reported . , @xmath26 ( a ) , and @xmath11 ( b ) for the @xmath0o and @xmath1o samples of lbco-1/8 .
the vertical lines mark the sc transition temperature @xmath0@xmath27 as well as the structural phase transition temperatures @xmath0@xmath2 and @xmath0@xmath18 . ]
@xmath28/@xmath28 , @xmath29@xmath26/@xmath26 ( a ) , and @xmath29@xmath11/@xmath11 ( c ) for the lattice parameters @xmath28 , @xmath26 , and @xmath11 of lbco-1/8 .
the vertical lines mark the sc transition temperature @xmath0@xmath27 as well as the structural phase transition temperatures @xmath0@xmath2 and @xmath0@xmath18 . ] in this letter we report a new negative oie on @xmath2 in lbco-1/8 determined by means of high - resolution neutron powder diffraction with an oie exponent @xmath7 = - 0.35(5 ) , which has the same sign as @xmath8 = -0.57(6 ) found for the spin - stripe order temperature @xmath9 @xcite . since the ltt transition is accompanied by static charge order , we expect an oie also to be present for the co temperature
. these results establish that all transitions observed in the stripe phase of lbco-1/8 are sensitive to oxygen ion related lattice vibrations , indicating that these are relevant for stripe formation in the cuprates .
new polycrystalline samples of oxygen - substituted ( @xmath0o/@xmath1o ) la@xmath3ba@xmath4cuo@xmath5 with @xmath6 = 1/8 were prepared in the same way as described in @xcite .
magnetization , @xmath30sr , and powder neutron diffraction experiments were performed on samples from the same batch in order to study the oie s on @xmath27 @xcite , @xmath9 , and @xmath2 and to make a direct comparison between the oie s on various quantities in the stripe phase of lbco-1/8 .
the oie exponents for @xmath27 and @xmath9 are @xmath31 = 0.46(6 ) and @xmath8 = -0.56(9 ) , respectively ( see supplemental materials for details ) , being in perfect agreement with our previous results @xcite .
the neutron diffraction experiments were carried out at the high - resolution powder diffractometer hrpt @xcite at the swiss spallation neutron source sinq at the paul scherrer institut ( psi ) , switzerland , with a wavelength of @xmath32 = 0.1494 nm .
= 2(@xmath33)/(@xmath34 ) for the @xmath0o and @xmath1o samples of lbco-1/8 .
the arrows denote the ltt transition temperatures @xmath0@xmath2 and @xmath1@xmath2 for the @xmath0o and @xmath1o samples , respectively .
the peculiarities at higher temperatures are artefacts connected with the exchange of the data treatment procedure . ]
the oie s on the ltt structural phase transition temperature @xmath2 and on the lattice parameters @xmath28 , @xmath26 , and @xmath11 of lbco-1/8 were studied by neutron diffraction .
the crystal structure of lbco-1/8 at room temperature ( rt ) was refined with the tetragonal @xmath124/@xmath13 space group using the program fullprof @xcite .
lattice parameters are @xmath28 = 0.5355(2 ) nm and @xmath11 = 1.3275(7 ) nm at rt .
the temperature dependence of the lattice parameters for the @xmath0o and @xmath1o samples of lbco-1/8 are shown in fig .
1a and 1b .
the lto and ltt structural phase transitions are clearly visible in the temperature dependences of @xmath28 and @xmath26 . at @xmath18 @xmath19 230 k a splitting of @xmath28 and @xmath26 reveals the transition to the lto structure which smoothly evolves from the htt phase . around @xmath35 @xmath19 55 k
these lattice parameters merge together , and a first - order transition from the lto to the ltt phase takes place @xcite .
the @xmath11-axis smoothly contracts with decreasing temperature ( see fig .
1b ) .
the values for @xmath18 and @xmath2 ( marked by the vertical lines in fig . 1 ) and those for the lattice parameters for the @xmath0o sample are in good agreement with published values @xcite .
the oie s on the lattice constants are displayed in fig .
2a and 2b where the temperature dependence of the relative isotope shifts of the lattice parameters @xmath29@xmath36/@xmath36 = ( @xmath1q-@xmath0q)/@xmath0q ( @xmath36 = @xmath28 , @xmath26 , @xmath11 ) are shown . in all structural phases the in - plane lattice parameters @xmath28 and @xmath26 for lbco-1/8 decrease only slightly upon replacing @xmath0o by @xmath1o .
the quantity @xmath29@xmath28/@xmath28 increases with decreasing temperature , reaching a maximum around @xmath19 100 k and decreases below . at room temperature @xmath29@xmath11/@xmath11
is nearly the same as @xmath29@xmath28/@xmath28 .
however , upon lowering the temperature @xmath29@xmath11/@xmath11 monotonically increases to be by a factor of 6 larger at the base temperature than at rt @xcite . of the ltt and
lto structural phases for the @xmath0o ( a ) and @xmath1o ( b ) samples of lbco-1/8 .
the vertical lines mark the ltt transition temperatures @xmath0@xmath2 and @xmath1@xmath2 for the @xmath0o and @xmath1o samples , respectively . ] in order to illustrate the oie on @xmath2 , we plot the orthorhombic strain @xmath37 = 2(a - b)/(a+b ) @xcite for the two isotope samples as a function of temperature ( see fig . 3 ) .
@xmath37 sharply drops to zero at @xmath35 @xmath19 55 k for the @xmath0o sample , consistent with the first order nature of the ltt transition .
the oie on @xmath2 is clearly visible as a pronounced shift to higher temperature with increasing isotope mass .
the results for the oie on @xmath2 are : @xmath0@xmath2 = 54.5(3 ) k , @xmath1@xmath2 = 56.5(3 ) k , @xmath29@xmath2 = @xmath1@xmath2 - @xmath0@xmath2 = 2.1(2 ) k , yielding an oie exponent @xmath7 = -0.36(5 ) .
the oie on the ltt transition can also be obtained from the relative weight fractions @xmath38 of the lto and ltt phases as a function of temperature , evidencing the coexistence of the lto and the ltt phases over a temperature interval of @xmath19 8 k. this is consistent with the first order nature of the ltt transition . in fig .
4a and b
we plot the temperature dependences of @xmath38 of the lto and the ltt phase for the @xmath0o and @xmath1o samples , respectively .
for the @xmath0o sample , the ltt phase appears below @xmath19 63 k and @xmath38 increases with decreasing temperature .
the transition to the ltt phase is complete below 54 k , and between 54 k and 63 k the ltt phase coexists with the lto one .
@xmath2 is defined as the temperature where @xmath38 of the lto and the ltt phase are equal , yielding @xmath0@xmath2 = 58.5(3 ) k , @xmath1@xmath2 = 60.7(3 ) k , and @xmath7 = -0.37(5 ) , in excellent agreement with @xmath7 = -0.36(5 ) obtained from the orthorhombic strain ( fig .
3 ) .
this demonstrates that the two independent methods for the determination of @xmath7 yield consistent results .
the oie on the structural phase transition temperature @xmath2 is accompanied by a clear shrinkage of the @xmath11-axis with the heavier isotope in the temperature range from 2 k to 300 k ( see fig .
2b ) .
both phase transitions , the one from htt to lto and lto to ltt are accompanied by substantial phonon softening @xcite , where the first one is associated with the condensation of a degenerate pair of transverse - optic ( to ) phonon modes at the x - point wave vector q=(1/2 , 1/2 , 0 ) , while the low temperature one is driven by a soft z - point phonon mode .
this mode emerges from the high temperature soft mode , which splits into two modes below @xmath18 , corresponding to a zone boundary mode at the ( 3,0,2 ) z point and a zone center one at the ( 0,3,2 ) @xmath39 point .
while the latter of the two modes hardens , the z point one continues to soften up to @xmath2 .
an understanding of this behavior and the transitional properties has been given in terms of a landau free energy expansion , where symmetry invariant combinations of a degenerate order parameter @xmath40 ( @xmath41 = 1,2 ) and the orthorhombic strain @xmath37 have been considered @xcite : @xmath42 @xmath43 is the transition temperature which depends on the coefficients @xmath28 , @xmath44 , @xmath45 , @xmath11 , and @xmath46 , and refers either to @xmath47 or @xmath2 . within this approach it is possible to construct the complex phase diagram , however , isotope effects are absent .
these are obtained within a dynamic anharmonic electron lattice interaction hamiltonian which reduces in the limit that only the soft modes are taken into account , to @xcite : @xmath48 with @xmath49 being the momenta in cell @xmath41 , @xmath50 the oxygen ion mass , @xmath51 , @xmath52 the harmonic and fourth order electron - ion interaction constants , @xmath53 the relative electron - ion displacement coordinate , and @xmath54 the long range interactions between sites @xmath41 , @xmath55 .
the fourth order term proportional to @xmath52 is treated within the self - consistent phonon approximation ( spa ) , and a phase transition takes place at @xcite @xmath56^{-1}\ ] ] ( a ) and the relative isotope shift of the dynamical relative oxygen - copper displacement @xmath57/@xmath58 ( b ) of lbco-1/8 , plotted together with @xmath29@xmath11/@xmath11 , taken from fig .
2b . ]
in order to determine the correct symmetry of the soft mode , the free energy [ eq .
( 1 ) ] has to be employed for the corresponding order parameter .
this is included in eq .
( 2 ) by solving the equations of motion and take the zero solution for the soft mode at the appropriate @xmath36-value from which also @xmath43 is obtained .
a simplified solution of @xmath43 is given by eq .
( 3 ) from which an isotope effect on @xmath43 is derived with the parameter values of ref . 3 for @xmath51 , @xmath52 , and @xmath54 as appropriate for the @xmath6 = 1/8 doping , yielding @xmath0@xmath43 @xmath19 54.8 k and @xmath1@xmath43 @xmath19
57.0 k. this corresponds to an oie exponent of @xmath59 @xmath19 -0.39 , in good agreement with the experimental value @xmath7 = -0.36(5 ) .
importantly , the calculated values for @xmath0@xmath43 and @xmath1@xmath43 are consistent with the experimental values @xmath0@xmath2 and @xmath1@xmath2 given above .
the z - point related soft phonon mode is derived within the spa from the equations of motion for the two isotopes .
fig .
5a shows the calculated temperature dependence of the z - point related phonon mode frequency @xmath60 . in this case
the transition temperatures are @xmath0@xmath43 @xmath19 52.1 k and @xmath1@xmath43 @xmath19 54.8 k , which are slightly shifted to lower temperatures as compared to the experimental values .
however , the corresponding oie exponent @xmath59 @xmath19 -0.49 is in agreement with the experiment . in order to achieve an additional correspondence between experiment and theory , the relative mean square cu - o displacement @xmath61 along the @xmath11-axis is calculated for both isotopes , @xmath0o and @xmath1o , respectively , and their relative difference @xmath29(@xmath58)/@xmath58 = ( @xmath1@xmath58-@xmath0@xmath58)/@xmath0@xmath58 is shown in fig .
5b . the relative displacement for @xmath1o is smaller than for @xmath0o in agreement with the data for the lattice parameter @xmath11 . in fig .
5b the temperature dependence of the relative isotope shift @xmath29(@xmath58)/@xmath58 is shown together with @xmath29@xmath11/@xmath11 taken from fig .
2b .
while a temperature independent systematic shift in absolute values is visible , there is salient agreement in the corresponding temperature dependences , providing strong evidence that the ltt structural phase transition is driven by phonon mode softening caused by anharmonic electron - lattice interactions @xcite . in conclusion ,
oxygen isotope effects on lattice properties in the static stripe phase of lbco-1/8 were investigated by means of neutron diffraction .
the low - temperature tetragonal phase transition temperature @xmath2 exhibits a large negative oie with @xmath7 = - 0.36(5 ) .
in addition , the temperature dependent contraction of the @xmath11-axis with the heavier isotope is observed .
since the ltt transition is linked to static charge order , we expect an oie to be present for the co temperature as well .
these results combined with the previously observed oie on @xmath9 evidence that all transitions observed in the stripe phase of lbco-1/8 are sensitive to oxygen lattice vibrations .
the theoretical lattice dynamical calculations based on a dynamic anharmonic electron lattice interaction hamiltonian consistently reproduce the experimental results .
this structural instability is driven by phonon mode softening stemming from anharmonic electron - lattice interactions , which are essential for the stripe formation in the cuprates .
we acknowledge elvezio morenzoni , vladimir pomjakushin and jonathan white for helpful discussions .
this work was supported by the swiss national science foundation , the nccr manep , the scopes grant no .
iz74z0 - 137322 , and the georgian national science foundation grant rnsf / ar/10 - 16 . 22 j.g . bednorz and k.a .
mller , z. phys .
b * 64 * , 189 ( 1986 ) .
j.d .
jorgensen , h.b .
schttler , d.g .
hinks , d.w .
capone ii , k. zhang , m. b. brodsky , and d. j. scalapino , phys .
rev .
lett . * 58 * , 1024 ( 1987 ) .
a. bussmann - holder , a. migliori , z. fisk , j. l. sarrao , r. g. leisure , and s .- w .
cheong , phys .
rev .
lett . * 67 * , 512 ( 1991 ) .
j.d .
axe , a.h .
moudden , d. hohlwein , d.e .
cox , k.m .
mohanty , a.r .
moodenbaugh , and y. xu , phys .
rev .
lett . * 62 * , 2751 ( 1989 ) .
m. hcker , m.v .
zimmermann , g.d .
gu , z.j .
xu , j.s .
wen , guangyong xu , h.j .
kang , a. zheludev , and j.m .
tranquada , phys .
rev .
b * 83 * , 104506 ( 2011 ) .
a.r .
moodenbaugh , y. xu , m. suenaga , t.j .
folkerts , and r.n .
shelton , phys .
rev .
b * 38 * , 4596 ( 1988 ) .
g.m .
luke , l.p .
le , b.j .
sternlieb , w.d .
wu , y.j .
uemura , j.h .
brewer , t.m .
riseman , s. ishibashi , and s. uchida , physica c * 185 - 9 * , 1175 ( 1991 ) .
m.k .
crawford , m. kunchur , w.e .
farneth , e.m .
mccarron , and s.j .
poon , physica c * 162 - 164 * , 755 ( 1989 ) .
j.m .
tranquada , b.j .
sternlieb , j.d .
axe , y. nakamura , and s. uchida , nature ( london ) * 375 * , 561 ( 1995 ) .
j.m .
tranquada , j.d .
axe , n. ichikawa , y. nakamura , s. uchida , and b. nachumi , phys .
rev .
b * 54 * , 7489 ( 1996 ) .
p. abbamonte , a. rusydi , s. smadici , g.d .
gu , g.a . sawatzky , and d.l .
feng , nat . phys . * 1 * , 155 ( 2005 ) .
m. hcker , m. v. zimmermann , m. debessai , j.s .
schilling , j.m .
tranquada , and g.d .
gu , phys .
rev .
lett . * 104 * , 057004 ( 2010 ) .
s.a .
kivelson , i.p .
bindloss , e. fradkin , v. oganesyan , j.m .
tranquada , a. kapitulnik , and c. howald , rev .
mod .
phys . * 75 * , 1201 ( 2003 ) .
m. vojta , adv .
phys . * 58 * , 699 ( 2009 ) .
y. kohsaka , c. taylor , k. fujita , a. schmidt , c. lupien , t. hanaguri , m. azuma , m. takano , h. eisaki , h. takagi , s. uchida , and j. c. davis , science * 315 * , 1380 ( 2007 ) .
j.m .
tranquada , g.d .
gu , m. hcker , q. jie , h .- j .
kang , r. klingeler , q. li , n. tristan , j.s .
wen , g.y .
xu , z.j .
xu , j. zhou , and m. v. zimmermann , phys . rev .
b * 78 * , 174529 ( 2008 ) .
t. valla , a. v. federov , j. lee , j. c. davis , and g. d. gu , science * 314 * , 1914 ( 2006 ) .
q. li , m. hcker , g.d .
gu , a.m. tsvelik , and j.m .
tranquada , phys .
rev .
lett . * 99 * , 067001 ( 2007 ) .
e. berg , e. fradkin , e .- a .
kim , s. a. kivelson , v. oganesyan , j. m. tranquada , and s. c. zhang , phys . rev .
lett . * 99 * , 127003 ( 2007 ) .
r .- h .
he , k. tanaka , s .- k .
mo , t. sasagawa , m. fujita , t. adachi , n. mannella , k. yamada , y. koike , z. hussain , and z .- x .
shen , nat .
phys . * 5 * , 119123 ( 2009 ) .
j.m .
tranquada , aip conference proceedings * 1550 * , 114 ( 2013 ) .
z. guguchia , a. maisuradze , g. ghambashidze , r. khasanov , a. shengelaya , and h. keller , new journal of physics * 15 * , 093005 ( 2013 ) .
z. guguchia , r. khasanov , m. bendele , e. pomjakushina , k. conder , a. shengelaya , and h. keller , phys .
rev .
lett . * 113 * , 057002 ( 2014 ) .
k.a .
mller , j. phys .
condens .
matter * 19 * , 251002 ( 2007 ) .
h. keller , a. bussmann - holder , and k.a .
mller , materials today * 11 * , 9 ( 2008 ) ; h. keller , and a. bussmann - holder , advances in condensed matter physics ( volume 2010 , article i d 393526 , 17 pages , doi:10.1155/2010/393526 ) .
a. shengelaya , guo - meng zhao , c.m .
aegerter , k. conder , i.m .
savi , and h. keller , phys .
rev .
lett . * 83 * , 24 ( 1999 ) .
r. khasanov , a. shengelaya , d. di castro , e. morenzoni , a. maisuradze , i.m .
savi , k. conder , e. pomjakushina , a. bussmann - holder , and h. keller , phys .
rev .
lett .
* 101 * , 077001 ( 2008 ) .
a. lanzara , guo - meng zhao , n.l .
saini , a. bianconi , k. conder , h. keller , and mller , j. phys . :
condens matter * 11 * , l541 ( 1999 ) .
d. rubio temprano , j.mesot , s. janssen , k. conder , a. furrer , h. mutka , and k.a.mller , phys .
rev .
lett . * 84 * , 1990 ( 2000 ) .
p. s. hfliger , s. gerber , r. pramod , v. i. schnells , b. dalla piazza , r. chati , v. pomjakushin , k. conder , e. pomjakushina , l. le dreau , n. b. christensen , o. f. syljusen , b. normand , and h.m .
rnnow phys .
rev .
b * 89 * , 085113 ( 2014 ) .
g.y .
wang , j.d .
zhang , r.l .
yang , and x.h .
chen , phys .
rev .
b * 75 * , 212503 ( 2007 ) .
suryadijaya , t. sasagawa , and h. takagi , physica c * 426 * , 402 ( 2005 ) .
m.k .
crawford , r.l .
harlow , e.m .
mccarron , w.e .
farneth , j.d .
axe , h. chou , and q. huang , phys .
rev .
b * 44 * , 7749 ( 1991 ) .
b. bchner , m. breuer , a. freimuth , and a.p .
kampf , phys .
rev .
lett . * 73 * , 1841 ( 1994 ) .
p. fischer , g. frey , m. koch , m.konnecke , v. pomjakushin , j. schefer , r. thut , n. schlumpf , r.burge , u. greuter , s. bondt , and e. berruyer , physica b * 276 - 278 * , 146 ( 2000 ) .
j. rodrguez - carvajal , physica b * 192 * , 55 ( 1993 ) .
note that some peculiar features close to @xmath2 are visible .
however , due to the fact that close to the ltt to lto structural phase transition both phases coexist ( see discussion below ) these pecularities are likely related to artefacts caused by the exchange of the data treatment procedure .
h. kimura , y. noda , h. goka , m. fujita , k. yamada , and g. shirane , j. phys .
. jpn . * 74 * , 445 ( 2005 ) .
b. keimer , r. j. birgenau , a. cassanho , y. endoh , m. greven , m. a. kastner , and g. shirane , z. phys .
b * 91 * , 373 ( 1993 ) .
wu ting , k. fossheim , and t. laegreid , sol .
st .
comm . * 75 * , 727 ( 1990 ) .
a. bussmann - holder , j. phys . :
condens matter * 24 * , 273202 ( 2012 ) .
a. bussmann - holder , a. simon , and a. r. bishop , zeitschrift fr kristallographie * 226 * , 177 ( 2011 ) .
g. mahan , phys .
rev .
b * 56 * , 8322 ( 1997 ) .
j. zhong and h .- b .
schttler , phys .
rev .
lett . * 69 * , 1600 ( 1992 ) .
a. bussmann - holder and a. bishop , phys .
rev .
b * 44 * , 2853 ( 1991 ) .
j. hirsch , phys .
rev .
b * 47 * , 5351 ( 1993 ) . |
we consider only undirected graphs without loops or multiple edges . our terminology and notation will be standard except as indicated , and a good reference for any undefined terms or notation is @xcite .
a _ degree sequence _ of a graph on @xmath6 vertices is any sequence @xmath7 consisting of the vertex degrees of the graph .
in contrast to @xcite , we will usually assume the sequence is in nondecreasing order .
we generally use the standard abbreviated notation for degree sequences , e.g. , @xmath8 will be denoted @xmath9 . a sequence of integers @xmath10
is called _ graphical _ if there exists a graph @xmath11 having @xmath0 as one of its degree sequences , in which case we call @xmath11 a _ realization _ of @xmath0 .
if @xmath12 and @xmath13 are two integer sequences , we say @xmath14 _ majorizes _
@xmath0 , denoted @xmath15 , if @xmath16 for @xmath17 .
if @xmath18 is a graphical property ( e.g. , @xmath1-connected , hamiltonian ) , we call a graphical degree sequence _ forcibly _ ( respectively , _ potentially _ ) _ @xmath18 graphical _ if every ( respectively , some ) realization of @xmath0 has property @xmath18 . historically , the degree sequence of a graph has been used to provide sufficient conditions for a graph to have a certain property , such as @xmath1-connected or hamiltonian . sufficient conditions for a degree sequence to be forcibly hamiltonian were given by several authors , culminating in the following theorem of chvtal @xcite in 1972 .
[ thm : chvatal ] let @xmath19 be a graphical degree sequence , with @xmath20 . if @xmath21 implies @xmath22 , then @xmath0 is forcibly hamiltonian graphical . unlike its predecessors
, chvtal s theorem has the property that if it does not guarantee that a graphical degree sequence @xmath0 is forcibly hamiltonian graphical , then @xmath0 is majorized by some degree sequence @xmath14 which has a nonhamiltonian realization .
as we ll see , this fact implies that chvtal s theorem is the strongest of an entire class of theorems giving sufficient conditions for @xmath0 to be forcibly hamiltonian graphical . a _ factor _ of a graph @xmath11 is a spanning subgraph of @xmath11 .
a _ @xmath1-factor _ of @xmath11 is a factor whose vertex degrees are identically @xmath1 . for a recent survey on graph factors ,
see @xcite . in the present paper
, we develop sufficient conditions for a degree sequence to be forcibly @xmath1-factor graphical
. we note that previous work relating degrees and the existence of factors has focused primarily on sufficient conditions for @xmath0 to be potentially graphical .
the following obvious necessary condition was conjectured to be sufficient by rao and rao @xcite , and this was later proved by kundu @xcite .
[ kundu ] the sequence @xmath10 is potentially @xmath1-factor graphical if and only if to10mm(1)@xmath23 is graphical , and to10mm(2)@xmath24 is graphical .
kleitman and wang @xcite later gave a proof of theorem [ kundu ] that yielded a polynomial algorithm constructing a realization @xmath11 of @xmath0 with a @xmath1-factor .
lovsz @xcite subsequently gave a very short proof of theorem [ kundu ] for the special case @xmath3 , and chen @xcite produced a short proof for all @xmath25 .
in section [ s2 ] , we give a theorem for @xmath0 to be forcibly graphical with deficiency at most @xmath26 ( i.e. , have a matching missing at most @xmath26 vertices ) , and show this theorem is strongest in the same sense as chvtal s hamiltonian degree theorem .
the case @xmath27 gives the strongest result for @xmath0 to be forcibly 1-factor graphical . in section [ s3 ] , we give the strongest theorem , in the same sense as chvtal , for @xmath0 to be forcibly 2-factor graphical .
but the increase in the number of nonredundant conditions which must be checked as we move from a 1-factor to a 2-factor is notable , and we conjecture the number of such conditions in the best monotone theorem for @xmath0 to be forcibly @xmath1-factor graphical increases superpolynomially in @xmath1 . thus it would be desirable to find a theorem for @xmath0 to be forcibly @xmath1-factor graphical in which the number of nonredundant conditions grows in a more reasonable way . in section [ s4 ] ,
we give such a theorem for @xmath28 , based on tutte s well - known factor theorem . while our theorem is not best monotone , it is nevertheless tight in a precise way , and we provide examples to illustrate this tightness .
we conclude this introduction with some concepts which are needed in the sequel .
let @xmath18 denote a graph property ( e.g. , hamiltonian , contains a , etc . )
such that whenever a spanning subgraph of @xmath11 has @xmath18 , so does @xmath11 .
a function @xmath29 such that @xmath30 implies @xmath0 is forcibly @xmath18 graphical , and @xmath31 implies nothing in this regard , is called a _
forcibly @xmath18 function_. such a function is called _ monotone _
if @xmath32 and @xmath30 implies @xmath33 , and _ weakly optimal _
if @xmath31 implies there exists a graphical sequence @xmath32 such that @xmath14 has a realization @xmath34 without @xmath18 . a forcibly @xmath18 function which is both monotone and weakly optimal is the best monotone forcibly @xmath18 function , in the following sense .
if @xmath35,@xmath36 are monotone , forcibly @xmath18 functions , and @xmath36 is weakly optimal , then @xmath37 , for every graphical sequence @xmath0 .
* proof : * suppose to the contrary that for some graphical sequence @xmath0 we have @xmath38 . since @xmath36 is weakly optimal , there exists a graphical sequence @xmath32 such that @xmath14 has a realization @xmath34 without @xmath18 , and thus @xmath39 .
but @xmath32 , @xmath30 and @xmath39 imply @xmath35 can not be monotone , a contradiction .
@xmath40 a theorem @xmath41 giving a sufficient condition for @xmath0 to be forcibly @xmath18 corresponds to the forcibly @xmath18 function @xmath42 given by : @xmath43 if and only if @xmath41 implies @xmath0 is forcibly @xmath18 .
it is well - known that if @xmath41 is theorem [ thm : chvatal ] ( chvtal s theorem ) , then @xmath42 is both monotone and weakly optimal , and thus the best monotone forcibly hamiltonian function in the above sense . in the sequel , we will simplify the formally correct ` @xmath42 is monotone , etc . ' to ` @xmath41 is monotone , etc .. '
in this section we present best monotone conditions for a graph to have a large matching .
these results were first obtained by las vergnas @xcite , and can also be obtained from results in bondy and chvtal @xcite . for the convenience of the reader
, we include the statement of the results and short proofs below .
the _ deficiency _ of @xmath11 , denoted @xmath44 , is the number of vertices unmatched under a maximum matching in @xmath11 .
in particular , @xmath11 contains a 1-factor if and only if @xmath45 .
we first give a best monotone condition for @xmath0 to be forcibly graphical with deficiency at most @xmath26 , for any @xmath46 .
[ thm : def ] let @xmath11 have degree sequence @xmath47 , and let @xmath48 with @xmath49 . if @xmath50 then @xmath51 .
the condition in theorem [ thm : def ] is clearly monotone .
furthermore , if @xmath0 does not satisfy the condition for some @xmath52 , then @xmath0 is majorized by @xmath53 . but @xmath14 is realizable as @xmath54 , which has deficiency @xmath55 .
thus theorem [ thm : def ] is weakly optimal , and the condition of the theorem is best monotone . * * proof of theorem [ thm : def]:**suppose @xmath0 satisfies the condition in theorem [ thm : def ] , but @xmath56 .
( the condition @xmath57 guarantees that @xmath58 is always even . )
define @xmath59 , with degree sequence @xmath60 @xmath61 .
note that the number of vertices of @xmath34 is odd .
suppose @xmath34 has a hamilton cycle .
then , by taking alternating edges on that cycle , there is a matching covering all vertices of @xmath34 except one vertex , and we can choose that missed vertex freely .
so choose a matching covering all but one of the @xmath62 new vertices . removing the other @xmath26 new vertices as well ,
the remaining edges form a matching covering all but at most @xmath26 vertices from @xmath11 , a contradiction .
hence @xmath34 can not have a hamilton cycle , and @xmath14 can not satisfy the condition in theorem [ thm : chvatal ] .
thus there is some @xmath63 such that @xmath64 subtracting @xmath62 throughout this equation gives @xmath65 replacing @xmath66 by @xmath67 we get @xmath68 thus @xmath0 fails to satisfy the condition in theorem [ thm : def ] , a contradiction .
@xmath40 as an important special case , we give the best monotone condition for a graph to have a 1-factor .
[ thm:1-factor ] let @xmath11 have degree sequence @xmath19 , with @xmath69 and @xmath6 even . if @xmath70 then @xmath11 contains a 1-factor .
we note in passing that is chvtal s best monotone condition for @xmath11 to have a hamiltonian path @xcite .
we now give a best monotone condition for the existence of a 2-factor . in what follows
we abuse the notation by setting @xmath71 .
[ thm:2-factor ] let @xmath11 have degree sequence @xmath19 , with @xmath20 .
if to10mm(i ) @xmath6 odd @xmath72 @xmath73 ; to10mm(ii ) @xmath6 even @xmath72 @xmath74 or @xmath75 ; to10mm(iii ) @xmath76 and @xmath77 @xmath72 @xmath78 or @xmath79 , for @xmath80 ; to10mm(iv ) @xmath81 and @xmath82 @xmath72 @xmath83 or @xmath84 , for @xmath85 , then @xmath11 contains a 2-factor . the condition in theorem [ thm:2-factor ] is easily seen to be monotone . furthermore ,
if @xmath0 fails to satisfy any of ( i ) through ( iv ) , then @xmath0 is majorized by some @xmath14 having a realization @xmath34 without a 2-factor . in particular ,
note that @xmath86if ( i ) fails , then @xmath0 is majorized by @xmath87 , having realization @xmath88 ; @xmath86if ( ii ) fails , then @xmath0 is majorized by @xmath89 , having realization @xmath90 ; @xmath86if ( iii ) fails for some @xmath66 , then @xmath0 is majorized by @xmath91 @xmath92 , having realization @xmath93 together with an edge joining @xmath94 and @xmath95 ; @xmath86if ( iv ) fails for some @xmath66 , then @xmath0 is majorized by @xmath96 @xmath97 , having realization @xmath98 together with three independent edges joining @xmath99 and @xmath100 .
it is immediate that none of the above realizations contain a 2-factor .
hence , theorem [ thm:2-factor ] is weakly optimal , and the condition of the theorem is best monotone . * * proof of theorem [ thm:2-factor]:**suppose @xmath0 satisfies ( i ) through ( iv ) , but @xmath11 has no
. we may assume the addition of any missing edge to @xmath11 creates a 2-factor .
let @xmath101 be the vertices of @xmath11 , with respective degrees @xmath102 , and assume @xmath103 are a nonadjacent pair with @xmath104 as large as possible , and @xmath105 .
then @xmath106 must be adjacent to @xmath107 and so @xmath108 similarly , @xmath109 must be adjacent to @xmath110 , and so @xmath111 since @xmath112 has a 2-factor , @xmath11 has a spanning subgraph consisting of a path @xmath18 joining @xmath106 and @xmath109 , and @xmath113 cycles @xmath114 , all vertex disjoint .
we may also assume @xmath115 and @xmath18 are chosen such that if @xmath116 are any nonadjacent vertices with @xmath117 and @xmath118 , and if @xmath119 is any @xmath120-path such that @xmath121 has a , then @xmath122 . otherwise , re - index the set of vertices of degree @xmath123 ( resp .
, @xmath124 ) so that @xmath125 ( resp . , @xmath126 )
is given the highest index in the set .
since @xmath11 has no , we can not have independent edges between @xmath127 and two consecutive vertices on any of the @xmath128 , @xmath129 .
similarly , we can not have @xmath130 , since otherwise @xmath131 is hamiltonian and @xmath11 contains a .
this means @xmath132
\text{and}\quad d_p(v_j ) + d_p(v_k ) \leq |v(p)|-1 .
\end{array}\ ] ] it follows immediately that @xmath133 we distinguish two cases for @xmath134 .
case 1:@xmath135 .
using , we obtain @xmath136 take @xmath137 so that @xmath138 , where @xmath139 . by case 1
we have @xmath140 . since also @xmath141 and @xmath142 , condition ( iii ) implies @xmath143 or @xmath144 .
in either case , @xmath145 adding @xmath146 to , we obtain @xmath147 but @xmath148 and together give @xmath149 , hence @xmath150 . on the other hand ,
gives @xmath151 , hence @xmath152 , a contradiction .
@xmath153 case 2:@xmath154 . in this case we have equality in ,
hence all the inequalities in become equalities .
in particular , this implies that every cycle @xmath128 , @xmath155 , satisfies one of the following conditions : to10mm(a ) every vertex in @xmath128 is adjacent to @xmath106 ( resp . , @xmath109 ) , and
none are adjacent to @xmath109 ( resp . ,
@xmath106 ) , or to10mm(b)@xmath156 is even , and @xmath103 are both adjacent to the same alternate vertices on @xmath128 .
we call a cycle of type ( a ) a _ @xmath157-cycle _
( resp . , _ @xmath1-cycle _ ) , and a cycle of type ( b ) a _ @xmath158-cycle_. set @xmath159 , @xmath160 , and @xmath161 , and let @xmath162 , @xmath163 , and @xmath164 .
vertices in @xmath165 which are adjacent to both ( resp . ,
neither ) of @xmath115 will be called _ large _ ( resp . ,
_ small _ ) vertices .
in particular , the vertices of each @xmath158-cycle are alternately large and small , and hence there are @xmath166 small and @xmath166 large vertices among the @xmath158-cycles . by the definitions of @xmath167 , noting that a cycle has at least 3 vertices , we have the following . [ obs1 ]
we have @xmath168 or @xmath169 , @xmath170 or @xmath171 , and @xmath172 or @xmath173 . by the choice of @xmath115 and @xmath18
, we also have the following observations . [ obs2 ] to10mm(a)if @xmath174 , then @xmath175 ; if @xmath176 , then @xmath177 .
to10mm(b)a vertex in @xmath178 has degree at most @xmath179 .
to10mm(c)a vertex in @xmath180 has degree at most @xmath181 .
to10mm(d)a small vertex in @xmath182 has degree at most @xmath179 . * * proof:**part ( a ) follows directly from the choice of @xmath115 as nonadjacent with @xmath183 maximal . for ( b ) ,
consider any @xmath184 , with say @xmath185 .
since @xmath186 , we have @xmath187 by the maximality of @xmath104 , and so @xmath188 . if @xmath189 , then since each vertex in @xmath178 is adjacent to @xmath106 , we can combine the path @xmath18 and the @xmath157-cycle @xmath128 containing @xmath190 ( leaving the other cycles @xmath128 alone ) into a path @xmath119 joining @xmath190 and @xmath109 such that @xmath121 has a and @xmath191 , contradicting the choice of @xmath18 .
thus @xmath192 , proving ( b ) .
parts ( c ) and ( d ) follow by a similar arguments .
let @xmath193 , and let us re - index @xmath18 as @xmath194 . by the case assumption , @xmath195 .
assume first that @xmath196 .
then @xmath197 and @xmath198 , so that @xmath199 . moreover , @xmath200 and there are @xmath201 large vertices and @xmath166 small vertices .
if @xmath202 , the large vertex @xmath203 is not adjacent to a vertex in @xmath178 or to a small vertex in @xmath182 , or else @xmath11 contains a 2-factor .
thus @xmath203 has degree at most @xmath204 , and by observations 2(b , c , d ) , @xmath0 is majorized by @xmath205 setting @xmath206 , so that @xmath207 , @xmath208 becomes @xmath209 since @xmath208 majorizes @xmath0 , we have @xmath210 , @xmath211 , @xmath212 , and @xmath213 , and @xmath0 violates condition ( iii ) .
hence @xmath170 by observation 1 , and a fortiori @xmath168 .
but if @xmath214 , then @xmath215 , @xmath6 is odd , and by observation 2(d ) , @xmath0 is majorized by @xmath216 since @xmath217 majorizes @xmath0 , we have @xmath218 , and @xmath0 violates condition ( i ) .
hence we assume @xmath219 .
we make several further observations regarding the possible adjacencies of @xmath115 into the path @xmath18 .
[ obs3 ] for all @xmath220 , @xmath221 , we have @xmath222 if and only if @xmath223 .
* * proof:**if @xmath224 then , @xmath225 , since otherwise @xmath131 is hamiltonian and @xmath11 has a 2-factor .
the converse follows since @xmath226 .
@xmath153 [ obs4 ] if @xmath227 for some @xmath220 , @xmath228 , then we have @xmath229 .
* * proof:**if @xmath230 , then @xmath231 by observation [ obs3 ] .
but since @xmath232 , this means that @xmath131 would have a 2-factor consisting of the cycles @xmath233 and @xmath234 , and thus @xmath11 would have a , a contradiction .
observation [ obs4 ] implies that if @xmath235 is adjacent to consecutive vertices @xmath236 for some @xmath237 , then @xmath235 is adjacent to all of the vertices @xmath238 .
[ obs5 ] if @xmath239 for some @xmath240 , then we have @xmath241 .
* * proof:**if @xmath242 , then @xmath243 by observation [ obs3 ] .
so if also @xmath244 , then @xmath131 would have a 2-factor as in the proof of observation [ obs4 ] , leading to the same contradiction .
observation [ obs5 ] implies that if @xmath235 is not adjacent to two consecutive vertices @xmath245 on @xmath18 for some @xmath246 , then @xmath235 is not adjacent to any of @xmath247 . by observation [ obs3 ] , the adjacencies of @xmath235 into @xmath18 completely determine the adjacencies of @xmath248 into @xmath18 . but combining observations [ obs4 ] and [ obs5 ]
, we see that the adjacencies of @xmath235 and @xmath248 into @xmath18 must appear as shown in figure 1 , for some @xmath249 .
in summary , @xmath235 will be adjacent to @xmath250 consecutive vertices @xmath251 ( where @xmath252 is taken to be empty if @xmath253 ) , @xmath248 will be adjacent to @xmath254 consecutive vertices @xmath255 , and @xmath256 are each adjacent to the vertices @xmath257 .
note that @xmath258 implies @xmath259 , and @xmath260 implies @xmath261 . on @xmath18.,scaledwidth=60.0% ]
counting neighbors of @xmath235 and @xmath248 we get their degrees as follows .
@xmath262 a+c+p-2,&\text{if $ r = p-2 $ , $ \ell=0$},\\[4pt ] a+c+r+\tfrac12(p - r-\ell-1);&\text{otherwise};\end{array}\right.\\[2 mm ] d_k = d_g(w_p)&=\left\{\begin{array}{ll } b+c+p-2,&\text{if $ \ell = p-2 $ , $ r=0$},\\[4pt ] b+c+1,&\text{if $ r = p-2 $ , $ \ell=0$},\\[4pt ] b+c+\ell+\tfrac12(p - r-\ell-1);&\text{otherwise}.\end{array}\right .
\end{aligned}\ ] ] we next prove some observations to limit the possibilities for @xmath263 and @xmath264 .
[ obs7 ] if @xmath265 ( resp . ,
@xmath266 , then we have @xmath170 ( resp . , @xmath168 ) .
* * proof:**if @xmath267 , there exists a @xmath1-cycle @xmath268 .
but if also @xmath265 , then @xmath269 and @xmath270 would be a 2-factor in @xmath271 , implying a 2-factor in @xmath11 .
the proof that @xmath272 implies @xmath168 is symmetric .
@xmath153 from observation [ obs6 ] , we have @xmath273 3-p,&\text{if $ r = p-2 $ , $ \ell=0$},\\[4pt ] \ell - r,&\text{otherwise}.\end{array}\right.\\[2mm]\ ] ] from this , we obtain [ obs8 ] @xmath274 . *
* proof:**suppose first @xmath275 .
if @xmath276 , then @xmath277 since @xmath278 by . but @xmath279 implies @xmath265 , and thus @xmath170 by observation [ obs7 ] , a contradiction .
suppose then @xmath280 .
then @xmath277 , since @xmath281 by .
since @xmath279 , we have the same contradiction as in the previous paragraph .
@xmath153 [ obs9 ] if @xmath282 , then @xmath283 .
* * proof:**else we have @xmath284 , and @xmath285 would be a hamiltonian cycle in @xmath286 . thus @xmath11 would have a 2-factor , a contradiction .
@xmath153 observations [ obs8 ] and [ obs9 ] together limit the possibilities for @xmath287 to @xmath288 and @xmath289 with @xmath290 .
we also can not have @xmath291 , since @xmath248 is always adjacent to @xmath292 , and so we would have @xmath258 in that case .
and we can not have @xmath293 , since then @xmath294 is odd , violating observation [ obs6 ] . to complete the proof of theorem [ thm:2-factor
] , we will deal with the remaining possibilities in a number of cases , and show that all of them lead to a contradiction of one or more of conditions ( i ) through ( iv ) . before doing so ,
let us define the spanning subgraph @xmath295 of @xmath11 by letting @xmath296 consist of the edges in the cycles @xmath128 , @xmath297 , or in the path @xmath18 , together with the edges incident to @xmath235 or @xmath248 .
note that the edges incident to @xmath235 or @xmath248 completely determine the large or small vertices in @xmath11 . in the proofs of the cases below
, any adjacency beyond those indicated would create an edge @xmath298 such that @xmath299 , and a fortiori @xmath11 , contains a .
case 2.1:@xmath300 .
since @xmath301 , we have @xmath214 , by observation [ obs7 ] . using observation [ obs6 ]
this means that @xmath302 , and hence @xmath6 is odd .
additionally , there are @xmath303 small vertices .
each of these small vertices has degree at most @xmath123 by observation [ obs2](a ) , and so @xmath0 is majorized by @xmath304 but @xmath305 ( a fortiori @xmath0 ) violates condition ( i ) .
@xmath153 case 2.2:@xmath306 . by observation [ obs6 ] , @xmath307 and @xmath308 , so that @xmath309
. also , there are @xmath310 large and @xmath311 small vertices . @xmath86by observation [ obs2](b , c ) , each vertex in @xmath178 ( resp . , @xmath180 ) has degree at most @xmath312 ( resp . , @xmath313 ) .
@xmath86each small vertex is adjacent to at most the large vertices ( otherwise @xmath11 contains a 2-factor ) , and so each small vertex has degree at most @xmath310 .
@xmath86the vertex @xmath203 ( resp . , @xmath292 ) is adjacent to at most the large vertices and @xmath235 ( resp .
, @xmath248 ) ( otherwise @xmath11 contains a 2-factor ) , and so @xmath203 , @xmath292 each have degree at most @xmath314 .
thus @xmath0 is majorized by @xmath315 setting @xmath316 , so that @xmath317 , the sequence @xmath318 becomes @xmath319 if @xmath320 , then since @xmath318 majorizes @xmath0 , we have @xmath210 , @xmath321 , @xmath322 , and @xmath323 , and @xmath0 violates condition ( iii ) . if @xmath324 , then @xmath6 is odd , and @xmath318 reduces to @xmath325 since @xmath326 majorizes @xmath0 , we have @xmath327 , and @xmath0 violates condition ( i ) . @xmath153 case 2.3:@xmath328 by observation [ obs7 ] , @xmath168 , and thus by observation [ obs6 ] , @xmath329 and @xmath330 .
also , there are @xmath331 large and @xmath332 small vertices . if @xmath333 then @xmath334 , a contradiction , and hence @xmath335 .
@xmath86by observation [ obs2](c ) , each vertex in @xmath180 has degree at most @xmath336 .
@xmath86each small vertex is adjacent to at most the large vertices , and so each small vertex has degree at most @xmath331 . @xmath86the vertex @xmath292 is adjacent to at most @xmath248 and the large vertices , and so @xmath292 has degree at most @xmath337 .
thus @xmath0 is majorized by @xmath338 setting @xmath339 , so that @xmath340 , @xmath341 becomes @xmath342 if @xmath343 , then since @xmath341 majorizes @xmath0 , we have @xmath210 , @xmath211 , @xmath322 , and @xmath344 , and @xmath0 violates condition ( iii ) . if @xmath345 , then @xmath6 is even , and @xmath341 reduces to @xmath346 since @xmath347 majorizes @xmath0 , we have @xmath348 and @xmath349 , and @xmath0 violates condition ( ii ) .
@xmath153 case 2.4:@xmath350 , _ where _
@xmath351 we have @xmath168 by observation [ obs7 ] , and @xmath352 by case 2.4 . by observation [ obs6 ] ,
@xmath353 and @xmath354
. moreover , there are @xmath355 large vertices including @xmath203 , and @xmath356 small vertices .
@xmath86by observation [ obs2](c ) , each vertex in @xmath180 has degree at most @xmath357 .
@xmath86each small vertex other than @xmath358 is adjacent to at most the large vertices except @xmath203 , and so each small vertex other than @xmath358 has degree at most @xmath356 .
@xmath86the vertex @xmath358 is not adjacent to @xmath248 , and so by observation [ obs2](a ) , @xmath358 has degree at most @xmath353 .
@xmath86the vertex @xmath292 is adjacent to at most @xmath248 and the large vertices except @xmath203 , and so @xmath292 has degree at most @xmath355 .
@xmath86each @xmath359 , @xmath360 , is adjacent to at most @xmath248 , the large vertices , the vertices in @xmath180 , and @xmath361 .
hence each such @xmath359 has degree at most @xmath362 .
@xmath86the vertex @xmath203 is adjacent to at most @xmath235 , @xmath248 , the other large vertices , the vertices in @xmath180 , and @xmath363 .
hence @xmath203 has degree at most @xmath364 .
@xmath86the vertex @xmath365 is not adjacent to @xmath235 , and so by observation [ obs2](a ) , vertex @xmath365 has degree at most @xmath354 .
thus @xmath0 is majorized by @xmath366 setting @xmath367 , so that @xmath368 , @xmath369 becomes @xmath370 since @xmath369 majorizes @xmath0 , we have @xmath371 , @xmath372 , @xmath373 , and @xmath374 , and thus @xmath0 violates condition ( iv ) .
@xmath153 case 2.5:@xmath375 we have @xmath168 , by observation [ obs7 ] . by observation [ obs6 ] , we then have @xmath376 and @xmath377 . if @xmath378 , then condition ( iii ) with @xmath379 implies @xmath380 , which means there are at least 2 vertices adjacent to all other vertices , a contradiction .
hence @xmath381 , and so @xmath382 by observation 1 .
finally , there are @xmath201 large vertices including @xmath203 , and @xmath166 small vertices . @xmath86by observation [ obs2](a ) , the vertices in @xmath180 have degree at most @xmath383 .
@xmath86by observation [ obs2](d ) , the small vertices in @xmath182 have degree at most @xmath384 .
@xmath86the vertex @xmath203 is not adjacent to the small vertices in @xmath182 , and so @xmath203 has degree at most @xmath385 .
@xmath86the vertices @xmath386 have degree at most @xmath383 by observation [ obs2](a ) , since none of them are adjacent to @xmath387 .
thus @xmath0 is majorized by @xmath388 setting @xmath389 , so that @xmath390 , @xmath391 becomes @xmath392 since @xmath391 majorizes @xmath0 , we have @xmath210 , @xmath211 , @xmath322 , and @xmath344 , and @xmath0 violates condition ( iii ) .
@xmath153 the proof of theorem [ thm:2-factor ] is complete .
the increase in complexity of theorem [ thm:2-factor ] ( @xmath4 ) compared to corollay [ thm:1-factor ] ( @xmath3 ) suggests that the best monotone condition for @xmath0 to be forcibly @xmath1-factor graphical may become unwieldy as @xmath1 increases . indeed , we make the following conjecture .
[ conj : superpoly ] the best monotone condition for a degree sequence @xmath0 of length @xmath6 to be forcibly @xmath1-factor graphical requires checking at least @xmath393 nonredundant conditions ( where each condition may require @xmath394 checks ) , where @xmath393 grows superpolynomially in @xmath1 .
kriesell @xcite has verified such rapidly increasing complexity for the best monotone condition for @xmath0 to be forcibly @xmath1-edge - connected .
indeed , kriesell has shown such a condition entails checking at least @xmath395 nonredundant conditions , where @xmath395 denotes the number of partitions of @xmath1 .
it is well - known @xcite that @xmath396 .
the above conjecture suggests the desirability of obtaining a monotone condition for @xmath0 to be forcibly @xmath1-factor graphical which does not require checking a superpolynomial number of conditions .
our goal in this section is to prove such a condition for @xmath28 .
since our condition will require tutte s factor theorem @xcite , we begin with some needed background .
belck @xcite and tutte @xcite characterized graphs @xmath11 that do not contain a @xmath1-factor . for disjoint subsets @xmath397 of @xmath398 , let @xmath399 .
we call a component @xmath295 of @xmath400 _ odd _ if @xmath401 is odd . the number of odd components of @xmath400 is denoted by @xmath402
. define @xmath403 we call any disjoint pair @xmath404 for which @xmath406 a _ @xmath1-tutte - pair _ for @xmath11 .
note that if @xmath408 is even , then @xmath397 is a @xmath1-tutte - pair for @xmath11 if and only if @xmath409 moreover , for all @xmath410 we have @xmath411 , so @xmath412 .
thus for each @xmath1-tutte - pair @xmath397 we have @xmath413 our main result in this section is the following condition for a graphical degree sequence @xmath0 to be forcibly @xmath1-factor graphical .
the condition will guarantee that no @xmath1-tutte - pair can exist , and is readily seen to be monotone .
we again set @xmath71 .
* proof*:let @xmath6 and @xmath428 be integers with @xmath408 even .
suppose @xmath0 satisfies ( i ) and ( ii ) in the theorem , but has a realization @xmath11 with no @xmath1-factor .
this means that @xmath11 has at least one @xmath1-tutte - pair . for ( b ) ,
clearly @xmath442 .
if @xmath443 for some @xmath435 , move @xmath125 to @xmath178 , and consider the change in each term in @xmath444 : @xmath445{0pt}{0pt}}_{\textrm{increases by $ k$ } } + \underbrace{{\textstyle\sum\limits_{u \in b } } d_{g - a}(u)\rule[-5mm]{0pt}{0pt}}_{\textrm{decreases by $ e(v , b ) \geq k$}}\!\!\!\!-\ ;
k|b| - \underbrace{odd_k(a , b)\rule[-5mm]{0pt}{0pt}}_{\textrm{decreases by $ \leq 1$}}\!.\ ] ] so by theorem [ thm:5.2](a ) , @xmath446 is also a @xmath1-tutte - pair in @xmath11 , contradicting the assumption that @xmath397 is a @xmath1-tutte - pair with @xmath178 as large as possible . and for ( c )
, suppose that @xmath447 for some @xmath448 .
this implies that @xmath449 .
now move @xmath450 to @xmath451 , and consider the change in each term in @xmath444 : @xmath452{0pt}{0pt}}_{\substack{\textrm{decreases by } \\[1pt ] d_{g - a}(t)\,\geq\ , k+odd_k(a , b ) - 1 } } \ ! -\ !
\underbrace{k|b|\rule[-5mm]{0pt}{0pt}}_{\textrm{decreases by $ k$ } } - \underbrace{odd_k(a , b)\rule[-5mm]{0pt}{0pt}}_{\textrm{decreases by $ \leq odd_k(a , b)$}}\!\!\!\!\!\!\!\!\!\!.\ ] ] so by theorem [ thm:5.2](a ) , @xmath453 is also a @xmath1-tutte - pair for @xmath11 , contradicting the minimality of @xmath397 . @xmath153 we introduce some further notation . set @xmath162 , @xmath454 , @xmath455 , @xmath456 , @xmath457 , and @xmath458 . using this notation , can be written as @xmath459 by lemma [ lem:5.4](a ) we have @xmath460 .
since @xmath180 is disjoint from @xmath178 , we trivially have @xmath417 . and
since the number of odd components of @xmath451 is at most the number of elements of @xmath451 , we are also guaranteed that @xmath461 . finally ,
since for all vertices @xmath125 we have @xmath462 , we get from that @xmath463 , hence @xmath464 .
it follows that @xmath415 satisfy the conditions in theorem [ thm : k - factor](ii ) .
if @xmath466 ( i.e. , if @xmath467 ) , let @xmath220 be the size of a largest component of @xmath400 . then , using lemma [ lem:5.4](b ) , for all @xmath468 we have @xmath469 clearly @xmath470 .
if @xmath471 , then @xmath472 , since @xmath451 has at least @xmath473 components .
thus @xmath474 . combining this
all gives @xmath475 next notice that we can not have @xmath476 , because otherwise @xmath477 and @xmath478 , and becomes @xmath479 , a contradiction . from and
we see that each of the @xmath480 vertices in @xmath481 has degree at most @xmath482 , and so @xmath483 .
if @xmath425 , then each of @xmath487 vertices in @xmath451 has degree at most @xmath488 by , and so @xmath426 .
this also holds if @xmath489 , since we set @xmath71 and @xmath490 since @xmath428 and @xmath461
. how good is theorem [ thm : k - factor ] ?
we know it is not best monotone for @xmath4 .
for example , the sequence @xmath491 satisfies theorem [ thm:2-factor ] , but not theorem [ thm : k - factor ] ( it violates @xmath422 when @xmath492 , @xmath493 and @xmath494 , with @xmath495 and @xmath496 ) . and
it is very unlikely the theorem is best monotone for any @xmath497 .
nevertheless , theorem [ thm : k - factor ] appears to be quite tight . in particular , we conjecture for each @xmath28 there exists a @xmath19 such that informally , for each @xmath28 , there exists a pair @xmath501 with @xmath14 ` just below ' @xmath0 such that theorem [ thm : k - factor ] detects that @xmath0 is forcibly @xmath1-factor graphical , while @xmath14 is not forcibly @xmath1-factor graphical . for example
, let @xmath502 and @xmath503 , and consider the sequences @xmath504 and @xmath505 .
it is easy to verify that the unique realization of @xmath506 fails to have a @xmath1-factor , for @xmath507 . on the other hand ,
we have programmed theorem [ thm : k - factor ] , and verified that @xmath508 satisfies theorem [ thm : k - factor ] with @xmath509 for all values of @xmath6 up to @xmath510 .
we conjecture that @xmath511 satisfies theorem [ thm : k - factor ] for all @xmath503 with @xmath512 .
there is another sense in which theorem [ thm : k - factor ] seems quite good .
a graph @xmath11 is _
@xmath450-tough _ if @xmath513 , for every @xmath514 with @xmath515 , where @xmath516 denotes the number of components of @xmath517 . in @xcite , the authors give the following best monotone condition for @xmath0 to be forcibly @xmath450-tough , for @xmath518 .
let @xmath19 be graphical , and let @xmath28 be an integer with @xmath523 and @xmath408 even .
if @xmath0 is forcibly @xmath1-tough graphical by theorem [ tough ] , then @xmath0 is forcibly @xmath1-factor graphical by theorem [ thm : k - factor ] . | we consider sufficient conditions for a degree sequence @xmath0 to be forcibly graphical .
we note that previous work on degrees and factors has focused primarily on finding conditions for a degree sequence to be potentially @xmath1-factor graphical .
we first give a theorem for @xmath0 to be forcibly 1-factor graphical and , more generally , forcibly graphical with deficiency at most @xmath2 .
these theorems are equal in strength to chvtal s well - known hamiltonian theorem , i.e. , the best monotone degree condition for hamiltonicity .
we then give an equally strong theorem for @xmath0 to be forcibly 2-factor graphical .
unfortunately , the number of nonredundant conditions that must be checked increases significantly in moving from @xmath3 to @xmath4 , and we conjecture that the number of nonredundant conditions in a best monotone theorem for a @xmath1-factor will increase superpolynomially in @xmath1 . this suggests the desirability of finding a theorem for @xmath0 to be forcibly graphical whose algorithmic complexity grows more slowly . in the final section , we present such a theorem for any @xmath5 , based on tutte s well - known factor theorem . while this theorem is not best monotone , we show that it is nevertheless tight in a precise way , and give examples illustrating this tightness .
* keywords * : @xmath1-factor of a graph , degree sequence , best monotone condition * ams subject classification * : 05c70 , 05c07 |
the asteroid and kuiper belts lie in regions of the solar system that are stable over long timescales to the gravitational perturbations of the planets ( lecar et al .
dust produced in collisions between planetesimals in these belts inhabits a broader distribution , but one that is strongly influenced by the same gravitational perturbations ( dermott et al . 1994 ; liou & zook 1999 ; moro - martn & malhotra 2003 ) .
these belts are thought to have started off with more than 200 times their present mass ( stern 1996 ; bottke et al . 2005 ) and it is speculated that the belts were depleted in the same event that caused the late heavy bombardment ( lhb ) .
the lhb was a period of @xmath4 myr duration that occurred 700 myr after the solar system formed when the terrestrial planets were bombarded with a large influx of comets and asteroids , possibly due to restructuring of the planetary system in a dynamical instability ( gomes et al .
the evolution of these planetesimal belts since this event is thought to have been relatively slow , with mass falling off due to collisional processing , but with large spikes in the dust content of the inner solar system occurring when two large asteroids collided ( nesvorn et al .
2003 ; farley et al .
knowledge of the planetesimal and dust content of the solar system and how it evolves thus provides a rich source of information about both the planetesimals themselves and the planetary system in which they reside .
it is important to determine the evolution of the planetesimal belts around other stars , since this may be equally illuminating as to the nature and evolution of their planetary systems .
surveys have been undertaken to determine this evolution , by measuring the infrared flux from dust produced in the collisional destruction of those planetesimals ( aumann et al .
the observations are used to derive basic properties of the dust belts , such their infrared luminosity @xmath5 ( quoted as fractional luminosity , @xmath6 ) , and the distance of the dust from the star ( which is either the same as , or strongly dependent on , the planetesimal belt radius @xmath7 ) , and these are then compared with stellar properties , such as age and spectral type .
most studies have focussed on the evolution of dust luminosity , showing that fractional luminosity decreases with age ; e.g. , spangler et al . (
2001 ) found the mean dust luminosity of disks of similar age falls off @xmath8 , while greaves & wyatt ( 2003 ) inferred from detection statistics that the mass falls off @xmath9 , najita & williams ( 2005 ) found the mass of the detected disks falls off @xmath10 , and the upper limit in 24 @xmath1 m of the a star disks was found to fall off @xmath10 ( rieke et al .
2005 ) , while disk mass from far - ir statistics was inferred to be relatively age independent in the range 30 - 1000 myr ( rhee et al .
these results are thus in general agreement with the results of theoretical models , which show that a planetesimal belt evolving in quasi - steady state would lose mass due to collisional grinding down giving a disk mass ( and dust luminosity ) that falls off @xmath10 ( dominik & decin 2003 ; wyatt et al .
2007 ) .
however , decin et al .
( 2003 ) noted that the maximum luminosity of debris disks remains constant at @xmath11 even for the oldest stars , and this was explained by dominik & decin ( 2003 ) as a consequence of delayed stirring , in which dust is not produced in the planetesimal belt until pluto - sized objects form and ignite a collisional cascade .
since planet formation models predict that such massive bodies take longer to form further from the star ( kenyon & bromley 2004 ) , that interpretation makes the prediction that the radius of the belts should increase with stellar age , and more recent surveys have also considered the evolution of radius . while in sub - mm surveys there is as yet no evidence of any evolution of radius with age ( najita & williams 2005 ; williams & andrews 2006 ) , the latest far - ir results for a stars do appear to show some evidence that radius is increasing with age , based on the fact that excess emission is detected at much later ages at 70 @xmath1 m than at 24 @xmath1 m
( su et al . 2006 ) .
it has also been suggested that much of the dust we see in some systems is produced episodically in collisions between large planetesimals , much in the same way dust in the zodiacal cloud increases when collisions occur in the asteroid belt , and that such collisions contribute to the large spread in the infrared luminosities seen in disks around a stars of similar age ( rieke et al . 2005 ) .
this scenario is supported by the discovery of a population of dust grains around vega in the process of removal by radiation pressure which indicates that this system can not have remained in steady - state for the full 350 myr age of the star ( su et al .
more recently it has also been shown that the few sun - like stars with hot dust at a few au ( gaidos 1999 ; beichman et al .
2005 ; song et al . 2005 ; wyatt et al . 2005 ) must be transient , since they are too bright for their age to be planetesimal belts that have evolved due to steady - state collisional processing ; such systems may be undergoing periods analogous to the lhb in the solar system ( wyatt et al . 2007 ) .
a stochastic element to disk evolution would introduce additional free parameters in models and would complicate interpretation of the observed statistics if it plays a major role .
it has thus not yet been possible to arrive at a consistent picture for debris disk evolution . the fact that surveys are undertaken for different spectral types , at different wavelengths , with different sensitivities to detections ,
has also added confusion to the interpretation of the statistics . in this paper
we revisit a simple analytical model for the steady - state collisional evolution of planetesimal belts which was originally explored in dominik & decin ( 2003 ) and recast with slightly modified assumptions in wyatt et al .
this model , including how it is applied in this paper to populations of debris disks , is given in
[ s : model ] . in [ s : rieke ] we build a model for the population of disks around a stars which is constrained using the 24 @xmath1 m statistics presented in rieke et al .
the properties of the model population are then compared in [ s : rsst05pred ] with those of a star disks detected in surveys at 70 @xmath1 m ( su et al .
2006 ) and the model parameters fine - tuned .
these results indicate that , with the exception of a few outlier systems , the available statistics can be fitted by steady - state collisional evolution without having to invoke a major role for stochasticity or delayed stirring .
finally the model population is used to make predictions for future debris disk surveys in the sub - mm ( [ s : pred ] ) .
the conclusions are given in
[ s : conc ] .
this section describes the simple analytical model for the steady - state collisional evolution of planetesimal belts that was derived in wyatt et al .
( 2007 ) ( [ ss : stst ] ) , and illustrates how that model is applied in this paper to populations of debris disks ( [ ss : pop ] ) .
the planetesimal belt is assumed to be in collisional equilibrium with a size distribution defined by : @xmath12 where @xmath13 is 11/6 for an infinite collisional cascade ( dohnanyi 1969 ) , and @xmath14 is the diameter of the planetesimals in km . that distribution is assumed to hold from the largest planetesimal in the disk , of diameter @xmath15 ( in km ) , down to the size below which particles are blown out by radiation pressure as soon as they are created , @xmath16 ( in @xmath1 m ) .
if @xmath13 is in the range 5/3 to 2 then most of the mass is in the largest planetesimals while the cross - sectional area is dominated by the smallest particles such that for spherical particles of density @xmath17 in kg m@xmath18 : @xmath19 where @xmath20 is in au@xmath21 and @xmath22 is in @xmath23 and the rest of the units are as described above and are used consistently throughout the paper . the planetesimal belt is assumed to be at a radius , @xmath7 , and have a width @xmath24 , both in units of au .
assuming that the grains act like black bodies , and so absorb all the radiation they intercept , the fractional luminosity , @xmath6 , of the dust emission is given by @xmath25 in other words , in this model @xmath20 , @xmath22 , and @xmath26 are all proportional to each other and just one is needed to define the scaling factor @xmath27 in equation ( [ eq : nd ] ) . assuming the particles act like black bodies also allows us to write down some other simple relations : @xmath28 where @xmath29 and @xmath30 are in solar units . also the emission from the disk at a wavelength @xmath31
is given by : @xmath32 where d is the distance to the star in pc , giving @xmath33 in jy if the planck function @xmath34 is in jy sr@xmath35 .
the dust temperature can be worked out from : @xmath36 in k. the emission from the star at the same wavelength is given by : @xmath37 where @xmath38 is in k. in a collisional cascade the amount of material within a given size range @xmath14 to @xmath39 decreases as these planetesimals are destroyed in collisions with other members of the cascade , but is replaced at the same rate by fragments created from the collisional destruction of larger objects .
the long - timescale evolution is thus determined by the removal of mass from the top end of the cascade . in this model
the scaling factor @xmath27 ( and so the total mass and fractional luminosity ) decreases as the number of planetesimals of size @xmath15 decreases .
the loss rate of such planetesimals is determined by their collisional lifetime , which was derived in wyatt et al .
( 2007 ) to be : @xmath40^{-0.5}}{(18 - 9q)g(q , x_{\rm{c } } ) } \right ) \label{eq : tcmtot}\ ] ] in myr , where @xmath41 and @xmath42 are the mean eccentricities and inclinations of the planetesimals orbits , and the factor @xmath43 is given by @xmath44 , \label{eq : qgxc}\end{aligned}\ ] ] and @xmath45 , where @xmath46 is the smallest planetesimal that has enough energy to catastrophically destroy a planetesimal of size @xmath15 .
the factor @xmath47 can be worked out from the dispersal threshold , @xmath48 , defined as the specific incident energy required to catastrophically destroy a particle such that ( wyatt & dent 2002 ) : @xmath49^{1/3 } , \label{eq : xc}\ ] ] where @xmath48 is in j kg@xmath35 . assuming that collisions are the only process affecting the evolution of the disk mass @xmath50 , then that mass ( or equivalently for @xmath27 , @xmath20 , or @xmath26 ) can be worked out by solving @xmath51 to give : @xmath52 , \label{eq : mtott}\ ] ] where @xmath53 is the initial disk mass and @xmath54 is the collisional lifetime at that initial epoch ; this solution is valid as long as mass is the only time variable property of the disk .
this results in a disk mass that is constant at @xmath53 for @xmath55 , but that falls off @xmath10 for @xmath56 . as noted in wyatt et al .
( 2007 ) , one property of this evolution is that , since the expression for @xmath54 includes a dependence on @xmath53 , the disk mass at late times is independent of initial disk mass .
this is because more massive disks process their mass faster .
this means that for any given age , @xmath57 , there is a maximum disk mass @xmath58 ( and also infrared luminosity , @xmath59 ) that can remain due to collisional processing : @xmath60^{-0.5}}{(18 - 9q)g(q , x_{\rm{c } } ) } \right ) , \label{eq : mmax1 } \\
f_{\rm{max } } & = & \left ( \frac{10^{-6}r^{1.5}(dr / r)}{4\pi m_\star^{0.5 } t_{\rm{age } } } \right ) \times \nonumber \\ & & \left ( \frac{2[1 + 1.25(e / i)^2]^{-0.5}}{g(q , x_{\rm{c } } ) } \right ) \left ( \frac{d_{\rm{bl}}}{d_{\rm{c } } } \right)^{5 - 3q } , \label{eq : fmax1}\end{aligned}\ ] ] where again this mass is in units of @xmath23 . to apply this evolution to populations of debris disks ,
the following assumptions were made .
all stars have a planetesimal belt that is undergoing a collisional cascade ( although this is not necessarily detectable ) .
the initial conditions of these belts are as described in
[ ss : stst ] using the parameters @xmath53 , @xmath7 , @xmath24 , @xmath17 , @xmath15 , @xmath16 ( which is defined by @xmath29 and @xmath30 ) and @xmath13 .
the subsequent evolution is determined by @xmath48 , @xmath41 and @xmath42 ( [ ss : stst ] ) . to simplify the problem we set
@xmath61 , @xmath62 kg m@xmath18 , @xmath63 and @xmath64 for all disks and did not consider the effects of changing these parameters .
the parameters @xmath48 , @xmath41 and @xmath15 were assumed to be the same for all disks , and their values were constrained by a fit to the observed properties of the known disks ; the consequence of choosing different values is discussed .
the remaining parameters , @xmath53 , @xmath7 , @xmath29 and @xmath30 were chosen from distributions for a population of a large number of disks ( 10,000 in this case ) .
for each disk the stellar spectral type was chosen at random from the appropriate range ( i.e. , b8-a9 for comparison with the sample of rieke et al .
2005 ) , thus defining @xmath29 , @xmath30 and so @xmath16 . the distribution of initial disk masses , @xmath53 , was based on the results of andrews & williams ( 2005 ) who did a sub - mm study of proto - planetary disks in the taurus - auriga star forming region , i.e. , for @xmath65 stars , from which they derived that this population has a log - normal distribution of dust masses that is centred on @xmath66 and has a @xmath67 width of 1.14 dex .
we used the same distribution but allowed @xmath68 to be different from that found in taurus - auriga , and this was a free parameter . for the distribution of disk radii ,
a power law distribution @xmath69 in the range 3 - 120 au was adopted , where @xmath70 is the number of disks with radii in the range @xmath7 to @xmath71 .
the range 3 - 120 au was chosen from the range of radii inferred for the sample of 46 a star disks with radius estimates ( table [ tab:2470sample ] ) , and the exponent @xmath72 was determined from a fit to the observations .
the mass evolution of each disk was then completely defined using equations ( [ eq : tcmtot ] ) and ( [ eq : mtott ] ) , and its properties at the current epoch were determined by assigning an age , @xmath57 , randomly from the appropriate range ( i.e. , 0 - 800 myr for a stars ) .
the stars were also assigned a random distance in the range 0 - 45 pc assuming an isotropic distribution ( i.e. , @xmath73 ) ; given that there are @xmath4 a stars within 45 pc , the model is overpopulated by a factor of around 100 compared with the real population ( although note that distance only becomes important in this paper in [ s : pred ] ) .
the free parameters in this population model are thus : @xmath68 , @xmath72 , @xmath15 , @xmath41 , and @xmath48 .
here the debris disk population model of [ s : model ] is applied to the statistics of the incidence of disks around a stars presented in rieke et al .
( 2005 ) based on observations at 24 @xmath1 m . there are two main aims .
the first , which is described in the rest of this section , is to determine if these statistics can be reproduced with steady - state evolution of the disks , and if so to set constraints on the physical properties and starting conditions of the disks in that population .
the second is to use the resulting model population to make predictions for the properties of a star disks found in surveys at other wavelengths and to determine if this simple model can explain the trends seen in the data .
this is described in [ s : rsst05pred ] and [ s : pred ] , where the model parameters are also fine - tuned .
rieke et al .
( 2005 ) searched a sample of 76 individual main sequence a stars and a number of young stellar clusters for excess 24 @xmath1 m emission above photospheric levels using spitzer and combined the data with archival results from iras for a total sample size of 266 a stars .
they also determined the ages of all stars in a consistent manner thus providing a very useful sample from which to study the evolution of debris around a stars .
their main result was a plot of @xmath74 versus age for all objects which showed a decrease in the upper envelope @xmath75 myr@xmath76 ( see fig .
[ fig : rsst05 ] ) .
the statistics were further quantified by splitting the sample into three age bins ( @xmath77 myr , @xmath78 myr and @xmath79 myr ) , and classifying sources as either having small excess ( @xmath80 ) , medium excess ( @xmath81 ) or large excess ( @xmath82 ) ( see their fig . 3 and our middle fig .
[ fig : rsst05 ] ) .
data on 12 objects was also presented at 70 @xmath1 m ( see their fig . 2 and our bottom fig . [
fig : rsst05 ] for a different version of this data ) . [ cols="^ " , ] bottom fig .
[ fig : scuba2 ] shows the histogram of the 850 @xmath1 m fluxes of the model population ( noting that this population should be representative of the scuba-2 survey sources , since all are within 45 pc ) .
the model predicts that 17 of the 100 stars in the scuba-2 a star sample should have a flux above the detection threshold of 2 mjy .
of those 17 : five would be expected above 10 mjy , and eight in the range 3 - 10 mjy , with the remaining four detected with @xmath83 confidence at 2 - 3 mjy .
the histogram also shows the sub - mm flux distribution for the 3 a stars already detected , showing that we already know of 2 stars within 45 pc with fluxes above 10 mjy ( hd172167 , hd216956 ) , while we know of just one in the range 3 - 10 mjy ( hd14055 ) .
thus the scuba-2 survey is expected to discover three new bright disks @xmath84 mjy , but can expect to find a further seven new disks at 3 - 10 mjy levels .
the @xmath84 mjy sources are particularly interesting , since at such a level of flux it is possible to resolve the structure of the disks ( holland et al .
1998 ; wyatt et al . 2005 ) , and indeed the model predicts that 60% of these will have an angular size of @xmath85 ( noting that all may be larger than this given that the black body assumption underestimates the true disk size by a factor of up to three , schneider et al .
it is also possible to extract some information on the spatial distribution of the 3 - 10mjy sources , e.g. , giving a direct measure of the size and inclination of the disk ( greaves et al .
2004 ) , with around 23% of these predicted to be larger than 6 .
it may even be possible to extract information from the 83 stars without detected disks , since they are predicted to have a mean flux of 0.60 mjy , of which 0.26 mjy is the mean disk flux and 0.34 mjy is the mean stellar flux .
by co - adding their data , the statistical noise level would be reduced to @xmath86 mjy , suggesting that it should be possible to get a positive detection of the photospheric emission , and also to discern if there is an excess coming from the disks with @xmath87 confidence . however , contamination from extragalactic sources that lie within the scuba-2 beam would have a signal of comparable magnitude to the mean disk flux and would reduce our ability to make a conclusive detection of disk emission from this population .
the model predicts that there will be little age dependence within the sample of disks detected by scuba-2 , in agreement with the result of rhee et al .
( 2007 ) who estimated the mass evolution based on far - ir observations .
for example , the model predicts that the mean age of the detected sample is @xmath88 myr ( compared with 400 myr for the whole population ) , with a roughly even number of disks ( 8 - 9 ) detected in both the @xmath89 myr and @xmath90 myr age bins , which will be indistinguishable from that expected if there were no age dependence ( 6 in the @xmath89 myr and 11 in the @xmath90 myr samples ) .
naturally scuba-2 is inclined to detect disks with low values of @xmath91 , i.e. , those for which their emission is more like black body ( which would be inferred to be composed of large grains ) .
however , given the small range in @xmath91 observed , this dependence is very slight and would affect mainly those disks detected close to the sensitivity threshold .
there is also predicted to be a small spectral type dependence , in that 19% of stars later than a3v are predicted to be detected by scuba-2 , compared with 15% of those earlier than a3v .
this arises because , while disks of lower fractional luminosity can be detected around earlier type stars , @xmath92 ( equation [ eq : fdet2 ] ) , the fractional luminosities of their dust belts are also much lower because of the higher dust blow - out radius , @xmath93 ( equation [ eq : fmax1 ] ) .
we do not anticipate that this would be detectable within the a star sample , however , it suggests that the detectability of sun - like stars could be high , should their disk population be described by the same parameters as for the a stars .
top figure [ fig : scuba2 ] also shows the 24 and 70 @xmath1 m detection limits of the spitzer survey of su et al . ( 2006 ) .
this shows that almost without exception ( 99.9% ) the disks that can be detected by scuba-2 at 850 @xmath1 m are also detectable by spitzer observations at 70 @xmath1 m with a detection threshold of @xmath94 , while most ( 61% ) of those detectable by scuba-2 are also detectable by spitzer at 24 @xmath1 m . in other words ,
ten of the 100 a stars in the scuba-2 sample should be detectable at 24 , 70 and 850 @xmath1 m , while seven should be detectable at 70 and 850 @xmath1 m , but not at 24 @xmath1 m ; the combination of sub - mm and far - ir data would be very important for constraining the dust temperature in these disks .
while this suggests that the survey could equally well be done at 70 @xmath1 m , there are three important points to consider . *
( i ) * first , not all of the sample has been observed with spitzer - the su et al .
( 2006 ) sample includes 42 sources within 45 pc ( of which 20 were detected at 70 @xmath1 m ) , which assuming even coverage across the sky means that 34 of these should fall in the scuba-2 survey ( which has declination limits of @xmath95 and @xmath96 ) .
this implies that two thirds of the scuba-2 a star sources have not previously been searched for disks in the far - ir ( although bright disks may have turned up in iras surveys which have a detection threshold well above that shown in fig . [ fig : scuba2 ] ) .
it is also not possible to scrutinize all stars with spitzer to the level of @xmath97 due to cirrus confusion .
thus it is predicted that the scuba-2 survey will turn up a significant number of genuinely new discoveries . *
( ii ) * second , it is important to emphasize the unbiased nature of the scuba-2 survey .
several sources have recently been identified as possessing excess emission in the sub - mm even when no excess is present in far - ir surveys ( wyatt et al . 2003 ; holmes et al . 2003 ; najita & williams 2005 ) .
these disks must be cool ( @xmath98k ) and at large radius ; e.g. , extrapolation of the detection thresholds in fig .
[ fig : scuba2 ] to larger radii indicates that the sub - mm is more sensitive than 70 @xmath1 m surveys to disks that are @xmath99 au at 5pc for a0v stars and @xmath100 au at 5 pc for a9v stars .
since the model population was constrained by 24 and 70 @xmath1 m surveys , there was no need to invoke a population of disks of large radius , although we know that dust exists around young a stars to hundreds of au ( kalas & jewitt 1995 ; clampin et al .
thus the predicted detection statistics should be viewed as a lower limit , with many large disks being detected at 850 @xmath1 m which can not be detected in the far - ir .
previous unbiased sub - mm surveys suggest that @xmath101% of stars could have disks too cold to detect in the far - ir , which could push the detection rate to @xmath102% for the scuba-2 a star survey . *
( iii ) * even discounting the ultra - cool disks discussed in ( ii ) , the sub - mm statistics will set important constraints on the distribution of the planetesimal belt radii , since the predictions made here are very sensitive to that distribution .
for example , with a radius distribution @xmath103 , which is not ruled out in the present study at the @xmath67 confidence level , the model would predict that 22 of the 100 a stars would be detected at @xmath104 mjy of which 7 would be @xmath84 mjy .
a simple model for the steady - state evolution of dust luminosity for planetesimal belts evolving due to collisions was described in [ s : model ] .
this section also described how the model could be applied to determine the properties of the disks of a population of stars , given that these disks would have had a range of initial properties .
this was applied in [ s : rieke ] to the population of a stars that was searched for evidence of dust emission at 24 @xmath1 m by rieke et al .
( 2005 ) , and it was shown that their detection statistics as a function of age could be reproduced with a model population in which the largest planetesimals have a size of @xmath105 km , and all planetesimals have a strength of @xmath106 j kg@xmath35 and an eccentricity of @xmath107
. however , these parameters should not be over - interpreted , since the observable properties of the model population remain the same as long as the combination @xmath108 is unchanged .
thus more detailed models of the collisional evolution of planetesimal belts ( e.g. , krivov , sremcevi & spahn 2005 ) are needed to interpret these parameters , e.g. , in terms of the composition of the planetesimal belts .
for now we note that these are reasonable parameters based on planetesimal ( benz & asphaug 1999 ) and planet formation models ( kenyon & bromley 2002 ) . the model used as input a distribution of planetesimal belt starting masses that is log - normal with the same width ( @xmath67 of 1.14 dex ) as inferred for proto - planetary disks ( andrews & williams 2005 ) , and found such
a distribution would have to be centred on @xmath109 .
this mass is consistent with that expected for proto - planetary disks around a stars , although the observable properties of the model population are also reproduced as long as @xmath110 is unchanged . in other words , the observable 24 @xmath1 m properties of main sequence
a star disks can be explained by the steady - state evolution of a disk population with realistic starting parameters ( although more extreme solutions are also possible ) . in [ s : rsst05pred ] this model
was tested against its predictions for the 70 @xmath1 m properties of the disks in this population .
these were found to reproduce the statistics of the su et al .
( 2006 ) survey which showed a much longer decay timescale for the 70 @xmath1 m excess than for that at 24 @xmath1 m .
the model was also used to predict the properties of the sub - sample of disks that could be detected at both 24 and 70 @xmath1 m , and this was compared with a sample of 46 a stars compiled from the literature for which excess emission has been detected at both 24 and 70 @xmath1 m .
the model reproduces and explains the distribution of the observed disks on the @xmath26 vs @xmath7 ( fig .
[ fig : fvsr ] ) , @xmath26 vs @xmath57 ( fig .
[ fig : fvst ] ) and @xmath7 vs @xmath57 ( fig .
[ fig : rvst ] ) plots , including an upper envelope in the @xmath26 vs @xmath7 plot that increases @xmath111 , a decay in the mean luminosity of disks with age @xmath112 , and a mean radius of detected disks that , if anything , increases with age @xmath113 .
thus it appears that the wide range in typical disk properties can be explained without appealing to stochasticity .
the large spread in 24 @xmath1 m fractional excess at any age ( rieke et al . 2005 ) can occur naturally from a spread in initial disk masses and radii , and the slower fall - off at 70 @xmath1 m and more detailed distributions of the properties of disks detected at 24 and 70 @xmath1 m ( su et al .
2006 ) could be the consequence of the detection bias .
this bias is best explained on the plot of @xmath26 vs @xmath7 , on which it is possible to plot the lines above which disks must lie to be detected at 24 and 70 @xmath1 m , and below which they must lie for a given age .
one illustrative example of this bias is the fact that while in the model the planetesimal belt of any individual star does not change in radius , the mean radius of detected planetesimal belts is predicted to increase with age .
the reason is that planetesimal belts that are closest to the star are processed faster and so fall below the detection threshold much faster than planetesimal belts that are farther from the star .
thus if such a trend is seen it would not necessarily be evidence for delayed stirring , in which the collisional cascade is only initiated once enough time has elapsed for pluto - sized objects to form , a process which takes longer further from the star ( e.g. , kenyon & bromley 2002 ; dominik & decin 2003 ) .
rather it could be evidence that the processing of the planetesimal belt is occurring inside - out , as inferred by su et al .
( 2006 ) .
this does not rule out the possibility that some , or even all , a star debris disks in the rieke et al .
( 2005 ) and su et al .
( 2006 ) samples are undergoing transient events on top of the overall trend established by the model .
in fact , there are a few disks that may ( rather than must ) require this mechanism to explain their properties .
the potentially transient systems identified in this paper are : hd3003 , hd38678 , and hd172555 , which have high luminosity for their age ; as well as hd115892 , which has unusually an low radius for its age .
all of these systems are identifiable by their high value of @xmath2 , where @xmath59 is the maximum possible fractional luminosity that a disk can have given its age and radius ( assuming the planetesimal belt properties inferred for the rest of the population ) .
we also found that the disks of hd2262 and hd106591 have unusually low radii for their age , and suggested that this may be attributable to the action of p - r drag on dust produced in their dust belts , since this starts to become important when @xmath114 , and these belts ( along with hd115892 and hd19356 ) can be identified from the sample as having @xmath115 .
neither do the results in this paper imply that the collisional cascade in all a star planetesimal belts must have been initiated soon ( @xmath116 myr ) after the stars formed , rather than the collisional cascade being initiated after a delay of several 10s or 100s of myr in some systems ( i.e. , the delayed stirring model ) . indeed ,
if the lack of large radii disks at young ages is confirmed ( rhee et al . 2007 ) , this would support some role for delayed stirring in debris disk evolution .
analysis of the distribution of @xmath2 of the 46 a star disks detected at both 24 and 70 @xmath1 m shows a distribution that may be attributed to the planetesimal belts not having exactly the same properties , but having a distribution in the combination of parameters given by @xmath108 ( which is also the combination in the parameter @xmath59 ) that is log - normal with a @xmath67 width of 1 dex .
this means two things : first , that the distribution of parameters of these a stars is quite uniform .
this is perhaps to be expected if the planetesimals in all belts are of similar composition , and grow to a similar maximum size before perturbations from the largest member of the cascade ( around 2000 km ) stir the rest of the planetesimal belt resulting in eccentricities of similar magnitude ( e.g. , kenyon & bromley 2002 ) .
second , that the anomalous systems identified in the last paragraph are only @xmath117 anomalies in that they may be explained as disks with higher than average planetesimal strength or size , or lower than average eccentricity . in contrast , this analysis strengthens the result of wyatt et al .
( 2007 ) , in which the dust emission from several sun - like stars , such as hd69830 , was inferred to be transient , since if the planetesimal belts of sun - like stars are described by similar properties to those of a stars , then all of the disks they inferred to be transient have @xmath118 and each one would be a @xmath119 anomaly ( and yet such disks are found around @xmath120% of stars ) . however , given current limits in definitive detection of transient events , the derived incidence is probably a lower limit .
one particularly interesting outcome of the model is the distribution of radii of the planetesimal belts .
the model assumed a power law distribution of radii in the range 3 - 120 au , and found from the distribution of radii of the sample detected at 24 and 70 @xmath1 m , that this follows @xmath3 .
it was found to be important to take the detection bias into account when interpreting the observed distribution .
the origin of this distribution is not considered in this model , which assumes that all stars have relatively narrow planetesimal belts with @xmath121 .
the inner hole in these belts has been inferred in other studies to be caused by the presence of inner planets ( e.g. , roques et al . 1994 ; wyatt et al . 1999 ; wilner et al .
2002 ; wyatt 2003 ; quillen 2006 ) . if these are kuiper belt analogues ( wyatt et al .
2003 ) , then this radius distribution would be indicative of the orbital radius of the outermost planet in its planetary system . however , these could also be asteroid belt analogues , i.e. , planetesimal belts in the midst of a planetary system .
if so the interpretation of this radius distribution is less clear , and could , like the solar system for example , be indicative of the orbital radius of the most massive giant planet in these systems . in any case , this distribution provides an important and unique constraint on the outcome of planet formation models .
another particularly powerful application of the model is that , regardless of the physical reason , it does explain the far - ir properties of a star disks and so it can also be used to make predictions for what surveys at other wavelengths ( or with different detection thresholds ) will see .
such predictions are made much easier when the survey detection threshold is uniform , or at least can be readily characterized , and this study emphasizes the importance of a characterizable threshold for future surveys .
one such survey is the scuba-2 debris disks survey ( matthews et al .
2007 ) , which will include an unbiased sub - mm survey of the 100 nearest a stars to a uniform @xmath122 sensitivity of 2 mjy at 850 @xmath1 m .
the model predicts that 17 of the 100 will be detected above 2 mjy , including 5 above 10 mjy , and that 5 of the detected disks should be resolvable in sub - mm imaging on size scales @xmath85 .
however , the detected fraction could be higher than this , since it depends on the radius distribution , which these observations would set strong constraints on .
it is also unknown how many a stars possess disks too cold to detect in the far - ir ( wyatt , dent & greaves 2003 ; najita & williams 2005 ) .
this suggests that the scuba-2 survey will be particularly fruitful , and the model presented here provides a framework which can be used to interpret the results of this survey , and those of future surveys . | in this paper a simple analytical model for the steady - state evolution of debris disks due to collisions is confronted with spitzer observations of dust around main sequence a stars .
all stars are assumed to have planetesimal belts with a distribution of initial masses and radii . in the model disk mass
is constant until the largest planetesimals reach collisional equilibrium whereupon the mass falls off @xmath0 . using parameters that are reasonable within the context of planet formation models and observations of proto - planetary disks , the detection statistics and trends seen at both 24 and 70 @xmath1
m can be fitted well by the model . while there is no need to invoke stochastic evolution or delayed stirring to explain the detection statistics of dust around a stars , the model is also consistent with a moderate rate of stochastic events .
potentially anomalous systems are identified by their high ratio of observed dust luminosity to the maximum permissible in the model given their radii and ages , @xmath2 ; these are hd3003 , hd38678 , hd115892 , and hd172555 .
it is not clear if their planetesimals have unusual properties ( e.g. , high strength or low eccentricity ) , or if their dust is a transient phenomenon .
there are also well - studied examples from the literature where transient phenomena are favored ( e.g. , vega , hd69830 ) .
however , the overall success of our model , which assumes planetesimals in all belts have the same strength , eccentricity and maximum size , suggests there is a large degree of uniformity in the outcome of planet formation .
the distribution of the radii of the planetesimal belts , once corrected for detection bias , is found to follow @xmath3 in the range 3 - 120 au . since the inner edge of a belt
is often attributed to an unseen planet , this provides a unique constraint on the planetary systems of a stars .
it is also shown that the effect of p - r drag on the inner edge of a star disks may need to be considered for those close to the spitzer detection threshold , such as hd2262 , hd19356 , hd106591 , and hd115892 .
predictions are made for the upcoming scuba-2 survey , including that at least 17 of the 100 a stars should be detectable above 2 mjy at 850 @xmath1 m , illustrating how this model can be readily applied to the interpretation of future surveys . |
knowledge of the fourier transform of a wave - function @xmath2 , i.e. _
0()_i e^i _ i(_i ) , [ intensitys ] is sufficient in order to obtain @xmath2 .
however , if only the absolute value @xmath10 is known , the information on the phase is lost and the inverse fourier transform can not be applied .
phase retrieval algorithms allows one to recover the phase provided additional constrains on a wave - function can be imposed .
assume that the support @xmath5 of @xmath2 is known and it is finite , that is usually the case for trapped atomic gases ( @xmath5 should be chosen so that the probability of the detection of an atom outside @xmath5 is negligibly small ) .
then , apart from the fact that the sought wave - function belongs to a set of all functions with the same modulus of its fourier transform @xmath6 , it also belongs to a set of functions with the same support @xmath5 .
the phase retrieval algorithms seek for the intersection of these two sets , where @xmath2 is located @xcite . in the error reduction ( er ) algorithm successive projection of an estimate of a wave - function onto the sets allows one to find the desired solution @xcite .
that is , starting with , e.g. , randomly chosen phases @xmath74 of @xmath75 , the following iteration is performed ^(n+1)=p_sp_m ^(n ) , where p_s ^(n)()= \ { . , and @xmath76 the symbol @xmath77 denotes the inverse fourier transform operator .
ideally , the iteration of the er algorithm should converge to @xmath41 for which the quantity q = d^2k||(k)|^2-|_0(k)|^2| , [ q ] equals zero .
however , it is known that the er algorithm tends to stuck at a local minimum of @xmath78 and the hybrid input - output ( hio ) algorithm , where ^(n+1)()= \ { . , is more efficient @xcite .
the parameter @xmath79 is usually chosen between @xmath80 and @xmath81 . in practice the best convergence
is achieved when after every 20 iterations of the hio algorithm there is an iteration of the er one , see @xcite .
it is possible to construct simple homometric examples of @xmath50 that correspond to different non - equivalent @xmath75 ( i.e. wave - functions which are not related to each other by e.g. a change of a global phase or origin shift ) .
however , for more complex structures the non - uniqueness is very unlikely in 2d and higher dimensions and does not occur in practise , see discussion in @xcite .
for very low temperatures , density fluctuations are strongly suppressed in trapped atoms with repulsive interactions @xcite .
then , the modulus of a wave - function @xmath13 can be quite accurately estimated if parameters of an external trapping potential and number of atoms are known .
this information can be used in the phase retrieval algorithms .
indeed , the support projection operator @xmath82 in the er algorithm can be substituted with the projection p_|| ^(n)= |()| e^i ^(n ) ( ) , where ^(n)()=^(n ) ( ) . such a modified er procedure is applied every 20 iterations of the hio algorithm .
if in an experiment atomic densities are measured after very long time - of - flight , the relation between an initial wave - function in the presence of an optical lattice and a measured image reduces to the fourier transform eq .
( [ intensitys ] ) , where @xmath83 .
however , if @xmath1 is not very long , the near - field corrections have to be taken into account , i.e. |_0()|^2 |_i e^i _ i(_i ) e^-i^2_i |^2 , where @xmath19 @xcite .
the presence of the phase factor @xmath20 can be easily included in the phase retrieval algorithm . indeed , after the inverse fourier transform eq .
( [ invfft ] ) is applied , one has to perform an additional transformation @xmath21 only .
the phase retrieval algorithm may stuck at local minimum .
therefore , in order to find the desired solution , the algorithm should be initiated many times with different , randomly chosen , phases @xmath74 . in the present publication , the algorithm has been initiated 30 - 100 times .
we have observed that in the near - field regime more extensive sampling has to be used than in the far - field limit . | we show that the phase of a bose - einstein condensate wave - function of ultra - cold atoms in an optical lattice potential in two - dimensions can be detected .
the time - of - flight images , obtained in a free expansion of initially trapped atoms , are related to the initial distribution of atomic momenta but the information on the phase is lost .
however , the initial atomic cloud is bounded and this information , in addition to the time - of - flight images , is sufficient in order to employ the phase retrieval algorithms .
we analyze the phase retrieval methods for model wave - functions in a case of a bose - einstein condensate in a triangular optical lattice in the presence of artificial gauge fields .
the development of experimental techniques in ultra - cold atomic gases is tremendous @xcite . both trapping potentials for atoms and mutual atom interactions
can be controlled and engineered . the artificial magnetic fields or even the non - abelian gauge potentials that are experienced by the neutral particles
can be created in a laboratory @xcite .
similar rapid progress is observed in detection methods .
nowadays , distributions of atoms can be measured even with a single atom resolution @xcite .
a typical bose - einstein condensate ( bec ) of ultra - cold atoms can be described by a real condensate wave - function ( order parameter ) and the appearance of a non - uniform phase of the condensate reflects thermal fluctuations only @xcite . however , in the presence of artificial gauge potentials or in multi - orbital superfluid phase of ultra - cold atoms in optical lattices , the order parameter may become intrinsically complex @xcite . in these cases , extracting information on the phase of the condensate would be invaluable . in particular , one would be able to reconstruct berry s phases related to synthetic abelian gauge fields .
for example , the fully frustrated triangular lattice problem reveals a doubly degenerate ground state which corresponds to two distinct configurations on the @xmath0flux lattice @xcite .
in such a system there is a theoretical possibility of spatial domains formation where different parts of the system are in two different groundstate configurations .
information on the condensate phase would allow experimentalists to reveal the domains ( for experiments with ferromagnetic domain formation see @xcite ) .
atomic density measurement after time - of - flight ( tof ) is a standard detection technique in ultra - cold atoms expierments @xcite . for sufficiently long tof
the density images reflect the initial distributions of atomic momenta .
the distributions alone are not sufficient to invert the fourier transform and obtain the condensate wave - function in the configuration space as the phase information is lost .
however , we will show that the phase of the condensate wave - function can be reconstructed if a system is initially bounded , which is always the case for trapped atoms . in crystallography ,
electron microscopy and astronomical imaging , computationally retrieving the phase of diffraction patterns is remarkably successful @xcite .
the examples range from the biological cells imaging to the evaluation of the aberrations in the hubble space telescope @xcite . on the other hand , in the present letter
it is not an external wave that diffracts on a measured object and is subsequently detected , but the matter wave itself is the _ object _ to be reconstructed . in the following we concentrate on atomic gases prepared in an optical lattice potential @xcite .
after a sudden turn off of the potential and a period of nearly free expansion , the atomic density replicates the initial momentum distribution of the system . in the far field regime ,
i.e. for a very long time - of - flight @xmath1 , the atomic density detected by a ccd camera reads i ( ) |_0()|^2|_i e^i _ i(_i)|^2 , = , [ intensity ] where @xmath2 is the initial condensate wave - function , @xmath3 s denote positions of the lattice sites and @xmath4 is the atomic mass .
multidimensional system of equations ( [ intensity ] ) , quadratic in @xmath2 , is difficult to solve .
phase retrieval algorithms try to find the condensate wave - function iteratively by imposing some extra conditions such as a finite support @xmath5 of @xmath2 @xcite .
the momentum distribution , i.e. the squared modulus of the fourier transform of the object @xmath2 , is known .
consequently , the sought wave - function belongs to a set of all objects with the same modulus of its fourier transform @xmath6 .
it also belongs to a set of objects with the same support @xmath5 .
phase retrieval algorithms seek for the intersection of these two sets , where the solution is located . in a basic version of the algorithm , starting with a randomly chosen @xmath7 that is localized on @xmath5 ,
the following procedure is applied : ( i ) the fourier transform is used resulting in @xmath8 , ( ii ) @xmath9 is substituted with @xmath10 , eq .
( [ intensity ] ) , and the inverse fourier transform is applied , ( iii ) a new @xmath11 is obtained by setting zeros in the result of the inverse transform for @xmath12 and the entire procedure is repeated until the wave - function converges .
if additional information , e.g. the modulus @xmath13 , is available , it can be used in the phase retrieval algorithms , see @xcite for all technical details . in the presence of a trap , even if the atomic density can not be measured in situ , it can be quite accurately estimated . in the case of an optical lattice potential
combined with a shallow trap , the modulus of the condensate wave - function corresponding to the ground state of repulsively interacting atoms reads |()| |_tf()|_i |w_0(-_i)| , [ trapden ] where @xmath14 is the thomas - fermi ( tf ) envelope and @xmath15 is the lowest band wannier function localized at @xmath3 @xcite . at low temperatures ,
the density fluctuations are suppressed due to the repulsive particle interactions @xcite and the density profile is identical to the ground state case , eq .
( [ trapden ] ) .
this additional information can be used in the phase retrieval algorithms .
indeed , the projection on the support @xmath5 can be substituted with the projection on the modulus of the wave - function .
that is , in the item ( iii ) described in the previous paragraph , at each iteration of the algorithm in order to get @xmath16 the modulus of the function obtained in the inverse fourier transform is substituted with eq .
( [ trapden ] ) @xcite .
so far we have assumed that images of atomic densities obtained after tof correspond to the far field regime .
this regime is rarely reached in an experiment where typically @xmath1 is about 20 - 30 ms . in a case of the finite expansion time of the atomic cloud released from an optical lattice potential , the density ( [ intensity ] )
has to be modified @xcite , i.e. @xmath17 where @xmath18 is the fourier transform of the wannier function and @xmath19 . a new phase factor @xmath20 accounts for the deformation and widening of bragg peaks and
can be easily included in our phase retrieval algorithm .
indeed , at each iteration of the algorithm after the inverse fourier transform is applied , one has to perform an additional transformation @xmath21 only cite supplement .
the initial stage of atomic gas expansion is influenced also by the atomic interactions .
this effect is weaker than the near field corrections @xcite , nevertheless , it can be taken into account by a suitable choice the wannier functions so that the envelope @xmath22 of the image is properly reproduced . to summarize , an experimentally measured image after the tof , theoretically estimated density @xmath23 and its support @xmath5 are information used in our phase retrieval algorithm which allows us to find unknown phases @xmath24 and consequently obtain the entire wave - function @xmath25 .
in ref .
@xcite we describe in detail how the information about @xmath23 and its support @xmath5 is used in order to obtain the results presented in the following .
let us apply the described phase microscopy to a two - dimensional ( 2d ) problem of a bec in a triangular optical lattice which can be described by the bose - hubbard hamiltonian @xcite .
primitive vectors of the bravais lattice read @xmath26 and @xmath27 where @xmath28 is the lattice constant .
the presence of a shallow harmonic trap is also assumed .
the periodic shaking of the lattice allows for the modification of tunneling amplitudes of the bose - hubbard model @xcite .
the modification of the kinetic part of the hamiltonian indicates the presence of a gauge vector potential @xcite .
especially for the negative tunneling amplitudes we deal with staggered magnetic fluxes .
the corresponding dispersion relation reveals two degenerate non - equivalent minima at @xmath29 where @xmath30 .
the degenerated ground states of the system can be written as @xmath31 where we approximate the wannier functions by 2d gaussian distributions with the standard deviation @xmath32 .
the non - zero quasi - momentum vectors of the ground states imply that the order parameter is complex .
assume that atomic gas , prepared in the ground state @xmath33 , performs the tof expansion and its density is measured in the far field regime , i.e. @xmath34 .
figure [ overlaps ] illustrates the influence of experimental imperfections on the accuracy of the phase microscopy . after adding gaussian white noise to the _ detected _ image , the overlap of the retrieved wave - functions with the exact ground state is diminished ( the overlaps are calculated between wave - functions which are firstly projected on the wannier basis vectors ) .
fig .
[ overlaps]a shows how the overlap changes with signal to noise ratio @xmath35 , where @xmath36 is the average value of @xmath37 , calculated on a region where @xmath38 , and @xmath39 is the standard deviation of the noise distribution .
each point of the plot has been obtained by running the phase retrieval algorithm for 30 sets of initially random phases @xmath40 and then the best result is chosen ( i.e. the result @xmath41 with minimal value of @xmath42 ) .
we see that even substantial amount of noise allows one to retrieve the wave - function with the reasonable overlap . in fig .
[ overlaps]b we present an example of a noisy image corresponding to @xmath43 which leads to the squared overlap of 0.83 .
the finite resolution of an imaging system is another important source of distortion of experimental density images . to analyze its impact on the phase microscopy we have convoluted the previously considered model image with the gaussian distribution .
standard deviation @xmath44 of the distribution is a measure of the resolution . in fig .
[ overlaps]c we show the squared overlap of the retrieved wave - functions with the exact state versus @xmath44 .
the overlap drops quite quickly which seems to be not very surprising because broadening of the bragg peaks reduces artificially the coherence length of the system .
the phase microscopy can be applied even if experimental images do not correspond to the far - field regime because the important additional phase factor @xmath45 can be easily included in the algorithm @xcite .
however , the presence of this phase factor makes the basin of attraction of a desired solution smaller .
therefore , the algorithm has to be run for greater number of randomly chosen phases @xmath40 in order to find the desired wave - function .
the performed analysis of experimental imperfections allows us to estimate requirements that have to be fulfilled in order to apply the phase microscopy .
assume that experimental images are obtained in the far - field limit . according to fig .
[ overlaps ] , the resolution of an imaging system has to fulfill @xmath46 , where @xmath47 , so that the retrieved wave - function has the squared overlap with the exact solution greater than 0.8 . in the near - field regime
the bragg peaks can be much wider than @xmath48 , but then they may possess small additional structures superimposed on the bell profiles which must not be blurred by an imaging system .
therefore , the requirement for the resolution should not be weakened in the near - field regime .
the presence of noise additionally worsens accuracy of the retrieved wave - function .
however , the discrimination of the noisy image with the threshold of @xmath49 ( i.e. @xmath50 is set 0 if it is smaller than @xmath51 ) can help significantly .
for example , we have checked that for @xmath52 , if we apply the discrimination in the noisy image ( @xmath53 ) , the overlap is smaller by 0.1 only as compared to the case without the noise .
the phase microscopy is very useful in detection and visualization of spatial domains . in order to demonstrate it
we have prepared a wave - function which consists of 5 spatially separated regions where either @xmath54 or @xmath55 is present . to visualize domains corresponding to the @xmath54 and
@xmath55 states let us define quantities @xmath56 . in the egions dominated by @xmath54 (
let us call them regions a ) we should get , e.g. , @xmath57 where @xmath58 is a constant global phase . however , in the regions where the state @xmath55 exists ( regions b ) , @xmath59 where @xmath60 is a constant phase .
let us now average @xmath61 over a small volume , @xmath62 where @xmath63 denotes the sum over @xmath64 and the nearest neighbors of the @xmath65-site ( one may consider also averaging extended to next to the nearest neighbors or even further ) . in regions
a we expect @xmath66 and @xmath67 while in regions b @xmath68 and @xmath69 . in fig .
[ modeldomain]a we show @xmath70 calculated for the exact model state whose momentum distribution is presented in fig .
[ modeldomain]d .
five spatially separated domains of different sizes are clearly visible in fig .
[ modeldomain]a .
if we add substantial noise ( @xmath71 ) to the momentum distribution , see fig .
[ modeldomain]e , the phase retrieval algorithm results in the domain picture depicted in fig .
[ modeldomain]b .
the noisy image has been discriminated with the threshold @xmath72 , before the algorithm is used . when the original momentum distribution is convoluted with a gaussian distribution characterized by the standard deviation @xmath73 ,
see fig .
[ modeldomain]f , the phase microscopy leads to the domain picture shown in fig .
[ modeldomain]c . comparing figs .
[ modeldomain]a - c one can see that despite the considerable imperfections assumed in the initial data , the largest domains are quite well reproduced by the phase retrieval algorithm .
small domains may either glue together ( fig .
[ modeldomain]b ) or disappear ( fig .
[ modeldomain]c ) especially if they are localized on the border where the probability density is very small .
in order to estimate the areas occupied by the different kinds of the domains we have calculated the squared overlap of the exact model wave - function with the ground states @xmath54 and @xmath55 which turn out to be 0.35 and 0.16 , respectively . in a case of the noisy image the corresponding overlaps are 0.5 and 0.26 while for the convoluted image we obtain 0.28 and 0.11 , respectively .
these numbers illustrate the accuracy of the phase microscopy in determination of domain sizes in the case when substantial experimental imperfections are present . to conclude
, we have demonstrated that the phase of a bose - einstein condensate wave - function of ultra - cold atomic gases in optical lattice potentials in 2d can be detected .
knowledge of an atomic density after time - of - flight expansion and a theoretical estimate for the initial atomic density in the presence of an external potential are used in phase retrieval algorithms .
we analyze influence of experimental imperfections on such a phase microscopy .
it seems that the most important requirement is the sufficiently high resolution of an imaging system .
the phase microscopy is particularly useful in situations when the order parameter of a system is complex , for example in the presence of artificial gauge potentials @xcite or in multi - orbital superfluid phase in optical lattices @xcite .
additionally , the phase microscopy can be applied to observe long range phase fluctuations . in the present letter we concentrate on a bec in a triangular optical lattice subjected to artificial gauge fields .
the phase microscopy allows for visualization of spatial domains whose presence is discussed in the literature @xcite .
recently in ref .
@xcite a different method for visualization of weiss domains has been proposed . in this method , however , phase relation between domains is lost . on the contrary ,
the method presented in this letter allows one to retrieve all phase informations .
we are grateful to : j. struck , m. weinberg , c. lschlger , p. windpassinger , j. simonet , and k. sengstock for fruitful discussions and useful comments during preparation of the manuscript .
we acknowledge support of polish national science center via project dec-2012/04/a / st2/00088 .
ak acknowledges support in a form of a special scholarship of marian smoluchowski scientic consortium `` matter energy future '' from know funding
. 99 i. bloch , j. dalibard , and w. zwerger , rev .
mod .
phys . * 80 * , 885 ( 2008 ) .
v. galitski , and i. b. spielman , nature * 494 * , 49 ( 2013 ) .
w. s. bakr , j. i. gillen , a. peng , s. flling , and m. greiner , nature * 462 * , 74 ( 2009 ) .
w. s. bakr , a. peng , m. e. tai , r. ma , j. simon , j. i. gillen , s. flling , l. pollet , and m. greiner , science * 329 * , 547 ( 2010 ) .
j. f. sherson , c. weitenberg , m. endres , m. cheneau , i. bloch , s. kuhr , nature * 467 * , 68 ( 2010 ) .
s. dettmer , d. hellweg , p. ryytty , j. j. arlt , w. ertmer , k. sengstock , d. s. petrov , g. v. shlyapnikov , h. kreutzmann , l. santos , m. lewenstein , phys .
rev .
lett . * 87 * 160406 ( 2001 ) .
d. hellweg , l. cacciapuoti , m. kottke , t. schulte , k. sengstock , w. ertmer , and j. j. arlt , phys .
rev .
lett . * 91 * , 010406 ( 2003 ) .
j. struck , c. lschlger , r. le targat , p. soltan - panahi , a. eckardt , m. lewenstein , p. windpassinger , k. sengstock , science * 333 * , 996 ( 2011 ) .
g. wirth , m. lschlger , and a. hemmerich , nature physics * 7 * , 147 ( 2011 ) .
p. soltan - panahi , d. s. luhmann , j. struck , p. windpassinger , and k. sengstock , nature phys . *
8 * , 71 ( 2012 ) .
m. lschlger , t. kock , g. wirth , a. ewerbeck , c. morais smith and a. hemmerich , new j. phys .
* 15 * 083041 ( 2013 ) .
k. jimnez - garcia , l. j. leblanc , r. a.williams,m . c. beeler , a. r. perry , and i. b. spielman , phys .
rev .
lett . * 108 * , 225303 ( 2012 ) .
j. struck , m. weinberg , c. lschlger , p. windpassinger , j. simonet , k. sengstock , r. hppner , p. hauke , a. eckardt , m. lewenstein , l. mathey , nature physics ( 2013 ) doi:10.1038/nphys2750 . c. v. parker , li - chung ha , c. chin , nature physics ( 2013 ) doi:10.1038/nphys2789 .
j. r. fienup , josa a , * 4 * , 118 ( 1987 ) .
j. miao , d. sayre , acta cryst .
a * 56 * , 596 ( 2000 ) .
s. marchesini , rev .
sci .
instrum . * 78 * , 011301 ( 2007 ) .
see supplemental material .
f. gerbier , s. trotzky , s. flling , u. schnorrberger , j. d. thompson , a. widera , i. bloch , l. pollet , m. troyer , b. capogrosso - sansone , n. v. prokofev , b. v. svistunov , prl , * 101 * , 155303 ( 2008 ) .
k. sacha , k. targoska , and j. zakrzewski , phys .
rev .
a * 85 * , 053613 ( 2012 ) .
j. struck , c. lschlger , m. weinberg , p. hauke , j. simonet , a. eckardt , m. lewenstein , k. sengstock , and p. windpassinger , phys .
rev .
lett . * 108 * , 225304 ( 2012 ) .
p. hauke , o. tieleman , a. celi , c. lschlger , j. simonet , j. struck , m. weinberg , p. windpassinger , k. sengstock , m. lewenstein , and a. eckardt , phys .
rev .
lett . * 109 * , 145301 ( 2012 ) .
a. kosior , k. sacha , phys .
rev .
a * 87 * , 023602 ( 2013 ) .
a. eckardt , c. weiss , and m. holthaus , phys .
rev .
lett . * 95 * , 260404 ( 2005 ) .
a. eckardt , p. hauke , p. soltan - panahi , c. becker , k. sengstock , m. lewenstein , europhys .
lett . *
89 * , 10010 ( 2010 ) .
e. arimondo , d. ciampini , a. eckardt , m. holthaus , and o. morsch , adv . at .
mol .
opt .
phys .
61 , 515 ( 2012 ) . |
the non - abelian nature of quantum chromodynamics ( qcd ) the theory of strong interaction gives rise to the expectation that its gauge bosons , the gluons , form composite objects denoted as glueballs @xcite .
these states would have access to various quantum numbers @xmath0 , where @xmath1 denotes the total spin , @xmath2 the parity and @xmath3 the charge conjugation ; the corresponding spectrum in the qcd yang - mills sector has been determined in numerical simulations @xcite but the identification of glueballs in experimental data has proven to be a challenge , particularly in the @xmath4 ( scalar ) channel . + + there are at least two reasons to study glueball states .
firstly , their masses are generated solely by the strong interaction ; no influence of the higgs mechanism providing , e.g. , quarks with a current mass is present in the glueball mass generation .
secondly , due to their structure glueballs must possess integer spin assigning them to mesons ; our understanding of the physical meson spectrum would not be complete without glueballs
. + + glueball ground state is scalar @xcite ; listings of the particle data group ( pdg ) @xcite contain five isospin - zero scalar states in the energy region below 1 gev : @xmath5 or @xmath6 , @xmath7 , @xmath8 , @xmath9 and @xmath10 .
while all of them possess the quantum numbers of the ground state , there are strong reasons to focus on resonances above 1 gev as candidates for the scalar glueball : ( @xmath11 ) simulations in lattice qcd determine the ground - state mass at between approximately 1.65 gev and 1.8 gev @xcite ; ( @xmath12 ) various effective approaches to low - energy qcd arrive at an analogous result for the mass while describing the overall dynamics correctly @xcite .
complication is that @xmath13 resonances will most certainly have contributions not only from the ( pure ) glueball states but also from those containing quarks ( @xmath14 @xcite , @xmath15 @xcite and others ) .
this leads to various issues in both theory and experiment @xcite and represents the main reason why the scalar glueball has still not been clearly identified .
+ + in this article , the question of the scalar glueball is tackled by a holographic approach to non - perturbative qcd .
such approaches are based on the idea of a duality between strongly coupled quantum gauge field theories and weakly coupled supergravity / superstring theories in one dimension higher , pursuing maldacena s groundbreaking conjecture of a complete equivalence between the supergravity limit of type - iib string theory on an @xmath16 space and the large-@xmath17 limit of an @xmath18 @xmath19 supersymmetric and conformal @xmath20 gauge theory on its boundary ( ads / cft correspondence @xcite ) . in ref .
@xcite , witten proposed an analogous duality in type - iia string theory , with supersymmetry and conformality not present in line with their absence in qcd .
the supersymmetry is broken by compactification on a circle ( @xmath21 ) ; for a vanishing circle radius , the 5-dimensional pure yang - mills theory is reduced to a 4-dimensional one .
however , the supergravity approximation requires finite circle radius ( whose inverse is defined as the kaluza - klein mass @xmath22 ) , and also a large coupling .
thus constructed holographic approaches are referred to as _ top - down _
models @xcite ; there are also more phenomenological _ bottom - up _ constructions
see ref .
@xcite and references therein .
this articles describes the implications of the top - down witten - sakai - sugimoto model @xcite for the glueball spectroscopy ; more details can be found in ref .
while witten s model contained only gauge fields , the novel feature of witten - sakai - sugimoto model is the inclusion of chiral quarks introduced by @xmath23 ( number of flavours ) probe d8 and anti - d8 branes [ inducing @xmath24 chiral symmetry ] that extend along all dimensions of the 10-dimensional space with the exception of a ( kaluza - klein ) circle .
the branes are usually antipodally separated with regard to this @xmath21 . the space geometry , however , is such that the branes and antibranes merge at a certain point interpreted as realisation of the chiral - symmetry breaking .
+ + up to a chern - simons term , the corresponding action for d8-branes reads @xmath25 where @xmath26 ( and @xmath27 , with @xmath28 the string coupling ) , @xmath29 is the metric of the d - brane world volume , @xmath30 is the dilaton field and @xmath31 a field strength tensor whose components are , upon dimensional reduction , identified as meson fields of interest .
no backreaction of the witten - model background to d8-branes is considered ; consequently @xmath23 is fixed and , as can be argued @xcite , significantly smaller than the number of colours ( large-@xmath32 limit ) .
+ + the above action is expanded up to the second order in fields as @xmath33\ ] ] where @xmath34 @xcite , @xmath35 is the t hooft coupling ( and @xmath36 the 4-dimensional coupling ) , @xmath37 is the holographic radial coordinate ( and @xmath38 ) and @xmath39 is the flat metric diag@xmath40 . the kaluza - klein mass @xmath22 sets the model scale ; beside the scale , the model contains only one unknown quantity : the coupling @xmath41 .
they are usually determined such that the mass of the rho meson and the pion decay constant correspond to their physical values .
this yields @xmath42 mev and @xmath43 ; alternative methods for their determination do not alter model conclusions @xcite .
+ + once @xmath22 and @xmath41 are known , the model describes various experimental quantities surprisingly well @xcite .
glueballs are of interest in this article ; building on the work of ref .
@xcite , the following masses in the scalar channel are obtained : @xmath44 two types of glueball states are present : one , denoted with @xmath45 , is predominantly dilaton while the other , denoted with @xmath46 , involves a graviton polarisation in a fourth spatial dimension , unlike the dilaton , and has therefore been termed _ exotic _
excited states are denoted with an asterisk .
we observe that the exotic ground state is approximately 50% lighter than the expectation from lattice qcd ; the dilaton mode mass is only approximately 15% smaller than the lattice result .
numerical simulations also indicate the mass of the first excited scalar state to be @xmath47 2600 mev with errors amounting to @xmath47 300 mev @xcite ( although mass corrections in the unquenched case may be substantial @xcite ) ; the excited dilaton state is within errors consistent with the lattice result while the excited exotic mode is approximately 15% too light .
hence already from the mass results the indication is that the exotic mode could be discarded while the dilaton mode appears compatible with simulations of the yang - mills sector of qcd .
+ + this is corroborated by the decay ratios of the modes .
the corresponding lagrangians are obtained by inserting 10-dimensional metric fluctuations ( whose explicit forms together with further details are presented in ref .
@xcite ) into the action for d8 branes and integrating over the bulk coordinates ( see also ref .
@xcite ) ; ratios of decay widths @xmath48 and the respective masses @xmath49 read @xmath50 the two main candidates for the scalar glueball are the @xmath9 and @xmath10 resonances ( see ref .
@xcite and references therein ) .
experimental data imply @xmath51 @xcite and @xmath52
@xmath53 @xcite .
the exotic mode is thus too broad , again indicating that its interpretation as a physical state is uncertain ; contrarily , the ratio @xmath54 for the dilaton mode is within the experimental interval for @xmath10 . +
similar is true for the excited states : @xmath55 is larger than @xmath56 and nicely comparable to widths of @xmath13 states near and above 2 gev ; however , the full decay width of @xmath57 having contributions from @xmath58 , @xmath59 , @xmath60 and other channels is unphysically large ( @xmath61 1 gev , see ref .
contrarily , @xmath56 is small but the full decay width of @xmath62 is of the order of 460 mev @xcite and thus significantly closer to the data @xcite whose current uncertainties unfortunately do not allow for a clear identification of an excited scalar glueball .
in this article , a top - down holographic approach to low - energy qcd witten - sakai - sugimoto model has been presented and its implications for phenomenology of scalar glueballs have been discussed .
the model offers two sets of glueball states , a dilaton and an _ exotic _
mode that , unlike the dilaton , involves a graviton polarisation in a fourth spatial dimension .
the exotic ground state has a mass approximately 50% smaller than the value expected in lattice qcd ; its @xmath63 decay width is substantially larger than that of the two prime candidates for the scalar glueball , the resonances @xmath9 and @xmath10 .
contrarily , the mass of the dilaton mode (= 1487 mev ) is quite close to masses of both mentioned resonances ; its @xmath63 decay width is within the experimental range for @xmath10 , which therefore appears to be the preferred candidate for the glueball ground state . in the excited channel ,
the exotic state is unphysically broad while the dilaton width @xmath64 mev is close to the ( still ambiguous ) data on @xmath13 states near / above 2 gev .
+ nonetheless , further pursuit of glueball dynamics in holography is called for , particularly in light of expectations from the planned panda experiments at fair @xcite .
i am grateful to f. brnner and a. rebhan for collaboration and to d. bugg and s. janowski for extensive discussions .
i am also grateful to the workshop organisers for their generous support of my participation .
this work is supported by the austrian science fund fwf , project no .
p26366 .
a. salomone , j. schechter and t. tudron , phys .
d * 23 * , 1143 ( 1981 ) ; c. rosenzweig , a. salomone and j. schechter , nucl .
b * 206 * , 12 ( 1982 ) [ erratum - ibid .
b * 207 * , 546 ( 1982 ) ] ; a. a. migdal and m. a. shifman , phys .
b * 114 * , 445 ( 1982 ) ; h. gomm and j. schechter , phys .
b * 158 * , 449 ( 1985 ) ; r. gomm , p. jain , r. johnson and j. schechter , phys .
d * 33 * , 801 ( 1986 ) .
c. j. morningstar and m. j. peardon , phys .
d * 60 * , 034509 ( 1999 ) [ hep - lat/9901004 ] ; y. chen _ et al .
_ , phys .
d * 73 * , 014516 ( 2006 ) [ hep - lat/0510074 ] ; m. loan , x. q. luo and z. h. luo , int .
j. mod .
a * 21 * , 2905 ( 2006 ) [ hep - lat/0503038 ] .
e. gregory , a. irving , b. lucini , c. mcneile , a. rago , c. richards and e. rinaldi , jhep * 1210 * , 170 ( 2012 ) [ arxiv:1208.1858 [ hep - lat ] ] . g. b. west , phys . rev .
* 77 * , 2622 ( 1996 ) [ hep - ph/9603316 ] .
k. a. olive _ et al_. ( particle data group ) , chin . phys .
c , * 38 * , 090001 ( 2014 ) .
f. e. close and a. kirk , eur .
j. c * 21 * , 531 ( 2001 ) [ hep - ph/0103173 ] ; f. giacosa , t. gutsche and a. faessler , phys .
c * 71 * , 025202 ( 2005 ) [ hep - ph/0408085 ] ; f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
b * 622 * , 277 ( 2005 ) [ hep - ph/0504033 ] ; f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , phys .
d * 72 * , 094006 ( 2005 ) [ hep - ph/0509247 ] ; h. y. cheng , c. k. chua and k. f. liu , phys . rev .
d * 74 * , 094005 ( 2006 ) [ hep - ph/0607206 ] ; v. mathieu , n. kochelev and v. vento , int . j. mod .
e * 18 * , 1 ( 2009 ) [ arxiv:0810.4453 [ hep - ph ] ] ; s. janowski , d. parganlija , f. giacosa and d. h. rischke , phys .
d * 84 * , 054007 ( 2011 ) [ arxiv:1103.3238 [ hep - ph ] ] ; s. janowski , f. giacosa and d. h. rischke , phys . rev .
d * 90 * , no .
11 , 114005 ( 2014 ) [ arxiv:1408.4921 [ hep - ph ] ] .
d. parganlija , f. giacosa and d. h. rischke , phys . rev .
d * 82 * , 054024 ( 2010 ) [ arxiv:1003.4934 [ hep - ph ] ] ; d. parganlija , p. kovacs , g. wolf , f. giacosa and d. h. rischke , phys .
d * 87 * , 014011 ( 2013 ) [ arxiv:1208.0585 [ hep - ph ] ] .
r. l. jaffe , phys .
d * 15 * , 267 ( 1977 ) ; a. h. fariborz , r. jora and j. schechter , phys . rev .
d * 72 * , 034001 ( 2005 ) [ arxiv : hep - ph/0506170 ] ; f. giacosa , phys .
d * 75 * , 054007 ( 2007 ) [ hep - ph/0611388 ] ; a. heinz , s. struber , f. giacosa and d. h. rischke , phys .
d * 79 * , 037502 ( 2009 ) [ arxiv:0805.1134 [ hep - ph ] ] ; a. heinz , s. struber , f. giacosa and d. h. rischke , acta phys .
supp . * 3 * , 925 ( 2010 ) [ arxiv:1006.5393 [ hep - ph ] ] .
d. v. bugg , phys .
rept . *
397 * , 257 ( 2004 ) [ hep - ex/0412045 ] ; e. klempt and a. zaitsev , phys .
rept . * 454 * , 1 ( 2007 ) [ arxiv:0708.4016 [ hep - ph ] ] .
d. parganlija , j. phys .
* 426 * , 012019 ( 2013 ) [ arxiv:1211.4804 [ hep - ph ] ] ; d. parganlija , j. phys .
ser . * 503 * , 012010 ( 2014 ) [ arxiv:1312.2830 [ hep - ph ] ] .
e. witten , adv .
theor . math .
* 2 * , 505 ( 1998 ) [ hep - th/9803131 ] .
i. r. klebanov and m. j. strassler , jhep * 0008 * , 052 ( 2000 ) [ hep - th/0007191 ] ; j. babington , j. erdmenger , n. j. evans , z. guralnik and i. kirsch , phys .
d * 69 * , 066007 ( 2004 ) [ hep - th/0306018 ] ; m. kruczenski , d. mateos , r. c. myers and d. j. winters , jhep * 0405 * , 041 ( 2004 ) [ hep - th/0311270 ] .
t. sakai and s. sugimoto , prog .
theor . phys .
* 113 * , 843 ( 2005 ) [ hep - th/0412141 ] ; t. sakai and s. sugimoto , prog .
phys . * 114 * , 1083 ( 2005 ) [ hep - th/0507073 ] ; t. imoto , t. sakai and s. sugimoto , prog .
phys . * 124 * , 263 ( 2010 ) [ arxiv:1005.0655 [ hep - th ] ] ; f. brnner , d. parganlija and a. rebhan , acta phys .
* 7 * , no .
3 , 533 ( 2014 ) [ arxiv:1407.6914 [ hep - ph ] ] ; a. rebhan , arxiv:1410.8858 [ hep - th ] . | identification of glueballs bound states of gauge bosons in quantum chromodynamics ( qcd ) is a very important open question in dynamics of the strong interaction .
the search for the glueball ground state , carrying scalar quantum numbers , poses a particular challenge due to the existence of ( _ i _ ) several candidates for its realisation in the physical spectrum and ( _ ii _ ) inevitable mixing of the pure glueball state with those comprised of quarks . in this article ,
i discuss implications of an approach in holographic qcd where , among others , the mass and the two - pion decay of the pure scalar glueball can be studied . |
the problem of computing a function from distributed information arises in many different contexts ranging from auctions and financial trading to sensor networks . in order to compute the desired target function , communication between the distributed users
is required .
if this communication takes place over a shared medium , such as in a wireless setting , the channel introduces interactions between the transmitted signals .
this suggests the possibility to harness these signal interactions to facilitate the task of computing the desired target function .
a fundamental question is therefore whether by jointly designing encoders and decoders for communication and computation , we can improve the efficiency of distributed computation . in this paper
, we explore this question by considering computation of a function over a two - user multiple - access channel ( mac ) . in order to focus on the impact of the structural mismatch between the target and channel functions on the efficiency of computation , we ignore channel noise and consider only _ deterministic _ macs here .
more formally , the setting consists of two transmitters observing a ( random ) variable @xmath0 and @xmath1 , respectively , and a receiver aiming to compute the function @xmath2 of these variables .
the two transmitters are connected to the destination through a deterministic mac with inputs @xmath3 and output @xmath4 , where @xmath5 describes the actions of the channel .
a straightforward achievable scheme for this problem is to separate the tasks of communication and computation : the transmitters communicate the values of @xmath6 and @xmath7 to the destination , which then uses these values to compute the desired target function @xmath8 .
this requires the receiver to decode @xmath9 message bits .
however , the mac itself also computes a function @xmath10 of the two inputs @xmath11 , creating the opportunity of taking advantage of the structure of @xmath5 to calculate @xmath12 .
this is trivially possible when @xmath5 and @xmath12 are _ matched _
, i.e. , compute the same function on their inputs . in such cases , performing the tasks of communication and computation jointly results in significantly fewer bits to be communicated . indeed , in the matched case only the @xmath13 bits describing the function value are recovered at the receiver .
this could be considerably less than the @xmath9 bits resulting from the separation approach .
naturally , in most cases the channel @xmath5 and the target function @xmath12 are _ mismatched_. the question is thus whether we can still obtain performance gains over separation in this mismatched situation . in other words , we ask if in general the natural computation done by the channel can be harnessed to help with the computation of the desired target function .
we consider two cases : i ) _ one - shot communication _
, where the mac is used only once , but the channel input alphabet @xmath14 and output alphabet @xmath15 are allowed to vary as a function of the domain @xmath16 of the target function . in this case ,
performance is measured in terms of the scaling needed for the channel alphabets with respect to the computation alphabets , i.e. , how @xmath17 grow with @xmath18 .
this is closer to the formulation in the computer science literature .
ii ) _ multi - shot communication _ , where the channel alphabets @xmath19 are of fixed size , but the channel can be used several times . in this case , performance is measured in terms of computation rate , i.e. , how many channel uses are needed to compute the target function .
this is closer to the formulation considered in information theory . as the main result of this paper
, we show that separation between computation and communication is essentially optimal for most with given domain @xmath16 and range @xmath20 , and all channel functions @xmath21 with given input alphabet @xmath14 and output alphabet @xmath15 , separation is optimal except for at most an exponentially small ( in domain size @xmath18 ) fraction of pairs . ]
pairs @xmath22 of target and channel functions .
in other words , the structural mismatch between the functions @xmath12 and @xmath5 is in general too strong for joint computation and communication designs to yield any performance gains .
we illustrate this with an example for one - shot communication .
assume that the variables @xmath23 at the transmitters take on a large range of values , say @xmath24 , and the receiver is only interested in knowing if @xmath25 , i.e. , in a binary target function .
then for most macs and one - shot communication , a consequence , it follows from the proofs that for a given domain @xmath16 the statements hold for all but an exponentially small ( in @xmath18 ) fraction of channel functions . ] of theorems [ thm : identityoneuse ] and [ thm : balanced ] in section [
sec : main ] ( illustrated in example [ eg : greater ] ) is that the transmitters need to convey the _
entire _ values of @xmath23 to the destination , which then simply compares them .
thus , even though the destination is interested in only a _ single _
bit about @xmath26 , it is still necessary to transmit @xmath27 bits over the channel .
more generally , theorems [ thm : identityoneuse ] and [ thm : balanced ] in section [ sec : main ] together demonstrate that for most target functions separation of communication and computation is asymptotically optimal for most macs .
example [ eg : equality ] illustrates that only for special functions like an equality check ( i.e. , checking whether @xmath28 ) can we significantly improve upon the simple separation scheme . intuitively , this is because the structural mismatch between most target and channel functions is too large to allow for any possibility of direct computation of the target function value without resorting to recovering the user messages first .
the technical ideas that enable these observations are based on a connection with results in extremal graph theory such as existence of complete subgraphs and matchings of a given size in a bipartite graph .
these connections might be of independent interest . similarly , for multi - shot communication , where we repeatedly use a fixed channel , theorem [ thm : converse_n ] in section [ sec :
main ] shows that for most functions , the computation rate is necessarily as small as that for the identity target function describing the entire variables @xmath23 at the destination . in other words ,
separation of communication and computation is again optimal for most target and channel functions . to prove this result
, the usual approach using cut - set bound arguments is not tight enough .
indeed , example [ eg : n ] shows that the ratio between the upper bound on the computation rate obtained from the cut - set bound and the correct scaling derived in theorem [ thm : converse_n ] can be unbounded .
rather , the structures of the target and channel functions have to be analyzed jointly .
these results show that , in general , there is little or no benefit in joint designs : computation - communication separation is optimal for most cases .
we thus advocate in this paper that separation of computation and communication for multiple - access channels is not just an attractive option from an implementation point of view , but , except for special cases , actually entails little loss in efficiency .
the problem of distributed function computation has a rich history and has been studied in many different contexts . in computer science , it has been studied under the branch of communication complexity , for example see @xcite and references therein .
early seminal work by yao @xcite considered interactive communication between two parties . among several other important results ,
the paper showed that the number of exchanged bits required to compute most target functions is as large as for the identity function . in the context of information theory ,
distributed function computation has been studied as an extension of distributed source coding in @xcite .
for example , krner and marton @xcite showed that for the computation of the finite - field sum of correlated sources linear codes can outperform random codes .
this was extended to large networks represented as graphs in @xcite and references therein .
randomized gossip algorithms @xcite have been proposed as practical schemes for information dissemination in large unreliable networks and were studied in the context of distributed computation in @xcite among several others . in most of these works , communication channels are represented as orthogonal point - to - point links .
when the channel itself introduces signal interaction , as is the case for a mac , there can be a benefit from jointly handling the communication and computation tasks as illustrated in @xcite .
function computation over macs has been studied in @xcite and references therein .
there is some work touching on the aspect of structural mismatch between the target and the channel functions . in @xcite ,
an example was given in which the mismatch between a linear target function with integer coefficients and a linear channel function with real coefficients can significantly reduce efficiency . in @xcite ,
it was conjectured that , for computation of finite - field addition over a real - addition channel , there could be a gap between the cut - set bound and the computation rate . in @xcite , mismatched computation when the network performs linear finite - field operations was studied . to the best of our knowledge ,
a systematic study of channel and computation mismatch is initiated in this work .
the paper is organized as follows . in section [ sec : problem ] , we formally introduce the questions studied in this paper .
we present the main results along with illustrative examples in section [ sec : main ] .
most of the proofs are given in section [ sec : proofs ] .
throughout this paper , we use sans - serif font for random variables , e.g. , @xmath29 .
we use bold font lower and upper case to denote vectors and matrices , e.g. , @xmath30 and @xmath31 .
all sets are typeset in calligraphic font , e.g. , @xmath14 .
we denote by @xmath32 and @xmath33 the logarithms to the base @xmath34 and @xmath35 , respectively . a discrete , memoryless , deterministic two - user mac consists of two _ input alphabets _ @xmath36 and @xmath37 , an _ output alphabet _ @xmath15 , and a deterministic _ channel function _ @xmath38 .
given channel inputs @xmath39 , the output of the mac is @xmath40 each transmitter @xmath41 has access to an independent and uniformly distributed _ message _ @xmath42 .
the objective of the receiver is to compute a _
target function _
@xmath43 of the user messages , see fig .
[ fig : mac ] . formally , each transmitter @xmath44
consists of an _ encoder _ @xmath45 mapping the message @xmath46 into the channel input @xmath47 the receiver consists of a _ decoder _ @xmath48 mapping the channel output @xmath49 into an _ estimate _ @xmath50 of the target function @xmath8 .
the _ probability of error _ is @xmath51 we point out that this differs from the ordinary communication setting , in which the decoder aims to recover both messages @xmath26 . instead , in the setting here , the decoder is not interested in @xmath26 , but only in the value @xmath8 of the target function . in the following
, it will often be convenient to represent the target function @xmath12 and the channel @xmath5 by their corresponding matrices @xmath52 and @xmath53 , respectively .
in other words , @xmath54 for @xmath55 , denote by @xmath56 the @xmath57-fold use of the same channel matrix @xmath31 .
in other words , the matrix @xmath56 describes the actions of the ( memoryless ) channel @xmath31 on the sequence @xmath58,x_2[1 ] ) , ( x_1[2],x_2[2 ] ) , \ldots , ( x_1[n],x_2[n ] ) \bigr)\ ] ] of length @xmath57 of channel inputs .
a pair @xmath59 of target and channel functions is _
@xmath60-feasible _ , if there exist encoders @xmath61 and a decoder @xmath62 computing the target function @xmath63 over @xmath31 with probability of error at most @xmath60 .
we will often consider pairs @xmath64 , in which case the definition of @xmath60-feasibility allows for coding over @xmath57 uses of the channel @xmath31 . without loss of generality ,
we assume that the target function @xmath63 has no two identical rows or two identical columns , since we could otherwise simply eliminate one of them . for ease of exposition , we will focus on the case
@xmath65 to simplify notation , we assume without loss of generality that @xmath66 finally , to avoid trivial cases , we assume that all cardinalities are strictly bigger than one , and that @xmath67 .
we denote by @xmath68 the collection of all target functions @xmath69 .
similarly , we denote by @xmath70 the collection of all channels @xmath71 .
the next example introduces several target functions @xmath63 and channels @xmath31 that will be used to illustrate results in the remainder of the paper .
[ eg : fnmac ] we start by introducing four target functions @xmath12 . *
let @xmath72 .
the _ identity _
target function is @xmath73 for all @xmath74 . since we will refer to the identity target function repeatedly
, we will denote it by the symbol @xmath75 .
* let @xmath76 .
the _ equality _
target function is @xmath77 for all @xmath78 .
* let @xmath79 .
greater - than _ target function is @xmath80 for all @xmath78 .
* a _ random _ target function corresponds to the matrix @xmath81 being a random variable , with each entry chosen independently and uniformly over @xmath20 .
the matrix @xmath81 is generated before communication begins and is known at both the transmitters and at the receiver .
we now introduce three channels @xmath5 .
* let @xmath82 and @xmath83 . the _ binary adder _
mac is given by @xmath84 for all @xmath85 , and where @xmath86 denotes ordinary addition .
* let @xmath82 and @xmath87 .
boolean @xmath88 _ or _ boolean or _
mac is @xmath89 for all @xmath85 . *
a _ random _ channel corresponds to the matrix @xmath90 being a random variable , with each entry chosen independently and uniformly over @xmath15 .
the matrix @xmath90 is generated before communication begins and is known at both the transmitters and at the receiver .
the emphasis in this paper is on the asymptotic behavior for large function domains , i.e. , as @xmath91 .
we allow the other cardinalities @xmath92 , @xmath93 and @xmath94 to scale as a function of @xmath95 .
we use the notation @xmath96 for the relation @xmath97 and analogously for @xmath98 .
similarly , we use @xmath99 for the relation @xmath100 and analogously for @xmath101 . finally , @xmath102 is short hand for @xmath103 for example , @xmath104 is equivalent is @xmath105 as @xmath91 stands for @xmath106 . ] to @xmath107 as @xmath91 . with slight abuse of notation , we will write @xmath108 to mean that @xmath109 for _ some _ finite @xmath110 . throughout this paper , we are interested in efficient computation of the target function @xmath12 over the channel @xmath5 . in theorems [ thm : identityoneuse ] and
[ thm : balanced ] only a single use of the channel is permitted , and efficiency is expressed in terms of the required cardinalities @xmath92 and @xmath93 of the channel alphabets as a function of @xmath95 . in theorems [ thm : identitymultipleuse ] and [
thm : converse_n ] , multiple uses of the channel are allowed , and efficiency is then naturally expressed in terms of the number of required channel uses @xmath111 as a function of @xmath95 .
finally , all results are stated in terms of the fraction of channels ( in theorems [ thm : identityoneuse ] and [ thm : balanced ] ) or target functions ( in theorem [ thm : converse_n ] ) for which successful computation is possible .
the proofs of all the theorems are based on probabilistic methods by using a uniform distribution over choices of channel @xmath5 or target functions @xmath12 .
let @xmath112 be the identity target function introduced in example [ eg : fnmac ] , and let @xmath31 be an arbitrary channel matrix . consider any other target function @xmath113 over the same domain @xmath114 , but with possibly different range @xmath20 .
assume @xmath115 is @xmath60-feasible .
then @xmath116 is also @xmath60-feasible , since we can first compute @xmath75 ( and hence @xmath117 and @xmath118 ) over the channel @xmath31 and then simply apply the function @xmath63 to the recovered messages @xmath117 and @xmath118 . this architecture , separating the computation task from the communication task , is illustrated in fig .
[ fig : separation ] . as a concrete example , let @xmath63 be the greater - than target function introduced in example [ eg : fnmac ] . the range @xmath119 of @xmath63 has cardinality two . on the other hand ,
the identity function @xmath75 has range @xmath114 of cardinality @xmath120 .
in other words , for large @xmath95 , the identity target function is considerably more complicated than the greater - than target function . as a result
, one might expect that the separation - based architecture in fig .
[ fig : separation ] is highly suboptimal in terms of the computation efficiency as described in section [ sec : problem ] . as the main result of this paper ,
we prove that this intuition is wrong in most cases . instead
, we show that for most pairs @xmath116 of target function and mac , separation of computation and communication is close to optimal .
we discuss the single channel - use case in section [ sec : main_single ] , and the @xmath57 channel - uses case in section [ sec : main_multiple ] . in this section
, we will focus on the case where the target function needs to be computed using just one use of the channel .
the natural value of the upper bound on the probability of error is @xmath122 in this case . in other words
, we will be interested in @xmath123-feasibility .
we start by deriving conditions under which computation of the identity target function over a mac is feasible .
equivalently , these conditions guarantee that _ any _ target function with same domain cardinality @xmath95 can be computed over a mac by separating communication and computation as discussed above . [
thm : identityoneuse ] let @xmath112 be the identity target function , and assume [ eq : identity1 ] @xmath124 then , @xmath125 the proof of theorem [ thm : identityoneuse ] is reported in section [ sec : proofs_identity ] .
recall that @xmath70 is the collection of all channels @xmath31 of dimension @xmath126 and range of cardinality @xmath127 .
theorem [ thm : identityoneuse ] ( together with the separation approach discussed earlier ) thus roughly implies that any target function with a domain of cardinality @xmath95 can be computed over most macs of input cardinality @xmath92 of order at least @xmath95 and output cardinality @xmath93 of order at least @xmath128 .
the precise meaning of `` most '' is that the fraction of channels @xmath31 in @xmath70 for which the statement holds goes to one as @xmath91 .
a look at the proof of the theorem shows that the convergence to this limit is , in fact , exponentially fast . in other words ,
the fraction of channels for which the theorem fails to hold is exponentially small in the domain cardinality @xmath95 .
since the achievable scheme is separation based , this conclusion holds regardless of the cardinality @xmath94 of the range of the target function .
similarly , since it is clear that the channel input has to have at least cardinality @xmath92 of order @xmath95 for successful computation , we see that the condition on @xmath92 in theorem [ thm : identityoneuse ] is not a significant restriction . what is significant , however , is the restriction that @xmath93 is at least of order @xmath128 .
the next result shows that this restriction on @xmath93 is essentially also necessary .
before we state the theorem , we need to introduce one more concept .
consider a target function @xmath129 .
for a set @xmath130 , consider @xmath131 for @xmath132 $ ] , the target function @xmath12 is said to be _
@xmath133-balanced _ if there exist a partition @xmath134 of @xmath20 such that @xmath135 for all @xmath41 .
most functions are @xmath133-balanced for any @xmath136 and @xmath94 as long as @xmath95 is large enough .
indeed , choosing @xmath137 and @xmath138 shows that @xmath139 where we recall that @xmath68 denotes the collection of all target functions @xmath63 of dimension @xmath140 and range of cardinality @xmath141 .
in fact , the convergence in is again exponentially fast in @xmath95 .
moreover , many functions of specific interest are balanced .
[ eg : balanced ] consider the target functions introduced in example [ eg : fnmac ] . *
the identity and the greater - than target functions are @xmath133-balanced for any constant @xmath142 and @xmath95 large enough . *
the equality target function is _ not _
@xmath133-balanced for any constant @xmath143 as @xmath91 . indeed , since @xmath144 in this case , the only choice ( up to labeling ) is to set @xmath145 and @xmath146
. then @xmath147 and @xmath148 , which is not @xmath133-balanced for any constant @xmath143 as @xmath91 .
we have the following converse result to theorem [ thm : identityoneuse ] for balanced target functions .
[ thm : balanced ] fix a constant @xmath149 $ ] independent of @xmath95 .
assume @xmath150 and [ eq : balanced0 ] @xmath151 let @xmath152 be any @xmath133-balanced target function .
then @xmath153 the proof of theorem [ thm : balanced ] is reported in section [ sec : proofs_balanced ] .
recall that the notation @xmath154 is used to indicate that @xmath92 grows at most polynomially in @xmath95an assumption that is quite mild .
thus , theorem [ thm : balanced ] roughly states that regardless of the value of @xmath94 , if the cardinality @xmath93 of the channel output is order - wise less than @xmath128 , then any balanced target function with a range of cardinality @xmath94 can not be computed over most macs . here the precise meaning of `` most '' is again that the fraction of channel matrices with at most @xmath93 channel outputs for which successful computation is possible converges to zero , and a look at the proof reveals again that this convergence is , in fact , exponentially fast in @xmath95 .
comparing this to theorem [ thm : identityoneuse ] , we see that the same scaling of @xmath93 allows computation of a target function using a separation based scheme ( i.e. , by first recovering the two messages @xmath155 and then applying the target function to compute the estimate @xmath156 ) .
thus , for the computation of a given balanced function over most macs , separation of computation and communication is essentially optimal .
moreover , since most functions are balanced by , the same also holds for most pairs @xmath116 of target and channel functions .
[ eg : greater ] let @xmath63 be the greater - than target function of domain @xmath157 introduced in example [ eg : fnmac ] .
note that this target function has range of cardinality @xmath144 , i.e. , @xmath63 is binary .
from example [ eg : balanced ] , we know that @xmath63 is balanced for any constant @xmath158 and @xmath95 large enough
. thus theorem [ thm : balanced ] applies , showing that , for large @xmath95 and most macs @xmath31 , separation of computation and communication is essentially optimal .
observe that the receiver is interested in only a _
single _ bit of information about @xmath26 .
nevertheless , the structure of the greater - than target function is complicated enough that , in order to recover this single bit , the decoder is essentially forced to learn @xmath26 itself . in other words , in order to compute the single desired bit , communication of @xmath159 message bits is essentially necessary .
theorem [ thm : balanced ] is restricted to balanced functions . even though only a vanishingly small fraction of target functions is not balanced , it is important to understand this restriction .
we illustrate this through the following example .
[ eg : equality ] assume @xmath144 and [ eq : equality0 ] @xmath160 let @xmath161 be the equality target function introduced in example [ eg : fnmac ] .
then @xmath162 the proof of the above statement is reported in section [ sec : proofs_equality ] .
this result shows that the equality function can be computed over a large fraction of macs with output cardinality @xmath93 of order at least @xmath95 .
this contrasts with output cardinality @xmath93 of order @xmath128 that is required for successful computation of balanced functions in theorem [ thm : balanced ] .
recall from example [ eg : balanced ] that the equality target function is _ not _
@xmath133-balanced for any @xmath143 and @xmath95 large enough .
thus , does not contradict theorem [ thm : balanced ] .
it does , however , show that for unbalanced functions separation of communication and computation can be suboptimal .
in this section , we allow multiple uses of the mac .
our emphasis will again be on the asymptotic behavior for large function domains @xmath91 .
however , in this section we keep the mac @xmath164 , and hence also the cardinalities of the channel domain @xmath14 and channel range @xmath15 , fixed .
instead , we characterize the minimum number @xmath165 of channel uses required to compute the target function .
we begin by stating a result for the identity target function introduced in example [ eg : fnmac ] .
equivalently , this result applies to _ any _ target function ( with same domain cardinality @xmath95 ) by using a scheme separating communication and computation .
let @xmath166 denote the entropy of a random variable @xmath167 .
[ thm : identitymultipleuse ] fix a constant @xmath168 independent of @xmath95 , and assume that @xmath169 and @xmath127 are constant . let @xmath112 be the identity target function , and let @xmath170 be any mac .
consider any joint distribution of the form @xmath171 , where @xmath172 is specified by the channel function @xmath31 .
then , for any @xmath111 satisfying [ eq : identityconstraints ] @xmath173 @xmath174 is @xmath60-feasible for @xmath95 large enough .
the result follows directly from the characterization of the achievable rate region for ordinary communication over the mac , see for example ( * ? ? ?
* theorem 14.3.3 ) .
using separation , theorem [ thm : identitymultipleuse ] implies that , for large enough @xmath95 , any target function of domain cardinality @xmath95 can be reliably computed over @xmath111 uses of an mac @xmath31 as long as it satisfies the constraints in .
the next result states that for most functions these restrictions on @xmath111 are essentially also necessary .
[ thm : converse_n ] assume that as @xmath91 stands for @xmath175 . ] @xmath176 as @xmath91 , that @xmath177 , and that @xmath169 and @xmath127 are constant .
let @xmath170 be any mac .
then , for any @xmath111 satisfying @xmath178 we must have @xmath179 for some joint distribution of the form @xmath180 , where @xmath172 is specified by the channel function @xmath31 .
the proof of theorem [ thm : converse_n ] is presented in section [ sec : proofs_n ] .
recall that @xmath68 denotes the collection of all target functions @xmath63 of dimension @xmath140 and range of cardinality @xmath141 .
together , theorems [ thm : identitymultipleuse ] and [ thm : converse_n ] thus show that , for any deterministic mac and most target functions , the smallest number of channel uses @xmath181 that enables reliable computation is of the same order as that needed for the identity function .
moreover , they show that for most such pairs , separation of communication and computation is essentially optimal even if we allow multiple uses of the channel and nonzero error probability . here
the precise meaning of `` most '' is that the statement holds for all but a vanishing fraction of functions .
moreover , the proof of the theorem shows again that this fraction is , in fact , exponentially small in @xmath95 .
[ eg : n ] let @xmath31 be the binary adder mac introduced in example [ eg : fnmac ] .
define @xmath182 where the maximization is over all independent random variables @xmath183 taking values in the channel input alphabet @xmath14 .
@xmath184 denotes the maximum entropy that can be induced at the channel output . for the binary adder mac @xmath31
, it follows from in theorem [ thm : identitymultipleuse ] that the identity function can be reliably computed over @xmath31 if the number of channel uses @xmath185 .
on the other hand , theorem [ thm : converse_n ] shows that for most functions @xmath63 of domain @xmath140 and range cardinality @xmath186 , the smallest number of channel uses @xmath181 required for reliable computation is of order @xmath187 .
thus , near - optimal performance can be achieved by separating computation and communication .
in other words , even though the receiver is only interested in @xmath188 function bits , it is essentially forced to learn the @xmath159 message bits as well .
this example also illustrates that the usual way of proving converse results based on the cut - set bound is not tight for most @xmath116 pairs .
for example , ( * ? ? ?
* lemma 13 ) shows that for reliable computation we need to have @xmath189 where @xmath190 denotes entropy .
since @xmath63 has range of cardinality @xmath94 , we have @xmath191 for @xmath192 and @xmath193 as considered here , the tightest bound that can in the _ best case _ be derived via the cut - set approach is thus @xmath194 however , we know that the correct scaling for @xmath181 is @xmath187 . hence , the cut - set bound is loose by an unbounded factor as @xmath91 .
we now prove the main results .
the proofs of theorems [ thm : identityoneuse ] and [ thm : balanced ] are reported in sections [ sec : proofs_identity ] and [ sec : proofs_balanced ] respectively .
the proof of in example [ eg : equality ] is presented in section [ sec : proofs_equality ] .
finally , the proof of theorem [ thm : identitymultipleuse ] is covered in section [ sec : proofs_n ] .
we start by presenting some preliminary observations in section [ sec : proofs_preliminaries ] .
recall our assumption that no two rows or two columns of the target function @xmath63 are identical . as a result
, @xmath63 can be computed over the channel @xmath31 with zero error , i.e. , @xmath116 is @xmath123-feasible , if and only if there exists a @xmath195 submatrix ( with ordered rows and columns ) @xmath196 of @xmath31 such that any two entries @xmath197 and @xmath198 with @xmath199 satisfy @xmath200 , see fig . [
fig : scheme ] . on the other hand
, this is not necessary if a probability of error @xmath168 can be tolerated .
as an example , consider the equality function . for any positive @xmath60 , a trivial decoder that always outputs
@xmath123 computes the equality function with probability of error @xmath201 .
as @xmath91 , the probability of error is eventually less than @xmath60 .
this motivates the following definition .
given a target function @xmath202 , a function @xmath203 with @xmath204 is said to be a _
@xmath60-approximation _ of @xmath12 if there exist two mappings @xmath205 and @xmath206 such that @xmath207 in words , the target function @xmath12 is equal to the approximation function @xmath208 for at least a @xmath209 fraction of all message pairs . as before , a @xmath60-approximation function @xmath210 can be represented by a @xmath211 matrix @xmath212 .
we have the following straightforward relation .
[ lemma : translation ] consider a target function @xmath63 and mac @xmath31 .
if @xmath116 is @xmath60-feasible , then there exists a @xmath60-approximation @xmath212 of @xmath63 such that @xmath213 is @xmath123-feasible .
let @xmath214 and @xmath62 be the encoders and the decoder achieving probability of error at most @xmath60 for @xmath116 .
let @xmath215 be the range of @xmath216 , and set @xmath217 for all @xmath218 .
then @xmath219 is a @xmath60-approximation of @xmath12 , and @xmath220 is @xmath123-feasible .
we will make frequent use of the chernoff bound , which we recall here for future reference .
let @xmath221 be independent random variables , and let @xmath222 by markov s inequality , @xmath223 assume furthermore that each @xmath224 takes value in @xmath225 , and set @xmath226 then , for any @xmath227 , @xmath228 see for example ( * ? ? ?
* theorem 4.1 , theorem 4.2 ) .
a scheme can compute the identity target function with zero error if and only if the channel output corresponding to any two distinct pairs of user messages is different . in
what follows , we will show that such a scheme for computing the identity target function over _ any _ mac
@xmath31 exists whenever the elements of @xmath31 take at least @xmath229 distinct values in @xmath15 .
we then argue that , if the assumptions on @xmath92 and @xmath93 in are satisfied , a _ random _
mac @xmath90 ( as introduced in example [ eg : fnmac ] ) of dimension @xmath230 has at least @xmath229 distinct entries with high probability as @xmath91 .
together , this will prove the theorem .
note that implies that @xmath231 and @xmath232 for @xmath95 large enough .
we will prove the result under these two weaker assumptions on @xmath92 and @xmath93 . since we can always choose to ignore part of the channel inputs
, we may assume without loss of generality that @xmath233 . in order to simplify notation
, we suppress the dependence of @xmath234 and @xmath235 on @xmath95 in the remainder of this and all other proofs .
given an arbitrary mac @xmath31 , create a bipartite graph @xmath236 as follows ( see fig .
[ fig : identity ] ) .
let the vertices of @xmath236 on each of the two sides of the bipartite graph be @xmath14 .
now , consider a value @xmath237 appearing in @xmath31 .
this @xmath238 corresponds to a collection of vertex pairs @xmath239 such that @xmath240 .
pick exactly one arbitrary such vertex pair @xmath239 and add it as an edge to @xmath236 .
repeat this procedure for all values of @xmath238 appearing in @xmath31 .
thus , the total number of edges in the graph @xmath236 is equal to the number of distinct entries in the channel matrix @xmath31 .
observe that any complete @xmath195 bipartite subgraph @xmath241 of the bipartite graph @xmath236 corresponds to a computation scheme for the identity function . indeed , by construction each edge in @xmath236 corresponds to a different channel output .
hence by encoding @xmath242 and @xmath243 as the left and right vertices , respectively , of the subgraph @xmath241 , we can uniquely recover @xmath244 from the channel output @xmath245 .
this problem of finding a bipartite subgraph @xmath241 in the bipartite graph @xmath236 is closely related to the zarankiewicz problem , see for example ( * ? ? ?
* chapter vi ) .
formally , the aim in the zarankiewicz problem is to characterize @xmath246 , the smallest integer @xmath247 such that every bipartite graph with @xmath57 vertices on each side and @xmath247 edges contains a subgraph isomorphic to @xmath248 . the kvri - ss - turn theorem ,
see for example ( * ? ? ?
* theorem vi.2.2 ) , states that @xmath249 using , we now argue that the bipartite graph @xmath236 defined above contains a complete @xmath195 bipartite subgraph @xmath241 if the number of edges in @xmath236 is at least @xmath250 . by definition
, @xmath236 contains @xmath241 if there are at least @xmath251 edges in @xmath236 . by , @xmath252 using the inequality @xmath253 for @xmath254 $ ] and that @xmath255 by assumption , we have @xmath256 combining this with shows that @xmath257 thus we have shown that the identity target function can be computed over _ any _ channel @xmath31 with @xmath255 if it has at least @xmath258 distinct entries .
consider now a _
channel @xmath90 .
the next lemma shows that @xmath90 satisfies this condition with high probability as @xmath259 .
[ lemma : distinctentries ] let @xmath260 be the number of distinct entries in the random channel matrix @xmath90 , and assume @xmath261 .
then @xmath262 for @xmath169 large enough .
the proof of lemma [ lemma : distinctentries ] is reported in appendix [ sec : appendix_distinct ] .
lemma [ lemma : distinctentries ] shows that with probability at least @xmath263 the identity target function can be computed with zero error over the random mac @xmath90 . since @xmath264 so that @xmath259 as @xmath91 , the statement of the theorem follows .
let @xmath265 and @xmath266 be the two sets in the definition of a balanced function . for a mac @xmath31 , a @xmath195 ordered submatrix @xmath196 corresponds to a valid code for computing @xmath63 with zero error only if there are no common values between the entries in @xmath196 corresponding to @xmath267 and @xmath268 .
consider then a random @xmath90 ( as introduced in example [ eg : fnmac ] ) and one such ordered submatrix @xmath269 .
observe that the selection of rows and columns of @xmath90 in @xmath269 is fixed the matrix @xmath269 is random because its entries are derived from the random matrix @xmath90 .
let @xmath270 denote the number of distinct values among the entries corresponding to @xmath267 in @xmath269 .
we have the following bound on @xmath270 .
[ lemma : distinctentries2 ] assume @xmath271 , and set @xmath272 then @xmath273 the proof of lemma [ lemma : distinctentries2 ] is reported in appendix [ sec : appendix_distinct2 ] .
the submatrix @xmath269 corresponds to a valid code for computing the target function @xmath63 only if all the entries corresponding to @xmath268 take values from the @xmath274 channel outputs not present in the entries corresponding to @xmath267 .
thus the probability of the submatrix @xmath269 being a valid code for computing the target function @xmath63 is at most @xmath275 where @xmath276 follows from lemma [ lemma : distinctentries2 ] and @xmath277 , @xmath278 follows since @xmath63 is @xmath133-balanced , and @xmath279 follows from the definition of @xmath280 .
the pair @xmath281 is @xmath123-feasible if and only if there exists some valid ordered submatrix @xmath269 with dimension @xmath282 of the @xmath283 channel matrix @xmath90 .
there are at most @xmath284 ways to choose the rows and columns of this submatrix .
hence , from the union bound and , the probability that @xmath281 is @xmath123-feasible is at most @xmath285 now , implies that @xmath286 and @xmath287 for some finite positive @xmath247 and @xmath95 large enough .
hence , @xmath288 as needed to be shown
. a scheme computes the equality target function @xmath289 with zero error if and only if the channel outputs corresponding to the set of message pairs @xmath290 are all distinct from those corresponding to the message pairs @xmath291 . in what follows
, we will first show that this is guaranteed if the channel matrix @xmath31 satisfies certain properties .
we then argue that a random channel matrix @xmath90 ( as introduced in example [ eg : fnmac ] ) satisfies these properties with high probability . from
, we can assume that @xmath292 and @xmath293 for @xmath95 large enough .
we will prove the result under these two weaker conditions .
since the encoders can always choose to ignore some of the channel inputs , we can assume without loss of generality that @xmath294 .
throughout this proof , we denote by @xmath295 the largest integer such that @xmath296 .
note that this implies @xmath297 given an arbitrary mac @xmath31 , create a bipartite graph @xmath236 as follows .
let the vertices on the two sides of the bipartite graph correspond to the @xmath169 different row and column indices of the channel matrix @xmath31 .
fix an arbitrary subset @xmath298 of cardinality @xmath295 of @xmath15 .
place an edge in the bipartite graph @xmath236 between a node @xmath299 on the left and a node @xmath300 on the right if @xmath301 , see fig .
[ fig : equality ] .
an _ induced matching _
@xmath302 in a bipartite graph @xmath236 is a subset of edges such that i ) no pair of edges in @xmath302 share a common endpoint and ii ) no pair of edges in @xmath302 are joined by an edge in @xmath236 .
note that an induced matching @xmath302 of size @xmath95 in @xmath236 corresponds to a zero - error computation scheme for the equality function @xmath289 .
this follows from the observation that the induced matching provides a subset of channel inputs @xmath303 and @xmath304 such that the only pairs of channel inputs for which the channel output is in @xmath298 are given by @xmath305 .
the decoder thus simply maps all channel outputs in @xmath298 to @xmath306 and all other channel outputs to @xmath123 .
a _ strong edge - coloring _ of a graph @xmath236 is an edge - coloring in which every color class is an induced matching , i.e. , any two vertices belonging to distinct edges with the same color are not adjacent .
the strong chromatic index @xmath307 is the minimum number of colors in a strong edge - coloring of @xmath236 .
a simple argument in @xcite shows that for any graph @xmath236 , @xmath308 where @xmath309 is the maximum degree of @xmath236 .
thus , a graph @xmath236 contains an induced matching of size at least @xmath310 where @xmath311 denotes the number of edges in @xmath236 .
consider again the bipartite graph @xmath236 constructed from @xmath31 for some fixed subset @xmath298 . from and the above discussion , we see that @xmath312 is @xmath123-feasible if @xmath313 we now show that this holds with high probability for a random channel matrix @xmath90 . since we consider a random @xmath90 , the graph @xmath314 is itself also random .
we start by deriving a lower bound on the number of edges @xmath315 in @xmath314 .
the event that a particular pair of vertices has an edge in @xmath314 is equivalent to the corresponding entry in the channel matrix @xmath90 being an element of @xmath298 , which happens with probability @xmath316 .
since the @xmath317 entries of @xmath90 are independent , this implies that the number of edges @xmath315 is given by a binomial random variable with mean @xmath318 .
thus , using the chernoff bound , that @xmath319 by assumption , and that @xmath320 by , @xmath321 which converges to zero as @xmath91 .
we continue by deriving an upper bound on the maximum degree @xmath322 of @xmath314 .
let @xmath323 and @xmath324 denote the maximum degree among the left - side and right - side vertices , respectively .
note that @xmath323 and @xmath324 are identically distributed , as the maximum value among @xmath169 independent binomial random variables with mean @xmath325 .
let @xmath326 be one such binomial random variable . using the chernoff bound , @xmath327 by the union bound , and using that @xmath328 by assumption , that @xmath320 by , and that @xmath329 , we have @xmath330 which converges to zero as @xmath91 . using @xmath296 by and the union bound , @xmath331 which , by and , converges to one as @xmath91 .
combined with , this shows that @xmath332 thus proving the claim .
consider an arbitrary target function @xmath63 and an arbitrary channel function @xmath31 .
recall the definition of a @xmath60-approximation function in . from lemma [ lemma :
translation ] , @xmath64 is @xmath60-feasible only if there exists some @xmath60-approximation @xmath333 of @xmath63 such that @xmath334 and @xmath335 is @xmath123-feasible . from the construction in lemma [ lemma :
translation ] , we can assume without loss of generality that @xmath336 . furthermore , we can assume without loss of generality that no two rows and no two columns of @xmath337 are identical .
hence , @xmath338 is @xmath123-feasible only if there exists a @xmath339 ordered submatrix @xmath196 of @xmath340 computing @xmath337 , as described in section [ sec : proofs_preliminaries ] . in the following ,
denote by @xmath341 the mapping corresponding to @xmath196 .
consider now such a @xmath211 ordered submatrix @xmath196 of @xmath56 .
for any @xmath342 , let @xmath343 denote the range of @xmath344 , with the arguments restricted to the subset @xmath345 .
let @xmath346 and @xmath347 be independent random variables uniformly distributed over @xmath348 and @xmath349 , respectively .
consider the random vector @xmath350 & \cdots & \msf{y}[n ] \end{pmatrix } \defeq s(\bm{\msf{v}}_1,\bm{\msf{v}}_2).\ ] ] then , for any @xmath351 , we have @xmath352 and let @xmath353 denote the corresponding entropy of the random vector @xmath354 .
the next result proves the existence of a `` typical '' set .
[ lemma : typical ] let @xmath196 be a @xmath339-dimensional ordered submatrix of @xmath56 , and let @xmath341 be the corresponding mapping . for any @xmath355
, there exists a set @xmath356 such that [ eqn : typical ] @xmath357 the proof of lemma [ lemma : typical ] is presented in section [ sec : proofs_n_typical ] .
consider now the event that @xmath358 is @xmath60-feasible for the random target function @xmath81 ( as introduced in example [ eg : fnmac ] ) . as we have seen before , this implies the existence of a @xmath60-approximation @xmath359 of @xmath81 such that @xmath360 is @xmath123-feasible .
let @xmath196 be the corresponding ordered submatrix of @xmath56 specifying the encoders , and let @xmath62 be the corresponding decoder .
for fixed @xmath361 , let @xmath362 be the typical set associated with @xmath196 , as guaranteed by lemma [ lemma : typical ] .
since @xmath62 correctly computes @xmath363 for all elements of @xmath364 , it does so in particular for all elements of @xmath345 .
more formally , @xmath365 for all @xmath366 .
fix an ordered submatrix @xmath196 of @xmath56 of dimension @xmath339 .
let @xmath345 be the typical set corresponding to @xmath196 .
consider a random @xmath81 , and let @xmath367 be the event that there exists a @xmath60-approximation @xmath359 of dimension @xmath368 and a mapping @xmath369 such that @xmath365 for all @xmath366 . from the discussion in the last paragraph , we have @xmath370 we continue by upper bounding the probability of the event @xmath367 .
fix a mapping @xmath62 and let @xmath371 denote the set of distinct @xmath211 matrices @xmath337 with entries in @xmath20 such that @xmath372 for all @xmath373 . using the union bound
, we have @xmath374 the number of matrices in @xmath371 is at most @xmath375 where the second inequality follows from in lemma [ lemma : typical ] .
since there are at most @xmath376 mappings @xmath62 from @xmath377 to @xmath20 , we have @xmath378 where the last inequality follows from in lemma [ lemma : typical ] .
consider then a fixed matrix @xmath337 .
the next lemma upper bounds the probability that this fixed @xmath337 is , in fact , a @xmath60-approximation of the random target function @xmath81 .
[ lemma : approximationmatrix ] fix @xmath379 and an arbitrary matrix @xmath337 of dimension @xmath339 with @xmath380 and range of cardinality @xmath141 .
let @xmath81 be the random target function of dimension @xmath140 and range of cardinality @xmath141 .
then @xmath381 with @xmath382 the proof of lemma [ lemma : approximationmatrix ] is presented in section [ sec : proofs_n_approximation ] . combining , , and lemma [ lemma : approximationmatrix ] shows that for any @xmath196 , @xmath383 substituting the above into , we have @xmath384 where the last inequality follows by noting that there are at most @xmath385 ordered submatrices @xmath196 of @xmath56 of dimension at most @xmath140 .
now , set @xmath386 and note that @xmath387 as @xmath91 since @xmath388 as @xmath91 .
recall that @xmath389 by assumption .
this implies that @xmath390 hence , the right - hand side of converges to zero as @xmath391 if the following two conditions hold , @xmath392 in particular , since @xmath393 without loss of generality , and since @xmath394 as @xmath395 , this is the case whenever @xmath396 thus , if @xmath397 then either @xmath398 or there must exist a submatrix @xmath196 of of @xmath56 of dimension at most @xmath140 such that @xmath399 assume that the latter is true . let @xmath400 denote independent random variables corresponding to the channel inputs of the two users , as specified by the submatrix @xmath196
. then we have @xmath401 which implies that @xmath402 similarly , we have @xmath403 from , , , it follows that there exists a joint distribution on @xmath404 of the form
@xmath405 | v_1[i ] , v_2[i])\ ] ] which satisfies @xmath406 ) } , \nonumber\\ u & \dotleq 2^ { h(\bm{\msf{y } } | \bm{\msf{v}}_1 ) } \leq 2^{\sum_{i=1}^{n } h(\msf{y}[i ] | \msf{v}_1[i ] ) } , \nonumber\\ u & \dotleq 2^ { h(\bm{\msf{y } } | \bm{\msf{v}}_2 ) } \leq 2^{\sum_{i=1}^{n } h(\msf{y}[i ] | \msf{v}_2[i])}.\end{aligned}\ ] ] we can then single - letterize the right - hand side of the above inequalities in the usual way , see for example the proof of ( * ? ? ?
* theorem 14.3.3 ) .
then it follows that there exists a joint distribution of the form @xmath407 such that @xmath408 thus , if holds , then the above inequalities provide a list of necessary conditions for @xmath409 to be @xmath60-feasible with positive probability .
it follows easily that the conditions above are also implied if the alternate condition in holds , thus concluding the proof .
it remains to prove lemmas [ lemma : typical ] and [ lemma : approximationmatrix ] .
consider a variable - length binary huffman code for the random variable @xmath354 distributed according to , and let @xmath410 be the length of the codeword associated with @xmath354 . by
* theorem 5.4.1 ) , the expected length @xmath411 of the code satisfies @xmath412 let @xmath413 denote the set of @xmath30 such that @xmath414 for some @xmath355 , and define @xmath415 as the elements in @xmath416 that are mapped into @xmath417 .
we have @xmath418 where @xmath276 follows since there are at most @xmath419 binary sequences of length at most @xmath420 and each of them can correspond to at most one value @xmath421 , and @xmath278 follows from . on the other hand
, we have @xmath422 where @xmath276 follows from and @xmath278 follows from markov s inequality . together
, this shows the existence of a set @xmath345 with the desired properties .
fix an arbitrary function @xmath423 with @xmath424 , and fix arbitrary maps @xmath425 and @xmath426 .
denote by @xmath427 the indicator variable of the event @xmath428 in words , @xmath429 if the target function @xmath430 is correctly approximated by @xmath208 at @xmath244 .
since the entries of @xmath81 are uniformly distributed over @xmath20 , we have @xmath431 since the entries of @xmath81 are independent , the number of message pairs for which the target function @xmath432 is correctly computed using the approximation function @xmath208 is then described by a binomial random variable @xmath433 with mean @xmath434 .
thus the probability that for fixed maps @xmath435 , the function @xmath208 is a @xmath60-approximation of the random target function @xmath432 is given by @xmath436 where @xmath276 follows from the chernoff bound and @xmath278 from the definition of @xmath437 .
the number of possible maps @xmath61 are at most @xmath438 and @xmath439 , respectively . by the union bound ,
the probability that the function @xmath208 is a @xmath60-approximation of the random target function @xmath432 is then at most @xmath440 thus concluding the proof .
we prove this result by posing it in the framework of the coupon - collector problem , see , e.g. , ( * ? ? ?
* chapter 3.6 ) . in each round ,
a collector obtains a coupon uniformly at random from a collection of @xmath127 coupons .
let @xmath326 denote the number of rounds that are needed until the first time @xmath258 distinct coupons are obtained .
then the event that @xmath326 is at most @xmath317 is equivalent to the number of distinct entries in the @xmath283 random channel matrix @xmath90 being at least @xmath258 .
let @xmath441 for the coupon collector problem , the minimum number of rounds @xmath326 needed to collect @xmath280 distinct coupons can be written as @xmath442 where the @xmath224 s are independent random variables and @xmath443 represents the geometric distribution with parameter @xmath444 .
observe that @xmath445 . from the chernoff bound , @xmath446 we have @xmath447 for @xmath448 .
since the right - hand side is decreasing in @xmath444 , we have @xmath449 for every @xmath450 .
this implies that @xmath451 since @xmath261 by assumption , we have @xmath452 from .
assume that this is the case in the following , and set @xmath453 in .
then @xmath454 now , since @xmath455 , we have @xmath456 , and thus @xmath457 here , the last inequality follows by setting @xmath458 in the inequality @xmath459$},\end{aligned}\ ] ] which follows from the observation that the left - hand side evaluates to zero at @xmath460 and is monotonically increasing for @xmath461 $ ] . substituting into and using the definition of @xmath280 yields @xmath462 thus concluding the proof .
consider again the coupon collector problem as in appendix [ sec : appendix_distinct ] , and let @xmath326 denote the number of rounds required to collect @xmath280 distinct coupons .
then the event that @xmath326 is at least @xmath463 is equivalent to @xmath270 being at most @xmath280 . following the proof of lemma [ lemma : distinctentries ] , we have from the chernoff bound that @xmath464 where the last inequality follows since @xmath465 by assumption , and with @xmath466 m. p. wilson , k. narayanan , h. d. pfister , and a. sprintson , `` joint physical layer coding and network coding for bidirectional relaying , '' _ ieee trans .
inf . theory _ ,
56 , no . 11 , pp . 56415654 , nov . | we consider the problem of distributed computation of a target function over a two - user deterministic multiple - access channel .
if the target and channel functions are matched ( i.e. , compute the same function ) , significant performance gains can be obtained by jointly designing the communication and computation tasks . however , in most situations there is mismatch between these two functions . in this work ,
we analyze the impact of this mismatch on the performance gains achievable with joint communication and computation designs over separation - based designs .
we show that for most pairs of target and channel functions there is no such gain , and separation of communication and computation is optimal . |
the coherent manipulation of quantum spins in either silicon or diamond represents two promising approaches to the development of a raft of quantum technologies including not only quantum computing but also sensing , metrology and magnetometry . in diamond ,
the remarkable properties of nitrogen vacancy ( nv ) colour centres for spin - dependent optical read - out and polarization are the cornerstones of a large number of proposed applications at the single spin level.@xcite in silicon , shallow donors ( mainly group v atoms including phosphorus ) provide coupled electron - nuclear spin systems which form the basis for the seminal proposal of kane for scalable silicon - based quantum computing.@xcite there has been much recent progress in single - spin detection and read - out,@xcite complementing studies on ensembles .
@xcite strong mixing between the donor electronic and nuclear spins leads to `` sweet - spots '' of enhanced electronic spin coherence , even in natural silicon , first investigated theoretically , @xcite and also with experiments.@xcite natural silicon comprises mostly spin - free @xmath4si isotopes , but @xmath5 of sites hold @xmath6si impurities with @xmath7 nuclear spins .
similarly , natural diamond is mostly spin - free @xmath8c , but @xmath9 of the atomic sites are occupied by @xmath10c nuclei for which @xmath7 . in these samples ,
@xmath6si or @xmath10c nuclei represent the major source of decoherence.@xcite however , interest in these impurities has now moved far beyond their role as a destructive source of decoherence to applications ranging from sensing of a few nuclear spins,@xcite to the very recent demonstrations of quantum registers combining the central electronic qubit with proximate nuclear spins.@xcite a pair of proximate nuclear spins can interact not only via direct dipolar coupling , but also via longer - ranged interactions mediated by the central electronic spin . in either case ,
the nuclear spins of the pair may ` flip - flop ' and the resulting magnetic noise provides a well - known source of decoherence for both electronic and nuclear qubits .
however , in the case of strong hyperfine coupling between the nuclear impurity pair and the electron spin , the resulting energy detuning on each partner of the pair overwhelms the dipolar coupling , suppressing spin flip - flopping and the associated decoherence within a so - called `` frozen core '' region .
the frozen core @xcite is a well - established concept in electron spin resonance ( esr ) studies but is now attracting new interest as a reservoir of protected qubits.@xcite the boundary radius @xmath11 of the frozen core is commonly set as the distance at which hyperfine coupling strengths have decreased to values comparable to the dipolar interactions between neighbouring nuclear spins.@xcite representative values of the latter may be inferred from measured linewidths ; for example , the @xmath12 hz linewidth of @xmath0si in natural silicon @xcite corresponds to an estimated @xmath13 for the phosphorus donor system ( si : p ) which exemplifies our study .
hyperfine couplings of proximate @xmath6si _ sites _ in an ensemble were already resolved spectroscopically in 1969 by continuous wave esr,@xcite and their coherence can now be investigated with , for example , pulsed endor techniques.@xcite for @xmath6si nuclear spins far outside the frozen core the value of the echo decay time @xmath14 ms has been measured.@xcite very recently,@xcite the @xmath3 of a proximate _ single _
@xmath6si spin was measured for the case of an ionized donor to be 6.4 ms , close to the measured ensemble / bulk value reported in ref . .
however , for the single spin , @xmath3 in the presence of the neutral donor ( i.e. within the frozen core due to the electron ) was not measured .
this parameter is however of interest to potential future realisations using @xmath6si nuclear spin registers in combination with electronic qubits , analogous to the recent studies of nv@xmath15 centres with @xmath10c nuclear spins.@xcite exceeds the strongest intrabath coupling @xmath16 ; it thus corresponds to the usual definition of the `` frozen - core '' region . however , electronic spin decoherence is dominated by an active zone ( purple colour ) of pairs of nuclear spins which are actually _ within _ the blue _ strongly _ detuned region , with @xmath17 for si : p .
the reason is that , while for large @xmath18 flip - flop amplitudes are strongly damped , qubit state - dependence of the quantum bath evolution , essential for the entanglement between the electronic spin and bath which produces decoherence , is also proportional to @xmath19.,width=259 ] : there is now potentially a very large _ state - independent _ component @xmath20 which simply damps the bath noise , in addition to a _ state - dependent _ component @xmath21 which leads to qubit - bath entanglement and thus decoherence . * ( a ) far bath model * for large @xmath22 , the bath spin interaction with both the electron spin and nuclear qubit is dipolar , thus @xmath23 so very weak contributions from an extremely large bath of @xmath24 pairs for @xmath25 must be combined to obtain a converged decay . *
( b ) equivalent pairs model*. in the frozen core there are comparatively few spin impurities . for equivalent pairs however , @xmath26 so @xmath27 .
their density is determined by the symmetry of the electronic wavefunction .
the requirement for strong state - selective detuning implies also that one member of the pair must be close enough to the qubit to allow appreciable direct dipolar coupling ( as opposed to long - range coupling between nuclear spins mediated by the electron ) .
pairs which also satisfy this requirement ( exemplified by the upper , but not the lower , equivalent pair ) are rare but even a few dozen suffice to exceed the contribution of the @xmath28 far - bath spin pairs.,title="fig:",width=244 ] : there is now potentially a very large _ state - independent _ component @xmath20 which simply damps the bath noise , in addition to a _ state - dependent _ component @xmath21 which leads to qubit - bath entanglement and thus decoherence . *
( a ) far bath model * for large @xmath22 , the bath spin interaction with both the electron spin and nuclear qubit is dipolar , thus @xmath23 so very weak contributions from an extremely large bath of @xmath24 pairs for @xmath25 must be combined to obtain a converged decay . *
( b ) equivalent pairs model*. in the frozen core there are comparatively few spin impurities . for equivalent pairs
however , @xmath26 so @xmath27 .
their density is determined by the symmetry of the electronic wavefunction .
the requirement for strong state - selective detuning implies also that one member of the pair must be close enough to the qubit to allow appreciable direct dipolar coupling ( as opposed to long - range coupling between nuclear spins mediated by the electron ) .
pairs which also satisfy this requirement ( exemplified by the upper , but not the lower , equivalent pair ) are rare but even a few dozen suffice to exceed the contribution of the @xmath28 far - bath spin pairs.,title="fig:",width=244 ] although it has long been known that the large energy detunings in the frozen core drastically suppress nuclear dynamics , this regime has not to date been investigated using more recently developed quantum - bath models , where decoherence is understood in terms of entanglement between the central spin and bath .
this leads to clear differences between decoherence of the electronic and proximate qubits , even if in both cases the same nuclear bath is responsible for the decoherence .
the differences are summarised in figs .
[ fig1 ] and [ fig2 ] and motivate a more careful look at what is meant by the frozen core and where , precisely , its boundaries lie .
for example , for the electronic qubits , decoherence is in fact dominated @xcite by impurities which lie within the usual definition of the frozen core , as illustrated in fig .
[ fig1 ] , since the detuning fully contributes to entanglement . in the present study
, we consider two decoherence models for the proximate nuclear qubits , either of which , given certain assumptions , might contribute .
both arise from pair flip - flops of the nuclear impurities , but under quite extreme conditions , not encountered in typical qubit decoherence studies . _
( 1 ) far bath model .
_ in this model , we consider the decoherence from distant nuclear spin pairs , which are outside the frozen core and thus can flip - flop appreciably .
we show that the typical contribution is so weak that we must include of order @xmath24 flip - flopping pairs outside the frozen core , at distances @xmath29 from the donor site , in order to obtain results converged with respect to bath size .
in contrast , typical quantum - bath calculations of electronic decoherence require @xmath30 pairs to obtain convergence . _ ( 2 ) equivalent pairs model . _ in this model ,
the dephasing noise arises from a few dozen nuclear spin pairs , well within the frozen core , for which : ( i ) the members of the pair are symmetrically sited relative to the central spin and thus have equivalent values of the hyperfine detuning ( ii ) at least one member is sufficiently close to the nuclear qubit to have a significant dipolar interaction , while the other can be remote .
they interact via a long - ranged hyperfine interaction mediated by the electron .
the indirect flip - flopping of these equivalent pairs ( eps ) is found to be most significant , but we include also the rarer contribution of direct flip - flops between the nuclear central spin and any equivalent partner it might have .
we obtain @xmath3 values in the seconds timescale both for individual realisations ( relevant to single donor experiments ) and also for ensemble averages over many realisations . though not previously considered , this source of dephasing is quite generic : equivalent sites may play a role in any solid state qubit system with a sufficiently dense surrounding nuclear bath .
quantum - bath decoherence calculations do not include all combinatorially allowed spin clusters : there are in principle @xmath31 @xmath6si spin pairs within @xmath32 .
fortunately , only a smaller fraction are physically significant .
these are found by numerical search of each randomly populated lattice realization by restricting the selection to , for instance , pairs within a certain distance and coupling strength .
however , applying normal distance / coupling strength thresholds turned out to be unreliable in the frozen core , so a different strategy had to be adopted to ensure convergence , given the importance of carefully locating the few dozen or so most important eps which can be quite widely separated .
there is a drastic difference between the choice of spin clusters which must be included in the quantum bath for each case ( a few dozen for the ep model , @xmath28 for the far bath model ) ; surprisingly , although the decay curves have a different shape , the @xmath3 values are comparable .
if electronic symmetries are important , then a strategy for breaking such symmetries , with external fields might be considered to obtain an even longer @xmath3 ; if the far bath is dominant , partial isotopic enrichment might be more useful ( consideration of @xmath6si nuclear spin registers in the present study naturally precludes full enrichment ) .
either way , the combined effect is still a @xmath33 s even without any other strategy for ameliorating decoherence , such as dynamical decoupling control .
we note that below 5 k , the electronic relaxation time @xmath34 is also above 1 s and would not limit the nuclear spin coherence . in section
ii , we review briefly quantum bath decoherence , for either nuclear or electron qubits . in section iii.a .
, we investigate the far bath model and in section iii.b we present the ep model .
finally , we compare @xmath3 values for both models and conclude in section iv .
the decoherence of the central electronic spin qubit is extremely well - studied.@xcite a dephasing process arising from noise due to flip - flopping nuclear spin pairs is responsible as illustrated in fig . 1 .
this basic pair flip - flop mechanism also underlies the decoherence of the proximate nuclear spins which are the subject of the present study .
thus it is useful to review briefly the pair decoherence mechanism within the quantum bath approach , for both cases .
the decoherence numerics here employ the cce ( cluster correlation expansion ) method for solving for quantum bath decoherence,@xcite which in general can combine clusters of all sizes , not just pairs .
echo decays of donor electronic spins , in weak electron - nuclear mixing regimes , can be well simulated with just the pair correlation.@xcite where the central and bath spins are the same species , higher correlations arising from larger clusters may be required for high accuracy,@xcite but in both our models would represent only a minor quantitative correction .
it is only in certain regimes of donors with strong mixing of the central - spin states , @xcite that a truly qualitative effect on the coherence times , arising from larger clusters , is evident . for either electronic @xmath35 or nuclear @xmath36 spin decoherence ,
dephasing arises from entanglement with a quantum bath .
the qubit is prepared in a superposition of its upper @xmath37 and lower @xmath38 states by an applied resonant @xmath39 pulse at @xmath40 : @xmath41 where @xmath42 denotes the initial spin state which is _ not _ resonant with the external control pulses , while @xmath43 denotes an initial state of the bath .
we now consider the effect of a single nuclear impurity pair ( both @xmath7 spins ) ; then , @xmath43 is one of the four thermal states : @xmath44 or @xmath45 .
the joint dynamics of the qubit spins and bath are given by : @xmath46 where the qubit hamiltonian @xmath47 represents the zeeman terms for the electron and nuclear qubit ( labelled @xmath48 ) , either one of which is resonant with any applied control field . here
we limit ourselves to the unmixed cases where any coupling between the qubit and the host nuclear spin is neglected .
the interaction hamiltonian : @xmath49 represents the hyperfine coupling @xmath50 between the central spin and bath spins and the dipolar interaction @xmath51 between the resonant nuclear qubit and the remainder of the 2-spin bath . for simplicity , we do not include explicitly the term @xmath52 coupling the electron to the resonant nucleus , which is only significant in the specific ( but minor ) contribution from direct flip - flopping processes . finally , the intrabath coupling @xmath53 , represents the dipolar coupling between the bath nuclei in secular form ( see appendix a for details ) . under the action of the full hamiltonian eq .
( [ htot ] ) the product state in eq .
( [ product ] ) evolves into an entangled state : @xmath54 \tfrac{\left| \phi(t ) \right\rangle_{e , n } } { \sqrt{2}}. \label{wf}\ ] ] the coherence decay is calculated from either @xmath55 or @xmath56 . for both cases , @xmath57 , so disregarding a constant prefactor @xmath58 or @xmath59 , the coherence has a ( real ) decaying envelope for the normalized @xmath60-th bath spin pair contribution : @xmath61 the measured decays , whether hahn echos or more complex dynamical decoupling sequences , are obtained from the combined contributions from many spin pairs : @xmath62 where the average is over all possible initial bath states , and if required , an ensemble average over different spatial realizations of the bath .
an accurate and insightful simplification arises if @xmath63 or if @xmath64 where @xmath65 denote typical values of the qubit - bath interaction . in that case , the interaction hamiltonian in eq .
( [ hint ] ) reduces to the ising form @xmath66 , a simple diagonal energy shift in @xmath67 terms relative to @xmath68 or a detuning of bath spin 1 relative to bath spin 2 . referred to as the pseudospin model,@xcite in this case , @xmath45 bath states
do not contribute while the dynamics of the @xmath44 states is given by an effective state - dependent hamiltonian : @xmath69 with the detuning @xmath70 for an electronic qubit and @xmath71 for a nuclear qubit . in the latter case , the electronic detuning @xmath72 represents a potentially large contribution which is not sensitive to the qubit state . from eq .
( [ equa8 ] ) decays ( indirect and direct flip - flops ) are deduced analytically and for the hahn echo case we obtain : @xmath73 where @xmath74 , @xmath75 while @xmath76 and the eigenvalues @xmath77 , and we have dropped the @xmath60 indices for convenience .
the larger @xmath78 , the larger the amplitudes of the flip - flopping of nuclear spin - pairs which drives the decoherence .
when considering the contribution of flip - flopping pairs which are within the frozen core ( the ep model below ) , we obtained excellent agreement between the pseudospin equations above and full numerical cce provided that the well - known perturbative correction for the non - ising ( anisotropic ) hyperfine terms,@xcite was added to the dipolar coupling when using eq .
( [ decay ] ) .
in other words , for the @xmath60-th cluster , in equations ( [ equa8 ] ) and ( [ decay ] ) we take : @xmath79 where , @xmath80 is the dipolar interaction between nuclear spins , allowing bath spins a long - ranged interaction mediated by the central electron , the second term in eq .
( [ mediate ] ) . for a full cce calculation , @xmath81 since
the additional effective long - ranged interaction emerges naturally if the full hyperfine interaction is included . for
non - equivalent pairs in the frozen core , @xmath82 , thus @xmath83 and flip - flops become too strongly suppressed .
the qubit state sensitivity enters in eq .
( [ decay ] ) mainly through the @xmath84 prefactor and is also suppressed by @xmath85 .
this imposes the further condition @xmath86 for a single individual pair to contribute apppreciably to the decay .
central to our modelling is the identification of spin clusters within the frozen core which can contribute non - negligibly to the decoherence of a proximate spin .
we now consider two models and apply them to the particular case of natural si : p .
from numerical simulations with a very large spin bath , we find that distant spin pairs outside the frozen core radius @xmath87 individually make an extremely small contribution to decoherence : the @xmath88 prefactor scales the coherence decays in eq .
( [ decay ] ) , since @xmath89 .
we can also show that the approximate weight of the @xmath60-th pair , is of order @xmath90@xcite assuming also the temporal character of the associated magnetic noise is relevant : in other words , flip - flop frequencies @xmath91 for the given pair can not be orders of magnitude different from @xmath92 . for a non - negligible contribution we would expect that @xmath93 where @xmath94 is a representative number of contributing spin pairs . from the pseudospin model ( leaving out the @xmath60 labels ) , @xmath95 , since @xmath96 .
the first factor ( @xmath97 ) determines whether the pair can flip - flop appreciably and is significant if @xmath98 . the second factor , @xmath99 determines state distinguishability . for the far spins , the hyperfine mediated
correction plays little role since @xmath100 and @xmath101 are small . for distances
@xmath102 , where the fermi contact component of the hyperfine interaction becomes small , the dipolar electronic - nuclear hyperfine interaction still makes a contribution to the detuning which is much larger than @xmath103 . here , @xmath104 thus the contribution of each such far bath spin pair is @xmath105 , so only a far bath with @xmath106 contributing spin pairs can produce significant decay . at very large @xmath22 , however , @xmath107 but there is a minimum value of the interaction @xmath108 where @xmath109 sets a timescale below which the bath noise is too slow to contribute .
as @xmath110 , then @xmath111 thus there is a maximum radius @xmath112 beyond which the far bath does not contribute significantly to decoherence .
hz and a bath of spins within @xmath113 of the origin , combining the contributions from @xmath114 spin pairs .
calculations were performed for the case of si : p , for @xmath115-band and magnetic field orientation @xmath116 $ ] , yielding a @xmath3 of 2 s for a single nuclear @xmath0si spin sited at the origin .
this represents an estimate for the upper bound for the coherence time if the far bath is the dominant process . due to the large nuclear spin bath
, the coherence decays are insensitive to the choice of random spatial realization of the bath .
, width=316 ] we tested this analysis numerically by means of cce calculations using a very large bath of nuclear spin pairs ( excluding eps ) and testing the effect on coherence decays of increasing the size of the bath .
figure [ fig3 ] shows convergence with respect to bath size for hahn echo decays , for a nuclear spin at the origin ( thus expected to give an upper bound on the coherence ) .
the @xmath117 hz bound indicates that the pairs are within @xmath118 of each other and the calculation is converged with respect to bath size if we include @xmath114 spin pairs within @xmath119 of the origin .
the scale of the bath is remarkable , in comparison with comparable electronic decoherence calculations with @xmath120 pairs .
although it is computationally feasible to solve for a bath of this magnitude by cce2 or pseudospin methods , the uniformity of the bath means that it is reliable to evaluate @xmath121 in a smaller but geometrically representative sample of the bath .
in addition no averaging over bath realisations was required : the results are insensitive to whether one has a single spin or an ensemble . the isotropic part of the hyperfine interaction is modeled using the kohn - luttinger ( kl ) wavefunction which is an approximation of the ground state wavefunction of the donor electron developed onto bloch wavefunctions and parametrized by the experimental ionization energy ; for phosphorus donors , @xmath122 ev ( see appendix a for details ) .
we can estimate the local densities of suitable eps in the isotropic case before considering effects from any anisotropies in the hyperfine coupling .
, but the innermost proximate spins typically interact with fewer eps . ]
the si crystal structure can be described by a simple cubic lattice with a parameter @xmath123 and an 8-site basis .
all atomic sites are represented by an integer vector @xmath124 $ ] . in our simulations ,
the full lattice size ranges over @xmath125 $ ] cubic cells for each dimension , resulting in 8@xmath126 unit cells and hence 64@xmath126 total atomic sites . owing to the symmetry of the system
, each site possesses several potential _ equivalent _
partners , for which positions can be deduced from any allowed permutations of @xmath127 $ ] .
we can assign each vector @xmath128 to a shell @xmath129 comprising @xmath130 or @xmath131 partners and we first obtain @xmath132 , the number of shells comprising @xmath133 partners within a radius of @xmath134 from the center .
the ranges were adjusted to ensure summation over _ complete _ shells ( see appendix b for details ) and we obtain estimates for @xmath132 : @xmath135 while @xmath136 . then assuming a binomial distribution , taking an abundance of @xmath137 for nuclear spin impurities in natural silicon , the estimated average number of significant ep in each shell is : @xmath138 where @xmath139 is the binomial coefficient .
for the two dominant shells @xmath140 and @xmath141 . in these cases
, it is quite likely that any impurity spin has an equivalent partner _ somewhere _ , albeit remotely located .
nevertheless , due to the long - range electron - mediated coupling , a @xmath16 of about tens of hz is present . within a sphere of radius @xmath142 cubic cells
, we expect the total number of ep to be simply : @xmath143 for instance , within a radius of @xmath144 , we find @xmath145 . for a proximate nucleus however , one member of the pair must be dipolar - coupled to the resonant spin ( cf .
caption of fig .
[ fig1 ] ) which is relevant within about @xmath146 cubic cells .
each nuclear qubit thus interacts with other nuclei in the neighbouring @xmath147 cells .
we can define a density of spin pairs : @xmath148 which gives the mean number of eps in each cubic cell as a function of distance , @xmath134 from the electron .
we see in fig .
[ fig4 ] that the mean number for large @xmath22 is about @xmath149 pairs per cubic cell ( cf .
left panel ) , thus each nuclear qubit interacts with @xmath150 potential eps if anisotropy is neglected .
mhz and * ( b ) * @xmath151 mhz ; blue : isotropic hyperfine coupling ; red : includes anisotropy correction . * ( c ) * calculated @xmath3 values .
there is a weak trend for @xmath3 to increase as the hyperfine coupling increases ( red line is a fit ) , possibly indicative of the decreasing density of eps as @xmath152 . in the far bath model ( grey dots ) ,
the slight increase in decoherence with lower @xmath153 ( grey line ) reflects the fact that the lower @xmath153 proximate spins are slightly closer to the far bath .
coherence times were obtained from decays averaged over 100 spatial realizations of the bath , but typical single realizations gave the same timescale of decoherence .
, width=316 ] for the numerical calculations of the echo decays , we carried out a careful search , retaining about 500 equivalent spin pairs and averaging over 100 realisations of a randomly generated lattice population with @xmath5 of sites occupied by @xmath6si spins .
two sets of calculations of the hahn echo decays were carried out .
the first employed only the isotropic contact interaction and neglected anisotropic components of the hyperfine interaction .
these calculations provide a lower bound for the @xmath3 and predicted decay rates @xmath154 s for different values of @xmath153 ( see appendix c for further details ) . a second set of calculations attempts to account for the anisotropy , which is less easy to calculate reliably .
we assumed that any degree of anisotropy detunes spin pairs so much that their contribution became negligible . in effect , this model provides an upper bound for the expected @xmath3 as not all shells are affected by anisotropy .
thus to remain an equivalent pair we required that spins have the same @xmath155 , where @xmath156 is the direction of the magnetic field .
the effect is to reduce the symmetries but to increase the number of shells , i.e. for @xmath157 $ ] and the main shells with @xmath158 partners , we have @xmath159 and therefore @xmath160 ( cf . fig .
[ fig4 ] right panel ) .
we have calculated coherence decays for a representative set of proximate nuclear spins by comparing the effect of a large far bath model with a new model introduced here , based on the effect of symmetrically - sited spin pairs , the equivalent pairs ( eps ) .
the corresponding @xmath3 are shown in fig .
[ fig5 ] along with the comparable far bath results .
we found that , in both cases , @xmath3 is of order 1 second with a weak dependence on @xmath153 in both cases , the coherence times tend to increase with larger @xmath153 , a trend also seen in recent experiments.@xcite for the ep model we treat the anisotropic correction simply as a symmetry lowering effect ; this is plausible as the resultant detuning would be extremely large .
presently , it is not possible to fully include anisotropy using the kl wavefunction ; typically , the dipolar correction within this framework is included with a heaviside function,@xcite and is thus neglected for @xmath161 .
given other uncertainties , these two ep calculations provide an upper and a lower bound to @xmath3 .
as both results are on the seconds timescales , they suffice for the practical aim of establishing the proximate nuclear spins as useful qubits . if the si : p wavefunction exhibits a degree of spatial symmetry comparable with the kl wavefunction , then the eps could be the dominant mechanism , albeit only slightly .
however , it is likely that such symmetries are at least partly broken ; in that case , the far bath would limit @xmath3 . given the uncertainties in the kl wavefunction , at present it is not possible to determine accurately the contributions of eps relative to the far bath , but as fortuitously the timescales are comparable , one can still conclude that the resulting @xmath3 is about 1 second . to facilitate comparison with ensemble experiments ,
the ep results are averaged over many realizations ( the far bath model coherence decays are fully insensitive to ensemble averaging ) . in the ep model , decoherence is primarily due to an indirect flip - flop process and this arises from several dozen such eps .
thus , although results from single donors fluctuate between realisations , the corresponding order of magnitude for @xmath162 remains on the one second timescale , whether ensemble averaging is carried out or not .
the exception is the atypical realization where the central spin happens to have an equivalent site it can directly flip - flop with . for proximate central spins , usually in inner shells with @xmath163
this is unlikely .
we find that the small subset of such realizations decohere rapidly .
they contribute little to the ensemble averaged @xmath33 s values but would clearly be unsuitable as qubit registers unless some strategy to exploit the degeneracy is envisaged .
we have investigated the coherence of nuclear spins lying within the so - called `` frozen core '' within a quantum bath framework . we have calculated the coherence using a very large far bath of spins lying outside the frozen core .
we also introduced here a new model , based on equivalent pairs ( eps ) deep within the so - called `` frozen core '' , which we argue would limit the phase coherence of proximate nuclear spins provided the electronic wavefunctions has the symmetries of the kl wavefunction ( or even an alternative model with comparable levels of symmetry ) . within the ep model , decoherence is primarily due to an indirect flip - flop process arising from a few dozen such eps .
our quantitative results are indicative : a more refined investigation should consider improved wavefunctions , more accurate than the kl wavefunction.@xcite experimental investigation of the behaviour including dependence on symmetry - breaking mechanisms ( such as crystal orientation and strain ) will be useful to test this proposal .
our models predicts @xmath0si nuclear spin coherence times of order 1 second ( using only a hahn echo ) , which is consistent with recent experimental measurements,@xcite showing such spins could be useful as potential qubits .
_ acknowledgements _ we are indebted to kohei itoh , steve lyon , brendon lovett , john morton , giuseppe pica , jarryd pla and martin uhrin for useful discussions .
r. g. is supported by epsrc collaborative computational project q grant ep / j010529/1
. s. j. b. is supported by the stocklin - selmoni studentship through the ucl impact programme .
the dipolar interaction strength @xmath164 between two nuclear spins @xmath165 and @xmath166 with gyromagnetic ratios @xmath167 and @xmath168 , defined in the main text ( @xmath16 ) is given by @xmath169 where @xmath170 the inter - nuclear separation , @xmath171 na@xmath172 and @xmath173 denotes the angle between the magnetic field and the line connecting the two nuclei.@xcite the hyperfine interaction is the magnetic interaction between an electronic spin @xmath174 and a localized nucleus @xmath175 . due to the large mismatch between electronic and nuclear gyromagnetic ratios and a sufficiently strong magnetic field ,
it can be written in secular form:@xcite @xmath176 where @xmath177 is the electron - nuclear separation vector , @xmath178 and the electron - nuclear dipolar interaction @xmath179 is non - zero for @xmath180 ( @xmath181 for si : p ) . the first term in eq .
( [ fullhyperfine ] ) is the _ isotropic _ fermi contact interaction . for a donor electron interacting with a nucleus , the strength @xmath153 is given by ( see e.g. ref . ) @xmath182^{2}.\end{aligned}\ ] ] here , @xmath183 and @xmath184 with lattice constant @xmath185 .
the prefactor @xmath186 contains the electronic and nuclear gyromagnetic ratios ( @xmath187 and @xmath188 ) , and the charge density on each atomic site @xmath189 , @xmath190 the relevant envelope functions of the kohn - luttinger ( kl ) model wavefunction are : @xmath191 } } { \sqrt{\pi(na)^{2}(nb)}},\ ] ] @xmath192 @xmath193 where @xmath194 and @xmath195 are lengths characteristic to the donor and @xmath196 with the electron ionization energy @xmath197 in ev .
c c c c @xmath133 & 48 & 24 & 12 + class & & & + 1 & @xmath198 & @xmath199 & @xmath200 + & & & + 2 & @xmath201 & @xmath202 & @xmath200 + & & & + 3 & & @xmath203 & @xmath204 + ( a ) , @xmath205 ( b ) , @xmath206 ( c ) , @xmath207 ( d ) , @xmath208 ( e ) , @xmath151 ( f ) mhz .
the blue lines correspond to isotropic coupling only and yield @xmath209 s ; red lines show the effect of symmetry reduction due to the anisotropy of couplings : we compare the effect of the further desymmetrisation if we constrain ep to have in addition the same orientation condition ( same @xmath155 as discussed in main text ) .
the effect is to produce @xmath3 in the seconds timescale.,width=324 ]
equivalent sites are those with the same hyperfine interaction which we obtain using the kl model wavefunction described above .
we begin by considering the allowed coordinates of the impurities in the crystal .
the si crystal structure can be described by a simple cubic structure with lattice parameter @xmath123 and an 8-site basis .
all atomic sites are represented by an integer vector @xmath124 $ ] which are obtained from translations ( modulo 4 ) in all directions of the 8 basis vectors [ 0,0,0 ] , [ 0,2,2 ] , [ 2,0,2 ] , [ 2,2,0 ] , [ 3,3,3 ] , [ 3,1,1 ] , [ 1,3,1 ] , [ 1,1,3 ] . for simplicity
, we can sort these vectors into three classes : class 1 contains [ 0,2,2 ] , [ 2,0,2 ] , [ 2,2,0 ] , class 2 contains [ 0,0,0 ] and class 3 [ 3,3,3 ] , [ 3,1,1 ] , [ 1,3,1 ] , [ 1,1,3 ] . to ensure counting over complete shells described by the basis vectors
, summations must range between @xmath125 $ ] for the 2 coordinates and @xmath210 $ ] for the 0 coordinate of class 1 giving @xmath211 number of sites ; between @xmath125 $ ] for class 2 giving @xmath212 number of sites and between @xmath210 $ ] for class 3 giving @xmath213 number of sites . owing to the symmetry of the system
, each site possesses several _ equivalent _ partners with positions which can be deduced by permutations of @xmath214 $ ] and which lie on the surface of shells of radius @xmath215 . by consideration of the symmetries of the kl wavefunction we can assign each vector @xmath128 to a group of @xmath130 or @xmath131 partners . for each class , the contribution to a shell comprising @xmath133 partners within a radius of @xmath134 of the center as a function of @xmath142 is summarized in table [ tab1 ] .
additionally , class 2 contributes as @xmath216 to @xmath217 and as @xmath218 to @xmath219 and class 3 contributes as @xmath216 to @xmath220 .
n. zhao , j. honert , b. schmid , m. klas , j. isoya , m. markham , d. twitchen , f. jelezko , r .- b .
liu , h. fedder and j. wrachtrup , nature nano .
* 7 * 657662 ( 2012 ) . s. kolkowitz , q. p. unterreithmeier , s. d. bennett and m. d. lukin , phys . rev . lett . * 109 * 137601 ( 2012 )
. h. bernien , b. hensen , w. pfaff , g. koolstra , m. s. blok , l. robledo , t. h. taminiau , m. markham , d. j. twitchen , l. childress and r. hanson , nature * 497 * 8690 ( 2013 ) .
a. morello , j. j. pla , f. a. zwanenburg , k. w. chan , k. y. tan , h. huebl , m. mttnen , c. d. nugroho , changyi yang , j. a. van donkelaar , a. d. c. alves , d. n. jamieson , c. c. escott , l. c. l. hollenberg , r. g. clark and a. s. dzurak nature * 467 * 687 ( 2010 ) .
juha t. muhonen , juan p. dehollain , arne laucht , fay e. hudson , takeharu sekiguchi , kohei m. itoh , david n. jamieson , jeffrey c. mccallum , andrew s. dzurak , andrea morello , nature nanotechnology * 9 * 986 ( 2014 )
. p. cappellaro , l. jiang , j. s. hodges and m. d. lukin , phys . rev
. lett . * 102 * 210502 ( 2009 ) .
g. waldherr , y. wang , s. zaiser , m. jamali , t. schulte - herbrggen , h. abe , t. ohshima , j. isoya , j. f. du , p. neumann and j. wrachtrup , nature * 506 * 204207 ( 2014 ) . | hybrid qubit systems combining electronic spins with nearby ( `` proximate '' ) nuclear spin registers offer a promising avenue towards quantum information processing , with even multi - spin error correction protocols recently demonstrated in diamond .
however , for the important platform offered by spins of donor atoms in cryogenically - cooled silicon , decoherence mechanisms of @xmath0si proximate nuclear spins are not yet well understood .
the reason is partly because proximate spins lie within a so - called `` frozen core '' region where the donor electronic hyperfine interaction strongly suppresses nuclear dynamics .
we investigate the decoherence of a central proximate nuclear qubit arising from quantum spin baths outside , as well as inside , the frozen core around the donor electron .
we consider the effect of a very large nuclear spin bath comprising many ( @xmath1 ) weakly contributing pairs outside the frozen core .
we also propose that there may be an important contribution from a few ( of order @xmath2 ) symmetrically sited nuclear spin pairs ( `` equivalent pairs '' ) , which were not previously considered as their effect is negligible outside the frozen core .
if equivalent pairs represent a measurable source of decoherence , nuclear coherence decays could provide sensitive probes of the symmetries of electronic wavefunctions . for the phosphorus donor system
, we obtain @xmath3 values of order 1 second for both the `` far bath '' and `` equivalent pair '' models , confirming the suitability of proximate nuclei in silicon as very long - lived spin qubits . |
cuprates@xcite , first discovered in 1986 , and iron - based superconductors@xcite(ibs ) , discovered in early 2008 , are two classes of unconventional high temperature superconductors who share many common features .
both of them are quasi - two - dimensional and the phase diagrams are similar in which superconductivity develops after a magnetic order is suppressed@xcite .
all the cuprates share a common structure element cuo@xmath0 plane , where cu atoms form a square lattice@xcite . the ibs share a common fe@xmath0x@xmath0 ( x = as and se ) layered structure unit , which possesses an anti - pbo - type ( anti - litharge - type ) atom arrangement .
the fe@xmath0x@xmath0 layers consist of a square lattice sheet of fe coordinated by x above and below the plane to form face sharing fex@xmath2 tetrahedra@xcite .
one of major questions in the field of high @xmath1 superconductors is whether cuprates and ibs share a common superconducting mechanism@xcite .
the answer to this question may be obtained if we can integrate the characteristics of both superconductors into a single compound so that their relations can be exclusively addressed .
as both structures are featured with layered square lattices with similar in - plane lattice constants , it is possible to design a compound containing both building blocks , cu - o layers of cuprates and fe - as(se ) layers of ibs .
a similar material design has been adopted and a target material nd@xmath2cuo@xmath3fe@xmath0as@xmath0 has been theoretically investigated@xcite .
recently , two new materials ba@xmath0mo@xmath0ag@xmath0se@xmath0 ( m = co , mn)@xcite have been synthesized via solid - state reaction in experiment .
these two compounds , whose structures belong to i4/mmm space group , consist of infinite mo@xmath0 square planes and antifluorite - type ag@xmath0se@xmath0 layers separated by barium . in this paper , motivated by the fact that the mo@xmath0 plane resembles the cuo@xmath0 plane in cuprates and
the ag@xmath0se@xmath0 layer resembles the fe@xmath0x@xmath0 layer in ibs , we consider the substitution of cu and fe for m and ag in the compounds respectively to obtain two possible new compounds , ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 .
we perform density functional calculations to study the stability and basic electronic structures of these new compounds .
we find that these materials integrate basic electronic characteristics of both high @xmath1 superconductors . in ba@xmath0cuo@xmath0fe@xmath0as@xmath0 , the cuo@xmath0 layers are electron - doped while the fe@xmath0as@xmath0 layers are hole - doped .
the situation is reversed in k@xmath0cuo@xmath0fe@xmath0se@xmath0 .
such doping configurations suggest that many possible mixed phases , in particular , magnetic and superconducting phases , may be realized in these materials by introducing additional carriers and applying external pressure .
our calculations are performed using density functional theory ( dft ) employing the projector augmented wave ( paw ) method encoded in the vienna ab initio simulation package ( vasp ) @xcite . both of the local density approximation ( lda ) and generalized - gradient approximation ( gga)@xcite for the exchange correlation functional are used . throughout the work ,
the cutoff energy is set to be 500 ev for expanding the wave functions into plane - wave basis . in the calculations of magnetic properties ,
the lda+u method is used with the effective on - site coulomb u being 7 ev for cu @xmath4 states@xcite . in the calculations ,
the brillouin zone is sampled in the @xmath5 space within monkhorst - pcak scheme@xcite .
the number of these @xmath6 points are depending on the lattice : @xmath7 for the general unit cell with 4 fe atoms and 2 cu atoms and @xmath8 for the @xmath9 unit cell .
we relax the lattice constants and internal atomic positions with both lda and gga , where forces are minimized to less than 0.01 ev / .
the phonon dispersions are calculated using finite displacement method@xcite as implemented in the phonopy code@xcite .
ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 crystallize in a body - centered tetragonal lattice , shown in figure [ structure ] .
the ba ( k ) spacer layer separates the tetrahedra6l fe@xmath0as@xmath0 ( fe@xmath0se@xmath0 ) layers and the square cuo@xmath0 layers .
there are double fe@xmath0x@xmath0 ( x = as and se ) and cuo@xmath0 layers in a unit cell , similar to bafe@xmath0as@xmath0 . in order to predict the structures of ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 ,
the lattices are fully optimized based on the experimental structural parameters of ba@xmath0coo@xmath0ag@xmath0se@xmath0 .
the optimized and experimental structural parameters are summarized in table [ tab_parameter ] .
both of the calculated lattice constants using lda and gga are close to those of ba@xmath0coo@xmath0ag@xmath0se@xmath0 in experiment , which validates the adopted substitution .
as the calculated lattice parameters for ba@xmath0coo@xmath0ag@xmath0se@xmath0 using gga are closer to the experimental data , we perform the following calculations for the two materials using the lattice parameters obtained by relaxation with gga . the binding energy is usually calculated to estimate the stability of new structures . here , the binding energy per atom , @xmath10 , is defined as @xmath11 , in which @xmath12,@xmath13,@xmath14,@xmath15,@xmath16,@xmath17 and @xmath18 are the respective energies per atom of elemental ba , k , cu , o , fe , as and se in the states at standard ambient temperature and pressure .
@xmath19 is the calculated total energy of a unit cell of ba@xmath0cuo@xmath0fe@xmath0as@xmath0 ( k@xmath0cuo@xmath0fe@xmath0se@xmath0 ) .
the obtained binding energies are 1.61 ev and 1.23 ev per atom for ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 , respectively .
both of the binding energies are very close to that of ba@xmath0coo@xmath0ag@xmath0se@xmath0(1.318 ev ) , indicating that the two structures are energetically favorable in experiment and may be synthesized using similar methods . to further test the stability of these two structures
, we calculate their phonon dispersions , shown in figure [ phonon_dispersion ] .
no imaginary frequencies are observed throughout the whole brillouin zone in phonon dispersions , confirming their dynamically structural stability .
cuo@xmath0fe@xmath0as@xmath0 ( k@xmath0cuo@xmath0fe@xmath0se@xmath0 ) .
[ structure ] , scaledwidth=30.0% ] .optimized structural parameters of ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 , using lda and gga in the paramagnetic phase .
the 6th column is the experimental structural parameters of ba@xmath0coo@xmath0ag@xmath0se@xmath0@xcite . [
cols="^,^,^,^,^,^ " , ] cuo@xmath0fe@xmath0se@xmath0 using gga relaxed parameters in the paramagnetic state . [ k_pm ] , scaledwidth=40.0% ] cuo@xmath0fe@xmath0se@xmath0 using gga relaxed parameters in the paramagnetic state .
[ fs_k_pm ] , scaledwidth=30.0% ] compared with the fermi surfaces of ba@xmath0cuo@xmath0fe@xmath0as@xmath0 , we find that the hole pockets at @xmath20 point of k@xmath0cuo@xmath0fe@xmath0se@xmath0 are much smaller while the electron pockets at m point are much larger ( see figure [ fs_k_pm ] ) , which indicate heavy electron doping in the fe@xmath0se@xmath0 layer . from the size of the pockets
, we can estimate that the electron doping concentration with respect to fese is roughly 0.16 electrons per fe atom .
similarly , we can check the consistency required by the charge conservation .
the size of the hole pockets at m point from cuo@xmath0 layers is about 0.32 holes for each cuo@xmath0 layer from their half filling configuration .
the dos at the fermi level attributed to fe 3d orbitals is significantly decreased with such a large electron doping , which may suggest that magnetic order in fe@xmath0se@xmath0 layers is strongly suppressed .
compared with ba@xmath0cuo@xmath0fe@xmath0as@xmath0 , we consider an additional bicollinear magnetic state , which is the ground state in fete@xcite .
the ground state is paramagnetic on fe@xmath0se@xmath0 layers and checkboard afm ordered for cuo@xmath0 layers with a energy gain of 196.0 mev / f.u . and a spin moment of 0.59 @xmath21 per cu .
the band structure and dos are shown in figure [ k_afm ] and the system remains metallic . in this case , the fe 3@xmath22 , cu 3@xmath22 and o 2@xmath23 states dominate the fermi level .
the dos at the fermi level is 7.04 ev@xmath24/f.u . for spin up channel .
cuo@xmath0fe@xmath0se@xmath0 using gga relaxed parameters in the afm state .
[ k_afm ] , scaledwidth=40.0% ]
it is well known that the pairing symmetry is @xmath22-wave for the cuprates and @xmath25-wave for iron pnictides with a potential sign change between the hole and electron fermi pockets .
as the proposed new materials contain both cuo@xmath0 layers and fe@xmath0x@xmath0(x = as , se ) layers and the electronic structures are similar to both , it can be expected that two different pairing symmetries may coexist in a single material .
the proximity effect between two types of layers with different pairing symmetries can result in interesting novel phenomena .
the time reversal symmetry may be broken with mixed s - wave and d - wave pairing symmetries .
introducing additional carriers or applying external pressure in these materials can maximize @xmath1 . for ba@xmath0cuo@xmath0fe@xmath0as@xmath0 ,
the doping of fe@xmath0as@xmath0 layer is 0.15 holes per fe and that of cuo@xmath0 is 0.3 electrons per cu .
the doping for fe@xmath0as@xmath0 layer is near the optimal level but that for cuo@xmath0 layer is unfortunately not in the superconducting region . to realize superconductivity for both layers
, we can introduce hole doping with partial replacement of ba@xmath26 ions by k@xmath27 ions . from the phase diagrams of cuprates and ( ba , k)fe@xmath0as@xmath0
, the above case can be realized in ba@xmath28k@xmath29cuo@xmath0fe@xmath0as@xmath0 ( 0.24 @xmath30 ) by assuming that the fe@xmath0as@xmath0 and cuo@xmath0 layers are equally hole doped . in this doping region ,
the t@xmath31 for fe@xmath0as@xmath0 layers shows little change .
@xmath32 corresponds the optimal doping value for electron - doped cuprates@xcite . for k@xmath0cuo@xmath0fe@xmath0se@xmath0 ,
however , the fe@xmath0se@xmath0 layers are electron doped and the hole doping for cuo@xmath0 layers is not in superconducting region but very close to it .
the superconductivity for both layers can be realized in k@xmath28ba@xmath29cuo@xmath0fe@xmath0se@xmath0 ( 0.10 @xmath33 ) .
this doping region is much wider than that in ba@xmath0cuo@xmath0fe@xmath0as@xmath0 due to the wide hole - doped superconducting region in cuprates .
the optimal doping value for cuo@xmath0 layers is @xmath34 .
the pressure is also an effective way to tune the superconductivity .
in conclusion , we identify two hypothetical compounds ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 containing fe@xmath0x@xmath0 layers and cuo@xmath0 layers , which are the basic structural units of ibs and cuprates , respectively .
the metallic spacer ba(k ) separates the basic units .
the calculations of binding energies and phonon spectrums indicate that they are dynamically stable , ensuring that they may be experimentally synthesized .
the fermi surfaces derived from fe@xmath0as@xmath0(fe@xmath0se@xmath0 ) layers and cuo@xmath0 layers are very similar to those of iron pnictides ( iron chalcogenides ) and cuprates , respectively . with heavy self - doping
, the ground state of ba@xmath0cuo@xmath0fe@xmath0as@xmath0 is determined to be e - type collinear afm state in the fe@xmath0as@xmath0 layer and paramagnetic state in the cuo@xmath0 layer , while k@xmath0cuo@xmath0fe@xmath0se@xmath0 favors checkboard afm state in cuo@xmath0 layers and paramagnetic state in fe@xmath0se@xmath0 layers .
without external doping , superconductivity can only be achieved in fe@xmath0x@xmath0 layers . however , with external doping through substitution , superconductivity for both fe@xmath0x@xmath0 and cuo@xmath0 layers can be simultaneously achieved in ba@xmath28k@xmath29cuo@xmath0fe@xmath0as@xmath0 ( 0.24 @xmath30 ) and k@xmath28ba@xmath29cuo@xmath0fe@xmath0se@xmath0 ( 0.10 @xmath33 ) .
the synthesis of these new compounds will not only provide us a unique opportunity to explore exotic properties in cuprates and ibs simultaneously but also help us to understand the mechanism of high t@xmath31 superconductivity .
_ acknowledgments : _ the work is supported by the ministry of science and technology of china 973 program ( no .
2015cb921300 ) , national science foundation of china ( grant no .
nsfc-1190020 , 11334012 ) , and the strategic priority research program of cas ( grant no .
xdb07000000 ) . j. g. bednorz and k. a. muller , z. phys .
b * 64 * , 189 ( 1986 )
. k. kishio , k. kitazawa , s. kanbe , i. yasuda , n. sugii , h. takagi , s. uchida , k. fueki and s. tanaka , chem . lett .
* 2 * , 429(1987 ) .
r. j. cava , r. b. dover , b. batlogg and e. a. rietman , phys .
lett . , * 58 * , 408 ( 1987 ) .
r. m. hazen , l. w. finger , r. j. angel , c. t. prewitt , n. l. ross , h. k. mao and c. g. hadidiacos , phys .
b * 35 * , 7238 ( 1987 ) .
m. k. wu , j. r. ashburn , c. j. torng , p. h. hor , r. l. meng , l. gao , z. j. huang , y. q. wang and c. w. chu , phys .
lett . * 58 * , 908 ( 1987 ) .
h. maeda , y. tanaka , m. fukutomi and t. asano , jap .
* 27 * , l209 ( 1988 ) .
z. z. sheng and a. m. hermann , nature * 332 * , 55 ( 1988 ) .
a. schilling , m. cantoni , j. d. guo and h. r. ott , nature * 363 * , 56 ( 1993 ) .
y. kamihara , t. watanabe , m. hirano and h. hosnon , j. am .
soc . * 130 * , 3296 ( 2008 ) .
j. h. tapp , z. tang , b. lv , k. sasmal , b. lorenz , p. c. w. chu , and a. m. guloy , phys .
b * 78 * , 060505(r ) ( 2008 ) . m. rotter , m. tegel , d. johrendt , i. schellenberg , w. hermes and r. pottgen , phys . rev .
b * 78 * , 020503(r ) ( 2008 ) .
f. c. hsu , j. y. luo , k. w. yeh , t. k. chen , t. w. huang , p. m. wu , y. c. lee , y. l. huang , y. y. chu , d. c. yan and m. k. wu , proc .
u s a * 105 * , 14262(2008 ) .
j. g. guo , s. f. jin , g. wang , s. c. wang , k. x. zhu , t. t. zhou , m. he and x. l. chen , phys .
b * 82 * , 180520(r ) ( 2008 ) .
a. damascelli , z. hussain and z. x. shen , rev .
phys . * 75 * , 473 ( 2003 ) .
d. n. basov and a. v. chubukov , nature phys . * 7 * , 272 ( 2011 ) . j. p. hu and h. ding , sci
. rep . * 2 * , 381 ( 2012 ) .
n. barisic , m. k. chan , y. li , g. yu , x. zhao , m. dressel , a. smontara and m. greven , proc .
110 * , 12235 ( 2013 ) .
j. paglione and r. l. greene , nature phys .
* 6 * , 645 ( 2010 ) .
d. j. scalapino , rev . mod
. phys . * 84 * , 1383 ( 2012 ) a. ricci , b. joseph , n. poccia , w. xu , d. chen , w. s. chu , z. y. wu , a. marcelli , n. l. saini and a. bianconi , supercond .
* 23 * , 052003 ( 2010 ) .
t. t. zhou , y. m. wang , s. f. jin , d. d. li , s. f. lai , t. p. ying , h. zhang , s. j. shen , w. j. wang and x. l. chen , inorg . chem . * 53 * , 4154 ( 2014 ) .
g. kresse and j. hafner , phys .
b * 47 * , 558 ( 1993 ) .
g. kresse and j. furthmuller , comput .
* 6 * , 15 ( 1996 ) .
g. kresse and j. furthmuller , phys .
b * 54 * , 11169(1996 ) .
j. p. perdew , k. burke and m. ernzerhof , phys .
. lett . * 77 * , 3865 ( 1996 ) .
m. t. czyzyk and g. a. sawatzky , phys .
b * 49 * , 14211 ( 1994 ) h. j. monkhorst and j. pack , phys . rev .
b * 13 * , 5188 ( 1976 ) .
d. alfe , computer physics communications , * 180 * , 2622 ( 2009 ) .
a. togo , f. oba and i. tanaka , phys .
b , * 78 * , 134106 ( 2008 ) .
a. togo and i. tanaka , scr .
, * 108 * , 1 ( 2015 ) .
z. p. yin , s. lebegue , m. j. han , b. p. neal , s. y. savrasov and w. e. pickett , phys .
* 101 * , 047001 ( 2008 ) .
i. i. mazin , m. d. johannes , l. boeri , k. koepernik and d. j. singh , phys . rev .
b * 78 * , 085104 ( 2008 ) .
d. j. singh , phys .
b. * 78 * , 094511 ( 2008 ) .
d. j. singh and m. h. du , phys .
lett . * 100 * , 237003 ( 2008 ) . h. ding , p. richard , k. nakayama , t. sugawara , t. arakane , y. sekiba , a. takayama , s. souma , t. sato , t. takahashi , z. wang , x. dai , z. fang , g. f. chen , j. l. luo and n. l. wang , europhys . lett . * 83 * , 47001 ( 2008 ) .
l. zhao , h. y. liu , w. t. zhang , j. q. meng , x. w. jia , g. d. liu , x. l. dong , g. f. chen , j. l. luo , n. l. wang , g. l. wang , y. zhou , y. zhu , x. y. wang , z. x. zhao , z. y. xu , c. t. chen and x. j. zhou , chin .
* 25 * , 4402 ( 2008 ) .
k. nakayama , t. sato , p. richard , y. m. xu , y. sekiba , s. souma , g. f. chen , j. l. luo , n. l. wang , h. ding and t. takahashi , europhys .
* 85 * , 67002 ( 2009 ) . c. liu , g. d. samolyuk , y. lee , n. ni , t. kondo , a. f. santander - syro , s. l. budko , j. l. mcchesney , e. rotenberg , t. valla , a. v. fedorov , p. c. canfield , b. n. harmon and a. kaminski , phys .
lett . * 101 * , 177005 ( 2008 ) .
y. m. xu , p. richard , k. nakayama , t. kawahara , y. sekiba , t. qian , m. neupane , s. souma , t. sato , t. takahashi , h. q. luo , h. h. wen , g. f. chen , n. l. wang , z. wang , z. fang , x. dai and h. ding , nat
* 2 * , 392 ( 2011 ) .
e. akturk and s. ciraci , phys .
b. * 79 * , 184523 ( 2008 ) .
a. subedi , l. zhang , d. j. singh and m. h. du , phys .
b * 78 * , 134514 ( 2008 ) . m. c. lehman , a. llobet , k. horigane and d. louca , j. phys .
: conf . ser . *
251 * , 012009 ( 2010 ) .
y. zhang , l. x. yang , m. xu , z. r. ye , f. chen , c. he , h. c. xu , j. jiang , b. p. xie , j. j. ying , x. f. wang , x. h. chen , j. p. hu , m. matsunami , s. kimura and d. l. feng , nat .
* 10 * , 273 ( 2011 ) .
f. j. ma , w. ji , j. p. hu , z. y. lu and t. xiang , phys .
lett . * 102 * , 177003 ( 2009 ) . | we propose two possible new compounds , ba@xmath0cuo@xmath0fe@xmath0as@xmath0 and k@xmath0cuo@xmath0fe@xmath0se@xmath0 , which hybridize the building blocks of two high temperature superconductors , cuprates and iron - based superconductors .
these compounds consist of square cuo@xmath0 layers and antifluorite - type fe@xmath0x@xmath0 ( x = as , se ) layers separated by ba / k .
the calculations of binding energies and phonon spectrums indicate that they are dynamically stable , which ensures that they may be experimentally synthesized .
the fermi surfaces and electronic structures of the two compounds inherit the characteristics of both cuprates and iron - based superconductors .
these compounds can be superconductors with intriguing physical properties to help to determine pairing mechanisms of high @xmath1 superconductivity . |
understanding how stars interact with their environment is a general problem important for many astrophysical systems .
many types of stars are surrounded by shells , envelopes or disks of material .
such circumstellar material often participates in accretion , polar outflows , winds and disk - star interactions . in young stars ,
these processes directly effect star and planet formation and evolution . even for the closest young stars we study here ,
the stellar radius is an important spatial scale that is smaller than 0.1 milliarcseconds and it will not be imaged directly even by the next generation of telescopes .
other methods like interferometry can yield useful constraints , but the indirect techniques of spectroscopy and spectropolarimetry have much to offer .
in particular , spectropolarimetric measurements can put unique constraints on the geometry of the system as well as thermodynamic and radiative environments of circumstellar material .
high - resolution linear spectropolarimetry measures the change in linear polarization across a spectral line .
typical spectropolarimetric signals are very small , often a few tenths of a percent change in polarization across a spectral line .
measuring these signals requires very high signal to noise observations and careful control of systematics to measure signals at this level . in order to address these issues
, we built a dedicated spectropolarimeter for the high - resolution visible and infrared spectrograph ( hivis ) on the 3.7 m aeos telescope and performed a telescope polarization calibration and cross - instrument comparison ( harrington et al .
2006 , harrington & kuhn 2008 ) .
in addition , new and archival observations with the espadons spectropolarimeter will be presented which will complement our hivis observations .
many models show spectropolarimetry is a potentially useful probe of circumstellar environments at small spatial scales .
circumstellar disks , rotationally distorted winds , magnetic fields , asymmetric radiation fields ( optical pumping ) , and in general , any scattering asymmetry can produce a change in linear polarization across a spectral line ( cf .
mclean 1979 , wood et al .
1993 , harries 2000 , ignace et al .
2004 , vink et al .
2005a , kuhn et al .
these spectropolarimeric signatures can directly constrain the physical properties of the circumstellar environment once sufficiently detailed models are developed .
there have been several studies of young stars to date using medium resolution spectropolarimetry ( vink et al .
2002 , 2005b , mottram et al .
these studies focused on young stars and are based on only a few nights of observations . though spectropolarimetric signatures have been detected in many classes of star , a many - epoch , high - resolution comparison of different stellar classes has yet to be done .
we collected a large number of stellar observations on over 100 nights from 2004 to 2008 , monitoring 29 herbig ae / be stars and 30 be and emission - line stars .
these stellar types have very different circumstellar environments with various combinations of winds , disks , accretion , and jets ( cf . porter & rivinius 2003 and waters & waelkens 1998 ) . however , there is still much debate over the details of these processes .
thomson scattering theory is the current framework used to describe linear spectropolarimetric signatures .
a depolarization " scattering theory of spectropolarimetric line profiles was originally developed for be stars ( mclean 1979 ) .
another scattering model we will call the disk - scattering " model has been advanced for other star types .
however , neither model significantly matches most of our herbig ae / be observations , and many observations of other star types . the observing campaign and basic data processing will be presented in section 2 along with the spectroscopy of the herbig ae / be stars . a quick overview of current spectropolarimetric theories will be presented in section 3 .
the hae / be spectropolarimetry will be detailed in section 4 .
all the hivis and espadons data will be presented with individual examples for each star .
section 5 will present the spectroscopy and spectropolarimetry of the be and emission - line stars for comparison . a comparison with other stellar types
will also be made .
these observations will then be compared and contrasted in the discussion of section 6 .
spectropolarimetry is a photon - hungry technique and the aeos telescope suffers from significant pointing - dependent polarization properties .
the robust results presented here have been obtained only after many epochs of observations on most targets
( harrington et al .
2006 , harrington & kuhn 2008 ) .
we collected observations of 29 herbig ae / be stars from 2004 to 2008 .
the minimum acceptable signal - to - noise ratio to detect a 1% signature is roughly 300 .
our hivis observations have a raw signal - to - noise ratio of 100 - 1000 at a spectral resolution of 13000 with a 1.5 slit . in using the 1.5 slit , the detector significantly over - samples the spectral orders ( see harrington & kuhn 2008 ) .
thorium - argon lines have a 16-pixel full - width - half - max near h@xmath1 .
this data is typically binned to lower sampling while preserving full spectral resolution in order to achieve high precision measurements at full spectral resolution . in some targets there
is obvious need for higher resolution spectropolarimetry .
our hivis data has been confirmed and extended using our own and archival espadons observations .
these have even higher signal - to - noise ratios of 700 - 3000 and will be presented unbinned at a full spectral resolution of 68000 .
this large dataset with high resolution is reliable for the detection of a diverse set of stellar spectropolarimetric signatures and robust detection statistics . in this section , the observing campaign ,
spectroscopy and information about individual herbig ae / be stars is presented .
the hivis instrument properties , telescope polarization calibration and reduction methods are outlined completely in harrington & kuhn 2008 .
all the espadons data presented was reduced with the dedicated libre - esprit reduction software at cfht ( donati et al .
1999 ) .
the spectropolarimetric survey was conducted over three years while simultaneously doing instrument development and engineering . in early 2004
the spectropolarimeter was being assembled .
some test observations were done during the engineering of the hivis spectropolarimeter in late 2004 . during this period ,
the original science camera failed and was pulled off for repairs . in the summer of 2005
the reassembled camera was mounted .
there was a large polarization calibration effort in this season associated with the spectropolarimetry of comet 9p / tempel 1 for the deep impact event ( meech et al .
2005 , harrington et al .
2007 ) . in mid-2006
there were a number of instrument fixes to the spectrograph .
there were then two major herbig ae / be observing seasons once the instrument configuration stabilized :
september 2006 to february 2007 and july 2007 to january 2008 .
table [ obslog ] lists the observing and engineering dates .
each line gives a description of what was done on each night .
there were a number of nights spent on component testing or experimental programs such as the initial spectrograph , spectropolarimeter and ir spectrograph testing .
there were also tests of the ir dichroic , new spectrograph components , and sensitivity experiments .
there are also notes in the table showing when the various improvements , tests , and repairs took place . in the first part of the survey ,
few bright targets were chosen for close monitoring .
ab aurigae and mwc 480 were monitored almost every night over the 2006 - 2007 winter . during the same period , mwc 120 , mwc 158 and hd 58647 were less well covered , but still monitored heavily .
ab aurigae or mwc 480 were observed continuously for several hours on some nights , and all 5 targets intermittently on others .
there are a total of 148 polarization measurements for ab aurigae , 58 for mwc 480 , 24 for mwc120 , 39 for mwc 158 , 19 for hd 58647 , plus another 33 unpolarized standard star observations taken over the 40 nights in winter 2006 - 2007 .
the h@xmath1 line for these five stars showed significant variability in intensity , width , and profile shape that is entirely consistent with other spectroscopic variability studies ( cf .
beskrovnaya et al .
1995 , beskrovnaya & pogodin 2004 , catala et al .
1999 ) . on some nights ,
ab aurigae and mwc480 showed dramatic spectroscopic variation on a timescale of minutes to hours , mostly in the blueshifted absorption trough . some general line width and line strength variability
was also seen on short timescales .
all nights showed much smaller but significant variation .
the campaign was broadened in july 2007 to include many other herbig ae / be stars . during the summer , hd 163296 , hd 179218 , hd 150193 and v1295 aql were chosen for closer study .
another 111 unpolarized standard star observations were also taken since the instrument had changed significantly .
table [ tab - obs ] shows a list of the herbig ae / be stars and their properties taken from the literature . during the fall of 2007 and winter of 2008 , a much larger target list of herbig ae / be stars was used as well as bright be and emission - line stars for comparison . throughout the survey
mwc 361 was used as a detailed stable comparison between hivis and espadons .
all of the h@xmath1 line profiles for the herbig ae / be stars are shown in figures [ fig : haebe - lprof1 ] and [ fig : haebe - lprof2 ] [ cols="<,<,<,<",options="header " , ] our observing campaign first focused on @xmath2 cas in november of 2006 but was then widened to include 30 targets .
the h@xmath1 line profiles for these stars show a wide variety of profiles .
figures [ fig : be - lprof1 ] and [ fig : be - lprof2 ] show all the profiles compiled for each stars .
some show strong disky " signatures like @xmath3 per and mwc 143 .
others show a more simple symmetric emission - line such as 11 cam or 31 peg .
table [ tab - obsbe ] lists the basic stellar properties .
essentially all the stars are b - type and are 5th magnitude or brighter .
although there is only a month or two baseline between observations for variability studies , most systems showed only small profile changes , if any .
the most variable stars were @xmath4 tau , with its strong emission , @xmath5 cas and mwc 92 with their weaker emission lines .
many stars showed small variability of the emission , such as 25 ori , @xmath6 ori , @xmath5 cma , or 18 gem , but the overall structure of the emission line did not change .
this shows that during the observing window , the circumstellar region responsible for the emission did not change its structure significantly . for the purposes of this study , which is concerned with the circumstellar material
, the stars can be broken into subclasses based on their h@xmath1 line profiles .
there are strong , symmetric emission lines such as 11 cam in figure [ fig : be - lprof2 ] that have line : continuum ratios of 5 - 8 and do nt show evidence for intervening absorption .
there are the classical disky " systems such as mwc 143 that have strong emission with a strong , narrow central absorptive component .
another class will be distinguished in table [ tab - obsbe ] called disk * - these show less absorption than the classical disky " systems , but do show central , often variable absorption .
the last type does not show significant emission , such as @xmath7 cyg or qr vul . + + + + + + + + + + + + + + + + in contrast to the detected hae / be morphologies , many be and emission - line star signatures showed broad , smooth polarization changes spanning the entire width of the line with the polarization change centered on the line .
although none of these stars show p - cygni profiles and direct comparison with many hae / be detections is not possible , there are several disky " systems provide a robust comparison .
the magnitude of the polarization change in these stars was also smaller on average , with 0.5% or less being typical .
some stars showed signatures over 1% , but the polarization signature was much broader than the line and extended out into the line wings . even though many of the h@xmath1 lines in systems with broad spectropolarimetric detections had strong absorption features , there were fewer stars that showed strong deviations from the broad polarization morphology . in 10/30 stars ,
very clear detections of the broad polarization change can be seen . of these 10
, four had significant spectropolarimetric effects in and around absorptive features .
the compiled spectropolarimetry is shown in figures [ fig : be - specpol1 ] and [ fig : be - specpol2 ] .
the detections are outlined in table [ be - res ] .
there are 30 total targets . of these ,
ten stars that show the broad signature , five stars show narrow low - amplitude polarization effects , and the other 15 stars are non - detections at the 0.05% to 0.1% level .
this detection rate can be broken down into subclasses .
the broad signatures range in magnitude from a barely detectable 0.1% to 1.1% .
all of the stars with detected signatures show evidence of intervening absorption in their line profiles , either by the flattened or notched line centers , or by having a full - blown disky " line .
the stars with broad spectropolarimetric effects do show morphological deviations from the typical depolarization profile , but the polarization is quite different from the polarization - in - absorption of the herbig ae / be systems . in four of the 10 broad - signature stars ,
more complex effects in and around absorptive components of the line can be seen .
these effects can be large in amplitude but would not have been observed at low spectral resolution .
these are observed in addition to the broad `` depolarization '' effect , unlike the herbig ae / be stars .
the five narrow detections do not fit the depolarization description and will be discussed in more depth later .
@xmath2 cas is a very well - studied target that has an extended envelope and a strong h@xmath1 emission line .
quirrenbach et al .
1993 resolved the h@xmath1 emission region using interferrometry .
an elliptical gaussian model fits the observations with an axial ratio of 0.74 and an angular fwhm of 3.2mas .
this flattened envelope is exactly the kind imagined in the original studies on continuum polarization from circumstellar envelopes .
@xmath2 cas was monitored quite heavily over a year and we can use this target as a clear example for the entire class .
this star did not show very significant h@xmath1 line - profile or spectropolarimetric variations throughout the monitoring period .
all the spectropolarimetric observations were taken without binning and aligned in qu - space to maximize stokes q in the central 50 pixels .
these polarized spectra all had similar morphology but slightly varying magnitudes because of the varying telescope polarization properties at the many different telescope pointings used .
the resulting 22 spectra were then averaged to make a single polarized spectra with only 0.07% rms error at full spectral resolution ( no bin - by - flux was applied to the observations ) .
figure [ fig : bebroadex ] shows the spectropolarimetry .
the polarization signature is almost entirely in the + q spectrum .
though the spectra have been arbitrarily rotated to maximize + q , this figure shows that the signature is almost entirely linear in qu - space . in the qu - loop plot
, this polarization change becomes a simple horizontal excursion .
the polarization change is roughly 0.3% and is morphologically broader than the line - profile itself .
there is a very small width to the extension which comes from a very slight variation in the u spectrum on the red side of the emission line , as seen in the middle panel of the spectropolarimetry in figure [ fig : bebroadex ] .
a decrease in polarization of the same 0.2% magnitude had been reported in commissioning runs with the william - wehlau spectropolarimeter ( eversberg et al .
1998 ) , though at lower spectral resolution . since the signature is broad and is a single linear excursion in qu - space , this star fits well with the traditional depolarization " effect from scattering theory .
the change in polarization is smaller in magnitude than the known intrinsic continuum polarization .
the initial polarization studies all found a similar depolarization in continuum polarizations of : 1.0% mclean & brown 1978 , 0.8% mclean 1979 , 0.5% in poeckert & marlborough 1977 .
cas as a broad - effect example . *
a ) * the average @xmath2 cas spectropolarimetry at full resolution .
all observations have been rotated to maximize stokes q in line - center then averaged * b ) * the corresponding qu plot - a linear extension .
the + q orientation is an artifact of the arbitrary alignment.,title="fig : " ] + cas as a broad - effect example . *
a ) * the average @xmath2 cas spectropolarimetry at full resolution .
all observations have been rotated to maximize stokes q in line - center then averaged * b ) * the corresponding qu plot - a linear extension .
the + q orientation is an artifact of the arbitrary alignment.,title="fig : " ] + + + + the other nine targets with broad detections show very similar depolarization " morphology .
all of them show broad changes that extend to the wings of the line as seen in figure [ fig : bebroad ] .
however , a more detailed examination is necessary .
the example of @xmath2 cas illustrates the shape of these broad signatures , but the overall magnitude and morphology varies between the targets . for example , mwc 143 has a very large 1.1% deviation ( 0.9% q , 0.6% u ) but omi pup is barely detectable at 0.1% .
however , the main point that must be emphasized is that there is a very clear polarimetric morphological difference between these stars and herbig ae / be stars .
these stars tend to show a polarization variation over the broad line profile , if there is any polarization signature , while the often heavily obscured herbig ae / be systems show obvious
polarization - in - absorption .
" there is a very broad signature apparent in 10 of 30 stars with many showing some additional morphological features .
another 5 show more complex signatures worth describing in detail .
this is probably not surprising since these systems have presumably weaker nearby extinction as they show only subtle absorptive polarization variations .
there are some morphological differences that are worth mentioning specifically . in 25 ori ,
the change in stokes q is flat - topped " across the line center .
the star @xmath4 tau shows a noticeable asymmetry , being larger in the blue side of the line .
there is a small , narrow asymmetric spectropolarimetric signature in the absorptive part of the @xmath3 per disky " profile . on the blue side of the central absorption
there is a relative decrease with respect to the broad signature and a relative increase on the red side of the absorption .
these morphological effects show that there is more to be learned about circumstellar environment .
qualitatively we might expect that a combination of scattering and absorptive linear polarization might account for these results .
other spectropolarimetric studies of some of these stars have been done as well .
@xmath6 ori was a non - detection in oudmaijer & drew 1999 with a continuum polarization of 0.30% and a non - detection in vink et al .
2002 with a continuum of 0.27% .
it should be noted that these effects extend to the other hydrogen lines .
for instance , oudmaijer et al .
2005 report a broad 0.4% change across pa@xmath8 at 1.28@xmath9 m in @xmath4 tau .
this is only moderately less than the signature observed in h@xmath1 .
there are interesting interferrometric constraints on some of these stars as well .
@xmath4 tau is a very well - studied target .
quirrenbach et al .
1994 resolved the h@xmath1 emission region using interferrometry and created a maximum - entropy map of the circumstellar region .
an elliptical gaussian model fits the observations with an axial ratio of 0.30 and an angular fwhm of 3.55mas .
the map shows a highly elongated structure that was interpreted as a near edge - on disk . in a later interferrometric and spectropolarimetric survey ,
quirrenbach et al .
1997 report resolving h@xmath1 emission regions in @xmath2 cas , @xmath10 per , @xmath11 tau , @xmath4 tau , and @xmath12 cmi with angular sizes of 1.5 to 3.5mas .
these measurements show that the h@xmath1 emission is quite extended and is much larger than the stellar - radii scale .
all of the stars were point - sources in the nearby continuum interferrometric measurements . in light of these observations , the depolarization effect from extended and less - scattered h@xmath1 emission is extremely likely .
however , the simple models used to date are not enough to fit the morphologies observed here completely .
other stars show smaller , more complex signatures .
there were five stars with detected spectropolarimetric signatures of very small magnitude : 18 gem , @xmath13 col , @xmath12 cmi , r pup , and @xmath11 tau .
there were a number of 18 gem and @xmath11 tau observations that were non - detections , as seen in figure [ fig : otherbe ] , but the signal - to - noise was not as good .
the 18 gem signature is a small antisymmetric change in stokes u across the center of the line .
the signature for @xmath11 tau is a small drop in stokes u across line - center , as seen in figure [ fig : otherbe ] . in @xmath13
col and @xmath12 cmi , the detection was antisymmetric , spanned the line and was stable over two nights observations at the same telescope pointing .
both nights of data are shown in figure [ fig : otherbe ] .
r pup showed a small drop in q across the emission blue side of the simple emission line .
all of these detections do not fit in to the broad " morphology of the typical be depolarization signature .
the detections in @xmath13 col and @xmath12 cmi are extremely small , having an amplitude of @xmath140.1% with antisymmetric components .
however , they do not fit in to the scheme of the disk - scattering theory either .
the signatures are not double - peaked and symmetric and are not significantly wider than the h@xmath1 line .
all that can be conclude is that more detailed modeling and investigation is necessary .
the espadons archive has many stars with linear spectropolarimetric detections . in the context of this paper , comparing and contrasting spectropolarimetric signatures from different stellar systems
, it is very useful to show some of the morphologies for different types of stars .
the stars discussed here range from a to k spectral types .
most of the systems we present are consistent with polarization being affected by absorptive components of the line profiles .
3 pup ( mwc 570 ) is an emission line star of spectral type a3iab ( simbad ) with a disky " h@xmath1 line .
this star provides yet another comparison between the different star types as it has an a spectral type because it has a distinctly different spectropolarimetric profilie .
there is good quality archival espadons data from february 7th and 8th 2006 showing a very large complex h@xmath1 signature .
the polarization change , shown in figure [ fig : swp-3pup - esp ] is centered on the disky " emission line but spans the entire width of the line .
the polarization in the two emissive peaks is more than 0.5% increasing to over 2% in the central absorption .
the qu - plot shows a very complex structure .
there are multiple components of narrow - wavelength range in stokes q with a strongly antisymmetric stokes u. in qu - space this translates into multiple loops . in the corresponding hivis observations , shown in figure [ fig : swp-3pup - esp ] , there is a similar overall morphology and strength , but with a slightly different exact shape .
there has been no rotation applied to the hivis observations , but the alignment is close .
the qu - loops show the differences much more clearly , regardless any rotation that could be applied to the data .
the overall amplitude of the hivis observations is lower , and there is much more symmetry in the hivis loops .
there double - loop structure is preserved , with a pattern somewhat resembling a figure-8 shape . in the espadons archive observations ,
the small third loop near ( 1.5,0.5 ) is quite clear whereas a simple cluster of points is only seen in the hivis loops .
however , the loss of the loop could be entirely caused by the bin - by - flux routine that is automatically applied to the hivis observations , while it was not applied to the espadons data .
the third loop at a q of 1.5% is at the wavelengths of the central absorption , where hivis binning is roughly 5 pixels ( since the threshold was 5x continuum ) .
+ in the context of the scattering theories , this star is quite difficult to explain in any framework currently available .
the polarization is very strong compared to most other observations .
the form is also very complex , having strong deviations and asymmetries .
the polarization is also strong across the entire width of the line : absorption , emission and wings .
the overall intensity of the h@xmath1 line is quite low , being only 3 times continuum , and the central absorption is not extremely strong either , returning only to near continuum .
this example provides yet another good example for the need for new , more detailed theories .
the archival spectropolarimetry for several additional targets is worth investigating for evidence of polarization - in - absorption .
figure [ fig : swp - astar - esp ] shows six systems with clear spectropolarimetric detections .
epsilon aurigae ( hr1605 , hd31964 ) is an algol - type eclipsing binary of spectral type a8iab ( simbad ) . the h@xmath1 line for this star has a complex shape with evidence for emission and overlying absorption .
there is good quality archival espadons data from february 7th and 8th 2006 also showing a strong and complex h@xmath1 signature .
the polarization change is mostly symmetric about line - center and spans the entire width of the absorptive component of the line .
the polarization has an amplitude of roughly 1% almost exactly at line - center .
ac her ( hd 170756 ) is an rv - tau type variable star with a spectral type f4ibpv ( simbad ) .
there is archival data from february 7th 2006 shown in figure [ fig : swp - astar - esp ] .
the h@xmath1 line has a peak intensity of roughly twice continuum and is double peaked with absorptive effects clearly overlying emission .
the spectropolarimetric effect is roughly 0.8% in magnitude ( + 0.7% q , -0.3% u ) and is concentrated almost entirely in the central absorptive component .
this is very similar to some haebe stars such as hd 58647 that show a simple morphology in absorption .
v856 sco ( hd 144668 , hr 5999 ) is a delta - scut type variable star of spectral type a7ive ( simbad ) .
there is archival data from august 14th 2006 shown in figure [ fig : swp - astar - esp ] .
the h@xmath1 line is fairly wide with an amplitude of roughly two times the continuum .
there is a clear absorptive component that creates an asymmetric , double peaked profile with absorptive effects clearly overlying emission .
the spectropolarimetric effect is quite large in amplitude , being over 1% .
this star is reminiscent of the herbig star mwc 158 with the strong anti - symmetric signatures being almost entirely confined to the central absorptive component .
ss lep ( 17 lep , hd 41511 , mwc 519 , hr 2148 ) is listed as an emission - line star of spectral type apsh ( simbad ) . there is archival data from february 9th 2006 shown in figure [ fig : swp - astar - esp ] .
the h@xmath1 line has a p - cygni type profile with an amplitude of 2.5 times the continuum .
there is a strong , wide blue - shifted absorptive component .
the spectropolarimetric effect is large in amplitude , being 0.5% .
the effect is definitely strongest in absorption , but does span the entire width of the line and extends quite far to the red side of the line .
u mon ( hd 59693 ) is an rv - tau type variable star with a k0ibpv spectral type ( simbad ) .
there is archival data from february 7th and 8th 2006 shown in figure [ fig : swp - astar - esp ] .
the h@xmath1 line is fairly wide with an amplitude of roughly three times the continuum .
there is a clear , mildly red - shifted absorptive component that creates an asymmetric , double peaked profile .
the spectropolarimetric effect is clearly complex , being roughly 0.5% in stokes u at peak amplitude .
this star is also has a relatively narrow effect around the absorptive component .
89 her ( hd 163506 , hr 6685 ) is a post - agb star of spectral type f2ibe ( simbad ) .
there is archival data from february 7th 2006 shown in figure [ fig : swp - astar - esp ] .
the h@xmath1 line has a very wide , strong absorptive component and a small red - shifted emissive component .
the spectropolarimetric effect is complex , being antisymmetric in q and monotonic in u with a very large amplitude of over 3%(-3% q , -1.5% u ) .
this polarization effect is clearly strongest in the absorptive component , though the stokes u spectrum does show a gradual decrease toward zero extending to the red - shifted component of the line .
all of these detections in various stellar types show that there are a wealth of spectropolarimetric morphologies and that the polarization in the absorptive components must be understood in order to make meaningful constraints on the circumstellar environment .
+ + the be and emission line spectropolarimetry shows remarkably different spectropolarimetric profiles when compared to the herbig ae / be stars even with the same shape and magnitude h@xmath1 line profile .
the be signatures are smaller in magnitude and typically span the entire line ( 10/30 ) .
the morphology of the be spectropolarimetry does not seem to be only tied to the strong absorptive components of the line directly because the polarization effect is present across the entire line .
there is some morphological change tied to the presence or strength of the absorption in four of the 10 broad spectropolarimetric profiles , but this has a smaller amplitude than detected in the herbig ae / be systems .
these small amplitude effects would not be observed at lower spectral resolution and deserve more detailed analysis .
systems with both strong and weak absorption in both strong and weak h@xmath1 lines showed the broad spectropolarimetric effects .
however , be stars with either symmetric unobscured emission lines or absorption lines did not show any detectable polarization changes .
herbig ae / be stars in contrast show a polarization that is much stronger and is directly tied to the absorptive component of the line . in windy systems with
p - cygni type line profiles the polarization can be over 2% in the blue - shifted absorption trough . in many cases , especially ab aurigae and mwc 480 , the polarization at the emissive peak is identical to the continuum polarization within the 0.1% noise . in both be and hae / be stars , there are a wealth of spectropolarimetric morphologies even for the same line type .
in other stellar types , such as post - agb and rv - tau type stars , there is very significant absorptive polarization with a complex morphology .
a direct comparison between six different systems is shown in figure [ fig : swp - sidebyside ] . though all stars have h@xmath1 lines with absorption features , very different spectropolarimetric morphologies are seen .
mwc 158 and mwc 480 illustrate the typical herbig ae / be effects .
the polarization changes are very strong in the absorptive components of the line and the qu - loop morphologies are dominated by the absorptive regions .
mwc 143 is a be star that shows a broad clear spectropolarimetric signature that is wider than the emission line itself and shows no significant deviation from this broad spectropolarimetric trend in the central absorption .
this is a `` disky '' system that shows no effect in absorption , a clear change from the typical herbig ae / be systems .
the be star @xmath13 col illustrates the 5/30 stars that show complex , more narrow effects that have not been seen in be stars at lower spectral resolution or lower precision .
51 oph is classified as a herbig ae / be star but shows antisymmetric , lower - amplitude components similar to be stars .
there is not a pronounced effect in the absorptive component of the line .
this is unlike most of the herbig be stars such as hd 58647 which show a strong spectropolarimetric signature only in the very center of the central absorption .
the star 3 pup is an a - type star not in either be or haebe classes and shows a very strong , complex signature that is stable during the epochs of our observations .
the polarization signature is as broad as the emission line and has many complex components .
comparison of these three stars highlights the range of effects seen in this survey .
the detections show that , although the broad spectropolarimetric effect seems common ( 10/30 ) in be and other b - type emission - line stars , the effect is certainly not common in the herbig ae / be stars .
one common thread in those herbig ae / be stars was the consistent presence of spectropolarimetric signatures in and around absorptive components of the h@xmath1 line . in many of the stars , there were very large polarization changes in absorptive components while the wavelengths of strongest emission showed no polarization deviations from continuum .
the be and emission - line stars do have their own morphological complexities that are worth investigating , and 5 of 30 detections that do not follow the broad morphological description and 4 of 10 `` broad '' signatures show additional absorptive effects .
these more complex detections , in some cases , look very similar to some herbig ae / be detections . now that a very clear morphological difference has been established between herbig ae / be stars and other emission - line stars , a new theory will be explored that may better fit the polarization - in - absorption " morphology of the herbig ae / be stars .
this paper presents a large amount of hivis observations taken on over 100 nights on 29 herbig ae / be stars , 30 be and emission - line targets as well as several alternatively classified objects to compile a linear spectropolarimetric survey far larger in scope than any survey performed to date .
archival data as well as our own observations with the espadons spectropolarimeter were used to supplement the hivis survey and to prove the robustness of the combined dataset .
the espadons observations represent an order - of - magnitude gain in resolution from any survey done to date .
the observing campaign from 2004 to 2008 was outlined in section 2 .
the spectroscopic observations of the herbig ae / be stars showed , as expected , a wealth of line morphologies .
the h@xmath1 line of these stars is typically strong and quite variable .
the emission lines can be a few to thirty times continuum with many different absorptive components on top of the emission .
spectroscopic measurements for these stars over many time - scales were presented and discussed .
a few short - term variability studies were reported illustrating the dynamic circumstellar environment in these stars .
current theories of spectropolarimetric line profiles , based on disk or envelope scattering effects were discussed in section 3 .
the current disk - scattering theory predicts symmetric , broad , double - peaked spectropolarimetric signatures in the orbiting thin - disk case .
the corresponding qu - loops show position - angle changes that are equally broad in wavelength . in the case of outward radial motion ,
the disk - scattering theory predicts a shift of the spectropolarimetric profiles toward the red side of the spectral line .
the depolarization theory is essentially a dilution of continuum polarization by unpolarized emission .
this effect produces broad signatures that are roughly inversely proportional to the amount of emission . in section 4 ,
the spectropolarimetric survey of the herbig ae / be stars was presented .
the survey showed that roughly two thirds of the haebe stars showed polarization effects in the absorptive components of the h@xmath1 line .
both the windy - type and disky - type stars showed this polarization - in - absorption effect in the blue - shifted or line - center absorptive components respectively .
the magnitude of the effect was typically 0.1% to 2% with several stars showing complex morphologies and complex geometries in qu - space .
only one star , hd 179218 , showed a spectropolarimetric effect across the entire h@xmath1 line .
these observations are difficult to explain using current scattering theories .
the depolarization effect causes a broad , monotonic change in polarization that is roughly inversely proportional to emission and is a linear extension in qu - space .
the disk - scattering effect shows broad , symmetric morphology for orbiting material and a shift toward red wavelengths for out - flowing material .
both of these effects do not fit the observed morphology of polarization - in - absorption on the blue - shifted side of the windy - stars or the polarization - in - central - absorption - only for the disky - stars .
the amplitude of the observed effects is also a consideration .
scattering theory has no natural amplitude , but most of the detections fall in the 0.3% to 1.5% range .
a survey of 30 be and emission - line stars was performed and presented in section 5 .
this was also complemented by studies of 6 other - type stars such as post - agb and rv - tau types , as well as an in - depth look at 3-pup .
these be and emission - line stars had a very different type of spectropolarimetric morphology . in 10 of 30 stars there was a broad , monotonic spectropolarimetric signature present across the entire h@xmath1 line .
this type of signature is typical of the depolarization effect .
however , in these 10 broad detections , four showed more complex morphologies in addition to the simple depolarization tied to absorptive components of the emission line . in another 5 of 30 stars there were antisymmetric or other more complex spectropolarimetric signatures that are inconsistent with depolarization theory .
this smaller survey of be stars is in good agreement with the large collection of previous work done to date on these stars at lower resolution .
the broad spectropolarimetric morphology is reproduced , but the higher resolution , higher signal - to - noise and larger sample of this survey shows that there are still some additional uncertainties about the exact form of the spectropolarimetric effects in some of these stars .
one third of the detections ( 5/15 ) do not fit the simple depolarization framework .
four of 10 broad signatures show significant deviations of the line from the simple depolarization morphology in absorptive components that would only have been seen at high resolution with high signal - to - noise .
the inability of scattering theories to fit the polarization - in - absorption " seen in the herbig ae / be stars is good evidence for intrinsic absorptive polarization effects .
this survey has already inspired optical pumping calculations ( kuhn et al . 2007 ) that demonstrate such a mechanism .
detailed models based on any paradigm assuming any realistic circumstellar environments do not yet predict spectropolarimetric line profiles .
the best example of this is the polarization - in - absorption " seen primarily in the herbig ae / be stars or the more complicated profiles observed in be and other emission - line stars .
unusual linear spectropolarimetric profiles were seen in both haebe and be classes as well as in other a - type stars such as 3 pup and @xmath15 aur .
an intrinsic absorptive polarization mechanism seems essential for explaining many of these observations as the herbig ae / be detections are clearly inconsistent with disk - scattering theories .
however , as this large survey shows , there is a broad range of linear spectropolarimetric `` signatures '' in these stellar systems waiting to be deciphered .
this program was partially supported by the nsf ast-0123390 grant , the university of hawaii and the airforce research labs ( afrl ) .
some of this research used the facilities of the canadian astronomy data centre operated by the national research council of canada with the support of the canadian space agency .
this archive was very helpful by providing the released espadons data .
this program also made use of observations obtained at the canada - france - hawaii telescope ( cfht ) which is operated by the national research council of canada , the institut national des sciences de lunivers of the centre national de la recherche scientifique of france , and the university of hawaii .
these cfht observations were reduced with the dedicated software package libre - esprit made available by j. -f .
the simbad data base operated by cds , strasbourg , france was very useful for compiling stellar properties .
ashok n. m. , et .
al . , 1999 ,
7130 baines d. et al . , 2006 ,
mnras , 367 , 737 beskrovnaya n.g .
et al . , 1994 ,
a&a , 287 , 564 beskrovnaya n.g .
et al . , 1995 , a&a , 298 , 585 beskrovnaya n.g .
et al . , 1998 , a&as , 127 , 243 beskrovnaya n.g .
et al . , 1999 , a&a , 343 , 163 beskrovnaya n.g . & pogodin m.a . , 2004 ,
a&a , 414 , 955 bjorkman k.s .
et al . , 1998 , apj , 509 , 904 bohm t. & catala c. , 1993 , a&as , 101 , 629 bohm t. & catala c. , 1995 , a&a , 301 , 155 bohm t. et al . , 1996 ,
, 120 , 431 bouret j .- c . , catala c. , & simon t. , 1997 , a&a , 328 , 606 brown j.c . & mclean i.s . , 1977 ,
a&a , 57 , 141 brown j.c .
et al . , 1978 ,
a&a , 68 , 415 brown j.c .
et al . , 1989 , a&a , 68 , 415 carmona a. et al . , 2007 ,
apj , 344 , 341 cassinelli j.p .
, 1987 , apj , 317 , 290 catala c. & kunasz p.b . , 1987 , a&a , 174 , 158 catala c. et al . 1989 , a&a , 221 , 273 catala c. et al . , 1999 ,
a&a , 345 , 884 catala c. et al .
, 2007 , a&a , 462 , 293 chavero c. et al . , 2006 ,
a&a , 452 , 921 devine d. et al . , 2000 ,
apj , 542 , l115 domiciano de souza a. et al . , 2003 ,
a&a , 407 , l47 donati j.f .
et al . , 1999 ,
a&as , 134 , 149 eversberg t. et al . , 1998 ,
pasp , 110 , 1356 fernndez m. et al . , 1995 , a&as , 114 , 439 fu h .-
, 1997 , aj , 114 , 1623 fukagawa m. et al .
, 2004 , apj , 605 , l53 garcia lopez r. et al . , 2006 ,
a&a , 459 , 837 grady c.a .
et al . , 1993 ,
apj , 415 , l39 grady c.a .
2000 , apj , 544 , 895 grady c.a .
et al . , 2005 , apj , 630 , 958 happer , w. 1972 , rev .
44 , 169 harries t.j . ,
2000 , mnras , 315 , 722 harrington d.m .
et al . , 2006 , pasp , 118 , 845 harrington d.m .
, 2007 , icarus , 187 , 177 harrington d.m . &
2007 , apjl , 667 , l89 harrington d.m . &
kuhn j.r . , 2008 , pasp , 120 , 89 hillenbrand l.a .
et al . , 1992 , apj , 397 , 613 hillier d.j . , 1990 , a&a , 231 , 116 hillier d.j . , 1991 ,
a&a , 247 , 455 hillier d.j . , 1994 ,
a&a , 289 , 492 hillier d.j . , 1996 , a&a , 308 , 521 hubrig s. et al . , 2006 ,
a&a , 446 , 1089 hubrig s. et al . , 2006b , mnras , 371 , 1953 ignace r. et al . , 2004 ,
apj , 609 , 1018 ismailov n.z . &
aliyeva a.a . , 2005 ,
ibvs , 5634 jaschek c. & andrillat t. , 1998 , a&as , 128 , 475 kozlova o.v .
et al . , 2003 ,
astrofizika , 46 , 331 kozlova o.v . , 2004 ,
astrofizika , 47 , 287 kozlova o.v . , 2006 , astrofizika , 49 , 81 kuhn j.r .
et al . , 2007 , apjl , 668 , l63 liu w.m .
et al . , 2007 ,
apj , 658 , 1164 mannings v. et al . , 1997 , nature , 388 , 555 manoj p. et al . , 2006 ,
apj , 653 , 657 marconi m. et al . , 2000 a&a , 355 , l35 mclean i.s . &
brown j.c . , 1978 , a&a , 69 , 291 mclean i.s . &
clarke d. , 1979 , mnras , 186 , 245 mclean i.s .
, 1979 , mnras , 186 , 265 meech k.j .
et al . , 2005 , science , 310 , 265 miroshnichenko a.s .
et al . , 1998 ,
pasp , 110 , 883 miroshnichenko a.s .
et al . , 1999 ,
mnras , 302 , 612 miroshnichenko a.s .
et al . , 2004 ,
a&a , 427 , 937 mottram j.c .
et al . , 2007 , mnras , 377 , 1363 oudmaijer r.d . , & drew j.e .
, 1999 , mnras , 305 , 166 oudmaijer r.d . , et al . , 2001 , a&a , 379 , 564 oudmaijer r.d .
et al . , 2005 , mnras , 364 , 725 patel m. et al . , 2006 , mnras , 373 , 1641 pirzkal n. et al . , 1997 ,
apj , 481 , 392 poeckert r. & marlborough l.m .
, 1976 , apj , 206 , 182 poeckert r. & marlborough l.m .
, 1977 , apj , 218 , 220 poeckert r. & marlborough l.m . , 1978 ,
apj , 220 , 940 poeckert r. & marlborough l.m . , 1979 , apj , 233 , 259 pogodin m.a . , 1994 ,
a&a , 282 , 141 pogodin m.a . , 1997 , a&a , 317 , 185 pogodin m.a .
et al . , 2000 ,
a&a , 359 , 299 pogodin m.a .
et al . , 2004 , a&a , 417 , 715 pogodin m.a .
et al . , 2005 ,
a&a , 438 , 239 pontefract m. et al . , 2000 ,
mnras , 319 , l19 porter j.m . & rivinius r. , 2003 , pasp , 115 , 1153 quirrenbach a. et al . , 1993 , apj , 416 , l25 quirrenbach a. et al . , 1994 , a&a , 283 , l13 quirrenbach a. et al . , 1997 ,
apj , 479 , 477 rodgers b.m . , 2001 ,
washington phd thesis sitko m.l .
, 1981 , apj , 247 , 1024 skinner s.l .
et al . , 1993 ,
apjs 87 , 217 th p.s .
et al . , 1994 , a&as , 104 , 315 van den ancker m.e .
et al . , 1998 ,
a&a , 330 , 145 vink j.s .
et al . , 2002 ,
mnras , 337 , 356 vink , j.s . , 2005a , a&a , 430 , 213 vink j.s .
et al . , 2005b , mnras , 359 , 1049 wade g.a .
et al . , 2007 , mnras , 376 , 1145 waters l.b.f.m .
& waelkens c. , 1998 , ann . rev .
a&a , 36 , 233 whittet d.c.b .
et al . , 1992 , apj , 386 , 562 wood k. et al . , 1993 ,
a&a , 271 , 492 wood k. & brown j.c .
, 1994 , a&a , 291 , 202 wood k. et al . , 1997 ,
apj , 477 , 926 | the polarization of light across individual spectral lines contains information about the circumstellar environment on very small spatial scales .
we have obtained a large number of high precision , high resolution spectropolarimetric observations of herbig ae / be , classical be and other emission - line stars collected on 117 nights of observations with the hivis spectropolarimeter at a resolution of r=13000 on the 3.67 m aeos telescope .
we also have many observations from the espadons spectropolarimeter at a resolution of r=68000 on the 3.6 m cfh telescope .
in roughly @xmath02/3 of the so - called `` windy '' or `` disky '' herbig ae / be stars , the detected h@xmath1 linear polarization varies from our typical detection threshold near 0.1% to over 2% . in all but one hae / be star the detected polarization effect is not coincident with the h@xmath1 emission peak but is detected in and around the obvious absorptive part of the line profile .
the qu - loops are dominated by the polarization in this absorptive region . in several stars
the polarization varies in time mostly in the absorptive component and is not necessarily tied to corresponding variations in intensity .
this is a new result not seen at lower resolution . in the be and
emission - line stars , 10 out of a sample of 30 show a typical broad depolarization effect but 4 of these 10 show weaker effects only visible at high resolution .
another 5 of 30 show smaller amplitude , more complex signatures .
six stars of alternate classification showed large amplitude ( 1 - 3% ) absorptive polarization effects .
these detections are largely inconsistent with the traditional disk - scattering and depolarization models . |
quantum information theory explores the potential of quantum mechanics in order to process and transmit information . a two - level system , a qubit , constitutes the unit resource for storing information .
similarly , a unitary operation on one qubit can be regarded as a basic unit of information processing @xcite . in this paper
we explore the possibility of storing quantum dynamics , in particular unitary transformations , in the state of a quantum system , in a manner that the transformation can be performed at a later time and on another system almost perfectly
. the problem we address can be well - posed in the context of quantum circuitry .
we will say that the _ program _ state @xmath0 of some _ program register _ stores the one - qubit transformation @xmath1 , if some `` fixed '' protocol employing the state @xmath0 is able to perform @xmath1 on an arbitrary _ data _ state @xmath2 of a single qubit _ data register_. here , a `` fixed '' protocol means that the manipulation of the joint state _u [ estat ] does not require knowing the operation @xmath1 nor the state @xmath2 .
a device able to transform state ( [ estat ] ) into uu^_,u , [ universal ] where @xmath3 is just some residual state , is known as a _ programmable _ quantum gate @xcite . in a similar fashion as modern ( classical ) computers take both the program to be executed and the data to be processed as input bit strings , a programmable or universal quantum gate is a device whose action @xmath1 on an arbitrary data state @xmath2 is determined by the program state @xmath0 .
nielsen and chuang @xcite analyzed the possibility of constructing such a programmable quantum gate .
its total dynamics was described by means of a fixed unitary operator g according to g [ ] = ( u ) , [ perfect ] where only pure data states @xmath4 were considered because this already warranties the mixed state case .
notice that the program state @xmath5 and the residual state @xmath6 which was showed to be independent of @xmath4 can always be taken to be pure , by extending the program register with an ancillary system if needed .
nielsen and chuang proved that any two inequivalent operations @xmath1 and @xmath7 require orthogonal program states , that is @xmath8 .
thus , in order to perfectly store a given operation @xmath9 from some set @xmath10 , a vector state @xmath11 from an orthonormal basis @xmath12 has to be used . the operation @xmath13 can then be implemented by , say , measuring the program register to obtain the value @xmath14 , and gauging correspondingly some convenient experimental device . since the set of unitary operations is infinite ,
their result implied that no universal gate can be constructed using finite resources , that is , with a finite dimensional program register .
the aim of this work is to present programmable quantum gates with a finite program register , and thus physically feasible .
a finite register turns out to be sufficient if a degree of imperfection , no matter how small , is allowed in performing the unkwon operation @xmath1 .
we will construct a family of _ probabilistic _ programmable quantum gates , that is programmable quantum gates which work with a given prior probability @xmath15 of a successful implementation of @xmath1 .
such a one - qubit gate with @xmath16 was already described in @xcite . here
we will achieve any arbitrarily small @xmath17 .
we will also consider _ approximate _ programmable quantum gates , which perform an operation @xmath18 very similar to the desired @xmath1 , that is @xmath19 for some transformation fidelity @xmath20 .
the second main result is a lower bound on the dimension of the program register of the programmable gate in terms of its degree of imperfection @xmath21 .
it implies that the orthogonality result of @xcite is robust
. we will discuss its implications in the context of secure secret computation .
finally , operations stored in a quantum state can be teleported .
this leads to a new scheme for quantum remote control @xcite that only requires unidirectional communication .
we start by showing how to store and reimplement , in an imperfect but feasible fashion , an arbitrary one - qubit unitary operation of the form u _ ( i_z ) , [ unialpha ] where @xmath22 .
notice that a general one - qubit operation @xmath23 can be obtained by composing three operations of the form of eq .
( [ unialpha ] ) with some fixed unitary operations , for instance as @xmath24@xmath25@xmath26@xmath27 .
let us consider the state ( e^i+e^-i ) , which someone , say alice , can prepare by applying @xmath28 on a qubit in the standard state @xmath29 .
suppose she also prepares , along with @xmath30 , another qubit in some arbitrary state @xmath31 and provides bob , who does nt know @xmath32 nor the complex coefficients @xmath33 and @xmath34 , with the two qubits in state @xmath35 .
alice challenges now bob to obtain the state @xmath36 .
what bob can do in order to implement the unknown @xmath1 with some probability of success is to perform a c - not operation taking the data qubit in state @xmath4 as the control and the program qubit in state @xmath30 as the target .
this will constitute the basic part of our simplest programmable quantum gate .
recalling that the c - not gate , i + _ x , permutes the @xmath37 and @xmath38 states of the target ( second qubit ) only if the control ( first qubit ) is in state @xmath38 , it is easy to check that the two - qubit state is transformed according to ( u _ + u_^ ) .
therefore , a projective measurement in the @xmath39 basis of the program register will make the data qubit collapse either into the desired state @xmath36 or into the wrong state @xmath40 , with each outcome having prior probability @xmath41 .
that is , we have already constructed a probabilistic programmable quantum gate with error rate @xmath42 ( see figure ( [ fig1 ] ) ) .
notice that a single qubit has been sufficient for alice to store an arbitrary unitary @xmath28 , i.e. , one from an infinite set , although its recovery only succeeds with probability @xmath41 .
if bob obtains @xmath40 instead of @xmath36 , then not only he fails at performing the wished operation , but in addition he does no longer have the initial data state @xmath4 .
how can we construct a more efficient programmable gate ?
notice that in case of failure , a second go of the previous gate can correct @xmath40 into @xmath36 . indeed , bob needs only apply the gate of fig .
( [ fig1 ] ) to @xmath40 , inserting a new program state , namely @xmath43 , which alice can prepare by performing twice the operation @xmath28 on @xmath44 .
therefore , if alice supplies the state @xmath45 to bob , he can perform the operation @xmath28 with probability @xmath46 .
figure ( [ fig2 ] ) displays a more compact version of this second probabilistic programmable gate , which requires a two - qubit program register and has a probability of failure @xmath47 . in case of a new failure ,
the state of the system becomes @xmath48 .
bob can insert again this state , together with state @xmath49 , into the elementary gate .
if bob has no luck and keeps on obtaining failures , he can try to correct the state as many times as he wishes , provided that the state @xmath50 is available for the @xmath51th attempt . therefore , for any @xmath52 , the @xmath52-qubit state @xmath53 can be used to implement the transformation @xmath28 with probability @xmath54 .
resembles that used in @xcite to implement a non - local unitary operation . in the present context
all intermediate measurements and conditional actions can be substituted by a single unitary operation , as described in figures 2 and 3 . in this section
we have first presented the several - measurement version for pedagogical reasons . ] the corresponding probabilistic programmable gate ( see figure ( [ fig3 ] ) ) , consists of the unitary transformation of @xmath55 into ( u _
+ u_^(2^n-1 ) ) [ nqubits ] and of a posterior measurement of the program register ( either in state @xmath56 or @xmath57 , @xmath58 ) .
its failure probability , @xmath59 , decreases exponentially with the size @xmath52 of the program register .
it is interesting to look at how long the program needs to be , on average , until bob succeeds to perform @xmath28 with certainty . with probability
@xmath60 he succeeds after using a single - qubit program ; with probability @xmath61 a two - qubit program is sufficient ; etc .
the average length @xmath62 of the required program is thus |n= _
[ average ] that is , a two qubit register is sufficient , on average , to store an arbitrary @xmath28 so that it can be performed with certainty .
a probabilistic programmable gate may either succeed or fail , depending on the result of the final measurement on the program register .
an approximate gate , instead , performs a transformation only similar to the desired one , but it is always successful .
suppose we want to apply the unitary transformation @xmath1 on @xmath63 but instead another ( general ) transformation @xmath64 is actually performed .
a possible way of quantifying how similar these two operations are is by applying both operations to the same state @xmath63 , and then computing the fidelity between the two transformed states , @xmath65 and @xmath66 . when averaged over all possible @xmath63 this reads f(,u ) du^()u .
[ fid ] suppose now that after the transformation ( [ nqubits ] ) of the previous probabilistic gate we decide to ignore the state of the @xmath52-qubit program register .
then the programmable gate works approximately , implementing an operation @xmath67 , where @xmath68 .
the average fidelity of performance ( [ fid ] ) satisfies @xmath69 .
so far we have explicitly constructed programmable quantum gates that perform , either probabilistically or approximately , some class of one - qubit unitary operations @xmath28 .
but the previous protocols also allow alice to codify with finite resources any unitary operation @xmath70 acting on an arbitrary number @xmath51 of qubits .
indeed , as already mentioned , alice can codify an arbitrary one - qubit unitary operation using only @xmath71 @xmath28 s , and then also combine several of those with c - not gates to obtain @xmath70 . in view of these results
, one may wonder whether quasi - perfect programmable gates can be applied , in the context of quantum cryptography and computation , to secretly compute some unitary operation @xmath70 , for instance a precious algorithm , on some initial @xmath51-qubit state @xmath72 .
the idea is that alice gives a program state @xmath73 and the data state @xmath72 to bob , who operates a programmable quantum gate array but ignores @xmath70 .
bob is required to compute @xmath74 , but alice does not want bob to know what program he is running on his quantum computer . if the gate is perfect as in ( [ perfect ] ) , then bob can in principle distinguish @xmath73 from any other program state , since they are orthogonal .
therefore he can , imperceptibly to alice , make an illegal copy of the program , perform the required transformation using the original program state , and give the computed state to alice . however , when the gate is slightly imperfect , different programs need no longer be orthogonal .
now bob can not determine perfectly well which program he is to run in his computer .
if he tries to estimate @xmath73 , then in addition he will necessarily modify the program state , which will result in an improper performance of the gate and then alice who may have simply been testing bob s integrity can , in principle , detect it .
that is , it is not possible for bob to copy , even in an approximate form , the program state and at the same time perform the operation alice has commended him with , without this being detectable .
we next derive lower bounds on the size @xmath52 of the program register of any quasi - perfect ( i.e. with @xmath75 ) programmable gate , and on the degree of orthogonality between its program states @xmath5 corresponding to similar operations , in terms of its failure parameter @xmath21 .
these bounds represent a severe limitation on the degree of reliability that a security scheme based on the above ideas can offer .
they indicate that the program vectors @xmath5 and @xmath76 are significantly non - orthogonal ( that is , non - distinguishable ) only when the imperfection parameter @xmath21 makes them effectively equivalent .
let us consider a generic imperfect programmable gate acting on a @xmath77 system , so that it can be programmed to perform some or all @xmath78 .
it can be described by a unitary operator @xmath79 according to g _ [ ] = ( u ) + , [ rotllo ] where the wrong state @xmath80 is not required to fulfill any requirement for an _ approximate _ gate , whereas it must satisfy @xmath81 ( here @xmath82 is the null vector of the data register ) for any two inequivalent operations @xmath7 and @xmath1 and any two data states @xmath83 and @xmath84 for a _
probabilistic _ gate .
this last condition is necessary for bob , who ignores both @xmath1 and the data , to be able to know whether the transformation @xmath1 has been successfully performed by measuring the program register .
we first notice that the state @xmath85 only depends on @xmath86 through a contribution of order @xmath87 , where from now on @xmath88 for approximate gates and @xmath89 for probabilistic ones . indeed , for any program state @xmath5 , the scalar product of ( [ rotllo ] ) corresponding to any two data states @xmath83 and @xmath84 reads = + o(^p ) , from which , by fixing @xmath83 and considering any @xmath84 , we find that @xmath90 .
that is = + o(^ ) . keeping this in mind , we now consider , for any given @xmath4 , the scalar product of ( [ rotllo ] ) corresponding to two unitary operations @xmath1 and @xmath7 , which turns out to read = u^v + o(^ ) .
[ productuv ] the scalar product @xmath91 does not depend on @xmath4
. therefore the dependence of @xmath92 on @xmath4 has to be of order @xmath87 , at most .
suppose @xmath1 and @xmath7 are very close . that is , u^ v = e^il = i + il - l^2 + o(l^3 ) , where all the eigenvalues @xmath93 of the traceless ( @xmath94 ) hermitian operator @xmath95 @xmath96 , are very small .
the largest variation of @xmath97 in ( [ productuv ] ) for two different vectors @xmath4 is @xmath98 .
we introduce a distance on the set of operators on @xmath99 , d_u , v ( [ ( u - v)^(u - v)])^. then @xmath100 .
subtracting ( [ productuv ] ) for @xmath101 from itself for @xmath102 we conclude that @xmath103 , which finally implies || .
[ bound1 ] this bound says that in a programmable quantum gate with a small error rate @xmath104 , two transformations @xmath1 and @xmath105 will have program states with significant overlap @xmath91 ( states @xmath5 and @xmath76 are indistinguishable ) only if @xmath1 and @xmath7 are also very close to each other , @xmath106 .
that is , only if @xmath1 and @xmath7 process the data very similarly , then a dishonest bob is unable to distinguish between the corresponding programs . the previous result can also be used to derive a lower bound on the dimension of the program register of an imperfect programmable gate with error @xmath21 .
for simplicity we will assume that the gate can only be programmed to perform the one - qubit transformations @xmath28 from eq .
( [ unialpha ] ) . consider a discrete subset of such transformations , namely those with @xmath107 , @xmath108 , and apply the previous bound to @xmath109 and @xmath110 .
we obtain || k^pm , t0 , [ bo ] where @xmath111 is some unimportant constant .
we need the following lemma . * lemma : * let @xmath112 be a set of @xmath113 ( normalized ) vectors such that their scalar products @xmath114 satisfy @xmath115 for @xmath116 . then the @xmath113 vectors are linearly independent . _
proof : _ the rank of the set @xmath112 is equal to the rank of the matrix @xmath52 , @xmath117 , which has ones in all diagonal entries .
the modulus of any entry of the matrix @xmath118 is smaller than @xmath119 .
let @xmath120 be a normalized eigenvector of @xmath118 , with eigenvalue @xmath121 .
then @xmath122_i=[(n\!-\!i){\mbox{$| \varphi \rangle$}}]_i = \sum_{j=1}^q \nu_{ij } [ { \mbox{$| \varphi \rangle$}}]_j$ ] , where @xmath123_j$ ] denotes the @xmath124th vector component .
let @xmath14 be such that @xmath125_i| \geq |[{\mbox{$| \varphi \rangle$}}]_j|$ ] @xmath126 .
then @xmath127_j| \leq \sum_{j=1}^q |\nu_{ij}||[{\mbox{$| \varphi \rangle$}}]_i|$ ] @xmath128 @xmath125_i|$ ] , that is , @xmath129 , and since this holds for all the eigenvalues of @xmath130 , @xmath52 has @xmath113 positive eigenvalues or , equivalently , rank @xmath113 .
* remark : * for @xmath113 sufficiently large , if all @xmath131 are of order @xmath132 and the components of the eigenvector @xmath120 are relatively equally weighted , then it is plausible that the eigenvalue @xmath133_j/[{\mbox{$|
\varphi \rangle$}}]_i$ ] is of the order @xmath134 ( random walk ) .
this suggests that in order for the set @xmath112 to have rank close to @xmath113 , it is sufficient that @xmath135 , instead of @xmath115 as required in the lemma
. let us set @xmath136 .
then ( [ bo ] ) becomes @xmath137 , and this means , because of the lemma , that at least an @xmath138-dimensional hilbert is required to contain @xmath139 .
that is , the program register must consists of at least @xmath140 qubits .
notice that the previous remark suggests that this bound may be reduced to @xmath141 qubits , in which case the probabilistic programmable gate of figure ( [ fig3 ] ) would require , asymptotically , the smallest possible program register .
for a general programmable gate implementing some or all transformations @xmath78 it is straightforward to obtain a similar lower bound on the dimensions of the program register , which also says that its number @xmath52 of qubits grows proportionally to the logarithm of the inverse of the rate error , @xmath142 , for some positive constant @xmath143 .
we have shown how to store an arbitrary unitary transformation in the pure state of a finite quantum register , in such a way that it can be performed quasi - perfectly at a later time . once the unknown operation has been encoded in a quantum state ,
it can of course be processed using _ any _ known state manipulation technique .
an interesting application of our results is in the context of quantum remote control . as introduced by huelga _
_ in @xcite , let us suppose bob wishes to manipulate some data state according to an unknown operation alice , a distant party , can implement by using some device .
if the state of alice s device can not be teleported , then the optimal protocol @xcite is to use standard teleportation @xcite to send the data from bob to alice , who will use the device to process it and will teleport it back to bob .
but we now know how to efficiently store operations in quantum states , which can then be teleported .
this leads to a new scheme for quantum remote control : alice stores the operation in a quantum state and applies standard teleportation to send it to bob . remarkably enough , in this protocol only one - way communication is required in addition to entanglement in order for alice to remotely manipulate bob s data , as opposed to the two - way classical communication of the scheme presented in @xcite .
this implies that the operation can be teleported independently of whether bob s data state is already available .
more specifically , we find that a @xmath52-qubit program @xmath144 can be teleported from alice to bob by using up @xmath52 ebits of entanglement and by sending @xmath52 classical bits from alice to bob ( recall that the classical communication cost of quantum teleportation of equatorial states , as @xmath30 , require only one bit per state @xcite ) . eq .
( [ average ] ) implies that , on average , @xmath145 ebits of entanglement between alice and bob , and @xmath145 classical bits from alice to bob are sufficient for alice to teleport an arbitrary @xmath28 to bob , so that he , ignoring @xmath28 , can perform it with certainty .
the storage of quantum transformations turns out to be useful in several other contexts .
if state estimation techniques are applied to the system that stores an unknown operation , then we obtain the scheme for estimation of quantum dynamics recently exploited by acn _
et al _ @xcite .
cirac _ et al _ @xcite have recently explored the possibilities of encoding operations in quantum states in the context of non - local transformations of a composite system .
in particular , they have shown how to implement non - local unitary transformations using less than one ebit of entanglement . in an extension of their work ,
dr _ et al _ @xcite have considered alternative schemes for storing and manipulating quantum transformations .
we have presented a scheme for storing unitary operations in the quantum state of a finite dimensional program register .
the operations can be implemented at a later time with some associated error @xmath21 , which decreases exponentially with the number of qubits of the program register .
we have presented both probabilistic and approximate programmable quantum gates , and have discussed the possibility of using them to make a secrete computation on a public quantum computer .
finally , a unidirectional scheme for remote manipulation of quantum states has also been put forward .
we thank w. dr for useful comments .
g.v . acknowledges a marie curie fellowship ( hpmf - ct-1999 - 00200 , european community ) .
this work was also supported by the sfb project 11 on `` control and measurement of coherent quantum systems '' ( austrian science foundation ) , the institute for quantum information gmbh and the project equip ( contract ist-1999 - 11053 , european community ) .
indeed , an arbitrary unitary operation @xmath70 on @xmath51 qubits can decomposed as a fixed that is , @xmath70-independent sequence of ( @xmath70-dependent ) one - qubit unitary operations and a specific kind of two - qubit unitary operations , for instance c - not gates since the sequence is fixed , @xmath70 is specified by the one - qubit transformations only ; an arbitrary von neumann measurement , given by orthogonal projectors @xmath146 , can be decomposed into a general unitary operation @xmath7 , with @xmath147 , a von neumann measurement on a fixed set of projectors @xmath148 and a second unitary operation @xmath149 .
finally , a general quantum operation on an arbitrary system can be achieved by appending an ancilla in some blank state to the system , performing a unitary transformation @xmath1 followed by a von neumann measurement , @xmath146 , and disposing of part of the system and/or ancilla .
therefore any possible transformation of a quantum system can be essentially specified by means of one - qubit unitary operations . | we show how quantum dynamics ( a unitary transformation ) can be captured in the state of a quantum system , in such a way that the system can be used to perform , at a later time , the stored transformation almost perfectly on some other quantum system .
thus programmable quantum gates for quantum information processing are feasible if some small degree of imperfection is allowed .
we discuss the possibility of using this fact for securely computing a secret function on a public quantum computer .
finally , our scheme for storage of operations also allows for a new form of quantum remote control . |
single crystals of yb(rh@xmath11co@xmath12)@xmath13si@xmath13 ( @xmath50 , 0.07 , 0.12 , 0.38 , 0.68 , 1.00 ) , ybir@xmath13si@xmath13 , and ybni@xmath14p@xmath13 have been grown and characterized as described previously @xcite .
the sharpness of the laue diffraction pattern taken from the samples suggests high crystalline quality which for ybrh@xmath13si@xmath13 and ybco@xmath13si@xmath13 is also confirmed by excellent arpes and stm data @xcite .
the physical properties of the samples have been extensively studied with e.g. specific heat , resistivity and magnetic susceptibility measurements . from these data
a consistent picture of the kondo scale in the yb(rh@xmath11co@xmath12)@xmath13si@xmath13 has emerged showing a gradual decrease of the single ion kondo temperature @xmath0 from about 25k for @xmath51 to well below 1k for @xmath52 @xcite . the onset of coherence effects in the resistivity which is typical for kondo lattices scales similarly from about 120k for @xmath51 to @xmath531.5k for @xmath52 .
the rxes data have been taken with the x - ray emission spectrometer of the former esrf beamline id16 , now upgraded and reopened as id20 , looking at the yb @xmath16-@xmath15 emission line .
measuring with energy resolution in the scattered beam significantly reduces the life - time broadening of the spectra and furthermore can be used to enhance the weak yb@xmath10 contributions in the kondo lattices @xcite . for each sample
we have taken the yb @xmath16 xas spectrum , then fixed the incident photon energy to the maximum of the yb@xmath10 resonance and measured the yb @xmath15 rxes spectrum . the experimental procedure and the data analysis
are described in details in @xcite .
the data has been taken twice for each sample during two different experiments and using two different sample growth batches .
the results of the first and the second run of experiments are exactly the same showing perfect reproducibility of our experimental findings . as a hard x - ray photon - in photon - out technique yb
@xmath16 edge rxes is a bulk sensitive technique free of surface effects .
the available temperature range was 3k to 300k and the data was taken starting from low temperature .
properties of ce systems at higher temperatures including the effect of cef have been theoretically studied using a scaling approach @xcite .
the result is that at higher temperatures , when the first excited and the second excited cef doublets get thermally populated , the low temperature kondo scale @xmath54 , where only the ground state doublet is taken into account , is replaced by an effective high temperature kondo scale @xmath55 and @xmath56{t_{k,0}\,\delta_1\,\delta_2}$ ] , respectively , where @xmath57 and @xmath58 are the energies of the first and the second excited cef doublets , respectively .
thus , as long as @xmath59 , the effective kondo scale at high @xmath2 is mostly determined by the cef splitting , and of the size of the cef splitting . further on , the lower @xmath54 , the larger will be the difference between @xmath54 and @xmath60 .
for trivalent yb the effect is even stronger , because the ground state @xmath40 multiplet is @xmath34 instead of @xmath61 for ce .
thus the cef leads to four doublets instead of three . accordingly
the high temperature kondo scale will be @xmath62{t_{k,0}\,\delta_1\,\delta_2\,\delta_3}$ ] .
usually the kondo effect is considered to become less and less effective upon increasing @xmath2 , but there are some properties for which a strengthening of the effect of the kondo interaction upon increasing @xmath2 due to the involvement of excited cef levels is well known and theoretically and experimentally well documented , e.g. the resistivity @xcite . for the simple , but theoretically more accessible single kondo ion model , the magnetic contribution to the resistivity @xmath63 first decreases logarithmically upon increasing @xmath2 , with parameters defined by @xmath54 .
but when @xmath2 approaches @xmath57 and/or @xmath58 , @xmath64 increases markedly , passes through a maximum , and then decreases again logarithmically , but with a larger slope than at lower @xmath2 . the larger slope at high @xmath2 is directly connected to the higher effective degeneracy of the local moment at high @xmath2 @xcite .
a similar behavior is also expected for the kondo lattice @xcite .
furthermore , the high temperature kondo scale defined above , @xmath65 and @xmath60 , have been shown to be relevant for the description of the magnetic susceptibility at high @xmath2 @xcite .
50ifxundefined [ 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 http://stacks.iop.org/0034-4885/44/i=1/a=001 [ * * , ( ) ] link:\doibase 10.1080/00018738400101681 [ * * , ( ) ] `` , '' ( , , ) p. link:\doibase
10.1103/physrevlett.35.1779 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.256403 [ * * , ( ) ] link:\doibase 10.1103/physrevx.5.011028 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.43.1892 [ * * , ( ) ] link:\doibase 10.1007/bf00710351 [ * * , ( ) ] http://stacks.iop.org/0953-8984/13/i=17/a=103 [ * * , ( ) ] link:\doibase 10.1126/science.aaa9733 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.88.196403 [ * * , ( ) ] http://dx.doi.org/10.1038/srep17937 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.71.1419 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.035120 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.035111 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.115114 [ * * , ( ) ] link:\doibase 10.7566/jpscp.1.012117 [ * * , ( ) ] link:\doibase 10.3938/jkps.62.1778 [ * * , ( ) ] link:\doibase 10.1103/physrevb.75.035113 [ * * , ( ) ] http://dx.doi.org/10.1038/nphys1299 [ * * , ( ) ] link:\doibase 10.1103/physrevb.84.245114 [ * * , ( ) ] link:\doibase 10.1103/physrevb.83.144405 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.035119 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.256402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.85.626 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.physb.2006.01.030 [ * * , ( ) ] link:\doibase 10.7566/jpscp.3.011001 [ * * , ( ) ] http://stacks.iop.org/1367-2630/13/i=10/a=103014 [ * * , ( ) ] link:\doibase 10.1126/science.1230583 [ * * , ( ) ] link:\doibase 10.1103/physrevb.36.2036 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.81.5225 [ * * , ( ) ] link:\doibase 10.1143/jpsj.81.034705 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0304-8853(90)90718-6 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.60.2331 [ * * , ( ) ] link:\doibase 10.1103/physrevb.39.3065 [ * * , ( ) ]
link:\doibase 10.1103/physrevlett.105.237601 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.165122 [ * * , ( ) ] link:\doibase 10.1143/jpsj.79.114717 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.106407 [ * * , ( ) ] http://dx.doi.org/10.1038/nature10148 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.176401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.085113 [ * * , ( ) ] @noop link:\doibase 10.1080/14786435.2012.669066 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.jcrysgro.2016.05.042 [ * * , ( ) ] link:\doibase 10.1143/ptp.71.450 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0304-8853(85)90438-x [ * * , ( ) ] link:\doibase 10.1103/physrevb.64.184438 [ * * , ( ) ] link:\doibase 10.1103/physrevb.5.4541 [ * * , ( ) ] link:\doibase 10.1103/physrevb.31.7424 [ * * , ( ) ] | * the kondo model predicts both the valence at low temperatures and its temperature dependence to scale with the characteristic energy @xmath0 of the kondo interaction . we have studied the evolution of the @xmath1 occupancy in a series of yb kondo lattices as a function of temperature using resonant x - ray emission spectroscopy .
here we show that while the low temperature value of the valence scales indeed very nicely with the @xmath0 determined from thermodynamic measurements , the @xmath2 induced increase of the valence towards the 3 + state occurs on a temperature scale @xmath3 nearly independent of @xmath0 despite the fact that the kondo temperatures of the materials included in our study span almost four orders of magnitude .
this observation is in remarkable contradiction to both the naive expectation and to precise theoretical predictions of the kondo model .
we suggest excited crystal electric field levels and lattice vibrations to be possible origins of this unexpected result . *
while most of the rare earth ( re ) elements show a very stable trivalent state in solids , some of them , especially ce , yb and eu , may display different valence states depending on the chemical composition , the temperature , and the applied pressure or magnetic field @xcite .
this valence instability , which is caused by the @xmath4 shell being almost empty ( ce ) , almost full ( yb ) or almost half - filled ( eu ) , is connected with a hybridization between localized @xmath1 and delocalized valence states @xcite .
weak hybridization results in a stable trivalent re state with large local magnetic moments which order at low @xmath2 due to rkky exchange . with increasing hybridization the local moment gets screened by the conduction electrons via kondo interaction @xcite .
as a result magnetic order disappears leaving a paramagnetic ground state with strongly renormalized conduction electrons , referred to as kondo lattices and heavy fermion systems . in these systems
the occupation of the @xmath4 shell , i.e. the number of @xmath1 electrons @xmath5 for ce or @xmath1 holes @xmath6 for yb , shows a weak decrease with increasing hybridization and the valence starts to deviate from the trivalent state @xmath7 .
models based on the kondo interaction have proven to be very appropriate to describe this regime @xcite . in the limit of strong hybridization ,
one enters the intermediate valence regime where a strong quantum mechanical mixing of different valence state occurs @xcite . in this case ,
@xmath5 and the valence can considerably deviate from 1 and 3 + , respectively . since the competition between the rkky interaction on one hand and the kondo interaction and valence fluctuations on the other hand leads to very unusual states and properties , such system
have attracted a lot of attention @xcite . particularly fascinating are phase transitions at absolute zero temperature , called quantum phase transitions , which are often connected to unconventional superconductivity .
knowledge of the re valence and of the occupation of the @xmath4 level is crucial for an understanding of the exotic properties of such systems . the simplicity of the x - ray absorption / emission measurements at the @xmath8 edges of rare earth elements and the large x - ray penetration depth providing bulk specific results made this technique a standard for the study of the valence in kondo lattice and intermediate valent systems as a function of composition and temperature .
hence a large number of studies have been performed to determine @xmath5 as well as the kondo or valence fluctuating energy scale using the @xmath2 dependence of @xmath5 @xcite .
however , the accuracy to which @xmath5 could be determined with standard xas or photoemission measurements was limited to maybe a few percent @xcite . since in kondo
lattices the deviation from the trivalent state is small and typical changes in @xmath5 between 3k and 300k are only of the order of a few percent , these studies of the @xmath2 dependent changes concentrated on systems with a relatively high @xmath9k . however , the most interesting cases are those where the deviation from trivalent is small , since this is the regime where the transition from a magnetic ordered to a paramagnetic ground state occurs @xcite .
the strong development of resonant x - ray emission spectroscopy ( rxes ) has opened new possibilities for studying even minute valence changes in kondo lattices and to determine @xmath5 with very high accuracy @xcite .
the yb case is especially favorable for rxes because the weak signal of the `` minority '' yb@xmath10 state is on the low energy side of the absorption edge , i.e. in a region with very low background by other emission lines .
rxes allows to selectively enhance this yb@xmath10 signal by at least one order of magnitude @xcite .
we have used this high accuracy for a high precision study of the composition and the @xmath2 dependence of @xmath5 and its relation to bulk properties in yb - based kondo lattices . the extended and detailed knowledge accumulated on the kondo alloy systems
yb(rh@xmath11co@xmath12)@xmath13si@xmath13 and yb(rh@xmath11ir@xmath12)@xmath13si@xmath13 makes them perfect candidates for a systematic investigation of the relation between @xmath5 and the kondo scale @xcite . with increasing co doping the kondo scale is gradually reduced by almost two orders of magnitude from about 25k for ybrh@xmath13si@xmath13 to less than 1k for ybco@xmath13si@xmath13 @xcite . in contrast substituting ir for rh increases @xmath0 by about a factor of two up to 45k @xcite .
ybni@xmath14p@xmath13 is a kondo lattice system just at the verge of the transition from a ferromagnetic to a paramagnetic ground state , with a kondo temperature of 8 k which increases upon substituting as for p @xcite .
-@xmath15 rxes spectra as a function of temperature for four different yb kondo lattice systems .
the incident photon energy was set to the maximum of the 2 + resonance of the yb @xmath16 edge .
all spectra are normalized to the intensity of the yb@xmath17 emission line . ] in figure [ fig : fig1 ] the obtained rxes spectra as a function of temperature are shown for four selected compounds .
the data has been normalized to the peak intensity of the yb@xmath17 emission line . for ybir@xmath13si@xmath13 and ybrh@xmath13si@xmath13 with a relatively high kondo temperature
the temperature effects are sizeable and easily seen by eye . in ybco@xmath13si@xmath13 where the kondo scale is small the deviation from integer valent yb@xmath17 is very small and consequently the intensity of the yb@xmath10 line is very weak .
however , even in this case the variation of the yb@xmath10 intensity with temperature can be measured and quantified with extremely high accuracy thanks to the excellent performance of the spectrometer yielding very high quality data .
si@xmath13 and ybco@xmath13si@xmath13 rxes spectra at low temperature.,width=283 ] for the quantitative analysis of the spectra we fitted the data with three components for the yb@xmath17 ( blue filling ) and yb@xmath10 emission line ( green ) and the weak quadrupole transitions e2 ( white ) in the pre - edge region , respectively @xcite .
this is shown in figure [ fig : fig2 ] for the low temperature spectra of ybir@xmath13si@xmath13 and ybco@xmath13si@xmath13 .
the line shape of the yb@xmath17 and yb@xmath10 lines is not symmetric and has been determined from the experimental data by looking at the difference spectrum between high and low @xmath2 data .
the yb valence can then be determined from the ratio of the yb@xmath10 and yb@xmath17 integrated intensities knowing that the yb@xmath17 emission line is reduced to almost a tenth of its maximum intensity when the incident photon energy is set to the yb@xmath10 resonance of the @xmath16 absorption edge @xcite .
the experimentally determined deviation from the trivalent ( one hole ) yb state @xmath18 as a function of temperature is shown in figure [ fig : fig3]a for all studied yb kondo lattices .
the absolute changes in the @xmath1 occupancy and , in particular , the low temperature values @xmath19 are consistent with the expectation that @xmath20 increases as @xmath0 increases .
surprisingly however , the temperature scale @xmath3 on which the valence changes occur seems to be very similar in all measured compounds despite the very different kondo scales , @xmath0 .
here we arbitrarily identify @xmath3 with the temperature at which @xmath21 following @xcite .
the dashed lines are fits to the data using the empirical form @xmath22^s$ ] , with @xmath20 and @xmath3 free parameters and @xmath23 @xcite , which approximates the observed @xmath2 dependences well . in order to be able to better compare the temperature dependence for the different samples we show the changes in @xmath24 normalized to their low - temperature values in fig .
[ fig : fig3]b .
apparently , there is no significant difference in the characteristic temperature scale @xmath3 of the valence changes neither within the co doping series yb(rh@xmath11co@xmath12)@xmath13si@xmath13 nor with respect to the other measured yb kondo lattices .
it should be noted that the temperature dependences obtained here also resemble those reported in the literature for other yb kondo lattices , some with a @xmath0 as high as several hundred kelvin @xcite .
also in these cases a clear @xmath2 dependence in the yb valence between 3 and 300 k , the range most easily accessible to x - ray spectroscopies , was observed . for comparison
we show those data in fig .
[ fig : fig3]b together with our results .
the plot clearly demonstrates that even though the kondo scale differs by up to four orders of magnitude between the different compounds , the yb valence shows a very similar @xmath2 response regardless of @xmath0 .
in contrast , the measured low temperature values of the yb valence correlate very well with the kondo scale determined in other experiments .
we found only one case in the literature , data reported on ybalb@xmath14 , where the valences changes also set in around 50k but then seem to happen over a much larger temperature range @xcite
. holes normalized to the low - temperature value vs temperature .
the dashed lines are fits to the data from which the temperature scale of the valence fluctuations , @xmath3 , was obtained .
previously reported results on other yb kondo lattices @xcite are shown in ( b ) for comparison . ]
our findings are summarized in fig .
[ fig : fig4 ] where we plot for each kondo lattice compound the experimentally measured low temperature deviation from the trivalent yb state , i.e. @xmath20 , against @xmath3 , the temperature scale of the valence changes , and @xmath0 , the kondo energy scale inferred from magnetic susceptibility , entropy or mssbauer measurements @xcite .
for comparison , we again included the results reported in the literature @xcite in panel ( b ) and also ( a ) if @xmath3 has been determined from the x - ray spectroscopy data .
while there is a clear correlation between @xmath20 and @xmath0 , @xmath3 seems to be independent of either of these two . against ( a ) the characteristic temperature scale of the valence changes , @xmath3 , extracted from the data shown in fig .
[ fig : fig3 ] and ( b ) the kondo energy scale , @xmath0 , obtained from thermodynamic measurements .
the dashed lines are guides to the eye . ] within the usually used approximations the kondo model predicts that both the low temperature valence , @xmath25 , and the temperature scale , @xmath3 , on which the valence changes occur will increase with increasing kondo scale @xmath0 ( see e.g. @xcite ) .
fig .
4b clearly shows that there is indeed a very nice correlation between @xmath20 measured with rxes and the thermodynamically determined @xmath0 .
the linear relation between @xmath26 and @xmath27 suggests a power law dependence between both properties , i.e. @xmath28 . the dashed line in fig .
[ fig : fig4]b corresponds to @xmath29 and @xmath30 . within the kondo model both @xmath20 and @xmath0
depend on several parameters .
therefore the exact relation between both properties is a complex and non - universal one .
however , in a simple approximation , @xmath20 is proportional to @xmath31 , where @xmath32 stands for the hybridization width of the ground state doublet @xcite . @xmath0 itself depends exponentially on @xmath32 which results in a weak logarithmic dependence of @xmath32 on @xmath0 . in total that should lead to a sublinear dependence of @xmath20 on @xmath0 .
therefore , given the approximations made , the @xmath29 power law we observed is in reasonable agreement with theoretical expectations .
the observed perfect scaling between @xmath20 and @xmath0 is strongly indicating that the kondo effect is the dominant interaction responsible for @xmath33 .
it is therefore a big surprise that the @xmath2 induced increase of the yb valence towards 3 + occurs on almost the same temperature scale @xmath3 for all compounds ( fig .
4a ) , independently of @xmath20 and @xmath0 .
the kondo model has proven to be very successful in explaining even fine details of the properties of yb - kondo lattices @xcite , but it predicts the @xmath2 scale of the recovery of the 3 + state to be proportional to @xmath0 , an intuitively trivial relation .
one therefore needs to look for possible mechanisms that could modify the @xmath2 dependence expected within the simple approximation usually used for kondo systems .
a first candidate is the crystal electric field ( cef ) , which is not taken into account in those simple approximations , but which is known to be important in rare earth compounds .
cef leads to a splitting of the @xmath34 multiplet of yb@xmath17 into 4 doublets ( or one quartet and 2 doublets in case of a cubic local yb symmetry ) .
usually for kondo physics only the lowest doublet is considered but at higher temperatures , when @xmath2 becomes comparable to the cef splitting , excited cef levels get thermally populated and become relevant .
notably the overall cef splitting of yb compounds is typically of the order of 200k , the size of @xmath3 .
however , even for a single kondo ion , an exact calculation of @xmath6 including cef is extremely demanding , and therefore to the best of our knowledge has only been done for one very specific case @xcite . hence the effect of cef on the @xmath2 dependence of @xmath6 is not very clear at the moment .
naively , because the kondo scale depends exponentially on the degeneracy @xmath35 of the effective local moment , involvement of excited cef states corresponds to an increase in the effective degeneracy which should result in an increase of the effective kondo scale with increasing @xmath2 . because of the exponential dependence of @xmath0 on @xmath35 , this effect would become stronger upon decreasing @xmath0 of the ground state .
thus it would provide a way for making @xmath3 independent of the ground state @xmath0 ( for a more detailed argument , see appendix [ app : cef ] ) .
a similar effect would also be expected if the hybridization of an excited cef level is stronger than the hybridization strength of the ground state doublet .
different hybridization strength for different cef levels have rarely been considered so far .
but recent theoretical and experimental results indicate that the hybridization depends on the symmetry of the cef level @xcite , and that this might result in unexpected phenomena @xcite . in ybrh@xmath13si@xmath13 , where the hybridization of the cef ground state seems to be relatively weak , such an effect might become important .
an stm study of ybrh@xmath13si@xmath13 indeed provided evidence that in this compound the evolution of the kondo lattice at higher temperatures is strongly influenced by excited cef levels @xcite .
another factor which could be relevant for @xmath36 is the lattice .
it seems that thermal expansion does not have a significant effect on @xmath36 . in yb kondo
lattices increasing the volume results in a shift of the valence towards the 2 + state , because of its larger atomic volume @xcite .
therefore the normal thermal expansion should lead to an increase of @xmath24 with increasing @xmath2 , opposite to the experimental observation . on the other hand
, @xmath3 is about 300k to 400k , which is similar to the debye temperatures @xmath37 of these compounds .
@xmath37 does not show large variations between the studied yb compounds , typically differing by not more than @xmath38 .
these differences would be even smaller if one would consider a @xmath37 only corresponding to the vibration of yb atoms .
thus it is tempting to discuss the universal @xmath3 scale in terms of a phonon - induced mechanism .
the renormalization of the quasiparticles at the fermi level @xmath39 due to electron - phonon coupling is indeed expected to vanish on the scale of @xmath37 which would lead to a decrease of the renormalized electronic density of states at @xmath39 on the same @xmath2 scale . as the kondo energy is proportional to the renormalized electronic density of states
, one would therefore expect @xmath0 to decrease with increasing temperature on a @xmath2 scale related to the debye temperature .
but at the same time , the exchange coupling itself should also be affected by thermally excited phonons and a preliminary analysis indicates that it might increase with increasing @xmath2 , thus compensating the effect of the change in renormalization .
therefore the effect of enhanced thermal vibration is not yet clear and requires more theoretical and experimental studies . on a general level , in strongly correlated electron systems the effect of the lattice on the electronic properties
is usually just ignored .
however , recent studies indicate that taking the lattice into account might severely change the expected electronic properties @xcite . very recently
, a study of the effect of zero point and thermal ionic motion on the superexchange @xmath40 in cuprates indicates a strong motional broadening of @xmath40 giving rise to substantial effects on some electronic and magnetic properties @xcite .
our results imply that a determination of the kondo scale based on the @xmath2 dependence of the valence determined with x - ray spectroscopy , as frequently done , is at least questionable .
instead this @xmath2 dependence might not necessarily be related to the kondo scale .
rxes is intimately related to standard @xmath8 edge xas and other x - ray and electron spectroscopic techniques .
previous studies comparing the different techniques showed that hard and soft x - ray photoemission yield the same low @xmath2 valences and @xmath2 dependences as x - ray absorption / emission techniques for yb compounds @xcite .
part of the previously reported results on other yb compounds which we included in fig .
3 and 4 have been obtained with x - ray spectroscopies other than rxes and show the same @xmath2 dependence .
thus the surprising @xmath2 dependent behavior reported here is seen independent of the employed spectroscopic technique and seems to be universal for yb kondo lattice systems .
it would be interesting to see whether the valence changes in ce kondo lattices with temperature follow a similar universal temperature curve independent of the kondo scale @xmath0 @xcite .
such a result might provide some further insight into the origin of the unexpected temperature dependence of @xmath24 .
the high sensitivity and accuracy in determining @xmath33 with rxes and the nice relation observed between @xmath0 and @xmath20 suggest some immediate application in the field of kondo lattices . in kondo lattices where @xmath0 is of the order of or smaller than the magnetic ordering temperature , thermodynamic , transport and magnetic data can not reliably determine @xmath0 since all these properties are then controlled by the rkky energy scale .
in such systems , a rxes measurement of @xmath20 can provide a much more reliable estimation of @xmath0 using the power law scaling between @xmath0 and @xmath20 .
this can be easily understood since @xmath20 , in contrast to many other properties , is not affected by the rkky exchange . in summary
, we used resonant x - ray emission spectroscopy ( rxes ) to study the evolution of the valence in the kondo alloy yb(rh@xmath11co@xmath12)@xmath13si@xmath13 and a few further yb based kondo lattices as a function of composition and temperature .
the kondo temperature @xmath0 in the studied compounds cover the range from very small values @xmath41k up to quite large values @xmath42k . with present rxes techniques
the yb valence @xmath43 in such systems can be determined with very high accuracy .
this allows to study the deviation from a strict trivalent state , i.e. the deviation in the number of holes @xmath6 in the 4f shell from unity , @xmath44 , as function of composition and temperature with a precision of the order of 0.1% . combining the present results and literature data , we reveal an almost perfect linear correlation between @xmath45 determined from rxes at @xmath46k and @xmath27 determined from thermodynamic properties .
this linear correlations extends over 3 orders of magnitude , from @xmath41k and @xmath47 up to @xmath48k and @xmath49 . upon increasing @xmath2
, all systems evolve towards a pure trivalent state , but surprisingly the @xmath2 scale on which @xmath33 decreases with increasing @xmath2 is almost the same for all compounds , an thus does not scale at all with @xmath0 . looking for the origin of this unexpected universal @xmath2 scale we identify two possible candidates , excited cef levels and lattice vibrations .
however the effect of both properties on @xmath36 has not been studied yet and is thus presently not clear .
therefore our results urge for more experimental and theoretical studies on how excited cef levels and lattice vibrations affect the @xmath2 dependence of the valence .
we thank l. simonelli for her support during the experiments at the esrf .
this work was supported by the german research foundation ( dfg ) ( grants vy64/1 - 3 , ge602/2 - 1 , grk1621 and sfb1143 ) . |
the recent discovery of a higgs - like boson at the lhc @xcite represents a remarkable success for the standard model ( sm ) of particle physics .
however , within the sm the higgs mass is subject to additive renormalization , implying that a large hierarchy between the electroweak scale and the planck scale is technically unnatural @xcite .
one of few motivated models beyond the standard model ( bsm ) addressing this so - called fine - tuning problem is the framework of composite higgs models @xcite with the higgs as a pseudo - nambu - goldstone - boson associated with the spontaneous breakdown of an approximate global symmetry of a sector which becomes strongly coupled at a scale @xmath4 .
the global symmetry is explicitly broken by yukawa couplings of the higgs to the quarks and their composite partner states .
integrating out the quark partners yields an effective potential for the higgs .
the effective potential strongly depends on the embedding of the top quark partners into the global symmetry group as well as the top partner mass , while the lighter quark partners typically play a minor role .
concrete realizations of composite higgs models @xcite , electroweak precision constraints @xcite , and top - partner phenomenology @xcite have already been studied , while only few articles focussed on bottom partners @xcite or partners of other light quarks @xcite .
although typically not majorly contributing to electroweak symmetry breaking , light quarks need to be accompanied by composite partners in order to generate ( small but non - vanishing ) masses of the sm - like light quarks . the quark partner phenomenology depends on the embedding of the quarks and their partners into the global symmetry group of the composite sector , as well as their masses and couplings . in ref .
@xcite , quark partners of the up and charm quarks were studied within the minimal composite higgs model based on @xmath0 .
the right - handed quark partners were embedded in the @xmath5 of @xmath6 , which comprises of a @xmath7 and a singlet under the @xmath1 which is unbroken by strong dynamics .
while bounds for the partners in the @xmath7 were obtained , the singlet partner remained unconstrained because of its suppressed couplings to electroweak gauge bosons .
it was furthermore shown , that the presence of a light singlet state can substantially weaken the constraints on the partners in the fourplet , due to a combination of smaller production rates of fourplet states , and the opening of cascade decay processes of fourplet states via the singlet state . in this article , we focus on constraints of quark partners in the @xmath1 singlet in composite higgs models based on the minimal coset @xmath0 . in sec .
[ sec : models ] we discuss the two basic phenomenologically viable setups for composite quarks in the @xmath1 singlet , where the right - handed quark is either realized as an elementary quark which mixes with a composite partner ( partial compositeness ) or as a chiral state of the composite sector ( full compositeness ) .
we establish effective lagrangians for both these models and use those in order to discuss their lhc phenomenology in in sec .
[ sec : pheno ] . in particular we show that both setups can result in a substantial increase of di - higgs production above the sm background if the quark partners are light . in the current absence of direct bounds on the di - higgs production channel at the lhc
, we use the atlas bounds on differential cross sections of the higgs di - photon decay in order to obtain constraints on the partially and fully composite light quark models in sec .
[ sec : results ] , where we present our results in terms of the effective models discussed in sec .
[ sec : models ] .
we conclude in sec .
[ sec : conclusions ] .
as a generic setup , we consider the fermion sector of the minimal composite higgs model @xcite based on the coset structure @xmath6@xmath8@xmath1 . we follow the conventions and notation of ref .
@xcite and use the setups presented there for an initial discussion .
we start the discussion with the embedding of up - type partners . for concreteness , we embed the left - handed elementary quark doublet @xmath9 in an incomplete @xmath5 of @xmath6 @xcite as the @xmath9 carries a @xmath10 charge of @xmath11 .
the @xmath10 is included in order to generate the correct hyper charge of the quarks , which is obtained by gauging @xmath12 , where @xmath13 is the diagonal generator of the @xmath14 in @xmath15 . ) , [ defpsi ] which can be decomposed as a fourplet , @xmath16 , with a mass scale @xmath17 and a singlet , @xmath18 , with a mass scale @xmath19 of the @xmath1 which is unbroken by the strong dynamics . in this article
, we are studying the singlet partner and therefore take the limit @xmath20 in which the fourplet partners decouple , while the singlet partner @xmath21 remains as the only bsm particle .
the interactions of the @xmath21 depend on the embedding of the right - handed quarks .
they can be either embedded as a chiral composite @xmath6 singlet @xmath22 ( in `` fully composite '' models ) @xcite or as an incomplete representation of @xmath6 ( in partially composite " models ) which is a singlet state in terms of the @xmath1 . for concreteness
, we embed @xmath22 into the @xmath5 of @xmath6 as compositeness , because the mass eigenstate is a linear combination of the `` elementary '' quarks and the composite partner @xmath21 . in both fully and partially composite embeddings , the right - handed quarks have a @xmath10 charge of @xmath11 .
down - type partners can be embedded analogously with a @xmath10 charge of @xmath23 for @xmath24 , @xmath25 and @xmath26 .
and @xmath26 read @xmath27 and @xmath28 . ]
the above embedding is not unique for composite higgs models based on @xmath6@xmath8@xmath1 .
other embeddings of left - handed and partner quarks discussed in the literature include the symmetric @xmath29 representation of @xmath6 @xcite . representation .
however , the @xmath30 decomposes into @xmath31 in terms of @xmath1 representations and does not contain an @xmath1 singlet partner as is considered , here .
] however , we will be focussing on the @xmath1 singlet partner @xmath21 ( or the equivalent partner for down - type quarks ) .
our results are to a large part independent of the @xmath6 representation in which the quarks are embedded in , as long as it contains the @xmath1 singlet . in the following ,
we derive an effective description for the @xmath1 singlet , using partners in the @xmath32 as a guideline and comment on how this can be applied to other representations of quark partners .
we start with the partially composite model outlined in ref .
@xcite . using the callan - coleman - wess - zumino ( ccwz ) formalism @xcite ,
the fermion lagrangian of the partially composite model is = _
comp+_el , mix with _ comp = i |q(d_+i e _ ) ^q + i |uu -m_4|qq -m_1|+(i c |q^i ^d^i_+ ) , where @xmath33 and @xmath34 are the ccwz connections ( appendix a of ref .
@xcite for the explicit expressions ) , and _ el , mix = i |q_lq_l+i |u_ru_r - y_l f|q^5_l ( u_gs)_r -y_r f|_l ( u_gs)u_r+,[lelmixpc ] where @xmath35 and @xmath36 are functions of the goldstone matrix u_gs= ( ) , [ gmatru ] with appropriate @xmath6 index contractions such that the action is @xmath6 invariant . for @xmath37 , @xmath38 and @xmath22 in the @xmath32 these are simply @xmath39 . in the limit @xmath20 in which the @xmath1 fourplet states @xmath16 decouple , one obtains & = & i |uu -m_1| + i |q_lq_l + i |u_ru_r + & & -[pclag1 ] + & = & i |uu -m_1| + i |q_lq_l + i |u_ru_r + & & -.in the above , the functions @xmath40 and @xmath41 are @xmath42 and @xmath43 for @xmath37 , @xmath38 and @xmath22 in the @xmath32 . in order to obtain the final expression
, we expanded the lagrangian in @xmath44 which yields m_2&= & f f ( ) = - f ( ) , + m_3&= & y_r f g ( ) = y_r f ( ) , + _ 2&= & f ( ) = - ( ) , + _ 3&= & y_r g ( ) = -y_r ( ) . ) .
diagonalizing the mass matrix yields the mass squared eigenvalues m^2_u_l , u_h&= & + & = & ( m_1 ^ 2+m_2 ^ 2+m_3 ^ 2 ) . requiring that the lighter eigenvalue reproduces
the quark mass square of the sm quarks which is ( much ) smaller than @xmath19 for ( 1st and 2nd ) 3rd family quarks thus requires the last term in the square root to be small . expanding in it yields m^2_u_l&= & + m^2_u_h&=&(m_1 ^ 2+m_2 ^ 2+m_3 ^ 2 ) . matching the lighter mass eigenvalue to
the sm quark mass thus requires @xmath45 .
note that the @xmath46 and the @xmath47 arise from mass terms with inherently different symmetry properties .
the @xmath46 term links a fundamental 4-plet to a composite @xmath1 singlet while the @xmath47 term links a fundamental singlet to a composite 4-plet .
therefore , @xmath46 and @xmath47 are independent parameters which are not required to be of the same order of magnitude by naturalness . for @xmath48 ,
the left - handed part of the light mass eigenstate becomes a linear combination of the elementary @xmath37 and the composite @xmath49 ( left - handed partial compositeness ) .
such a mixing is strongly constraint because in this case , the coupling of the lighter quark mass eigenstate to the @xmath50 boson is modified .. for @xmath51 quark partners @xmath52 is modified at tree level . ]
we therefore focus on right - handed partial compositeness , where @xmath53 . in what follows ,
we take @xmath47 as a free @xmath54 parameter and neglect the effect of @xmath46 apart from fixing it in order to reproduce the correct quark mass in @xmath55 in the above . in this limit ,
the heavy quark mass becomes m_u_h= , [ pcmassh ] and the bi - unitary transformation which diagonalizes the mass matrix is the identity on the left - handed up - quarks and a rotation by @xmath56 on the right - handed up - quarks where = . ) , [ zlag ] in the mass eigenbasis @xmath57 .
note that the couplings arising from the @xmath10 gauge couplings are universal , and a rotation into the mass eigenbasis of these terms does not induce any `` mixed '' interactions of the @xmath50 to @xmath58 and @xmath59 .
such an interaction can only arise from the @xmath60 component of the @xmath50 in the first term , but as the mixing in the left - handed sector is small ( of order @xmath61 ) , the `` mixing '' couplings are negligible .
the same holds for mixing couplings of the @xmath62 gauge boson which follow from the analogous charged current interactions .
the higgs couplings to the quark mass eigenstates follow from _ yuk = -_3 h |_l u_r+ = -^eff_mixh|u_h , lu_l , r- ^effh|u_h , lu_h , r+ with ^eff_mix&=&_3 ( ) = y_r g(),[lambdapceff ] + ^eff&=&_3 ( ) = y^2_r g()g ( ) .
the coupling @xmath63 does not contribute to the production or decay of the partner quarks , and we neglect it for the effective description of the model . collecting all other terms , the effective partially composite quark model lagrangian is given by _
eff=_sm+|u_h(i+ e -g+ g_3 ) u_h - m_u_h|u_hu_h - .
[ lpceff ] the lagrangian and the definition of the effective coupling has been derived for up - type partners .
the analogous calculation for down - type partners yields the same lagrangian with the charge factors @xmath11 being replaced by @xmath23 as directly follows from the @xmath10 charge assignments . for illustration we embedded @xmath9 , @xmath38 and @xmath22 in the @xmath32 of @xmath6 in the above , but from the derivation it is apparent , that the effective lagrangian holds more generally .
as long as the quark partner multiplet contains one @xmath1 singlet , and in the limit in which all other partner states are decoupled , the only dependence on the chosen representation enters through the functions @xmath35 and @xmath36 in eq . via the functions @xmath40 and @xmath41 in eq . into the effective coupling @xmath64 in eq . .
as an illustration , table [ table : pc ] shows the corresponding functions and the effective coupling for an embedding of @xmath38 and @xmath22 in the @xmath32 and @xmath9 in the @xmath65 .
the extension to other embeddings which contain an @xmath1 singlet partner is straight forward .
we emphasize that for concrete realizations of partially composite quarks , the parameters @xmath64 and @xmath66 are correlated .
in particular , the mass of the quark partner is @xmath67 , such that a light partner mass @xmath68 implies an upper bound on @xmath19 and @xmath69 which in turn implies an upper bound on @xmath64 ( for fixed @xmath66 and @xmath4 , and depending on the embedding chosen ) . in sec .
[ sec : phenopc ] and [ sec : pcresults ] , we study the phenomenological implications of the more general effective model defined by eq . in terms of the parameters
@xmath64 and @xmath66 and point out the implications for specific quark embeddings .
.relations between the effective lagrangian and parameters given in eqs .
( [ pclag1],[lambdapceff],[lpceff ] ) and explicit embeddings of the quark states into the @xmath70 and @xmath71 of @xmath6 . [ cols="<,<,<,<,<,<",options="header " , ] in sec .
[ sec : results ] we conservatively did not include the @xmath72 and @xmath73 overflow bin data into our analysis . in this appendix
we repeat our analysis , including the overflow bins .
of course , the overflow bin data should be taken on a different footing .
, and bounds presented in this appendix are to be taken with a grain of salt .
the main purpose of this analysis is to demonstrate the impact of including high @xmath72 and @xmath73 events in searches for higgses from bsm events which like in our case originate from decays of bsm particles .
the unfolding correction factor for overflow bins are not specified in ref .
the specified correction factors for the other @xmath72 bins vary between 150% and 152% , while the correction factors for @xmath73 vary between 138% and 159% . for our analysis
we assumed the unfolding correction factor for overflow bins to be identical the correction factor of the highest @xmath74 bin of the respective distribution ( 150% for @xmath72 and 143% for @xmath73 ) .
[ fig : pcresultoverflow ] shows the 95% cl exclusion bounds on partners of partially composite light quarks in the @xmath75 vs. @xmath66 parameter space , which result from the highest @xmath72 bin , the highest @xmath73 bin , the @xmath76 , and the combined bound .
the flavor and @xmath64 independent bound arising from qcd pair produced quark partners is increased to m_u_h 385 as can be seen , the strongest constraint arises from the @xmath72 overflow bin , followed by the @xmath73 overflow bin . in sec .
[ sec : results ] the dominant bound arose from the @xmath77 distribution . this demonstrates the impact of including the high @xmath72 and @xmath73 events . for fully composite quark models
we find , that the bound is enhanced due to the @xmath72 overflow bin data while the @xmath73 overflow bin excess is too small to give an exclusion bound . combining the constraints obtained from @xmath72 and @xmath78 excesses we find a bound of m_u_h 240 g. t hooft , in proc .
of 1979 cargse institute on _ recent developments in gauge theories _ , p. 135 , plenum press , new york 1980 ; k. g. wilson , phys .
d * 3 * ( 1971 ) 1818 ; e. gildener , phys .
d * 14 * , 1667 ( 1976 ) ; s. weinberg , phys .
b * 82 * , 387 ( 1979 ) .
d. b. kaplan , h. georgi , phys .
* b136 * ( 1984 ) 183 .
d. b. kaplan , h. georgi , s. dimopoulos , phys .
* b136 * ( 1984 ) 187 .
h. georgi , d. b. kaplan , p. galison , phys .
* b143 * ( 1984 ) 152 . t. banks , nucl . phys .
* b243 * ( 1984 ) 125 .
h. georgi , d. b. kaplan , phys .
* b145 * ( 1984 ) 216 .
m. j. dugan , h. georgi , d. b. kaplan , nucl .
* b254 * ( 1985 ) 299 . n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) [ hep - ph/0206021 ] ; r. contino , y. nomura and a. pomarol , nucl .
b * 671 * , 148 ( 2003 ) [ hep - ph/0306259 ] .
k. agashe , r. contino and a. pomarol , nucl .
b * 719 * , 165 ( 2005 ) [ hep - ph/0412089 ] .
o. matsedonskyi , g. panico and a. wulzer , jhep * 1301 * , 164 ( 2013 ) [ arxiv:1204.6333 [ hep - ph ] ] . m. redi and a. tesi , jhep * 1210 * , 166 ( 2012 ) [ arxiv:1205.0232 [ hep - ph ] ] .
d. marzocca , m. serone and j. shu , jhep * 1208 * , 013 ( 2012 ) [ arxiv:1205.0770 [ hep - ph ] ] .
a. pomarol and f. riva , jhep * 1208 * , 135 ( 2012 ) [ arxiv:1205.6434 [ hep - ph ] ] .
g. panico , m. redi , a. tesi and a. wulzer , jhep * 1303 * , 051 ( 2013 ) [ arxiv:1210.7114 [ hep - ph ] ] .
c. grojean , o. matsedonskyi and g. panico , jhep * 1310 * , 160 ( 2013 ) [ arxiv:1306.4655 [ hep - ph ] ] .
n. vignaroli , phys .
d * 86 * ( 2012 ) 075017 [ arxiv:1207.0830 [ hep - ph ] ] .
a. azatov , m. salvarezza , m. son and m. spannowsky , arxiv:1308.6601 [ hep - ph ] .
n. vignaroli , jhep * 1207 * ( 2012 ) 158 [ arxiv:1204.0468 [ hep - ph ] ] . c. delaunay , t. flacke , j. gonzalez - fraile , s. j. lee , g. panico and g. perez , [ arxiv:1311.2072 [ hep - ph ] ] .
m. redi and a. weiler , jhep * 1111 * , 108 ( 2011 ) [ arxiv:1106.6357 [ hep - ph ] ] .
o. gedalia , j. f. kamenik , z. ligeti and g. perez , phys .
b * 714 * , 55 ( 2012 ) [ arxiv:1202.5038 [ hep - ph ] ] .
g. cacciapaglia , c. csaki , j. galloway , g. marandella , j. terning and a. weiler , jhep * 0804 * , 006 ( 2008 ) [ arxiv:0709.1714 [ hep - ph ] ] .
a. l. fitzpatrick , g. perez and l. randall , phys .
lett . * 100 * , 171604 ( 2008 ) [ arxiv:0710.1869 [ hep - ph ] ] .
d. pappadopulo , a. thamm and r. torre , jhep * 1307 * , 058 ( 2013 ) [ arxiv:1303.3062 [ hep - ph ] ] .
m. montull , f. riva , e. salvioni and r. torre , phys .
d * 88 * , 095006 ( 2013 ) [ arxiv:1308.0559 [ hep - ph ] ] .
a. azatov and j. galloway , phys .
d * 85 * , 055013 ( 2012 ) [ arxiv:1110.5646 [ hep - ph ] ] . c. delaunay , t. golling , g. perez and y. soreq , arxiv:1310.7029 [ hep - ph ] . j. m. butterworth , a. r. davison , m. rubin and g. p. salam , phys . rev
. lett . * 100 * , 242001 ( 2008 ) [ arxiv:0802.2470 [ hep - ph ] ] ; l. g. almeida , s. j. lee , g. perez , g. f. sterman , i. sung and j. virzi , phys .
d * 79 * , 074017 ( 2009 ) [ arxiv:0807.0234 [ hep - ph ] ] ; l. g. almeida , s. j. lee , g. perez , g. sterman and i. sung , phys .
d * 82 * , 054034 ( 2010 ) [ arxiv:1006.2035 [ hep - ph ] ] ; t. plehn , g. p. salam and m. spannowsky , phys .
* 104 * , 111801 ( 2010 ) [ arxiv:0910.5472 [ hep - ph ] ] ; d. e. soper and m. spannowsky , jhep * 1008 * , 029 ( 2010 ) [ arxiv:1005.0417 [ hep - ph ] ] ; j. -h .
kim , phys .
d * 83 * , 011502 ( 2011 ) [ arxiv:1011.1493 [ hep - ph ] ] ; l. g. almeida , o. erdogan , j. juknevich , s. j. lee , g. perez and g. sterman , phys .
d * 85 * , 114046 ( 2012 ) [ arxiv:1112.1957 [ hep - ph ] ] ; m. backovic , j. juknevich and g. perez , jhep * 1307 * , 114 ( 2013 ) [ arxiv:1212.2977 [ hep - ph ] ] .
j. r. espinosa , c. grojean , m. muhlleitner and m. trott , jhep * 1205 * , 097 ( 2012 ) [ arxiv:1202.3697 [ hep - ph ] ] .
d. carmi , a. falkowski , e. kuflik , t. volansky and j. zupan , jhep * 1210 * , 196 ( 2012 ) [ arxiv:1207.1718 [ hep - ph ] ] .
g. moreau , phys .
d * 87 * , 015027 ( 2013 ) [ arxiv:1210.3977 [ hep - ph ] ] .
c. delaunay , c. grojean and g. perez , jhep * 1309 * , 090 ( 2013 ) [ arxiv:1303.5701 [ hep - ph ] ] .
a. alloul , n. d. christensen , c. degrande , c. duhr and b. fuks , arxiv:1310.1921 [ hep - ph ] . c. degrande , c. duhr , b. fuks , d. grellscheid , o. mattelaer and t. reiter , comput .
commun .
* 183 * ( 2012 ) 1201 [ arxiv:1108.2040 [ hep - ph ] ] .
n. christensen , b. fuks , c. duhr , + http://feynrules.irmp.ucl.ac.be/wiki/standardmodel s. heinemeyer _ et al .
_ [ lhc higgs cross section working group collaboration ] , arxiv:1307.1347 [ hep - ph ] .
j. alwall , m. herquet , f. maltoni , o. mattelaer and t. stelzer , jhep * 1106 * ( 2011 ) 128 [ arxiv:1106.0522 [ hep - ph ] ] .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky and w. k. tung , jhep * 0207 * ( 2002 ) 012 [ hep - ph/0201195 ] .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * ( 2006 ) 026 [ hep - ph/0603175 ] .
j. de favereau , c. delaere , p. demin , a. giammanco , v. lematre , a. mertens and m. selvaggi , arxiv:1307.6346 [ hep - ex ] .
m. cacciari and g. p. salam , phys .
b * 641 * ( 2006 ) 57 [ hep - ph/0512210 ] . | in composite higgs models , the generation of quark masses requires the standard model - like quarks to be partially or fully composite states which are accompanied by composite quark partners . the composite quark partners decay into a standard model - like quark and an electroweak gauge boson or higgs boson , which can be searched for at the lhc .
in this article , we study the phenomenological implications of composite quarks in the minimal composite higgs model based on the coset @xmath0 .
we focus on light quark partners which are embedded in the @xmath1 singlet representation . in this case , a dominant decay mode of the partner quark is into a higgs boson and a jet , for which no experimental bounds have been established so far .
the presence of @xmath1 singlet partners leads to an enhancement of the di - higgs production cross section at the lhc .
this will be an interesting experimental signature in the near future , but , unfortunately , there are no direct bounds available yet from the experimental analyses . however , we find that the currently available standard model - like higgs searches can be used in order to obtain the first constraints on partially and fully composite quark models with light quark partners in the @xmath1 singlet .
we obtain a flavor- and composition parameter independent bound on the quark partner mass of @xmath2 for partially composite quark models and @xmath3 for fully composite quark models . |
this work was supported by ibs under the project code , ibs - r018-d1 [ kc and shi ] , and by pusan national university research grant , 2015 [ ksj ] , and the research fund program of research institute for basic sciences , pusan national university , korea , 2015 , project no .
ribs - pnu-2015 - 303 [ ksj ] . for a recent review ,
see j. e. kim and g. carosi , rev .
phys . * 82 * , 557 ( 2010 ) ; a. ringwald , phys .
dark univ .
* 1 * ( 2012 ) 116 ; m. kawasaki and k. nakayama , ann . rev .
nucl . part .
sci . * 63 * , 69 ( 2013 ) .
t. hiramatsu , m. kawasaki , k. i .
saikawa and t. sekiguchi , phys .
d * 85 * , 105020 ( 2012 ) [ erratum - ibid .
d * 86 * , 089902 ( 2012 ) ] ; jcap * 1301 * , 001 ( 2013 ) ; m. axenides , r. h. brandenberger and m. s. turner , phys . lett .
b * 126 * , 178 ( 1983 ) ; d. seckel and m. s. turner , phys .
d * 32 * , 3178 ( 1985 ) ; a. d. linde , phys .
b * 158 * , 375 ( 1985 ) ; d. h. lyth , phys . lett .
b * 236 * , 408 ( 1990 ) ; m. s. turner and f. wilczek , phys .
lett . * 66 * , 5 ( 1991 ) .
p. fox , a. pierce and s. d. thomas , hep - th/0409059 ; m. beltran , j. garcia - bellido and j. lesgourgues , phys .
d * 75 * , 103507 ( 2007 ) ; m. p. hertzberg , m. tegmark and f. wilczek , phys .
d * 78 * , 083507 ( 2008 ) ; k. j. mack and p. j. steinhardt , jcap * 1105 * , 001 ( 2011 ) ; k. j. mack , jcap * 1107 * , 021 ( 2011 ) .
d. j. e. marsh , d. grin , r. hlozek and p. g. ferreira , phys .
lett . * 113 * , 011801 ( 2014 ) ; l. visinelli and p. gondolo , phys .
lett . * 113 * , 011802 ( 2014 ) ; a. g. dias , a. c. b. machado , c. c. nishi , a. ringwald and p. vaudrevange , jhep * 1406 * , 037 ( 2014 ) .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron . astrophys . * 571 * , a22 ( 2014 ) . n. arkani - hamed , l. motl , a. nicolis and c. vafa , jhep * 0706 * , 060 ( 2007 ) .
a. d. linde , phys .
b * 259 * , 38 ( 1991 ) .
h. murayama , h. suzuki and t. yanagida , phys .
b * 291 * , 418 ( 1992 ) ; k. choi , e. j. chun and j. e. kim , phys .
b * 403 * , 209 ( 1997 ) .
e. j. chun , k. dimopoulos and d. lyth , phys .
d * 70 * , 103510 ( 2004 ) .
k. choi , k. s. jeong and m. s. seo , jhep * 1407 * , 092 ( 2014 ) .
e. j. chun , phys . lett .
b * 735 * , 164 ( 2014 ) .
k. choi , k. s. jeong , k. i. okumura and m. yamaguchi , jhep * 1106 * , 049 ( 2011 ) ; g. honecker and w. staessens , fortsch . phys . *
62 * , 115 ( 2014 ) ; e. i. buchbinder , a. constantin and a. lukas , phys .
d * 91 * , no . 4 , 046010 ( 2015 ) .
k. s. jeong and f. takahashi , phys .
b * 727 * , 448 ( 2013 )
. t. higaki , k. s. jeong and f. takahashi , phys .
b * 734 * , 21 ( 2014 ) .
m. fairbairn , r. hogan and d. j. e. marsh , phys .
d * 91 * , no . 2 , 023509 ( 2015 ) .
n. kitajima and f. takahashi , jcap * 1501 * , no .
01 , 032 ( 2015 ) . j. e. kim and h. p. nilles , phys .
b * 138 * , 150 ( 1984 ) . d. h. lyth and e. d. stewart , phys .
lett . * 75 * , 201 ( 1995 ) ; d. h. lyth and e. d. stewart , phys .
d * 53 * , 1784 ( 1996 ) .
i. affleck and m. dine , nucl .
b * 249 * , 361 ( 1985 ) ; m. dine , l. randall and s. d. thomas , nucl .
b * 458 * , 291 ( 1996 ) .
k. choi , s. h. i m and k. s. jeong , work in progress . | we propose a new mechanism to suppress the axion isocurvature perturbation , while producing the right amount of axion dark matter , within the framework of supersymmetric axion models with the axion scale induced by supersymmetry breaking .
the mechanism involves an intermediate phase transition to generate the higgs @xmath0-parameter , before which the weak scale is comparable to the axion scale and the resulting stronger qcd yields an axion mass heavier than the hubble scale over a certain period .
combined with that the hubble - induced axion scale during the primordial inflation is well above the intermediate axion scale at present , the stronger qcd in the early universe suppresses the axion fluctuation to be small enough even when the inflationary hubble scale saturates the current upper bound , while generating an axion misalignment angle of order unity .
the non - observation of the neutron edm requires the cp violating qcd angle to be as tiny as @xmath1 , causing the strong cp problem .
an appealing solution of this puzzle is to introduce a spontaneously broken global peccei - quinn ( pq ) symmetry @xcite .
then @xmath2 corresponds to the vacuum value of the associated nambu - goldstone boson , the axion , which is determined to be vanishing by the low energy qcd dynamics @xcite .
an interesting consequence of this solution is that axions can explain the dark matter in our universe .
yet , the prospect for axion dark matter depends on the cosmological history of the pq phase transition .
a possible scenario is that the spontaneous pq breaking occurs after the primordial inflation is over .
in such a case , the model is constrained to have the domain - wall number @xmath3 , where @xmath4 corresponds to the integer - valued u@xmath5su@xmath6su@xmath7 anomaly coefficient .
then axions are produced mainly by the annihilations of axionic strings and domain - walls , which would result in the right amount of axion dark matter for the axion scale @xmath8 gev @xcite .
however it appears to be difficult to realize this scenario within the framework of a fundamental theory such as string theory , since it requires a pq symmetry with @xmath9 , as well as a restored pq phase until some moment after the primordial inflation .
another scenario which we will focus on in this paper is that u@xmath10 is spontaneously broken during the primordial inflation and never restored afterwards .
then the model is not subject to the condition @xmath9 , but is constrained by the axion isocurvature perturbation @xcite . for instance , from the observed cmb power spectrum , one finds @xcite , [ iso - constraint ] ( ) _ iso ( ) < 3.810 ^ -6 , where @xmath11 and @xmath12 denote the average misalignment angle and the angle fluctuation , respectively , for the axion field right before the conventional qcd phase transition when @xmath13 with a temperature @xmath14 gev .
the relic axion density is given by [ axion - density ] 1.7 ^2_mis ( ) ^1.19 , with @xmath15 being the total dark matter fraction .
here we have assumed that @xmath16 and there is no significant evolution of @xmath17 from @xmath18 to the present time @xmath19 so that @xmath20 . in inflationary cosmology , the primordial quantum fluctuation of the axion field results in [ angle - fluctuation ] ( t_qcd ) = ( t_i)= , where @xmath21 and @xmath22 denote the axion scale and the hubble parameter , respectively , during the primordial inflation epoch @xmath23 , and the factor @xmath24 is introduced to take into account the evolution of @xmath12 from @xmath23 to @xmath18 . note that the inflationary hubble scale @xmath22 is bounded by the tensor - to - scalar ratio of the cmb perturbation as [ hubble - bound ] r 0.16()^2 < 0.11,and the weak gravity conjecture @xcite suggests that generic axion scales are bounded as [ weak - gravity ] f_a o(m_pl),where @xmath25 gev is the reduced planck mass . to discuss the implication of the isocurvature constraint ( [ iso - constraint ] )
, one needs to specify the cosmological evolution of the axion scale after the primordial inflation is over .
if @xmath26 as has been assumed in most of the previous studies , it requires that either @xmath22 is smaller than its upper bound @xmath27 gev by at least five orders of magnitude , so that the cmb tensor mode is too small to be observable , or @xmath12 should experience a large suppression after the primordial inflation , which appears to be difficult to be implemented .
the above observation suggests a more attractive scenario realizing @xmath28 @xcite in a natural manner . indeed supersymmetric axion models offer a natural scheme to realize such a scenario , generating the axion scale through the competition between the tachyonic susy breaking mass term and a supersymmetric , but planck - scale - suppressed higher dimensional term in the scalar potential @xcite .
one then finds [ axion - scale ] f_a(t_0 ) & ~ & , + f_a(t_i ) & ~ & , which explains elegantly the origin of an intermediate axion scale at present , while giving a hubble - induced inflationary axion scale well above the present axion scale , if the supersymmetry ( susy ) breaking mass @xmath29 at present is around tev scale .
furthermore , this type of axion models can be successfully embedded into string theory .
specifically , they can be identified as a low energy limit of string models involving an anomalous u@xmath30 gauge symmetry with vanishing fayet - illiopoulos term @xcite . in such string models ,
the u@xmath30 gauge boson is decoupled from the low energy world by receiving a heavy mass @xmath31 through the stckelberg mechanism , while leaving the global part of u@xmath30 as an unbroken pq symmetry in the supersymmetric limit .
once susy breaking is introduced properly , in both the present universe and the inflationary early universe , the residual pq symmetry can be spontaneously broken to generate the axion scales as ( [ axion - scale ] ) . in this paper
, we discuss a novel mechanism to suppress the axion isocurvature perturbation , while producing the right amount of axion dark mater , within the framework of supersymmetric axion models with the axion scales given by ( [ axion - scale ] ) . the isocurvature constraint ( [ iso - constraint ] ) and
the relic axion density ( [ axion - density ] ) suggest that for @xmath22 near the current upper bound @xmath27 gev , the allowed amount of axion dark matter is maximal when @xmath32@xmath33 gev , while @xmath21 nearly saturates the weak gravity bound ( [ weak - gravity ] ) , e.g. @xmath34@xmath35 gev .
interestingly , the axions scales generated by susy breaking as ( [ axion - scale ] ) automatically realize such pattern if @xmath29 is around tev scale . more specifically , for the case f_a(t_0)/f_a(t_i ) , the isocurvature bound ( [ iso - constraint ] ) reads off < ( ) ^2 ( ) ( ) ^0.8 ( ) , when combined with ( [ axion - density ] ) .
this implies that a high scale inflation scenario with @xmath36@xmath37 gev , which would give an observable tensor - to - scalar ratio @xmath38@xmath39 in the cmb perturbation , can be compatible with the axion dark matter @xmath40 , if the axion field fluctuation experiences just a mild suppression after @xmath23 , e.g. @xmath41@xmath39 in ( [ angle - fluctuation ] ) . to suppress @xmath12 through its cosmological evolution , one needs a period with @xmath42 well before @xmath18 . on the other hand ,
usually this is not easy to be realized because the axion mass should be generated mostly by the qcd anomaly in order for the strong cp problem solved by the pq mechanism .
( see refs .
@xcite for an alternative possibility . ) in the following , we propose a simple scheme to achieve such a cosmological period by having a phase of stronger qcd in the early universe
. our scheme is based on a phase transition at @xmath43 , which will be called the @xmath0-transition in the following as it generates the higgs @xmath0-parameter through the superpotential term @xcite , [ kim - nilles ] ( x)h_uh_d , where @xmath44 is a pq - charged gauge - singlet superfield .
specifically , x(tt_)=0 , x(t > t_)~ , so that ( tt_)=0 , ( t > t _ ) ~m_susy . with this transition ,
the weak scale experiences an unusual evolution in a way that the weak scale before the @xmath0-transition is comparable to the axion scale ( [ axion - scale ] ) , as will be discussed below . to proceed ,
let us discuss first the key features of the scheme , and later present an explicit model to realize the whole ingredients . including the hubble - induced contribution , the mass of the @xmath45-flat higgs direction @xmath46 is generically given by m_^2 = c_h^2 + _ m_susy^2 + 2||^2 ( ^2h_uh_d ) , where @xmath47 and @xmath48 are model - dependent parameters of order unity . in our scheme
, both @xmath47 and @xmath48 are assumed to be negative , so @xmath49 before the @xmath0-transition . then @xmath50 is stabilized by the competition between the tachyonic @xmath51 and a supersymmetric term of @xmath52 in the scalar potential , which results in f_a(t_i)&~ & ( t_i)~ , + f_a(t_)&~ & ( t _ ) ~ . on the other hand , after the @xmath0-transition , @xmath53 due to @xmath54 .
the resulting weak scale and axion scale at present are given by ( t_0 ) & = & o(100 ) gev , + f_a(t_0 ) & ~ & x(t_0 ) ~ .
a simple consequence of the above evolution of @xmath46 is that the weak scale is comparable to the axion scale before the @xmath0-transition : ( tt_)~ f_a f_a(tt _ ) .
this results in a higher qcd scale , i.e. a stronger qcd , and therefore a heavier axion mass which might be even bigger than the hubble scale for a certain period .
let us estimate the qcd scale @xmath55 before the @xmath0-transition , which is defined as the scale where the 1-loop qcd coupling blows up , as well as the resulting axion mass @xmath56 . for the case with @xmath57 , where @xmath58 denotes the gluino mass before the @xmath0-transition
, we find _ qcd & & 23tev ( ) ^2/11 + & & ( ) ^3/11 ( ) ^6/11 , where @xmath59 at present , and @xmath60 for the gluino mass @xmath61 at present . here we assume that @xmath62 and @xmath63 for the qcd coupling and the quark yukawa couplings . when the temperature @xmath64 , the axion mass during the period of stronger qcd is estimated to be [ axion - mass-1 ] m_a ^2_qcd / f_a . on the other hand , if @xmath65 , the resulting qcd scale is estimated as _
qcd & & 21 tev ( ) ^1/3 ( ) ^2/3 , with the axion mass [ axion - mass-2 ] m_a m^1/2_g ^3/2_qcd / f_a .
here @xmath66 denotes the scale where the stronger qcd becomes nonperturbative , i.e. around @xmath67 with @xmath68 .
note that the axion potential for the axion mass ( [ axion - mass-2 ] ) can be obtained by a single insertion of the susy breaking spurion @xmath69 to the nonperturbative superpotential @xmath70 induced by the gluino condensation .
if the stronger qcd scale @xmath71 is high enough , there could be a period with @xmath72 well before the conventional qcd phase transition .
as is well known , in such a period the axion field experiences a damped oscillation , with an amplitude @xmath73 ( averaged over each oscillation period ) evolving as |a r^-3/2(t ) , where @xmath74 is the scale factor of the expanding universe .
then the spatially averaged vacuum value of the axion field is settled down at the minimum of the axion potential induced by the stronger qcd , while the axion angle fluctuation is diluted according to [ angle - fluc ] = ( ) ^3/2 , where @xmath75 denotes the moment when the damped axion oscillation begins , and @xmath76 is the moment when it is over . note that , after the @xmath0-transition , the weak scale and the qcd scale quickly roll down to the present values , so the axion mass becomes negligible compared to @xmath77 until @xmath78 when the universe undergoes the conventional qcd phase transition . also , the minimum of the axion potential induced by the stronger qcd is generically different from the minimum of the axion potential at present . as a result , our scheme generates an axion misalignment angle of order unity : _ mis - = o(1 ) , together with an intermediate axion scale at present , so gives rise to @xmath40 in a natural way . in our case
, the damped axion oscillation induced by the stronger qcd begins at a temperature @xmath79 as @xmath56 is highly suppressed by thermal effects for @xmath80 .
on the other hand , the scalar field @xmath44 generating @xmath0 through ( [ kim - nilles ] ) is trapped at the origin by thermal effects until the universe cools down to a temperature @xmath81 .
in fact , our scheme involves a variety of dimensionless parameters which affect the naive estimate of the involved scales .
we find that there is a large fraction of the natural parameter region where the axion mass m_a 0.4 mev ( ) ^-1 ( ) ^2 is larger than the hubble scale [ hubble ] h(t _ ) 0.2 mev ( ) , over the period @xmath82 with a temperature ratio : t(t_)/t(t_i)= o(10 ^ -110 ^ -2 )
. then the resulting @xmath12 given by ( [ angle - fluc ] ) can be small enough to satisfy the isocurvature bound ( [ iso - constraint ] ) even when @xmath22 saturates its upper bound @xmath27 gev .
note that during @xmath82 , ( t)-(t_0)~ x(t)-x(t_0 ) ~ , so the corresponding vacuum energy density @xmath83 .
this means that in this period the universe is dominated by the vacuum energy density with the hubble scale given by ( [ hubble ] ) , which is often called the thermal inflation @xcite .
it should be stressed that in our scheme the axion isocurvature perturbation is suppressed by two steps .
the first suppression is due to @xmath84 , and the second is due to the stronger qcd dynamics before the @xmath0-transition , yielding a further suppression by @xmath85 . to illustrate the result
, we depict in fig .
[ fig : iso - bound ] the upper bound on the inflationary hubble scale @xmath22 resulting from the isocurvature constraint ( [ iso - constraint ] ) for @xmath40 . to make a comparison
, we depict the results for three distinct cases : @xmath86 the conventional scenario of @xmath87 without a stronger qcd , @xmath88 a scheme with @xmath89 , but without a stronger qcd , @xmath90 our scheme with @xmath89 and a stronger qcd before the @xmath0-transition .
l let us now present an explicit model implementing the mechanisms discussed above . as a simple example , we consider a model with the following superpotential , [ model-1 ] & & w = ( ) + y^c + & & + + + , where @xmath44 and @xmath91 are pq - charged gauge singlets responsible for the @xmath0-transition , @xmath92 is the mssm lepton doublet , and @xmath93 are @xmath94-charged exotic matter fields introduced to give a thermal mass to @xmath91 .
then the scalar potential for the @xmath0-transition is given by [ mu - transition ] v_1 & = & m_x^2|x|^2+m_y^2|y|^2 + ( xy^3+h.c . ) + & & + ( |y|^6 + 9|x|^2|y|^4 ) , where m_x^2 & = & c_xh^2+_xm_susy^2 + 4|_x|^2 , + m_y^2 & = & c_yh^2+_ym_susy^2 + _ yt^2 , for @xmath95 .
here @xmath96 are the hubble - induced masses , @xmath97 are the susy breaking masses at zero temperature , and @xmath98 is the thermal mass of @xmath91 induced by the coupling @xmath99 , which is of @xmath100 for @xmath101 . for simplicity
, we will assume that all the dimensionless parameters appearing in the superpotential and the susy breaking scalar masses are of order unity .
however it should be noted that these parameters can have a variation of @xmath102@xmath103 easily . in particular , the superpotential parameters @xmath104 can have a much wider variation without invoking fine - tuning .
this gives us a rather large room to get an enough suppression of the axion angle fluctuation @xmath12 through a stronger qcd before the @xmath0-transition . at any rate , assuming that @xmath105 , @xmath106 and @xmath107 , the scalar potential ( [ mu - transition ] ) indeed yields the desired @xmath0-transition as [ vev1 ] & & x = y=0 tt _ , + & & x~y ~ t > t _ , with @xmath81 .
now the higgs and slepton fields can have a nontrivial evolution along the following flat direction : & & h^t_d = ( _ d,0),l^t = ( _ , 0 ) , + & & h^t_u = ( 0 , ) , which satisfies the @xmath108 and @xmath45 flat conditions .
the relevant terms of the scalar potential of @xmath109 are given by v_2&= & m_i^2|_i|^2 + ( b_d + h.c . ) + & + & ( |_i|^2 ) ( + h.c . ) + & + & ( |_i|^2 ) ( |_d|^4 + 4|_d_|^2+|_|^4 ) , for @xmath110 , where m__d^2 & = & c_d h^2 + _ d m_susy^2 + 2||^2 , + m__^2&= & c_h^2 + _ m_susy^2 + ||^2 . again , assuming @xmath111 and @xmath112 , but @xmath113 due to @xmath114 , the above scalar potential yields [ vev2 ] f_a(t_i)~_d,(t_i)~ , + f_a(t_)~_d,(t_)~ , and [ vev3 ] _
d(t_0 ) & = & o(100 ) gev , _
( t_0 ) = 0 , + f_a(t_0 ) & ~ & x(t_0 ) ~ y(t_0)~
. to summarize , under a reasonably plausible assumption on the susy breaking during the primordial inflation and in the present universe , the model with the superpotential ( [ model-1 ] ) can successfully realize the desired cosmological evolution of the three relevant scales : the axion scale , the weak scale , and the qcd scale as given by ( [ vev1 ] ) , ( [ vev2 ] ) and ( [ vev3 ] ) . being generated by susy breaking , an inflationary axion scale @xmath115 is determined to be well above the present axion scale @xmath116 , and a stronger qcd in the early universe is realized to yield an enough suppression of the axion angle fluctuation even when @xmath22 saturates its upper bound .
we note that the minimum of the axion potential induced by the stronger qcd depends on @xmath117 , but not on @xmath118 , while the minimum of the axion potential at present depends on @xmath118 , but not on @xmath117 . as a result
, the stronger qcd generates an axion misalignment angle @xmath119 , so that the axion dark matter with @xmath40 arises naturally in our scheme .
there is a remaining issue which should be addressed to complete our scheme .
as we have noticed , the @xmath0-transition is foregone by a late - time thermal inflation .
this suggests that the scheme should be accompanied by a late - time baryogenesis operating after the @xmath0-transition .
in fact , the model of ( [ model-1 ] ) offers an elegant mechanism to generate the baryon asymmetry through the rolling flat direction @xmath120 @xcite .
more detailed cosmology of our scheme , including the leptogenesis by rolling @xmath120 , will be discussed elsewhere @xcite . |
the prediction of extra dimensions , which the brane - confined standard model is oblivious to , suggests that early enough in the history of the universe a dimensional selection would have taken place , implying the breaking of the lorentz symmetry at the level of the higher dimensional bulk spacetime and the possible generation of a different geometry along the directions orthogonal to the brane .
the breaking of the lorentz symmetry can be realized by spontaneous breaking of a symmetry in the presence of a vector field @xmath0 living in the bulk and subject to a potential @xmath1 the acquisition by the bulk vector field @xmath0 of a non - vanishing vacuum value and the consequent assignment to the bulk vacuum of an intrinsic direction determined by @xmath2 induce the breaking of rotation invariance and thus of lorentz symmetry in the bulk .
the brane , regarded as the locus of the observable universe embedded in the higher dimensional bulk spacetime , would be expected to share a subgroup of the symmetries of the bulk preserved in a manner akin to the goldstone mechanism .
the overly tested lorentz symmetry of the observable universe imposes that the brane must stand as a vacuum solution where the lorentz invariance would be a residual symmetry of the spontaneously broken symmetry @xcite .
the possibility of violation of lorentz invariance has nevertheless been widely discussed in the recent literature ( see e.g. @xcite ) .
spontaneous breaking of lorentz symmetry may arise in the context of string / m - theory due to the existence of non - trivial solutions in string field theory @xcite , in loop quantum gravity @xcite and in noncommutative field theories @xcite , or via the spacetime variation of fundamental couplings @xcite .
lorentz violation modifications of the dispersion relations via five dimensional operators for fermions have also been considered and constrained @xcite .
consequently , this putative breaking will have astrophysical @xcite implications , and in what concerns ultra - high energy cosmic rays , one can establish that lorentz symmetry holds up to about @xmath3 @xcite . the way to relate the breaking of lorentz symmetry to gravity has been considered in ref .
@xcite , and solutions as well as implications have been discussed in ref .
@xcite . in ref .
@xcite , a connection between the cosmological constant and the violation of lorentz invariance has been conjectured . for a general discussion on the relation between spacetime symmetries and higher dimensions ,
the reader is referred to ref .
@xcite . in this contribution
we report on a recent study on how spontaneous lorentz violation in the bulk has repercussions on the brane and how it can be constrained .
we consider a bulk vector field coupled non - minimally to the graviton which , upon acquiring a non - vanishing expectation value in the vacuum , introduces spacetime anisotropies in the gravitational field equations through the coupling with the graviton @xcite .
first we derive the bulk field equations and project them parallel and orthogonal to the brane .
we then establish how to derive brane quantities from bulk quantities by adopting fermi normal coordinates with respect to the directions on the brane and continuing into the bulk using the gauss normal prescription .
finally , we obtain the conditions for the lorentz invariance to be a symmetry on the brane .
we begin by parameterizing the world - sheet in terms of coordinates @xmath4 intrinsic to the brane . using the chain rule
, we may express the brane tangent and normal unit vectors in terms of the bulk basis as follows : @xmath5 with @xmath6 where @xmath7 is the bulk metric @xmath8 to obtain the metric induced on the brane we expand the bulk basis vectors in terms of the coordinates intrinsic to the brane and keep the doubly brane tangent components only .
it follows that @xmath9 is the @xmath10-dimensional metric induced on the brane by the @xmath11-dimensional bulk metric @xmath12 the induced metric with upper indices is defined by the relation @xmath13 it follows that we can write any bulk tensor field as a linear combination of mutually orthogonal vectors on the brane , @xmath14 and a vector normal to the brane , @xmath15 we illustrate the example of a vector @xmath0 and a tensor @xmath16 bulk fields as follows @xmath17 derivative operators decompose similarly .
we write the derivative operator @xmath18 as @xmath19 fixing a point on the boundary , we introduce coordinates for the neighbourhood choosing them to be fermi normal .
all the christoffel symbols of the metric on the boundary are thus set to zero , although the partial derivatives do not in general vanish .
the non - vanishing connection coefficients are @xmath20 as determined by the gaussian normal prescription for the continuation of the coordinates off the boundary .
for the derivative operator @xmath21 we find that @xmath22 + g^{nn}\left [ ( n^\mu\nabla_\mu)(n^\nu\nabla_\nu ) -n^\mu(\nabla_\mu n^\nu)\nabla_\nu \right ] \cr & = & g^{ab}\left [ \nabla_a \nabla_b + k_{ab}\nabla_n\right ] + \nabla_n \nabla_n.\end{aligned}\ ] ] we can now decompose the riemann tensor , @xmath23 along the tangent and the normal directions to the surface of the brane as follows @xmath24 from which we find the decomposition of the einstein tensor , @xmath25 obtaining the gauss - codacci relations @xmath26
in order to study the gravitational effects of the breaking of lorentz symmetry in a braneworld scenario , we consider a bulk vector field @xmath27 with a non - minimal coupling to the graviton in a five - dimensional anti - de sitter space .
the lagrangian density consists of the hilbert term , the cosmological constant term , the kinetic and potential terms for @xmath27 and the @xmath27graviton interaction term , as follows @xmath28 where @xmath29 is the tensor field associated with @xmath0 and @xmath30 is the potential which induces the spontaneous global symmetry breaking when the @xmath27 field is driven to the minimum at @xmath31 , @xmath32 being a real positive constant . here , @xmath33 @xmath34 is the five - dimensional planck mass and @xmath35 is a dimensionless coupling constant that we have inserted to track the effect of the interaction which enabled a simplification of the results without loss in generality of the purpose of the paper . here
, however , we shall keep @xmath35 free . ] . in the cosmological constant term @xmath36
we have included both the bulk vacuum value @xmath37 and that of the brane @xmath38 described by a brane tension @xmath39 localized on the locus of the brane , @xmath40 by varying the action with respect to the metric , we obtain the einstein equation @xmath41 where @xmath42\end{aligned}\ ] ] are the contributions from the interaction term and @xmath43 \quad\end{aligned}\ ] ] is the contribution from the vector field for the stress - energy tensor . for the equation of motion for the vector field @xmath44 obtained by varying the action with respect to @xmath45 we find that @xmath46 where @xmath47 we now proceed to project the equations parallel and orthogonal to the surface of the brane .
following the prescription used in the derivation of the gauss - codacci relations , we derive the components of the stress - energy tensor and of the interaction terms .
for the stress - energy tensor we find that @xmath48 , \cr t_{an } & = & b_{ac}b_{n}{}^{c } + 4v^{\prime}b_{a}b_{n},\cr t_{nn } & = & b_{nc}b_{n}{}^{c } + 4v^{\prime}b_{n}b_{n } + g_{nn}\left [ -{1\over 4}\left ( b_{cd}b^{cd } + 2b_{cn}b^{cn}\right ) -v\right],\end{aligned}\ ] ] and for the interaction source terms that @xmath49\cr & & { } -b_{a}\left [ b^{c}\left ( r^{(ind)}_{cb } + 2k_{ce}k_{eb } -k_{cb}k -k_{cb , n}\right ) + b^{n}\left ( k_{eb;e } -k_{;b}\right)\right]\cr & & -\left [ \left ( r^{(ind)}_{ac } + 2k_{ae}k_{ec } -k_{ac}k -k_{ac , n}\right)b^{c } + \left ( k_{ea;e } -k_{;a}\right)b^{n}\right]b_{b } , \cr l_{an}&= & { } -b_{a}\left [ b^{c}\left ( k_{ec;e } -k_{;c}\right ) + b^{n}\left ( k_{eh}k_{eh } -k_{,n}\right)\right ] \cr & & { } -\left [ \left ( r^{(ind)}_{ac } + 2k_{ae}k_{ec } -k_{ac}k -k_{ac , n}\right)b^{c } + \left ( k_{ea;e } -k_{;a}\right)b^{n}\right]b_{n},\cr l_{nn}&= & { } -2b_{n}\left [ b^{c}\left ( k_{ec;e } -k_{;c}\right ) + b^{n}\left ( k_{eh}k_{eh } -k_{,n}\right)\right],\end{aligned}\ ] ] and @xmath50 \cr & & { } + 2k_{ab}\left ( \nabla_{c}(b_{n}b_{c } ) + \nabla_{n}(b_{n}b_{n})\right ) + k\left ( \nabla_{a}(b_{b}b_{n } ) + \nabla_{b}(b_{a}b_{n})\right ) \cr & & { } -2k_{ac}\left ( \nabla_{c}(b_{b}b_{n } ) -\nabla_{n}(b_{b}b_{c})\right ) -2k_{bc}\left ( \nabla_{c}(b_{a}b_{n } ) -\nabla_{n}(b_{a}b_{c})\right ) -k\nabla_{n}(b_{a}b_{b } ) \cr & & { } -g_{ab}\left [ 2k\nabla_{c}(b_{c}b_{n } ) -2k_{cd}\nabla_{c}(b_{d}b_{n } ) + k\nabla_{n}(b_{n}b_{n } ) -k_{cd}\nabla_{n}(b_{c}b_{d})\right ] \cr & & { } + ( \nabla_{a}k)b_{b}b_{n } + ( \nabla_{b}k)b_{a}b_{n } \cr & & { } -(\nabla_{c}k_{ca})b_{n}b_{b } + ( \nabla_{n}k_{ca})b_{c}b_{b } -(\nabla_{c}k_{cb})b_{a}b_{n } + ( \nabla_{n}k_{cb})b_{a}b_{c}\cr & & { } -g_{ab}\left [ ( \nabla_{c}k)b_{c}b_{n } -(\nabla_{n}k_{cd})b_{d}b_{c } + ( \nabla_{n}k)b_{n}b_{n}\right ] \cr & & { } + 2k_{ab } ( -k_{cd}b_{c}b_{d } + kb_{n}b_{n})\cr & & { } -2k_{ac}k_{bc}b_{n}b_{n } -2k_{ac}k_{bd}b_{c}b_{d } + k ( k_{ac}b_{c}b_{b } + k_{bc}b_{a}b_{c})\cr & & { } -g_{ab}\left [ ( 2k_{ce}k_{ed } -kk_{cd})b_{c}b_{d } -(2k_{cd}k_{dc } -k^2)b_{n}b_{n } \right ] \big]\cr \cr \sigma_{an}&= & { 1\over 2}\big [ \nabla_{a}\nabla_{c}(b_{n}b_{c } ) + \nabla_{a}\nabla_{n}(b_{n}b_{n } ) + \nabla_{n}\nabla_{c}(b_{a}b_{c } ) -\nabla_{c}\nabla_{c } ( b_{a}b_{n } ) \cr & & { } -2k_{ac}\left [ \nabla_{d}(b_{d}b_{c } ) + \nabla_{c}(b_{n}b_{n})\right ] -k_{cd}\nabla_{a}(b_{c}b_{d } ) + k\nabla_{a}(b_{n}b_{n } ) + k_{ac}\nabla_{n}(b_{c}b_{n } ) \cr & & { } -(\nabla_{a}k_{cd})b_{d}b_{c } + ( \nabla_{a}k)b_{n}b_{n } + ( \nabla_{n}k)b_{a}b_{n } + ( \nabla_{c}k_{cd})b_{a}b_{d } \cr & & { } -(\nabla_{c}k_{ca})b_{n}b_{n } + ( \nabla_{n}k_{ac})b_{c}b_{n } \cr & & { } -k_{ac}kb_{c}b_{n } -k_{cd}k_{dc}b_{a}b_{n } \big],\cr \cr \sigma_{nn}&= & { 1\over 2}\big [ \nabla_{n}\nabla_{c}(b_{n}b_{c } ) -\nabla_{c}\nabla_{d}(b_{c}b_{d } ) -\nabla_{c}\nabla_{n}(b_{c}b_{n } ) -\nabla_{c}\nabla_{c } ( b_{n}b_{n } ) \cr & & { } -k_{cd}\nabla_{n}(b_{d}b_{c } ) -2k\nabla_{c}(b_{n}b_{c } ) + 2k_{cd}\nabla_{c}(b_{n}b_{d } ) \cr & & { } -(\nabla_{n}k_{cd})b_{d}b_{c } + ( \nabla_{n}k)b_{n}b_{n } + 2(\nabla_{c}k_{cd})b_{d}b_{n } -(\nabla_{c}k)b_{c}b_{n } \cr & & { } + kk_{cd}b_{d}b_{c } -k^2b_{n}b_{n } \big].\end{aligned}\ ] ] the equation of motion for the vector field @xmath27 decomposes similarly as follows , respectively parallel @xmath51 = 0 , \label{eqn : b_a}\end{aligned}\ ] ] and orthogonal to the brane @xmath52 = 0 .
\label{eqn : b_n}\end{aligned}\ ] ] next we proceed to derive the induced equations of motion for both the metric and the vector field in terms of quantities measured on the brane .
the induced equations on the brane are the @xmath53 projected components after the singular terms across the brane are subtracted out by the substitution of the matching conditions . considering the brane as a @xmath54-symmetric shell of thickness @xmath55 in the limit @xmath56 derivatives of quantities discontinuous across the brane generate singular distributions on the brane .
integration of these terms in the coordinate normal to the brane relates the induced geometry with the localization of the induced stress - energy in the form of matching conditions . from the @xmath57 symmetry it follows that @xmath58 but that @xmath59 and consequently that @xmath60 and @xmath61 moreover , @xmath62 implies that @xmath63 first we consider the einstein equations and then the equations of motion for @xmath27 which , due to the coupling of @xmath27 to gravity , will also be used as complementary conditions for the dynamics of the metric on the brane . combining the gauss - codacci relations with the projections of the stress - energy tensor and the interaction source terms
, we integrate the @xmath53 component of the einstein equation in the coordinate normal to the brane to obtain the matching conditions for the extrinsic curvature across the brane , i.e. the israel matching conditions .
we find that @xmath64=-g_{ab}\sigma \cr & + & { \xi \over 2}\big [ \nabla_{a}(b_{b}b_{n } ) + \nabla_{b}(b_{a}b_{n } ) -\nabla_{n}(b_{a}b_{b } ) \cr & & { } + 4 ( b_{a}b_{c}k_{cb } + k_{ac}b_{c}b_{b } ) -2k_{ab}b_{n}b_{n}\cr & & { } + g_{ab}\left ( -2\nabla_{c}(b_{c}b_{n } ) -\nabla_{n}(b_{n}b_{n } ) + k_{cd}b_{c}b_{d } -kb_{n}b_{n}\right ) \big].\label{eqn : imc}\end{aligned}\ ] ] these provide boundary conditions for ten of the fifteen degrees of freedom .
five additional boundary conditions are provided by the junction conditions for the @xmath65 and @xmath66 components of the projection of the einstein equations . from inspection of the @xmath65 component
, we note that @xmath67 which vanishes due to conservation of the induced stress - energy tensor @xmath68 on the brane . from integration of the @xmath66 component in the normal direction to the brane
, we find the following junction condition @xmath69 which we substitute back in , obtaining @xmath70 \cr & + & { \xi\over 2}\bigl [ -\nabla_{c}\nabla_{d}(b_{c}b_{d } ) -\nabla_{c}\nabla_{c}(b_{n}b_{n } ) + { 12\over { 2\xi -1}}b_{n}\left ( \nabla_{c}\nabla_{c}b_{n } -2v^{\prime}b_{n}\right)\cr & & { } + 2\left ( k_{cd } -g_{cd}k\right)\nabla_{c}(b_{d}b_{n } ) + 2k_{cd}b_{d}(\nabla_{c}b_{n})+k_{cd;c}b_{d}b_{n}\cr & & { } + \left ( 7k_{cd}k_{cd } -k^2\right)b_{n}b_{n } + \left ( ( 6\xi + 1)k_{ce}k_{ed } + kk_{cd}\right)b_{c}b_{d}\bigr ] .
\label{eqn : g_nn : induced}\end{aligned}\ ] ] however , the israel matching conditions also contain terms which depend on the prescription for the continuation of @xmath27 out of the brane and into the bulk , namely @xmath71 and @xmath72 .
the five additional boundary conditions required are those for the vector field * b. * in ref . @xcite the boundary conditions for bulk fields were derived subject to the condition that modes emitted by the brane into the bulk do not violate the gauge defined in the bulk . here
, however , we integrate the ( a ) and ( n ) components of the equation of motion for @xmath44 eq .
( [ eqn : b_a ] ) and eq .
( [ eqn : b_n ] ) respectively , to find the corresponding junction condition for @xmath73 and for @xmath74 across the brane . from eq .
( [ eqn : b_a ] ) we have that @xmath75 = 0.\end{aligned}\ ] ] if @xmath76 is sufficiently small , the difference between @xmath77 and @xmath78 is negligibly small .
it follows that , in the limit where @xmath56 we can assume that @xmath79 .
it then follows that @xmath80 similarly , from eq .
( [ eqn : b_n ] ) we find that @xmath81=0,\end{aligned}\ ] ] which becomes @xmath82 the junction conditions eq .
( [ eqn : b_a : junction ] ) and eq .
( [ eqn : b_n : junction ] ) offer the required @xmath11 boundary conditions respectively for @xmath73 and @xmath74 on the brane . substituting the junction condition for @xmath73 back in eq .
( [ eqn : b_a ] ) and using the result from @xmath83 we find for the induced equation of motion for @xmath73 on the brane that @xmath84 similarly , substituting the junction condition for @xmath74 back in eq .
( [ eqn : b_n ] ) we obtain @xmath85 = 0 .
\label{eqn : b_n : induced}\end{aligned}\ ] ] thus , eq . ( [ eqn : b_a : junction ] ) provides the value at the boundary for @xmath71 whereas eq .
( [ eqn : b_n : induced ] ) provides that for @xmath72 . using the results derived above in the israel matching conditions we find that @xmath64=-g_{ab}~\sigma \cr & + & { \xi\over 2}\big [ ( \nabla_{a}b_{b})b_{n } + ( \nabla_{b}b_{a})b_{n } \big]\cr & + & \xi b_{a}b_{c}k_{cb } + \xi k_{ac}b_{c}b_{b } -k_{ab}b_{n}b_{n}\cr & + & \xi g_{ab}\biggl [ -\nabla_{c}(b_{c}b_{n } ) + { 1\over 2}k_{cd}b_{c}b_{d } -{1\over 2}kb_{n}b_{n}\cr & & { } + { 1\over { 2\xi -1}}{b_{n}\over k } \bigl ( \nabla_{c}\nabla_{c } b_{n } -2v^{\prime}b_{n } + ( 2\xi -1)k_{cd}k_{cd}b_{n}\bigr ) \biggr].\end{aligned}\ ] ] which provide a second order equation for the trace of the extrinsic curvature , @xmath86 finally , using eq .
( [ eqn : g_nn : induced ] ) in the @xmath53 einstein equation , we find for the einstein equation induced on the brane that @xmath87 + 2g_{ab}\lambda_{(5)}\cr & + & { } { 1\over 2}\left [ -b_{ac}b_{bc } -4v^{\prime}b_{a}b_{b } + g_{ab}\left ( { 1\over 4}b_{cd}b_{cd } + 4v^{\prime}b_{n}b_{n } + 2v\right)\right]\cr & = & { \xi\over 2}\biggl [ \nabla_{a}\nabla_{c}(b_{b}b_{c } ) + \nabla_{b}\nabla_{c}(b_{a}b_{c } ) -\nabla_{c}\nabla_{c}(b_{a}b_{b})\cr & & { } -2k_{ac}\nabla_{c } ( b_{b}b_{n } ) -2k_{ac}(b_{b}\nabla_{c}b_{n } + b_{c}\nabla_{b}b_{n})\cr & & { } -2k_{bc}\nabla_{c } ( b_{a}b_{n } ) -2k_{bc}(b_{a}\nabla_{c}b_{n } + b_{c}\nabla_{a}b_{n})\cr & & { } -{8\over { 2\xi -1}}{b_{n}\over k } k_{ab } ( \nabla_{c}\nabla_{c}b_{n } -2v^{\prime}b_{n } + ( 2\xi -1)k_{cd}k_{cd}b_{n } ) + kb_{n}(\nabla_{a}b_{b } + \nabla_{b}b_{a } ) \cr & & { } -2b_{a}b_{c}\left ( r^{(ind)}_{cb } + 2k_{cd}k_{bd } + ( \xi -1)k_{cb}k\right)\cr & & { } -2b_{b}b_{c}\left ( r^{(ind)}_{ca } + 2k_{ad}k_{cd } + ( \xi -1)k_{ac}k\right)\cr & & { } + ( k_{ac;b } + k_{bc;a }
-2k_{ab;c})b_{n}b_{n } -4k_{ab}kb_{n}b_{n}\cr & & { } + ( k_{ac}b_{b } + k_{bc}b_{a } ) ( -5\xi k_{dc}b_{d } + kb_{c } ) -(4\xi + 2)k_{ac}k_{bd}b_{c}b_{d } \biggr]\cr & + & { \xi\over 2}g_{ab}\biggl [ -2\nabla_{c}\nabla_{d}(b_{c}b_{d } ) -\nabla_{c}\nabla_{c}(b_{n}b_{n } ) + { 12\over { 2\xi -1}}b_{n}\left ( \nabla_{c}\nabla_{c}b_{n } -2v^{\prime}b_{n}\right ) \cr & & { } + 4(k_{cd } -g_{cd}k)\nabla_{d}(b_{c}b_{n } ) + 4k_{cd}b_{d}(\nabla_{c}b_{n } )
\cr & & { } + b_{c}b_{d}r^{(ind)}_{cd } + ( 9k_{cd}k_{cd } -2k^2)b_{n}b_{n } + ( 14\xi k_{ce}k_{de } + kk_{cd})b_{c}b_{d } \biggr].\end{aligned}\ ] ] the results obtained above show both the coupling of the bulk to the brane and the coupling of the vector field @xmath27 to the geometry of the spacetime .
the first is manifested in the dependence on normal components in the induced equations ; the latter is manifested in the presence of terms of the form @xmath88 terms of the form @xmath89 illustrate both couplings , where @xmath74 relates with @xmath90 and @xmath73 by eq .
( [ eqn : b_n : junction ] ) . the directional dependence on the @xmath91 direction is encapsulated in the extrinsic curvature . in the penultimate line we can substitute the israel matching condition found above .
however , the derivatives of the extrinsic curvature along directions parallel to the brane which appear in the ninth line are not reducible to quantities intrinsic to the brane .
in this section we particularize the formalism developed above for the case when the bulk vector field @xmath27 acquires a non - vanishing vacuum expectation value by spontaneous symmetry breaking .
the vacuum value generates the breaking of the lorentz symmetry by selecting the direction orthogonal to the plane of the brane .
this selection can be achieved by assigning to the parallel or the orthogonal components of the bulk vector field a non - vanishing vacuum expectation value .
the minimum of the potential occupied by the vacuum value is assumed to be also a zero of the potential . here
we consider the case when the parallel component of the vector field with respect to the brane acquires a non - vanishing expectation value , @xmath94 , whereas the expectation value of the normal component is chosen to vanish on the brane , @xmath93 .
the junction conditions from the equations for @xmath95 @xmath96 @xmath97 and @xmath98 reduce respectively to @xmath99 & = & -g_{ab}~\sigma + \xi \left <
b_{a}\right>\left < b_{c}\right > k_{cb } + \xi \left < b_{b}\right>\left < b_{c}\right > k_{ac},\end{aligned}\ ] ] and the induced equations of motion become @xmath100 for @xmath95 @xmath101 + { \xi\over 2}\left [ -\nabla_{c}\nabla_{d}(\left
< b_{c}\right>\left < b_{d}\right > ) + ( 6\xi + 1)k_{ce}k_{ed}\left < b_{c}\right>\left < b_{d}\right > \right]\end{aligned}\ ] ] for @xmath97 and finally @xmath102 + g_{ab}\lambda_{(5 ) } -{1\over 2}\left < b_{ac}\right>\left < b_{bc}\right > \cr & = & { 1\over 2}\left ( { 5\over 4 } + { 1\over \xi}\right)\left [ \left < b_{a}\right>\nabla_{c}\left ( \nabla_{c}\left < b_{b}\right > -\nabla_{b}\left < b_{c}\right>\right ) + \left < b_{b}\right>\nabla_{c}\left ( \nabla_{c}\left < b_{a}\right >
-\nabla_{a}\left < b_{c}\right>\right ) \right]\cr & + & { \xi\over 2}\biggl [ \nabla_{a}\nabla_{c}(\left < b_{b}\right>\left
< b_{c}\right > ) + \nabla_{b}\nabla_{c}(\left
< b_{a}\right>\left < b_{c}\right > ) -\nabla_{c}\nabla_{c}(\left < b_{a}\right>\left < b_{b}\right>)\cr & & { } + \left ( { 5\over 2 } -2 + { 2\over \xi}\right)\left ( \left < b_{a}\right>\left < b_{c}\right > r^{(ind)}_{cb } + \left < b_{b}\right>\left < b_{c}\right > r^{(ind)}_{ac}\right ) -(4\xi + 2)k_{ac}k_{bd}\left < b_{c}\right>\left < b_{d}\right > \biggr]\cr & + & { 1\over 2}(1 -2(\xi -1))g_{ab}k\sigma + { \xi\over 2}g_{ab}\biggl [ \left < b_{c}\right>\left < b_{d}\right > r^{(ind)}_{cd } + 2(\xi -1)k_{ce}k_{ed}\left
< b_{c}\right>\left < b_{d}\right>\biggr ] \label{eqn : einsteinvev1}\end{aligned}\ ] ] from @xmath103 where we used also the previous results , namely the @xmath97 equation , the israel matching condition and the @xmath73 equation . imposing the condition for the covariant conservation of the vacuum expectation value of the field @xmath44
@xmath104 @xcite , it follows that @xmath105 which enables us to further simplify eq .
( [ eqn : einsteinvev1 ] ) : @xmath102 + g_{ab}\lambda_{(5 ) } \cr & = & { \xi\over 2}\biggl [ \left ( { 5\over 2 } -2 + { 2\over \xi}\right)\left ( \left
< b_{a}\right>\left < b_{c}\right > r^{(ind)}_{cb } + \left < b_{b}\right>\left < b_{c}\right > r^{(ind)}_{ac}\right ) -(4\xi + 2)k_{ac}k_{bd}\left
< b_{c}\right>\left < b_{d}\right > \biggr]\cr & + & { 1\over 2}(1 -2(\xi -1))g_{ab}k\sigma + { \xi\over 2}g_{ab}\biggl [ \left
< b_{c}\right>\left < b_{d}\right > r^{(ind)}_{cd } + 2(\xi
-1)k_{ce}k_{ed}\left < b_{c}\right>\left < b_{d}\right>\biggr ] .
\label{eqn : einsteinvev2}\end{aligned}\ ] ] hence , in order to obtain a vanishing cosmological constant and ensure that lorentz invariance holds on the brane , we must impose respectively that @xmath106 and that @xmath107 \cr & = & { \xi\over 2}\biggl [ \left ( { 5\over 2 } -2 + { 2\over \xi}\right)\left ( \left < b_{a}\right>\left < b_{c}\right >
r^{(ind)}_{cb } + \left < b_{b}\right>\left < b_{c}\right > r^{(ind)}_{ac}\right ) -(4\xi + 2)k_{ac}k_{bd}\left < b_{c}\right>\left < b_{d}\right > \biggr]\cr & + & { \xi\over 2}g_{ab}\biggl [ \left < b_{c}\right>\left
< b_{d}\right > r^{(ind)}_{cd } + 2(\xi -1)k_{ce}k_{ed}\left < b_{c}\right>\left < b_{d}\right>\biggr ] \label{eqn : lorentzbrane},\end{aligned}\ ] ] which for @xmath108 reduce to the results presented in ref . @xcite .
we observe that there is close relation between the vanishing of the cosmological constant and the maintenance of the lorentz invariance on the brane .
these conditions are enforced so that the higher dimensional signatures encapsulated in the induced geometry of the brane cancel the lorentz symmetry breaking inevitably induced on the brane , thus reproducing the observed geometry .
the first condition , eq .
( [ eqn : vanishingcc ] ) , can be modified to account for any non - vanishing value for the cosmological constant , as appears to be suggested by the recent wilkinson microwave anisotropy probe data , by defining the observed cosmological constant @xmath109 such that @xmath110 a much more elaborate fine - tuning , however , is required for the lorentz symmetry to be observed on the brane , as described by the condition in eq .
( [ eqn : lorentzbrane ] ) . to our knowledge
this is a new feature in braneworld models , as in most models lorentz invariance is a symmetry shared by both the bulk and the brane .
notice that a connection between the cosmological constant and lorentz symmetry has been conjectured , on different grounds , long ago @xcite .
choosing instead @xmath111 and @xmath112 we would also expect to violate lorentz symmetry on the brane .
however , for @xmath113 the boundary conditions imply that we must have @xmath93 and thus rendering the vacuum lorentz symmetric . if , on the other hand , we allow @xmath114 then the israel matching conditions yield that @xmath115 , rendering the brane inexistent , with @xmath116 being but the five dimensional gravitational constant @xmath117 if we consider the general case , with both @xmath73 and @xmath74 acquiring different , constant non - vanishing vacuum expectation values along the directions parallel to the brane , i.e. @xmath92 and @xmath112 and such that @xmath118 , we find that for @xmath113 we must have that @xmath119 thus obtaining the same boundary conditions as those found in subsection [ 1 ] .
should we allow @xmath114 then we find , as in subsection [ 2 ] , that @xmath115 and that @xmath120 is defined in terms of both @xmath121 and @xmath116 according to @xmath122/k_{ab } + \left < b_{n}\right>\left < b_{n}\right>$ ] .
in this contribution we analysed the spontaneous symmetry breaking of lorentz invariance in the bulk and its effect on the brane . for this purpose
, we considered a bulk vector field subject to a potential which endows the field with a non - vanishing vacuum expectation value , thus allowing for the spontaneous breaking of the lorentz symmetry in the bulk . this bulk vector field
is directly coupled to the ricci tensor so that , after the breaking of lorentz invariance , the loss of this symmetry is transmitted to the gravitational sector . we assign a non - vanishing vacuum expectation value first separately to the parallel and orthogonal components of the vector field , finding then that the case where both components attained non - vanishing vacuum expectation values reduced to the previous two cases . the complex interplay between matching conditions and the lorentz symmetry
breaking terms was examined .
we found that lorentz invariance on the brane can be made exact via the dynamics of the graviton , vector field and the extrinsic curvature of the surface of the brane .
as a consequence of the exact reproduction of lorentz symmetry on the brane , we found a condition for the matching of the observed cosmological constant in four dimensions .
this tuning does not follow from a dynamical mechanism but is instead imposed by phenomenological reasons only . from this point of view , both the value of the cosmological constant and the induced brane lorentz symmetry seem to be a consequence of a complex fine tuning .
we shall examine further implications of this mechanism in a forthcoming publication where we will also discuss the inclusion of a bulk scalar field @xcite .
* acknowledgments * 0.2 cm c. c. thanks the portuguese agency , fundao para a cincia e a tecnologia ( fct ) , for financial support under the fellowship /bpd/18236/2004 .
c. c. thanks martin bucher , georgios kofinas and rodrigo olea for useful discussions .
the work of o.b .
is partially supported by the fct project poci / fis/56093/2004
. 0.3 cm
0.3 cm cpt and lorentz symmetry iii , alan kosteleck , ed .
( world scientific , singapore , 2005 ) ; o. bertolami , gen .
gravitation * 34 * ( 2002 ) 707 ; o. bertolami , lect .
notes phys .
* 633 * ( 2003 ) 96 , hep - ph/0301191 ; d. mattingly , liv .
* 8 * ( 2005 ) 5 , gr - qc/0502097 ; t. jacobson , s. liberati and d. mattingly , ann
* 321 * ( 2006 ) 150 , astro - ph/0505267 ; r. lehnert , `` cpt- and lorentz - symmetry breaking : a review '' , hep - ph/0611177 .
carroll , j.a .
harvey , v.a .
kosteleck , c.d . lane and t.
okamoto , phys .
* 87 * ( 2001 ) 141601 ; o. bertolami and l. guisado , phys . rev .
* d67 * ( 2003 ) 025001 ; jhep * 0312 * ( 2003 ) 013 ; o. bertolami , mod .
phys . lett . *
a20 * ( 2005 ) 1359 . | we propose the mechanism of spontaneous symmetry breaking of a bulk vector field as a way to generate the selection of bulk dimensions invisible to the standard model confined to the brane . by assigning a non - vanishing vacuum value to the vector field ,
a direction is singled out in the bulk vacuum , thus breaking the bulk lorentz symmetry .
we present the condition for induced lorentz symmetry on the brane , as phenomenologically required . |
the scalar channels in the region up to 1 gev became a stumbling block of qcd .
the point is that both perturbation theory and sum rules do not work in these channels because there are not solitary resonances in this region . at the same time
the question on the nature of the light scalar mesons is major for understanding the mechanism of the chiral symmetry realization , arising from the confinement , and hence for understanding the confinement itself .
the qcd lagrangian is given by + @xmath8=@xmath9 + @xmath10 is a diagonal matrix of quark masses , @xmath11=@xmath12=@xmath13 @xmath10 mixes left and right spaces .
but in chiral limit , @xmath14@xmath150 , these spaces separate realize @xmath16 flavour symmetry , which , however , is broken by the gluonic anomaly up to @xmath17 as experiment suggests , confinement forms colourless observable hadronic fields and spontaneous breaking of chiral symmetry with massless pseudoscalar fields .
there are two possible scenarios for qcd at low energy .
1 . non - linear @xmath7 model .
2 . linear @xmath7 model ( lsm ) .
the experimental nonet of the light scalar mesons , @xmath18 ( or @xmath19 ) , @xmath20(700 - 900 ) , @xmath21 and @xmath1 mesons , suggests the @xmath16 lsm .
hunting the light @xmath7 and @xmath20 mesons had begun in the sixties already and a preliminary information on the light scalar mesons in pdg reviews had appeared at that time .
but long - standing unsuccessful attempts to prove their existence in a conclusive way entailed general disappointment and an information on these states disappeared from pdg reviews .
one of principal reasons against the @xmath7 and @xmath20 mesons was the fact that both @xmath22 and @xmath23 scattering phase shifts do not pass over @xmath24 at putative resonance masses .
situation changes when we showed that in lsm there is a negative background phase which hides the @xmath7 meson in @xmath2 .
it has been made clear that shielding wide lightest scalar mesons in chiral dynamics is very natural .
this idea was picked up and triggered new wave of theoretical and experimental searches for the @xmath7 and @xmath20 mesons .
our approximation is as follows ( see fig.1 ) : @xmath26 = @xmath27 + @xmath28,\ t^0_0=t_0^{0(tree)}$ ] @xmath29= [ e^{2i\left(\delta_{bg}+\delta_{res}\right)}$]@xmath30/(2i\rho_{\pi\pi } ) $ ] = @xmath31@xmath32 = @xmath33 @xmath34 = @xmath35/[m^2_{res}-s+ \mbox{re}\pi_{res}(m^2_{res})-\pi_{res}(s ) ] = ( e^{2i\delta_{res}}-1)/(2i\rho_{\pi\pi})\,,\ t_{bg}=(e^{2i\delta_{bg}}-1)/(2i\rho_{\pi\pi } ) = \lambda(s)/[1-i\rho_{\pi\pi}\lambda(s)]$ ] , @xmath36 $ ] , @xmath37=92.4mev , + @xmath38= @xmath39 , @xmath40= @xmath41 , @xmath42=@xmath43 , @xmath44=@xmath45 , @xmath46=@xmath47 , @xmath48=@xmath49=@xmath50 .
+ figure 1 .
the graphical representation of the @xmath51 wave @xmath52 @xmath22 scattering amplitude @xmath53 .
@xmath54 $ ] , + @xmath55=@xmath56$]=@xmath57 .
@xmath58=@xmath59 gev , @xmath60=@xmath61 gev , @xmath62=@xmath63 gev , @xmath64=@xmath65 @xmath66 @xmath67=0.33 , @xmath68=@xmath69 , @xmath70=@xmath71 , @xmath72=@xmath73 , @xmath74=@xmath75
+ figure 2 . the @xmath7 model .
our approximation . @xmath76 .
( @xmath77)=latexmath:[$32\pi(|t^0_0|^2 , the chiral shielding of the @xmath0 meson is illustrated in fig .
2(a ) with the help of the @xmath22 phase shifts @xmath79 , @xmath80 , and @xmath81 , and in fig .
2(b ) with the help of the corresponding cross sections .
@xmath82 @xmath83=@xmath84 gev@xmath85 , @xmath86 + considering the residue of the @xmath7 pole in @xmath53 as the square of its coupling constant to the @xmath22 channel is not a clear guide to understand the @xmath7 meson nature for its great obscure imaginary part .
@xmath87=@xmath88@xmath89+@xmath90@xmath91 $ ] . + the @xmath7 meson self - energy @xmath92 is caused by the intermediate @xmath22 states ,
that is , by the four - quark intermediate states if we keep in mind that the @xmath25 lsm could be the low energy realization of the two - flavour qcd .
this contribution shifts the breit - wigner ( bw ) mass greatly @xmath93=0.50 gev .
so , half the bw mass is determined by the four - quark contribution at least .
the imaginary part dominates the propagator modulus in the region 300mev@xmath94600 mev .
so , the @xmath7 field is described by its four - quark component at least in this energy region .
@xmath96 + @xmath97 + @xmath98 , + @xmath99 + @xmath100 , + @xmath101 @xmath102 our results are shown in fig .
3 . @xmath103@xmath15@xmath104@xmath15@xmath105=@xmath106@xmath15@xmath104@xmath107 + @xmath108 , where @xmath109@xmath15@xmath104@xmath15@xmath110=@xmath111 @xmath112 ; see fig .
so , the the @xmath113 decay is described by the triangle @xmath104 loop diagram @xmath114 .
consequently , it is due to the four - quark transition because we imply a low energy realization of the two - flavour qcd by means of the the @xmath25 lsm . as fig .
4 suggests , the real intermediate @xmath104 state dominates in @xmath115 in the @xmath7 region @xmath1160.6gev .
thus the picture in the physical region is clear and informative .
but , what about the pole in the complex @xmath89 plane ? does the pole residue reveal the @xmath7 indeed ?
@xmath117 , + @xmath118=@xmath119 , @xmath120 , @xmath121=@xmath122 it is hard to believe that anybody could learn the complex but physically clear dynamics of the @xmath113 decay from the residues of the @xmath7 pole .
30 m. gell - mann and m. levy , nuovo cimento * 16 * , 705 ( 1960 ) .
n.n . achasov and g.n .
shestakov , phys .
d * 49 * , 5779(1994 ) .
n.n . achasov and g.n .
shestakov , phys .
lett . * 99 * , 072001 ( 2007 ) .
i. caprini , g. colangelo , and h. leutwyler , phys .
lett . * 96 * , 132001 ( 2006 ) .
jaffe , phys .
d * 15 * , 267 , 281 ( 1977 ) .
achasov et al . , phys . lett . *
108b * , 134 ( 1982 ) ; z. phys .
c * 16 * , 55 ( 1982 ) .
n.n . achasov and a.v .
kiselev , phys .
d * 73 * , 054029 ( 2006 ) ; yad.fiz . *
70 * , 2005 ( 2007 ) .
g. colangelo , j. gasser , and h. leutwyler , nucl .
b * 603 * , 125 ( 2001 ) .
achasov , yad.fiz . * 65 * , 573 ( 2002 ) .
achasov , nucl .
phys . a * 728 * , 425 ( 2003 ) .
n.n . achasov and v.n .
ivanchenko , nucl .
b * 315 * ( 1989 ) 465 .
achasov and v.v .
gubin , phys .
d * 56 * , 4084 ( 1997 ) .
n.n . achasov and v.v .
gubin , phys .
d * 63 * , 094007 ( 2001 ) .
n.n . achasov and a.v .
kiselev , phys .
d * 68 * , 014006 ( 2003 ) .
achasov et al . , z. phys .
c * 27 * , 99 ( 1985 ) .
achasov , and g.n .
shestakov , z. phys .
c * 41 * , 309 ( 1988 ) .
t. mori et al . , j. phys .
. jap . * 76 * , 074102 ( 2007 ) ; k. abe et al .
, arxiv : 0711.1926 [ hep - ex ] .
n.n . achasov and g.n .
shestakov , phys .
d * 72 * , 013006 ( 2005 ) .
n.n . achasov and g.n .
shestakov , phys .
d * 77 * , 074020 ( 2008 ) . | outline : 1 .
introduction , 2 .
confinement , chiral dynamics and light scalar mesons , 3 .
chiral shielding of the @xmath0 , chiral constraints ( the cgl band ) , the @xmath0 and the @xmath1 in @xmath2 , @xmath3 , @xmath4 , 4 .
the @xmath5 meson radiative decays on light scalar resonances , 5 .
light scalars in @xmath6 collisions . evidence for four - quark components of light scalars is given .
the priority of quantum field theory in revealing the light scalar mystery is emphasized . |
_ derivators _ were introduced by grothendieck @xcite as an intermediate step between a ( suitable ) category @xmath1 with weak equivalences and its homotopy category @xmath2 , obtained by formally inverting the weak equivalences .
in addition to the homotopy category @xmath2 , the derivator @xmath3 contains all the homotopy categories of diagrams @xmath4 indexed by the opposite of a finite direct category @xmath5 together with the data of the various derived functors associated to the change of indexing category . formally , a derivator @xmath6 is a @xmath7-functor @xmath8 satisfying a set of axioms inspired by the example @xmath9 . here
@xmath10 is the @xmath7-category of small categories and @xmath11 is the @xmath7-full sub-@xmath7-category of finite direct categories , i.e. those categories whose nerve has a finite number of non - degenerate simplices .
maltsiniotis , after franke @xcite , added an extra axiom to define _ triangulated derivators _
an important example that was worked out by keller @xcite is the derivator of the category @xmath12 of bounded complexes in an exact category @xmath13 , where the weak equivalences are the quasi - isomorphisms in the abelian envelope of @xmath13 .
this derivator will be denoted @xmath14 instead of @xmath15 .
maltsiniotis defined the @xmath0-theory @xmath16 of a triangulated derivator @xmath6 , and stated three conjectures : _ comparison _ , _ localization _ and _ additivity_. the proof of the additivity conjecture by cisinski and neeman @xcite showed that maltsiniotis s @xmath0-theory produces a @xmath0-theoretically sound construction .
the combination of the other two conjectures claims that derivator @xmath0-theory , as a functor from a suitably defined category of small triangulated derivators to spaces , recovers quillen s @xmath0-theory of exact categories and satisfies localization .
schlichting @xcite showed that there can not exist a functor from small triangulated categories to spaces that has these two properties .
the comparison conjecture says that a certain natural map from quillen s to maltsiniotis s @xmath0-theory space , @xmath17 is a weak homotopy equivalence for every exact category @xmath13 , i.e. , the induced homomorphisms @xmath18 , @xmath19 are isomorphisms for every @xmath20 .
maltsiniotis proved this for @xmath21 @xcite .
the conjecture is also true for @xmath22 @xcite .
moreover , the following more general version holds .
if @xmath1 is a waldhausen category with cylinders and a saturated class of weak equivalences , then @xmath3 is a _
right pointed derivator _ @xcite .
garkusha @xcite considered the derivator @xmath0-theory of this wider class of derivators .
there is a natural comparison map @xmath23 from waldhausen s to garkusha s @xmath0-theory , and so natural comparison homomorphisms @xmath24 , @xmath25 , @xmath26 such that maltsiniotis s comparison map is obtained by precomposing with the gillet
waldhausen equivalence @xcite , @xmath27 the homomorphisms @xmath28 are isomorphisms for @xmath29 @xcite .
ton and vezzosi showed in @xcite that @xmath30 can not be an equivalence for all @xmath1 in all higher dimensions .
more precisely , taking @xmath31 to be the category of finite @xmath32-complexes @xmath33 relative to a space @xmath34 equipped with a retraction @xmath35 @xcite , they noticed that the natural comparison map @xmath36 can not be a weak homotopy equivalence in general , since the action of the topological monoid of self - equivalences of @xmath34 on @xmath37 does not factor through an aspherical space .
despite the fact that the derivator encodes a lot of the homotopy theory of @xmath1 , it still forgets the higher homotopical information of the self - equivalences of @xmath1 , and in some sense that is exactly everything that it forgets @xcite .
thus the failure of the generalized comparison conjecture in this case is due to the homotopical complexity of the algebraic @xmath0-theory of spaces .
this could suggest that it may however be true if we restrict to categories of algebraic origin such as , for instance , the category @xmath38 of perfect right modules over a differential graded algebra @xmath39 .
note that @xmath40 is a triangulated derivator while @xmath41 is not .
our first theorem shows that the generalized comparison conjecture is false even in this algebraic context .
moreover , it shows that @xmath42 need not be isomorphic to @xmath43 even abstractly as abelian groups .
this question was so far open .
let @xmath44 $ ] be the graded algebra of laurent polynomials with a degree @xmath45 indeterminate @xmath46 , @xmath47 , over the field with @xmath48 elements , @xmath48 a prime number , endowed with the trivial differential @xmath49 .
let @xmath50 be the graded algebra over the integers @xmath51 generated by @xmath52 and @xmath53 in degree @xmath54 with the following additional relations , @xmath55 endowed with the degree @xmath56 differential @xmath57 defined by @xmath58 notice that @xmath59 .
[ counterexample ] the differential graded algebras @xmath39 and @xmath50 have equivalent derivator @xmath0-theory , @xmath60 but different waldhausen @xmath0-theory @xmath61 in dimension @xmath62 if @xmath63 , in dimension @xmath64 if @xmath65 , and in dimension @xmath66 if @xmath67 .
this is something close , although not actually a counterexample to maltsiniotis s original comparison conjecture .
nevertheless , we succeed in showing that the comparison conjecture can not be true if the localization conjecture holds .
the localization conjecture says that any pair of composable pseudo - natural transformations between triangulated derivators @xmath68 that yields a short exact sequence of triangulated categories @xmath69 at any finite direct category @xmath5 , induces a homotopy fibration in @xmath0-theory spaces , @xmath70 [ disagreement ] derivator @xmath0-theory does not satisfy both the comparison and localization conjectures .
let @xmath71 be a stable pointed model category @xcite .
an object @xmath34 in the triangulated category @xmath72 is _ compact _ if the functor @xmath73 preserves coproducts . in most cases , the waldhausen category @xmath1 will essentially be the full subcategory of compact cofibrant objects in some @xmath71 . for any direct category @xmath5
we equip the diagram category @xmath74 with the model structure where weak equivalences and fibrations are defined levelwise ( * ? ? ?
* theorem 5.1.3 ) .
the poset @xmath75 is the free category generated by the following dynkin quiver of type @xmath76 , @xmath77 a compact cofibrant object in @xmath78 is a sequence of cofibrations between compact cofibrant objects in @xmath71 , @xmath79 objects in the category @xmath80 are such sequences together with choices of cofibers @xmath81 , @xmath82 , such that @xmath83 is the distinguished zero object of @xmath1 .
the morphisms are defined as commutative diagrams in the obvious way .
it is clear that @xmath80 is equivalent to the full subcategory of @xmath78 that consists of the compact cofibrant objects , the equivalence being the functor forgetting the choice of cofibers .
the categories @xmath84 and @xmath85 are also equivalent . for @xmath21 ,
@xmath86 is the category with one object @xmath87 and one morphism ( the identity ) .
these categories assemble to a simplicial waldhausen category @xmath88 where @xmath89 and @xmath90 are induced by the usual functors @xmath91 and @xmath92 , @xmath93 , the degeneracy @xmath94 adds @xmath87 at the beginning of the sequence , @xmath95 and the face @xmath96 uses the choice of cofibers @xmath97 waldhausen s @xmath0-theory space of @xmath1 is defined to be the loop space of the geometric realization of the simplicial subcategory of weak equivalences @xcite , @xmath98 in most cases of interest , the waldhausen category @xmath1 produces a right pointed derivator @xmath3 @xcite . the @xmath0-theory space of @xmath3 is naturally weakly equivalent to @xmath99 where @xmath100 denotes the subcategory of isomorphisms , compare @xcite .
the natural comparison map is induced by the canonical simplicial functor @xmath101 , which takes weak equivalences to isomorphisms , and therefore gives rise to a map @xmath102
for any commutative frobenius ring @xmath103 we denote @xmath104 the _ stable category of @xmath103-modules _ ,
i.e. , the quotient of the category @xmath105 of @xmath103-modules by the ideal of morphisms which factor through a projective object .
we denote @xmath106 the stable pointed model category structure on @xmath105 where fibrations are the epimorphisms , cofibrations are the monomorphisms , and weak equivalences are those homomorphisms which project to an isomorphism in @xmath104 ( * ? ? ? * theorem 2.2.12 ) .
the homotopy category is @xmath107 and the compact objects are essentially the finitely generated @xmath103-modules .
we denote @xmath108 the waldhausen subcategory of finitely generated @xmath103-modules , whose underlying category is denoted by @xmath109 .
its homotopy category @xmath110 is the full subcategory of @xmath104 spanned by the finitely generated @xmath103-modules . in this section
we will compute @xmath111 for @xmath103 a commutative noetherian local ring with principal maximal ideal @xmath112 such that @xmath113 .
we denote @xmath114 the residue field .
the ring @xmath103 is frobenius by ( * ? ? ?
* theorems 15.27 and 16.14 ) , since the socle of @xmath103 is @xmath115 .
the upshot of the computation will be that the category @xmath111 is determined up to equivalence by the field @xmath116 .
the first important observation , for @xmath21 , is that the full inclusion of the category of finite dimensional @xmath116-vector spaces @xmath117 induced by the unique ring homomorphism @xmath118 gives rise to an equivalence of categories @xmath119 for any @xmath25 , this functor induces a section of the canonical functor @xmath120 , since the following composite is an equivalence , @xmath121 the category @xmath122 is well known since it is the category of finite - dimensional representations of the quiver @xmath123 over the field @xmath116 . the monoid of isomorphism classes of objects is freely generated by the following set of indecomposable representations , @xmath124 given a category @xmath125 and a set of objects @xmath126 in @xmath125 we denote @xmath127 the full subcategory spanned by the objects in @xmath126 .
we now describe @xmath128 .
consider the poset @xmath129 and the corresponding @xmath116-linear category @xmath130 $ ] .
we refer to @xcite for the basics on the ring and module theory of linear categories .
then there is an isomorphism of @xmath116-linear categories @xmath131^{\op}/((i-1,i-1)\leq ( i , i))_{i=1}^{n}\st{\cong}\to{\mathrm{mod}(k)}^{{\mathbf{n}}}_{e_{n}}\ ] ] sending @xmath132 to @xmath133 and @xmath134 to the morphism @xmath135 given by @xmath136 equivalently , @xmath128 is isomorphic to the @xmath116-linear category of the following quiver @xmath137 e.g. for @xmath138 @xmath139&\\ & & ( 1,3)\ar[ru]\ar[rd]&&(0,2)\ar[rd]\\ & ( 2,3)\ar[rd]\ar[ru]&&(1,2)\ar[ru]\ar[rd]&&(0,1)\ar[rd]\\ ( 3,3)\ar[ru]&&(2,2)\ar[ru]&&(1,1)\ar[ru]&&(0,0)\\ } \end{array}\ ] ] with relations : * the square @xmath140&\\ ( i , j)\ar[ru]\ar[rd]&&(i-1,j-1)\\ & ( i , j-1)\ar[ru]&}\end{array}$ ] is commutative for @xmath141 , * the composite @xmath142&\\ ( i , i)\ar[ru]&&(i-1,i-1)\\ & & } \end{array}$ ] is trivial for @xmath143 .
this category will be denoted by @xmath144 since it is ( isomorphic to ) the _ mesh category _ of the quiver @xmath145 with translation @xmath146 , @xmath147 , which is the _ auslander - reiten quiver _ of @xmath123 , see @xcite . the arrows @xmath148 and @xmath149 in @xmath137 correspond to the following morphisms , @xmath150 recall that given a @xmath116-linear category @xmath125 and a @xmath125-bimodule @xmath151 the _ semidirect product _
@xmath152 is the category with the same objects as @xmath125 , morphism @xmath116-modules @xmath153 and composition defined by @xmath154 let @xmath155 be the @xmath144-bimodule with generators @xmath156 and relations @xmath157 this @xmath144-bimodule is @xmath158 generated by @xmath159 and zero otherwise .
it is a well - known fact from representation theory that , under the isomorphism @xmath160 , we have a bimodule isomorphism @xmath161 sending @xmath162 to the element represented by the following extension @xmath163 compare @xcite , hence @xmath164 [ iso1 ] there is an isomorphism of categories @xmath169 which is defined on the second factor by the isomorphism @xmath170 and the splitting , and on the first factor it sends @xmath171 , @xmath172 , to the inverse of @xmath173 in the homotopy category composed with the morphism @xmath174 below , @xmath175 the arrows @xmath176 and @xmath177 in and lift to the following two morphisms @xmath178 and @xmath179 between cofibrant replacements , respectively , @xmath180 given @xmath181 we denote @xmath182 the unique composite of s and s , and of s and s , respectively .
we have @xmath183 .
the morphism is trivial in @xmath111 since it factors through a contractible object as follows , @xmath195 on the contrary can not be trivial in @xmath111 since it contains a commutative square of the form @xmath196 which induces the non - trivial morphism @xmath197 in @xmath198 on cofibers ( i.e. cokernels ) of horizontal arrows ( which are cofibrations ) . in order to prove that the functor in the statement is well defined we only have to check that @xmath200 is a square - zero ideal .
the only non - trivial cases to check are the composites of generators @xmath202 such a composite , precomposed with the isomorphism @xmath184 in the homotopy category , is @xmath203 the morphism @xmath204 is @xmath205 therefore @xmath206 , and hence also the composite @xmath207 .
now that we have proved that the functor is well defined , the previous computation of @xmath200 automatically shows that the functor is actually an isomorphism , see the descriptions of the bimodule @xmath155 .
moreover , @xmath199 and its splitting correspond to the projection and the inclusion of the second factor of @xmath208 , respectively . for any @xmath116-linear category @xmath125
we denote @xmath209 the @xmath116-linear category obtained by formally adding a zero object @xmath87 .
this is the left adjoint of the forgetful functor from @xmath116-linear categories with zero object to @xmath116-linear categories .
we define the simplicial @xmath116-linear category with zero object @xmath210 as @xmath211 , @xmath212 , @xmath213 , faces and degeneracies defined on objects by @xmath214 on morphisms in @xmath144 as @xmath215 and on generators of @xmath155 as @xmath216 we leave the reader the tedious but straightforward task to check that @xmath210 is well defined .
notice that this simplicial category only depends on the field @xmath116 .
actually , it is functorial in @xmath116 .
moreover , the inclusion of the second factors @xmath217 define a simplicial @xmath116-linear subcategory with zero object @xmath218 , however the face @xmath96 does not take all elements in @xmath155 to elements in @xmath219 . consider the following set of objects of @xmath220 , @xmath221 with the choices of cofibers indicated in the following short exact sequences , @xmath222 note that the full inclusions @xmath223 define a simplicial subcategory with zero object @xmath224 .
[ iso2 ] there is a unique isomorphism @xmath225 such that the cofibrant replacements @xmath226 define natural isomorphisms indicated with double arrow in the following diagram , @xmath82 , @xmath227 ^ -{\xi(r,\alpha)_{n+1}}_-{\cong}_<(.5){\begin{array}{c}\vspace{5pt}\end{array}}="a"\ar[d]_-{\eqref{iso1}}^-{\cong } & \ho(s_{n+1}{\mathscr{w}}_{r})_{\tilde{e}_{n}\cup\{0\}}\ar[d]^-{\text{full inclusion}}\\ \ho({\mathscr{w}}_{r}^{{\mathbf{n}}})_{e_{n}\cup\{0\}}\ar[r]_-{\text{full inclusion}}^<(.1 ) { \begin{array}{c}\vspace{5pt}\end{array}}="b"&\ho({\mathscr{w}}_{r}^{{\mathbf{n } } } ) \ar@{=>}"a";"b"^-{\scriptscriptstyle\cong}}\ ] ] the _ additivization _ of a @xmath116-linear category @xmath125 is the category @xmath228 whose objects are sequences @xmath229 of finite length @xmath213 of objects in @xmath125 , and whose morphisms are matrices , @xmath230 this is an additive @xmath116-linear category with zero object @xmath231 the empty sequence ( of legth @xmath21 ) .
the additivization is the left adjoint ( in the weak @xmath7-categorical sense ) of the forgetful @xmath7-functor from small additive @xmath116-linear categories to small @xmath116-linear linear categories ( weak because the counit is given by the choice of direct sums @xmath232 and zero object @xmath87 in @xmath125 ) .
moreover , if @xmath126 is a set of generators of the monoid of isomorphism classes of objects of @xmath125 then an adjoint @xmath233 of the full inclusion @xmath234 is an equivalence .
[ iso3 ] the isomorphism @xmath235 in lemma [ iso2 ] induces a simplicial @xmath116-linear functor @xmath236 which is a levelwise equivalence .
in particular , restricting to isomorphisms we obtain a simplicial functor which is a levelwise equivalence , @xmath237 let @xmath48 be a prime . the rings @xmath238/\varepsilon^{2}$ ] and @xmath239
are commutative noetherian local with non - trivial maximal ideal generated by @xmath240 and @xmath48 , respectively , and such that @xmath113 , therefore they satisfy the assumptions of this section .
dugger and shipley proved that @xmath241/\varepsilon^{2}}$ ] and @xmath242 are quillen equivalent to the categories @xmath243 and @xmath244 of right modules over the differential graded algebras defined in the introduction , respectively , where weak equivalences are quasi - isomorphisms and fibrations are levelwise surjections ( * ? ? ?
* corollary 4.4 ) ( we here use cohomological degree , unlike dugger and shipley ) .
this , together with the approximation theorem in @xmath0-theory ( e.g. see ( * ? ? ?
* thorme 2.15 ) , ( * ? ? ?
* corollary 3.10 ) ) shows that we can replace everywhere @xmath245 and @xmath246 with @xmath247/\varepsilon^{2}}$ ] and @xmath248 , respectively .
now the equivalence in the statement of the theorem is obtained by taking the loop space of the geometric realization of the second map in the previous corollary .
the last claim for @xmath63 is a computation made in @xcite , see also ( * ? ? ?
* remark 4.9 ) .
for @xmath67 we can apply the computations @xmath249 , @xmath250 ( * ? ? ? * theorem 8) , @xmath251/\varepsilon^{2})=0 $ ] @xcite , and @xmath252 ( * ? ? ?
* corollary 4.4 ) to the long exact sequences arising from the homotopy fibration in @xcite in order to obtain exact sequences , @xmath253/\varepsilon^{2}})\r 0,\qquad & \mathbb{z}/3\r k_{3}({\mathscr{w}}_{\mathbb{z}/4})\r \mathbb{z}/2\r 0 . \end{array}\ ] ] therefore @xmath254 contains a cyclic group of order @xmath7 unlike @xmath255/\varepsilon^{2}})$ ] , so they can not be isomorphic .
for @xmath65 we apply the calculations @xmath256/\varepsilon^{2})=0 $ ] and @xmath257 in @xcite together with the well known @xmath258 and @xmath259 from ( * ? ? ?
* theorem 8) . in this case
we obtain the following exact sequences , @xmath260/\varepsilon^{2}})\r 0,\qquad & \mathbb{z}/26\r k_{5}({\mathscr{w}}_{\mathbb{z}/9})\r \mathbb{z}/3\r 0 . \end{array}\ ] ] hence @xmath261 contains a cyclic group of order @xmath262 and @xmath263/\varepsilon^{2}})$ ] does not , so they are not isomorphic .
note that the approximation theorem in @xmath0-theory implies that the stable model categories @xmath241/\varepsilon^{2}}$ ] and @xmath242 can not be connected by a zigzag of quillen equivalences . by a result of renaudin ( * ? ? ?
* thorme 3.3.2 ) , it follows that the associated triangulated derivators are not equivalent .
however , by corollary [ iso3 ] , their difference can not be detected on the part that is involved in the definition of derivator @xmath0-theory .
given a commutative noetherian ring @xmath103 , let @xmath109 be the abelian ( in particular exact ) category of finitely generated @xmath103-modules and @xmath264 the inclusion of the full exact subcategory of projectives .
there is an induced morphism between the associated triangulated derivators , @xmath265 which is given for every @xmath5 in @xmath266 by the inclusion of a thick subcategory , @xmath267 the pointwise verdier quotients define a prederivator , @xmath268 let @xmath52 denote the terminal category and , given @xmath39 in @xmath266 and an object @xmath269 in @xmath39 , let @xmath270 be the inclusion of @xmath269 .
[ conservative ] for every finite direct category @xmath39 , the collection of functors @xmath271 , @xmath269 in @xmath39 , is conservative , i.e. , a morphism @xmath272 in @xmath273 is an isomorphism if and only if @xmath274 is an isomorphism for every @xmath269 in @xmath39 .
let @xmath275 be a bounded above complex of projective @xmath103-modules which is quasi - isomorphic to a bounded complex of projective @xmath103-modules @xmath276 with @xmath277 for @xmath278 .
then @xmath279={\operatorname{ker}}[d^{m+1}\colon p^{m}\r p^{m+1}]$ ] is a projective @xmath103-module for all @xmath280 . both @xmath275 and @xmath276 are fibrant and cofibrant in the model category @xmath281 of complexes of @xmath103-modules (
* lemma 2.3.6 and theorem 2.3.11 ) , therefore there exists a quasi - isomorphism @xmath282 .
the mapping cone @xmath283 of @xmath272 is a bounded above acyclic complex of projective @xmath103-modules , therefore @xmath283 is contractible ( * ? ? ?
* lemma 2.3.8 ) , in particular @xmath284={\operatorname{ker}}[d^{m+1}\colon c_{f}^{m}\r c_{f}^{m+1}]\ ] ] is a projective @xmath103-module for all @xmath285 .
the complex @xmath286 coincides with @xmath275 in degrees @xmath280 , hence we are done .
let @xmath287 be a morphism in @xmath273 such that @xmath274 is an isomorphism for all objects @xmath269 in @xmath39 .
let @xmath288 be a morphism in @xmath289 whose image under the localization functor @xmath290 is isomorphic to @xmath272 .
since @xmath291 is a triangulated functor , the distinguished triangle @xmath292\ ] ] in @xmath293 maps to a distinguished triangle isomorphic to @xmath294\ ] ] and so , by the assumption , there is an object @xmath295 in @xmath296 and an isomorphism @xmath297 for every object @xmath269 in @xmath39 .
let @xmath298 be an integer such that @xmath299 for every @xmath300 and @xmath269 in @xmath39 .
we consider the closed model category , in the sense of quillen @xcite , of bounded above complexes @xmath301 of finitely generated @xmath103-modules , where the weak equivalences are the quasi - isomorphisms and the fibrations are the levelwise surjective morphisms .
in the derivable category @xmath302 @xcite , there is a cofibrant resolution given by a pointwise quasi - isomorphism @xmath303 where @xmath304 is a diagram of bounded above complexes of finitely generated projective @xmath103-modules . by the previous lemma ,
replacing the complex @xmath305 with its functorial truncation @xmath306 gives a diagram of bounded complexes of finitely generated projective @xmath103-modules such that the inclusion @xmath307 is a pointwise quasi - isomorphism .
hence @xmath308 is isomorphic in @xmath309 to an object in @xmath310 , so @xmath311 is an isomorphism , as required to show .
let @xmath312 be the waldhausen category with cylinders with underlying category @xmath313 where cofibrations are the monomorphisms and weak equivalences are the morphisms whose mapping cone is quasi - isomorphic to a bounded complex of finitely generated projective @xmath103-modules .
that is precisely the strongly saturated class of morphisms that map into isomorphisms under the functor @xmath314 .
let @xmath5 be a finite direct category .
the universal property of localizations implies that the dotted arrows can be uniquely filled in the following commutative diagram @xmath317 \ar[d ] & { d^b}({\mathrm{mod}(r)}^{i^{\op } } ) \ar[r ] \ar@{.>}[dl ] & { d^b}({\mathrm{mod}(r)}^{i^{\op}})/{d^b}({\mathrm{proj}(r)}^{i^{\op } } ) .
\ar@{.>}[dll ] \\
\ho(\tilde{{\mathscr{w}}}_{r}^{i^{\op } } ) & & } \ ] ] similarly in the following diagram , using proposition [ conservative ] , @xmath317 \ar[d ] & { d^b}({\mathrm{mod}(r)}^{i^{\op}})/{d^b}({\mathrm{proj}(r)}^{i^{\op } } ) \\ \ho(\tilde{{\mathscr{w}}}_{r}^{i^{\op } } ) \ar@{.>}[ur ] & } \ ] ] by the universal property , the functors @xmath318 are mutually inverse isomorphisms , natural in @xmath5 , hence the result follows . by (
* theorem 2.1 ) , @xmath321 induces an equivalence @xmath322 the functor @xmath321 is a right exact functor between strongly saturated right derivable categories , therefore by ( * ? ? ?
* thorm 3.19 ) it induces an equivalence @xmath323 of triangulated derivators . as in the previous section ,
let @xmath103 be a commutative noetherian local ring with principal maximal ideal @xmath112 such that @xmath113 , @xmath114 the residue field , and @xmath118 the natural projection .
we consider the full simplicial subcategory @xmath324 spanned by the zero object and @xmath325 , @xmath93 .
the following set @xmath326 is a basis of the monoid of isomorphism classes of objects in @xmath327 , @xmath328 with choices of cofibers as in .
there is defined an isomorphism of simplicial categories preserving the zero object @xmath329 sending @xmath330 to @xmath331 this isomorphism defines a levelwise equivalence of simplicial categories @xmath332 fitting into a commutative diagram @xmath333 ^ -{\sim}\ar@{^{(}->}[d ] & & s_{\bullet}({\mathrm{mod}(k)})\ar[d]_{\text{induced by } q}\\ \mathcal{e}_{\bullet}(k)^{{\operatorname{add}}}\ar@{^{(}->}[d ] & & s_{\bullet}({\mathrm{mod}(r)})\ar[d]_{\text{canonical } } \ar[rr]^- { \begin{array}{c } \text{\scriptsize complexes}\vspace{-3pt}\\ \text{\scriptsize concentrated}\vspace{-3pt}\\ \text{\scriptsize in degree $ 0 $ } \end{array } } & & \ho(s_{\bullet}(c^{b}({\mathrm{mod}(r)})))\ar[d]\\ \mathcal{d}_{\bullet}(k)^{{\operatorname{add}}}\ar[rr]^-{\sim}_-{\tilde{\xi}(r,\alpha ) } & & \ho(s_{\bullet}({\mathscr{w}}_{r}))\ar[rr]^-{\sim}_{\text{proposition \ref{auxiliary2}}}&&\ho(s_{\bullet}(\tilde{{\mathscr{w}}}_{r}))}\ ] ] considering the subcategories of isomorphisms and taking the loop space of the geometric realizations in this diagram , we obtain the following commutative diagram , @xmath334 ^ -{\sim}\ar@{^{(}->}[dd ] & k({\mathrm{mod}(k)})\ar[d]^{\text{d\'evissage}}_{\sim}\\ & k({\mathrm{mod}(r)})\ar[d ] \ar[r]^-{\rho } & k({\mathbb{d}}({\mathrm{mod}(r)}))\ar[d]\\ \omega|{\operatorname{iso}(\mathcal{d}_{\bullet}(k)^{{\operatorname{add}}})}|\ar[r]^-{\sim } & k({\mathbb{d}}({\mathscr{w}}_{r}))\ar[r]^-{\sim}_-{\text{props .
\ref{auxiliary1 } and \ref{auxiliary2}}}&k(\underline{{\mathbb{d}}}(r))}\ ] ] assume that derivator @xmath0-theory satisfies both the localization and comparison conjectures .
then @xmath335 is an equivalence and the short exact sequence of triangulated derivators @xmath336 induces a homotopy fibration of @xmath0-theory spaces @xmath337 therefore @xmath338 is the homotopy fiber of the map @xmath339 which is independent of @xmath103 ( it only depends on @xmath116 ) , but this contradicts the fact that , for any prime number @xmath48 , @xmath238/\varepsilon^{2}$ ] and @xmath239 have inequivalent quillen @xmath0-theories .
more precisely , if @xmath63 , @xmath340/\varepsilon^{2})&\cong \mathbb{z}/p\oplus \mathbb{z}/p \oplus\mathbb{z}/(p^{2}-1)\\&\ncong \mathbb{z}/p^{2 } \oplus\mathbb{z}/(p^{2}-1)\cong k_3(\mathbb{z}/p^{2 } ) , & \text{\cite{allps , k3wvff}},\\ k_4(\mathbb{f}_{3}[\varepsilon]/\varepsilon^{2})\cong 0 & \ncong \mathbb{z}/3\cong k_4(\mathbb{z}/9 ) , & \text{\cite{otaktozpn}},\\ k_2(\mathbb{f}_{2}[\varepsilon]/\varepsilon^{2})\cong0 & \ncong\mathbb{z}/2\cong k_2(\mathbb{z}/4 ) , & \text{\cite{k2nd , k2dvr}}.\end{aligned}\ ] ] although the suspension functor of the triangulated category @xmath341 can be easily seen to be naturally isomorphic to the identity functor ( e.g. see @xcite ) , this is not true in general for the triangulated categories @xmath111 .
for example , the suspension of the morphism in @xmath342 defined by the square @xmath343^{\scriptstyle \alpha } \ar[d ] & r \ar[d]^{\scriptstyle q } \\ 0 \ar[r ] & k}\ ] ] is isomorphic to the morphism @xmath343 \ar[d]_{\scriptstyle \alpha } & 0 \ar[d ] \\
r \ar[r]^{\scriptstyle q } & k } \ ] ] these are isomorphic to the morphisms @xmath343^{\scriptstyle \alpha } \ar[d]_-{\scriptstyle 0 } & r \ar[d]^{\scriptstyle q } \\
r \ar[r]^{\scriptstyle q } & k}\ ] ] and @xmath343^{\scriptstyle \alpha } \ar[d]_-{\scriptstyle \alpha } & r \ar[d]^{\scriptstyle 0 } \\
r \ar[r]^{\scriptstyle q } & k}\ ] ] respectively .
then note that the sum of the two morphisms represents the zero morphism from @xmath344 to @xmath345 in @xmath342 .
thus the argument of @xcite does not apply to show that the additivity theorem for derivator @xmath0-theory @xcite and the agreement on @xmath346 @xcite contradict the localization conjecture .
b. keller , _ appendice : le drivateur triangul associ une catgorie exacte _ , categories in algebra , geometry and mathematical physics , contemp .
431 , amer .
soc . , providence , ri , 2007 , pp .
369373 .
g. maltsiniotis , _ la @xmath0-thorie dun drivateur triangul _ , categories in algebra , geometry and mathematical physics , contemp . math .
431 , amer .
soc . , providence , ri , 2007 , pp . 341368 .
r. w. thomason and t. trobaugh , _ higher algebraic @xmath350-theory of schemes and derived categories _ , the grothendieck festschrift , a collection of articles written in honor pf the 60th birthday of alexander grothendieck , progress in math .
iii , brikhuser , boston , 1990 , pp . | in this paper we show an example of two differential graded algebras that have the same derivator @xmath0-theory but non - isomorphic waldhausen @xmath0-theory .
we also prove that maltsiniotis s comparison and localization conjectures for derivator @xmath0-theory can not be simultaneously true . |
the search for the materials exhibiting large dielectric permittivity is a thrust area of research , because these materials have been recognized as promising candidates for the applications in electroceramic devices such as capacitors and random access memories .
recent reports on giant dielectric permittivity have directed considerable attention to several new materials such as percolative batio@xmath2 -ni composites @xcite , ( li , ti)-doped nio @xcite , epitaxial thin films of ca@xmath9la@xmath10mno@xmath2 @xcite and nonferroelectric cacu@xmath2ti@xmath11o@xmath12 @xcite .
the origins of large dielectric response in these materials still remain elusive .
for example , several possible mechanisms have been proposed to interpret the large dielectric constant in cacu@xmath2ti@xmath11o@xmath12 such as fluctuations of lattice - distortion - induced dipoles in nanosize domains @xcite , electrode polarization effects due to the different work functions of electrode and sample @xcite , internal barrier layer capacitor ( iblc ) effects originating from the insulating grain boundaries surrounding semiconducting grains @xcite , intragranular insulating barrier effects @xcite , etc .
currently , a considerable amount of research is devoted for understanding the origin of such high dielectric response which triggers special attention for fundamental research as well as for the development of new materials having large dielectric response .
rare earth manganites with perovskite structure have been recognised as promising candidates for exhibiting multiferroicity @xcite and colossal magnetocapacitive response @xcite which have been suggested due to the intrinsic phase separation in the grain interior .
recent investigations on the magnetic properties of lamn@xmath0fe@xmath13o@xmath2 exhibit the evidence of intrinsic phase separation between ferromagnetic and spin - glass regions in the grain interior for @xmath6 = 0.30 and 0.50 @xcite .
although detailed magnetic properties are available for lamn@xmath0fe@xmath13o@xmath2 @xcite , the transport mechanism has not been probed explicitly as a result of fe substitution .
it is worthwhile to note that the magnetic phase coexistence might be the origin of electrical inhomogeneity in the grain interior of the compounds . in this paper
, we investigate dc conductivity and dielectric response for different compositions ( @xmath6 ) and oxygen stoichiometry in lamn@xmath9fe@xmath10o@xmath2 .
the results indicate the signature of grain interior electrical heterogeneity in the compounds which leads to a considerable change in the magnitude of temperature and frequency independent dielectric constant at low temperature .
the polycrystalline compounds lamn@xmath0fe@xmath13o@xmath2 with @xmath6 = 0 , 0.15 , 0.30 , 0.50 , 0.70 , and 1.0 were prepared by a standard sol - gel technique which is described in our earlier report . @xcite the final heat treatments were performed in the form of pellets ( 1.1 cm in diameter and approximately 1 mm in thickness ) at 1000@xmath14c for 12 h in air followed by furnace cooling down to room temperature .
a part of each of the samples were further annealed at 1000@xmath14c for 12 h in a closed oxygen atmosphere .
the single phase of the crystal structure was confirmed by a bruker axs x - ray powder diffractometer ( model no .
8d - advance ) .
x - ray powder diffraction patterns are shown in figure 1 for the oxygen annealed samples of all compositions . for @xmath8 0.50
the x - ray patterns could be indexed by rhombohedral structure ( @xmath15 ) whereas orthorhombic structure ( @xmath16 ) is observed for @xmath5 0.70 which is consistent with the previous reports @xcite . in accordance with the previous report @xcite the lattice parameters
remain almost unaltered in the rhombohedral structure for @xmath8 0.30 .
the maximum intensed lines are highlighted in the inset of figure 1 where one can note that the peak positions remain almost unchanged for @xmath8 0.30 . at room temperature @xmath17fe mssbauer results
indicate the absence of precipitation of any iron oxide phase in the compounds .
capacitance ( @xmath18 ) measurements were carried out in the frequency range 20 hz to 2 mhz using an agilent e4980a lcr meter which was fitted with a computer for acquiring the data .
the measurement was carried out in the temperature range 20 - 300 k using a low temperature cryo - cooler ( advanced research systems , usa ) wired with coaxial cables .
all samples were circular discs in shape and the electrical contacts were fabricated using air drying silver paint .
the silver electrodes along with the sample were cured at 150 @xmath14c for 4 h. the dc resistivity measurements were carried out in a liquid nitrogen cryostat with a trace of helium as an exchange gas .
four probe connection was used for the measurement with air drying silver paint for connecting leads to the sample .
0.0 cm 0.0 cm 0.0 cm we have measured the real part of complex dielectric permittivity ( @xmath19 ) as a function of temperature ( @xmath20 ) and frequency ( @xmath21 ) for all the compositions .
the magnitude of @xmath19 for each sample shows a temperature as well as frequency independent behavior at low temperature which is defined as the static dielectric constant ( @xmath4 ) .
top panel of figure 2 highlights the temperature variation of @xmath19 at @xmath21 = 10 khz up to 45 k exhibiting nearly temperature independent values for oxygen annealed samples with all compositions .
the @xmath19 is also independent of frequency in the low temperature region .
the bottom panel of figure 2 shows the temperature dependence of @xmath19 measured at @xmath21 = 1 , 10 , 100 and 1000 khz for a representative composition at @xmath6 = 0.50 .
the plot exhibits nearly temperature and frequency independent value of @xmath19 below @xmath7 55 k. the moderately large value of @xmath19 ( @xmath7 10@xmath22 ) is observed at room temperature which is almost frequency independent and it shows weak temperature dependence down to @xmath7 150 k at @xmath21 = 1 khz .
figure 3 shows the variation of @xmath4 with @xmath6 .
the data for @xmath6 = 0 is not included in the figure , because temperature and frequency independent region did not develop while cooling down to @xmath7 20 k up to which our experimental facility is limited .
the reported value of @xmath4 is found in the range 17.6 - 18 for lamno@xmath2 @xcite which is close to the value of the oxygen annealed sample for @xmath6 = 0.15 .
initially , the value of @xmath4 rises with @xmath6 reaching a maximum at @xmath6 = 0.5 and then it decreases sharply at @xmath6 = 0.70 showing a further increasing trend with @xmath6 .
we note a considerable increase of @xmath4 @xmath7 43 % for the increase of @xmath6 from 0.15 to 0.50 .
the sharp drop of @xmath4 at @xmath6 = 0.70 is correlated with the structural change from rhombohedral to orthorhombic structure .
the filled and the unfilled symbols represent the data for the oxygen annealed and the as - synthesized samples , respectively , exhibiting similar dependence with @xmath6 where @xmath4 is increased further due to the oxygen annealing for @xmath8 0.50 . at @xmath6 = 0.50 the value of @xmath4
is increased up to 17 % due to the oxygen annealing .
the origin of considerable change of @xmath4 with fe substitution and oxygen annealing will be discussed elsewhere .
0.0 cm 0.0 cm we measured dc resistivity ( @xmath23 ) for both the oxygen annealed and as - synthesized samples . semiconducting temperature dependence of resistivity is observed for all the compounds where @xmath23 at room temperature increases monotonically with @xmath6 .
the resistivity data are fitted using small polaron hopping ( sph ) as well as variable range hopping ( vrh ) models .
figure 4 shows the dc resistivity data fitted to sph model defined as , @xmath24 \propto \frac{e_{\rm a}}{k_{\rm b}t}\ ] ] where only limited high temperature region is found to fit satisfactorily .
top panel shows the data corresponding to the oxygen annealed samples .
bottom panel shows the plot of the activation energies ( @xmath25 ) with @xmath6 obtained from the sph fitting corresponding to both the groups of samples where the values of @xmath26 do not change considerably due to the oxygen annealing for @xmath8 0.70 while it differs markedly at @xmath6 = 1.0 .
the values are higher for the oxygen annealed sample than the as - synthesized sample .
in addition , @xmath25 shows a small increasing trend until @xmath6 = 0.50 above which @xmath25 increases sharply .
we note that the sharp increase of @xmath25 above @xmath6 = 0.50 is associated with the structural change for @xmath5 0.70 .
top panel of fig 5 shows the dc resistivity data of the oxygen annealed samples fitted to the vrh model defined as @xmath27 \propto \left[\frac{t_0}{t}\right]^{1/4}\ ] ] which fits satisfactorily in the low temperature region .
the constants @xmath28 obtained from the slopes of the plots following vrh mechanism are shown as a function of @xmath6 for both the groups of the samples in the bottom panel of figure 5 .
in contrast to the behaviour of resistivity in the high temperature region satisfying sph mechanism , @xmath28 obtained from vrh fitting is different even at @xmath6 = 0 due to the oxygen annealing while the difference increases with increasing @xmath6 .
overall temperature dependent resistivity results for all the compositions clearly demonstrate that the effect of fe substitution and oxygen annealing strongly influences the low temperature resistivity following vrh mechanism and does not influence considerably the high temperature mechanism following sph model .
increase of oxygen stoichiometry always increases the hopping barrier .
cm 0.0 cm recent investigations on the transport properties of hole doped manganites , la@xmath0ca@xmath1mno@xmath2 @xcite and la@xmath0sr@xmath1mno@xmath2 @xcite demonstrate the strong correlation between dc conductivity and dielectric response at low temperature where low temperature dielectric constant was found to be correlated with the low temperature hopping mechanism of dc conductivity satisfying vrh process @xcite . in the present investigation a strong correlation between low temperature conductivity and dielectric constant is noticed where vrh hopping barrier increases associated with an increase in @xmath4 while @xmath6 is increasing in the range 0.15 @xmath3 0.50 . a sharp drop in @xmath4
is observed at @xmath6 = 0.70 where a structural change is noticed which is consistent with the results observed in the low hole doped la@xmath0ca@xmath1mno@xmath2 @xcite . at low temperature
the electrode polarization effect , due to different work functions of the electrode and sample , is negligible when the dielectric response is independent of frequency and temperature .
the values of @xmath4 as a function of @xmath6 in figure 3 is thus attributed to the grain interior mechanism or grain interior and grain boundary mechanism . in order to the identify the origin ,
the complex impedance and modulus plane plots have been used at temperatures where sample - electrode contribution to the response is negligible . the complex-@xmath29 and complex-@xmath30 plots at 48 k are shown in figure 6 for oxygen annealed sample with @xmath6 = 0.30 where dielectric constant is almost independent of frequency and temperature .
the @xmath30-plot indicates a portion of a single semicircle ( bottom panel of figure 6 ) whereas the @xmath29-plot ( top panel of figure 6 ) clearly resolves two semicircles .
top panel of figure 7 shows two other @xmath29-plots in the same temperature regime at 43 and 52 k together with the plot at 48 k. the onset of the arcs at high-@xmath21 are marked by the arrows which are highlighted in the inset . in order to identify the microstructural contributions to the semicircles in the complex modulus plots we examine similar plots at little higher temperatures at t = 62 , 72 and 82 k which are shown in the bottom panel of figure 7 .
one can note that as the temperature is increased a third arch gradually develops at the low frequency side .
the arch is fully developed in the plot at 82 k where the signature of crossover among different arcs are marked by the arrows . with the further increase in temperature
the high-@xmath21 arc gradually shifts out of the frequency window . for an interpretation of the experimental data , it is convenient to have a model equivalent circuit that provides a realistic representation of the electrical properties .
the simplest equivalent circuit with a serial array of parallel @xmath31 elements is used to interpret the complex plane plot where modulus plot gives emphasis to those elements with smaller capacitances and impedance plot highlights those with largest resistance @xcite .
moreover , it s well known that modulus formalism has the advantage of eliminating the contact contributions .
this plot resolves semicircles when @xmath18-values are comparable in magnitude @xcite . thus we analyze the plot using two segments of parallel @xmath32 and @xmath18 elements in series ( inset of figure 6 ) .
the capacitance values are calculated from the intercepts of the arcs on the real axis of the complex-@xmath29 plot using the formula @xmath18
= @xmath33/@xmath34 , where @xmath18 is the estimated capacitance , @xmath33 is the free space permittivity and @xmath34 is the intercept on the real-@xmath29 axis @xcite .
the values are @xmath35 = 0.53 nf and @xmath36 = 1.03 nf which are comparable to each other .
typical capacitive value of the sample - electrode interface is @xmath7 10@xmath37 f. here small values of @xmath35 and @xmath36 are attributed to the grain interior microstructure where appearance of a new low-@xmath21 arc at little higher temperatures for @xmath38 62 ( bottom panel of figure 7 ) is suggested due to the grain boundary effect .
the corresponding values of @xmath39 and @xmath40 must be calculated similarly from the intercepts in the complex impedance ( @xmath30 ) plot where the single arch represents the component with resistance much larger than the other component @xcite . here
the complex-@xmath30 plot shows only a portion of a single arc at the high-@xmath21 region within our available frequency window where , @xmath32 = 4.2 m@xmath41 .
since the high-@xmath21 arc generally corresponds to the intrinsic contribution of the sample , high @xmath32-value obtained from the analysis gives us the high resistive intrinsic component at 48 k for @xmath6 = 0.3 in lamn@xmath0fe@xmath1o@xmath2 .
the analysis of the complex-@xmath29 and @xmath30 plane plots indicates that the two high-@xmath21 arcs in complex-@xmath29 plot correspond to two different intragrain regions having comparable capacitance values but with largely different conductivities .
we note that the values of resistivity at 300 k are 2.2 and 3.2 @xmath42 10@xmath43 @xmath41 cm for lamno@xmath2 and lafeo@xmath2 , respectively where the substitution of fe in lamn@xmath0fe@xmath1o@xmath2 significantly increases the resistivity as well as the vrh hopping barrier at low temperature .
the considerable increase of @xmath4 associated with the increase of vrh hopping barrier with increasing fe substitution ( @xmath8 0.50 ) is suggested to be a result of the intragranular electrical inhomogeneity between semiconducting and insulating conductivities where fe substitution increases the insulating phase resulting in the increase of @xmath4 .
moreover , oxygen annealing further increases the insulating phase leading to a further rise of vrh hopping barrier and @xmath4 as compared to the as - synthesized samples .
recent reports on the magnetic properties of the sample with @xmath6 = 0.30 in lamn@xmath0fe@xmath1o@xmath2 clearly demonstrate the grain interior phase separation between ferromagnetic and spin - glass phases @xcite .
the magnetodielectric measurements on the sample @xmath6 = 0.30 do not show any signature of change in the dielectric constant due to the application of external magnetic field up to 30 koe .
nevertheless , it is understood that the conductivities of the ferromagnetic phases are much higher than that of the spin - glass phases which might be correlated with the electrical inhomogeneities in the grain interior of the compounds .
the low temperature static dielectric constant ( @xmath4 ) is increased considerably ( @xmath7 43 % ) with the fe substitution in lamn@xmath0fe@xmath1o@xmath2 for @xmath8 0.50 .
the value is further increased due to oxygen annealing for @xmath8 0.50 where @xmath4 is increased up to @xmath7 17 % at @xmath6 = 0.50 .
a sharp change of @xmath4 is observed for @xmath44 0.50 which is associated with the structural change .
the low temperature dc resistivities following vrh mechanism are found to be correlated with the increase of @xmath4 as a result of fe substitution for @xmath8 0.50 and due to the effect of oxygen annealing .
a simplified equivalent circuit with a serial array of parallel @xmath31 elements is used to interpret the plots of complex impedance and modulus planes which suggest the existence of intragranular electrical inhomogeneities in the compounds . * acknowledgement * s.g .
wishes to thank department of science and technology , india ( project no .
sr / s2/cmp-46/2003 ) for the financial support .
99 pecharroman c , betegon f e , bartolome j f , esteban s l and moya j s 2001 _ adv .
_ * 13 * 1541 wu j , nan c -w , lin y and deng y 2002 _ phys .
lett . _ * 89 * 217601 cohn j l , peterca m and neumeier j j 2005 _ j. appl .
phys . _ * 97 * 034102 subramanian m a , li d , duan n , reisner b a , sleight a w 2000 _ j. solid state chem . _ * 151 * 323 homes c c , vogt t , shapiro m , wakimoto s , ramirez a p. _ science _ 2001 * 293 * 673 lunkenheimer p , fichtl r , ebbinghaus s g , loidl a 2004 _ phys .
_ b * 70 * 172102 sinclair d c , adams t b , morrison f d , west a r 2002 _ appl .
* 80 * 2153 ; adams t b , sinclair d c , west a r 2006 _ phys .
_ b * 73 * 094124 fu d , taniguchi h , taniyama t , itoh m and koshihara s 2008 _ chem .
mater . _ * 20 * 1694 rao c n r and serrao c r 2007 _ j. mater .
* 17 * 4931 rairigh r p , singh - bhalla g , tongay s , dhakal t , biswas a , hebard a f 2007 _ nature physics _ * 3 * 551 patra m , de k , majumdar s and giri s 2007 _ eur . phys . j. _
b * 58 * 367 thakur m , patra m , majumdar s and giri s 2008 _ j. phys . : condens . matter _ * 20 * 195215 de k , patra m , majumdar s and giri s 2008 _ j. phys .
* 41 * 175007 liu x j , lia z q , yu a , liu m l , li w r , li b l , wu p , bai h l and jiang e y 2007 _ j. magn .
magn . mater _
* 313 * 354 bhame s d , joseph joly v l and joy p a 2005 _ phys .
_ b * 72 * 054426 de k , ray r , panda r n , giri s , nakamura h and kohara t 2005 _ j. magn .
_ * 288 * 339 zhou x -d , pederson l r , cai q , yang j , scarfino b j , kim m , yelon w b , james w j , anderson h u and wang c 2006 _ j. appl .
* 99 * 08m918 tong w , zhang b , tan s and zhang y 2004 _ phys
014422 cohn j l , peterca m and neumeier j j 2004 _ phys .
_ b * 70 * 214433 neupane k p , cohn j l , terashita h and neumeier j j 2006 _ phys .
_ b * 74 * 144428 pimenov a , hartinger ch , loidl a , mukhin a a , ivanov v y and balbashov a m 1999 _ phys .
_ b * 59 * 12419 macdonald j r 1987 _ impedance spectroscopy emphasizing solid materials and systems _
( john wiley & sons , singapore ) hodge i m , ingram m d and west a r 1975 _ j. electroanal .
_ * 58 * 429 hodge i m , ingram m d and west a r 1976 _ j. electroanal .
* 74 * 125 sinclair d c and west a r 1989 _ j. appl .
_ * 66 * 3850 | bulk dc resistivity and dielectric constant measurements with temperature and frequency have been performed in lamn@xmath0fe@xmath1o@xmath2 ( 0 @xmath3 1.0 ) as a result of fe substitution both for the as - synthesized and oxygen annealed samples .
temperature dependence of real part of dielectric constant at different frequencies show a frequency and temperature independent value ( @xmath4 ) at low temperature for @xmath5 0.15 where @xmath4 rises with @xmath6 reaching a maximum at @xmath6 = 0.5 and then it decreases sharply at @xmath6 = 0.70 showing a further increasing trend with the further increase in @xmath6 .
the sharp drop of @xmath4 at @xmath6 = 0.70 is correlated with the structural change from rhombohedral to orthorhombic structure .
a considerable increase of @xmath4 @xmath7 43 % is observed for an increase of @xmath6 from 0.15 to 0.50 .
furthermore , @xmath4 is increased considerably ( up to @xmath7 17 % at @xmath6 = 0.5 ) due to the oxygen annealing for @xmath8 0.50 .
the low temperature resistivities satisfying variable range hopping model are found to be related with the increase of @xmath4 with @xmath6 for @xmath8 0.50 .
the analysis of the complex impedance and modulus planes at low temperature indicates the electrical inhomogeneities in the grain interior of the compounds . |
the quantum volume operator is one of the most studied objects in the field of loop quantum gravity and of crucial importance for the construction of dynamics within this approach @xcite . in the literature
, one finds traditionally two versions of such an operator , due to rovelli and smolin @xcite , and to ashtekar and lewandowski @xcite , respectively .
their properties and interrelations have been intensively investigated @xcite , including a third proposal for a volume operator by bianchi , dona , and speziale @xcite .
the latter one is closer to the concept of spin foams @xcite and relies on an older geometric theorem due to minkowski @xcite .
volume operators are usually considered in connection with polyhedra .
the most elementary objects of this kind are tetrahedra consisting of four faces which are represented by angular momentum operators coupling to a total spin singlet @xcite . here
all three definitions of the volume operator coincide . among the most recent developments , bianchi and
haggard have performed a bohr - sommerfeld quantization of the volume using an appropriate parameterization of the classical phase space of a tetrahedron , and the obtained semiclassical eigenvalues agree amazingly well with exact numerical data @xcite . the purpose of the present communication is to point out that , in the sector of large eigenvalues , the volume operator of such a quantum tetrahedron is accurately described by a quantum harmonic oscillator .
our presentation will continue as follows : after briefly summarizing important features of the quantum tetrahedron and its volume operator in section [ tet ] , we derive our central result , starting from numerical observations , in section [ vol ] .
we give explicit formulae for the large - eigenvalue sector of the ( square of the ) volume operator and also analyze its scaling behavior as a function of the tetrahedron size . in section [ rec ]
we discuss the role of different angular momentum coupling schemes , and in section [ concl ] we close with an outlook .
a quantum tetrahedron consists of four angular momenta @xmath0 , @xmath1 representing its faces and coupling to a vanishing total angular momentum @xcite , i.e. the hilbert space consists of all states @xmath2 fulfilling @xmath3 in what follows we will adopt the coupling scheme where both pairs @xmath4 and @xmath5 couple first to two irreducible su(2 ) representations of dimension @xmath6 each , which are then added to give a singlet .
thus , the quantum number @xmath7 ranges as @xmath8 with @xmath9 leading to a total dimension of @xmath10 .
the volume operator can be formulated as @xmath11 where the operators @xmath12 @xmath1 represent the faces of the tetrahedron with @xmath13 and @xmath14 being the immirzi parameter . as seen form eq .
( [ defv ] ) it is useful to consider the operator @xmath15 which , in the basis of the states @xmath2 , can be represented as @xcite @xmath16 with @xmath17 here @xmath18 is the area of a triangle with edges @xmath19 expressed via herons formula , @xmath20 note that @xmath21 couples only basis states @xmath2 with neighboring labels . in the following
it will be convenient to readjust the phases of these states via the unitary matrix @xmath22 such that the resulting operator becomes real , @xmath23 since @xmath21 is antisymmetric , the spectrum of @xmath21 and , in turn , @xmath24 consists for even @xmath25 of pairs of eigenvalues @xmath26 differing in sign .
moreover , because of @xmath27 with @xmath28 , the corresponding eigenstates @xmath29 , @xmath30 fulfill @xmath31 for odd @xmath25 an additional zero eigenvalue occurs whose eigenvector ( with respect to @xmath21 ) has the unnormalized form @xcite @xmath32 which is , as it must be , an eigenstate of @xmath33 .
( filled circles ) for small @xmath34 and a typical choice of angular momentum quantum numbers .
the solid lines are oscillator wave functions @xmath35 according to eq .
( [ oscwf ] ) . ] ( filled circles ) for small @xmath34 and @xmath36 .
the solid lines are oscillator wave functions @xmath35 according to eq .
( [ oscwf ] ) . ]
we denote by @xmath37 , @xmath38 , the eigenstates of @xmath24 in descending order of eigenvalues with @xmath39 being the state with the largest eigenvalue . in the above basis
they can be expanded as @xmath40 where the coefficients @xmath41 can be viewed as the `` wave function '' of the state @xmath37 with respect to the `` coordinate '' @xmath7 .
[ fig1 ] shows this data for small @xmath34 and a typical choice of angular momentum quantum numbers ( all being of order a few ten ) . as seen there , the functions @xmath41 show the characteristic features of wave functions of the harmonic oscillator for low - lying states .
indeed , the solid lines in fig
. [ fig1 ] are gauss - hermitian oscillator wave functions for parameters to be determined a few lines below .
such properties of the functions @xmath41 occur for arbitrary sufficiently large angular momentum quantum numbers @xmath42 and sets in when all @xmath42 exceed a value of about five . for illustration ,
[ fig2 ] displays the data for the case @xmath36 where the oscillator - like features of the wave functions gradually disappear with increasing @xmath34 .
the observation made in figs .
[ fig1 ] and [ fig2 ] can be explained as follows : fig .
[ fig3 ] shows the matrix elements @xmath43 as a function of @xmath7 for several arbitrary choices of angular momentum lengths including the situation of fig .
[ fig1 ] . in all cases
, minima occur at @xmath44 with a unique maximum in between at @xmath45 determined by @xmath46 where we have considered @xmath7 as a continuous variable .
the above features can also be established by a detailed analytical discussion of the function @xmath43 .
now , since the operator @xmath24 couples only states with neighboring label @xmath7 , the wave functions with large eigenvalues will have predominantly support around the maximum of @xmath43 .
we therefore expand the matrix elements of @xmath24 between arbitrary states @xmath47 , @xmath48 ( lying predominantly in the sector of large eigenvalues ) around @xmath49 , i.e. @xmath50 , where the decrement of @xmath51 accounts for the fact that @xmath52 couples states of the form @xmath53 and @xmath54 . in doing so , we obtain @xmath55 here we have introduced the notations @xmath56 , @xmath57 , and additionally performed a continuum approximation to the latter function according to @xmath58 from eq .
( [ qmat ] ) one easily reads off an effective operator having the form of a harmonic oscillator , @xmath59 \label{qeff}\ ] ] with @xmath60 and @xmath61 for various choices of angular momentum quantum numbers .
the left panel includes the situation of fig .
[ fig1 ] . in all cases
minima occur at @xmath44 with a unique maximum in between . ] the eigenstates @xmath62 of @xmath63 are just the well - known wave functions diagonalizing the harmonic oscillator in real - space representation , @xmath64 where @xmath65 are the usual hermite polynomials .
these functions @xmath66 are plotted as solid lines in fig .
[ fig1 ] and are remarkably accurate approximations to the coefficients @xmath41 . the corresponding eigenvalues are @xmath67 .the largest eigenvalues @xmath68 of @xmath24 obtained by numerical diagonalization of the operator , and the corresponding approximate eigenvalues @xmath69 according to eq .
( [ oscev ] ) .
the choice of angular momentum quantum numbers is the same as in fig .
the exact and the approximate data agree within a few per mille . [ table1 ] [ cols="^,^,^,^",options="header " , ] table [ table1 ] compares the largest eigenvalues of @xmath24 obtained via exact numerical diagonalization with the approximate results eq .
( [ oscev ] ) .
both data coincide within a few per mille , in accordance with our previous findings regarding the corresponding eigenvectors .
note that under a rescaling of all four angular momenta , @xmath70 , @xmath71 scales in leading order as @xmath72 , while for the frequency one finds @xmath73 . in summary
, we have constructed an effective operator describing the sector of large eigenvalues of the square of the volume operator of a quantum tetrahedron .
this operator has the form of a harmonic oscillator with a coordinate@xmath74 and a momentum@xmath75 fulfilling the commutation relation @xmath76=-i\ ] ] which is part of the bedrock of quantum theory . the approximate data shown in fig . [ fig1 ] ( solid lines ) and
table [ table1 ] was generated by first finding numerically the maximum position @xmath49 of @xmath43 and inserting this value into an analytical expression of @xmath77 to obtain @xmath78 via eq .
( [ omega ] ) .
thus , no adjustable parameter is involved .
closed analytical results for @xmath49 are possible if the four angular momenta come in two pairs of equal length , and the expressions become particularly simple in the case of a regular tetrahedron , @xmath79 . here one has @xmath80 such that the parameters entering the effective operator ( [ qeff ] ) are given , to leading orders in @xmath81 , by @xmath82 @xmath83 @xmath84 thus , the eigenvalues ( [ oscev ] ) of the effective operator read to the first leading orders in @xmath81 @xmath85 in particular , from eq .
( [ omegaj ] ) we see that the width @xmath86 of the wave functions ( [ oscwf ] ) is proportional to @xmath87 .
moreover , for the largest eigenvalue of the volume operator , one finds from eq .
( [ defv ] ) @xmath88 here the leading term @xmath89 is exactly the classical volume of a regular tetrahedron whose faces have area @xmath81 , and the subleading correction is , after a redefinition of the face area , identical to the one found in ref . @xcite using bohr - sommerfeld quantization .
furthermore , our findings here suggest that the classical volume of a general tetrahedron with face areas @xmath90 is , to leading order in all @xmath42 , given by @xmath91 as already discussed in section [ tet ] , the large eigenvalues @xmath68 have counterparts @xmath92 with the the same modulus but negative sign , and according to eq .
( [ signchange ] ) , the pertaining eigenvectors can be obtained from the previous ones by changing the sign of any other component . regarding the wave functions ( [ oscwf ] ) one could try to mimic this behavior by attaching an appropriate phase factor , @xmath93 however , the above functions are clearly not eigenfunctions of the effective operator ( [ qeff ] ) .
in fact , an operator having @xmath94 as eigenstates with eigenvalues @xmath95 can be constructed as follows : @xmath96\,.\label{qeff2}\end{aligned}\ ] ] this operator is not invariant under a time reversal@xmath97 which corresponds to the fact that the eigenfunctions ( [ oscev ] ) can not be chosen to be real . moreover , the operators @xmath98 and @xmath99 are , along with their eigenfunctions , obviously just related by a u(1 ) gauge operation , apart from the global minus sign on the r.h.s of eqs .
( [ qeff1 ] ) and ( [ qeff2 ] ) .
however , since @xmath99 is merely a consequence of the rather phenomenological ansatz ( [ oscev ] ) , a more rigorous effective description of eigenstates with negative eigenvalue is desirable
. work in this direction could possibly build upon ideas of ref .
@xcite where the quantity @xmath100 was considered as an effective potential for states with eigenvalues of both sign .
there are obviously alternatives to the coupling scheme of angular momenta we have used so far . for instance , instead of the previous procedure , @xmath101 and @xmath102 could first be coupled to two irreducible representations of dimension @xmath103 each , which are then combined to a total singlet . the operator @xmath24 is then expressed in a form analogous to eq .
( [ defq ] ) with matrix elements @xmath104 given by the r.h.s of eq .
( [ alpha ] ) and obvious interchanges of labels . as seen before
, @xmath104 has a unique maximum at some @xmath105 .
thus , putting again @xmath106 , the eigenstates with large eigenvalues will again be accurately approximated by oscillator wave functions @xmath107 according to eq .
( [ oscwf ] ) with @xmath108 and the corresponding approximate eigenvalues read @xmath109 with @xmath110 .
since the exact spectrum of @xmath24 is of course independent of the coupling scheme used , this holds as well , to an excellent degree of approximation , for the approximate eigenvalues , as it is easily checked by numerics . for instance , for the parameters of fig .
[ fig1 ] and table [ table1 ] we find @xmath111 , @xmath112 and @xmath113 , @xmath114 .
thus , we have , as an again excellent approximation , @xmath115 which in particular means that the wave functions @xmath116 and @xmath107 can be taken as identical .
moreover , since switching to another coupling scheme implies just a change of basis in the hilbert space , the above two gauss - hermitian wave functions should be related by a unitary transformation , @xmath117 with some phase factor @xmath118 , @xmath119 .
an obvious solution is given by @xmath120 and @xmath121 , while another possibility follows from the well - known fact that the fourier transform of a gauss - hermitian function is a function of that same type : here one has @xmath122 and @xmath123 thus , up to the scale factor @xmath78 occurring in eq .
( [ trafo2 ] ) , changing from one coupling scheme to another just corresponds to a fourier transform of the approximating oscillator wave functions .
this observation is of course strongly reminiscent of switching from real space to momentum representation in standard quantum mechanics . on the other hand , treating @xmath7 and @xmath124 again as discrete state labels , the basis states in both coupling schemes are related by @xmath125 using wigner @xmath126-symbols in the standard convention of prefactors @xcite . for the case of all angular momenta being large
compared to unity , ponzano and regge @xcite have devised the following asymptotic expression for such quantities ( for more recent developments , see also refs .
@xcite ) , @xmath127 here @xmath128 is the volume of a tetrahedron having edge lengths @xmath129 , @xmath130 where edges occurring in the same column of the @xmath126-symbol are opposite to each other , i.e. do not have a common vertex , and @xmath131 is the external dihedral angle between faces joining at edge @xmath42 .
the cosine occurring in the above equation bears some similarity to the exponential in the transformation ( [ trafo2 ] ) .
moreover , under a rescaling of all angular momenta , @xmath132 , @xmath128 scales obviously as @xmath72 , such that the prefactor of @xmath133 in eq .
( [ recoup ] ) is proportional to @xmath134 , which is the same scaling behavior as in ( [ trafo2 ] ) .
however , we leave it to future studies to more deeply investigate the possible relationship between the transformations ( [ trafo2 ] ) and ( [ recoup ] ) .
we have shown that the ( square of the ) volume operator of a quantum tetrahedron is , in the sector of large egenvalues , accurately desribed by a quantum harmonic oscillator .
this finding is a consequence of the fact that ( i ) the volume operator couples only neighboring states of its standard basis , and ( ii ) its matrix elements show a unique maximum as a function of state labels .
the ingredients of the harmonic oscillator constructed here are an appropriate coordinate variable and a momentum operator defined by the corresponding derivative .
these two quantities fulfill the canonical commutation relation .
we give explicit formulae for the large eigenvalues of the volume operator in terms of the equidistant harmonic oscillator spectrum .
it is an interesting speculation whether or not these linear excitations of space are related to gravitational waves .
moreover , in this limit the quantum tetrahedron is naturally described semiclassically by oscillator coherent states , in contrast to other approaches where tensor products of su(2 ) coherent states projected onto the singlet subspace are used @xcite .
we have also analyzed the scaling properties of the oscillator parameters as a function of the size of the tetrahedron . for a regular tetrahedron
we reproduce recent findings @xcite on the largest volume eigenvalue and generalize them to the next smaller eigenvalues . in terms of classical geometry ,
our approach here also suggests an interesting expression given in eq .
( [ vcl ] ) for the volume of a general tetrahedron . to further investigate this conjecture
might be , from a more mathemaitcal perspective , a route for future studies ( possibly starting from numerical tests ) .
here we have shown the result only for the very special case of a regular tetrahedron .
finally , we have discussed the role of different angular momentum coupling schemes .
one might argue that the findings here on the tetrahedral volume operator are in fact very general : expanding a classical system described by just one pair of canonical variables @xcite around an extremum will generically lead to an effective harmonic oscillator .
an interesting point here is that this oscillator - like behavior sets in at already quite moderate lengths of the involved angular momenta ( being about five ) .
moreover , the quantum number resulting from the coupling of angular momenta has an immediate interpretation in terms of the oscillator coordinate .
the present work exclusively deals with tetrahedra , i.e. , in the language of spin networks , 4-valent nodes @xcite .
an obvious and interesting question is how the results found here translate to higher nodes .
recent work , in a similar spirit as here , on the semiclassical properties of pentahedra includes refs .
i thank hal haggard for useful correspondence .
c. rovelli , _ quantum gravity _ , cambridge university press 2004 . | it is shown that the volume operator of a quantum tetrahedron is , in the sector of large eigenvalues , accurately described by a quantum harmonic oscillator .
this result relies on the fact that ( i ) the volume operator couples only neighboring states of its standard basis , and ( ii ) its matrix elements show a unique maximum as a function of internal angular momentum quantum numbers .
these quantum numbers , considered as a continuous variable , are the coordinate of the oscillator describing its quadratic potential , while the corresponding derivative defines a momentum operator .
we also analyze the scaling properties of the oscillator parameters as a function of the size of the tetrahedron , and the role of different angular momentum coupling schemes . |
the depinning and dynamics of the vortex lattice ( vl ) in type ii superconductors is exemplary for the behavior of driven , periodic media in presence of a pinning potential @xcite .
other examples range from sliding surfaces exhibiting static and dynamic friction and absorbed monolayers @xcite to charge density waves ( cdw s ) @xcite , wigner crystals @xcite and ( magnetic ) bubble arrays @xcite .
vortex matter offers the advantage that the periodicity @xmath12 of the hexagonal lattice can be tuned by changing the magnetic induction @xmath13 .
in addition , the effect of various types of pinning potentials can be studied .
this pinning potential , arising from inhomogeneities in the host material , can be completely random , as in most natural materials , or can be arranged in periodic arrays using nano - fabrication techniques @xcite . in a variety of cases
correlated inhomogeneities occur naturally in a material , such as twin boundaries and the layered structure of the high @xmath14 superconductors @xcite .
depinning of the vl in a _ random potential
_ generally involves regions of plastic deformations @xcite , i.e. coexistence of ( temporarily ) pinned domains with moving domains .
for very weak pinning the typical domain size can exceed the correlation length @xmath15 of the vl ( see @xcite ) and the weak collective pinning theory @xcite can be successfully used to estimate the critical current density @xmath16 @xcite . however , as either the ratio of the vl shear modulus @xmath17 and the elementary pinning strength or the number of pins per correlated volume decreases , plastically deformed regions start to have a noticeable effect on @xmath16 .
recent imaging experiments @xcite have shown directly that the rise in @xmath16 in weak pinning materials near the upper critical field @xmath18 , known as the peak effect @xcite , originates from such , rather sudden , enhancement of the defect density .
this strong reduction of the vl correlation length is also accompanied by a qualitative change in the nature of depinning : for strong pinning , depinning proceeds through a dense network of quasi - static flow channels ( filaments ) such that the typical width of both static and moving domains has approached the lattice spacing @xcite .
depinning transitions via a sequence of static , channel like structures have also been observed experimentally via transport experiments @xcite . in superconductors with _
periodic _ pinning arrays ( ppa s ) , matching effects between the lattice and the ppa become important .
as shown first by daldini and martinoli @xcite , when the vortex spacing coincides with the periodicity of the potential , pronounced maxima can occur in @xmath16 , while at mismatch defects ( discommensurations ) appear which gives rise to a reduced @xmath16 .
the last decade , many more studies of vl s in ppa s have appeared , both experimentally and numerically .
pronounced commensurability effects were found in films with 2-d periodic pinning @xcite for flux densities equal to ( integers of ) the density of dots . in these systems , vortex chains at interstitial positions of the periodic arrays ( e.g. at the second matching field of a square pinning array ) can exhibit quasi one - dimensional motion under the influence of the interaction with neighboring , pinned vortices @xcite , as has also been observed in numerical simulations @xcite .
in addition these simulations have revealed that , depending on the vortex interactions and the symmetry or strength of the ppa , a rich variety of other states and dynamic transitions can occur , often leading to peculiar transport characteristics . besides the above examples , the phenomenon of vortex channelling can also arise from the presence of grain boundaries in the sample .
historically , the shear depinning of vortices in grain boundaries in low @xmath14 materials received considerable attention @xcite because it could explain the quadratic decrease of @xmath16 near @xmath18 in practically relevant poly - crystalline superconductors .
more recently , the interesting issue of channelling of mixed abrikosov - josephson vortices in grain boundaries in high @xmath14 materials was addressed in detail by gurevich @xcite . a system in which channel motion and its dependence on the structural properties of vortex matter
can be studied systematically is that of narrow , weak pinning flow channels in a superconducting film @xcite , see fig .
[ plot_sketch ] .
the samples are fabricated by etching straight channels of width @xmath19 nm through the top layer of an a - nbge / nbn double layer . with a magnetic field applied perpendicular to the film ,
vortices penetrate both the strong pinning nbn in the channel edges ( ce s ) and the remaining nbge weak pinning channels .
the strongly pinned ce vortices provide confinement to the vortices inside the channel as well as the pinning ( shear ) potential which opposes the lorentz force from a transport current @xmath20 applied perpendicular to the channel . by changing the applied field
@xmath21 one can tune the commensurability between the vortex lattice constants and the channel width , allowing a detailed study of the shear response and threshold for plastic flow as a function of the mismatch and the actual microstructure in the channel .
phenomenologically , plastic flow in the channel occurs when the force density @xmath22 ( with @xmath23 the vortex density ) exceeds @xmath24 , where @xmath25 is the flow stress at the edge ( the factor 2 is due to both ce s ) and @xmath1 is the effective width between the first pinned vortex rows , defined in fig .
[ plot_sketch ] .
thus , the critical force density is given by : @xmath26 the parameter @xmath27 describes microscopic details of the system : it depends on lattice orientation , ( an)harmonicity of the shear potential , details of the vortex structure in the ce s and the microstructure of the array inside the channel .
critical current measurements as function of applied field reflected this change in microstructure through oscillations of @xmath28 , shown in fig .
[ plot_introfs ] for a channel with @xmath29 nm .
note that in such a narrow channel , the pinning strength due to intrinsic disorder in the a - nbge is at most @xmath30 of @xmath28 ( except for @xmath31 mt ) and does not affect the oscillations .
since the natural row spacing of the vl is @xmath32 , with @xmath33 , and in our geometry @xmath34 one can check that the periodicity of the oscillations corresponds to transitions from @xmath3 to @xmath35 with @xmath4 integer , i.e. the principal lattice vector @xmath36 is oriented along the channel ( fig .
[ plot_sketch ] ) .
the envelope curve represents eq.([phenomenological ] ) with brandt s expression for the vl shear modulus @xcite : @xmath37 ( @xmath38 is the reduced field and @xmath39 is the penetration depth ) and a value @xmath40 .
this value for @xmath27 is close to the value @xmath41 for the relative displacements at the crossover from elastic to plastic deformations as obtained from measurements on the peak effect @xcite .
this led to a qualitative interpretation of the reduction of @xmath28 at minima as being due to defects in the channel , which develop at incommensurability .
however , recent developments @xcite have shown that ( strong ) structural disorder may be present in the ce arrays , in which case the interpretation can drastically differ .
in this article we present numerical and analytical studies of the threshold force and dynamics of vortices in the channel system for various degrees of edge disorder . in an early paper @xcite we studied the commensurability effects in the idealized case with periodic arrays in the ce s .
in this situation @xmath28 at matching ( @xmath3 ) is equal to the ideal lattice strength @xmath42 ( the value @xmath43 follows from frenkels considerations @xcite ) , while at mismatch dislocations develop , leading to @xmath44 .
the resulting series of delta - like peaks in @xmath28 versus matching parameter differed considerably from the experimental results , which could not be explained by thermal fluctuations or intrinsic disorder inside the channel .
therefore we investigated the effect of positional disorder in the ce arrays on @xmath28 near commensurability ( @xmath45 ) @xcite . in this regime
the behavior is dominated by the longitudinal displacements of vortices in the chains , i.e. quasi one - dimensional ( @xmath46d ) , and @xmath28 is controlled by defects with burgers vector along the channel . at weak disorder , we found a clear reduction of @xmath28 at commensurability caused by nucleation of defects at threshold , while the existing defects at incommensurability become pinned by disorder , leading to an increase of @xmath28 in the mismatching case .
the present paper first describes in detail these quasi 1d phenomena near commensurability and/or for weak disorder . using a generalized sine - gordon model ,
we quantitatively describe how the structure and transport properties depend on the vortex interaction range and on weak disorder in the ce s .
besides the connection to our system , these results also provide a background for understanding quasi-@xmath46d vortex states and matching effects in artificial ppa s , including the effects of disorder which these ppa s may contain due to fabrication uncertainties .
the 1d model shows that , above a certain disorder strength , spontaneously ( disorder ) induced defects along the ce s dominate over incommensurability induced defects .
the commensurability peak in @xmath28 is then completely smeared out with a value of @xmath28 at matching ( @xmath3 ) saturating at @xmath7 of the ideal lattice strength . in the more general case of wider channels ,
the _ transverse _ degrees of freedom , especially away from matching ( @xmath47 ) , lead to new phenomena : under the influence of disorder , the channel array may split up in regions with @xmath4 and @xmath48 rows , involving dislocations with burgers vector strongly _ misaligned _ with the ce s . at sufficient disorder strength ,
such dislocations lead to a more effective pinning of the array then the aligned dislocations around matching .
@xmath28 then exhibits a smooth oscillation as function of @xmath11 , similar to fig .
[ plot_introfs ] , with yield strength maxima occurring _ around mismatch_.
this behavior resembles the classical peak effect , i.e. at mismatch the enhanced ability of the arrays to sample configuration space allows better adjustment to the random ce s . in the last part of the paper
we show detailed simulations of both static and dynamical aspects of this behavior , including a study of reordering phenomena at large drive .
we find an ordering velocity of the arrays with a channel width dependence similar to that of the threshold force .
using a modified version of the dynamic ordering theory in @xcite it is shown that such behavior can be explained by a reduction of the energy for formation of misaligned defect pairs away from matching .
the numerical results at strong disorder are also in good qualitative agreement with recent mode - locking experiments on the channel system @xcite .
the outline of the paper is as follows . in sec.[secmodel ]
the channel geometry and the simulation procedure are discussed .
the first part of the paper deals with channels having hexagonal , ordered arrays in the ce s : in sec.[secord1d ] we present the sine - gordon description and numerical results for a single 1d vortex chain in an ordered channel ; in section [ secord2d ] we show how the 1d behavior extends to wider channels with multiple rows and ordered ce s .
the second part of the paper deals with disordered channels : section [ secdis1d ] describes the effects of weak ce disorder on the behavior of a 1d chain , both analytically and using numerical simulations .
the effects of weak disorder in wider channels are discussed in sec.[secweakdis2d ] .
section [ secstrongdis2d ] describes the static and dynamic properties of wider channels in presence of strong disorder , including an analysis of the reordering phenomena in this situation . a comparison with the dynamic ordering theory , the confrontation with experiments and a summary of the results are presented in secs.[secdiscuss ] and [ secsum ] .
we consider straight vortices at @xmath49 in the geometry as illustrated in fig .
[ plot_geometry ] for the case of 1 row in the channel . the approximation @xmath49 is well justified over a considerable range of the experiments ( see sec.[secdiscuss ] ) .
the ce s are formed by two semi - infinite static arrays .
the distance between the first vortex rows on both sides of the channel is @xmath50 , with @xmath1 the _ effective _ channel width .
the vortices are assumed to be fixed by columnar pins in the ce s .
the principal axis of the pinned arrays is along the channel direction @xmath51 .
a relative shift @xmath52 is allowed between the arrays . in ( a )
the simplest configuration is shown , where ce vortices form a perfect triangular lattice . for @xmath53 ,
their coordinates are : @xmath54a_{0},m[b_0+(w - b_0)/2|m| ] ) , \label{ordedgecoord}\ ] ] for @xmath55 and @xmath56 denotes the remainder of @xmath57 .
disorder is incorporated in the model by adding random shifts @xmath58 to the coordinates of the ordered arrays : @xmath59 as shown in fig .
[ plot_geometry](b ) .
the amplitudes of the random shifts are characterized by disorder parameters @xmath60 and @xmath61 as follows : transverse relative displacements @xmath62 are chosen independently from a box distribution @xmath63 $ ] .
the longitudinal shifts @xmath64 are chosen such that the _ strain _
@xmath65 along the channel is uniformly distributed in the interval @xmath66 $ ] .
the latter provides a simple way of implementing loss of long range order along the ce s . for @xmath60 and @xmath61
we study the following specific cases : @xmath67 in secs .
[ secord1d],[secord2d ] , @xmath68 in sec.[secdis1d ] and @xmath69 in secs.[secweakdis2d],[secstrongdis2d ] . to study the commensurability effects , the effective width of the channel is varied from a value @xmath70 .
we assume that the vortex density in and outside the channel are the same . the number of vortices in the channel is then given by @xmath71 with @xmath72 the channel length . in a commensurate situation
one has @xmath73 and both the row spacing and ( average ) longitudinal vortex spacing in the channel match with the vortex configuration in the ce s . when @xmath5 these spacings become different , leading to generation of topological defects .
while this model differs from the experimental case where the applied field drives the incommensurability , the method offers a simple way of introducing geometrical frustration and study various ( mis)matching configurations .
with a uniform transport current @xmath20 applied perpendicular to the channel , the equation of motion for vortex @xmath74 in the channel reads ( in units of n / m ) : @xmath75 @xmath76 is the vortex - vortex interaction potential , @xmath77 labels other vortices inside the channel , the damping parameter @xmath78 is given by @xmath79 with @xmath80 the flux flow resistivity , and @xmath81 is the drive along the channel .
for films which are not too thin compared to the penetration depth @xmath39 and magnetic fields small compared to the upper critical field @xmath18 , the interaction @xmath76 is given by the london potential : @xmath82 where @xmath83 and @xmath84 is the flux quantum . in the simulations we integrate eq.([eqm ] )
numerically for all vortices in the channel .
we use a runge - kutta method with variable time steps such that the maximum vortex displacement in one iteration was @xmath85 .
distances were measured in units @xmath12 ( @xmath86 ) , forces in units @xmath87 and time in units @xmath88 .
following ref .
@xcite , the london potential was approximated by : @xmath89 with a cut - off radius @xmath90 corresponding to @xmath91 .
we performed most simulations for @xmath92 .
periodic boundary conditions in the channel direction were employed . for each @xmath11 we relaxed the system to the ground state for @xmath93 .
we found that this is best achieved by starting from a uniformly stretched or compressed @xmath4 or @xmath48 configuration .
for an initial configuration with @xmath94 vortices distributed randomly in the channel , relaxation resulted in ( slightly ) metastable structures , even when employing a finite temperature , simulated annealing method .
some peculiarities associated with such structures are mentioned in secs.[secord2d ] and [ secweakdis2d ] .
after the @xmath93 relaxation , the average velocity versus force ( _ v - f _ ) curve was recorded by stepwise varying the force from large @xmath95 ( occasionally @xmath96 was used to check for hysteresis ) . at each force
we measured @xmath97 after the temporal variations in @xmath98 became @xmath99 ( ignoring transients by discarding the data within the first @xmath100 ) .
in addition , at each force we measured several other quantities , e.g. the temporal evolution of @xmath101 and the time dependent velocity @xmath102 .
the first relevant issue for plastic flow and commensurability effects in the channel is to understand the influence of periodically organized vortices in the ce s ( see fig . [
plot_geometry]a ) .
the characteristic differences between commensurate and incommensurate behavior can be well understood by focusing on a 1-d model in which only a single vortex chain is present in the channel .
therefore the ce s are assumed to be symmetric with respect to @xmath103 ( i.e. @xmath53 in fig .
[ plot_geometry ] ) and only the longitudinal degrees of freedom of the chain are retained . at commensurability , @xmath104 , the longitudinal vortex spacing @xmath105 . for @xmath106 the average spacing @xmath107 does not match with the period @xmath12 in the edges and interstitials or vacancies develop in the channel .
their density @xmath108 is given by @xmath109 .
we first consider the interaction of a vortex in the channel with the periodic arrays in the ce s . as shown in app.[appa ] , when @xmath110 and @xmath111 the edge potential arising from this interaction is : @xmath112 where @xmath113 . for @xmath104 and @xmath103 , the associated sinusoidal force caused by the edge has an amplitude which we denote by @xmath114 : @xmath115 next we consider a static chain of vortices _ inside _ the channel . the chain is most easily described in terms of a continuous displacement field @xmath116 , representing the deviations of vortices in the chain with respect to the commensurate positions , i.e. @xmath117 . the edge force is then given as @xmath118 . to describe the interaction between vortices _ within _ the chain , we assume that their relative displacements are small , @xmath119
. then one can use linear elasticity theory .
taking into account that the interaction range @xmath120 , the elastic force at @xmath121 is : @xmath122 .
\label{semidiscretefel}\ ] ] using the fourier transform of @xmath123 the force due to a displacement @xmath124re@xmath125 with wave vector @xmath126 is : @xmath127u(x ) .
\label{kappaderive}\ ] ] recasting this into a sum over reciprocal vectors @xmath128 and retaining only the @xmath129 term , one obtains the following dispersive elastic modulus of the chain : @xmath130 for deformations of scale @xmath131 , the elastic force is @xmath132 with a long wavelength stiffness @xmath133 .
the equation of motion for @xmath134 for a uniformly driven chain , is obtained by adding the driving force @xmath135 to the edge force and the intra - chain interactions resulting in : @xmath136 .
assuming for the moment that the long wavelength description is valid , the evolution of @xmath134 is given by the following sine gordon ( s - g ) equation : @xmath137 a useful visual representation of eq.([ordconteqm ] ) is an elastic string of stiffness @xmath138 with transverse coordinate @xmath116 in a tilted washboard potential @xmath139 .
the s - g equation ( [ ordconteqm ] ) has been thoroughly studied in different contexts ( e.g. @xcite ) . in the static case ( @xmath93 )
, it has the trivial solution @xmath140 , corresponding to a commensurate chain , or kinked , incommensurate , solutions in which @xmath116 periodically jumps by @xmath141 , each jump representing a point defect in the channel . in the context of long josephson junctions ( ljjs , @xcite )
, a kink corresponds to a josephson vortex where the phase difference across the junction changes by @xmath142 .
isolated _ defect is represented by the familiar soliton solution of the s - g model : @xmath143/\pi .
\label{kinkforms}\end{aligned}\ ] ] here @xmath144 denotes the center of the defect and the @xmath145 ( @xmath146 ) sign denotes a vacancy or interstitial ( kink or antikink ) .
the length @xmath147 represent the core size of the defect : @xmath148 with @xmath149 the dimensionless ratio between the chain stiffness and maximum curvature of the pinning potential : @xmath150 as follows from eq.([mudef]),([kappaexpression ] ) . for @xmath151 , @xmath147
thus considerably exceeds the lattice spacing .
the continuum approach is validated since @xmath152 . in fig .
[ plot_sgkink ] we have illustrated the characteristic defect shape eq.([kinkforms ] ) , along with numerical data from a later section .
the long wavelength limit is only valid when @xmath147 considerably exceeds @xmath39 .
since @xmath147 grows only as @xmath153 the dispersion in the elastic interactions becomes important beyond a certain value of @xmath154 .
this value is estimated by setting @xmath155 in eq.([kappaexpression ] ) , resulting in @xmath156 , in which case @xmath157 . for larger interaction range
one employs the following approach , first derived by gurevich @xcite for mixed abrikosov - josephson vortices in grain boundaries .
expression ( [ semidiscretefel ] ) for the elastic force can be written as an integral @xmath158 . for defects on a scale @xmath159 , only the short distance behavior of @xmath123 has to be retained : @xmath160 . integrating the expression for @xmath161 by parts and adding the edge force and
the drive , the equation of motion becomes : @xmath162 a static solution of eq.([nonlocalordconteqm ] ) for a single defect is @xcite : @xmath163 with @xmath164 , which is valid when @xmath165 .
the value for @xmath166 is nearly the same as the s - g core size @xmath147 for @xmath167 .
this means that upon approaching the nonlocal regime , the increase of core size saturates at @xmath168 , while only the tails of the defect are affected according to eq.([nonlocalkink ] ) , see fig . [ plot_sgkink ] .
a more accurate calculation of the onset of the nonlocal field regime using brandt s field - dependent vortex interaction ( app.[appa ] ) shows that non - locality is only relevant for a channel in a superconductor with @xmath169 .
so far , we discussed isolated defects . for finite defect density , the repulsive interaction between defects of the same sign causes a periodic superstructure in the chain .
when @xmath108 grows to @xmath170 , the defects start to overlap significantly . for the ( local ) s - g model ,
explicit solutions for @xmath134 have been obtained in terms of the jacobi elliptic functions , for which we refer to @xcite .
recently , also in the non - local limit where @xmath171 , the soliton chain has been described analytically @xcite , which we will not repeat here . with a uniform drive @xmath135
, the transport properties strongly depend on the presence and density of defects in the channel . at commensurability ( @xmath105 , @xmath172 ) , a threshold force @xmath173 is required , above which all vortices start moving uniformly .
their velocity is identical to that of an overdamped particle in a sinusoidal potential : @xmath174 @xcite .
the threshold @xmath114 coincides with the well known relation between shear strength and shear modulus of an ideal lattice by frenkel @xcite : for a harmonic shear interaction , a value @xmath175 applies in eq.([phenomenological ] ) . identifying @xmath176 for @xmath104 ,
one finds : @xmath177 which coincides with the familiar expression for the shear modulus in the london limit : @xmath178 . in app.[appa ] we generalize the expression for the ordered channel potential to higher field and show that also in that case the potential is harmonic and that @xmath179 holds for a commensurate channel . at incommensurability ,
depinning of the chain is governed by the threshold force to move a defect . in the present continuum approach such threshold
is absent .
however , taking into account the discreteness of the chain , in which case eq.([ordconteqm ] ) turns into a frenkel - kontorova ( fk ) model @xcite , a finite peierls - nabarro ( pn ) barrier exists to move a defect over one lattice spacing ( see e.g. @xcite ) .
the magnitude of the pn barrier has been studied for a variety of cases , including fk - models with anharmonic and/or long - range interactions @xcite . for @xmath180 ,
@xmath181 can amount to a considerable fraction of @xmath114 .
additionally , in this regime anharmonicity may renormalize @xmath149 @xcite and cause pronounced differences between the properties of kinks ( vacancies ) and antikinks ( interstitials ) . in our limit @xmath182 , where @xmath183 and harmonic elastic theory applies , these differences are small and the pinning force vanishes as @xmath184 .
hence , defects in an ordered channel give rise to an essentially vanishing plastic depinning current @xmath0 @xcite .
considering the dynamics , for small drive @xmath185 the motion of defects , each carrying a flux quantum , provides the flux transport through the channel .
when defects are well separated , for @xmath186 , the mobility of the chain is drastically reduced compared to free flux flow and the average velocity @xmath98 is proportional to the defect density : @xmath187 . here @xmath188 is the velocity of an isolated defect at small drive .
it can be calculated from the general requirement that the input power must equal the average dissipation rate : @xmath189 the last step arises from the space and time periodicity of @xmath134 . using @xmath190 and the kinkshape eq.([kinkforms ] ) ,
one obtains the flux flow resistivity at small defect density : @xmath191 where @xmath192 is the kink mobility in the s - g model @xcite . for larger defect density , where defects start to overlap this relation changes . the linear response for @xmath193
may then be obtained from the solutions for @xmath134 based on elliptic integrals @xcite . for larger drive @xmath194 ,
the tilt induced reduction of the ( washboard ) edge potential becomes important .
this leads to an expansion of the cores of the sliding defects and causes a nonlinear upturn in the _ v - f _ curves .
exact solutions of eq.[ordconteqm ] describing this behavior do not exist .
therefore we use a perturbative method similar to that in @xcite which is able to describe the full _ v - f _ curve over a wide range of defect densities .
it is convenient to define the displacements @xmath195 , where @xmath196 , with @xmath197 , is the continuous field describing the displacements of an undeformed incommensurate chain ( i.e. straight misoriented string in the washboard potential ) moving with velocity @xmath98 . in terms of @xmath198 , the equation of motion ( [ ordconteqm ] ) and eq.([dissiprate ] ) can be written as : @xmath199 @xmath200 the last term in eq.([fvdispl ] ) describes additional dissipation due to internal degrees of freedom in the chain . under influence of the potential
, @xmath198 acquires modulations with period @xmath201 in @xmath51 , i.e. period @xmath12 in @xmath202 .
these modulations are then expressed as a fourier series of modes with wavelength @xmath203 ( @xmath204 integer @xmath205 ) and amplitude @xmath206 : @xmath207+c.c . \label{persolansatz}\end{aligned}\ ] ] the overlap of defects and the core expansion for @xmath208 appears in the @xmath126 and @xmath98 dependence of @xmath198 .
both effects cause a reduction of the relative displacements @xmath198 .
an approximate solution for @xmath209 is obtained by substituting eq.([persolansatz ] ) into eq.([ordconteqm ] ) , yielding the coefficients @xmath206 ( details of the solution are deferred to app.[appb ] ) .
the _ v - f _ relation eq.([fvdispl ] ) attains the form : @xmath210}\right ] .
\label{fvresultana}\end{aligned}\ ] ] the additional friction force is represented in terms of the pinning frequency @xmath211 , the washboard frequency @xmath212 and @xmath213 which is the effective relaxation frequency for nonlinear deformations associated with a defect density @xmath214 , with @xmath215 given in app.[appb ] . at small @xmath98 , the elastic relaxation time @xmath216 for the chain to relax
is much smaller than the timescale @xmath217 between passage of maxima in the edge potential .
this corresponds to the linear sliding response of the static structure of ( overlapping ) defects . for large @xmath98 , @xmath218 meaning that the incommensurate chain is not given enough time to deform .
this leads to expanded defects described by a sinusoidal variation of @xmath198 with reduced amplitude ( see app.[appb ] ) .
the _ v - f _
curve then approaches free flux flow according to @xmath219 as for a single particle .
recently , exact solutions describing the nonlinear dynamics and core expansion of mixed abrikosov - josephson vortices based on the nonlocal eq.([nonlocalordconteqm ] ) have been derived in @xcite .
the resulting transport curves are very similar to those obtained from eq.([fvresultana ] ) , see fig .
[ plot_fvord1d ] .
we also note the similarity with the @xmath220 curves obtained from a model for kinked josepson strings @xcite in high @xmath14 superconductors with the field under an angle with respect to the insulating layers .
the simulations of symmetric channels ( @xmath53 ) for @xmath221 fully support the above findings .
the interaction with the ce s for @xmath104 provides a maximum restoring force with a value @xmath222 , independent of the interaction cut - off @xmath90 used in the numerics .
this value is in agreement with the dimensionless values for @xmath114 and @xmath17 in eq.([mudef ] ) and ( [ londonc66 ] ) : @xmath223 and @xmath224 . the data points in fig . [ plot_sgkink ]
show the displacement field of a single defect ( obtained by adding one vortex to a commensurate chain ) for three values of the cut - off @xmath90 ( i.e. various @xmath154 ) .
we conclude that up to @xmath225 ( @xmath167 ) the s - g kink shape eq.([kinkforms ] ) forms a good description of a defect in the chain .
the data points in fig .
[ plot_fvord1d ] show numerical results for the transport of a single chain in channels of various widths and @xmath92 .
the features discussed previously , i.e. the vanishing pn barrier and nonlinear transport , clearly appear in the data for incommensurate chains .
we also plotted the results according to eq.([fvresultana ] ) , with @xmath215 evaluated using the results in app.[appb ] for @xmath226 and taking into account that @xmath114 slightly depends on @xmath1 .
the analytical treatment gives a very reasonable description of the data .
finally , we show in fig .
[ displ97ord ] the numerical results and analytical results of app.[appb ] for the quasi - static and dynamic shape of the chain for @xmath227 ( @xmath228 ) .
the numerical results closely mimic the analytic results , both for the kinked shape at small @xmath98 and the core expansion with the associated reduction of @xmath198 for large @xmath98 .
to conclude this section we mention that , at incommensurability , due to the vanishing barrier for defect motion , the average velocity @xmath229 has a vanishing ac - component . only at commensurability
the washboard modulation is retained , the velocity at large drive being @xmath230 .
we now turn to the results for channels containing multiple vortex rows and ordered ce s .
the simulations are performed with the full @xmath231d degrees of freedom and @xmath92 .
we implemented an edge shift @xmath232 with a saw tooth shape ( @xmath233 ) .
this assures that , as we vary @xmath1 , a perfect hexagonal structure is retained for @xmath73 with @xmath234 an integer . however , for @xmath235 , the qualitative behavior did not depend on @xmath52 .
figure [ widecompurefv ] shows _ v - f _ curves of commensurate channels , @xmath236 with integer @xmath237 . in these cases
the arrays are perfectly crystalline and have a shear strength @xmath238 , inversely proportional to the channel width and in accord with eq.([phenomenological ] ) with @xmath239 .
this is consistent with the fact that only the first mobile chains within a distance @xmath240 from both ce s experience the periodic edge potential ( see eq.([ordvmb ] ) ) while the other chains provide an additional pulling force via the elastic interaction .
this interaction brings an additional feature to the dynamics , namely shear waves .
the shear displacements of rows @xmath4 in the bulk of the channel can be described in continuum form , @xmath241 , by the following equation of motion : @xmath242 at large @xmath98 the ce interaction can be represented by oscillating boundary conditions . as shown in app.[appc ] , this causes an oscillatory velocity component @xmath243 with @xmath244-dependent amplitude and phase describing periodic lagging or advancing of chains with respect to each other : @xmath245 here @xmath246 is the washboard frequency @xmath247 , @xmath248 , @xmath249 .
the length scale @xmath250 explicitly depends on @xmath98 and represents the distance over which the amplitude and phase difference decay away from the ce s .
although in principle eq.([comuhprimeyt ] ) is only valid for @xmath251 , it provides useful qualitative insight in the dynamics at all velocities : at small velocity @xmath252 is large , meaning that for all rows the velocity modulation and phase become similar .
hence , for @xmath253 the array may be described as a single vortex chain , which is the underlying origin of the fact that close to threshold the curves approach the 1d commensurate behavior @xmath254 with reduced threshold @xmath255 . at large velocity , @xmath252 eventually becomes less than the row spacing . in that limit only the two chains closest to the ce experience a significant modulation . in app.[appc ] we quantitatively analyze the friction force in this regime with the result : @xmath256 in the inset of fig .
[ widecompurefv ] this behavior is displayed for @xmath257 by the dotted line . in the high velocity regime
the result agrees well with the numerical data , at lower velocities eq.([widecomfvexpression ] ) underestimates the true friction .
next we discuss the behavior of incommensurate channels .
the static vortex configuration for a channel of width @xmath258 is shown in the upper part of fig .
[ 392tria352stat ] .
a delauney triangulation shows that the array consists of 4 rows with two pairs of 5,7-fold coordinated vortices at the ce constituting two misfit dislocations of opposite burgers vector @xmath259 and glide planes along @xmath51 . due to their mutual attraction ,
dislocations at the upper and lower ce are situated along a line with an angle of @xmath260 with @xmath261 .
the two edge dislocations thus form a transverse stacking fault ( tsf ) . in the lower part of fig .
[ 392tria352stat ] the structure for a channel with @xmath262 is shown . here
the density of stacks , given by @xmath263 , is enhanced .
the dislocations at one side of the ce repel each other and are equally spaced , like the periodic superstructure for a single chain in fig .
[ displ97ord ] .
the slight misalignement between the upper and lower dislocations of a pair is due to the relative shift between the ce s : the exact orientation of the pairs is determined by the choice of @xmath52 . for channel widths in the regime @xmath264 with integer @xmath4 , we find very similar structures but instead of tsf s consisting of vacancies , we now have tsf s consisting of _ interstitial vortices _ , again arranged in a periodic superstructure .
figure [ fvwideincom ] shows the transport curves associated with these structures . as for the single chain ,
the presence of misfit defects causes an essentially vanishing threshold force . for small drive , @xmath265 ,
a low mobility regime occurs associated with glide of the edge dislocation pairs along the ce .
this allows for elastic motion of a complete tsf , i.e. the vortices in the bulk of the channel remain @xmath266-fold coordinated .
it is interesting to study how the mobility due to the tsf s changes on increasing the number of rows . in fig .
[ plot_m_tsf ] we plot the mobility per stack , @xmath267 versus channel width .
@xmath268 around each peak decreases with increasing @xmath269 .
this is caused by overlap of the strain fields of the defects , in analogy with the behavior for a single chain .
the overall increase of the peak value of @xmath268 is related to a change in the size of an isolated tsf .
an extension of the analysis in sec.[secord1d ] allows to describe this change quantitatively . for small @xmath4
the longitudinal deformations do not vary strongly over the channel width .
this can be understood by considering shear and compression deformations related by the equation @xmath270 .
it follows that a longitudinal deformation on a scale @xmath271 along the channel varies over a scale @xmath272 perpendicular to the channel . in case @xmath273 ,
the transverse variation of @xmath274 is small and can be neglected so that @xmath138 in eq.([ordconteqm ] ) can be replaced by an effective stiffness @xmath275 due to @xmath4 rows .
similarly , the driving force is replaced by @xmath276 .
this results in the same equation ( [ ordconteqm ] ) with a rescaled edge force @xmath277 .
accordingly , the longitudinal size of a defect ( tsf ) is given by @xmath278 and the mobility of an isolated tsf by ( compare @xmath279 below eq.([linearkinkresist ] ) ) : @xmath280 as shown by the drawn line in fig .
[ plot_m_tsf ] , this form gives a reasonable description of the data up to @xmath281 .
working out the condition @xmath273 given above for the validity of eq.([m_tsfformula ] ) , one obtains @xmath282 , in agreement with the data . at larger @xmath4 , bulk - mediated elasticity @xcite leads to decay of the longitudinal deformations towards the channel center .
we also note that , due to the increase of @xmath283 with @xmath4 , the density @xmath269 for which defects are non - overlapping , decreases on increasing @xmath4 . in the _ v - f _ curves of fig .
[ fvwideincom ] we observe at larger velocity features very similar to the transport of the @xmath46d chain : for @xmath284 the effective barrier is reduced , leading to core expansion of the tsf s .
accordingly , the curves approach free flux flow behavior . as in the commensurate case
, this approach is initially more slowly then @xmath285 due to additional oscillating shear deformations in the channel for @xmath286 .
figure [ fcwpure ] summarizes the behavior of the shear force @xmath287 , taken at a velocity criterion @xmath288 , versus the matching parameter . at integer @xmath236 ,
the threshold is @xmath255 , but we note that it can be reduced due to a finite edge shift @xmath52 . at mismatch
@xmath287 is essentially vanishing .
near half filling , @xmath47 , where the arrays switch from @xmath4 to @xmath10 chains , a small enhancement of @xmath287 is observed . in this regime ,
the static ( @xmath93 ) structure was obtained by annealing from a random initial configuration , attempting to determine the exact switching point .
this results in _
metastable _ structures with coexisting @xmath4 and @xmath48 row regions ( or longitudinal stacking faults , lsf s ) bordered by dislocations with misoriented burgers vector , see @xcite .
the increase in @xmath287 is caused by the finite barrier for climb - like motion of these dislocations , via which an lsf can move as giant defect through the channel . for sufficiently large drive , ( part of ) the lsf s
are annealed which may result in hysteresis for up / down cycled _
v - f _ curves .
we will discuss these mixed @xmath289 structures in more detail in secs.[secweakdis2d ] and [ secstrongdis2d ] in the context of disordered ce s .
we also note that the integer chain structures with tsf s away from half filling differ from the results in @xcite .
the structures there , obtained from a random initial configuration , contained point defects unequally distributed among rows , yielding gliding dislocations within the channel .
such structures are also slightly metastable but the conclusion of vanishing @xmath287 for incommensurate , integer chain structures , drawn in @xcite , remains unaltered .
we will now consider the influence of disorder in the ce arrays on transport in the channels , focussing in this section on the characteristics of a single chain for @xmath290 with only longitudinal degrees of freedom .
the ce disorder is implemented with longitudinal random shifts as described in sec.[secmodel ] .
we note that both ce s remain in phase ; the effect of quenched phase slips or dislocations between the ce s will be treated in the discussion in sec.[secdiscuss ] .
first we consider the form of the channel potential in presence of weak disorder . to that end
we generalize eq.([vedge ] ) in app.[appa ] and express the ce potential at @xmath291 in terms of the vortex density @xmath292 in the ce s : @xmath293 with @xmath294 the fourier transform of @xmath292 . for weak disorder ( @xmath295 ) , this density can be expressed in terms of the displacement field @xmath58 in the ce as follows @xcite : @xmath296 where @xmath297 $ ] represents the microscopic modulation due to the lattice ( @xmath298 spans the reciprocal lattice ) while @xmath299 reflects density modulations . as described in app.[appd ] , this decomposition of @xmath292 leads to two contributions to the potential : @xmath300\cos[k_0(x - d)]/k_0 , \label{1ddisordpotform}\end{aligned}\ ] ] where in the second term @xmath301 .
the term @xmath302 represents long range potential fluctuations and is smooth on the scale @xmath303 .
its correlator @xmath304 is derived in app.[appd ] . assuming that @xmath305 has short range correlations ( on the scale @xmath306 ) and a variance @xmath307 as in the simulations , @xmath308 can be written as : @xmath309 the exponent @xmath310 and the prefactor @xmath311 depend on the disorder correlations between rows in the edge : @xmath312 when the strain @xmath313 is identical for all rows and @xmath314 when the strain is uncorrelated between rows .
the term @xmath315 in eq.([1ddisordpotform ] ) is the quasi - periodic potential arising from @xmath316 of the vortex rows nearest to the ce s .
the amplitude fluctuations @xmath317 are characterized by ( see app.[appd ] ) : @xmath318 to obtain the energy of the vortex chain and the equation of motion , the vortex density inside the channel , @xmath319 , is decomposed similar to @xmath292 : @xmath320 $ ] , where @xmath134 is the displacement field of the chain . as shown in app.[appd ] , in the limit @xmath120 the resulting interaction with the ce s can be written as @xmath321 where @xmath322 $ ] represents the sine - gordon functional for an ordered channel , and @xmath323 , @xmath324 are the disorder contributions due to amplitude fluctuations and random coupling to the strain : @xmath325 the term @xmath326 contains contributions from local and nonlocal strains .
the latter dominates for @xmath327 ( see app.[appd ] ) .
hence @xmath328 . the model described by @xmath329
is also used to describe ljjs or commensurate cdw s with weak disorder , however with different disorder correlations . in the former case ,
the term @xmath323 in eq.([textformhdistotal ] ) describes local variations in the junctions critical current @xcite . for cdw s ,
a disorder contribution of the form @xmath323 arises from so called backward scattering impurities , while the term @xmath324 originates from forward scattering impurities @xcite .
we also note that our model differs from the usual fukuyama - lee - rice model for cdw s @xcite , in which commensurability is ignored either due to strong direct random coupling to @xmath134 ( @xmath330 ) or due to large mismatch . in principle , the equation of motion for the chain is given by @xmath331 .
however , it has been shown in previous studies @xcite that in the moving state a convective term @xmath332 should be included .
while irrelevant for the depinning process , such term can be important for the dynamics and for completeness we include it @xcite .
the resulting equation of motion is : @xmath333 \sin(k_{0}u ) \nonumber \\ & & -\partial_xv_s-\gamma v\partial_x u. \label{contdiseqm}\end{aligned}\ ] ] in writing eq.([contdiseqm ] ) we have assumed , for simplicity , that the elastic deformations in presence of disorder can be described by the long wavelength stiffness @xmath138 . ignoring the last term , eq.([contdiseqm ] )
describes the transverse displacements @xmath116 of an elastic string in a tilted washboard potential with random amplitude @xmath334 and random phase @xmath335 .
the latter represents a @xmath134 independent random deformation of the chain . the influence of disorder on the threshold force and the dynamics of the chain are directly visible in numerical simulations .
the simulations were performed using @xmath336 and channels of length @xmath337 .
figure [ plot_fv1dweakdis ] shows several _ v - f _
curves for channels with @xmath338 at a disorder strength @xmath339 .
we first focus on the result for the commensurate case @xmath104 .
the disorder leads to a significant reduction of the threshold @xmath287 with respect to the pure value @xmath340 .
the reduction is enhanced on increasing @xmath341 , as shown for @xmath342 in the inset .
the origin of the reduction is that disorder lowers the energy barrier for formation of vacancy / interstitial ( kink / antikink ) pairs in the chain .
figure [ plot_nuccomdisord]a shows the time evolution of the displacements @xmath343 upon a sudden increase of @xmath135 to a value @xmath344 at @xmath345 . for @xmath346 @xmath134 is flat and the 2d crystal formed by the chain and the ce s is topologically ordered . at @xmath347 ,
the motion starts at an unstable site ( at @xmath348 ) by nucleation of a vacancy / interstitial pair visible as steps of @xmath141 in @xmath134 .
we henceforth denote the force at which this local nucleation occurs by @xmath349 .
the defects are driven apart by the applied force and when their spacing becomes @xmath350 a new pair nucleates at the same site .
this process occurs periodically with rate @xmath351 , leading to the formation of a domain with defect density @xmath352 and a net velocity @xmath353 with @xmath354 the average defect velocity . in the present case of weak disorder @xmath355
is essentially the same as for @xmath356 . for a further increase of the force to @xmath357 an increase of the nucleation rate
is observed . in ref.@xcite
we showed that in larger systems coarsening occurs in the initial stage of depinning due to a distribution of unstable sites .
however , after sufficiently long times the stationary state consists of one domain around the site with the largest nucleation rate ( smallest local threshold @xmath358 ) with vacancies travelling to the left and interstitials to the right .
it is interesting to compare this to a study of cdw s with competing disorder and commensurability pinning @xcite . using a coarse grained version of eq.([contdiseqm ] )
it was found in @xcite that in the pinning dominated , low velocity regime , the so - called interface width @xmath359 grows linearly with the system size @xmath72 .
the mechanism of defect nucleation which we observe naturally explains this phenomenon .
in addition , we found that at depinning the average velocity @xmath360 can be described by @xmath361 with a depinning exponent @xmath362 , similar as previously reported for 1d periodic media @xcite .
the defective flow profile does not persist up to a arbitrary large forces . in the commensurate _ v - f _ curve in fig .
[ plot_fv1dweakdis ] and its inset , a small kink is observed for @xmath193 .
associated with this kink we find a transition to a much more ordered state .
we have illustrated the temporal evolution of vortex displacements in this state in fig .
[ plot_nuccomdisord](b ) for @xmath363 .
the staircase structure has vanished and the relative vortex displacements are greatly reduced .
in fact , in the above mentioned study ref .
@xcite a very similar transition in the cdw dynamics was found , and was shown to be of first order .
we leave the precise dependence of this transition on disorder and vortex interactions in our channels for future studies .
we now turn to the incommensurate case .
the _ v - f _ curves with @xmath364 in fig .
[ plot_fv1dweakdis ] all exhibit a finite threshold instead of the vanishing threshold for the incommensurate channels without disorder ( fig . [ plot_fvord1d ] ) . with disorder the defects that are present in the channel , couple to the disorder in the ce , which causes a pinning barrier @xmath365 .
this barrier has a distribution along the channel @xmath366 and maximum value @xmath367 .
we now focus on the curves with small defect density @xmath368 for which the defects are individually pinned . in this regime
the threshold force @xmath287 satisfies @xmath369 .
the precise threshold behavior depends on the distribution of barriers @xmath366 , similar as for ljj and cdw systems @xcite . as an illustration
, we show in fig .
[ plot_uincomweakdis](a ) the evolution of displacements for a channel with @xmath370 for a force just above threshold @xmath371 .
the static configuration at @xmath372 ( @xmath373 ) shows that the disorder breaks the periodicity of the soliton chain . for @xmath374 , depinning starts with the defect at @xmath375 and proceeds via a collision - release process between the moving defect and its pinned neighbor .
thus , for @xmath376 strong local variations in the defect mobility exist and the overall chain velocity depends strongly on the distribution @xmath366 ( of which we show an example below ) .
however , as seen in the _
v - f _ curves in fig .
[ plot_fv1dweakdis ] , for @xmath377 these effects vanish and the mobility approaches @xmath378 with @xmath279 the defect mobility without disorder .
another feature in the _ v - f _
curves for small defect densities is the velocity upturn at a force @xmath379 .
it is caused by nucleation of _ new defect pairs _ in the incommensurate chain .
the start of such a process is illustrated in fig .
[ plot_uincomweakdis](b ) : at @xmath380 the chain is unstable against pair nucleation and the nucleated interstitials / vacancies are formed on top of the moving incommensurate structure .
this process only occurs at small defect densities when the time between passage of _ existing _ defects @xmath381 exceeds the nucleation time @xmath382 . for @xmath194 ,
the structure of both defects disappears again .
the resulting dynamic state resembles that of the large velocity profile shown in fig .
[ displ97ord ] , but with additional roughness due to the weak ce disorder .
the _ v - f _ curves in fig .
[ plot_fv1dweakdis ] at @xmath383 and @xmath384 , for which the defect density in the chain is larger , exhibit a smaller threshold force . in this regime
the interaction between defects starts to become important and @xmath287 is determined by _ collective _ pinning of the defects .
this situation was studied analytically for the case of josephson vortices in a disordered ljj in @xcite .
we will not consider this situation explicitly but we note here that , as the disorder and the typical pinning force on the defects increases , the onset of the collective pinning regime shifts to larger defect density , where defect interactions are stronger @xcite . in fig .
[ plot_fs1dversuswdel ] we show the dependence of @xmath287 on channel width , both for the weak disorder regime treated above and for larger disorder .
the data at @xmath339 exhibit a sharp peak at @xmath104 , reflecting the gap between minimum nucleation threshold and maximum defect pinning force .
larger disorder however rapidly smears the peak , being eventually completely suppressed for @xmath385 .
the origin of this behavior is a spontaneous nucleation of defects in the static chain at larger disorder .
this is conveniently illustrated via the changes in the distribution of the individual defect pinning force @xmath366 and that of the nucleation threshold @xmath386 for increasing disorder strength , shown in fig .
[ plot1dpinstatistics ] .
the data were obtained by simulating hundreds of short channels ( @xmath387 ) both with one vacancy , yielding @xmath366 , and without geometric defects , yielding @xmath386 . while for @xmath339 the distributions are separated ( formally , in infinite systems , such separation only exist for bounded disorder , see the next section ) , for larger disorder they start to overlap and they become nearly identical for @xmath388 .
this implies that nucleated defect pairs at @xmath104 can remain pinned , while at incommensurability defects may be nucleated before the geometrical defects are released . in other words ,
regardless of the matching condition the _ static _ configuration contains both kinks and antikinks .
while for bounded disorder static defects first appear at a disorder strength defined as @xmath389 , the complete collapse of the peak in @xmath287 is associated with the presence of a _ finite _ density of disorder induced defects , of the order of the inverse kinkwidth @xmath390 .
we define the disorder strength at which this occurs as @xmath391 , here @xmath392 .
an example of the displacement fields for this disorder strength is shown in fig .
[ plot_del15_1d_ucomdisord ] for @xmath104 .
the lower panel shows the displacements for @xmath393 , relative to the displacements in the ce .
clearly , the static configuration has numerous defects . in general
, the approach to the critical pinned state occurs by avalanches in which local nucleation and repinning , i.e. nonpersisting nucleation events , drive the rearrangements .
the upper panel displays the evolution of displacements above threshold , revealing a growth of mountains due to persistent nucleation , superimposed on a disordered background .
the effect of large disorder on the shape of the _ v - f _ curves is shown in the inset to fig .
[ plot_fs1dversuswdel ] .
all curves now exhibit essentially linear behavior @xcite , except in a small regime just above @xmath287 .
we note that also in this disordered regime a gradual transition to a smoother displacement field occurs at larger forces , similar to the dynamic transition found for cdw s .
going back to fig .
[ plot_fs1dversuswdel ] we should also mention the overall asymmetry of @xmath287 with respect to @xmath394 and the slight decrease of @xmath287 on increasing @xmath11 at large disorder .
these effects are unrelated to the competition between commensurability pinning and disorder discussed so far , but simply reflect the overal decrease of the edge potential for larger width . using the results of sec.[subdis1deqm ]
, we now analyze in more detail the dependence of the pinning force on disorder and the vortex interaction range .
we focus on the _ average _ pinning strength of isolated defects , which we derive here in a semi - quantitative fashion ( the formal calculation is deferred to app.[appd ] ) .
our analysis applies to the case of weak disorder , i.e. we assume that the defect shape is unaffected by disorder @xcite ) .
extrapolation to larger disorder provides a useful estimate for the crossover value @xmath391 at which the commensurability peak vanishes .
we conclude the section with a summary of previous results @xcite for the threshold forces @xmath358 and @xmath367 in the special case of bounded disorder .
the disorder correction eq.([textformhdistotal ] ) to the energy of the vortex chain consists of a term @xmath323 , due to amplitude fluctuations in the periodic potential , and a term @xmath324 , due to random coupling to the strain .
we first evaluate the typical pinning energy of a defect @xmath395 due to the amplitude fluctuations .
the local fluctuations are assumed to be uncorrelated on a length scale @xmath12 , and have a variance @xmath396 .
hence , for a defect in the chain , which extends over a range @xmath147 , the resulting random potential has a variance @xmath397 . the typical pinning force on a defect is then given by @xmath398 which reduces to : @xmath399 the typical pinning energy @xmath400 of a defect due to the term @xmath324 in eq.([textformhdistotal ] ) is estimated in a similar way .
the mean square energy due to coupling of a single fluctuation in @xmath401 to the strain of a defect is @xmath402 where @xmath403 is the range of @xmath404 , given below eq.([textformhdistotal ] ) , and @xmath405 represents the strain . on the scale of a defect , there are @xmath406 such fluctuations .
thus the associated random potential for a defect has a variance @xmath407 .
taking @xmath408 and using eq.([correlator ] ) , in which case @xmath409 , yields @xmath410 ( the factor @xmath231 comes from the refined calculation in app.[appd ] ) the typical pinning force @xmath411 due to random coupling to the strain is then given by : @xmath412 the ratio between the two characteristic defect energies is @xmath413 , which shows that , particularly for increasing @xmath154 , the dominant pinning is due to random coupling to the strain .
henceforth we use only this contribution .
we next estimate the disorder strength where the commensurability peak vanishes . as mentioned before , this collapse occurs when the density of disorder induced defects becomes @xmath390 , in other words , when the typical energy gain of a defect due to disorder becomes of the same magnitude as its bare elastic energy @xmath414 .
this leads to @xmath415 .
for the particular case of random strains that are identical for all rows ( @xmath312 ) , @xmath391 is given by : @xmath416 this can be compared to the numerical data in fig .
[ plot_fs1dversuswdel ] .
even though those results were obtained for @xmath226 ( @xmath417 ) , formally outside the regime of validity of our analysis , the predicted value @xmath418 is in reasonable agreement with the data . in the particular case of _
bounded _ random strains in the ce s , the distribution of the nucleation force @xmath386 at commensurability is bounded from below by @xmath358 and that of the defect pinning force @xmath366 is bounded from above by @xmath367 ( at weak disorder ) . for completeness
we give here the previously derived results @xcite for these extremal values : both occur due to disorder fluctuations on the same length scale as that on which the displacement field @xmath116 varies . for a defect this naturally corresponds to @xmath147 .
the associated maximum defect pinning force is @xmath419 ( for uniform strains @xcite ) . for nucleation , at @xmath193 ,
the appropriate length scale is @xmath420 , the extent of a so called small amplitude nucleus @xcite . due to the nonlinearity of the pinning force , @xmath420 itself
depends on the force , i.e. @xmath421 and it diverges for @xmath422 .
as shown in detail in @xcite , this leads to a minimum nucleation threshold given by @xmath423^{4/3}$ ] . from the condition
@xmath424 one then obtains the disorder strength @xmath389 at which pinned defects can first appear spontaneously in the system : @xmath425 .
we now consider how channels of larger width , in which vortices have the @xmath231d degrees of freedom , behave in the presence of weak edge disorder .
close to commensurability , the effects we find are similar to that for a single chain . however , around half filling the importance of the transverse degrees of freedom of the channel vortices become apparent . for the commensurate case , @xmath236 , weak ce disorder causes a reduction of the threshold with respect to the ideal value @xmath426 , see the data in fig .
[ plot_fvwide_ranampi10 ] .
the reduction originates from defect formation at threshold , as illustrated for @xmath427 in fig .
[ plot_w3nuc ] . in ( a )
, three snapshots of the displacements of individual rows inside the channel are displayed .
the first snapshot is for @xmath428 and yields the flat profile .
the subsequent snapshots for @xmath429 reveal simultaneous nucleation and motion of a pair of oppositely charged tsf s , each terminated at the ce s by a pair of edge dislocations ( see fig .
[ plot_w3nuc](b ) ) . the macroscopic , stationary motion of the array is governed by periodic repetition of this process at the least stable nucleation site . in ( c ) we show the vortex trajectories during nucleation of the tsf at the left in figs .
[ plot_w3nuc](a),(b ) .
very similar images were obtained in decoration experiments at the initial stage of vl depinning in nbse@xmath430 @xcite , implying that even for weakly disordered vl s , defects may nucleate at depinning ( see also @xcite ) .
we also note that in simulations of a rapidly moving @xmath231d vortex lattice @xcite the same nucleation mechanism as in fig .
[ plot_w3nuc ] , but relative to the comoving frame , was identified as source of velocity differences between chains . for incommensurate channels close to commensurability ,
the tsf s which are caused by the mismatch are pinned by the disorder . for the weak disorder strengths considered here
the random stress from the ce s is not sufficient to break up the tsf s .
consequently , at zero drive the array in the channel consists of a weakly disordered superlattice of tsf s .
the behavior of the transport curves is shown in fig .
[ plot_fvwide_ranampi10 ] for @xmath431 and @xmath432 .
it reveals features very similar to the curves of a single chain ( fig .
[ plot_fv1dweakdis ] ) . for @xmath433 , a low mobility regime in the _
v - f _ curves develops due to the motion of the tsf s . here
@xmath287 corresponds either to the rms pinning force of individual tsf s or to the collective pinning force for larger stack density . for forces @xmath434 the curves approach linear flow behavior again . the dependence of @xmath287 on the channel width for @xmath342 is shown in fig .
[ fcweakdisversusw ] for two disorder realizations . for larger channel widths ,
smaller channel lengths were used with @xmath435 .
the data around matching ( @xmath436 ) reflect the nucleation and pinning of tsf s as discussed above : @xmath287 _ at _ commensurability is reduced compared to the pure value @xmath437 and the commensurability peak is considerably broadened , particularly for larger @xmath4 , due to the pinning of tsf s . the apparent discontinuity in the peak
may be an artifact of the finite channel length ( see the discussion in sec.[subdis1dnum ] on the distribution of nucleation sites ) . a new , and robust feature , however , is that around half filling distinct maxima in @xmath287 appear .
the origin of these maxima is illustrated by the static structure near half filling .
an example is given for @xmath438 in for fig .
[ plot_weakdis_staticcoex ] .
the triangulation shows that , in addition to aligned dislocations with @xmath439 , also misaligned dislocations appear with burgers vector at an angle of about @xmath440 with the ce s .
these misaligned defects are locally stabilized by the disorder in the ce s and thus also pinned by the disorder which leads to the increase in the threshold force .
in addition , the projection of the driving force along the glide direction is always smaller for misaligned dislocations than for aligned dislocations .
it is seen that the misaligned dislocations separate regions with @xmath4 rows from regions with @xmath10 rows .
either of these regions may thus be considered as a longitudinal stacking fault ( lsf ) . in sec.[secord2d ] we mentioned that in absence of disorder lsf s are metastable , the structure with a single integer number of rows and a regular distribution of tsf s is energetically somewhat more favorable . in presence of weak disorder ,
however , the lsf are stable and have the lowest energy . as for the dynamics at @xmath429 , the dislocation structure and flow pattern are generally different from the static pattern of lsf s .
we illustrate this in fig .
[ plot_weakdis_dynacoex ] where the vortex trajectories at various ( increasing ) driving forces are shown . for small drive ( fig .
[ plot_weakdis_dynacoex](a ) , @xmath441 ) , a region of plastic motion _ within _ the channel is seen at about the location where the static pattern shows a @xmath442 row structure .
vortex transport through these fault zones occurs by repeated nucleation and annihilation of misaligned defects . at a fixed driving force
an lsf remains at the same position , although its boundaries fluctuate over a distance of at most @xmath443 .
this contrasts the situation in absence of disorder where lsf s can move along the channel via a climb like process ( see @xcite ) . for different driving forces
, the location and amount of either @xmath4 or @xmath10-row regions or fault zones is different , as shown in figs .
( b),(c ) and ( d ) . in this particular segment
the @xmath444 region expands on increasing drive but at other locations the reverse can occur .
moreover , after cycling the force , a different structure can occur at the same drive , i.e. no unique structure exists at a given force .
this may also lead to small hysteresis in the _ v - f curves_. overall , we see that the transverse degrees of freedom in the channel , in combination with disorder , give rise to an important new mechanism of yield strength enhancement . in the following section we analyze these structures in more detail in the context of strong disorder .
the behavior of @xmath287 versus @xmath11 in fig . [ fcweakdisversusw ] still exhibits considerable discrepancies with the experimental data in fig .
[ plot_introfs ] . clearly , the ce disorder underlying these experimental data ( see @xcite ) differs from the type of weak ce disorder considered so far .
motivated by recent imaging experiments @xcite , which showed glassy vortex configurations in the nbn edge material , we now consider the case of strong ce disorder . as will be shown , in this regime the effect of transverse degrees of freedom and the presence of misaligned defects provide the main mechanism for the critical current oscillations .
the simulations we discuss here in detail were performed at a large disorder strength @xmath445 .
the only remaining order in the ce arrays is their preferred orientation with the principle lattice vector along the ce s .
the system sizes were typically such that @xmath446 .
we also allowed for quenched defects between the two ce s ( uncorrelated longitudinal strains in the upper and lower ce arrays ) , but at these large disorder strengths this is not essential . in fig .
[ plot_statstrongdistria ] we show triangulations of the static vortex configuration for channels of width @xmath447 , @xmath448 and @xmath449 . as seen in figs.(a ) and
( c ) , even in the matching case dislocations are present due to the large random stresses from the ce .
while some of these may originate from quenched phase slips between upper and lower ce s , we checked that , also without these phase slips , the matching structure at this disorder strength always exhibits twisted bonds ( defined by two adjacent dislocations of opposite charge ) at the ce s as well as oppositely charged tsf s terminated by defect pairs at the ce .
while most defects are thus situated along the ce with @xmath450 , occasionally they can be located inside the channel and have misoriented burgers vector . turning to the mismatch case in ( b ) , it is seen that a region of @xmath442 rows ( left ) coexists with a @xmath451 row region ( right ) . in between these regions
, there must be dislocations with misaligned burgers vectors .
in addition , numerous other misaligned defects are visible , rendering local destruction of the chain alignment with the ce s ( although less frequent , the latter can also occur in the matching state , see fig .
( c ) ) . to further characterize the disorder in the @xmath93 structures
, we analyzed , for the regime @xmath452 , the average length of domains without misaligned dislocations , @xmath453 , as well as the total fraction , @xmath454 , of regions with @xmath281 rows .
the results are shown in fig .
[ plot_del20statlcor ] . as observed , both quantities are maximum at @xmath447 and decay considerably away from matching : for @xmath455 the average length of correlated @xmath281 regions in the static structure is no more than @xmath456 and they make up less than @xmath457 of the channel .
the remaining fraction @xmath458 contains misaligned dislocations and small regions with @xmath231 ( @xmath459 ) or @xmath442 ( for @xmath460 ) aligned chains . in fig .
[ plot_fsversuswranampo40 ] we show the behavior of the threshold force for @xmath445 .
the modulation of @xmath287 with channel width is still present but it has changed considerably compared to the weak disorder case .
the sharp maxima at integer @xmath11 have vanished , very similar to the case of the 1d chain at strong disorder , see fig .
[ plot_fs1dversuswdel ] . instead
, we now observe smooth oscillations , with maxima in @xmath287 for @xmath461 and minima for @xmath462 .
the maxima , although occurring slightly above half filling , are of similar nature as the local maxima at @xmath463 for weak disorder : they are related to numerous misaligned defects present in the structure around mismatch .
they enhance the flow stress compared to that of the structures around matching with predominantly aligned defects . the differences in threshold force are also reflected by the vortex trajectories at small velocity .
in fig . [ plot_thresholdtrajec_del20 ] we show these trajectories for channel widths @xmath464 , @xmath465 and @xmath466 , close to the extrema in @xmath287 .
the first thing to notice is that the trajectories at mismatch ( fig.[plot_thresholdtrajec_del20](b ) , @xmath467 ) are densely interconnecting , on a scale @xmath303 , i.e. the motion is fully plastic and creation and annihilation of misaligned defects occurs over nearly the full channel length . for the near matching cases in ( a ) and ( c ) ,
the motion occurs mainly in the form of integer chains .
however , for the small velocity considered here , the dynamics still exhibits a considerable amount of plastic motion .
this partly occurs due to vortices which remain stuck at the ce s ( also visible in ( b ) ) and partly due to narrow interconnecting regions .
in fact , comparing the average length of the regions without inter row switching in ( a ) with the data for the static structure in fig . [ plot_del20statlcor ] , it is seen that near matching the structure at small @xmath98 is more disordered than the corresponding static structure , reflecting nucleation of misaligned defects in regions which were free of such defects at @xmath93 .
we now consider in more detail the properties of the moving structures at @xmath468 .
we first show in fig .
[ plot_fvdel20plusdif](a ) two characteristic _ v - f _ curves associated with a minimum ( @xmath464 ) and a maximum ( @xmath467 ) in flow stress .
it is observed that the enhanced threshold in the latter case also translates in a larger dynamic friction , @xmath469 , of the driven structure .
in addition , the latter curve exhibits a small positive curvature in the small velocity regime @xmath470 .
this nonlinearity is related to the strong plastic nature of the motion in this regime . a convenient way to characterize
the amount of plasticity is to calculate the mean square displacement of vortices form their center of mass ( m ) positions @xcite : @xmath471 ^ 2/n_{ch } , \label{eq_wydefinition}\end{aligned}\ ] ] where @xmath472 , @xmath473 denotes longitudinal and transverse displacements , respectively . as shown by kolton _
et al _ for a bulk 2d vortex lattice @xcite , @xmath474 can be characterized by @xmath475 .
for example , when @xmath476 we have normal transverse diffusion ( caused by random switching of vortices between chains ) with @xmath477 the diffusion coefficient .
however , in the channels @xmath478 will become bounded at long times ( large @xmath479 ) due to the finite channel width . in the inset ( i ) to fig .
[ plot_fvdel20plusdif](a ) we have illustrated this behavior for @xmath467 and @xmath480 . for @xmath481 ( the point @xmath482 was chosen in the steady state after transients had disappeared ) , @xmath478 increases linearly as in usual diffusion , but for larger displacements ( times ) @xmath478 levels off and eventually saturates .
in addition , even when all vortices remain in their chain ( no transverse diffusion ) , @xmath478 initially increases to a value @xmath483 due to finite chain roughness. such behavior is observed for @xmath484 in the inset ( ii ) to fig .
[ plot_fvdel20plusdif](b ) where @xmath485 is shown for a more coherent flow situation at @xmath464 and @xmath486 . in practice , we found that @xmath483 is always reached for @xmath487 , while for the long time ( large distance ) behavior significant levelling of @xmath478 occurs only when @xmath488 .
the appropriate regime we use to characterize real diffusion is therefore given by @xmath489 ( i.e. @xmath490 ) .
further , @xmath491 itself does not directly reflect the density of chain - switching points along the channel . defining the 1d density of such switching points as @xmath492 , the _ rate _ of switching increases linearly both with @xmath493 and with the average velocity : @xmath494 .
hence , the diffusion constant is given by : @xmath495 .
being interested in @xmath493 , we therefore divide out the intrinsic velocity dependence of @xmath491 and calculate @xmath496 .
the results of @xmath493 versus @xmath135 are shown in fig .
[ plot_fvdel20plusdif](b ) for the two cases @xmath464 and @xmath467 . around the minimum in flow stress ( @xmath497 ) ,
the overall value of @xmath493 is considerably smaller than around the maximum ( @xmath498 ) , similar to what is seen in the trajectories in fig .
[ plot_thresholdtrajec_del20 ] . for both cases ,
@xmath493 is clearly reduced on increasing the force .
this reflects both a decrease of the number of fault zones in the moving structure as well as a suppression of switch events ( called transverse freezing in @xcite ) within regions already organized in @xmath4 moving rows . around matching
, @xmath493 smoothly vanishes at @xmath499 , indicating complete dynamic ordering into an @xmath281 row structure without transverse wandering . for @xmath467 an ordering transition
is also observed but it occurs at much larger drive ( @xmath500 , not shown ) and the array orders into an @xmath444 row configuration , with a reduced spacing @xmath501 between the chains and average vortex spacing @xmath502 within the chains . at the end of this section we will show how in this large drive regime the number of rows changes with @xmath11 .
the insets to fig .
[ plot_fvdel20plusdif ] also show the longitudinal mean square displacements . for strongly plastic flow at mismatch ( inset ( i ) )
, @xmath503 is large and @xmath504 is close to @xmath231 , as expected when some vortices remain stuck at the ce s . for the more coherent situation in ( ii ) , where transverse switching has nearly ceased , @xmath503 is smaller and @xmath505 .
we always find an exponent @xmath506 , similar to the results for 2d vl s in @xcite .
interestingly , even without transverse wandering ( @xmath507 ) , @xmath503 increases indefinitely ( with @xmath508 ) , indicating that the moving integer chain structure still exhibits slip events and ( local ) velocity differences between the chains .
we illustrate some more aspects of the structures at large drive in fig .
[ plot_triatrajecdel20largev ] , where the vortex trajectories and triangulations of a single snapshot are displayed for @xmath464 and @xmath509 , both at @xmath510 . for @xmath464 , we do not see any transverse wandering in fig .
[ plot_triatrajecdel20largev](a ) .
figure [ plot_triatrajecdel20largev](b ) shows that in this case the dynamic structure exhibits only dislocations with @xmath439 , mainly located at the ce but also occasionally between chains inside the channel .
the latter dislocations are possibly dynamically nucleated .
they are non - stationary in the comoving frame and lead to slip events and the growth of @xmath503 as was discussed above . turning to the mismatch case ( figs .
[ plot_triatrajecdel20largev](c ) and ( d ) ) , it is seen that the dynamic structure consists of @xmath281 and @xmath444 row regions coexisting in the channel . at the driving force considered here
, @xmath511 has a finite but small value @xmath512 , which is solely due to switching of vortices in the fault zones separating the @xmath451 and @xmath442-row regions .
_ within _ these regions transverse wandering is absent . at
yet larger forces the minority @xmath451-row regions vanish and complete ordering into @xmath442 rows occurs , as for the case @xmath467 .
for a given driving force , the @xmath4-row regions again remain quasi - static during motion .
the triangulation in ( d ) exhibits the expected misaligned dislocations at the fault zone but in general also aligned dislocations are present between the chains within an @xmath4-row region ( not shown in the figure ) .
it is also interesting to compare the velocities in the two coexisting regions with @xmath4 and @xmath48 rows . denoting the vortex velocity in an @xmath4-row region by @xmath513 and the longitudinal vortex spacing there by @xmath514
, flux conservation implies that @xmath515 .
we checked that , in both regions , the average flux density @xmath516 , with @xmath517 the row spacing , was equal to @xmath518 within @xmath519 .
therefore @xmath520 , and consequently the average vortex velocities are equal , @xmath521 . however , the local _ washboard frequency _
, @xmath522 , is different in both regions .
indeed , the spectrum of the velocity fluctuations in channels with dynamic coexistence of @xmath4 and @xmath523 rows showed two shallow fundamental peaks at frequencies @xmath524 .
we however note that , both around matching ( where a single peak occurs at @xmath525 ) and around mismatch , the amplitude of the washboard peak(s ) decays on increasing the channel length @xmath72 .
in addition , for large velocities the mixed @xmath526 structures ultimately anneal into a single @xmath4 or @xmath48 domain , causing the collapse of one of the peaks .
the simulations also allow to explicitly show the influence of the transverse degrees of freedom on the modulations of the dynamic friction force ( and , ultimately , the critical current ) . generalizing the expression for the friction force eq.([dissiprate ] ) in sec .
[ subord1dfv ] including transverse fluctuations leads to : @xmath527 \nonumber \\ & & = \gamma v+f_{fric}^x+f_{fric}^y \nonumber \\ & & \equiv \gamma v + f_{fric } , \label{eq_transvdissip}\end{aligned}\ ] ] where @xmath528 and @xmath529 denote the contribution to the total friction due to longitudinal and transverse fluctuations , respectively .
figure [ plot_del20dynpropversusw](a ) displays the numerical results for these quantities , normalized by the total friction force @xmath469 as obtained from the @xmath530 curves .
the sum of the two data is @xmath46 as it should be , confirming the correct numerical evaluation of these quantities .
the total friction is also shown for clarity ( fig.(b ) ) and is essentially the same as the data shown in fig .
[ plot_fsversuswranampo40 ] .
clearly , the _ relative _ contribution of longitudinal fluctuations to @xmath531 _ decreases _ on approaching a mismatch situation , while @xmath532 increases accordingly .
these qualitative features remain present also at larger velocities , where permeation modes between chains are being suppressed . finally , we consider the continuous modulation of structural ( dis)order in the moving arrays when varying the channel width . as was shown in fig .
[ plot_fvdel20plusdif](b ) , for a fixed velocity , the density of switching points @xmath493 is maximum around mismatch while for fixed channel width it decreases with velocity .
we can then define an ordering velocity @xmath533 operationally as the point at which @xmath493 is reduced below a certain threshold . in fig .
[ plot_del20dynpropversusw](c ) the behavior of @xmath533 for channel widths @xmath534 is shown for two criteria .
the lower curve ( @xmath535 ) , with @xmath536 , corresponds to a situation with @xmath537 of the channel length transversely frozen into integer chain regions of length @xmath538 .
the upper curve ( @xmath539 ) , with @xmath540 , corresponds to nearly fully annealed arrays .
as observed , @xmath535 increases smoothly by about an order of magnitude between a matching and mismatching situation , while @xmath539 increases by a factor @xmath541 ( except for @xmath542 ) .
regardless of the criterion , the amplitude of the modulation of @xmath533 is considerably larger than the amplitude of the @xmath287 modulation , which we will further discuss shortly .
for @xmath543 the arrays all completely anneal into a single @xmath4-row structure without permeation modes .
figure [ plot_del20dynpropversusw](d ) displays the number of rows @xmath4 of these structures versus @xmath11 .
the switching from @xmath4 to @xmath544 rows is seen to occur at half integer channel widths for @xmath545 but the transitions @xmath546 and @xmath547 take place below these points , at @xmath548 and @xmath549 , respectively . around the transition points , there are regions in which @xmath4 and @xmath48 rows coexist at smaller velocity @xmath550 .
the widths of these regions were found to be @xmath551 .
the results for @xmath287 versus @xmath11 at strong disorder in the previous section show strong resemblance to the measurements of the critical current versus field @xmath13 in the experimental channel system , see fig .
[ plot_introfs ] . for a more detailed comparison ,
we combine in fig .
[ plot_sim_exp_compare ] both data , represented in terms of the parameter @xmath27 ( see sec.[secintro ] ) : for the experiment , @xmath552 is determined from @xmath553 and is plotted versus @xmath554 , for the simulations , @xmath555 ( @xcite ) . as observed , the shapes of the oscillations are in very reasonable agreement , although the overall value of @xmath27 from the numerics is larger than the experimental value @xcite .
the important conclusion from the simulations here is thus that the maxima in the measured critical current do not correspond to traditional commensurability peaks , but are caused by enhanced plastic motion and transverse deviations in mismatching channels with strong edge disorder .
previously , we indeed obtained experimental evidence for this mechanism via mode - locking ( ml ) experiments @xcite in which an rf - drive is superimposed on the dc - drive .
interference between the former and collective modes of deformation in the moving array leads to plateaus in the current - voltage curves .
the plateau ( ml ) voltage then directly yields the number of moving rows @xmath4 and transitions from @xmath556 rows were observed to coincide with maxima in critical current @xcite .
more detailed ml - experiments provided a wealth of additional information on the dynamics of the arrays @xcite .
in particular , a minimum velocity was required to observe the ml - phenomenon , which we identified as ordering velocity .
this ordering velocity exhibited a strong upturn away from matching , similar to the behavior of @xmath533 in the simulations .
as proposed in @xcite , @xmath533 can be estimated from a modified version of the dynamic ordering theory of koshelev and vinokur ( kv ) @xcite .
firstly , instead of the 2d random potential in @xcite , for the channels @xmath533 is related to the short range correlated random stress from the ce , with r.m.s .
amplitude @xmath557 and @xmath558 the random strain .
secondly , when thermal fluctuations can be ignored compared to typical defect energies ( see below ) , @xmath533 is inversely proportional to the energy @xmath559 for creation of small defect pairs ( see also @xcite ) : @xmath560 , with @xmath561 ( per unit vortex length ) .
the typical pairs referred to are those with misaligned burgers vector , which are responsible for breakup of the chain structure .
hence , the increase of @xmath533 away from matching implies a decrease in the pair formation energy , which is accounted for by including the parameter @xmath562 in @xmath563 , while setting @xmath564 at matching .
experimentally , @xmath565 decreased to @xmath566 at mismatch .
the behavior of @xmath533 and @xmath565 in the simulations is analyzed using the dimensionless form of the above formula for @xmath533 : @xmath567 the data for @xmath535 around matching ( where @xmath568 ) yield as measure for the random strain @xmath569 and , near mismatch , a reduction of the defect pair creation energy by a factor @xmath570 .
the latter is in very reasonable agreement with the experiments , and the reduction of @xmath559 near mismatch also qualitatively agrees with the large number of disorder induced fault zones observed in the static structures near mismatch ( fig.[plot_statstrongdistria ] ) .
however , the result for @xmath558 is considerably larger than the value @xmath571 found experimentally .
this is also manifest in the fact that in the experiments the pinning frequency , defined as @xmath572 , always exceeds the ordering frequency @xmath573 , while in the simulations @xmath574 . within the modified kv theory , the pinning frequency and the ordering frequency
are directly related via @xcite : @xmath575 with @xmath576 .
experimentally , @xmath577 was found to be independent of the matching condition .
the decrease in the defect pair creation energy ( @xmath578 ) then relates to the increase in the pinning frequency ( yield strength ) away from matching ( @xmath579 ) as @xmath580 .
the numerical results for @xmath565 versus @xmath27 behave similarly and can be fitted by @xmath581 with @xmath582 , but data collected over the full range of @xmath11 show to much scatter to make a more detailed comparison .
nevertheless , the kv model qualitatively accounts for the enhanced amplitude of the @xmath533 modulation compared to that of the yield strength . as for the discrepancy between the experimental and numerical values for @xmath558 or @xmath533
, one should keep in mind that experimentally the ordering velocity is determined from the onset of an @xmath4-row ml plateau , i.e. it corresponds to the velocity at which coherent @xmath4-row regions first appear , while incoherent regions may still exist in other parts of the channel . in the simulations this may thus correspond to @xmath533 determined using a larger criterion for the density of switching points @xmath511 .
in addition , the superimposed rf - drive in the experiments may assist reordering of the structure , also leading to smaller values of @xmath533 . for future studies
it is interesting to directly compare numerical simulations of channels with mixed rf - dc drive with the experiments and test which criterion best represents the reordering phenomenon .
additionally , the experimental results indicated that in the large drive regime ( where the ml amplitude saturates at a constant value ) , the coherently moving fraction of vortices does not exceed @xmath583 ( at matching ) , while it was reduced on approaching mismatch .
this feature appears at odds with the simulations where eventually the arrays all order into a completely transversely frozen @xmath4-row domain at large drive , regardless of the matching condition . at present
we do not have a good explanation for this discrepancy .
finally we comment on the @xmath49 approximation made throughout this study .
when thermal fluctuations are important , not only does one expect the dynamic ordering behavior to be changed ( see @xcite ) , also the sharp threshold behavior in the _ v - f _ curves will be smeared and activated flow may occur .
the relevant energy scale for both phenomena is again the energy for formation of small defect pairs @xmath559 .
in the low magnetic field experiments in @xcite this energy was @xmath584 orders of magnitude larger than @xmath585 and for comparison with these results the @xmath49 approximation is justified . to compare with the experiments near the melting field in @xcite
, it would be required to include thermal fluctuations in the simulations .
we have presented a detailed study of the properties of vortices confined to narrow flow channels with pinned vortices in the channel edges . in the experimental system which motivated this work @xcite ,
the threshold force ( yield strength ) shows pronounced commensurability oscillations when the natural vortex row spacing is varied through integer fractions of the channel width . the analysis and simulations presented in this paper show that in a mesoscopic channel system the dependence of threshold on commensurability as well as the dynamics of vortices in the channels drastically vary with the amount of disorder in the confining arrays . at zero or weak disorder
the system behaves similar to 1d ljj - systems and defects at the ce s reduce the yield strength . at large disorder
the behavior involves transitions from quasi-1d to 2d structures , where an increase in the amount of plastic deformations enhances the yield strength similar to the situation in the classical peak effect in superconductors .
we first presented a generalized sine - gordon description for a 1d vortex chain in an ideally ordered channel . in this case ( or for channels with multiple chains near commensurability ) the threshold force has sharp peaks at commensurate widths , whereas it is essentially vanishing at incommensurability due to easy glide of aligned defects , i.e. defects with burgers vector along the ce .
the model was then extended to study the effects of weak disorder in the confining arrays .
simulations and analytical results showed that this disorder causes the sharp maxima in the threshold force at matching to be lowered and broadened due to nucleation ( at matching ) and pinning(at mismatch ) of edge defects .
apart from these defects , the arrays respond elastically in this regime , both near threshold and at large drive .
we studied numerically the relevant edge defect dynamics and , using the sine - gordon model , we analyzed the crossover to strong defect pinning on increasing the disorder strength . for large disorder in the ce s , matching between the longitudinal vortex spacings in and outside the channel becomes irrelevant and the peaks in threshold force around matching completely vanish with a saturated value for @xmath287 of about @xmath586 of the ideal lattice strength . around mismatch however , the arrays become susceptible to formation of defects with burgers vector misoriented with the channel direction .
such defects either locally break up the integer chain structure or exist at the boundaries of @xmath4 and @xmath10-row regions coexisting in the channel . at large disorder , they are strongly pinned and cause the threshold force to exceed that around matching . approaching a matching condition , the density of misaligned defects is reduced and a smooth modulation of @xmath287 results , with minima near matching .
the depinning transition always involves plastic deformations _ inside _ the channel , but the amount of plasticity drastically increases away from matching . using the density of transverse switching points ( obtained from the transverse diffusion in the moving structures ) as dynamic order parameter , we study the evolution of the moving structures on changing the channel width and the drive .
the arrays reorder ( partially ) into transversely frozen @xmath4-row regions at a velocity @xmath533 which shows a similar modulation with commensurability as the threshold force . finally , we compared the modulations of @xmath287 and @xmath533 at strong disorder with the available experimental results and with the dynamic ordering theory in @xcite and find good qualitative agreement .
r.b . and p.k .
are supported by the nederlandse stichting voor fundamenteel onderzoek der materie ( fom ) and v.m.v .
is supported by u.s.doe , office of science under contract @xmath587w-31 - 109-eng-38 .
in this appendix we calculate the edge potential and sine - gordon parameters in a symmetric , ordered channel with @xmath588 for arbitrary field . the interaction between a vortex at @xmath589 in the channel and the pinned vortices in the ce s at @xmath590
is given by : @xmath591 where @xmath592 is the fourier transform of the vortex - vortex interaction .
to obtain an expression valid over larger range of fields than the ( low field ) london - regime , we use a generalization of the london potential eq.([vlondon ] ) as proposed by brandt @xcite .
this generalization accounts for the reduction of superfluid density with field and an additional attractive interaction due to overlapping vortex cores .
the fourier transform of this interaction reads : @xmath593 , \label{ftfullvbrandt}\end{aligned}\ ] ] where @xmath38 , @xmath594 and @xmath595 the effective coherence length with @xmath596 .
it is convenient to split eq.([vedge ] ) for the total potential in terms of the contribution @xmath597 of row @xmath204 ( see fig .
[ plot_geometry ] ) . integrating over @xmath598 and using poisson summation yields @xmath599 where @xmath113 and the prefactors are : @xmath600 \label{ordvmbprefac}\end{aligned}\ ] ] with @xmath601 , @xmath602 and @xmath603 $ ] .
first of all , we neglect in eq.([ordvmb ] ) the @xmath129 terms which represent uniform ( @xmath51-independent ) interaction . secondly ,
when @xmath604 , as practically encountered in films , @xmath605 .
then the @xmath606 terms can be neglected , resulting in a sinusoidal potential @xmath607 .
further , in the summation over @xmath204 only the contributions from the @xmath608 terms are significant .
next we employ the relation @xmath609 and rewrite @xmath610 .
the resulting expression for the total edge potential is : @xmath611\nonumber\\ & & \times 2u_0(1-b)e^{-\pi\frac{w+b_0}{a_0}}\cos{k_0 x } , \label{ordvrfinal}\end{aligned}\ ] ] where @xmath612 . for a channel with @xmath104 , the maximum @xmath613 of the sinusoidal pinning force @xmath614 at @xmath103
is then given by : @xmath615 where we used @xmath616 and @xmath617 it can be checked that the associated shear modulus @xmath618 is very similar to the interpolation formula of brandt eq.([c66 ] ) .
additionally , the edge potential eq.([ordvrfinal ] ) is harmonic for all fields .
hence the ideal flow stress of a commensurate , ordered channel , is characterized by frenkels value @xmath619 for all fields @xcite . in the low field limit @xmath620 ( and @xmath621 )
the above expressions reduce to eq.([ordvrlowfield ] ) , ( [ mudef ] ) in sec .
[ subord1dana ] . using the field - dependent vortex interaction eq.([ftfullvbrandt ] )
, one can derive the parameters in the sine - gordon description of sec .
[ subord1dana ] generalized for higher field .
the equation for the chain stiffness becomes : @xmath622 .
\label{allfieldstiffness}\end{aligned}\ ] ] the reduced stiffness is obtained from @xmath623 : @xmath624 taking into account these refinements in eq.([semidiscretefel ] ) and eq.([nonlocalordconteqm ] ) , the defect width in the nonlocal regime becomes : @xmath625 one can obtain the typical crossover field @xmath626 ( or typical @xmath627 ) at which nonlocal behavior sets in for a chain in an ordered channel by equating eq.([ldnonlocb ] ) to the s - g value for the kink width @xmath628 .
approximating @xmath629 , one finds @xmath630 , which has the approximate solution @xmath631\label{bnonloc}\end{aligned}\ ] ] hence , the nonlocal regime is absent for a channel in a material with @xmath632 ( and thickness @xmath633 ) .
for @xmath634 , nonlocal behavior occurs for @xmath635 with @xmath636 .
this is to be compared with the estimate @xmath637 resulting from a simple london interaction , see sec .
[ subord1dana ] .
for a displacement field of the form eq.([persolansatz ] ) , expressed in modes with wave vector @xmath638 and amplitude @xmath206 , eq.([fvdispl ] ) for the _ v - f _
curve attains the form : @xmath639 .
\label{coeffric}\end{aligned}\ ] ] the amplitudes @xmath206 are obtained by inserting eq.([persolansatz ] ) into the equation of motion ( [ ordconteqm ] ) with the wavelength dependent elasticity @xmath640 from eq.([kappaexpression ] ) .
since @xmath198 can become of the order of a lattice spacing @xmath12 , one expands the sin term in ( [ ordconteqm ] ) up to second order in @xmath198 .
furthermore , we assume that @xmath206 decays rapidly upon increasing @xmath204 and we keep only the three lowest order contributions @xmath206 with @xmath641 .
collecting terms of equal wave number , one obtains the following set of approximate equations : @xmath642 @xmath643 @xmath644 where @xmath645 and @xmath646 denotes the complex conjugate of @xmath198 . at small @xmath98 ,
the real components of @xmath198 vanish and the amplitudes describing the shape of the quasi - static deformations are given by : @xmath647 @xmath648 @xmath649 for arbitrary @xmath98 , the @xmath206 s in eq.([hullset1],[hullset2],[hullset3 ] ) may be determined by a mathematical program . however , the most important effect of the coupling is that above a characteristic velocity @xmath650 ( see below ) , @xmath651 and @xmath652 decrease rapidly as @xmath653 and @xmath654 , so that only the first fourier mode survives .
this mode is described by ( @xmath655 ) : @xmath656 since eq.([largevelrel ] ) can be obtained by neglecting the elastic force terms @xmath657 , it describes the deviation from the average velocity of a single particle in a periodic potential
. further analysis of eq.([hullset1],[hullset2],[hullset3 ] ) shows that the crossover velocity @xmath650 is determined by the amplitude of the quasi - static result in eq.([quasistat1],[quasistat2],[quasistat3 ] ) : @xmath658 using the definition @xmath659 we rewrite eq.([velcriterion ] ) : @xmath660 here @xmath661 can be interpreted as the effective relaxation time for the non linear deformations in the chain , which is expressed through the relaxation times @xmath662 of linear modes ( phonons ) by eq.([kappacdef ] ) .
the velocity dependence of @xmath663 may then be written as : @xmath664}. \label{endresult}\end{aligned}\ ] ] this has the correct small @xmath98 behavior ( where higher modes play a role ) and large @xmath98 behavior ( where @xmath665 ) .
first order perturbation using only @xmath666 to first order in eq.([hullset1 ] ) yields the same functional form but with @xmath667 replaced by @xmath668 , supporting the analytical interpolation made in obtaining eq.([endresult ] ) . finally , using eq.([coeffric ] ) one arrives at eq.([fvresultana ] ) in section [ subord1dfv ] .
starting from eq.([elasheareqm ] ) we write @xmath134 as the sum of a spatially uniform dc - component and a non - uniform , time - dependent component @xmath669 .
the result for @xmath670 is the following modified diffusion equation : @xmath671 where @xmath672 .
the boundary condition is set by : @xmath673 \nonumber\\ & & + ( c / b_0 ^ 2)\delta h(t ) , \label{boundarycondition}\end{aligned}\ ] ] and similarly for @xmath674 . here
the discrete character of the array near the ce s is retained in @xmath675 .
for the pinning term ( last term on the r.h.s . ) we now use the large velocity expansion for the restoring force from one ce : @xmath676\simeq ( \mu/2)\sin[\omega_0 t]$ ] . then eqs.([comhytequation]),([boundarycondition ] ) become similar to those for heat diffusion in a rod with at both ends heat sources that vary sinusoidally in time . in our case
the frequency is the washboard frequency @xmath212 . by separation of variables
one finds : @xmath677\nonumber \\ & & -\frac{(f-\gamma v)}{2c}y^2 , \label{comuytsolution}\end{aligned}\ ] ] where @xmath678 and @xmath679 with a velocity ( frequency ) dependent healing length @xmath680 .
the factor @xmath681 normalizes the displacement at the ce to @xmath682 . the latter is obtained from the boundary condition eq.([boundarycondition ] ) . in the limit @xmath683 and @xmath684 ,
@xmath682 can be approximated by : @xmath685 the solution ( [ comuytsolution ] ) describes a periodic velocity modulation @xmath686 of each chain , with a @xmath244-dependent amplitude @xmath687 and phase shift .
the latter reflects periodic lagging and advancing of the inner rows with respect to the outer ones .
both the decay of @xmath687 away from the ce and the phase shift strongly increase with velocity through @xmath252 . in fig .
[ anawa0_8 ] we have illustrated this behavior for small velocity @xmath688 ( a ) and larger velocity @xmath689 ( b ) .
although the result in ( a ) is actually out of the regime of validity of the high velocity expansion , the in - phase behavior at small @xmath98 is a feature that qualitatively agrees with the simulations . for comparison
we show in fig .
[ vytcompurew9f0110 ] numerical results for the oscillating velocity component for @xmath690 and small drive , @xmath691 ( @xmath692 ) .
in addition to the small phase shift one observes that the modulation is highly non - sinusoidal , which is not captured by our approximate solution .
based on the above , one can also evaluate the dynamic friction force @xmath693 . using eq.([dissiprate]),([fvdispl ] ) , the expression for the elastic - continuum @xmath694 would be @xmath695 ^ 2 $ ] .
however , in the velocity regime where our solution applies , the length @xmath696 .
since we have a discrete number of vortex rows , it is the two outer rows at @xmath697 which give the dominant contribution . evaluating @xmath469 using eq.([heexpression ] ) then leads to eq.([widecomfvexpression ] ) in sec .
[ secord2d ] .
in this appendix we derive the disorder corrections to the channel potential and the effect on pinning of the chain . for the vortex interaction we use the london potential of eq.([vlondon ] ) . starting from eq.([vedgeformal ] ) we first split the potential in contributions from row @xmath204 with transverse coordinate @xmath244 : @xmath698 where @xmath699 and @xmath700 in addition , @xmath701 in eq.([disordvm ] ) is related to the longitudinal displacement field @xmath702 via @xcite @xmath703e^{-ikx } , \label{rhomk}\end{aligned}\ ] ] which contains the microscopic density modulation @xmath704)$ ] as well as the density variations @xmath705 due to long wavelength deformations . inserting eq.([rhomk ] ) in eq.([disordvm ] ) yields a total potential of the form @xmath706 where @xmath707 and @xmath708 . \label{vmpstart}\end{aligned}\ ] ] a constant offset arising from the dcpart of the density has been omitted in eq.([vml ] ) and in eq.([vmpstart ] ) only the @xmath709 components of the density are taken into account .
the latter represents a quasi - periodic potential with wavelength @xmath710 .
it is only significant for @xmath711 , i.e. for @xmath608 , as in the pure case .
therefore @xmath315 can be written as @xmath712\cos[k_0(x - d)]/k_0 , \label{vpformula}\end{aligned}\ ] ] where @xmath713 and @xmath714i k d(k)e^{ikx}\nonumber\\ & & \simeq \pi \sqrt{3 } \mu \partial_x d , \label{deltamurx}\end{aligned}\ ] ] where @xmath715 and the second line is valid for @xmath716 .
the local strains thus also provide variations in the potential height .
a similar conclusion holds when transverse shifts in the ce are included . in the simulations
the mean square strain is @xmath717 and short range ( @xmath306 ) correlated along @xmath51 with correlator @xmath718 .
accordingly , the amplitude fluctuations of the periodic potential are characterized by : @xmath719 the non - local contributions eq.([vml ] ) add up to a total contribution @xmath720 in @xmath721 .
@xmath302 thus originates from strains within a region @xmath722 around the channel and will be smooth on the scale @xmath12 ( we assume @xmath723 ) . transforming the sum into an integral and using eq.([vmsk ] )
, the correlator @xmath724 can be written as @xcite : @xmath725 the term @xmath726 can be approximated by @xmath727 in case the strains are uncorrelated between the rows @xmath728 and eq.([vrcorrelatorbasic ] ) can be approximated by @xmath729 with @xmath314 and @xmath730 . in case of uniform strains , @xmath731 is independent of @xmath732
. then again the correlator is given by the above formula , but with @xmath312 and @xmath733 .
to study the effect of the disorder on the pinning of vortices inside the channel , we write the total energy as a sum of an elastic and a pinning term : @xmath734 the dispersion of the elastic constant has been neglected and @xmath134 represents the displacements of vortices inside the channel . by writing the density of the chain as @xmath735 $ ] , the pinning term @xmath736 in eq.([htotal ] ) can be expressed as : @xmath737 where only the lowest fourier components of @xmath319 are retained , i.e. @xmath738 . this expression can be simplified as follows .
since we consider the limit where @xmath739 is nearly constant within @xmath12 , the cross - term of the quasi - periodic potential @xmath315 and @xmath739 can be neglected compared to the @xmath740 term .
the product @xmath741 is also oscillatory and can be neglected as well @xcite .
the other remaining term @xmath742 can be written as @xmath743 $ ] . shifting the argument via @xmath744 and writing @xmath745 , an extra term @xmath746
is generated in @xmath747 in eq.([htotal ] ) ( and also @xmath134-independent terms which can be neglected ) .
the total coupling to the strain @xmath739 then consists of @xmath748 .
the result for the total energy is @xmath749 .
\label{approxhtotal}\end{aligned}\ ] ] the potential @xmath401 has a correlator @xmath750 which is characterized by @xmath751 in which @xmath752 and the last term arises from cross - correlations . in eq.([approxhtotal ] ) , @xmath753 $ ] is the original sine - gordon energy functional of the pure model and the remaining terms reflect the corrections due to disorder .
finally , we denote the first correction , which is due to the amplitude fluctuations , as @xmath323 and the second , coupling to the strain , as @xmath324 . for weak disorder , we can now calculate the effect of disorder on a defect in the chain by assuming that the shape of a defect at @xmath51 , @xmath754/\pi$ ] , is unchanged by disorder @xcite .
the pinning energy of a defect due to the term @xmath323 is : @xmath755 the correlations of @xmath756 are given by @xcite : @xmath757 where @xmath758 with @xmath759 . for @xmath760 we can approximate @xmath761 in eq.([ebcorrelator ] ) by @xmath762 leading to : @xmath763 the effect of the coupling to the strain
is given by @xmath764 which has the following correlator : @xmath765 using @xmath766 , @xmath767 is given by : @xmath768 for @xmath769 the final result is : @xmath770 e^{-(\frac{2s}{l_d})^2 } , \label{efcorresult}\end{aligned}\ ] ] where the term @xmath771 in square brackets is due to the nonlocal contribution @xmath302 and the factor @xmath442 arises from the term @xmath772 in @xmath401 .
hence , for large @xmath154 the nonlocal term dominates in the coupling to the strain .
w. k. kwok , u. welp , g. w. crabtree , k. g. vandervoort , r. hulscher , and j. z. liu , phys .
lett . * 64 * , 966 ( 1990 ) ; w. k. kwok , u. welp , v. m. vinokur , s. fleshler , j. downey , and g. w. crabtree , phys . rev .
67 * , 390 ( 1991 ) ; k. harada , o. kamimura , h. kasai , t. matsuda , a. tonomura , and v. v. moshchalkov , science * 274 * , 1167 ( 1996 ) . m. velez , d. jaque , j. i. martn , m. i. montero , i. k. schuller , and j. l. vicent , phys . rev .
b * 65 * , 104511 ( 2002 ) .
e. rosseel , m. van bael , m. baert , r. jonckheere , v. v. moshchalkov , and y. bruynseraede , phys .
b * 53 * , 2983 ( 1996 ) ; l. van look , e. rosseel , m. j. van bael , k. temst , v. v. moshchalkov , and y. bruynseraede , phys . rev .
b * 60 * , 6998 ( 1999 ) . c. reichhardt , c.j .
olson , and f. nori , phys .
78 * , 2648 ( 1997 ) ; c. reichhardt , c.j .
olson , and f. nori , phys .
b * 58 * , 6534 ( 1998 ) ; c. reichhardt , g. t. zimnyi and n. gronbech - jensen , phys . rev .
b * 64 * , 014501 ( 2001 ) .
a. gurevich , phys . rev .
b * 46 * , 3187 ( 1992 ) ; phys .
b * 65 * , 214531 ( 2002 ) ; a. gurevich , m. s. rzchowski , g. daniels , s. patnaik , b. m. hinaus , f. carillo , f. tafuri , and d. c. larbalestier , phys . rev . lett .
* 88 * , 097001 ( 2002 ) .
braun , yu.s .
kivshar and i.i .
zelenskaya , phys .
b. * 41 * , 7118 ( 1990 ) ; o.m .
braun and y.s .
kivshar , phys .
b * 50 * , 13388 ( 1994 ) ; s.f .
mingaleev , y.b .
gaididei , e. majernikova , and s. shpyrko , phys .
e * 61 * , 4454 ( 2000 ) .
practically , discreteness effects can be neglected for @xmath773 .
one can then also consider the regime @xmath774 .
the reduced stiffness is @xmath775 where @xmath776 and @xmath777 is obtained from the full ( @xmath39 dependent ) expression for @xmath778 in app .
[ appa ] ( eq.([ordvmb ] ) and ( [ ordvmbprefac ] ) ) .
the result is that the continuum approach is valid for @xmath779 .
for the frequently occurring geometry of a vortex chain confined by vortices pinned in a square array one can also estimate the continuum regime . in this case one
replaces @xmath780 in eq.([ordvrlowfield ] ) by @xmath781 , yielding a periodic pinning force with @xmath782 .
correspondingly @xmath783 and lattice discreteness is unimportant for @xmath784 . h. fukuyama and p.a .
lee , phys .
b * 17 * , 535 ( 1978 ) ; p.a .
lee and t.m .
rice , phys .
b * 19 * , 3970 ( 1979 ) ; an early study of the competition between impurity and commensurability pinning was reported in h. fukuyama , j. phys .
japan * 45 * , 1474 ( 1978 ) .
the convective term @xmath785 arises naturally when realizing that the driving force density on a string segment is the product of @xmath135 and the average vortex density @xmath786 ) , see also @xcite .
a detailed study of the _ v - f _ curves at strong disorder showed that the friction force @xmath469 exhibited , besides the usual decrease on increasing @xmath98 above threshold , a shallow minimum and then a weak _ increase _ when further increasing @xmath98 .
these features did not depend on system size or simulation time step and varied little with commensurability .
the observed behavior is in marked contrast with analytical and numerical studies in standard cdw models , which find a correction @xmath787 at large velocity for a @xmath46d system , see l. sneddon , m.c .
cross , and d.s .
fisher , phys .
* 49 * , 292 ( 1982 ) and also h. matsukawa , j. ph . soc .
bf 56 , 1522 ( 1987 ) ; * 57 * 3463 ( 1988 ) .
the difference could arise from the additional @xmath134 independent random force or the remaining ( relatively small ) commensurability pinning which are present in our system , but absent in the previous studies .
this issue deserves further investigation . the data in fig .
[ plot_introfs ] have been taken in a field down experiment .
qualitatively similar data were obtained in field cooled experiments and both data can be interpreted in terms of strong edge disorder .
the results of field up measurements however are qualitatively different and indicate much weaker ce disorder in this case , see n. kokubo _
, in preparation . for the numerical data
, @xmath287 was obtained from a criterion @xmath788 , i.e. @xmath789 , while in the experiments a criterion @xmath790 mhz was used for the critical current .
the latter corresponds to a dimensionless criterion @xmath791 where @xmath792 was used , with @xmath793 the normal resistivity and @xmath18 the upper critical field .
taking @xmath794 @xmath114 m , @xmath795 t for the channel system under investigation , @xmath796 t and @xmath797 @xmath798 m , yields @xmath799 .
occasional checks using the latter criterion however yielded results for @xmath287 and @xmath800 essentially the same as in figs .
[ plot_fsversuswranampo40 ] and [ plot_sim_exp_compare ] .
there are several possible explanations for the discrepancy between @xmath552 and the numerical result for @xmath27 .
the experimental system consists of channels with walls of finite hight at the ce s .
screening currents along these walls may cause the average distance between the first mobile row and the first pinned rows to be larger than that in the simulation ( where it is @xmath2 at matching ) , leading to an overall reduced value of the shear interaction .
in addition , the precise amount of disorder in the experimental system is unknown ( imaging studies only exist on related geometries and outside the relevant field regime , see b.l.t .
plourde , d.j .
van harlingen , n. saha , r. besseling , m.b.s .
hesselberth , and p.h .
kes , phys .
b * 66 * , 054529 ( 2002 ) and m.v .
marchevsky , ph.d .
thesis , leiden university ( 1997 ) ) .
further , differences in the amount of longitudinal and transverse positional disorder may exist .
finally , compared to the experiments , the simulations were performed for relatively small ratio @xmath154 . in the experiments we found that the switching point @xmath801 together with the maxima and minima in @xmath0 can be shifted to different magnetic field by changing the field history .
this implies a non - trivial relation between the effective width @xmath1 and the physical channel width @xmath802 , depending on the amount of screening currents along the channel walls , see n. kokubo , r. besseling and p.h .
kes , physica c * 412 - 414 * , 362 ( 2004 ) . | we study numerically and analytically the behavior of vortex matter in artificial flow channels confined by pinned vortices in the channel edges ( ce s ) .
the critical current density @xmath0 for channel flow is governed by the interaction with the static vortices in the ce s . motivated by early experiments which showed oscillations of @xmath0 on changing ( in)commensurability between the channel width @xmath1 and the natural vortex row spacing @xmath2 , we study structural changes associated with ( in)commensurability and their effect on @xmath0 and the dynamics .
the behavior depends crucially on the presence of disorder in the arrays in the ce s . for ordered ce s , maxima in @xmath0
occur at commensurability @xmath3 ( @xmath4 integer ) , while for @xmath5 defects _ along _ the ce s cause a vanishing @xmath0 . for weak disorder ,
the sharp peaks in @xmath0 are reduced in height and broadened via nucleation and pinning of defects .
the corresponding structures in the channels ( for zero or weak disorder ) are quasi-1d @xmath4 row configurations , which can be adequately described by a ( disordered)sine - gordon model . for larger disorder , matching between the longitudinal vortex spacings in and outside
the channel becomes irrelevant and , for @xmath6 , the shear current @xmath0 levels at @xmath7 of the value @xmath8 for the ideal commensurate lattice . around
half filling ( @xmath9 ) the disorder leads to new phenomena , namely stabilization and pinning of _ misaligned _ dislocations and coexistence of @xmath4 and @xmath10 rows in the channel . at sufficient disorder
, these quasi-2d structures cause a _ maximum _ in @xmath0 around mismatch , while @xmath0 smoothly decreases towards matching due to annealing of the misaligned regions . near threshold ,
motion inside the channel is always plastic .
we study the evolution of static and dynamic structures on changing @xmath11 , the relation between the @xmath0 modulations and transverse fluctuations in the channels and find dynamic ordering of the arrays at a velocity with a matching dependence similar to @xmath0 .
we finally compare our numerical findings at strong disorder with recent mode - locking experiments , and find good qualitative agreement . |
one of the most influential random matrix theory ( rmt ) developments of the last decade was discovery in 2002 of the so - called tridiagonal @xmath0-ensembles by dumitriu and edelman in @xcite .
the tridiagonal random matrices of this ensemble have explicit and independent ( up to symmetry ) entries and their eigenvalues are distributed according to the equilibrium joint probability density function ( jpdf ) of charged particles in a one dimensional coulomb gas with electrostatic repulsion , confined in a quadratic potential and subject to a thermal noise at temperature @xmath8 for _ arbitrary _ @xmath9 .
more precisely , the jpdf of the eigenvalues is given by @xmath10 for @xmath11 ( respectively @xmath12 ) , this jpdf arises as the joint law of the eigenvalues of the classical gaussian orthogonal ( respectively unitary , symplectic ) ensembles , whose linear eigenvalues statistics were extensively studied in the literature ( see @xcite for a review of rmt and its applications ) . the introduction of the tridiagonal random matrices for _ arbitrary _
@xmath0-ensembles has led to considerable progress for the study of linear statistics of the point process with jpdf @xmath13 .
they have permitted to prove that the largest eigenvalues converge jointly in distribution to the low - lying eigenvalues of the random schrodinger operator , also called _ stochastic airy operator _ , @xmath14 , restricted to the positive half - line , where @xmath15 is a white noise on @xmath16 @xcite ( see also @xcite for a review on the top eigenvalue statistics of random matrices ) . in this paper ,
we are interested in the limiting marginal distributions of the bottom eigenvalues of the stochastic airy operator when the parameter @xmath0 tends to @xmath1 .
as mentioned above , the stochastic airy operator appears as the continuous scaling limit of @xmath0-ensembles at the edge of the spectrum and we shall see that the question we investigate is in fact related to the largest eigenvalues statistics of @xmath0-ensembles when the @xmath0 index scales with the dimension @xmath17 such that @xmath18 when @xmath6 .
a somehow related question was investigated in @xcite ( see also @xcite ) where the authors consider the empirical eigenvalues density of @xmath0-ensembles in the limit of large dimension @xmath17 and with @xmath19 .
the limiting family of probability density @xmath20 is computed explicitly in terms of parabolic cylinder functions and is proved to interpolate continuously between the gaussian shape ( obtained for @xmath21 ) and the wigner semicircle shape ( which is recovered when @xmath22 ) .
the @xmath0-wishart ensemble was handled similarly in @xcite .
the question of the characterization of an interpolation between the tracy - widom @xmath0 distribution ( which governs the typical fluctuations of the top eigenvalue as @xmath23 with @xmath9 fixed ) and the gumbel distribution ( which governs the typical fluctuations of the maximum of independent gaussian variables corresponding to the @xmath24 case ) was raised in @xcite and @xcite .
we answer this question proving that the tracy - widom @xmath0 distribution converges weakly ( when properly rescaled and centered ) to the gumbel distribution when @xmath0 goes to @xmath1 .
this is the content of theorem [ tracywidom.gumbel ] .
we use the characterization of the marginal distributions of the low lying eigenvalues of the stochastic airy operator in terms of the explosion times process of the associated riccati diffusion @xcite .
we show that the empirical measure of the explosion times converges weakly in the space of radon measures to an inhomogeneous poisson point process on @xmath16 with explicit intensity . the weak convergence of all the marginal distributions of the second , third , etc eigenvalues can be readily deduced .
although we expect the minimal eigenvalues to have poissonian statistics in the small @xmath0 limit , the convergences of the joint distribution of the @xmath2 bottom eigenvalues for any fixed index @xmath2 seem to be difficult to prove as there is not a simple characterization of this law in terms of a single diffusion .
it is still characterized in this setting in terms of a family of coupled diffusions but the interaction between those diffusions is complex and makes the analysis difficult ( see figures [ explosion1 ] and [ couplage_diffusions ] below ) . as an application ,
we investigate ( with heuristic and numerical arguments ) the weak convergence of the top eigenvalue of @xmath4-ensembles in the double scaling limit @xmath6 and @xmath5 . we revisit the ideas of @xcite which proposes that tridiagonal random matrices of @xmath0-ensembles
are properly viewed as finite difference schemes of the stochastic airy operator . from our heuristic discussion in section [ rmt ] , this relation seems to remain valid also in the regime @xmath5 and permits to establish the weak convergence of the top eigenvalue of @xmath4 ensembles to the gumbel distribution , for any sequence @xmath4 such that @xmath25 we explicit the scaling and centering of this convergence which are in fact the same as in the convergence of the tracy - widom @xmath0 distribution to the gumbel distribution .
again , the allied results on the second , third , fourth , etc eigenvalues can also be derived from our former results on the stochastic airy ensemble ( @xmath26 for short ) .
we finally mention that our derivation does not cover the case where @xmath27 , which is highly interesting as the interpolation for the empirical spectral distribution occurs on this range of @xmath4 @xcite .
for @xmath4 decreasing as slowly as @xmath28 or even more slowly , the typical fluctuations of the top eigenvalues seem to enter the gumbel regime in the sense that the centerings and scalings are found to be the same as in the classical setting of independent gaussian variables . to facilitate the reading , let us draw up a short outline of the paper . in section [ sae.review ] , we give a brief review on stochastic airy ensemble ( @xmath29 for short ) , recalling in particular the correspondence between the law of the eigenvalues and the law of the explosion times process of the associated riccati diffusion established by @xcite . in section [ exit.time.statio ]
, we revisit the classical problem of the exit time from a domain of a diffusion which evolves in a _
stationary _ potential .
this problem is the stationary counterpart of our main study and turns to be useful in the next sections to approximate the non stationary riccati diffusion and in particular its explosion times .
we provide a simple characterization of the law of the exit time which permits to prove its weak convergence to an exponential distribution , when the trap gets very deep in comparison to the noise .
then we consider the explosion times process of the stationary riccati diffusion and we prove that it converges to a ( homogeneous ) poisson point process .
finally , we discuss in view of subsection [ macro.density ] the fokker planck equation which relates between the transition probability distribution of a diffusion and the flux of probability in the system .
in section [ sae.results ] , we state our results on the convergence of the distribution of the minimal eigenvalue of the stochastic airy operator , i.e. of the tracy - widom @xmath0 law , to the gumbel law .
this is straightforwardly deduced from the convergence of the explosion time process of the diffusion .
the convergences of the marginal distributions of the other neighboring minimal eigenvalues can be deduced as well .
the proofs of those results appear in subsection [ main.proof ] . at the end of the section
, we compute with a perturbative _ heuristic _ method the empirical eigenvalue density of the stochastic airy operator as @xmath30 on the macroscopic scale , i.e. without any zooming in the minimal eigenvalues scaling region . as an application of our results
, we discuss in section [ rmt ] the marginal statistics of the minimal eigenvalues of @xmath4-ensembles in the double scaling limit @xmath31 .
we conjecture that they can be readily deduced from our results since the tridiagonal random matrices , when zooming in the edge scaling region , are well approximated , even when @xmath5 , by the stochastic airy operator .
some technical computations and auxiliary proofs are gathered into appendices .
* acknowledgments .
* we are very grateful to j .- p .
bouchaud for many enlightening discussions .
we have shared many ideas with him and benefited from his insight along this project .
we also thank edouard brzin for his explanations on @xcite , antoine levitt and alain comtet for useful comments on schrodinger operators and for pointing out reference @xcite as well as satya n. majumdar for interesting discussions related to problems at stake in section [ rmt ] .
ra 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 at the time this work was finished .
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 .
in the first subsection , we recall the recent results on the spectral statistics of the stochastic airy operator obtained in @xcite . then , we give a characterization of the first exit time ( also called blow - up time ) of the non homogeneous _ riccati _ diffusion associated to the stochastic airy operator .
the stochastic airy operator @xmath32 is defined formally ( see ( * ? ? ?
* section 2 ) for a precise definition ) for @xmath9 and @xmath33 as @xmath34 where @xmath35 is a white noise on @xmath16 .
following @xcite , denote by @xmath36 the space of functions @xmath37 satisfying @xmath38 and @xmath39 .
we will say that @xmath40 is an eigenfunction/ eigenvalue pair for @xmath32 if @xmath41 and if @xmath42 holds for all @xmath33 in the following integration by part sense , @xmath43 ramrez , rider and virg proved in ( * ? ? ?
* theorem 1.1 ) that , almost surely , for each @xmath44 , the set of eigenvalues of @xmath32 has a well defined @xmath45 st lowest element denoted @xmath46 . furthermore the law of any eigenvalue @xmath46 with given index @xmath2 is characterized in term of the explosion times of a stochastic process @xmath47 defined through the riccati change of functions @xmath48 .
this stochastic process is a diffusion process whose initial condition and langevin equation are obtained from the dirichlet boundary condition @xmath49 ( otherwise @xmath50 is identically @xmath1 ) and the two signs of @xmath51 are equal . ] and @xmath52 where @xmath53 is a standard brownian motion .
solutions of may blow up to @xmath54 at finite times , as will happen whenever @xmath50 vanishes . in this case
, the diffusion @xmath55 immediately restarts at @xmath56 at that time in order to continue the solution corresponding to the langevin equation .
the authors of @xcite prove that the operator @xmath32 satisfies a sturm - liouville like property in the sense that _ the number of eigenvalues of @xmath32 at most @xmath57 is equal to the total number of explosions of the diffusion @xmath58 on @xmath16_. has only a finite number of explosions in @xmath16 and thus that the number of eigenvalues below @xmath57 is almost surely finite . ] before explaining the idea behind this key relation , let us preliminary state the so - called _ increasing property _ of the coupled family of diffusions @xmath59 , which will be used many times in the paper .
the increasing property is rather intuitive and can be enunciated as follows : if @xmath60 , then the number of explosions of @xmath61 is stochastically bounded above by the number of explosions of @xmath62 on any compact interval @xmath63 $ ] .
equivalently , the diffusion @xmath62 remains below @xmath61 until its first explosion time and can not cross the trajectory of @xmath55 from below to above .
intuitively this property is rather obvious since the drift of @xmath62 pulls stronger downside than the one of @xmath61 .
note however that the comparison theorem for sdes ( see ( * ? ? ?
* proposition 2.18 ) or ( * ? ? ?
* theorem ( 3.7 ) , chapter ix ) ) does not apply directly ( the drifts are not lipschitz ) and one needs to use a localization argument before applying it .
now we explain the key relation .
first , we will look at the operator @xmath64 defined on the set of functions with support in the truncated interval @xmath65 $ ] with dirichlet boundary conditions at both endpoints .
this truncated operator is shown to approximate closely the operator @xmath32 when @xmath66 in a precise topology @xcite . at
fixed @xmath67 , the definitions of the eigenfunctions and their associated riccati transform imply that @xmath68 is an eigenvalue of the operator @xmath64 _ if and only if _ the riccati diffusion @xmath61 explodes precisely at the end point @xmath67 .
we can deduce that the number of eigenvalues of the operator @xmath64 smaller than @xmath57 equals the number of real numbers @xmath69 such that @xmath70 blows up exactly in @xmath67 .
indeed , if @xmath71 slowly decreases from @xmath57 to @xmath54 , we can see from the _ increasing property _ and the continuity of the whole trajectory @xmath72 with respect to @xmath71 that the explosion waiting times of @xmath62 will simultaneously get longer and longer to finally diverge to @xmath56 when @xmath71 gets too small for the diffusion to explode in @xmath16 ( see fig .
[ explosion1 ] and [ couplage_diffusions ] ) .
in particular , any explosion of @xmath61 which occurs before time @xmath67 will slowly translate continuously to occur at some point exactly in @xmath67 for some @xmath62 with @xmath73 .
thus the number of eigenvalues of @xmath64 smaller than @xmath57 is equal to the number of explosions of @xmath55 on the interval @xmath74 $ ] .
taking @xmath66 finally establishes this property for the operator @xmath32 . driven by the same brownian motion for several values of @xmath57 .
the values of @xmath57 are between @xmath75 and @xmath76 on a grid of mesh @xmath77 .
we took @xmath78 .
the smallest eigenvalue @xmath79 of the airy operator is between @xmath75 ( no explosion ) and @xmath80 ( at least one explosion ) on this event . ] and @xmath62 driven by the same brownian motion with @xmath81 ( blue curve ) and @xmath82 ( red curve ) , for @xmath78 .
we add in dashed lines the corresponding parabolas where the drifts cancel .
when @xmath83 , the diffusions converge to the upper part of their respective parabolas . on this event , we have @xmath84 and @xmath85 . ] the marginal laws of the eigenvalues are characterized in a rather simple way in term of the distribution of the explosion times process of one single riccati diffusion @xmath55 as a function of the parameter @xmath57 .
we shall use this characterization several times throughout the paper . in particular ,
the cumulative distribution of the lowest eigenvalue @xmath86 satisfies @xmath87 = \p\left [ x_\lambda(t ) \ , \textrm{blows up to $ - \infty$ in a finite time } \right ] \,.\ ] ] and for @xmath88 , we have @xmath89 = \p\left [ x_\lambda(t ) \ , \textrm{blows up to $ - \infty$ at least $ k+1 $ times } \right ] \,.\end{aligned}\ ] ] the joint distribution of the @xmath2 bottom eigenvalues can also be characterized in this setting but in a more complicated way from the law of the family of coupled diffusions @xmath90 all satisfying with the _ same _ driving brownian motion @xmath91 .
for instance , with @xmath92 , @xmath93 \notag\\ & = \p\bigg [ x_{\lambda } \mbox { blows up to $ - \infty$ at least $ k$ times } , x_{\lambda ' } \mbox { blows up to $ - \infty$ at most $ k$ times } \bigg]\ , .
\end{aligned}\ ] ] although the two diffusions @xmath55 and @xmath62 are driven by the same brownian motion , they have different drifts and ( to our knowledge ) this makes the right hand side probability of difficult to estimate in practice . in view of (
* theorem 1.1 ) which states that the top eigenvalue of the tridiagonal matrices properly centered and scaled converges in law to @xmath94 , the tracy - widom @xmath0 law ( @xmath95 for short ) has been defined for general @xmath9 as the law of the random variable @xmath94 . using the characterization in terms of the diffusion @xmath55 ,
the first two leading terms of the right large deviation tail of the tracy - widom @xmath0 distribution were rigorously obtained by the second author and virg in @xcite .
those two first terms were also computed by forrester in @xcite using a different method .
finally , the right tail of the tracy - widom @xmath0 law was computed to all orders by borot and nadal in @xcite using heuristic arguments .
their result valid in the @xmath96 limit reads @xmath97 = \frac{\gamma\left(\frac{\beta}{2}\right)}{(4\beta)^{\frac{\beta}{2 } } 2 \pi } \lambda^{-\frac{3\beta}{4 } } \exp\left ( -\frac{2}{3 } \,\beta \ , \lambda^{3/2 } \right ) \exp\left ( \sum_{m=1}^{+\infty } \frac{\beta}{2 } r_m(\frac{2}{\beta } ) \lambda^{-3m/2 } \right),\ ] ] where the @xmath98 are ( explicit ) polynomials of degree at most @xmath99 .
it is in fact possible to obtain an analytical characterization of the law of the minimal eigenvalue @xmath86 . in (
* theorem 1.7 ) , the authors prove that the cumulative distribution of @xmath86 is the unique solution of a boundary value problem .
using similar techniques , we can prove slightly more : the _ laplace transform _ of the first explosion time is the solution of a boundary value problem as well .
see remark [ rem.lt ] for more details .
but it turns out that this boundary value problem is hard to analyse , even in the limit @xmath30 , and this approach in spite of its robustness did not permit us to prove theorem [ th.conv.explosion.times ] .
we now revisit the classical problem of the exit time from a domain of a diffusion which evolves in a stationary potential .
the _ small noise limit _ was widely studied in the literature ( see e.g. @xcite ) using large deviation theory .
the results we derive in this section are not new and hold in a general setting .
the diffusion @xmath100 considered below in the small noise limit appears in the study @xcite of the law of the ground state ( minimal eigenvalue ) of the _ hill _
s operator , defined as @xmath101 where @xmath102 is as before a white noise on the segment @xmath103 , l>0 $ ] . in this context , due to the stationarity ( absence of the linear @xmath104 term ) , we need to restrict to a finite perimeter @xmath105 and we work with dirichlet boundary conditions @xmath106 for the eigenvectors such that @xmath107 . as in the previous section
, the law of the minimal eigenvalues @xmath108 of the operator @xmath109 can be characterized in terms of the family of diffusions @xmath110 obtained through the riccati transformation and defined by @xmath111 where @xmath112 is a fixed parameter and @xmath113 and the diffusion @xmath100 immediately restarts from @xmath56 whenever an explosion occurs .
the characterization of the marginal distribution of the minimal eigenvalue with index @xmath2 now reads @xmath114 = \p\left [ y_a \ , \textrm{blows up to $ - \infty$ at least $ k+1 $ times before time $ l$ } \right ] \,.\ ] ] the forthcoming study of the exit time distribution of the diffusion @xmath100 in the small noise limit will permit us to analyze the ( marginal ) distributions of the minimal eigenvalues of the stochastic operator @xmath115 in the limit @xmath116 ( which replaces the limit @xmath30 in the airy case ) . in view of subsection [ macro.density ]
, we also provide a discussion in subsection [ transition.pdf ] on the transition probability density of the diffusion @xmath100 , and relate this transition pdf with the limiting density of state of the operator @xmath115 .
this section corresponds to the stationary counterpart of the main study of this paper .
the problem is of course easier to solve for the hill
s operator @xmath115 than for the airy operator @xmath32 thanks to the stationarity .
we will actually see later that the law of the minimal eigenvalues of the operators @xmath115 and @xmath32 are similar in the respective limits @xmath116 and @xmath3 . in this section
, @xmath117 denotes the law of the diffusion @xmath100 when @xmath118 and @xmath119:=\p[\cdot|y_a(0)=y]$ ] is the law of the diffusion @xmath100 conditionally on @xmath120 .
in particular , @xmath121 .
we are interested in the distribution of the exit time ( blowup time ) @xmath122 and in particular in its limit in law when @xmath123 .
the diffusion evolves in a potential @xmath124 which presents a local minimum in @xmath125 and a local maximum in @xmath126 .
the potential barrier @xmath127 gets very large when @xmath123 while the noise remains constant ( see figure [ fig.potential ] ) . as a function of @xmath128 . ]
if the particle starts from above the potential well ( from @xmath129 , say ) at time @xmath1 , then it has to cross a very large barrier of size @xmath130 with a deep well in @xmath131 . from kramer s theory @xcite
, we expect the exit time to be distributed according to an exponential law with parameter @xmath132 .
we give a simple proof of this result by a laplace transform method ( see also @xcite ) .
let us introduce the laplace transform of the first exit time @xmath133 of the diffusion @xmath134 with initial position @xmath128 @xmath135\,.\ ] ] the following proposition characterizes the laplace transform @xmath136 as the unique solution to a boundary value problem .
the proof of this proposition cand be found in appendix [ proof.theo ] .
[ unicity.statio ] let @xmath137 .
then the function @xmath136 defined in is the unique bounded and twice continuously differentiable solution of the boundary value problem @xmath138
satisfying the additional boundary condition @xmath139 in addition , it satisfies the fixed point equation @xmath140 this proposition permits to derive a lot of information about the behavior of the diffusion . a first natural question one could ask is : what is the probability that the diffusion @xmath141 explodes in a finite time ? the answer is straightforward from proposition [ unicity.statio ] by taking @xmath142 and noting that @xmath143 = \p_y[\zeta < + \infty ] = 1\ , .
\end{aligned}\ ] ] the mean exit time of the diffusion starting at time @xmath144 from position @xmath128 can also be computed explicitly from proposition [ unicity.statio ] .
it suffices to differentiate @xmath145 with respect to @xmath146 and then take @xmath142 .
. then transforms into a second order differential equation which can be solved explicitly .
we find the mean exit time @xmath147 starting from position @xmath128 @xmath148 for @xmath129 , this expression @xmath149 ( simply denoted as @xmath150 in the sequel ) simplifies , after two changes of variables and a further gaussian integration , in a single integral expression @xmath151 this explicit integral form for @xmath150 is convenient to determine its asymptotic when @xmath123 using the saddle point method .
this is done in appendix [ moments.exit.time ] and the estimate will be useful throughout the paper . all the moments @xmath152 $ ] for @xmath153 of the exit time @xmath133 ( when the diffusion starts from @xmath128 at time @xmath1 ) can also be derived by iterating this argument ( see appendix [ proof.theo ] ) . in particular
all the moments are finite for all starting point @xmath128 and fixed @xmath154 .
they actually satisfy @xmath152 { \;\leqslant\;}n!\ , m(a)^n$ ] .
[ rem.lt ] as mentioned in the previous paragraph , it is interesting to note that we can derive a similar theorem for the laplace transform of the first exit time of the non - stationary diffusion under consideration in the latter section @xmath61 .
indeed , let us denote by @xmath155 the first explosion time of the diffusion process @xmath156 conditioned to start at time @xmath104 in @xmath157 and let @xmath158 for @xmath137 be its laplace transform . for @xmath142 , the laplace transform
is extended by continuity @xmath159\,.\ ] ] similarly to proposition [ unicity.statio ] , we can show that the function @xmath158 is the unique solution of the following boundary value problem :
@xmath160 satisfying the additional boundary conditions @xmath161 note that this characterization is also true for @xmath142 : we recover the result of ( * ? ? ?
* theorem 1.7 ( ii ) ) which permits to find the cumulative distribution of the minimal eigenvalue thanks to eq . which rewrites as @xmath162 = \lim_{x\to + \infty } \p[\zeta(x,0 ) = + \infty]\,.\ ] ] the generalization given here has the advantage to contain the full information about the law of the first exit time .
nevertheless , it did not permit us to derive the convergence of the point process of the explosion times .
we can now establish the weak convergence when @xmath123 of the explosion waiting time @xmath133 when rescaled by its mean @xmath150 to the exponential distribution with parameter @xmath163 .
the result is valid for a large range of starting points , mainly all the points which are _ in _ the potential well ( _ i.e . above _ the local maximum of the potential ) .
[ conv.exp ] let @xmath164 such that @xmath165 .
then , @xmath166 in particular , for any @xmath167 the first blowup time to @xmath54 of the diffusion @xmath168 starting from position @xmath128 at time @xmath1 , rescaled as @xmath169 , converges weakly when @xmath123 to an exponential law with parameter @xmath163 .
we recover here the prediction of kramer s theory .
indeed the exit time ( starting from a point inside the well ) is distributed in the limit @xmath123 according to an exponential law with parameter @xmath170 , for which we have found a logarithmic equivalent @xmath171 in appendix [ moments.exit.time ] . from theorem [ conv.exp ] and the characterization of the law minimal eigenvalues of the operator @xmath115
, we get back the result due to mckean in @xcite about the fluctuations of the ground state of the hill s operator .
[ conv - ground - hill ] in the limit @xmath116 , the fluctuations of the minimal eigenvalue @xmath172 of the stochastic hill operator @xmath115 are governed by the gumbel distribution .
more precisely , we have the convergence in law as @xmath116 , @xmath173 \rightarrow e^{-x } \exp(-e^{-x } ) \ , dx\ , .
\end{aligned}\ ] ] _ proof .
_ using the asymptotic estimate for @xmath150 as @xmath123 , the convergence in law of the exit time towards the exponential distribution provided by theorem [ conv.exp ] and the characterization , it is easy to see that @xmath174 \to_{l\to + \infty } 1-e^{-e^x}\ , , \end{aligned}\ ] ] which yields the result .
although the diffusion @xmath168 blows up to @xmath54 in a finite time almost surely , we can again define the trajectory for all time @xmath33 by restarting the diffusion in @xmath56 immediately after any blow up to @xmath54 .
endowed with this new definition , we introduce the empirical measure @xmath175 of the explosion times @xmath176 ( with a further rescaling ) defined for any borel set @xmath177 of @xmath16 as @xmath178 in particular , the random variable @xmath179)$ ] is the number of explosions of the diffusion in the interval @xmath180 $ ] .
we have seen ( see theorem [ conv.exp ] ) that the sequence of the waiting time @xmath181 until the next increment for @xmath175 converges in law when @xmath123 to an exponential distribution with parameter @xmath163 .
it is therefore easy to deduce the convergence of the point process @xmath175 to a poisson point process on @xmath16 .
more precisely , we have the following result . [ poisson.statio ]
let @xmath100 the diffusion with initial position @xmath182 .
the associated explosion times point process @xmath175 converges weakly ( in the space of radon measures on @xmath16 equipped with the topology of vague convergence @xcite ) when @xmath123 to a poisson point process with intensity @xmath163 on @xmath16 .
the end of this section is devoted the proof of theorem [ poisson.statio ] .
_ proof of theorem [ poisson.statio ] .
_ we will use the useful criterion from kallenberg @xcite , which states that it is sufficient to prove that , for any finite union @xmath183 of disjoint and bounded intervals , we have the following convergences when @xmath123 , @xmath184 & \longrightarrow |i|\ , , \label{cv1 } \\
\p[\mu_a(i ) = 0 ] & \longrightarrow \exp(- |i| ) \label{cv2 } \,,\end{aligned}\ ] ] where @xmath185 denotes the length of the set @xmath183 . towards , by linearity
we just need to prove that @xmath186 \to t$ ] .
the advantage here is that the starting point of the diffusion @xmath100 at time @xmath1 is @xmath56 so that the waiting time @xmath169 until the first explosion converges weakly to an exponential distribution according to theorem [ conv.exp ] .
we have @xmath187 = \sum_{k=0}^{+\infty } \p\left[\mu_a[0,t ] { \;\geqslant\;}k\right ] = 1 + \sum_{k=1}^{+\infty } \p\left[\frac{\zeta_k}{m(a ) } { \;\leqslant\;}t\right ] \
, , \end{aligned}\ ] ] where @xmath188 is the @xmath2 th exit time of the diffusion @xmath100 started at @xmath56 .
the strong markov property implies that , for each @xmath2 , the random variable @xmath189 is independent of @xmath190 and has the same distribution as @xmath191 .
it is then easy to deduce that for any fixed @xmath2 , @xmath192 converges in law when @xmath123 to the gamma distribution @xmath193 with shape and scale parameter @xmath2 and @xmath163 [ recall @xmath193 is simply the law of a sum of @xmath2 independent exponential random variables with parameter @xmath163 ] .
one can easily prove existence of a constant @xmath195 independent of @xmath2 such that @xmath196{\;\leqslant\;}c / k^2 $ ] using for instance chebyshev s inequality since we have @xmath197{\;\leqslant\;}1 $ ] for all @xmath198 or the cramer s large deviation principle which would give a much stronger bound .
the bounded convergence theorem finally applies and gives , when @xmath123 , @xmath199 \longrightarrow 1+\sum_{k=1}^{+\infty } \p\left[\gamma(k,1 ) { \;\leqslant\;}t\right ] = t.\end{aligned}\ ] ] the second equality is proved with the same idea . for any @xmath88 , using the strong markov property for the diffusion @xmath100 , we easily prove that the first @xmath2 explosions times converge jointly in law to the first @xmath2 occurrence times @xmath200 of a poisson point process with intensity @xmath163 ( the increments are independent and each of them converges to an exponential distribution thanks to theorem [ conv.exp ] ) _ i.e. _ @xmath201 the convergence follows by proving first that we can consider only the first @xmath2 explosion times for @xmath2 large enough using a large deviation argument as above ( in order to bound the probability of having more than @xmath2 explosion times before time @xmath202 ) and then by using the convergence on the overwhelming event @xmath203
. we aim at describing also the probability density @xmath204 of the diffusion @xmath134 at time @xmath104 and other related quantities .
the main tool is the fokker planck equation which gives the evolution of @xmath204 and writes as @xmath205\ , .
\end{aligned}\ ] ] this equation takes the form of a _
continuity equation_. indeed , introducing the flux @xmath206 eq .
rewrites as @xmath207 the equilibrium of the system can be characterized by finding the stationary solution @xmath208 to the fokker planck equation which satisfies @xmath209 where @xmath210 is a constant , which does not depend on @xmath128 or @xmath104 .
we can solve the ode eq .
explicitly and find the constant @xmath210 using the additional normalization constraint @xmath211 .
we obtain @xmath212 \,,\ ] ] where @xmath213^{-1 } \,.\ ] ] here we stress that @xmath214 is precisely equal to the inverse of the expected exit time of the diffusion starting from @xmath56 @xcite . in the limit of large @xmath67 , the number @xmath215 of blow - ups to @xmath54 of the diffusion @xmath110 during the interval @xmath74 $ ] ( or equivalently using the characterization
the number of @xmath115-eigenvalues lying below the level @xmath154 ) is proportional ( at leading order ) to the inverse of the expected exit time of the diffusion , @xmath216 we recover the formula of mckean @xcite for the limiting integrated density of states of the operator @xmath115 introduced above , as @xmath217 .
the scaling order @xmath218 for the fluctuations comes from the central limit theorem .
let us finally mention that this study corresponds to the @xmath219 case of a more general model of @xmath17 interacting particles . in this context
, we expect the number of explosions per unit of time to display a different fluctuation order when @xmath17 goes to infinity because of the non - trivial interaction .
see @xcite for more details about the model and this conjecture .
we first state our results on the limiting marginal distributions of the minimal eigenvalues of the stochastic airy operator @xmath32 when @xmath3 .
then , we provide a more global information on the spectrum by computing the empirical eigenvalue density on the macroscopic scale .
let us introduce a more convenient stochastic linear operator defined for @xmath220 as @xmath221 denoting by @xmath222 its eigenvalues , it is easy to show that in law @xmath223 where @xmath224 are the eigenvalues of the operator @xmath32
. indeed , follows after the change of function @xmath225 with @xmath226 in the differential equation satisfied by the eigenfunction @xmath227 of @xmath32 .
the riccati diffusion @xmath228 associated to the stochastic linear operator @xmath229 satisfies @xmath230 where @xmath177 is a brownian motion .
the law of the diffusion @xmath231 is denoted @xmath117 and @xmath232:=\p[\cdot|z_\ell(t)=z]$ ] is the law of the diffusion @xmath233 conditionally on @xmath234 .
recall that when blowing up to @xmath54 at some time @xmath104 , the diffusion @xmath228 immediately restarts in @xmath56 at this time @xmath104 and that we have the key relation @xmath235 = \p\left [ z_\ell \mbox { blows up to $ - \infty$ at least $ k$ times in $ { \mathbb{r}}_+$ } \right ] \,.\end{aligned}\ ] ] we investigate the minimal eigenvalues statistics of the linear stochastic operator @xmath229 when @xmath30 and in particular the convergence of the marginal distributions of the bottom eigenvalues .
when @xmath3 , we can check that for any _ fixed _
@xmath236 and @xmath2 , @xmath237\to 1 $ ] . indeed , if @xmath0 tends to @xmath1 while @xmath236 is fixed , the diffusion @xmath231 defined in converges in law to the diffusion @xmath100 of the previous section for @xmath238 .
the probability for the diffusion @xmath231 to explode @xmath2 times in @xmath16 will therefore tend to the corresponding probability for the diffusion @xmath100 , which is exactly equal to @xmath163 .
recalling , we have the claim . therefore if we look for a non trivial limit in law for the eigenvalue @xmath239 when @xmath3 , the parameter @xmath236 should decrease to @xmath54 as @xmath30 and we need to determine the rescaling of @xmath236 as a function of @xmath0 .
we can actually make the guess @xmath240 using the right tail asymptotic of the tracy - widom @xmath0 distribution and the relation between the laws of @xmath241 and @xmath242 . in agreement with this heuristic derivation ,
we fix @xmath243 and set @xmath244 we shall prove that the function @xmath245 $ ] converges to a non trivial cumulative distribution function on @xmath246 when @xmath30 . similarly to the previous section ,
we first consider the empirical measure @xmath247 of the explosions times @xmath248 of the diffusion @xmath249 after a further ( well chosen ) rescaling of time .
for a borel set @xmath250 , @xmath251 finally , recall the key point : for almost all @xmath157 , the number @xmath252 of explosions in @xmath16 is equal to the number of eigenvalues smaller than @xmath253 . [ th.conv.explosion.times ] the explosion times point process @xmath247 associated to the diffusion @xmath249 converges weakly ( in the space of radon measures equipped with the topology of vague convergence @xcite ) when @xmath3 to a poisson point process with inhomogeneous intensity @xmath254 .
it readily implies the following convergence : for any @xmath255 , @xmath256 , @xmath257 = k \right ] \longrightarrow_{\beta \to 0 } \exp\left ( - e^{x } \
, \int_{t } ^{t ' } e^{-s}\ , ds\right ) \frac{\left(e^{x } \int_{t } ^{t ' } e^{-s}\ , ds\right)^k}{k!}\,.\ ] ] it might be useful to compare the time scale of the measure @xmath247 with the one used in the definition of @xmath175 . using the asymptotic formula eq .
[ asymp.ma ] given in appendix [ moments.exit.time ] , we can check that @xmath258 in such a way , the two time scales are actually the same up to exchanging the parameters @xmath154 and @xmath259 .
more generally , note that for any @xmath260 , @xmath261 the scenario described in theorem [ th.conv.explosion.times ] is therefore fairly simple .
when @xmath3 , the diffusion @xmath262 feels the evolution in time due to the linear term @xmath263 in the drift but in a rather trivial way : setting @xmath264 to work in the appropriate time scale , the explosion times process of @xmath249 is somehow the same as the one of a stationary frozen " system which evolves with time @xmath104 _ adiabatically _ such that the parameter @xmath154 in the drift of the diffusion @xmath100 evolves slowly with time @xmath104 as @xmath265 finally let us come to the initial motivation of this paper which comes from interesting questions , recently asked in the literature in @xcite , about the existences and characterizations of possible distributions in extreme value theory which would interpolate between the tracy - widom @xmath266 ( maximum of _ highly _ correlated random variables ) and the gumbel law ( maximum of _ weakly _ correlated random variables ) .
the following theorem , obtained as a straightforward application of theorem [ th.conv.explosion.times ] , establishes and describes precisely the progressive deformation of the tracy - widom @xmath0 laws into a gumbel law when @xmath30 . [ tracywidom.gumbel ] when properly rescaled and centered , the tracy - widom @xmath0 law converges weakly to the gumbel law .
more precisely , when @xmath3 , the following convergence in law holds @xmath267\longrightarrow e^{-x } \exp(-e^{-x } ) \ , dx\,.\ ] ] comparing theorem [ tracywidom.gumbel ] to corollary [ conv - ground - hill ] , we see that the fluctuations of the ground state of the stochastic hill and airy operators are very similar , governed by the gumbel distribution , with the same scalings under the relation @xmath268 .
the convergences of the marginal distributions of the other minimal eigenvalues with general index @xmath2 can also be deduced from theorem [ th.conv.explosion.times ] which implies that @xmath269 = \p\left[\nu_\beta[0;+\infty ] { \;\geqslant\;}k+1\right ] \rightarrow_{\beta\to 0 } 1- e^{-e^{x } } \ , \sum_{i=0}^k \frac{(e^{x})^i}{i!}\ , .
\end{aligned}\ ] ] we can check that the function of @xmath157 in the right hand side is indeed a cumulative distribution function and the weak convergence of @xmath239 follows .
note also that the right hand side of corresponds exactly to the cumulative distribution of the point with index @xmath2 of a poisson point process on @xmath246 with intensity @xmath270 .
it is therefore natural to conjecture that the empirical measure @xmath271 of the eigenvalues of @xmath229 defined , for any @xmath272 borel set of @xmath246 , as @xmath273 converges weakly when considered on the _ microscopic scaling region of the minimal energies _ as in this paragraph to a poisson point process on @xmath246 with intensity @xmath274 , when @xmath30 .
let us finally consider the matching between the microscopic ( where one zooms in the measure @xmath275 on the microscopic region of the bottom eigenvalues ) and the macroscopic regime where the sets @xmath276 remains fixed as @xmath30 .
in section [ macro.density ] , we compute asymptotically when @xmath3 the empirical spectral measure on the macroscopic scale .
our result reads @xmath277 where @xmath278 is a constant of order @xmath163 as @xmath3 .
in particular , @xmath279-\infty;\ell])$ ] is the number of eigenvalues below the level @xmath236 . using the result obtained in appendix [ consistency.pt ] ,
where we check that the second order term @xmath278 in remains negligible compared to the leading order when @xmath30 and @xmath280-\infty ; \ell_\beta(x)]$] we see that for any fixed @xmath157 , @xmath281-\infty;\ell_\beta(x)]\right ) \sim_{\beta\to 0 } e^{x}\ , , \end{aligned}\ ] ] where we have also used the asymptotic .
therefore , if one considers a poisson point process @xmath282 on @xmath246 with intensity @xmath283 , then the probability that the @xmath284 bottom points of @xmath282 are below the level @xmath253 converges when @xmath3 as @xmath285-\infty ; \ell_\beta(x)]\right ) { \;\geqslant\;}k+1 \right ] \longrightarrow 1- e^{-e^{x } } \ , \sum_{i=0}^k \frac{(e^{x})^i}{i!}\ , .
\end{aligned}\ ] ] the matching between the two microscopic and macroscopic regimes , if the poisson point process convergence be true , would be smooth .
let us mention two articles @xcite from the literature on random schrodinger operators ( with _ stationary _ potential ) which establish that the bottom eigenvalues have poissonian statistics in the limit of infinite support for different kind of potential , including sums of dirac masses and smooth functions of diffusions .
let us fix @xmath286 and denote simply by @xmath287 the diffusion @xmath249 in this proof . from kallenberg s theorem @xcite
, we just need to see that , for any finite union @xmath183 of disjoint and bounded intervals , we have when @xmath3 , @xmath288 & \longrightarrow e^x \int_i e^{-t}\ , dt \
, , \label{cv1bis } \\
\p[\nu_\beta(i ) = 0 ] & \longrightarrow \exp(- e^x \int_i e^{-t}\ , dt ) \label{cv2bis } \ , .
\end{aligned}\ ] ] denote by @xmath289 $ ] the right most interval of @xmath183 , by @xmath290 the union of disjoint and bounded intervals such that @xmath291 $ ] and by @xmath292 the supremum of @xmath290 .
note that @xmath293 . to simplify notations , set @xmath294 for @xmath295 ( the times @xmath296 will correspond to the real time scale for the diffusion @xmath287 ) .
thanks to the linearity of the expectation , it is enough to prove for intervals @xmath183 of the form @xmath297 $ ] . on the other hand for , the simple markov property yields @xmath298 = \p\left[1_{\nu_\beta(j ) = 0 } \p_{z(s_0 ) , s_0 } \left[\nu_\beta[t_1;t_2]=0 \right ] \right ] \ , .
\end{aligned}\ ] ] showing therefore reduces to see that with probability going to @xmath163 as @xmath3 , @xmath299 = 0\right ] \longrightarrow _ { \beta\to 0 } \exp(- e^x \int_{t_1}^{t_2 } e^{-t}\ , dt ) \,.\ ] ] for both and , the idea is to decompose the interval @xmath300 $ ] ( with @xmath301 for ) into a finite number of small intervals of length @xmath302 and to approximate the number of explosions of @xmath287 on each small interval of the subdivision by those of stationary diffusions , thanks to the _ increasing property_. we will use in fact a random subdivision instead of a deterministic one to avoid technical issues due to special points . to this end , let us define a sequence @xmath303 of i.i.d .
random variables with uniform law in @xmath304 $ ] , independent of the diffusion @xmath287 .
let @xmath305 small enough such that @xmath306 .
then , we construct iteratively the sequence of random times @xmath307 such that @xmath308 , , we simply set @xmath309 . ]
@xmath310 for @xmath311 and the stopping times defined as : @xmath312 in the following @xmath313 refers to expectation with respect to both @xmath287 and the @xmath314 .
set @xmath315 such that the drift of @xmath287 at time @xmath316 simply equals @xmath317 . on each interval
@xmath318 $ ] of the random subdivision , we define the following two diffusions @xmath319 and @xmath320 ( independent of the times @xmath314 ) driven by the same brownian motion as @xmath287 . @xmath321 \,,\\ & m_k(s_k ) = z(s_k ) \ , , \quad \mbox{and } \quad dm_k(s ) = \left ( a + \frac{\beta}{4 } s_{k+1 } - m_k^2(s)\right ) ds + db_s \mbox { for } s\in [ s_k;s_{k+1}]\ , . \end{aligned}\ ] ] the _ increasing property _ implies that the number of explosions @xmath322 $ ] of the diffusion @xmath287 is stochastically dominated from above ( respectively from below ) by the number of explosions of the diffusion @xmath319 ( respectively @xmath320 ) . indeed , the drifts of the diffusion @xmath323 and @xmath324 are in increasing order for @xmath325 $ ] @xmath326 the drifts of the diffusions @xmath319 and @xmath320 do not depend on time @xmath316 so that the previous section applies on each @xmath327 $ ] .
let us introduce the so - called downside region @xmath328 and its complementary upside one @xmath329 we define as @xmath330 the choice @xmath331 is large enough for our purposes ( it tends to infinity ) and small enough such that the process will spend little time in the region @xmath328 ( see lemma [ lem.la_t ] ) . from the previous section , we know
that _ as long as the starting point @xmath332 belongs to the region @xmath329 _ ( which indeed is contained in the interval where the convergence applies ) , the two respective explosions point processes of the diffusions @xmath319 and @xmath320 converge weakly in the space of radon measure , when the time scale is renormalized respectively by @xmath333 and @xmath334 , to poisson point processes with intensity @xmath163 , independently of the exact location of @xmath332
. therefore we need to prove that @xmath335 has a probability going to @xmath163 when @xmath336 .
we first prove that the occupation time of the region @xmath328 by the particle @xmath287 on the region of times considered tends to @xmath1 in expectation .
this is the content of the following lemma whose proof is deferred at the end of this section .
[ lem.la_t ] fix @xmath337 and as before @xmath338,i=1,2 $ ] .
define @xmath339 then , for any @xmath340 there exists @xmath341 and a positive constant @xmath342 independent of @xmath0 such that for all @xmath9 , @xmath343 { \;\leqslant\;}c \ , \big(\ln \frac{1}{\beta}\big)^{-\eta } \,.\end{aligned}\ ] ] we are now ready to prove that @xmath344 . indeed , using the notations introduced above , the probability of @xmath345 is easily bounded from above by @xmath346 { \;\leqslant\;}\sum_{k=0}^{4(t_2-t_1)/{\varepsilon } } { \p}_{z(s_0 ) , s_0 } \left[z(s_k ) \in d \right]+ { \p}\big[\sum_{k=1}^{4(t_2-t_1)/{\varepsilon } } \tau_k { \;\leqslant\;}\frac{t_2-t_1}{{\varepsilon } } \big]\ , .
\label{union } \end{aligned}\ ] ] the second probability on the rhs of is easily bounded by noting simply that the @xmath314 have mean @xmath347 so that the empirical sum of the @xmath314 should be of order @xmath348 . from large deviation theory , we know that this probability is of order @xmath349 where @xmath350 is a positive constant independent of @xmath351 .
thanks to the independence between the diffusion @xmath287 and the sequence @xmath352 , the sum in can be bounded as @xmath353&= \sum_{k=1}^{4(t_2-t_1)/{\varepsilon } } { \e}_{z(s_0 ) , s_0}[\int_{s_{k-1}}^{s_{k-1}+\delta } 1_{\{z(\tau ) \in d \}}\ , \frac { d\tau}{\delta } ] \\ & { \;\leqslant\;}2\beta ( \frac{3}{8}\ln\frac{1}{\beta})^{1/3}\varepsilon^{-1 } \ , { \e}_{z(s_0 ) , s_0}\left[\int_{s_0}^{s_{4(t_2-t_1)/{\varepsilon}}}1_{\{z(\tau ) \in d\ } } \ , d\tau \right]\ , .
\end{aligned}\ ] ] using again the cramer s theorem of large deviation theory , we can show that , with probability going to @xmath163 when @xmath354 , the empirical sum @xmath355 is smaller than @xmath356 .
is arbitrary . ]
lemma [ lem.la_t ] , applied for fixed times @xmath357 such that @xmath358\subset [ s_1';s_2']$ ] and for @xmath359 , permits to conclude that there exist @xmath360 and two positive constants @xmath350 and @xmath195 both independent of @xmath361 such that @xmath362 { \;\leqslant\;}c \beta \ , ( \ln \frac{1}{\beta})^{1/3-\eta } + \ , e^{-c/{\varepsilon}}\ , .
\end{aligned}\ ] ] we are now ready to prove and . to simplify notations , set @xmath363 , @xmath364 and @xmath365 for the number of explosions of the diffusions @xmath366 and @xmath319 in the interval @xmath327 $ ] .
recall that a.s .
for all @xmath2 , @xmath367 we begin with .
similarly to the stationary case , we could take @xmath301 in this case for which the situation is slightly easier . indeed , at the starting point @xmath301 , @xmath368 and we do not need to introduce an independent random time just before @xmath369 to start from a nice position .
nevertheless , we treat the case @xmath370 to have consistent notations with the ones required in the proof of .
thanks to the increasing property , we can bound from above the mean number of explosions of @xmath287 as @xmath371 { \;\leqslant\;}\e\left[\sum_{k=0}^{k_2 - 1}\e\left [ n_k^+ \left| z(s_k)\in u\right.\right ] \right]+ \e[\nu_\beta[t_1,t_2 ] 1_{\mathcal{c}^c}]\ , .
\label{upper.bound.expect}\end{aligned}\ ] ] the second expectation in is negligible . to show this
, we use the cauchy schwartz inequality and the increasing property which permits to bound the second moment of @xmath372 $ ] by the second moment of the number of explosions in the interval @xmath373 $ ] of the diffusion @xmath374 studied in section [ exit.time.statio ] .
it is easily seen that this second moment is bounded independently of @xmath154 thanks to cramer s theorem for example as was done to bound the expectation of the number of explosions in the proof of theorem [ poisson.statio ] . gathering the above arguments
, we can deduce the convergence of the first term .
indeed , conditionally on @xmath375 , the explosion times process of @xmath319 converges when time is renormalized by @xmath376 , to a poisson point process with intensity @xmath163 .
here we must pay attention to the renormalization of time : noting that @xmath377 we can find the limit of the time scales ratio @xmath378 therefore , when @xmath3 , we have @xmath379 \longrightarrow \e[\tau_{k+1 } \exp(x - t_1-{\varepsilon}(-\tau_0 + \sum_{i=1}^{k } \tau_i ) ) ] \ , , \end{aligned}\ ] ] which finally gives taking @xmath354 and thanks to the convergence of the riemann sum associated to the subdivision @xmath380 that @xmath381 { \;\leqslant\;}e^{x } \int_{t_1}^{t_2 } e^{-u}\ , du \,.\end{aligned}\ ] ] for the lower bound , @xmath382= \e\left[\sum_{k=0}^{\infty}\e\left [ n_k^- \ , \ , 1_{\{s_1{\;\leqslant\;}s_k { \;\leqslant\;}s_{k+1 } { \;\leqslant\;}s_2\ } } 1_{\ { z(s_k)\in u \ } } \right ] \right ] \\ & { \;\geqslant\;}\e\left[\sum_{k=0}^{4(t_1-t_0)/{\varepsilon}}\e\left [ n_k^- \ , \ , 1_{\{s_1{\;\leqslant\;}s_k { \;\leqslant\;}s_{k+1 } { \;\leqslant\;}s_2\ } } \left| z(s_k)\in u\right.\right ] \right]\\ & - \e\left[\sum_{k=0}^{4(t_1-t_0)/{\varepsilon } } \e\left [ n_k^- \ , \
, 1_{\{s_1{\;\leqslant\;}s_k { \;\leqslant\;}s_{k+1 } { \;\leqslant\;}s_2\}}\left| z(s_k)\in u\right.\right ] \p\left[z(s_k)\in d\right]\right ] \,.\end{aligned}\ ] ] it suffices to show that the second term is negligible and apply the same method as was done above to obtain .
the second term is bounded from below by @xmath383\right ] \times \sup_{0{\;\leqslant\;}k { \;\leqslant\;}\frac{4(t_1-t_0)}{{\varepsilon } } } \frac{\p[z(s_k)\in d ] } { 1- \p[z(s_k)\in d ] } \ , .
\label{le.sup}\end{aligned}\ ] ] lemma [ lem.la_t ] applied for @xmath384 such that @xmath385 and @xmath386 gives a uniform ( independent of @xmath387 ) bound of @xmath388 $ ] which goes to @xmath1 as @xmath3 .
it easily follows that the quantity tends to @xmath1 ( we have already seen that the first term is bounded independently of @xmath154 ) .
we now turn to the convergence . by the increasing property
, we have @xmath389 = 0\right ] \\ & { \;\leqslant\;}{\p}_{z(s_0 ) , s_0}\left [ \prod_{k = k_1}^{k_2 - 2 } \p_{z(s_k ) , s_k } \left [ n_k^-=0 \left| z(s_k ) \in u , ( \tau_k)_{k\in { \mathbb{n } } } \right .
\right]\right ] + { \p}_{z(s_0 ) , s_0}[\mathcal{c}^c]\,.\end{aligned}\ ] ] using the previous arguments , we get @xmath390 = 0\right ] { \;\leqslant\;}{\e}\left [ \prod_{k = k_1}^{k_2 - 2 } \exp\left(- { \varepsilon}\tau_{k+1 } \exp(x - t_1-{\varepsilon}(-\tau_0 + \sum_{i=1}^{k+1 } \tau_i ) ) \right ) \right ] + e^{-c/{\varepsilon}}\ , .
\end{aligned}\ ] ] taking the limit @xmath354 , we obtain as before @xmath391 = 0\right ] { \;\leqslant\;}\exp(- e^{x } \int_{t_1}^{t_2 } e^{-u } \ , du ) \ , .
\end{aligned}\ ] ] for the lower bound , we use again the increasing property which allows to write @xmath392 = 0\right ] \\ { \;\geqslant\;}&{\p}_{z(s_0 ) , s_0}\left [ \prod_{k=0}^{k_2 - 1 } \p_{z(s_k ) , s_k } \left[n_k^+=0 \left| z(s_k ) \in u , ( \tau_k)_{k\in { \mathbb{n } } } \right .
\right]\right ] - { \p}_{z(s_0 ) , s_0}[\mathcal{c}^c]\,.\end{aligned}\ ] ] using the same arguments as above , we can deduce that @xmath393 = 0\right ] { \;\geqslant\;}\exp(- e^{x } \int_{t_1}^{t_2 } e^{-u } \ , du ) \
, , \end{aligned}\ ] ] and is proved .
_ proof of lemma [ lem.la_t ] .
_ the key estimate to prove this lemma is given by ( * ? ? ?
* proposition 10 ) , which is recalled in appendix [ proof.paper.laure ] .
let us recall it here .
we denote by @xmath394 the waiting time of the first explosion after @xmath395 .
there exists a constant @xmath396 independent of @xmath0 such that for all @xmath397$]$ ] indeed is contained in the region where lemma [ lemma.laure ] applies . ]
@xmath398 { \;\geqslant\;}\exp(-c\sqrt{\ln a})\,.\ ] ] we can rewrite the inequality in the following way , using the asymptotic expansion of @xmath154 as a function of @xmath0 : @xmath399 { \;\geqslant\;}\exp(-c\ , ( \ln \frac{1}{\beta})^{1/3})\,.\ ] ] eq . gives a lower bound of the probability for the diffusion @xmath287 starting from the right most point of the interval @xmath328 at time @xmath395 to explode to @xmath54 in a short time . for the sake of completeness ,
we rewrite the proof of the estimate eq . in appendix [ proof.paper.laure ] with more details than originally given in @xcite .
the idea to obtain the lemma is to use to relate between the time spent in the region @xmath328 by the process @xmath287 and the number of its explosions which is bounded on the interval @xmath400 $ ] . in the following ,
the probability @xmath117 is with respect to the diffusion starting from @xmath401 at time @xmath402 . from
, we can control for any given @xmath403 , the probability that the diffusion @xmath287 is in the region @xmath328 at time @xmath395 : indeed , we have @xmath404 = \p\left[z(u ) \in d , \zeta^{(u ) } { \;\leqslant\;}c\,(\ln \frac{1}{\beta})^{-\eta } \right ] + \p\left[z(u ) \in d , \zeta^{(u ) } > c\,(\ln \frac{1}{\beta})^{-\eta } \right]\notag \\ & { \;\leqslant\;}\p\left[z(u ) \in d,\zeta^{(u ) } { \;\leqslant\;}c\,(\ln \frac{1}{\beta})^{-\eta } \right ] + \left(1- \exp(-c\ , ( \ln \frac{1}{\beta})^{1/3 } ) \right ) \ , \p\left[z(u ) \in d \right ] \label{trick.proba1}\end{aligned}\ ] ] where we have used the simple markov property and the increasing property in the second line .
. readily rewrites as @xmath405 { \;\leqslant\;}\exp(c\ , ( \ln \frac{1}{\beta})^{1/3 } ) \ , \p\left[z(u ) \in d , \zeta^{(u ) } { \;\leqslant\;}c\,(\ln \frac{1}{\beta})^{-\eta } \right]\notag\\ & { \;\leqslant\;}\exp(c\ , ( \ln \frac{1}{\beta})^{1/3})\ , \p\left[\mbox{the interval } \left[u , u + c\,(\ln \frac{1}{\beta})^{-\eta } \right ] \label{trick.proba2 } \mbox { contains at least one explosion}\right]\,.\end{aligned}\ ] ] note that , denoting by @xmath406 $ ] the ( random ) number of explosions in the interval @xmath407 and as before by @xmath408 the explosion times , we have almost surely @xmath409 \ } } du { \;\leqslant\;}c\,(\ln \frac{1}{\beta})^{-\eta } \ , \nu_\beta[t'_1,t'_2]\,.\ ] ] therefore , integrating eq . with respect to @xmath395 in the interval @xmath407
, we finally obtain the inequality @xmath410 { \;\leqslant\;}c\,(\ln \frac{1}{\beta})^{-\eta } \ , \exp(c\ , ( \ln \frac{1}{\beta})^{1/3 } ) \ , \ , \e\left[\nu_\beta[t'_1,t'_2 ] \right]\ , . \end{aligned}\ ] ] in addition we easily check , using the increasing property and the convergence proved in the previous section , that the mean number of explosions @xmath411 $ ] is bounded independently of @xmath0 .
the lemma follows . in this subsection
, we derive the empirical spectral measure @xmath275 defined in of the stochastic linear operator @xmath229 on the macroscopic scale , i.e. _ without zooming _ in the minimal eigenvalues scaling region .
as mentioned above , we expect the number @xmath279-\infty;\ell])$ ] , which is almost surely finite for any @xmath9 ( see @xcite ) , to tend to @xmath56 when @xmath3 .
the minimal eigenvalues are indeed going to @xmath54 as @xmath412 and if one sets @xmath413 abruptly so that the linear confining term disappears , then the operator @xmath414 has an infinite number of eigenvalues below any given level @xmath415 ( see subsection [ exit.time.statio ] ) .
the spectral statistics of this operator have in fact been extensively studied in the literature ( see @xcite ) . for this study ,
the eigenfunctions are restricted to a finite interval @xmath74 $ ] with dirichlet boundary conditions in @xmath144 and @xmath416 so that the number of eigenvalues below a certain level remains finite .
the authors investigate the spectrum of @xmath417 in the large @xmath67 limit .
the minimal eigenvalues statistics of the operator @xmath229 when @xmath30 as described above are different from the statistics found in @xcite .
we shall see that the limiting empirical eigenvalue density differs as well ( to be compared with the density found in @xcite ) . to compute @xmath279-\infty;\ell])$ ] , we use again the fact that the number of @xmath229-eigenvalues strictly less than @xmath236 equals the number of explosions of the diffusion @xmath418 ( defined in eq . ) on @xmath16 .
counting the blow - ups along the trajectory @xmath418 can be done computing the flux of particles at @xmath419 in the system , i.e. the number of particles going through @xmath54 per unit of time . in this non stationary system ,
the flux depends on time @xmath104 and on position @xmath401 and will be denoted @xmath420 in the following .
as the particle immediately restarts in @xmath421 when blowing up to @xmath54 , the flux in @xmath419 is equal to the flux in @xmath421 .
we can again write the fokker planck equation which gives the evolution of the transition probability density @xmath422 of the diffusion @xmath423 as @xmath424 \,.\ ] ] this equality can in turn be rewritten as a continuity equation @xmath425 where @xmath426 is the flux of probability in @xmath401 at time @xmath104 . by definition of the flux ,
the number of explosions to @xmath54 of the diffusion @xmath418 on @xmath16 is @xmath427-\infty;\ell])= \int_0^{+\infty } j_\beta(-\infty , t ) \ , dt\,.\ ] ]
re scaling time as before by setting @xmath428 , eq . becomes @xmath429 \,.\ ] ] we now use perturbation theory in the limit @xmath430 in eq . in order to obtain an approximation for @xmath431 for small @xmath0 .
the method consists in searching a solution of eq .
valid at small @xmath0 under the form @xmath432 at leading order , we find the following ordinary differential equation for @xmath433 @xmath434 equation is the same as eq . of the previous section with the parameter @xmath154 replaced by @xmath435 .
the solution is @xmath436 where @xmath210 is the stationary flux given by eq .
such that @xmath437 .
hence , recalling that @xmath438 is the law of @xmath439 , we conclude that the flux in @xmath54 at time @xmath104 is @xmath440 and thus , @xmath441-\infty;\ell ] ) = \int_0^{+\infty } j_\beta\left(-\infty , t \right ) \ , dt = \frac{4}{\beta } \int_{-\infty}^{\ell } j_0\left(u\right ) \ , du + o(1 ) \,,\ ] ] where @xmath278 is a correction of order @xmath163 when @xmath3 .
it is now straightforward to deduce the empirical eigenvalue density @xmath442 this formula contrasts with the result found in @xcite where the empirical eigenvalue density of the operator @xmath417 restricted on a finite interval @xmath74 $ ] is proportional to the length @xmath67 of the time interval @xmath443 the computation leading to this result was in fact recalled above in .
it is interesting to study the behaviors at @xmath444 of the empirical eigenvalue density @xmath445 and of the integrated density @xmath279-\infty;\ell])$ ] .
this can be done from the integral forms with the saddle point route .
when @xmath446 , we have ( see again end of appendix [ moments.exit.time ] ) @xmath447-\infty;\ell ] ) \sim \frac{1}{\pi\beta } \exp\left(-\frac{8}{3}|\ell|^{3/2}\right)\,.\ ] ] on the other side , when @xmath448 , we obtain @xmath449-\infty;\ell ] ) \sim \frac{8}{3\pi\beta } \ , \ell^{3/2}\,.\ ] ] we note the correct matching of the tails of the empirical eigenvalue density @xmath445 as @xmath450 with those of the crossover density of eigenvalues ( in random matrix models ) near the edge scaling region of width @xmath451 found by bowick and brzin in @xcite ( see figure [ density.edge.scaling ] ) . as a function of @xmath57 . blue oscillatory curve : density @xmath452 at the edge of the spectrum for @xmath453 found by bowick and brzin .
the bumps of the @xmath453 curve are reminiscent to the electrostatic repulsion between the particles . ]
we discuss the validity of perturbation theory as applied here in appendix [ consistency.pt ] .
in this subsection , we use the connection between random matrices and stochastic linear operators to study the top eigenvalue statistics of @xmath0-ensembles with an index @xmath4 depending of @xmath17 such that @xmath454 example of such a sequence is @xmath455 where @xmath456 .
the tridiagonal random matrices introduced by dumitriu and edelman in @xcite , whose eigenvalues are distributed according to the jpdf @xmath13 defined in , can be written as @xmath457 where the @xmath458 are independent gaussian random variables with variance @xmath163 and where the @xmath459 are independent @xmath460 distributed random variables with @xmath461 degrees of freedom and scale parameter @xmath462 . in @xcite ,
the authors prove that the largest eigenvalues of those tridiagonal random matrices converge in distribution to the low lying eigenvalues of the stochastic airy operator @xmath32 introduced above .
their proof is rather technical , but a simple heuristic of this convergence can be found in @xcite . in the following
, we apply this heuristic to analyse the present case with a vanishing repulsion coefficient @xmath4 satisfying .
the main idea is that , under the assumption , the information on the top eigenvalues of the matrix @xmath463 is contained in the upper left sub matrices of @xmath463 . in order to analyze the law of those eigenvalues , it is sufficient to consider only the entries with column or line numbers very small compared to the dimension @xmath17 of the matrix .
developing for @xmath464 the @xmath460 variables as @xmath465 where the @xmath466 are independent gaussian variables , we obtain the following decomposition for the translated and rescaled matrix @xmath467 using the same argument as in @xcite , may be rewritten in short as @xmath468 in view of the previous section , it is important to keep the linear term @xmath469 ( even for @xmath470 ) so that the least eigenvalue of the operator on the right hand side of remains finite .
this term is small for small values of @xmath104 but gets large for values of @xmath471 and should not neglected .
the top eigenvalue @xmath472 of the matrix @xmath463 should therefore approximately satisfy , for large values of @xmath17 , the equality in law @xmath473 where @xmath474 is the least eigenvalue of the operator @xmath475 . using the previous results , eq .
can be rewritten @xmath476 where @xmath477 is a random variable distributed according to the gumbel distribution .
this result was checked numerically with very good agreement . from this proposal that the tridiagonal random matrices in the @xmath0 ensembles may be regarded at the edge of the spectrum as finite different schemes of the stochastic airy operator ,
we argue that the results proved in the continuous setting for the stochastic airy operator can be extended to the discrete setting of random matrices .
the convergence we obtained for the marginal distribution of any eigenvalue with fixed index @xmath2 in can be rewritten in the context of @xmath4 ensembles as was done in for the maximal eigenvalue . at this point , it is tempting to conjecture , as in the continuous setting of the stochastic airy operator , that the maximal eigenvalues of @xmath4-ensembles have poissonian statistics in the double scaling limit @xmath478
. nevertheless the transition between wigner/ airy and poisson statistics is not very well known up to now and many questions have been asked in the literature of rmt ( see e.g. @xcite for related questions on eigenvectors localization / delocalization ) .
the discussion of the previous subsection breaks down if @xmath0 scales with @xmath17 as @xmath480 where @xmath350 is a positive constant . in such cases
( and for even more rapidly decreasing @xmath4 ) , the lower part of the matrix @xmath463 is no longer negligible compared to the upper part .
one has to keep all the entries in the matrix @xmath463 and the approach through the stochastic linear operator is no longer valid . as mentioned in the introduction , the case @xmath19
was studied before . in @xcite ,
the authors prove that , if @xmath0 scales with @xmath17 as @xmath19 where @xmath350 is a positive constant , then the empirical eigenvalue density @xmath481 of the matrix @xmath463 converges in the large @xmath17 limit to a continuous probability density @xmath482 with non compact support .
the density @xmath482 is computed explicitly in @xcite and one can easily recover the gaussian tails of @xmath482 for @xmath483 @xmath484 where @xmath342 is an explicit multiplicative constant . the reader may wonder whether eq . is coherent with the results found in @xcite where the limiting empirical eigenvalue density of the matrix @xmath463 was computed in the double scaling limit @xmath485 with @xmath486 .
the developments of the previous subsection should a priori apply for any sequence @xmath4 such that @xmath5 and @xmath487 as @xmath6 . in particular , it should hold for @xmath488 .
for such a @xmath4 , eq .
rewrites under the form @xmath489 where @xmath490 are ( explicit ) constant ( their values are irrelevant in the present discussion ) .
this scaling form matches the one would find for the maximum of @xmath17 particles allocated according to the density @xmath482 derived in @xcite when @xmath485 .
we therefore conjecture that , even though @xmath472 has gumbel fluctuations for any sequences @xmath4 going to @xmath1 in the large @xmath17 limit , it has different scalings depending on whether * @xmath4 decreases faster than @xmath28 , then the centering and scaling in the convergence of @xmath472 are as in .
* @xmath4 decreases slower than @xmath28 , then the centering and scaling are as in eq . .
_ proof of proposition [ unicity.statio ] .
_ it is classical to show that the function @xmath136 defined in eq .
is a weak solution to the boundary value problem eq . and @xcite .
note simply that the differential equation is @xmath491 where @xmath492 is the infinitesimal generator associated to the diffusion process @xmath168 .
the boundary condition is satisfied by @xmath136 as defined in eq . since the diffusion ,
when starting from @xmath128 , will explode in a time going to @xmath1 when @xmath493 . at this point
, we do not know yet that the function @xmath136 is a @xmath494 function .
nevertheless , we can ( at least formally ) perform the change of function @xmath495 g_\alpha'(y)$ ] and show that the function @xmath136 satisfies the fixed point equation .
this computation can be made rigorous by considering regularizations of @xmath136 and taking the limit in the end .
it immediately follows from eq . that the function @xmath136 is @xmath496 .
we can now prove that there exists a unique bounded @xmath494 function @xmath136 satisfying the boundary value problem specified by eq . and .
if @xmath497 is another solution , it is straightforward to check from it s formula that the process @xmath498 is a martingale . besides
, it is bounded and hence we can apply the stopping time theorem which yields @xmath499 = \e_y[e^{-\alpha \zeta } ] = g_\alpha(y)\,,\ ] ] where we have used the boundary condition at @xmath500 satisfied by @xmath497 as well as the fact that @xmath501 almost surely . the unicity is proved .
_ proof of theorem [ conv.exp ] .
_ first note that @xmath502 is non increasing with respect to @xmath128 .
in addition we have @xmath503 and thus @xmath504 admits a limit when @xmath505 . we will denote this limit @xmath506 . from theorem [ unicity.statio ] ,
@xmath502 satisfies for any @xmath507 the fixed point equation @xmath508 let us define recursively a sequence @xmath509 such that @xmath510 and for @xmath511 @xmath512 to begin note that , for @xmath513 , and for all @xmath514 , @xmath515 { \;\leqslant\;}1\ ] ] using the upper bound of eq . , we obtain from eq .
a new lower bound @xmath516 using now the lower bound of eq .
, we obtain from eq . a new upper bound @xmath517 we can check iteratively that for all @xmath17 , @xmath518 in particular , we have by letting @xmath6 that , for any @xmath519 , @xmath520 note that permits to deduce the values of all the moments @xmath152= n ! \ , m(a)^n \ , r_n(y , a ) $ ] .
now it suffices to prove that for all fixed @xmath521 and for any @xmath522 , @xmath509 converges to @xmath163 when @xmath123 .
this is the content of the following lemma [ lemmern ] whose proof can be found at the end of this section .
[ lemmern ] let @xmath523 such that @xmath524 .
then , for each @xmath521 and for any @xmath167 , @xmath509 converges to @xmath163 when @xmath123 .
it is then straightforward to see that @xmath525 this convergence holds for any @xmath507 .
the theorem follows since @xmath526 is the laplace transform of an exponential law with parameter @xmath163 .
_ proof of lemma [ lemmern ] .
_ first note that for each @xmath527 , @xmath509 is increasing with respect to @xmath128 and uniformly bounded by @xmath163 .
this can be seen easily by induction over @xmath527 recalling the expression of @xmath528 given in .
it follows that the @xmath529 exists and is denoted as usual @xmath530 .
we proceed by induction over @xmath527 to show that @xmath509 converges to @xmath163 for any @xmath527 and for any @xmath167 . by monotonicity ,
it is enough to prove that both @xmath531 and @xmath530 converges to @xmath163 when @xmath123 . by the definitions of @xmath532 ( eq .
for @xmath533 ) and of @xmath150 ( see eq . ) , it is clear that , for @xmath522 fixed and @xmath534 .
moreover , by monotonicity arguments , we easily check that this convergence holds uniformly on the interval @xmath535 . at step
@xmath527 , performing two changes of variables in the recursive integral expression eq . of @xmath509
, we obtain for all @xmath536 , @xmath537 by monotonicity again , we simply need to consider @xmath538 . from the second line and provided that @xmath539 ( which is insured by assumption ) , we see that this integral is concentrated in the regions @xmath540 .
one can in fact show that the other regions bring negligible contributions . in this regime @xmath541
and we have , by the induction hypothesis , @xmath542 uniformly for @xmath543 , \epsilon > 0 $ ] small .
the convergence of @xmath531 and @xmath530 follow from the previous arguments and the asymptotic expression for @xmath150 given in .
setting @xmath238 to work with a positive parameter and writing @xmath544 instead of @xmath231 , recall the definition of the diffusion @xmath544 , @xmath545 recall also that @xmath546 refers to the law of the diffusion @xmath544 starting from @xmath401 at time @xmath104 .
we denote in this paragraph by @xmath547 the first blowup time after time @xmath395 of @xmath544 .
proposition 10 of @xcite yields that [ lemma.laure ] fix @xmath548 .
there exists a positive constant @xmath350 independent of @xmath154 and @xmath549 $ ] such that for all @xmath154 large enough and for all @xmath550 @xmath551 { \;\geqslant\;}\exp(-c\sqrt{\ln a})\ , . \end{aligned}\ ] ] the constant in front of @xmath552 on the lhs is not optimal and can be replaced by any value strictly greater than @xmath553 . for @xmath554 , set @xmath555 and let @xmath556 and @xmath557 . without loss of generality we can suppose that @xmath558 . moreover , to simplify the proof , we will restrict to the case @xmath559 .
the other cases follow the same lines ( the only change to make is the value of @xmath272 introduced below which should be replaced by @xmath560 ) . using the strong markov property and the increasing property @xcite ,
we can lower bound the left hand side of by @xmath561 \times \p_{-\sqrt{a}- \delta , \frac{1}{\sqrt{a}}}\left [ t_{-\sqrt{a}-{\varepsilon } } < \frac{\ln a}{8 \sqrt{a } } \wedge t_{-\sqrt{a}}\right ] \\
\times \p_{-\sqrt{a}- { \varepsilon } , \frac{\ln a}{8\sqrt{a}}}\left [ t_{-\infty } < \frac{3}{8 } \frac{\ln a}{\sqrt{a } } \right]\ , .
\end{aligned}\ ] ] @xmath562 the first probability gives the main cost . using the comparison theorem for sde ( see ( * ? ? ?
* proposition 2.18 ) or ( * ? ? ?
* theorem ( 3.7 ) , chapter ix ) ) , we see that the process @xmath544 starting from @xmath563 and until the stopping time @xmath564 is stochastically dominated for @xmath154 large enough by the drifted brownian motion @xmath565 .
more precisely , this means that for all @xmath566 , we have a.s .
@xmath567 therefore , using the fact that @xmath568 for all @xmath33 , we can lower bound the first probability of by @xmath569\,.\end{aligned}\ ] ] by the reflection principle and @xmath33 , @xmath570 = \p[b_t { \;\geqslant\;}2a - b]$ ] . ] and brownian scaling , this later probability is @xmath571 -\p [ b(1 ) { \;\geqslant\;}7 ( \ln a)^{1/4}]&= \p[-7 ( \ln a)^{1/4 } { \;\leqslant\;}b(1 ) { \;\leqslant\;}-5 ( \ln a)^{1/4 } ] \\ & { \;\geqslant\;}c ' \ , \exp(-\frac{25}{2}\ , \sqrt{\ln a})\ , , \end{aligned}\ ] ] where the last inequality holds for @xmath154 large enough and for a positive constant @xmath572 .
@xmath562 for the second part , using again the comparison theorem for sde ( * ? ? ?
* proposition 2.18 ) , we can see that the process @xmath544 starting from @xmath573 is almost surely below the brownian motion for @xmath574 $ ] .
more precisely , @xmath575 for all @xmath576 $ ] . therefore the second probability is bounded from below by @xmath577\ , .
\end{aligned}\ ] ] again from reflexion principle and brownian scaling property , this probability is equal to @xmath578 = _ { a\to + \infty } o ( \frac{1}{(\ln a)^{1/4}})\ , .
\end{aligned}\ ] ] this handles the second probability of which does not contribute to the main cost as it is much larger than the first probability .
@xmath562 for the last probability , the idea of @xcite is to compare the diffusion @xmath544 with the solution of an ( random ) autonomous differential equation which can be computed explicitly .
set @xmath579 denote @xmath580 for @xmath581 , we have @xmath582 the function @xmath583 satisfies the following ( random ) first order differential equation for @xmath584 @xmath585 now consider the event @xmath586 under which we are able control the solution @xmath587 and its explosion time .
note that this event happens with a positive probability @xmath588 independent of @xmath154 .
moreover , under this event , we have @xmath589 so that the inequality is valid for any @xmath590 .
thus , under the event @xmath591 and for @xmath592 , we easily check that @xmath593 introduce now the function @xmath594 solution of the ( random ) autonomous first order differential equation @xmath595 with @xmath596 it is easy to check ( with the argument used to prove the increasing property ) that , under the event @xmath591 , the function @xmath597 dominates the function @xmath587 a.s . for all @xmath592 , @xmath598 but eq .
can be integrated with respect to @xmath104 leading to @xmath599 where @xmath342 can be computed explicitly from the initial condition in @xmath144 . from this expression , we see that the function @xmath597 explodes after a finite time @xmath600 which can be computed explicitly and bounded as @xmath601 using and , we deduce that under the event @xmath591 , the diffusion @xmath544 explodes in a time smaller than @xmath602 . the lemma is proved .
with two consecutive change of variables , we obtain for @xmath536 , @xmath603 with a more precise analysis of the integral , we can in fact check that @xmath604 then it is plain to deduce that , with @xmath605 , @xmath606 and @xmath607 differentiating @xmath214 with respect to @xmath154 , we obtain for @xmath536 @xmath608
perturbation theory was used to obtain approximation at leading order of the solution of the partial differential equation eq . .
we compute the @xmath278 correction to the leading order in the expansion eq . of @xmath279-\infty;\ell])$ ] and show that , for all values of @xmath236 , perturbation theory leads to a consistent expansion at small @xmath0 of @xmath279-\infty;\ell])$ ] such that the @xmath278 correction is much smaller than the leading term of order @xmath609 .
in particular this holds for @xmath610 scaling with @xmath0 as @xmath611 when @xmath3 .
this was not obvious in the first place and needed to be checked in particular for @xmath612 in the scaling region of the minimal eigenvalues at small @xmath0 .
the @xmath278 correction denoted @xmath613 such that @xmath614-\infty;\ell ] ) = \frac{4}{\beta}\int_{-\infty}^\ell j_0(u)\ , du + \gamma_{\beta}^1(\ell ) + o(\beta ) \,.\ ] ] can be obtained by computing first the linear correction @xmath615 to the flux @xmath616 .
the flux @xmath617 is related to the function @xmath618 introduced in eq . and satisfies @xmath619 .
we thus need to compute @xmath618
. this can be done by identifying the linear terms in @xmath0 on both sides of equation , we obtain the following ordinary differential equation for @xmath620 @xmath621 = \frac{d}{dt } p_0^{\ell - t}(z)\ , .
\end{aligned}\ ] ] by a further integration with respect to @xmath401 , eq . becomes @xmath622 where @xmath623 is a constant which does not depend on @xmath401 . as mentioned above , we are interested only in @xmath624 , which can be computed easily using the normalization constraint @xmath625 . is the coefficient of the linear correction in @xmath0 to the probability density @xmath431 : it should not bring mass in the integral .
] we find @xmath626 we can finally deduce the correction @xmath613 as a function of the flux @xmath623 @xmath627 leading us to @xmath628 a careful analysis of this integral ( see appendix [ appendixb ] ) permits to extract the asymptotic behavior of @xmath613 when @xmath629 @xmath630 using this estimate , we can check that the asymptotic range @xmath253 for @xmath631 ( going to @xmath54 when @xmath3 ) of the minimal eigenvalue @xmath241 satisfies @xmath632 thus we see that the correction @xmath613 remains negligible compared to @xmath279-\infty;\ell])$ ] in the region @xmath633 . if @xmath634 , then @xmath635 in the limit @xmath636 , we can determine the leading order of this integral with the saddle point method .
we have to compute the maximum of the function @xmath637 on the domain of integration @xmath638 .
it is easily seen that it is reached for @xmath639 , @xmath640 , @xmath641 and is equal to @xmath642 .
let us perform the following change of variables @xmath643 up to small corrections , when @xmath644 , the integral is equivalent to @xmath645 two contributions have to be accounted for . the variables @xmath646 and @xmath647 being fixed of order @xmath163 , we look separately at the integration over @xmath648 with @xmath649 and then with @xmath650 .
@xmath651 integrating now with respect to @xmath646 and @xmath647 leads to a contribution to the total integral @xmath652 of order @xmath653 the region @xmath650 has to be treated differently : here we have to keep the correction term @xmath654 in the exponential which prevents the integral to be infinite .
@xmath655 hence the region @xmath650 leads to the same contribution as the region @xmath649 .
we finally have @xmath656 next we need to analyse the integral @xmath657 up to small corrections , the integral @xmath658 takes the form @xmath659 the integral in the second line can be analysed following the previous lines .
we check that the mass in the integral is carried this time exclusively by the region @xmath650 .
we obtain finally @xmath660 20 i. dumitriu and a. edelman . _ matrix models for beta ensembles _ j. math
. phys . * 43 * , 5830 - 5847 ( 2002 ) .
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 . , see theorem 9.2 .
_ random matrices _
( elsevier , new york , 2004 ) .
_ log gases and random matrices _ ( princeton university press , princeton , 2010 ) .
g. akemann , j. baik , and ph .
di francesco . _ the oxford handbook of random matrix theory _ ( oxford university press , new york , 2011 ) .
a. edelman and b. d. sutton .
_ from random matrices to stochastic operators_. j. stat
* 127 * 6 , 1121 - 1165 ( 2007 ) . j. a. ramrez , b. rider and b. virg . _
beta ensembles , stochastic airy spectrum , and a diffusion_. j. amer . math .
* 24 * 919 - 944 ( 2011 ) .
s. n. majumdar and g. schehr .
_ top eigenvalue of a random matrix : large deviations and third order phase transition_. j. stat . mech .
* 1 * 12 ( 2014 ) .
r. allez , j .- p .
bouchaud and a. guionnet .
_ invariant @xmath0-ensembles and the gauss - wigner crossover_. phys .
* 109 * , 094102 ( 2012 ) .
r. allez and a. guionnet .
_ a diffusive matrix model for invariant @xmath662-ensembles_. electron
. j. probab . * 18 * 62 , 1 - 30 ( 2013 ) .
r. allez , j .-
bouchaud , s. n. majumdar , p. vivo . _ invariant @xmath662-wishart ensembles , crossover densities and asymptotic corrections to the marchenko - pastur law_. j. phys . a : math
* 46 * 015001 ( 2013 ) .
k. johansson .
_ from gumbel to tracy - widom_. prob .
fields * 138 * 75 - 112 ( 2007 ) .
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 ) . c. texier .
_ individual energy level distributions for one - dimensional diagonal and off - diagonal disorder_. j. phys .
a : math . gen . * 33 * 6095 ( 2000 ) .
l. dumaz and b. virg . _ the right tail exponent of the tracy - widom - beta distribution_. ann .
h. poincar probab .
* 49 * , 4 , 915 - 933 , ( 2013 ) .
g. borot and c. nadal .
_ right tail expansion of tracy - widom beta laws_. random matrices : theory appl . * 1 * 1250006 ( 2012 ) .
p. j. forrester .
_ spectral density asymptotics for gaussian and laguerre ? - ensembles in the exponentially small region_. j. phys .
a math . gen .
* 45 * 075206 ( 2012 ) .
a. bloemendal and b. virag .
_ limits of spiked random matrices i_. probab .
theory relat .
fields ( online first ) ( 2012 ) .
m. freidlin and a. wentzell .
_ random perturbations of dynamical systems , 3rd edition_. grundlehren der mathematischen wissenschaften , springer ( 2012 ) .
a. dembo and o. zeitouni .
_ large deviations techniques and applications ( second edition ) _ , springer ( 1998 ) . s. varadhan .
_ large deviations and applications_. cole dt de probabilits de saint - flour , springer , academic press , new york , london ; akademie - verlag , berlin ( 1986 ) .
h. l. frisch and s. p. lloyd .
_ electron levels in a one - dimensional random lattice_. phys
. rev . * 120 * 1175 - 1189 ( 1960 ) .
b. i. halperin .
_ green s functions for a particle in a one - dimensional random potential_. phys
139 * a 104-a117 ( 1965 ) .
h. p. mckean .
_ a limit law for the ground state of hill s equation_. j. stat
. phys . * 74 * ,
1227 ( 1994 ) . m. h. holmes .
_ introduction to perturbation methods_. texts appl . math . ,
* 20 * , springer - verlag , new - york , ( 1998 ) .
p. hanggi , p. talkner and m. borkonec .
_ reaction - rate theory : fifty years after kramers_. rev .
phys . * 62 * 251 - 341 ( 1990 ) .
o. kallenberg .
_ random measures , 4th edition_. academic press , new york , london ; akademie - verlag , berlin ( 1986 ) .
r. allez and l. dumaz .
_ random matrices in non - confining potentials_. arxiv:1404.5265 ( 2014 )
. s. a. molcanov . _ the local structure of the spectrum of the one - dimensional schrdinger operator_. commun .
phys . * 78*,429 - 446 ( 1981 ) .
l. n. grenkova , s. a. molcanov , and ju .
n. sudarev .
_ on the basic states of one - dimensional disordered structures_. commun
. math . phys . * 90*,101 - 123 ( 1983 ) .
bouchaud and p. cizeau .
_ theory of lvy matrices_. phys .
e * 3 * , 1810 - 1822 ( 1994 ) . c. bordenave and a. guionnet .
_ localization and delocalization of eigenvectors for heavy - tailed random matrices_. probab .
theory relat .
fields ( online first ) ( 2013 ) v. bapst . _ the large connectivity limit of the anderson model on tree graphs_. arxiv:1303.4908 ( 2013 ) .
g. biroli , a. c. ribeiro - teixeira and m. tarzia .
_ difference between level statistics , ergodicity and localization transitions on the bethe lattice _ arxiv:1211.7334 ( 2013 ) .
p. j. forrester .
_ the spectrum edge of random matrix ensembles_. nucl .
b * 402 * , 709 - 728 ( 1993 ) .
k. it and h. p. mckean .
_ diffusion processes and their sample paths_. springer
. i. karatzas and s. e. shreve .
_ brownian motion and stochastic calculus , second edition_. graduate texts in mathematics , springer .
d. revuz and m. yor .
_ continuous martingales and brownian motion , third edition _ , springer . | we investigate the marginal distribution of the bottom eigenvalues of the stochastic airy operator when the inverse temperature @xmath0 tends to @xmath1 . we prove that the minimal eigenvalue , whose fluctuations are governed by the tracy - widom @xmath0 law , converges weakly , when properly centered and scaled , to the gumbel distribution .
more generally we obtain the convergence in law of the marginal distribution of any eigenvalue with given index @xmath2 .
those convergences are obtained after a careful analysis of the explosion times process of the riccati diffusion associated to the stochastic airy operator .
we show that the empirical measure of the explosion times converges weakly to a poisson point process using estimates proved in [ l. dumaz and b. virg .
ann .
inst .
h. poincar probab .
statist . *
49 * , 4 , 915 - 933 , ( 2013 ) ] .
we further compute the empirical eigenvalue density of the stochastic airy ensemble on the macroscopic scale when @xmath3 . as an application
, we investigate the maximal eigenvalues statistics of @xmath4-ensembles when the repulsion parameter @xmath5 when @xmath6 .
we study the double scaling limit @xmath7 and argue with heuristic and numerical arguments that the statistics of the marginal distributions can be deduced following the ideas of [ a. edelman and b. d. sutton .
j. stat .
phys . * 127 * 6 , 1121 - 1165 ( 2007 ) ] and [ j. a. ramrez , b. rider and b. virg .
j. amer .
math .
soc .
* 24 * 919 - 944 ( 2011 ) ] from our later study of the stochastic airy operator . |
statically stable density stratified shear layers are ubiquitous in the atmosphere and oceans .
such shear layers can become hydrodynamically unstable , leading to turbulence and mixing in geophysical flows .
turbulence and mixing strongly influence the atmospheric and oceanic circulation - processes known to play key roles in shaping our weather and climate .
hydrodynamic instability is characterized by the growth of wavelike perturbations in a laminar base flow .
such perturbations can grow at an exponential rate , transforming the base flow from a laminar to a turbulent state . in the present study
, we will theoretically investigate the underlying mechanism(s ) leading to modal ( exponential ) , as well as non - modal , growth of small wavelike perturbations in idealized homogeneous and stratified shear flows .
the classical method used to determine flow stability is the normal - mode approach of linear stability analysis @xcite . under the normal - mode assumption , a waveform can grow or decay but can not deform .
normal - mode perturbations are added to the laminar background flow profile , followed by linearizing the governing navier stokes equations . for inviscid , density stratified shear flows
, the normal - mode formalism leads to the celebrated taylor - goldstein equation , derived independently by @xcite and @xcite .
the taylor - goldstein equation is an eigenvalue problem which calculates the wave properties like growth - rate , phase - speed , and eigenfunction associated with each normal - mode . for stability analysis ,
the range of unstable wavenumbers and the wavenumber corresponding to the fastest growing normal - mode are of prime interest . in many practical scenarios ,
the taylor - goldstein equation is found to accurately capture the onset of instability @xcite , and it also provides a first order description of the developing flow structures @xcite . the normal - mode approach , however , has shortcomings .
firstly , it provides limited insight into the physical mechanism(s ) responsible for hydrodynamic instability .
the answer to why an infinitesimal perturbation vigorously grows in an otherwise stable background flow is provided in the form of mathematical theorems - rayleigh - fj@xmath0rtoft s theorem for homogeneous flows , and miles - howard criterion for stratified flows @xcite .
it is often difficult to form an intuitive understanding of shear instabilities from these theorems . since
linear instability is the first step towards understanding the more complicated and highly elusive non - linear processes like chaos and turbulence , it is desirable to explore alternative routes in order to provide additional insight . in this context sir g. i. taylor
writes : `` it is a simple matter to work out with the equations which must be satisfied by waves in such a fluid , but the interpretation of the solutions of these equations is a matter of considerable difficulty '' @xcite .
a second drawback of the normal - mode approach is the normal - mode assumption itself . the extensive work by @xcite , @xcite , @xcite and others have shown that shear allows rapid non - modal transient growth due to non - orthogonal interaction between the modes .
@xcite developed the `` generalized stability theory '' for obtaining the optimal non - modal growth from the singular value decomposition of the propagator matrix of a linear dynamical system .
lord rayleigh was probably the first to inquire about the mechanism behind shear instabilities , and conjectured the possible role of wave interactions @xcite .
lord rayleigh s hypothesis was later corroborated by sir g. i. taylor while he was theoretically studying three - layered flows in constant shear .
he explained the mechanism to be as follows : `` thus the instability might be regarded as being due to a free wave at the lower surface forcing a free wave at the upper surface of separation when their velocities in space coincide '' @xcite .
early explorations of the wave interaction concept have been reviewed in @xcite .
the first mathematically concrete mechanistic description of stratified shear instabilities was provided by @xcite .
using idealized velocity and density profiles , holmboe postulated that the resonant interaction between stable propagating waves , each existing at a discontinuity in the background flow profile ( density profile discontinuity produces interfacial gravity waves and vorticity profile discontinuity produces vorticity waves ) , yields exponentially growing instabilities .
he was able to show that rayleigh / kelvin - helmholtz ( hereafter , `` kh '' ) instability @xcite is the result of the interaction between two vorticity waves ( also known as rayleigh waves ) .
moreover , holmboe also found a new type of instability , now known as the `` holmboe instability '' , produced by the interaction between vorticity and interfacial gravity waves .
bretherton , a contemporary of holmboe , proposed a similar theory to explain mid - latitude cyclogenesis @xcite .
he hypothesized that cyclones form due to a baroclinic instability caused by the interaction between two rossby edge waves ( vorticity waves in a rotating frame of reference ) , one existing at the earth s surface and the other located at the atmospheric tropopause .
the theories proposed by holmboe and bretherton have been refined and re - interpreted over the years , see @xcite . as reviewed in @xcite , resonant interaction between two edge waves in an idealized homogeneous or stratified shear layer
occurs when these waves attain a phase - locked state , i.e. they are at rest relative to each other . maintaining this phase - locked configuration , the waves grow equally at an exponential rate .
there is also an alternative description of shear instabilities through wave interactions which was put forward by @xcite .
he introduced the concept of `` negative energy waves '' , which are stable modes , and their introduction into the flow causes a decrease in the total energy .
whether a given wave mode has positive or negative energy depends on the frame of reference used .
instability results when negative energy mode resonates with a positive energy mode , and this occurs when the waves have the same phase - speed and wavelength .
this can be identified by the crossing of dispersion curves for the positive and negative energy modes in a frequency - wavenumber diagram .
yet another mechanistic picture of shear instabilities was proposed by lindzen and co - authors ( summarized in @xcite ) .
this theory , known as the `` over - reflection theory '' , proposes that under the right flow configuration , over - reflection of waves can continuously energize an advective `` orr process '' @xcite which is finally responsible for the perturbation growth . the present paper focusses on studying shear instabilities in terms of wave interactions .
from the recent review by @xcite it can be inferred that the wave interaction theory is in its early phases of development .
in fact , there is no strong theoretical justification behind the argument that two progressive waves lock in phase and resonate , thereby producing exponentially growing instabilities in a shear layer .
many questions in this context remains unanswered , e.g. ( a ) is there a condition which determines whether two waves will lock in phase ?
( b ) starting from an initial condition , how long does it take for the waves to get phase - locked ?
( c ) if exponentially growing instabilities occur after phase - locking , then what kind of instabilities ( if any ) occur prior to phase - locking ?
a point worth mentioning here is that the phenomenon of phase - locking occurs in diverse problems ranging from biology to electronics @xcite .
in fact `` synchronization '' is an area of study ( in dynamical systems theory ) specifically dedicated to this purpose .
the history of synchronization goes back to the @xmath1th century when the famous dutch scientist christiaan huygens reported his observation of synchronization of two pendulum clocks .
we suspect that there may be an analogy between the fundamental aspects of synchronization theory and that of wave interaction based interpretation of shear instabilities .
if exploited successfully , this analogy could be beneficial in answering some of the key questions concerning the origin and evolution of shear instabilities . in recent years
, heifetz and co - authors @xcite have extensively studied the interaction between rossby edge waves .
they performed a detailed analysis of non - modal instability and transient growth mechanisms in idealized barotropic shear layers .
modal and non - modal growth occurring due to the interactions between surface gravity waves and a pair of vorticity waves ( existing at the interfaces of a submerged piecewise linear shear layer ) was recently studied by @xcite .
the goal of our paper is to frame a theoretical model of shear instabilities , which could be applied to different idealized ( broken - line ) shear layer profiles ( e.g. the classical profiles studied by rayleigh , taylor or holmboe ) .
the model should be sufficiently generalized in the sense that the constituent wave types need not be specified _ a priori_.
the model should be able to capture the transient dynamics , hence should not be limited to normal - mode waveforms .
the outline of the paper is as follows . in [ sec : linlin ] we provide a theoretical background of linear progressive waves , and focus on two types of waves - vorticity waves and internal gravity waves .
the wave theory in this section is more generalized than that usually reported in the literature . in
[ sec : waveint ] we investigate the mechanism of interaction between two progressive waves .
we undertake a dynamical systems approach to better understand the wave interaction problem , especially the resonant condition .
since the wave interaction formulation is not restricted to normal - mode type instabilities , we investigate non - modal / transient growth processes in
[ sec : matstuff ] . finally in
[ sec : inst ] we use wave interactions to analyze three well known types of shear instabilities - kh instability ( resulting from the interaction between two vorticity waves ) , taylor - caulfield instability ( resulting from the interaction between two interfacial internal gravity waves ) , and holmboe instability ( resulting from the interaction between a vorticity wave and an interfacial internal gravity wave ) .
consider a fluid interface existing at @xmath2 in an unbounded , inviscid , incompressible , two dimensional ( @xmath3-@xmath4 ) flow .
moreover , let the interface be perturbed by an infinitesimal displacement @xmath5 in the @xmath4 direction , given as follows : @xmath6 } \}. \label{eq : eta}\ ] ] this displacement manifests itself in the form of stable progressive wave(s ) , the amplitude and phase of which are @xmath7 and @xmath8 , respectively .
for example , when this interface is a vorticity interface , it produces a vorticity wave .
likewise , two oppositely traveling gravity waves are produced in case of a density interface .
we have assumed the interfacial displacement ( or the wave ) to be monochromatic , having a wavenumber @xmath9 .
moreover , the interface satisfies the _ kinematic condition _ - a particle initially on the interface will remain there forever .
the linearized kinematic condition is given by @xmath10 where @xmath11 is the background velocity in the @xmath3 direction and @xmath12 is the @xmath4-velocity at the interface .
we prescribe the latter to be as follows : @xmath13}\}. \label{eq : wi}\ ] ] here @xmath14 is the amplitude and @xmath15 is the phase of @xmath12 .
the interfacial displacement creates vorticity perturbation _ only _ at the interface , the perturbed velocity field is irrotational everywhere else in the domain .
thus @xmath16 in the above equation @xmath17 is the perturbation streamfunction . assuming @xmath18\}$ ] , and substituting it in ( [ eq : lapup ] ) we get @xmath19 the above equation yields @xmath20 ( where @xmath21 ) .
the vertical velocity @xmath22 is then given by @xmath23 thus , like the streamfunction , the vertical velocity decays exponentially away from the interface , and vanishes at infinity . substituting ( [ eq : eta ] ) and
( [ eq : wi ] ) in ( [ eq : kincon ] ) , we obtain @xmath24 here @xmath25 $ ] .
the frequency and the growth - rate of a wave are respectively defined as @xmath26 and @xmath27 ( overdot denotes @xmath28 ) . using these definitions in ( [ eq : singwave ] ) , we get @xmath29 where @xmath30 .
( [ eq:1freq ] ) shows that the frequency of a wave consists of two components - ( i ) the doppler shift @xmath31 , and ( ii ) the _ intrinsic frequency _ @xmath32 , where @xmath33 .
the phase - speed @xmath34 of the wave is found to be @xmath35 where @xmath36 denotes the _ intrinsic phase - speed_. noting that a wave in isolation can not grow or decay on its own , ( [ eq:13 ] ) demands that @xmath37 .
therefore for a stable wave , the vertical velocity field at the interface has to be in quadrature with the interfacial deformation .
another point worth mentioning is that an isolated wave can not accelerate or decelerate on its own , hence @xmath38 should be constant .
this in turn means @xmath39 is a constant quantity .
applying the quadrature condition to ( [ eq:1freq ] ) and ( [ eq:14 ] ) , the magnitudes of the intrinsic frequency and the intrinsic phase - speed become @xmath40 and @xmath41 , respectively .
the _ intrinsic direction of motion _ of the wave , however , is determined by @xmath42 . for waves moving to the left relative to the interfacial velocity ( @xmath43 ) , @xmath44 .
similarly for right moving waves , @xmath45 .
when such a stable progressive wave is acted upon by external influence(s ) ( e.g. when another wave interacts with the given wave , as detailed in [ sec : waveint ] ) , the quadrature condition is no longer satisfied , i.e. @xmath46 .
therefore , the wave may grow ( @xmath47 ) or decay ( @xmath48 ) , and its intrinsic frequency @xmath49 and phase - speed @xmath38 may change . in our analyses
we will consider two types of progressive interfacial waves - vorticity waves and internal gravity waves .
vorticity waves , also known as rayleigh waves , exist at a vorticity interface ( i.e. regions involving a sharp change in vorticity ) .
such interfaces are a common feature in the atmosphere and oceans . in a rotating frame ,
the analog of the vorticity wave is the rossby edge wave which exists at a sharp transition in the potential vorticity .
when rossby edge waves propagate in a direction opposite to the background flow , they are called `` counter - propagating rossby waves '' @xcite . in order to evaluate the frequency @xmath39 of vorticity waves ,
let us consider a velocity profile having the form @xmath50 here the constant @xmath51 is the vorticity , or the shear in the region @xmath52 .
( [ eq : v1 ] ) shows that the vorticity @xmath53 is discontinuous at @xmath2 .
this condition supports a vorticity wave .
an interfacial deformation @xmath5 adds vorticity @xmath54 to the upper layer and removes it from the lower layer , thereby creating a vorticity imbalance at the interface .
this imbalance is countered by the background vorticity gradient , which acts as a restoring force , and thereby leads to wave propagation . the horizontal component @xmath55 of the perturbation velocity field set up by the interfacial deformation undergoes a jump at the interface
, the value of which can be determined from stokes theorem ( see appendix [ app : b ] ) : @xmath56 by taking an @xmath3 derivative of ( [ eq : stokes ] ) and invoking the continuity relation , we get @xmath57 by substituting ( [ eq : eta ] ) and ( [ eq : w ] ) in ( [ eq : vort12 ] ) we obtain @xmath58 where @xmath59 is the sign function . from ( [ eq : vortwaverel ] ) @xmath49 of a vorticity wave is found to be @xmath60 .
the phase - speed @xmath61 can be evaluated by substituting ( [ eq : vortwaverel ] ) into ( [ eq:14 ] ) : @xmath62 if @xmath63 , the vorticity wave moves to the left relative to the background flow .
the conventional derivation of the frequency and phase - speed of a vorticity wave can be found in @xcite .
interfacial gravity waves exist at a density interface , i.e. a region involving sharp change in density .
the most common example is the surface gravity wave existing at the air - sea interface .
however , in this paper we will only consider waves existing at the interface between two fluid layers having small density difference .
such waves are known as interfacial _
gravity waves ( hereafter , gravity waves ) , and often arise in the pycnocline region of density stratified natural water bodies like lakes , estuaries and oceans . in the case of gravity waves ,
@xmath49 can be evaluated by considering the _ dynamic condition_. this condition implies that the pressure at the density interface must be continuous . let the density of upper and lower fluids be @xmath64 and @xmath65 respectively .
the background velocity is constant , and is equal to @xmath43 .
then the linearized dynamic condition at the interface @xmath2 after some simplification becomes ( ( 3.13 ) of @xcite ) : @xmath66 here @xmath67 is the reduced gravity and @xmath68 is the reference density . under boussinesq approximation @xmath69 . by taking an @xmath3 derivative of ( [ eq : dynbound ] ) and using the streamfunction relation @xmath70 , we get @xmath71 substitution of ( [ eq : eta ] ) and ( [ eq : wi ] ) in ( [ eq : gravrel ] ) yields @xmath72 the quantity @xmath73 . on substituting this relation in ( [ eq:12221 ] )
we obtain @xmath74 an important aspect of ( [ eq : gravwaverel ] ) is that it has been derived _
independent _ of the kinematic condition .
the presence of single or multiple interfaces _ does not _ alter the expression in ( [ eq : gravwaverel ] ) , implying that this equation provides a generalized description of @xmath39 .
inclusion of the kinematic condition yields an expression for @xmath39 which is simpler , but problem specific .
for example , when a single interface is present , inclusion of the kinematic condition in ( [ eq : gravwaverel ] ) produces the well known expression for gravity wave frequency .
this can be shown by substituting ( [ eq:1freq ] ) ( note that this equation has been derived from ( [ eq : kincon ] ) , which is the kinematic condition for a single interface . ) in ( [ eq : gravwaverel ] ) and considering only the positive value : @xmath75 the above equation is the dispersion relation for gravity waves .
substitution of ( [ eq : gravwaverel ] ) in ( [ eq:14 ] ) produces the well known expression for the phase - speed of a gravity wave : @xmath76 the above equation shows that each density interface supports two gravity waves , one moving to the left and the other to the right relative to the background velocity @xmath43 .
it is not uncommon in the literature to use the expressions ( [ eq : gravwaverel111 ] ) and ( [ eq : gravspeed ] ) _ even _ when multiple interfaces are present .
such usage certainly leads to erroneous results , especially when the objective is to study multiple wave interactions .
probably the confusion arises from the traditional derivation of gravity wave frequency and phase - speed @xcite .
this derivation strategy obscures the fact that the kinematic condition ( at an interface ) is influenced by the number of interfaces present in the flow , while the dynamic condition is not .
let us now consider a system with two interfaces , one at @xmath77 and the other one at @xmath78 .
the linearized kinematic condition at each of these interfaces is given by : @xmath79 it has been implicitly assumed that both waves have the same wavenumber @xmath9 . the r.h.s .
of ( [ eq : kin1])-([eq : kin2 ] ) reveal the subtle effect of wave interaction , and can be understood as follows .
the effect of @xmath80 extends away from the interface @xmath81 , hence it can be felt by a wave existing at another location , say @xmath82 .
therefore the vertical velocity of the wave at @xmath82 gets modified - it becomes the linear superposition of its own vertical velocity @xmath83 and the component of @xmath80 existing at @xmath82 .
this phenomenon is also known as `` action - at - a - distance '' , see @xcite . on substituting ( [ eq : eta ] ) and ( [ eq : wi ] ) in ( [ eq : kin1])-([eq : kin2 ] ) ,
we get @xmath84 proceeding in a manner similar to [ sec : linlin ] , the growth - rate @xmath85 and phase - speed @xmath61 of each wave are found to be @xmath86 \label{eq:24}\\ & & \gamma_{2}=\frac{a_{w_{2}}}{a_{\eta_{2}}}\cos\left(\triangle\phi_{22}\right)+\frac{a_{w_{1}}}{a_{\eta_{2}}}e^{-\alpha\left|z_{1}-z_{2}\right|}\cos\left(\triangle\phi_{21}\right ) \label{eq:25}\\ & & c_{2}=u_{2}-\frac{1}{\alpha}\left[\frac{a_{w_{2}}}{a_{\eta_{2}}}\sin\left(\triangle\phi_{22}\right)+\frac{a_{w_{1}}}{a_{\eta_{2}}}e^{-\alpha\left|z_{1}-z_{2}\right|}\sin\left(\triangle\phi_{21}\right)\right ] .
\label{eq:26}\end{aligned}\ ] ] here @xmath87 . when @xmath88 , the two waves decouple , and we recover ( [ eq:13])-([eq:14 ] ) for each wave .
as argued in [ sec : linlin ] , a wave in isolation can not grow or decay on its own .
therefore , the first term in each of ( [ eq:23 ] ) and ( [ eq:25 ] ) should be equal to zero , implying @xmath37 . in all our analyses
, we will be considering a system with a _ left moving top wave _ ( @xmath89 ) and a _
right moving bottom wave _ ( @xmath90 ) , the wave motion being relative to the background velocity at the corresponding interface .
therefore , we will _ only _ consider counter - propagating waves ( i.e. waves moving in a direction opposite to the background flow at that location ) .
@xcite provides a detailed explanation how counter - propagating vorticity waves in rayleigh s shear layer naturally satisfies rayleigh - fj@xmath0rtoft s necessary condition of shear instability .
let the phase - shift between the bottom and top waves be @xmath91 .
therefore @xmath92 and @xmath93 .
defining amplitude - ratio @xmath94 , we re - write ( [ eq:23])-([eq:26 ] ) to obtain @xmath95\label{eq:28}\\ & & \gamma_{2}=r\upomega_{1}e^{-\alpha\left|z_{1}-z_{2}\right|}\sin\phi\label{eq:29}\\ & & c_{2}=u_{2}+\frac{1}{\alpha}\left[\upomega_{2}-r\upomega_{1}e^{-\alpha\left|z_{1}-z_{2}\right|}\cos\phi\right].\label{eq:30 } \end{aligned}\ ] ] the quantity @xmath96 , redefined here for convenience , is constant according to the argument in [ sec : linlin ] . eqs .
( [ eq:27])-([eq:30 ] ) describe the linear hydrodynamic stability of the system . unlike the conventional linear stability analysis
, we did not impose normal - mode type perturbations ( they only account for exponentially growing instabilities ) in our derivation .
therefore the equation - set provides a _ non - modal _ description of hydrodynamic stability in multi - layered shear flows .
we refer to this theory as the `` wave interaction theory ( wit ) '' .
a schematic description of the wave interaction process is illustrated in figure [ fig : wave_int ] .
interestingly , there exists an analogy between wit and the theory behind _ synchronization _ of two coupled harmonic oscillators .
synchronization is the process by which interacting , oscillating objects affect each other s phases such that they spontaneously lock to a certain frequency or phase @xcite . weakly ( linearly ) coupled oscillators interact only through their phases , however more complicated interaction takes place when the amplitudes of oscillation can not be neglected . the first analytical step to include
the effect of amplitude in the synchronization problem is by assuming the coupling to be weakly non - linear . for the case of two
coupled harmonic oscillators , weakly non - linear theory yields a system of @xmath97 equations ( * ? ?
* equation ( 8.13 ) ) , which represents a dynamical system describing the temporal variation of amplitude and phase of each oscillator ( i.e. @xmath98 and @xmath99 ) .
once the highest order terms are neglected , these equations become analogous to our wit equation - set ( [ eq:27])-([eq:30 ] ) .
notice that the wit equation - set is expressed in terms of @xmath85 and @xmath61 , hence in order to see the correspondence between the two sets , we recall the following relations : @xmath27 and @xmath100 . observing the analogy with the theory of coupled oscillators , we re - frame the wave interaction problem into a dynamical systems problem
. subtracting ( [ eq:29 ] ) from ( [ eq:27 ] ) and ( [ eq:30 ] ) from ( [ eq:28 ] ) , we find @xmath101 .
\label{eq : dphidt}\end{aligned}\ ] ] the two parameters have the following range of values : @xmath102 and @xmath103 $ ] .
( [ eq : dphidt ] ) resembles the well known adler s equation in synchronization theory ( * ? ? ?
* equation ( 7.29 ) ) .
the equation - set ( [ eq : drdt])-([eq : dphidt ] ) represents a two dimensional , autonomous , non - linear dynamical system .
although the dynamical system is non - linear , the fact that it has only two dynamical variables ( @xmath104 and @xmath105 ) imply that the system is _ non - chaotic_. the _ two _ equilibrium points of the system , found by imposing the steady state condition @xmath106 in ( [ eq : drdt ] ) and @xmath107 in ( [ eq : dphidt ] ) , are given by @xmath108 where@xmath109 .
\label{eq : theta_nm_general}\end{aligned}\ ] ] for reasons that will become evident in [ sec : matstuff ] we have used the subscript nm , which is short for `` normal - mode '' .
( [ eq : theta_nm_general ] ) reveals that the equilibrium points exist only if @xmath110 it might appear that @xmath111 , where @xmath112 , trivially satisfies the steady state condition .
although ( [ eq : drdt ] ) is satisfied , substitution of @xmath111 in ( [ eq : dphidt ] ) yields only one valid expression of @xmath104 , which is ( [ eq : r_nm_general ] ) .
any other expression violates ( [ eq : necessary ] ) and yields imaginary solutions of @xmath104 for all @xmath113 .
the linear behaviour of the dynamical system around the equilibrium points is of interest . to understand this behaviour
, we evaluate the jacobian matrix , @xmath114 , at the equilibrium points : @xmath115 .
\label{eq : jacobian}\ ] ] eq .
( [ eq : jacobian ] ) shows that the two eigenvalues corresponding to each equilibrium point are equal .
further analysis reveals that every vector at the equilibrium point is an eigenvector .
the equilibrium point @xmath116 produces negative eigenvalues ( assuming @xmath117 ) , while the eigenvalues corresponding to @xmath118 are positive .
the dynamical system represented by ( [ eq : drdt])-([eq : dphidt ] ) is therefore `` source - sink '' type .
this means that a point in the phase space will move away from the `` source node '' @xmath118 , and following a unique trajectory , will finally converge to the `` sink node '' @xmath116 . in summary
, wit provides an understanding of hydrodynamic instability from two different perspectives - wave interaction and dynamical systems . according to the former , exponentially
growing instabilities signify resonant interaction between two waves . from dynamical systems
point of view , resonance implies a `` steady state or equilibrium condition '' ( @xmath107 and @xmath106 ) .
the wave interaction interpretation of each of the two components of the equilibrium condition is as follows : [ [ a ] ] ( a ) + + + _ phase - locking _ or @xmath107 : reduction in the phase - speed of each wave occurs through the interaction mechanism - the vertical velocity field produced by the distant wave acts so as to diminish the phase - speed of the given wave .
furthermore , if the waves are `` counter - propagating '' ( meaning , the direction of @xmath38 is opposite to the background flow ) , the background flow causes an additional reduction in the phase - speed .
both wave interaction and counter - propagation work synergistically until the two waves become `` phase - locked '' , i.e. stationary relative to each other ( @xmath107 ) .
the phase - shift at the phase - locked state , @xmath119 , is a unique angle dictated by the physical parameters of the system , as evident from ( [ eq : theta_nm_general ] ) .
_ any _ arbitrary initial condition ( say @xmath120 and @xmath121 ) finally leads to phase - locking , as evident from figure [ fig : summaryfig](a ) , provided ( [ eq : necessary ] ) is satisfied . when the condition ( [ eq : necessary ] ) is _ not _ satisfied , the interaction is too weak to cause phase locking
the two waves continue moving in opposite directions , hence the phase - shift angle oscillates between @xmath122 $ ] ( for example see figure [ fig : manyfigs](d ) ) . [ [ b ] ] ( b ) + + + _ mutual growth _ or @xmath106 : not only the two waves lock in phase , they also lock in amplitude , producing the unique steady state amplitude - ratio @xmath123 ; see figure [ fig : summaryfig](b ) .
( [ eq : drdt ] ) reveals that @xmath106 implies @xmath124 , which in turn signifies _ resonance _ between the two waves .
furthermore , ( [ eq:27 ] ) and ( [ eq:29 ] ) imply that at steady state @xmath125 , meaning that the wave amplitudes grow at an exponential rate .
in the previous section we used wit and dynamical systems theory to describe multi - layered inviscid hydrodynamic instabilities in terms of interacting interfacial waves .
we found that the `` equilibrium condition '' of the dynamical system basically signifies the `` resonant condition '' of wit .
a question which naturally arises is `` _ _ what do these conditions mean in terms of conventional hydrodynamic stability theory _ _ ? '' in this section we will draw parallels between wit ( and the dynamical systems formulation ) and eigenanalysis and singular value decomposition ( svd ) , which are conventionally used to study modal and non - modal instabilities , respectively . the linearized kinematic condition with two interfaces , ( [ eq : kin1])-([eq : kin2 ] ) , can be written in the matrix form by expressing @xmath12 in terms of @xmath5 ( by invoking the definition of @xmath39 ) , and imposing the condition for counter - propagation ( @xmath89 and @xmath90 ) : @xmath126 where @xmath127\,\,\,\,\,\,\,\,\,\textrm{and}\,\,\,\,\,\,\,\,\boldsymbol{\mathcal{m}}=-i\left[\begin{array}{cc } \alpha u_{1}-\upomega_{1 } & \upomega_{2}e^{-\alpha\left|z_{1}-z_{2}\right|}\\ -\upomega_{1}e^{-\alpha\left|z_{1}-z_{2}\right| } & \alpha u_{2}+\upomega_{2 } \end{array}\right].\ ] ] eq .
( [ eq : eettaa ] ) represents the first order perturbation dynamics , and @xmath128 is the linearized dynamical operator . since
our dynamical system is autonomous @xmath128 is time independent , and the solution is explicit : @xmath129\boldsymbol{\eta}\left(0\right)=\left[\boldsymbol{\mathcal{u}}\boldsymbol{\sigma}\boldsymbol{v}^{\dagger}\right]\boldsymbol{\eta}\left(0\right ) .
\label{eq : boroeqn}\ ] ] the matrix exponential @xmath130 in the above equation is the propagator matrix , which advances the system in time .
the transient dynamics of the system is solely governed by the normality of @xmath128 , i.e. whether or not @xmath128 commutes with its hermitian transpose @xmath131 @xcite .
if they commute ( @xmath132 ) then @xmath128 is normal and has complete set of orthogonal eigenvectors . in this case
the dynamics can be fully understood from the eigendecomposition of the propagator .
this basically means expressing @xmath130 as @xmath133 , where @xmath134 is a diagonal matrix containing the complex eigenvalues @xmath135 of @xmath128 ( arranged by the real part of the eigenvalues in the descending order of magnitude ) , and @xmath136 is the corresponding matrix of eigenvectors .
alternatively if @xmath128 is non - normal , the interaction between the discrete non - orthogonal modes of @xmath128 produces non - normal growth processes .
non - normality can be understood through svd of the propagator .
svd is a generalized matrix factorization technique and matches eigendecomposition _ only _ when @xmath128 is hermitian ( @xmath137 ) .
svd of the propagator yields @xmath138 , where @xmath139 contains the eigenvectors of @xmath140 , @xmath141 contains the eigenvectors of @xmath142 , and @xmath143 is a diagonal matrix containing the singular values ( @xmath144 , which are real and positive ) arranged in the descending order of magnitude . if @xmath128 commutes with its hermitian transpose , then @xmath145 .
otherwise @xmath146 , implying that growth - rate higher than the least stable normal - mode is possible @xcite .
if @xmath128 is a normal matrix , then eigendecomposition of the propagator matrix is sufficient to capture the dynamics .
if @xmath128 is non - normal , then eigenanalysis captures the asymptotic dynamics _ only _ for large times .
the eigenvalues of the matrix @xmath128 are as follows : @xmath147\pm\frac{1}{2}\sqrt{\mathscr{d } } , \label{eq : lamd}\ ] ] where @xmath148^{2}$ ] .
normal - mode instability can _ only _ occur if @xmath149 .
this basically gives rise to the condition ( [ eq : necessary ] ) , implying that ( [ eq : necessary ] ) denotes the _ necessary and sufficient _
( n&s ) _ condition _ for exponentially growing instabilities in idealized ( broken - line profiles ) , homogeneous and stratified , inviscid shear layers .
unstable _ eigenvalues of ( [ eq : lamd ] ) can therefore be written as @xmath147\pm\sqrt{\upomega_{1}\upomega_{2}}e^{-\alpha\left|z_{1}-z_{2}\right|}\sin\left(\phi_{nm}\right ) .
\label{eq : lamd1}\ ] ] the positive and negative eigenvalues respectively imply growing and decaying normal - modes of the discrete spectrum .
there is a direct relation between the normal - modes and the resonant condition of wit ( or the equilibrium condition of the dynamical system ( [ eq : drdt])-([eq : dphidt ] ) ) . _
each equilibrium point corresponds to a normal - mode _ :
@xmath116 corresponds to the growing normal - mode ( signifying exponential growth @xmath150 ) , and @xmath118 corresponds to the decaying normal - mode ( signifying exponential decay @xmath151 ) .
an analogy can be drawn with a system of two coupled harmonic oscillators .
such a system has two normal - modes of vibration , the in - phase synchronization mode ( zero phase - shift ) and the anti - phase synchronization mode ( phase - shift of @xmath152 ) .
the in - phase and anti - phase modes are respectively analogous to the growing and decaying normal - modes of wit .
wit is only relevant to problems where the discrete spectrum is of primary interest , and the effect of the continuous spectrum can be neglected . in other words ,
the applicability of wit is restricted to idealized multi - layered profiles .
real profiles are continuous and their eigenanalysis yields a continuous spectrum .
although the continuous spectrum consists of neutral modes , they interact to produce sheared disturbances .
such disturbances can be obtained by artificially changing the vorticities of fluid elements by small amounts , imposing a sinusoidal streamwise dependence , and then letting the system evolve freely .
this spectrum crucially contributes to the non - modal instability processes , and is responsible for large transient growths .
if @xmath128 is non - normal , then eigenanalysis fails to capture the dynamics in the limit @xmath153 . to understand the short - term dynamics
, one needs to perform svd analysis .
the eigenvalue matrix of @xmath140 ( or @xmath142 ) is equal to @xmath154 ; the maximum eigenvalue of @xmath143 is key in determining the transient dynamics . in the limit @xmath153 ,
taylor expansion of @xmath142 produces @xcite : @xmath155 where @xmath156 is the identity matrix .
the maximum eigenvalue of @xmath157 ( which is known as the _ numerical abscissa _ of @xmath128 ) and its associated eigenvector provide the maximum instantaneous growth - rate and structure .
the numerical abscissa is found to be @xmath158 it is straight - forward to check that @xmath159 the non - normality of the system can give rise to transient energy amplification .
even when the waves are decaying , the non - orthogonal superposition of eigenvectors may lead to short - time growth of energy ( or some other norm ) .
for the non - modal analysis we have chosen @xmath160 as a suitable norm .
the time evolution of @xmath161 is used to indicate the growth ( or decay ) of the system , and is calculated as follows : @xmath162 the rightmost side of ( [ eq : mejo ] ) is obtained by substituting ( [ eq:27 ] ) and ( [ eq:29 ] ) .
furthermore , the inequality @xmath163 .
hence @xmath164 @xmath165 is maximum when @xmath166 .
the corresponding @xmath161 is @xmath167 .
the time evolution of @xmath168 is given by @xmath169 the sign of the r.h.s .
of ( [ eq : emax ] ) dictates whether the system is growing ( positive sign ) or decaying ( negative sign ) .
since only @xmath170 is a signed quantity , it means that the instantaneous phase - shift governs the instantaneous growth or decay .
largest _ instantaneous growth occurs when @xmath171 ( this fact can also be verified from ( [ eq:27 ] ) and ( [ eq:29 ] ) ) .
figure [ fig : manyfigs](a ) shows an example of how @xmath172 ( = @xmath173 ) varies with time . during the initial period ,
the wave amplitude decays ( since the initial phase - shift @xmath174 ) and then grows at a rate faster than that predicted by the normal - mode theory . eventually , the growth - rate asymptotes to that predicted by normal - mode theory .
note that the flow becomes non - linear relatively quickly , and the linear theories ( e.g. wit and normal - mode ) start to deviate from the non - linear theory ; see figure 3 of @xcite .
the amplification or gain ( @xmath175 ) of the system is given by : @xmath176 .
\label{eq : gg}\ ] ] the quantity @xmath177 can be expressed in terms of @xmath178 by invoking ( [ eq : dphidt ] ) .
substituting @xmath179 and integrating , we obtain @xmath180 where @xmath181 , and @xmath182 is the value of @xmath105 at @xmath183 .
notice that @xmath184 denotes the n&s condition ( [ eq : necessary ] ) .
when the n&s condition is satisfied @xmath185 ; see ( [ eq : theta_nm_general ] ) .
hence ( [ eq : gg1 ] ) yields @xmath186 we find that @xmath187 when the final phase - shift @xmath188 .
the positive and negative values of @xmath119 respectively denote the sink node and the source node in phase space .
since trajectories move away from the source node , @xmath189 during the non - modal evolution process ( for example , see figure [ fig : kh](b ) ) .
the significance of the non - modal gain @xmath175 can be properly understood when compared with a hypothetical case in which the system undergoes modal growth between @xmath121 and @xmath119 . for this
we first compute the gain obtained from a normal - mode type perturbation growing over an interval @xmath190 : @xmath191 \nonumber \\ = \left|\frac{\sin\left(\frac{\phi_{nm}+\phi_{t}}{2}\right)}{\sin\left(\frac{\phi_{nm}-\phi_{t}}{2}\right)}\right|\left|\frac{\sin\left(\frac{\phi_{nm}-\phi_{0}}{2}\right)}{\sin\left(\frac{\phi_{nm}+\phi_{0}}{2}\right)}\right|.\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\end{aligned}\ ] ] like @xmath175 , the normal - mode gain @xmath192 when @xmath193 .
the gain - ratio @xmath194 is found to be @xmath195 when @xmath196 , the non - normal growth exceeds exponential growth within the time window @xmath197 $ ] .
for our calculations we have chosen the entire time window of non - modal evolution .
thus ( [ eq : gg8 ] ) becomes @xmath198 notice that unlike @xmath175 and @xmath199 , the value of @xmath200 is finite as @xmath201 .
figure [ fig : manyfigs](b ) depicts the variation of @xmath202 with @xmath9 and @xmath203 for kh instability .
the flow is linearly unstable for @xmath204 ( see [ sec : kh ] ) .
non - modal gain _ exceeds _
modal gain by _ several _ orders of magnitude in the neighbourhood of stability boundaries .
this behaviour is especially prominent near the lower boundary .
the fact that substantial transient growth occurs near stability boundaries has also been observed for non - modal holmboe instability @xcite .
the magnitude of @xmath200 is strongly dependent on the initial condition .
we observe higher values of @xmath200 when the component waves are long ( smaller values of @xmath9 ) , and initially out of phase ( @xmath203 away from @xmath205 ) .
this situation reverses near the upper boundary ; waves starting in - phase exhibit higher magnitudes of @xmath200 . when the n&s condition is not satisfied @xmath206 , and the flow is linearly stable according to the normal - mode theory .
the amplitude of a small perturbation is therefore expected to stay constant in this region .
however non - modal processes can produce transient growth in the parameter space for which the flow is deemed stable by the normal - mode analysis .
this can be shown by studying the evolution of the optimal perturbation .
using svd analysis @xcite have shown that optimal perturbation evolves such that the final phase - shift @xmath207 .
the phase - shift is symmetric in time about @xmath208 , maximizing @xmath170 in ( [ eq : gg ] ) .
thus the optimal gain @xmath209 is found to be @xmath210 the optimal gain is maximum when @xmath211 , and is given by @xmath212 @xmath213 is known as the _ global optimal gain_. we take kh instability as an example for studying the non - modal behaviour .
the condition @xmath206 in this case translates to @xmath214 ( see [ sec : kh ] ) . from figure
[ fig : manyfigs](c ) we find that the perturbation amplitude corresponding to @xmath215 grows to a maximum of @xmath216 times its initial value .
similar results have been obtained by @xcite , who studied the non - normal behaviour using the enstrophy norm .
figure [ fig : manyfigs](d ) shows the corresponding temporal variation of its phase - shift .
unlike figure [ fig : summaryfig](a ) , there is no phase - locking since n&s condition is not satisfied here . instead
, the two `` marginally stable '' ( as referred to in the classical theory ) wave - modes continue to propagate in opposite directions .
hence the phase - shift oscillates between @xmath205 and @xmath217 .
the wit formulation proposed in [ sec : waveint ] is quite general in the sense that we have not prescribed the types of the constituent waves .
different kinds of shear instabilities may arise depending on the wave types .
for example , the interaction between two vorticity waves results in kh instability ; taylor - caulfield instability results from the interaction between two gravity waves in constant shear , while the interaction between a vorticity and a gravity wave produces holmboe instability . in this section we will use wit to analyze the three above mentioned classical examples of shear instabilities .
let us consider a piecewise linear velocity profile @xmath218 this profile is a prototype of barotropic shear layers occurring in many geophysical and astrophysical flows @xcite .
it supports two vorticity waves , one at @xmath81 and the other at @xmath82 .
the shear @xmath219 .
we nondimensionalize the problem by choosing a length scale @xmath220 and a velocity scale @xmath221 . in a reference frame
moving with the mean flow @xmath222 , the non - dimensional velocity profile becomes @xmath223 this profile , along with the vorticity waves , is shown in figure [ fig : kh](a ) .
the top wave is left moving while the bottom wave moves to the right . using the classical normal - mode theory @xcite showed that the profile in ( [ eq : kh2 ] ) is linearly unstable for @xmath204 .
this instability is often referred to as the `` rayleigh s shear instability '' .
however we have addressed it as the `` rayleigh / kelvin - helmholtz '' instability , and used the acronym `` kh '' .
figure [ fig : kh](a ) reveals that the kh instability can be analyzed using wit framework . since the two waves involved in the kh instability are vorticity waves , we substitute ( [ eq : vortwaverel ] ) in ( [ eq:27])-([eq:30 ] ) , and after non - dimensionalization we obtain @xmath224\label{eq : khh2}\\ & & \gamma_{2}=\frac{r}{2}e^{-2\alpha}\sin\phi\label{eq : khh3}\\ & & c_{2}=-1+\frac{1}{2\alpha}\left[1-re^{-2\alpha}\cos\phi\right].\label{eq : khh4 } \end{aligned}\ ] ] eqs .
( [ eq : khh1])-([eq : khh4 ] ) provide a _
non - modal _ description of kh instability .
the equation - set is isomorphic to ( 14a)-(14d ) of @xcite and homomorphic to ( 7a)-(7d ) of @xcite . while the equation - set described by @xcite shows how counter - propagating rossby wave interactions lead to barotropic shear instability , the equation - set formulated by @xcite shows how baroclinic instability is produced through the interaction of temperature edge waves .
the generalized non - linear dynamical system given by ( [ eq : drdt])-([eq : dphidt ] ) in this case translates to @xmath225 the equilibrium points of this system are @xmath226 , where @xmath227 .
\label{eq : nor_mode_kh}\end{aligned}\ ] ] the phase portrait is shown in figure [ fig : kh](b ) .
it confirms that the dynamical system is indeed of source - sink type , as predicted in
[ sec : waveint ]
. the n&s condition for instability expressed via ( [ eq : necessary ] ) in this case reads @xmath228 the range of unstable wavenumbers obtained from the above equation corroborates rayleigh s normal - mode analysis . normal - mode theory also shows that the wavenumber of maximum growth is @xmath229 . the same answer can be obtained from wit by imposing the normal - mode condition and maximizing @xmath230 or @xmath231 with respect to @xmath9 . the fact that kh instability develops into a standing wave instability can be verified by applying the normal - mode condition in ( [ eq : khh2 ] ) and ( [ eq : khh4 ] ) . performing the necessary steps we find @xmath232 , i.e. the waves became stationary after phase - locking .
the waves stay in this configuration and grow exponentially , hence the shear layer grows in size .
the growth process eventually becomes non - linear , and the shear layer modifies into elliptical patches of constant vorticity @xcite .
let us consider a uniform shear layer with two density interfaces @xmath233 the shear @xmath54 is constant .
we choose @xmath234 as the density scale , @xmath220 as the length scale , and thereby nondimensionalize ( [ eq : tayl0 ] ) . the physical state of the system is determined by the competition between density stratification and shear , the non - dimensional measure of which is given by the bulk richardson number @xmath235 , where @xmath236 is the reduced gravity , and @xmath68 is the reference density .
the dimensionless velocity and density profiles therefore become @xmath237 this flow configuration is shown in figure [ fig : taylor](a ) .
notice that a _ homogeneous _ flow with @xmath238 is linearly stable .
moreover , a _ stationary _
fluid having a density distribution as in ( [ eq : tayl1 ] ) is gravitationally stable .
surprisingly , the flow obtained by superimposing the two stable configurations is linearly unstable .
@xcite was the first to report and provide a detailed theoretical analysis of this rather non - intuitive instability .
the first experimental evidence of this instability was provided by @xcite .
following @xcite we refer to it as the `` taylor - caulfield ( tc ) instability '' .
the interplay between the background shear and the gravity waves existing at the density interfaces produce the destabilizing effect .
@xcite found that for each value of @xmath239 , there exists a band of unstable wavenumbers ( and vice - versa ) , shown in figure [ fig : taylor](b ) .
this unstable range is given by ( see @xcite or ( 2.154 ) of @xcite ) : @xmath240 probably the only way to make sense of tc instability is through wave interactions , which has been described in @xcite and @xcite using normal - mode theory .
wit provides the framework for studying non - modal tc instability .
substituting ( [ eq : gravwaverel ] ) in ( [ eq:27])-([eq:30 ] ) , and performing non - dimensionalization , we obtain @xmath241 here @xmath242 , and by definition is a positive quantity . from ( [ eq : t2 ] ) and ( [ eq : t4 ] )
we construct a quadratic equation for @xmath243 : @xmath244 amongst the two roots , only the positive root is relevant .
equations ( [ eq : t1])-([eq : t4 ] ) provide a _ non - modal _ description of tc instability .
the equation - set is of coupled nature , and is therefore _ not _ homomorphic to ( [ eq:27])-([eq:30 ] ) .
the non - linear dynamical system in this case is given by @xmath245 at phase - locking @xmath246 .
substituting this value in ( [ eq : betaevol ] ) gives @xmath247 .
therefore @xmath248 and @xmath232 at resonance , which implies that tc instability , like kh , also evolves into a stationary disturbances .
the phase portrait is shown in figure [ fig : taylor](c ) .
the phase - shift @xmath119 is evaluated from ( [ eq : tphi ] ) : @xmath249 .
\label{eq : nor_mode_taylor}\ ] ] the n&s condition for tc instability is given by @xmath250 the latter result corroborates the classical normal - mode result given in ( [ eq : ta1 ] ) .
non - modal tc instability has also been studied by @xcite . in their paper
the non - modal equation - set is given by ( 3.11a , b)-(3.12a , b ) , which is quite different from our equation - set ( [ eq : t1])-([eq : t4 ] ) .
however , these two equation - sets should be identical since they are describing the same physical problem . using the equation - set of @xcite we obtained the following range of unstable wavenumbers : @xmath251 .
this bound is different from the correct stability bound ( [ eq : ta1 ] ) .
let us consider the following velocity and density profiles @xmath252 we nondimensionalize ( [ eq : holm0 ] ) exactly like the tc instability problem , which gives us the dimensionless velocity and density profiles : @xmath253 the vorticity interface at the top supports a vorticity wave , while the density interface at the bottom supports two gravity waves .
the interaction between the left moving vorticity wave at the upper interface and the right moving gravity wave at the lower interface leads to an instability mechanism , known as the `` holmboe instability '' ( note that the generation of ocean surface gravity waves by wind shear @xcite can be viewed as `` non - boussinesq holmboe instability '' , and can therefore be interpreted using wave interactions . ) .
the corresponding flow setting is shown in figure [ fig : holmboe](a ) .
@xcite was the first to consider the instability mechanism resulting from the interaction between vorticity and gravity waves . performing a normal - mode stability analysis
, holmboe discovered the existence of an unstable mode characterized by traveling waves .
the profile used by holmboe is more complicated than that which we are considering .
holmboe s profile , which included multiple wave interactions , was substantially simplified by @xcite by introducing the profile in ( [ eq : holm1 ] ) .
linear stability analysis shows that corresponding to each value of @xmath239 , there exists a band of unstable wavenumbers , shown in figure [ fig : holmboe](b ) .
the stability boundary has been evaluated in appendix [ app : a ] . in order to understand the holmboe instability in terms of wit , we substitute ( [ eq : vortwaverel ] ) and ( [ eq : gravwaverel ] ) in ( [ eq:27])-([eq:30 ] ) .
after non - dimensionalization , we obtain @xmath254 like the tc instability case , the holmboe equation - set is also of coupled nature , and is therefore not homomorphic to ( [ eq:27])-([eq:30 ] ) .
the non - linear dynamical system is given by : @xmath255 the equilibrium points of this system are @xmath226 , where @xmath256 .
\label{eq : nor_mode_holmboe}\end{aligned}\ ] ] the n&s condition for holmboe instability is found to be @xmath257 this provides the range of @xmath239 leading to holmboe instability , and is as follows : @xmath258 where @xmath259\\ & & c=\left(2\alpha-1+e^{-2\alpha}\right)\left(2\alpha-1\right)^{3}.\end{aligned}\ ] ] eq .
( [ eq : bound_holmboe ] ) corroborates the normal - mode result given in appendix [ app : a ] . the phase portrait of holmboe instability , corresponding to an unstable combination of @xmath9 and @xmath239 , is shown in figure [ fig : holmboe](c ) .
this phase portrait is slightly different from tc and kh instabilities , because @xmath260 in this case .
another feature of holmboe instability is that , unlike tc and kh instabilities , its phase - speed is non - zero at the equilibrium condition .
this phase - speed is found to be @xmath261 in the limit of large @xmath9 and @xmath239 , the two phase - locked waves move with a speed of unity to the positive @xmath3 direction .
shear instability plays a crucial role in atmospheric and oceanic flows . in the last 50 years
, significant efforts have been made to develop a mechanistic understanding of shear instabilities . using idealized velocity and density profiles
, researchers have conjectured that the resonant interaction between two counter - propagating linear interfacial waves is the root cause behind exponentially growing instabilities in homogeneous and stratified shear layers .
support for this claim has been provided by considering interacting vorticity and gravity waves of the normal - mode form . in this paper
we investigated the wave interaction problem in a generalized sense .
the governing equations ( [ eq:27])-([eq:30 ] ) of hydrodynamic instability in idealized ( broken - line profiles ) , homogeneous or density stratified , inviscid shear layers have been derived _ without _ imposing the wave type , or the normal - mode waveform .
we refer to this equation - set as wave interaction theory ( wit ) . using wit we showed in figures [ fig : summaryfig](a ) and [ fig :
summaryfig](b ) that two counter - propagating linear interfacial waves , having _ arbitrary _ initial amplitudes and phases , eventually _ resonate _ ( lock in phase and amplitude ) , provided they satisfy the n&s condition ( [ eq : necessary ] ) . in
[ subsub : eigenanalysis ] we showed that the n&s condition is basically the criterion for normal - mode type instabilities .
the waves which do not satisfy the n&s condition may exhibit non - normal instabilities , but will never resonate . in other words , such waves will never lock in phase and undergo exponential growth .
these facts have been demonstrated in figures [ fig : manyfigs](c ) and [ fig : manyfigs](d ) .
we considered three classic examples of shear instabilities - rayleigh / kelvin - helmholtz ( kh ) , taylor - caulfield ( tc ) and holmboe , and derived their discrete spectrum non - modal stability equations by modifying the basic wit equations ( [ eq:27])-([eq:30 ] ) .
these equations are respectively given by ( [ eq : khh1])-([eq : khh4 ] ) , ( [ eq : t1])-([eq : t4 ] ) and ( [ eq : h1])-([eq : h4 ] ) .
for each type of instability we validated the corresponding equation - set by showing that the n&s condition matches the predictions of the canonical normal - mode theory .
although validation is an important first step for checking the non - modal equation - sets , the aim of this work is _ not _ proving well - known results using an alternative theory .
the real strength of the non - modal formulation is in providing the opportunity to explore the transient dynamics .
non - orthogonal interaction between the two wave modes can lead to rapid transient growth .
we show in figure [ fig : manyfigs](b ) that depending on wavenumber and initial phase - shift , non - modal gain can exceed corresponding modal - gain by several orders of magnitude .
this implies that the flow may become non - linear during the initial stages of flow development , leading to an early transition to turbulence .
instability has been observed for wavenumbers which are deemed stable by the normal - mode theory .
all these facts are shown for the example of kh instability ; see figures [ fig : manyfigs](a)-(d ) . in order to be able to study the transient dynamics of holmboe and tc instabilities
, the analysis in [ sec : matstuff ] needs to be modified in parts .
this is because the equations developed in
[ sec : matstuff ] are restricted to systems whose non - modal equation - sets are _ homomorphic _ to ( [ eq:27])-([eq:30 ] ) .
some examples of homomorphic equation - sets are : ( [ eq : khh1])-([eq : khh4 ] ) representing kh instability , ( 14a)-(14d ) of @xcite representing barotropic shear instability , and ( 7a)-(7d ) of @xcite representing baroclinic instability of the eady model .
another objective of our work is to provide an intuitive description of shear instabilities , and to find connection with other physical processes .
how two waves can interact to produce instability has been discussed throughout the paper , and schematically represented in figure [ fig : wave_int ] .
we observed an analogy between wit equations and that governing the synchronization of two coupled harmonic oscillators . on the basis of this analogy , we re - framed wit as a non - linear dynamical system .
the resonant configuration of the wave equations translated into a steady state configuration of the dynamical system .
this dynamical system is of the source - sink type ; the source and the sink nodes being the two equilibrium points .
when interpreted in terms of canonical linear stability theory , the source and the sink nodes respectively correspond to the decaying and the growing normal - modes of the discrete spectrum .
the analogous two coupled harmonic oscillator system exhibits two normal modes of vibration - the in - phase and the anti - phase synchronization modes .
the growing normal - mode of wit is analogous to the in - phase synchronization mode , while the decaying normal - mode is analogous to the anti - phase synchronization mode .
although we studied wit in the context of shear instabilities , the framework is actually derived from ( [ eq:23])-([eq:26 ] ) , and is therefore applicable to both sheared and unsheared flows , as well as co - propagating and counter - propagating waves .
as an example , one can study the interaction between deep water surface gravity and internal gravity waves ( which gives rise to barotropic and baroclinic modes of oscillations in lakes ; see @xcite pg .
259 - 261 ) .
the framework can also be extended to the weakly non - linear regime to study multiple - wave interactions leading to resonant triads @xcite .
wit can also be applied to understand liquid - jet instability @xcite .
finally we focus on the implications of using broken - line profiles of velocity and/or density .
these idealizations have allowed us to concentrate only on the discrete spectrum dynamics and understand hydrodynamic instability in terms of interfacial wave interactions .
real profiles are always continuous , hence realistic analysis requires inclusion of the continuous spectrum .
using green s function , @xcite have shown that the continuous spectrum dynamics in a smooth homogeneous shear layer can be understood in terms of infinite number of interacting vorticity ( rossby edge ) waves .
@xcite used the same approach to understand the normal - mode continuous spectrum of smooth stratified shear layers .
these studies indicate that the growth process is better understood through the `` orr mechanism '' of shearing of waves @xcite , than edge wave interactions .
this implies that the intuition brought by wit formulation may become less apparent in continuous systems .
however the applicability of wit in continuous systems can only be properly judged when the current version is modified to include the effects of a continuous spectrum .
we speculate that the analogy between wit and the synchronization of coupled harmonic oscillators can be exploited to obtain an intuitive understanding of continuous spectrum dynamics .
our speculation is based on the work of @xcite , who studied the kuramoto model in the continuous limit , and observed superficial similarities between oscillator synchronization and instabilities in ideal plasmas and inviscid fluids .
we would like to acknowledge the useful comments and suggestions of prof .
neil balmforth of ubc , prof .
kraig winters of ucsd , prof .
michael mcintrye , prof .
colm caulfield and prof .
peter haynes of the university of cambridge ( damtp ) , prof .
eyal heifetz of tel aviv university , prof .
ishan sharma and prof .
mahendra verma of iit kanpur , and the anonymous referees .
stokes theorem relates the surface integral of the curl of a vector ( velocity in this case ) field @xmath262 over a surface @xmath263 to the line integral of the vector field over its boundary @xmath264 : @xmath265 @xcite used this theorem to relate the interfacial displacement @xmath5 with the difference in velocity perturbation ( @xmath266 ) produced at a vorticity interface ; see ( [ eq : stokes ] ) .
this equation is referred to as `` eq .
( 3.2 ) '' in his paper .
however , the relevant steps required to derive this equation were not provided . in order to understand how ( [ eq : stokes ] )
is obtained , we first graphically describe the problem in figure [ fig : appen ] .
the background velocity is such that the flow is irrotational when @xmath267 , and has a constant vorticity , say @xmath54 , when @xmath268 .
when the interface is disturbed by an infinitesimal displacement @xmath5 ( solid black curve in figure [ fig : appen ] ) , the velocity field also changes slightly - the perturbation velocity in the upper layer ( @xmath267 ) becomes @xmath269 and that in the lower layer ( @xmath268 ) becomes @xmath270 .
let us consider a circuit a - b - c - d .
applying stokes theorem , we obtain @xmath271 where @xmath272 is the area of a - b - c - d , and @xmath273 is the vorticity in this area
. therefore we obtain @xmath274 which is basically ( [ eq : stokes ] ) .
both interfaces in the holmboe profile ( [ eq : holm1 ] ) individually satisfy the kinematic condition : @xmath275 where @xmath276 is the stream function perturbation at the lower interface .
this interface being a density interface also satisfies the dynamic condition : @xmath277 we assume the perturbations to be of normal - mode form : @xmath278 , @xmath279 , and @xmath280 . here
the wave speed @xmath281 is generally complex . defining @xmath282^{t}$ ] , we obtain the following eigenvalue problem : @xmath283 where @xmath284.\ ] ] eq .
( [ eq : app2 ] ) generates the following characteristic polynomial : @xmath285 this equation produces complex conjugate roots only when the discriminant is negative . since the presence of complex roots signify normal - mode instability , negative values of the discriminant is of our interest .
the discriminant ( d ) in this case is given by : @xmath286-\left(1 - 2\alpha\right)^{3}\left(2\alpha-1+e^{-2\alpha}\right).\ ] ] imposing the condition @xmath287 , we find @xmath288 where @xmath259\\ & & c=\left(2\alpha-1+e^{-2\alpha}\right)\left(2\alpha-1\right)^{3}.\end{aligned}\ ] ] thus holmboe instability occurs only when the condition in ( [ eq : bbbb ] ) is satisfied . | @xcite postulated that resonant interaction between two or more progressive , linear interfacial waves produces exponentially growing instabilities in idealized ( broken - line profiles ) , homogeneous or density stratified , inviscid shear layers .
here we have generalized holmboe s mechanistic picture of linear shear instabilities by ( i ) not initially specifying the wave type , and ( ii ) by providing the option for non - normal growth .
we have demonstrated the mechanism behind linear shear instabilities by proposing a _ purely _ kinematic model consisting of two linear , doppler - shifted , progressive interfacial waves moving in opposite directions .
moreover , we have found a _
necessary _ and _ sufficient _ ( n&s ) _ condition _ for the existence of exponentially growing instabilities in idealized shear flows .
the two interfacial waves , starting from _ arbitrary _ initial conditions , eventually phase - lock and resonate ( grow exponentially ) , provided the n&s condition is satisfied .
the theoretical underpinning of our wave interaction model is analogous to that of synchronization between two coupled harmonic oscillators .
we have re - framed our model into a non - linear autonomous dynamical system , the steady state configuration of which corresponds to the resonant configuration of the wave - interaction model .
when interpreted in terms of the canonical normal - mode theory , the steady state / resonant configuration corresponds to the growing normal - mode of the discrete spectrum .
the instability mechanism occurring prior to reaching steady state is non - modal , favouring rapid transient growth .
depending on the wavenumber and initial phase - shift , non - modal gain can exceed the corresponding modal gain by many orders of magnitude .
instability is also observed in the parameter space which is deemed stable by the normal - mode theory . using our model
we have derived the discrete spectrum non - modal stability equations for three classical examples of shear instabilities - rayleigh / kelvin - helmholtz , holmboe and taylor - caulfield .
we have shown that the n&s condition provides a range of unstable wavenumbers for each instability type , and this range matches the predictions of the normal - mode theory . |
thermal dilepton production in the low - mass region is largely mediated by the light vector mesons @xmath1 , @xmath3 and @xmath4 . among these ,
the @xmath1(770 mev / c@xmath5 ) is the most important , due to its strong coupling to the @xmath6 channel and its short lifetime of only 1.3 fm / c , much shorter than the lifetime of the fireball .
theoretical predictions exist for changes both of the mass and the width which are tied directly @xcite or indirectly @xcite to chiral symmetry restoration .
while broadening of the @xmath1 is generally expected , changes of the mass are controversial . for masses @xmath70.9 gev / c@xmath5 , other hadronic processes set in , and partonic processes start to play a larger role as well . the first na60 results focused on the space - time averaged spectral function of the @xmath1 @xcite ; more details were added recently @xcite .
the present paper therefore concentrates on new developments : an analysis of the shape of the excess mass spectra @xcite , and in particular first results on acceptance - corrected @xmath2 and @xmath8 spectra for different mass windows .
mass and @xmath2 spectra are also compared to the latest theoretical developments @xcite .
details of the na60 apparatus are contained in @xcite , while the different analysis steps ( including the critical assessment of the combinatorial background from @xmath9 and @xmath10 decays through event mixing ) are described in @xcite .
the results reported here were obtained from the analysis of data taken in 2003 for in - in at 158 agev . the left part of fig .
[ fig1 ] shows the opposite - sign , background and signal dimuon mass spectra , integrated over all collision centralities . after subtracting the combinatorial background and the signal fake matches ,
the resulting net spectrum contains about 360000 muon pairs in the mass range 0 - 2 gev / c@xmath11 , roughly 50% of the total available statistics .
the associated average charged - particle multiplicity density measured by the vertex tracker is @xmath12 = 120 .
the vector mesons @xmath3 and @xmath4 are completely resolved ; the mass resolution at the @xmath3 is 20 mev / c@xmath5 .
a major part of the subsequent analysis is done in four classes of collision centrality ( defined through the charged - particle multiplicity density ) : peripheral ( 4@xmath1330 ) , semiperipheral ( 30@xmath13110 ) , semicentral ( 110@xmath13170 ) and central ( 170@xmath13240 ) .
the peripheral data can essentially be described by the expected electromagnetic decays of the neutral mesons , i.e. the 2-body decays of the @xmath14 , @xmath1 , @xmath3 and @xmath4 resonances and the dalitz decays of the @xmath14 , @xmath15 and @xmath3 @xcite .
this is not true in the more central bins , due to the existence of a strong excess . to isolate this excess with _ a priori unknown _ characteristics without any fits ,
we have used a novel procedure as shown in fig .
[ fig1 ] ( right ) , made possible by the high data quality . for each centrality bin ,
the cocktail of the decay sources is subtracted from the data by using solely _ local _ criteria @xcite .
the @xmath16 on the contrary , is not subtracted .
examples for the resulting difference spectra are contained in fig .
[ fig1 ] ( right ) for all centralities and in fig .
[ fig6 ] for the semicentral bin .
the qualitative features of the spectra are striking : a peaked structure is always seen , residing on a broad continuum with a yield strongly increasing with centrality , but remaining essentially centered around the position of the nominal @xmath1 pole .
further details on the subtraction procedure and the results , including a discussion of the systematic errors ( which reach up to 25% in the continuum region ) , can be found in @xcite .
a more quantitative analysis of the shape of the excess mass spectra vs. centrality has recently been performed , using a finer subdivision of the data into 12 centrality bins .
the data ( fig .
[ fig1 ] , right ) were subdivided into 3 windows , one central ( c ) in the mass range 0.64@xmath17@[email protected] gev / c@xmath5 , and two adjacent ones on the left ( l ) and the right ( r ) with equal widths .
the spectral shape can then roughly be characterized by two yields : c-(l+r)/2 for the @xmath1-like peak , and 3(l+r)/2 for the underlying continuum .
the ratios peak/@xmath1 , continuum/@xmath1 and peak / continuum ( where @xmath1 stands for the cocktail @xmath1 ) are plotted in the left part of fig .
[ fig2 ] , compared to the rms of the total mass interval 0.44@xmath17@[email protected] gev / c@xmath5 on the right .
three essential features should be noted .
first , the ratio peak/@xmath1 decreases from the most peripheral to the most central bin by nearly a factor of 2 , ruling out the naive view that the shape can simply be explained by the cocktail @xmath1 residing on a broad continuum , independent of centrality .
second , the sum of the ratios ( peak+continuum)/@xmath1 is the total enhancement factor relative to the cocktail @xmath1 ; it reaches about 5.5 in the most central bin .
third , the centrality dependences are largely consistent in the two plots .
in particular , the flat part up to dn@xmath19/dy=100 is followed by more rapid changes in all variables , statistically highly significant for the ratio continuum/@xmath1 and the rms . for different mass windows .
right : rapidity distribution of the excess data for the mass window 0.6@xmath17@[email protected] gev/@xmath20 and for three selected @xmath2 bins . the measured data ( full markers ) are reflected around midrapidity ( open markers).,title="fig:",scaledwidth=45.0% ] for different mass windows .
right : rapidity distribution of the excess data for the mass window 0.6@xmath17@[email protected] gev/@xmath20 and for three selected @xmath2 bins .
the measured data ( full markers ) are reflected around midrapidity ( open markers).,title="fig:",scaledwidth=45.0% ] dependences on @xmath21 were so far shown only as mass spectra associated with three different @xmath21 windows , without acceptance correction @xcite .
we have now obtained acceptance - corrected @xmath21 spectra associated with three different mass windows .
the na60 acceptance relative to @xmath22 as a function of @xmath21 is shown in fig .
[ fig3 ] ( left ) . in principle
, the acceptance correction requires a 3-dimensional grid in @xmath23 space .
however , the low populated bins in the phase space corners can introduce large errors once the correction is applied . to overcome this problem ,
the correction is performed in 2-dimensional @xmath24 space , using the measured rapidity distribution as an input .
the latter was determined with an acceptance correction found , in an iterative way , from monte carlo simulations matched to the data in mass and @xmath21 . on the basis of this rapidity distribution ,
0.1 gev/@xmath25 bins in mass and 0.2 gev/@xmath26 bins in @xmath21 were used to determine the remaining 2-dimensional correction .
once corrected , the results were integrated over the three extended mass windows 0.4@xmath17@[email protected] , 0.6@xmath17@[email protected] and 1.0@xmath17@[email protected] .
gev / c@xmath5 .
the rapidity distribution of the central mass window is shown in fig .
[ fig3 ] ( right ) for three different @xmath21 windows .
it is interesting to note that it resembles closely the distribution of inclusive pion production , as measured by na49 for pb - pb .
the results for the acceptance - corrected @xmath2 spectra are summarized in fig .
the errors shown are purely statistical .
systematic errors arise from the acceptance corrections including the rapidity distribution used , the subtraction of the cocktail , and the subtraction of the combinatorial background plus fake matches . for @[email protected] gev / c ,
the combinatorial background contributes most , ranging from 10 to 25% for semiperipheral up to central . for @xmath2@xmath71 gev / c ,
the statistical errors dominate .
the data show a significant dependence on mass , but hardly on centrality . to bear out the differences in mass more clearly
, the data were summed over the three more central bins and plotted vs. transverse mass @xmath27 in place of @xmath21 , as shown in fig .
[ fig5 ] ( left ) .
the inverse slope parameter @xmath28 as determined from differential fits of the @xmath8 spectra with @xmath29 , using a sliding window in @xmath2 , is plotted on the right .
the main features of the results are remarkable and somewhat unexpected . at very low @xmath8 ,
all spectra steepen rather than flatten , as expected from radial flow , equivalent to very small values of @xmath28 ( the @xmath4 resonance , placed just in between the upper two mass windows , _ does flatten _ as expected ) .
moreover , depending on the fit region , @xmath28 covers an unusually large dynamic range .
finally , the largest masses have the steepest @xmath8 spectrum , i.e. the smallest value of @xmath28 everywhere , again contrary to radial flow and to what is usually observed for hadrons .
all this suggests that different mass regions are coupled to basically different emission sources .
the qualitative features of the excess mass spectra shown above are consistent with an interpretation as direct thermal radiation from the fireball , dominated by @xmath0 annihilation .
as explained in detail in @xcite , the excess mass spectra , with _ no _ acceptance correction and no @xmath21 selection , can be approximately interpreted as the spectral function of the intermediate @xmath1 , averaged over momenta and the complete space - time history of the fireball . as already discussed in the first na60 publications ( see @xcite and earlier ref .
cited ibd . ) , the moving - mass scenario la brown / rho scaling @xcite was found to be incompatible with the experimental excess mass spectra . even considering fireball variations within extremes ,
this has basically remained unchanged .
other scenarios , characterized essentially by a broadening of the @xmath1 , have in the meantime been refined @xcite or newly developed @xcite .
a summary of the comparison of these latest theoretical results to the excess mass spectrum for the semicentral bin is contained in fig . [ fig6 ] .
hees / rapp @xcite have modified the fireball part of the original predictions for in - in in the 2@xmath9 region ( based on @xcite ) and have now also added 4@xmath9 processes , reflecting vector - axialvector mixing ; the latter makes a strong contribution for @xmath18@xmath71 gev / c@xmath5 .
renk / ruppert @xcite use the spectral function of eletsky et al . including baryonic effects in the 2@xmath9 region , coupled to a strong partonic contribution which dominates the yield for @xmath18@xmath71 gev / c@xmath5 .
dusling / zahed @xcite use a chiral virial approach including baryonic interactions , coupled again to some partonic contribution which is on a similar level as the hadronic yield for @xmath18@xmath71 gev / c@xmath5 .
all models achieve a reasonable description of the mass spectrum over the whole mass range , even in absolute terms . for @[email protected]
gev / c@xmath5 , the decisive role played by baryonic interactions in @xcite is seen in the upper right of fig .
[ fig6 ] by switching them off . for @xmath18@xmath71 gev / c@xmath5
, the agreement of all predictions with the data implies a low ( if any ) sensitivity to the dynamics here , equivalent to hadron - parton duality .
the @xmath27 spectra predicted by the models above are compared to the data in fig .
in contrast to the mass spectra , none of the predictions is presently able to provide a satisfactory description of the @xmath8 spectra over their full range .
it seems that the @xmath8 spectra are , in some sense , more sensitive to the dynamics than the mass spectra .
a quantitative understanding of the @xmath8 spectra may thus help to discriminate between the different sources and provide further insight beyond hadron - parton duality ( e.g. different radial flow effects for hadrons and partons ) .
summarizing , the previously measured excess mass spectra and the new acceptance - corrected @xmath2 and @xmath8 spectra present rather strict boundary conditions to theoretical modeling . beyond the @xmath1 spectral
function itself , this may also lead to a better understanding of the continuum part of the spectra . 00
r. d. pisarski , phys .
* 110b * , 155 ( 1982 ) g. e. brown , m. rho , phys . rept . *
363 * , 85 ( 2002 ) r. rapp and j. wambach , adv .
. phys . * 25 * , 1 ( 2000 ) r. arnaldi _ et al . _
( na60 collaboration ) , phys .
* 96 * ( 2006 ) 162302 s. damjanovic _ et al . _
( na60 collaboration ) ( 2006 ) nucl - ex/0609026 h. van hees and r. rapp , phys .
* 97 * ( 2006 ) 102301 t. renk and j. ruppert , arxiv : hep - ph/0612113 ; t. renk _ et al .
_ , in preparation k. dusling , d. teaney and i. zahed , arxiv : nucl - th/0604071 g. usai _ et al . _
( na60 collaboration ) , eur .
j. * c43 * , 415 ( 2005 ) m. keil _ et al .
_ , nucl .
instrum .
a539 * , 137 ( 2005 ) and * a546 * , 448 ( 2005 ) r. shahoyan _ et al . _
( na60 collaboration ) , nucl .
* a774 * , 677 ( 2006 ) | the na60 experiment at the cern sps has studied low - mass muon pairs in 158 agev in - in collisions . a strong excess of pairs is observed above the yield expected from neutral meson decays .
after subtraction of the decay sources , the shape of the resulting mass spectrum is largely consistent with a dominant contribution from @xmath0 annihilation .
the associated @xmath1 spectral function exhibits considerable broadening , but essentially no shift in mass .
the acceptance - corrected @xmath2 spectra have a shape atypical for radial flow .
they also significantly depend on mass , pointing to different sources in different mass regions .
both mass and @xmath2 spectra are compared to recent theoretical predictions . for the na60 collaboration : , relativistic heavy - ion collisions , quark - gluon plasma , lepton pairs 25.75.-q , 12.38.mh , 13.85.qk |
let s start with a game : `` hex '' is a board game for two players , invented by the ingenious danish poet , designer and engineer piet hein in 1942 @xcite , and rediscovered in 1948 by the mathematician john nash @xcite who got a nobel prize in economics in 1994 ( for his work on game theory , but not really for this game ) .
hex , in hein s version , is played on a rhombical board , as depicted in the figure .
@xmath0 the rules of the game are simple : there are two players , whom we call white and black .
the players alternate , with white going first .
each move consists of coloring one `` grey '' hexagonal tile of the board white resp .
white has to connect the white borders of the board ( marked @xmath1 and @xmath2 ) by a path of his white tiles , while black tries to connect @xmath3 and @xmath4 by a black path .
they ca nt both win : any winning path for white separates the two black borders , and conversely .
( this is nt hard to prove however , the statement is closely related to the jordan curve theorem , which is trickier than it may seem when judged at first sight : see exercise [ exer : jct ] . ) however , here we concentrate on the opposite statement : there is no draw possible when the whole board is covered by black and white tiles , then there always is a winner .
( this is even true if one of the players has cheated badly and ends up with much more tiles than his / her opponent !
it is also true if the board is nt really `` square , '' that is , if it has sides of unequal lenghts . )
our next figure depicts a final hex position sure enough one of the players has won , and the proof of the following `` hex theorem '' will give us a systematic method to find out which one .
@xmath5 [ the hex theorem ] if every tile of an @xmath6-hex board is colored black or white , then either there is a path of white tiles that connects the white borders @xmath1 and @xmath2 , or there is a path of black tiles that connects the black borders @xmath3 and @xmath4 .
our plan for this section is the following : we give a simple proof of the hex theorem .
we show that it implies the brouwer fixed point theorem , and even conversely : the brouwer fixed point theorem implies the hex theorem .
then we prove that one of the players has a winning strategy .
and then we see that on a square board , the first player can win , while on an uneven board , the player with the longer borders has a strategy to win .
all of this is really quite simple , but it nicely illustrates how a topological theorem enters the analysis of a discrete situation . we follow a certain path _ between _ the black and white tiles that starts in the lower left - hand corner of the hex board on the edge that separates @xmath1 and @xmath3 .
whenever this path reaches a corner of degree three , there will be both colors present at the corner ( due to the edge we reach it from ) , and so there will be a unique edge to proceed on that does have different colors on its two sides .
@xmath7 our path can never get stuck or branch or turn back onto itself : otherwise we would have found a vertex that has one or three edges that separate colors , whereas this number clearly has to be even at each vertex .
thus the path can be continued until it leaves the board that is , until it reaches @xmath2 or @xmath4 .
but that means that we find a path that connects @xmath1 to @xmath2 , or @xmath3 to @xmath4 , and on its sides keeps a white path of tiles resp . a black path .
that is , one of white and black has won ! now this was easy , and ( hopefully ) fun .
we continue with a re - interpretation of the hex board in nash s version that buys us two drinks for the price of one : * a @xmath8-dimensional version of the hex theorem , and * the connection to the brouwer fixed point theorem .
the @xmath8-dimensional _ hex board _ is the graph @xmath9 with vertex set @xmath10 , in which two vertices @xmath11 are connected if @xmath12 .
the _ colors _ for the @xmath8-dimensional hex game are @xmath13 , where we identify `` @xmath14white '' and `` @xmath15black . ''
the _ interior _ of the hex board is given by @xmath16 .
all the other vertices , in @xmath17 , form the _ boundary _ of the board .
the vertices in the boundary of @xmath9 get preassigned colors
@xmath18 @xmath19 our drawing depicts the @xmath20-dimensional hex board @xmath21 , which represents a dual graph for the @xmath22-board that we used in our previous figures , with the preassigned colors on the boundary .
the @xmath8-dimensional hex game is played between @xmath8 players who take turns in coloring the interior vertices of @xmath9 .
the @xmath23-th player _ wins _ if he achieves a path of vertices of color @xmath23 that connects a vertex whose @xmath23-th coordinate is @xmath24 to a vertex whose @xmath23-th coordinate is @xmath25 .
there is no draw possible for @xmath8-dimensional hex : if all interior vertices of @xmath26 are colored , then at least one player has won .
the proof that we used for @xmath20-dimensional hex still works : it just has to be properly translated for the new setting .
for this we first check that @xmath9 is the graph of a triangulation @xmath27 of @xmath28^d$ ] , which is given by the _ clique complex _ of @xmath9 : that is , a set of lattice points @xmath29 forms a simplex in @xmath27 if and only if the points in @xmath30 are pairwise connected by edges .
( to check this , verify that each point @xmath31^d$ ] lies in the relative interior of a unique simplex , which is given by @xmath32 & & \lfloor x_i - x_j\rfloor\le v_i - v_j\le\lceil x_i - x_j\rceil \mbox{\rm~for all $ i\neq j$}\big\}.\end{aligned}\ ] ] our picture illustrates the @xmath20-dimensional case . ) now every full - dimensional simplex in @xmath27 has @xmath33 vertices .
a simplex @xmath30 in @xmath27 is _ completely colored _ if it has all @xmath8 colors on its vertices .
thus each completely colored @xmath8-simplex in @xmath34 has exactly two completely colored facets , which are @xmath35-faces of the complex @xmath27 .
conversely , every completely colored @xmath35-face is contained in exactly two @xmath8-simplices if it is not on the boundary of @xmath28^d$ ] . with this the ( constructive ) proof that we gave before for the @xmath20-dimensional hex theorem generalizes to the following : we start at the @xmath8-simplex @xmath36 for which the facet ( @xmath35-face ) @xmath37 is completely colored .
this simplex is shaded in the following figure for @xmath21 , which depicts the same final position that we considered before .
@xmath38 now we construct a sequence of completely colored @xmath8-dimensional simplices that starts at @xmath39 : we find the second completely colored @xmath35-face of @xmath39 , find the second completely colored @xmath8-simplex it is contained in , etc .
thus we find a chain of completely colored @xmath8-simplices that ends on the boundary of @xmath40^d$]at a different simplex than the one we started from .
in particular , the last @xmath8-simplex in the chain has a completely colored facet ( a @xmath35-face ) in the boundary , and by construction this facet has to lie in a hyperplane @xmath41 .
( in fact , at this point we check that every completely colored @xmath35-simplex in the boundary of @xmath9 is contained in one of the hyperplanes @xmath42 , with the sole exception of the boundary facet of our starting @xmath8-simplex . ) and the chain of @xmath8-simplices thus provides us with an @xmath23-colored path from the @xmath23-colored vertex @xmath43 to the @xmath23-colored vertex in @xmath42 : so the @xmath23-th player wins .
our drawing illustrates the chain of completely colored simplices ( shaded ) and the sequence of ( white ) vertices for the winning path that we get from it .
now we will proceed from the discrete mathematics setting of the hex game to the continuous world of topological fixed point theorems . here
are three versions of the brouwer fixed point theorem .
[ t : brouwer ] the following are equivalent ( and true ) : every continuous map @xmath45 has a fixed point .
every continuous map @xmath46 has a fixed point .
every null - homotopic map @xmath47 has a fixed point .
@xmath48 ( the term _ null - homotopic _ that appears here refers to a map that can be deformed to a constant map ; see the proof below . )
( br1)@xmath49(br2 ) is trivial , since @xmath50 . for ( br2)@xmath49(br3 ) let @xmath51\to s^{d-1}$ ]
be a null - homotopy for @xmath52 , i.e. , a continuous map that interpolates between our original map @xmath52 and a constant map , with @xmath53 and @xmath54 for all @xmath55 .
from this we construct a continuous map @xmath56 that extends @xmath52 , by @xmath57 @xmath58 this map is continuous , and by ( br2 ) it has a fixed point , which must lie in the image , that is , in @xmath59 . for the converse , ( br3)@xmath49(br2 )
, let @xmath46 be continuous .
then the restriction @xmath60 is null - homotopic , since @xmath61 provides a null - homotopy .
thus by ( br3 ) @xmath60 has a fixed point , hence so does @xmath52 .
finally , we get ( br2)@xmath49(br1 ) : if @xmath45 has no fixed point , then we set @xmath62 .
this defines a map @xmath63 that has a fixed point @xmath64 by ( br2 ) , with @xmath65 . but this implies @xmath66 for @xmath67 , and this is impossible for @xmath64 . in the following we use the unit cube @xmath68^d$ ] instead of the ball @xmath69
: it should be clear that the brouwer fixed point theorem equally applies to self - maps of any domain @xmath70 that is homeomorphic to @xmath69 , resp . of the boundary @xmath71 of such a domain .
if @xmath72^d\to [ 0,1]^d$ ] has no fixed point , then for some @xmath73 we have that @xmath74 for all @xmath75^d$ ] ( namely , one can take @xmath76^d\}$ ] , which exists since @xmath68^d$ ] is compact ) . furthermore , any continuous function on the compact set @xmath68^d$ ] is uniformly continuous ( see e.g. munkres @xcite ) , hence there exists some @xmath77 such that @xmath78 implies @xmath79 .
we take @xmath80 ( without loss of generality ) , and then choose @xmath81 with @xmath82 . from @xmath52
, we now define a coloring of @xmath9 , by setting @xmath83 for the interior vertices @xmath84 , where @xmath85 denotes the @xmath23th component of @xmath52 .
this is well - defined , since @xmath86^d$ ] , and thus at least one component of @xmath87 has to be at least @xmath88 in its absolute value .
now the @xmath8-dimensional hex theorem guarantees , for some @xmath23 , a chain @xmath89 of vertices of color @xmath23 , where @xmath90 and @xmath91 .
furthermore , we know @xmath92 for @xmath93 . and at the ends we of the chain know the signs : * @xmath94^d$ ] implies @xmath95 and hence @xmath96 , and * @xmath97^d$ ] implies @xmath98 and hence @xmath99 .
hence , for some @xmath100 we must have a sign change : * @xmath101 and @xmath102 .
all this taken together provides a contradiction , since * @xmath103 whereas @xmath104 assume we have a coloring of @xmath9 .
we use it to define a map @xmath105^d\to [ 0,n]^d$ ] , as follows : on the points in @xmath106 we define @xmath107 & \text{that connects $ \vv$ to a vertex $ \ww$ with $ w_i=0$}\\ \vv-\ee_i & \text{if
$ \vv$ has color $ i$ , but there is no such path . }
\end{cases}\ ] ] if for the given coloring there is no winning path for hex , then these definitions do not map any point @xmath108 outside @xmath105^d$ ] .
hence this defines a simplicial map @xmath109^d\to [ 0,n]^d$ ] , by linear extension on the simplices of the triangulation @xmath27 that we have considered before .
the following two observations now give us a contradiction , showing that this @xmath52 can not have a fixed point : if @xmath110 is a simplex and @xmath111 is a linear map defined by @xmath112 , then @xmath52 has a fixed point on @xmath34 if and only if @xmath113 .
if @xmath114 are adjacent vertices , then we can not get @xmath115 and @xmath116 .
hence for each simplex of @xmath27 , all the vectors @xmath117 lie in one orthant of @xmath118 ! 1 . in the proof of the brouwer fixed point theorem ( thm .
[ t : brouwer ] , ( br2)@xmath49(br3 ) ) , we could simply have put @xmath119 .
is this continuous ?
so , who can win the @xmath20-dimensional hex game ?
a simple but ingenious argument due to john nash , known as `` stealing a strategy , '' shows that on a square board the first player ( `` white '' ) always has a winning strategy . in the following we first define winning strategies ,
then find that one of the players has one , and finally conclude that the first player has one .
still : the proof will be non - constructive , and we do nt know how to win hex .
so , the game still remains interesting
a _ strategy _ is a set of rules that tells a player which move to make ( i.e. , which tile to color ) for every legal position on the board .
a _ winning strategy _ here guarantees to lead to a win , starting from an empty board , for all possible moves of the opponent .
a _ position _ of the hex game is a board on which some tiles may have been colored white or black , together with the information who moves next ( unless all tiles are colored ) .
a position is _ legal _ if it can occur in a hex game : that is , if either white moves next , and the numbers of white and black tiles agree , or if black moves next , and white has one more tile . a _ winning position for white _ is a position such that white has a _ winning strategy _ that tells him how to proceed ( for arbitrary moves of black ) and guarantees a win .
similarly , _ a winning position for black _ has a _ winning strategy _ that guarantees to lead black to a win .
[ nodraw ] every ( legal ) position for hex is either a winning position for white or a winning position for black .
here we proceed by induction on the number @xmath120 of `` grey '' tiles ( i.e. , `` free '' positions on the board ) .
if no grey tiles are present @xmath121 , then one of the players has won by the hex theorem .
if @xmath122 and white is to move , then any move that white could make reduces @xmath120 , and thus ( by induction ) produces a winning position for one of the players .
if there is a move that leads to a winning position for white , then this is really nice and great for white : this makes the present position into a winning position for white , and any such move can be used for a winning position for white .
otherwise too bad : if every possible move for white produces a winning position for black , then we are at a winning position for black already .
and the same argument applies for @xmath122 if black is to move .
of course , the argument given here is _
much _ more general : essentially we have proved that for any finite deterministic 2-person game without a draw and with `` complete information '' there is a winning strategy for one of the players .
( this is a theorem of zermelo , which was rediscovered by von neumann and morgenstern ) .
furthermore , for games where a draw is possible either one player has a winning strategy , or _ both _ players can force a draw .
we refer to exercise [ ex : drawplay ] , and to blackwell & girshick @xcite . for hex
, lemma [ nodraw ] shows that at the beginning ( for the starting position , where all tiles are grey , and white is to move ) , there is a winning strategy either for white or for black .
but who is the winner ? our first attempt might be to follow the proof of lemma [ nodraw ] . only for the @xmath123 board
this can be done : in this drawing , you can decide for every position whether it is a winning position for white or for black , starting with the bottom row ( @xmath124 ) that has three winning positions for each player , ending at the top node ( @xmath125 ) , which turns out to be a winning position for white . for larger boards
, this approach is hopeless
after all , there are @xmath126 final positions to classify for `` @xmath124 , '' and from this one would have work one s way up to the top node of a huge tree ( of height @xmath127 ) .
nevertheless , people have worked out winning strategies for white on the @xmath128 boards for @xmath129 ( see gardner @xcite ) . for the hex game
played on a hex board with equal side lengths , white ( the first player ) has a winning strategy .
assume not : then by lemma [ nodraw ] black has a winning strategy .
but then white can start with an arbitrary move , and then using the symmetry of the board and of the rules just ignore his first tile , and follow black s winning strategy `` for the second player . ''
this strategy will tell white always which move to take . here
the `` extra '' white tiles can not hurt white : if the move for white asks to occupy a tile that is already white , then an arbitrary move is o.k .
for white .
but this `` stealing a strategy argument '' produces a winning strategy for white , contradicting our assumption ! gale s beautiful paper @xcite was the source and inspiration for our treatment of brouwer s fixed point theorem in terms of the hex game .
nash s analysis for the winning strategies for hex is from gardner s classical account in @xcite , some of which reappears in milnor s @xcite .
see also the accounts in jensen & toft ( * ? ? ?
17.14 ) , and in berlekamp , conway & guy @xcite , where other cases of `` strategy stealing '' are discussed .
( a theoretical set - up for this is in hales & jewett ( * ? ? ?
* sect . 3 ) .
) the traditional combinatorial approach to the brouwer fixed point theorem is via sperner s lemma @xcite ; see e.g. the presentation in @xcite .
a more geometric version of the combinatorial lemmas is given by mani @xcite .
stir your coffee cup .
show that the ( moving , but flat ) surface has at every moment at least one point that stands still ( has velocity zero ) .
2 . prove that if you tear a sheet of paper from your notebook , crumble it into a small ball , and put that down on your notebook , then at least one point of the sheet comes to rest exactly on top of its original position .
+ could it happen that there are exactly two such points ?
3 . for hex on a @xmath130 board ,
how large is the tree of possible positions ? 4 .
can you write a computer program that plays hex and wins ( sometimes ) @xcite ? 5 . for @xmath8-dimensional hex , is there always some `` short '' winning path ? show that for every @xmath131 there is a constant @xmath132 such that for all @xmath81 there is a final configuration such that only one player wins , but his shortest path uses more than @xmath133 tiles .
construct an algorithm that , for given @xmath134 and @xmath72 ^ 2\to [ 0,1]^2 $ ] , calculates a point @xmath135 ^ 2 $ ] with @xmath136 .
@xcite 7 .
[ ex : drawplay ] if in a complete information two player game a draw is possible , argue why either one of the players has a winning strategy , or _
both _ can force at least a draw
[ exer : jct ] prove that for @xmath20-dimensional hex , not both players can win ! for this , prove and use the `` polygonal jordan curve theorem '' : any simple closed polygon in the plane uniquely divides the plane into `` inside '' and `` outside . ''
+ ( the general jordan curve theorem for simple `` jordan arcs '' in the plane has extensive discussions in many books ; see for example munkres @xcite , stillwell ( * ? ? ?
* sect . 0.3 ) , or thomassen @xcite . ) 9 . on an @xmath137-board that is not square ( that is , @xmath138 ) , the player who gets the longer sides , and hence the shorter distance to bridge by a winning path , has a winning strategy .
( our figure illustrates the case of a @xmath139-board , where the claim is that black has a winning strategy . ) * show that for this , it is sufficient to consider the case where @xmath140 ( i.e. , the second player black , who gets the longer side , has a sure win ) . @xmath141 * show that in the situation of ( i ) , black has the following winning strategy .
label the tiles in the `` symmetric '' way that is indicated by the figure , such there are two tiles of each label .
the strategy for black is to always take the second tile that has the same label as the one taken by white .
why will this strategy always win for black ?
( hint : you will need the jordan curve theorem . )
+ ( this is in gardner @xcite and in milnor @xcite , but neither source gives the proof .
you ll have to work yourself ! )
let @xmath142 be a system of sets on a ground set @xmath143 ; both @xmath142 and @xmath143 may generally be infinite . the _ packing number _ of @xmath142 , usually denoted by @xmath144 and
often also called the _ matching number _ , is the maximum cardinality of a system of pairwise disjoint sets in @xmath142 : @xmath145 the _ transversal number _ or _
piercing number _ of @xmath142 is the smallest number of points of @xmath143 that capture all the sets in @xmath142 : @xmath146 a subsystem @xmath147 of pairwise disjoint sets is usually called a _ matching _ ( this refers to the graph - theoretical matching , which is a system of pairwise disjoint edges ) , and a set @xmath148 intersecting all sets of @xmath142 is referred to as a _ transversal _ of @xmath142 .
clearly , any transversal is at least as large as any matching , and so always @xmath149 in the reverse direction , very little can be said in general , since @xmath150 can be arbitrarily large even if @xmath151 . as a simple geometric example
, we can take the plane as the ground set of @xmath142 and let the sets of @xmath142 be lines in general position .
then @xmath152 , since every two lines intersect , but @xmath153 , because no point is contained in more than two of the lines .
one of the basic general questions in combinatorics asks for interesting special classes of set systems where the transversal number can be bounded in terms of the matching number . is the system of all open sets in a topological space
, @xmath150 is the minimum size of a dense set and is called the _ density _ , while @xmath144 is known as the _ souslin number _ or _
cellularity _ of the space . in 1920
, souslin asked whether a linearly ordered topological space exists ( the open sets are unions of open intervals ) with countable @xmath154 but uncountable @xmath155 .
it turned out in the 1970s that the answer depends on the axioms one is willing to assume beyond the usual ( zfc ) axioms of set theory .
for example , it is yes if one assumes the continuum hypothesis ; see e.g.@xcite .
] many such examples come from geometry . here
we restrict our attention to one particular type of systems , the _ @xmath8-intervals _ , where the best results have been obtained by topological methods . before introducing @xmath8-intervals
, we mention another important parameter of a set system , which always lies between @xmath154 and @xmath155 and often provides useful estimates for @xmath154 or @xmath155 .
this parameter can be introduced in two seemingly different ways . for simplicity , we restrict ourselves to finite set systems ( on possibly infinite ground sets ) .
a _ fractional packing _ for a finite set system @xmath142 on a ground set @xmath143 is a function @xmath156 $ ] such that for each @xmath157 , we have @xmath158 . the _ size _ of a fractional packing @xmath159 is @xmath160 , and the _ fractional packing number _ @xmath161 is the supremum of the sizes of all fractional packings for @xmath142 .
so in a fractional packing , we can take , say , one - third of one set and two - thirds of another , but at each point , the fractions for the sets containing that point must add up to at most 1 .
we always have @xmath162 , since a packing @xmath163 defines a fractional packing @xmath159 by setting @xmath164 for @xmath165 and @xmath166 otherwise .
similar to the fractional packing , one can also introduce a fractional version of a transversal .
a _ fractional transversal _ for a ( finite ) set system @xmath142 on a ground set @xmath143 is a function @xmath167 $ ] attaining only finitely many nonzero values such that for each @xmath168 , we have @xmath169 .
the size of a fractional transversal @xmath170 is @xmath171 , and the _ fractional transversal number _
@xmath172 is the infimum of the sizes of fractional transversals . by the duality of linear programming ( or by the theorem about separation of disjoint convex sets by a hyperplane ) , it follows that @xmath173 for any finite set system @xmath142 . when trying to bound @xmath155 in terms of @xmath154 , in many instances it proved very useful to bound @xmath174 as a function of @xmath154 first , and then @xmath155 in terms of @xmath175 .
the proof presented below follows a somewhat similar approach .
let @xmath176 be disjoint parallel segments of unit length in the plane .
a set @xmath177 is a _
@xmath8-interval _ if it intersects each @xmath178 in a closed interval .
we denote this intersection by @xmath179 and call it the _ @xmath23th component _ of @xmath180 .
the drawing shows a @xmath181-interval : intersection and piercing for @xmath8-intervals are taken in the set - theoretical sense , i.e. two @xmath8-intervals intersect if , for some @xmath23 , their @xmath23th components intersect , etc .
the @xmath182-intervals , which are just intervals in the usual sense , behave nicely with respect to packing and piercing : for any family @xmath183 of intervals , we have @xmath184 ( this is well - known and easy to prove ) . the following family @xmath183 of three 2-intervals has @xmath185 while @xmath186 . by taking multiple copies of this family ,
one obtains families with @xmath187 for all values of @xmath154 .
gyrfs and lehel @xcite showed by elementary methods that for any @xmath8 and any family @xmath183 of @xmath8-intervals , @xmath188 can be bounded by a function of @xmath189 ( also see @xcite ) .
their function was rather large ( about @xmath190 for @xmath8 fixed ) .
after an initial breakthrough by tardos @xcite , who proved @xmath191 for any family of 2-intervals , kaiser @xcite obtained the following result : [ t : d - int ] every family @xmath183 of @xmath8-intervals , @xmath131 , has a transversal of size at most @xmath192 . here
we present a proof using brouwer s fixed point theorem .
alon @xcite found a short non - topological proof of the slightly weaker bound @xmath193 .
let @xmath183 be a fixed system of @xmath8-intervals with @xmath194 , and let @xmath195 be a suitable ( yet undetermined ) integer .
the general plan of the proof is this : assuming that there is no transversal of @xmath183 of size @xmath196 , we show by a topological method that the fractional packing number @xmath197 is at least @xmath198 . then a simple combinatorial argument proves that the packing number @xmath189 is at least @xmath199 , which leads to @xmath200 .
an sharper combinatorial reasoning in this step leads to the slightly better bound in the theorem .
our candidates for a transversal of @xmath183 are all sets @xmath201 with each @xmath202 having exactly @xmath203 points ; so @xmath204 . for technical reasons
, we also permit that some of the @xmath203 points in @xmath178 coincide , so @xmath201 can be a multiset .
the letter @xmath201 could also abbreviate a _ trap_. the trap is set to catch all the @xmath8-intervals in @xmath183 , but if it is not set well enough , some of the @xmath8-intervals can escape . each of them escapes through a hole in the trap , namely through a _
@xmath8-hole_. the points of @xmath205 cut the segment @xmath178 into @xmath198 open intervals ( some of them may be empty ) , and these are the _ holes in @xmath178 _ ; they are numbered 1 through @xmath198 from left to right .
a @xmath8-hole consists of @xmath8 holes , one in each @xmath178 .
the _ type _ of a @xmath8-hole @xmath206 is the set @xmath207 , where @xmath208 $ ] is the number of the hole in @xmath178 contained in @xmath206 .
a @xmath8-interval @xmath209 _ escapes _ through a @xmath8-hole @xmath206 if it is contained in the union of its holes .
the drawing shows a @xmath181-hole , of type @xmath210 , and a @xmath181-interval escaping through it : let @xmath211 be the hypergraph with vertex set @xmath212\times [ t{+}1]$ ] and with edges being all possible types of @xmath8-holes ; for example , the hole in the picture yields the edge @xmath210 .
so @xmath211 is a complete @xmath8-partite @xmath8-uniform hypergraph ( we will meet such hypergraphs several times in this book ) . by saying that a @xmath209 escapes through an edge @xmath206 of @xmath211 , we mean that @xmath180 escapes through the @xmath8-hole ( uniquely ) corresponding to @xmath206 .
next , we define weights on the edges of @xmath211 ; these weights depend on the set @xmath201 ( and also on @xmath183 but this is considered fixed ) .
the weight of an edge @xmath206 is @xmath213 here @xmath214 and @xmath215 is the distance of the @xmath23th component of @xmath180 to the closest point of @xmath205 .
thus , @xmath216 can be interpreted as the slimmest margin by which a @xmath8-interval @xmath180 escaping through @xmath206 avoids being trapped .
if no members of @xmath183 escape through @xmath206 , we define @xmath216 as 0 .
note that this is the only case where @xmath217 ; otherwise , if anything escapes , it does so by a positive margin , since we are dealing with closed intervals . from the edge weights ,
we derive weights of vertices : the weight @xmath218 of a vertex @xmath219 is the sum of the weights of the edges of @xmath211 containing @xmath220 . these weights ,
too , are functions of @xmath201 ; to emphasize this , we can write @xmath221 .
[ l : samewt ] for any @xmath222 , @xmath223 , and any @xmath183 , there is a choice of @xmath201 such that all the vertex weights @xmath224 , @xmath225\times[t{+}1]$ ] , coincide .
it is this lemma whose proof is topological .
we postpone that proof and finish the combinatorial part .
let us suppose that a trap @xmath201 was chosen as in the lemma , with @xmath226 for all @xmath220 . if @xmath227 then @xmath201 is a transversal , since all edge weights are 0 and no @xmath209 escapes .
so suppose that @xmath228 .
let @xmath229 , the _ escape hypergraph _ of @xmath201 , consist of the edges of @xmath211 with nonzero weights .
note that @xmath230 indeed , given a matching @xmath163 in @xmath231 , for each edge @xmath232 choose a @xmath209 escaping through @xmath206this gives a matching in @xmath183 .
we note that the re - normalized edge weights @xmath233 determine a fractional packing in @xmath231 ( since the weights at each vertex sum up to 1 ) . for the size of this fractional packing , which is
the total weight of all vertices , we find by double counting @xmath234\times [ t{+}1]}\frac{w_v}w= \frac 1d \sum_{v } 1 = t+1.\ ] ] the last step is to show that @xmath235 can not be small if @xmath236 is large . here is a simple argument leading to a slightly suboptimal bound , namely @xmath237 . given a fractional matching @xmath238 of size @xmath198 in @xmath231 , a matching can be obtained by the following greedy procedure : pick an edge @xmath239 and discard all edges intersecting it , pick @xmath240 among the remaining edges , etc .
, until all edges are exhausted .
the @xmath238-weight of @xmath241 plus all the edges discarded with it is at most @xmath242 , while all edges together have weight @xmath198 .
thus , the number of steps , and also the size of the matching @xmath243 , is at least @xmath244 .
if we set @xmath245 , we get @xmath246 , which contradicts ( [ e : nuhnuf ] ) . therefore ,
for this choice of @xmath203 , all the vertex weights must be 0 and @xmath201 as in lemma [ l : samewt ] is a transversal of @xmath183 of size at most @xmath247 .
the improved bound @xmath248 for @xmath249 follows similarly using a theorem of fredi @xcite , which implies that the matching number of any @xmath8-uniform @xmath8-partite hypergraph @xmath231 satisfies @xmath250 .
( for @xmath251 , a separate argument needs to be used , based on a theoreom of lovsz stating that @xmath252 for all graphs @xmath253 . ) the tardos kaiser theorem [ t : d - int ] is proved .
let @xmath254 denote the standard @xmath203-dimensional simplex in @xmath255 , i.e. the set @xmath256 .
a point @xmath257 defines a @xmath203-point multiset @xmath258 $ ] , @xmath259 , by setting @xmath260 .
here is a picture for @xmath261 : a candidate transversal @xmath201 with @xmath203 points in each @xmath178 can thus be defined by an ordered @xmath8-tuple @xmath262 of points , @xmath263 , where @xmath264 determines @xmath205 .
such an ordered @xmath8-tuple can be regarded as a single point @xmath265 in the cartesian product @xmath266 . to each @xmath267 , we have thus assigned a candidate transversal @xmath268 . for each vertex @xmath219 of the hypergraph @xmath211
, we define the function @xmath269 by @xmath270 , where @xmath224 is the vertex weight .
this is a continuous function of @xmath265 .
we note that for each @xmath265 , the sum @xmath271 is independent of @xmath23 ; this is because @xmath272 equals the sum of the weights of all edges .
so we can write just @xmath273 instead of @xmath272 . if there is an @xmath267 with @xmath274 , then all the vertex weights @xmath275 are 0 and we are done .
otherwise , we define the normalized functions @xmath276 for each @xmath23 , @xmath277 are nonnegative and sum up to 1 , and so they are the coordinates of a point in the standard simplex @xmath254 .
all the maps @xmath278 together can be regarded as a map @xmath279 . to prove the lemma
, we need to show that the image of @xmath52 contains the point of @xmath280 with all the @xmath281 coordinates equal to @xmath282 .
the product @xmath280 is a convex polytope , and its nonempty faces are exactly all cartesian products @xmath283 , where @xmath284 are nonempty faces of @xmath254 ( exercise [ ex : polytprod ] ) .
we note that for any face @xmath285 of @xmath280 , we have @xmath286 . indeed , a face @xmath253 of @xmath254 has the form @xmath287 , for some index set @xmath288 , and the faces of @xmath280 are products of faces @xmath253 of this form .
so it suffices to know that @xmath289 whenever @xmath290 .
this holds , since @xmath290 means that the @xmath291th hole in @xmath178 is empty , so nothing can escape through that hole , and thus @xmath289 .
the proof of lemma [ l : samewt ] is now reduced to the following statement : [ l : polyt - surj ] let @xmath280 be a convex polytope and let @xmath279 be a continuous mapping satisfying @xmath286 for each face for each facet of @xmath280 ( that is , for each face of dimension @xmath292 ) , since each face is the intersection of some facets . ] @xmath285 of @xmath280 .
then @xmath52 is surjective .
since the condition is hereditary for faces , it suffices to show that each point @xmath293 in the interior of @xmath280 has a preimage . for contradiction , suppose that some @xmath294 is not in the image of @xmath52 . for @xmath267
, consider the semiline emanating from @xmath295 and passing through @xmath293 , and let @xmath296 be the unique intersection of that semiline with the boundary of @xmath280 .
this @xmath120 is a well - defined and continuous map @xmath297 , and by brouwer s fixed point theorem , there is an @xmath298 with @xmath299 .
the point @xmath300 lies on the boundary of @xmath280 , in some proper face @xmath285 .
but @xmath301 can not lie in @xmath285 , because the segment @xmath302 passes through the point @xmath293 outside @xmath285a contradiction .
it turns out that the bound in theorem [ t : d - int ] is not far from being the best possible .
in particular , for @xmath185 and @xmath8 large , the transversal number can be near - quadratic in @xmath8 , which is rather surprising . for all @xmath303 and @xmath8 , systems
@xmath183 of @xmath8-intervals can be constructed with @xmath194 and @xmath304 for a suitable constant @xmath305 ( matouek @xcite ) .
the construction involves an extension of a construction due to sgall @xcite of certain systems of set pairs .
here we outline a ( non - topological ! ) proof of a somewhat simpler result concerning families of _ homogeneous _
@xmath8-intervals , which are unions of at most @xmath8 closed intervals on the real line .
these are more general than the @xmath8-intervals , but an upper bound only slightly weaker than theorem [ t : d - int ] can be proved for them along the same lines ( exercise [ ex : homog ] ) : @xmath306 .
[ p : d - int - lb ] for every @xmath307 and @xmath308 , there exists a system @xmath183 of homogeneous @xmath8-intervals with @xmath194 and @xmath309 given @xmath8 and @xmath303 , we want to construct a system @xmath183 of homogeneous @xmath8-intervals .
clearly , it suffices to consider the case @xmath310 , since for larger @xmath303 , we can take @xmath303 disjoint copies of the @xmath183 constructed for @xmath310 .
thus , we want an @xmath183 in which every two @xmath8-intervals intersect and with @xmath188 large . in the construction , we will use homogeneous @xmath8-intervals of a quite special form : each component is either a single point or a unit - length interval .
first , it is instructive to see why we can not get a good example if all the components are only points . in that case
, the family @xmath183 is simply a @xmath8-uniform hypergraph ( whose vertices happen to be points of the real line ) .
we require that any two edges intersect , and thus any edge is a transversal and we have @xmath311 . for the actual construction , let @xmath81 and @xmath312 be integer parameters ( whose value will be set later ) .
let @xmath313 $ ] be an index set , and @xmath314 , for @xmath315 , be auxiliary pairwise disjoint unit intervals on the real line . in each @xmath314
, we choose @xmath312 distinct points @xmath316 , @xmath317 . the constructed system @xmath183 consists of homogeneous @xmath8-intervals @xmath318
. for each @xmath317 , we choose auxiliary sets @xmath319 , and we construct @xmath320 as follows : @xmath321 the picture shows an example of @xmath322 for @xmath323 , @xmath324 and @xmath325 : @xmath326 the heart of the proof is the construction of suitable sets @xmath327 and @xmath328 on the ground set @xmath329 . since the @xmath320 should be homogeneous @xmath8-intervals , we obviously require 1 . for all @xmath317 , @xmath330 and @xmath331 .
the condition that every two members of @xmath183 intersect is implied by the following : 1 . for all @xmath332 , @xmath333
, we have @xmath334 or @xmath335 ( or both ) .
finally , we want @xmath183 to have no small transversal .
since no two @xmath8-intervals of @xmath183 have a point component in common , a transversal of size @xmath203 intersects no more than @xmath203 members of @xmath183 in their point components , and all the other members of @xmath183 must be intersected in their interval components .
therefore , the transversal condition translates to 1 .
put @xmath336 for a sufficiently small constant @xmath305 , and let @xmath337 .
then @xmath338 , and consequently @xmath339 for any @xmath340 arising from @xmath341 by removing at most @xmath203 sets .
a construction of sets @xmath342 and @xmath343 as above was provided by sgall @xcite .
his results give the following : [ p : sgall ] let @xmath344 be a given integer , let @xmath345 for a sufficiently small constant @xmath305 , and let @xmath346 be @xmath344-element subsets of @xmath313 $ ] . then there
exist sets @xmath347 , with @xmath348 , @xmath349 , and such that ( c2 ) is satisfied . with this proposition , the proof of proposition [ p : d - int - lb ] is easily finished .
we set @xmath350 , @xmath351 , and we let @xmath346 be all the @xmath352 subsets of @xmath329 of size @xmath344 .
we have @xmath353 and condition ( c3 ) holds .
it remains to construct the sets @xmath327 according to proposition [ p : sgall ] ; then ( c1 ) and ( c2 ) are satisfied too .
the proof of proposition [ p : d - int - lb ] is concluded by passing from the @xmath327 and @xmath328 to the system @xmath183 of homogeneous @xmath8-intervals as was described above .
let @xmath354 be a graph on @xmath81 vertices of maximum degree @xmath344 with the following expander - type property : for any two disjoint @xmath344-element subsets @xmath355 , there is at least one edge @xmath356 connecting a vertex of @xmath357 to a vertex of @xmath3 .
( the existence of such a graph can be easily shown by the probabilistic method ; the constant @xmath358 arises in this argument .
see @xcite for references . ) for each @xmath23 , let @xmath359 be an ( arbitrary ) element of the set @xmath328 , and let @xmath360 where @xmath361 denotes the set of neighbors in @xmath253 of a vertex @xmath315 .
it is easy to check that @xmath349 , and some thought reveals that the condition ( c2 ) is satisfied .
kaiser and rabinovich @xcite investigated conditions on a family @xmath183 of @xmath8-intervals guaranteeing that @xmath183 can be pierced by a `` multipoint , '' i.e. @xmath362 and there is a transversal using one point of each @xmath178 .
they proved the following : [ t : multipoint ] let @xmath363 and let @xmath183 be a family of @xmath8-intervals such that any @xmath303 or fewer members of @xmath183 have a common point .
then @xmath183 can be pierced by a multipoint .
we use notation from the proof of theorem [ t : d - int ] .
we apply lemma [ l : samewt ] with @xmath364 , obtaining a set @xmath201 with one point in each @xmath205 such that all the @xmath365 vertices of the escape hypergraph @xmath366 have the same weight @xmath1 .
if @xmath227 we are done , so let us assume @xmath228 . by the assumption on @xmath183 ,
every @xmath303 edges of @xmath231 share a common vertex . we will prove the following claim for every @xmath367 : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
if every @xmath368 edges of @xmath231 have at least @xmath369 common vertices , then every @xmath367 edges of @xmath231 have at least @xmath370 common vertices . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for @xmath371 ,
the assumption holds with @xmath372 , and so by @xmath373-fold application of this claim , we get that every edge of @xmath231 `` intersects itself '' in at least @xmath374 vertices , i.e. @xmath375 . the claim thus implies the theorem .
the claim is proved by contradiction .
suppose that @xmath376 is a set of @xmath367 edges such that @xmath377 has at most @xmath378 vertices .
recall that the vertices of @xmath231 are pairs @xmath379
, @xmath380 $ ] .
let @xmath381 ( note that @xmath382 never contains both @xmath383 and @xmath384 , since no edge of @xmath231 does ) . by the assumption , @xmath385 plus
any other edge together intersect in at least @xmath369 vertices .
thus , any @xmath386 contains at least @xmath369 vertices of @xmath382 , and consequently no more than @xmath369 vertices of @xmath387 .
let @xmath1 be the total weight of the vertices in @xmath382 and @xmath388 the total weight of the vertices in @xmath387 .
the edges in @xmath385 contribute solely to @xmath1 , while any other edge @xmath206 contributes at least as much to @xmath1 as to @xmath388 , and so @xmath389 .
but this is impossible since all vertex weights are identical and @xmath390 .
the claim , and theorem [ t : multipoint ] too , are proved .
an interesting open problem is whether @xmath363 in theorem [ t : multipoint ] could be replaced by @xmath391 for some constant @xmath392 independent of @xmath8 .
the best known lower bound is @xmath393 .
tardos @xcite proved the optimal bound @xmath394 for 2-intervals by a topological argument using the homology of suitable simplicial complexes .
kaiser s argument @xcite is similar to the presented one , but he proves lemma [ l : samewt ] using a rather advanced borsuk ulam - type theorem of ramos @xcite concerning continuous maps defined on products of spheres .
the method with brouwer s theorem was used by kaiser and rabinovich @xcite for a proof of theorem [ t : multipoint ] .
alon s short proof @xcite of the bound @xmath395 for families of @xmath8-intervals applies a powerful technique developed in alon and kleitman @xcite .
for the so - called hadwiger
debrunner @xmath396-problem solved in the latter paper , the quantitative bounds are probably quite far from the truth .
it would be interesting to find an alternative topological approach to that problem , which could perhaps lead to better bounds .
see , for example , hell @xcite . the variant of the piercing problem for families of homogeneous @xmath8-intervals has been considered simultaneously with @xmath8-intervals ( @xcite , @xcite , @xcite , @xcite ) .
the upper bounds obtained for the homogeneous case are slightly worse : @xmath397 for homogeneous @xmath20-intervals , which is tight , and @xmath306 for homogeneous @xmath8-intervals , @xmath398 @xcite .
the reason for the worse bounds is that the escape hypergraph needs no longer be @xmath8-partite , and so fredi s theorem @xcite relating @xmath154 to @xmath174 gives a little worse bound ( for @xmath251 , one uses a theorem of lovsz instead , asserting that @xmath399 for any graph ) .
sgall s construction @xcite answered a problem raised by wigderson in 1985 .
the title of sgall s paper refers to a different , but essentially equivalent , formulation of the problem dealing with labeled tournaments .
alon @xcite proved by the method of @xcite that if @xmath201 is a tree and @xmath183 is a family subgraphs of @xmath201 with at most @xmath8 connected components , then @xmath193 .
more generally , he established a similar bound for the situation where @xmath201 is a graph of bounded tree - width ( on the other hand , if the tree - width of @xmath201 is sufficiently large , then one can find a system of connected subgraps of @xmath201 with @xmath152 and @xmath155 arbitrarily large , and so the tree - width condition is also necessary in this sense ) . a somewhat weaker bound for trees has been obtained independently by kaiser @xcite .
[ ex : polytprod ] let @xmath280 and @xmath400 be convex polytopes . show that there is a bijection between the nonempty faces of the cartesian product @xmath401 and all the products @xmath402 , where @xmath285 is a nonempty face of @xmath280 and @xmath253 is a nonempty face of @xmath400 .
show that the following `` brouwer - like '' claim resembling lemma [ l : polyt - surj ] is _ not _ true : if @xmath403 is a continuous map of the @xmath81-ball such that the boundary of @xmath404 is mapped surjectively onto itself , then @xmath52 is surjective .
[ ex : homog ] prove the bound @xmath405 for any family of _ homogeneous
_ @xmath8-intervals ( unions of @xmath8 intervals on a single line ) .
hint : follow the proof for @xmath8-intervals above , but encode a candidate transversal @xmath201 by a point of a simplex ( rather than a product of simplices ) .
fixed point theorems are `` global - local tools '' : from global information about a space ( such as its homology ) they derive local effects , such as the existence of special points where `` something happens . ''
of course , in application to combinatorial problems we need to combine them with suitable `` continuous - discrete tools '' : from continous effects , such as topological information about continuous maps of simplicial complexes , we have to find our way back to combinatorial information .
in addition to the usual game of graphs , posets , complexes and spaces , we will in the following exploit the deep topological effects caused by symmetry , that is , by finite group actions . a ( finite ) group @xmath253 _ acts _ on a ( finite ) simplicial complex for the polyhedron ( the geometric realization of a simplicial complex @xmath406 ) . ] @xmath406 if each group element corresponds to a permutation of the vertices of @xmath406 , where composition of group elements corresponds to composition of permutations , in such a way that @xmath407 is a face of @xmath406 for all @xmath408 and for all @xmath409 .
this action on the vertices is extended to the geometric realization of the complex @xmath406 , so that @xmath253 acts as a group of simplicial homeomorphisms @xmath410 .
the action is _ faithful _ if only the identity element in @xmath253 acts as the identity permutation . in general ,
the set @xmath411 is a normal subgroup of @xmath253 .
hence we get that the quotient group @xmath412 acts faithfully on @xmath406 , and we usually only consider faithful actions . in this case , we can interpret @xmath253 as a subgroup of the _ symmetry group _ of the complex @xmath406 .
the action is _ vertex transitive _ if for any two vertices @xmath413 of @xmath406 there is a group element @xmath408 with @xmath414 .
a _ fixed point _ ( also known as _ stable point _ ) of a group action is a point @xmath415 that satisfies @xmath416 for all @xmath408 .
we denote the set of all fixed points by @xmath417 . note : this is not in general a subcomplex of @xmath406 .
let @xmath418}$ ] be the complex of a triangle , and let @xmath419 be the cyclic group ( a proper subgroup of the symmetry group @xmath420 ) , acting such that a generator cyclically permutes the vertices , @xmath421 .
@xmath422{2 } } \put(5446,-3166){\makebox(0,0)[lb]{1 } } \put(7156,-5956){\makebox(0,0)[lb]{3 } } \end{picture}\ ] ] this is a faithful action ; its fixed point set consists of the center of the triangle only this is not a subcomplex of @xmath406 , although it corresponds to a subcomplex of the barycentric subdivision @xmath423 .
[ l : fps ] after replacing @xmath406 by its barycentric subdivision , we get that the fixed point set @xmath417 is a subcomplex of @xmath406 . after replacing @xmath406 by its barycentric subdivision once again
, we even get that the quotient space @xmath424 can be constructed from @xmath406 by identifying all faces with their images under the action of @xmath253 ; that is , the equivalence classes of faces of @xmath406 , with the induced partial order , form a simplicial complex that is homeomorphic to the quotient space @xmath424 .
we leave the proof as an exercise .
it is not difficult ; for details and further discussion see bredon ( * ? ? ? * sect .
iii.1 ) . a powerful tool on our agenda is hopf s trace theorem .
let @xmath329 be any finite - dimensional vector space @xmath329 , or a free abelian group of finite rank .
when we consider an endomorphism @xmath425 then the _ trace _ @xmath426 is the sum of the diagonal elements of the matrix that represents @xmath120 .
the trace is independent of the basis chosen for @xmath329 . in the case
when @xmath329 is a free abelian group , then @xmath426 is an integer .
[ the hopf trace theorem][t : hopf - trace ] let @xmath427 be a self - map , and denote by @xmath428 resp .
@xmath429 the maps that @xmath52 induces on @xmath23-dimensional chain groups resp .
homology groups . using an arbitrary field of coefficients @xmath303
, one has @xmath430 the same identity holds if we use integer coefficients , and compute the traces for homology in the quotients @xmath431 of the homology groups modulo their torsion subgroups ; these quotients are free abelian groups .
the proof for this uses the definition of simplicial homology , and simple linear algebra ; we refer to munkres ( * ? ? ?
22.1 ) or bredon ( * ? ? ?
iv.23 ) . for an arbitrary coefficient field @xmath303
, we define the _ lefschetz number _ of the map @xmath432 as @xmath433 similarly , taking integral homology modulo torsion , we define the _ integral lefschetz number _ as @xmath434 the universal coefficient theorems imply that one always has @xmath435 : thus the integral lefschetz number @xmath436 can be computed in rational homology , but it is an integer
. the _ euler characteristic _ of a complex @xmath406 coincides with the lefschetz number of the identity map @xmath437 , @xmath438 thus the hopf trace theorem yields that the euler - characteristic of a finite simplicial complex @xmath406 can be defined resp . computed without a reference to homology , simply as the alternating sum of the face numbers of the complex @xmath406 , where @xmath439 denotes the number of @xmath23-dimensional faces of @xmath406 : @xmath440 this is then a finite sum that ends with @xmath441 if @xmath406 has dimension @xmath8 .
thus the hopf trace theorem applied to the identity map just reproduces the euler
poincar formula .
this proves , for example , the @xmath8-dimensional euler polyhedron formula , not only for polytopes , but also for general spheres , shellable or not ( see ziegler @xcite ) . for us
the main consequence of the trace formula is the following theorem . [
the lefschetz fixed point theorem][t : lefschetzhopf ] let @xmath406 be a finite simplicial complex , and @xmath303 an arbitrary field .
if a self - map @xmath427 has lefschetz number @xmath442 , then @xmath52 and any map homotopic to @xmath52 have a fixed point . in particular , if @xmath406 is @xmath443-acyclic for some prime @xmath444 , then every continuous map @xmath427 has a fixed point . for a finite simplicial complex @xmath406 ,
the polyhedron @xmath445 is compact .
so if @xmath52 does not have a fixed point , then one has @xmath446 such that @xmath447 for all @xmath448 .
now take a subdivision into simplices of diameter smaller than @xmath449 , and a simplicial approximation of error smaller than @xmath450 , so that the simplicial approximation does not have a fixed point , either .
now apply the trace theorem , where the induced map @xmath451 in @xmath452-dimensional homology is the identity .
note that brouwer s fixed point theorem [ t : brouwer ] is the special case of theorem [ t : lefschetzhopf ] when @xmath406 triangulates a ball . for a reasonably large class of spaces ,
a converse to the lefschetz fixed point theorem is also true : if @xmath453 , then @xmath52 is homotopic to a map without fixed points .
see brown ( * ? ? ?
viii ) .
`` smith theory '' was started by p.a .
smith @xcite in the thirties we refer to bredon ( * ? ? ?
* chapter iii ) for a nice textbook treatment .
smith theory analyses finite group actions on compact spaces ( such as finite simplicial complexes ) , providing relations between the structure of the group to its possible fixed point sets .
here is one key result .
[ t : smith ] if @xmath280 is a @xmath444-group ( that is , a finite group of order @xmath454 for a prime @xmath444 and some @xmath455 ) , acting on a complex @xmath406 that is @xmath443-acyclic , then the fixed point set @xmath456 is @xmath443-acyclic as well .
in particular , it is not empty .
the key is that , with the preparations of lemma [ l : fps ] , the maps that @xmath52 induces on the chain groups ( with @xmath443 coefficients ) nicely restrict to the chain groups on the fixed point set @xmath456 .
passing to traces and using the hopf trace theorem , one can derive that @xmath456 is non - empty .
a more detailed analysis leads to the `` transfer isomorphism '' in homology , which proves that @xmath456 must be acyclic .
see bredon ( * ? ? ?
iii.5.2 ) and oliver @xcite , and also de longueville ( * ? ? ?
* appendix d and e ) . on the combinatorial side
, one has an euler characteristic relation due to floyd @xcite ( * ? ?
iii.4 ) : @xmath457 it implies that if @xmath280 is a @xmath444-group ( in particular , if @xmath458 ) , then @xmath459 using induction on @xmath203 , where @xmath454 .
[ t : cyclic - oliver ] if @xmath460 is a cyclic group , acting on a @xmath461-acyclic complex @xmath406 , then the action has a fixed point .
in fact , in this case the fixed point set @xmath417 has the euler characteristic of a point , @xmath462 .
the first statement follows directly from the lefschetz fixed point theorem : any cyclic group is generated by a single element @xmath120 , this element has a fixed point , this fixed point of @xmath120 is also a fixed point of all powers of @xmath120 , and hence of the whole group @xmath253 . for the second part , take @xmath463 to be a maximal prime power that divides @xmath81 , consider the corresponding subgroup isomorphic to @xmath464 , and use induction on @xmath203 and the transfer homomorphism , as for the previous proof .
unfortunately , results like these may give an overly optimistic impression of the generality of fixed point theorems for acyclic complexes .
there are fixed point free finite group actions on balls : examples were constructed by floyd & richardson and others ; see bredon ( * ? ? ?
i.8 ) . on the positive side
we have the following result due to oliver , which will play a central role in the following section .
[ oliver s theorem i ( * ? ? ? * prop .
i)][t : oliver ] if @xmath253 has a normal subgroup @xmath465 that is a @xmath444-group , such that the quotient @xmath466 is cyclic , acting on a complex @xmath406 that is @xmath443-acyclic , then the fixed point set @xmath417 is @xmath443-acyclic as well . in particular
, it is not empty .
this is as much as we will need in this chapter .
oliver proved , in fact , a more general and complete theorem that includes a converse .
[ t : oliver2 ] let @xmath253 be a finite group .
every action of @xmath253 on a @xmath443-acyclic complex @xmath406 has a fixed point if and only if @xmath253 has the following structure : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ @xmath253 has normal subgroups @xmath467 such that @xmath280 is a @xmath444-group , @xmath468 is a @xmath469-group ( for a prime @xmath469 that need not be distinct from @xmath444 ) , and the quotient @xmath470 is cyclic . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in this situation one always has @xmath471 . the lefschetz
hopf fixed point theorem was announced by lefschetz for a restriced class of complexes in 1923 , with details appearing three years later . the first proof for the general version was by hopf in 1929 .
there are generalizations , for example to absolute neighborhood retracts ; see bredon ( * ? ? ?
iv.23.5 ) and brown ( * ? ? ?
we refer to brown s book @xcite .
[ argument complexity of a set system ; evasiveness][d : ask ] in the following , we are concerned with a fixed , known set system @xmath472 , and with the complexity of deciding whether some set @xmath473 is in the set system . here
our `` model of computation '' is such that given , and known , : : is a set system @xmath472 , where @xmath474 is fixed , @xmath475 .
: : on the other hand , there is a fixed , but unknown : : subset @xmath476 .
we have to decide : : whether @xmath477 , using only questions : : of the type `` is @xmath478 ? '' ( it is assumed that we always get correct answers yes or no .
we only count the _ number _ of questions that are needed in order to reach the correct conclusion : it is assumed that it is not difficult to decide whether @xmath478 .
you can assume that some `` oracle '' that knows both @xmath357 and @xmath183 is answering . ) clearly @xmath480 .
the set system @xmath183 is _ trivial _ if @xmath481 : then no questions need to be asked ; this can only be the case if @xmath482 or if @xmath483 .
( otherwise @xmath183 is _ non - trivial_. ) for example , if @xmath485 , then @xmath484 : if we again and again get the answer no , then we have to test all the elements to be sure that @xmath486 .
so @xmath485 is an evasive set system : `` being empty '' is an evasive set property .
[ graph properties][d : gp ] for this we consider graphs on a fixed vertex set @xmath313 $ ] .
loops and multiple edges are excluded .
thus any graph @xmath487 is determined by its edge set @xmath357 , which is a subset of the set @xmath488 of `` potential edges . ''
we identify a _ property _
@xmath489 of graphs with the family of graphs that have the property @xmath489 , and thus with the set family @xmath490 given by @xmath491,a)\mbox { has property } \pp\}.\ ] ] furthermore , we will consider only graph properties that are isomorphism invariant ; that is , properties of abstract graphs that are preserved under renumbering the vertices . with the symmetry condition of definition [ d : gp ] , we would accept `` being connected '' , `` being planar , '' `` having no isolated vertices , '' and `` having even vertex degrees '' as graph properties . however ,
`` vertex @xmath182 is not isolated , '' `` @xmath492 is a triangle , '' and `` there are no edges between odd - numbered vertices '' are not graph properties .
having no edge : : : clearly we have to check every single @xmath356 in order to be sure that it is not contained in @xmath357 , so this property is evasive : its argument complexity is @xmath493 .
having at most @xmath303 edges : : : let us assume that we ask questions , and the answer we get is yes for the first @xmath303 questions , and then we get no answers for all further questions , except for possibly the last one . assuming that @xmath494
, this implies that the property is evasive .
otherwise , for @xmath495 , the property is trivial .
being connected : : : this property is evasive for @xmath496 .
convince yourself that for any strategy , a sequence of `` bad '' answers can force you to ask all the questions .
being planar : : : this property is trivial for @xmath497 but evasive for @xmath498 . in fact , for @xmath499 one has to ask all the questions ( in arbitrary order ) , and the answer will be @xmath500 unless we get a yes answer for all the questions including the last one .
this is , however , not at all obvious for @xmath501 : it was claimed by hopcroft & tarjan @xcite , and proved by best , van emde boas & lenstra ( * ? ? ?
* example 2 ) @xcite . a large star : : : let @xmath489 be the property of being a disjoint union of a star @xmath502 and an arbitrary graph on @xmath181 vertices , and
let @xmath183 be the corresponding set system .
@xmath503 then @xmath504 for @xmath505 . for
@xmath506 we can easily see this , as follows .
test all the @xmath507 edges @xmath508 with @xmath509 . that way we will find exactly one vertex @xmath303 with at least @xmath510 neighbors ( otherwise property @xmath489 can not be satisfied ) : that vertex @xmath303 has to be the center of the star .
we test all other edges adjacent to @xmath303 : we must find that @xmath303 has exactly @xmath511 neighbors .
thus we have identified three vertices that are not neighbors of @xmath303 : at least one of the edges between those three has not been tested .
we test all other edges to check that @xmath512,a)$ ] has property @xmath489 .
( this property was found by l. carter ( * ? ? ?
* example 16 ) . )
being a scorpion graph : : : a _ scorpion graph _ is an @xmath81-vertex graph that has one vertex of degree @xmath182 adjacent to a vertex of degree @xmath20 whose other neighbor has degree @xmath513 .
we leave it as an ( instructive ! ) exercise to check that `` being a scorpion graph '' is not evasive if @xmath81 is large : in fact , best , van emde boas & lenstra ( * ? ? ?
* example 18 ) @xcite have shown that @xmath514 .
@xmath515 from these examples it may seem that most `` interesting '' graph properties are evasive .
in fact , many more examples of evasive graph properties can be found in bollobs ( * ? ? ? * sect .
viii.1 ) , alongside with techniques to establish that graph properties are evasive , such as milner & welsh s `` simple strategy '' @xcite . why is this model of interest
? finite graphs ( similarly for digraphs and bipartite graphs ) can be represented in different types of _ data structures _ that are not at all equivalent for algorithmic applications .
for example , if a finite graph is given by an adjacency list , then one can decide fast ( `` in linear time '' ) whether the graph is planar , e.g. using an old algorithm of hopcroft & tarjan @xcite ; see also mehlhorn ( * ? ? ?
iv.10 ) and @xcite .
note that such a planar graph has at most @xmath516 edges ( for @xmath517 ) . however , assume that a graph is given in terms of its adjacency matrix @xmath518 where @xmath519 means that @xmath508 is an edge of @xmath253 , and @xmath520 says that @xmath508 is not an edge . here
@xmath253 is faithfully represented by the set of all @xmath521 superdiagonal entries ( with @xmath522 ) .
then one possibly has to inspect a large part of the matrix until one has enough information to decide whether the graph in question is planar .
in fact , if @xmath472 is the set system corresponding to all planar graphs , then @xmath479 is exactly the number of superdiagonal matrix entries that every algorithm for planarity testing has to inspect in the worst case .
the statement that `` being planar '' is evasive ( for @xmath498 ) thus translates into the fact that every planarity testing algorithm that starts from an adjacency matrix needs to read at least @xmath523 bits of the input , and hence its running time is bounded from below by @xmath524 .
this means that such an algorithm such as the one considered by fisher @xcite cannot run in linear time , and thus can not be efficient . for digraph properties we again use the fixed vertex set
@xmath313 $ ] .
loops and parallel edges are excluded , but anti - parallel edges are allowed .
thus any digraph @xmath487 is determined by its arc set @xmath357 , which is a subset of the set @xmath525 of all @xmath526 `` potential arcs '' ( corresponding to the off - diagonal entries of an @xmath128 adjacency matrix ) .
a _ digraph property _ is a property of digraphs @xmath512,a)$ ] that is invariant under relabelling of the vertex set .
equivalently , a digraph property is a family of arc sets @xmath527 that is symmetric under the action of @xmath528 that acts by renumbering the vertices ( and renumbering all arcs correspondingly ) .
a digraph property is _ evasive _ if the associated set system is evasive , otherwise it is _ non - evasive_. for bipartite graph properties we use a fixed vertex set @xmath529 of size @xmath530 , and use @xmath531 as the set of potential edges .
bipartite graph property _ is a property of graphs @xmath532 with @xmath533 that is preserved under renumbering the vertices in @xmath329 , and also under permuting the vertices in @xmath1 .
equivalently , a bipartite graph property on @xmath534 is a set system @xmath535 that is stable under the action of the automorphism group @xmath536 that acts transitively on @xmath534 .
having at most @xmath303 arcs : : : again , this is clearly evasive with @xmath484 if @xmath537 , and trivial otherwise .
having a sink : : : a _ sink _ in a digraph on @xmath81 vertices is a vertex @xmath303 for which all arcs going into @xmath303 are present , but no arc leaves @xmath303 , that is , a vertex of out - degree @xmath538 , and in - degree @xmath539 .
let @xmath183 be the set system of all digraphs on @xmath81 vertices that have a sink .
it is easy to see that @xmath540 . in particular , for @xmath517 `` having a sink '' is a non - trivial but non - evasive digraph property .
+ in fact , if we test whether @xmath541 , then either we get the answer yes , then @xmath23 is not a sink , or we get the answer no , then @xmath291 is not a sink .
so , by testing arcs between pairs of vertices that `` could be sinks , '' after @xmath542 questions we are down to one single `` candidate sink '' @xmath303 . at this point at least one arc adjacent to @xmath303 has been tested .
so we need at most @xmath543 further questions to test whether it is a sink .
[ history : the aanderaa rosenberg conjecture][arc ] originally , arnold l. rosenberg had conjectured that all non - trivial digraph properties have quadratic argument complexity , that is , that there is a constant @xmath544 such that for all non - trivial properties of digraphs on @xmath81 vertices one has @xmath545 .
however , s. aanderaa found the counter - example ( for digraphs ) of `` having a sink '' ( * ? ? ?
* example 15 ) @xcite .
we have also seen that `` being a scorpion graph '' is a counter - example for graphs .
hence rosenberg modified the conjecture : at least all _ monotone _ graph properties , that is , properties that are preserved under deletion of edges , should have quadratic argument complexity .
this is the statement of the _ aanderaa rosenberg conjecture _ @xcite .
richard karp considerably sharpened the statement , as follows .
we will prove this below for graphs and digraphs in the special case when @xmath81 a prime power ; from this one can derive the aanderaa
rosenberg conjecture , with @xmath546 .
similarly , we will prove that monotone properties of bipartite graphs on a fixed ground set @xmath547 are evasive ( without any restriction on @xmath548 and @xmath549 ) .
however , we first return to the more general setting of set systems .
any strategy to determine whether an ( unknown ) set @xmath357 is contained in a ( known ) set system @xmath183as in definition [ d : ask]can be represented in terms of a decision tree of the following form .
a _ decision tree _ is a rooted , planar , binary tree whose leaves are labelled `` yes '' or `` no , '' and whose internal nodes are labelled by questions ( here they are of the type `` @xmath550 '' ) .
its edges are labelled by answers : we will represent them so that the edges labelled `` yes '' point to the right child , and the `` no '' edges pointing to the left child .
a _ decision tree for @xmath472 _ is a decision tree such that starting at the root with an arbitrary @xmath473 , and going to the right resp .
left child depending on whether the question at an internal node we reach has answer yes or no , we always reach a leaf that correctly answers the question `` @xmath551 '' .
@xmath552 the root of a decision tree is at _ level _ @xmath452 , and the children of a node at level @xmath23 have level @xmath553 .
the _ depth _ of a tree is the greatest @xmath303 such that the tree has a vertex at level @xmath303 ( a leaf ) .
a decision tree for @xmath183 is _ optimal _ if it has the smallest depth among all decision trees for @xmath183 ; that is , if it leads us to ask the smallest number of questions for the worst possible input .
let us consider an explicit example .
@xmath554 the following figure represents an optimal algorithm for the `` sink '' problem on digraphs with @xmath555 vertices .
this has a ground set @xmath556 of size @xmath557 .
the algorithm first asks , in the root node at level @xmath452 , whether @xmath558 . in case
the answer is yes ( so we know that @xmath182 is not a sink ) , it branches to the right , leading to a question node at level @xmath182 that asks whether @xmath559 , etc . in case
the answer to the question @xmath558 is no ( so we know that @xmath20 is not a sink ) , it branches to the left , leading to a question node at level @xmath182 that asks whether @xmath560 , etc .
for every possible input @xmath357 ( there are @xmath561 different ones ) , after two questions we have identified a unique `` candidate sink '' ; after not more than @xmath562 question nodes one arrives at a leaf node that correctly answers the question whether the graph @xmath563 has a sink node : yes or no .
( the number of the unique candidate is noted next to each node at level @xmath20 . )
@xmath564 for each node ( leaf or inner ) of level @xmath303 , there are exactly @xmath565 different inputs that lead to this node .
this proves the following lemma .
another way to view a ( binary ) decision tree algorithm is as follows . in the beginning
, we do not know anything about the set @xmath357 , so we can view the collection of possible sets as the complete boolean algebra of all @xmath568 subsets of @xmath474 . in the first node ( at `` level @xmath452 '' ) we ask a question of the type `` @xmath550 '' ; this induces a subdivision of the boolean algebra into two halves , depending on whether we get answer yes or no .
each of the halves is an interval of length @xmath569 of the boolean algebra @xmath570 . at level @xmath182
we ask a new question , depending on the outcome of the first question .
thus we _ independently _ bisect the two halves of level @xmath452 , getting four pieces of the boolean algebra , all of the same size . @xmath571
this process is iterated .
it stops we do not need to ask a further question on the parts which we create that either contain only sets that are in @xmath183 ( this yields a yes - leaf ) or that contain only sets not in @xmath183 ( corresponding to no - leaves ) . thus the final result is a special type of partition of the boolean algebra into intervals . some of them are yes intervals , containing only sets of @xmath183 , all the others are no - intervals , containing no sets from @xmath183 .
if the property in question is monotone , then the union of the yes intervals ( i.e. , the set system @xmath183 ) forms an _ ideal _ in the boolean algebra , that is , a `` down - closed '' set such that with any set that it contains it must also contain all its subsets .
consider one interval @xmath576 in the partition of @xmath577 that is induced by any optimal algorithm for @xmath183 . if the leaf , at level @xmath303 , corresponding to the interval is reached through a sequence of @xmath578 yes - answers and @xmath579 no - answers ( with @xmath580 )
, then this means that there are sets @xmath581 with @xmath582 and @xmath583 with @xmath584 , such that @xmath585 in other words , the interval @xmath576 contains all sets that give yes - answers when asked about any of the @xmath578 elements of @xmath586 , no - answers when asked about any of the @xmath579 elements of @xmath587 , while the @xmath588 elements of @xmath589 may or may not be contained in @xmath357 .
thus the interval @xmath576 has size @xmath590 , and its counting polynomial is @xmath591 now the complete set system @xmath183 is a disjoint union of the intervals @xmath576 , and we get @xmath592 in particular , for an optimal decision tree we have @xmath593 and thus @xmath594 at every leaf of level @xmath303 , which means that all the summands @xmath595 have a common factor of @xmath596 .
we can now draw the conclusion , based only on simple counting , that most set families are evasive .
this can not of course be used to settle any specific cases , but it can at least make the various evasiveness conjectures seem more plausible .
let @xmath609 be any @xmath303-orbit of @xmath253 , that is , a collection of @xmath303-sets @xmath610 on which @xmath253 acts transitively .
while every set in @xmath609 contains @xmath303 elements @xmath356 , we know from transitivity that every element of @xmath474 is contained in the same number , say @xmath8 , of sets of the orbit @xmath609 . thus , double - counting the edges of the bipartite graph on the vertex set @xmath611 defined by `` @xmath478 '' ( displayed in the figure below ) we find that @xmath612 .
thus for @xmath613 we have that @xmath444 divides @xmath614 , while @xmath615 is one single `` trivial '' orbit of size @xmath182 , and @xmath616 does nt appear .
hence we have @xmath617 which implies evasiveness by corollary [ c : euler - char ] .
* @xmath623 , so we have @xmath624 * @xmath625 and all images under @xmath626 ,
that is , all singleton sets : @xmath627 , * @xmath628 and @xmath629 and all images under @xmath626 , so @xmath630 , * @xmath631 and @xmath632 and all their @xmath626-images , so @xmath633 , * @xmath634 and their @xmath626-images , so @xmath635 . an explicit decision tree of depth @xmath636 for this @xmath637 is given in our figure below . here
the _ pseudo - leaf _ `` yes(7,10 ) '' denotes a decision tree where we check all elements @xmath356 that have not been checked before , other than the elements @xmath638 and @xmath639 .
if none of them is contained in @xmath357 , then the answer is yes ( irrespective of whether @xmath640 or @xmath641 ) , otherwise the answer is no .
the fact that two elements need not be checked means that this branch of the decision tree denoted by this `` pseudo - leaf '' does not go beyond depth @xmath639 .
similarly , a pseudo - leaf of the type `` yes(7 ) '' represents a subtree of depth @xmath636 .
we now concentrate on the case where @xmath183 is closed under taking subsets , that is , @xmath183 is an abstract simplicial complex , which we also denote by @xmath645 . in this setting ,
the symmetry group acts on @xmath406 as a group of simplicial homeomorphisms .
if @xmath183 is a graph or digraph property , then this means that the action of @xmath253 is transitive on the vertex set @xmath474 of @xmath406 , which corresponds to the edge set of the graph in question .
again we denote the cardinality of the ground set ( the vertex set of @xmath406 ) by @xmath475 .
a complex @xmath646 is _ collapsible _ if it can be reduced to a one - point complex ( equivalently , to a simplex ) by steps of the form @xmath647 @xmath648 are faces of @xmath406 with @xmath649 , where @xmath650 is the _
unique _ maximal element of @xmath406 that contains @xmath651 .
our figure illustrates a sequence of collapses that reduces a @xmath20-dimensional complex to a point . in each case
the face @xmath651 that is contained in a unique maximal face is drawn fattened .
@xmath652{eps / collapse.eps}\ ] ] the first implication is clear : for a cone we do nt have to test the apex @xmath654 in order to see whether a set @xmath357 is a face of @xmath406 , since @xmath409 if and only if @xmath655 . the third implication is easy topology : one can write down explicit deformation retractions .
the middle implication we will derive from the following lemma , which uses the notion of a _ link _ of a vertex @xmath656 in a simplicial complex @xmath406 : this is the complex @xmath657 formed by all faces @xmath658 such that @xmath659 but @xmath660 .
@xmath406 is non - evasive if and only if either @xmath406 is a simplex , or it is not a simplex but it has a vertex @xmath656 such that both the deletion @xmath661 and the link @xmath657 are non - evasive . if no questions need to be asked ( that is , if @xmath662 ) , then @xmath406 is a simplex .
otherwise we have some @xmath656 that corresponds to the first question to be asked by an optimal algorithm .
if one gets a yes answer , then the problem is reduced to the link @xmath657 , since the faces @xmath663 correspond to the faces @xmath664 of @xmath406 for which @xmath478 . in the case of a no - answer
the problem similarly reduces to the deletion @xmath661 .
we use induction on @xmath369 , where @xmath372 is clear .
if the vertex @xmath656 corresponds to a good first vertex to ask , then we start with a sequence of collapses of the complex that correspond to a collapsing sequence for the link of @xmath656 in @xmath406 : this is possible by induction , since the link of @xmath656 is non - evasive and has at most @xmath569 vertices .
( a non - maximal face in the link @xmath657 that is contained in a unique maximal face provides the same type of face in the complete complex @xmath406 . )
thus we can apply collapses to @xmath406 until we get that @xmath665 .
then one further collapsing step ( with @xmath666 and @xmath667 ) takes us to the one - point complex .
if @xmath668 is the vertex set of @xmath406 , then any point @xmath448 has a unique representation of the form @xmath669 with @xmath670 and @xmath671 .
if the group action , with @xmath672 is transitive , then this means that for every @xmath673 there is some @xmath408 with @xmath674 .
furthermore , if @xmath675 is a fixed point , then we have @xmath416 for all @xmath408 , and hence we get @xmath676 for all @xmath673 . from this
we derive @xmath677 for all @xmath23 .
hence we get @xmath678 and this is a point in @xmath406 only if @xmath406 is the complete simplex with vertex set @xmath329 . alternatively : the fixed point set of any group action is a subcomplex of the barycentric subdivision , by lemma [ l : fps ] .
thus a vertex @xmath675 of the fixed point complex is the barycenter of a face @xmath357 of @xmath406 .
since @xmath675 is fixed by the whole group , so is its support , the set @xmath357 .
thus vertex transitivity implies that @xmath679 , and @xmath680 .
[ the evasiveness conjecture for prime powers : kahn , saks & sturtevant @xcite ] all monontone non - trivial graph properties and digraph properties for graphs on a prime power number of vertices @xmath681 are evasive .
we identify the fixed vertex set @xmath329 with @xmath682 .
corresponding to a non - evasive monotone non - trivial graph property we have a non - evasive complex @xmath406 on a set @xmath683 of @xmath684 vertices . by theorem [ t :
collapsible ] @xmath406 is collapsible and hence @xmath443-acyclic .
the symmetry group of @xmath406 includes the symmetric group @xmath685 , but we take only the subgroup of all `` affine maps '' @xmath686 that permute the vertex set @xmath329 , and ( since we are considering graph properties ) extend to an action on the vertex set @xmath683 of @xmath406 .
then we can easily verify the following facts : @xmath253 is doubly transitive on @xmath329 , and hence induces a vertex transitive group of symmetries of the complex @xmath406 on the vertex set @xmath683 ( interpret @xmath682 as a @xmath182-dimensional vector space , then any ( ordered ) pair of distinct points can be mapped to any other such pair by an affine map on the line ) ; taking these facts together , we have verified all the requirements of oliver s theorem [ t : oliver ] . hence @xmath253 has a fixed point on @xmath406 , and by lemma [ l : transitive - fixed ] @xmath406 is a simplex , and hence the corresponding ( di)graph property is trivial
. from this one can also deduce with a lemma due to kleitman & kwiatowski ( * ? ? ? * thm .
2)that every non - trivial monotone graph property on @xmath81 vertices has complexity at least @xmath691 .
( for the proof see ( * ? ? ?
* thm . 6 ) . )
this establishes the aanderaa
rosenberg conjecture [ arc ] . on the other hand , the evasiveness conjecture is still an open problem for every @xmath692 that is not a prime power .
kahn , saks & sturtevant ( * ? ? ?
4 ) report that they verified it for @xmath323 .
an interesting aspect of yao s proof is that it does not use a vertex transitive group .
in fact , let the cyclic group @xmath695 act by cyclically permuting the vertices in @xmath1 , while leaving the vertices in @xmath329 fixed .
the group @xmath253 satisfies the assumptions of oliver s theorem [ t : oliver ] , with @xmath696 .
it acts on the complex @xmath406 which is acyclic by theorem [ t : collapsible ] .
thus we get from oliver s theorem that the fixed point set @xmath417 is acyclic .
this fixed point set is not a subcomplex of @xmath406 ( it does not contain any vertices of @xmath406 ) , but it is a subcomplex of the order complex @xmath697 , which is the barycentric subdivision of @xmath406 ( lemma [ l : fps ] ) .
the bipartite graphs that are fixed under @xmath253 are those for which every vertex in @xmath329 is adjacent to none , or to all , of the vertices in @xmath1 ; thus they are complete bipartite graphs of the type @xmath698 for suitable @xmath303 .
our figure illustrates this for the case where @xmath557 , @xmath499 , and @xmath699 .
@xmath700 monotonicity now implies that the fixed graphs under @xmath253 are _ all _ the complete bipartite graphs of type @xmath698 with @xmath701 for some @xmath702 with @xmath703 .
( here @xmath704 is impossible , since then @xmath406 would be a simplex , corresponding to a trivial bipartite graph property . )
now we observe that @xmath417 is the order complex ( the barycentric subdivision ) of a different complex , namely of the complex whose vertices are the complete bipartite subgraphs @xmath705 , and whose faces are _ all _ sets of at most @xmath702 vertices . thus @xmath417 is the barycentric subdivision of the @xmath706-dimensional skeleton of an @xmath707-dimensional simplex . in particular , this space is not acyclic . even its reduced euler characteristic , which can be computed to be @xmath708 , does not vanish .
however , conjectures ( 5 ) and ( 4 ) fail for @xmath323 : a counterexample is provided by the six - vertex triangulation of the real projective plane ( see ( * ? ? ?
* section 5.8 ) ) .
even conjectures ( 3 ) and possibly ( 2 ) fail for @xmath712 : a counterexample by oliver ( unpublished ) , of dimension @xmath636 , is based on @xmath713 ; see lutz @xcite .
so , it seems that conjecture ( 1)the monotone version of the generalized aanderaa
rosenberg conjecture [ c : gar]may be the right generality to prove , even though its non - monotone version fails by proposition [ e : illies ] . 1 .
what kind of values of @xmath479 are possible for graph properties of graphs on @xmath81 vertices ?
for monotone properties , it is assumed that one has @xmath714 , and this is proved if @xmath81 is a prime power . in general , it is known that @xmath715 unless @xmath481 , by bollobs & eldridge @xcite , see ( * ? ? ?
2 . show that the digraph property `` has a sink '' has complexity @xmath716 can you also prove that for any non - trivial digraph property one has @xmath717 ? + ( this is stated in best , van emde boas & lenstra @xcite ; there are analogous results by bollobs & eldridge @xcite ( * ? ? ?
viii.5 ) in a different model for digraphs . ) 3 .
show that if a complex @xmath406 corresponds to a non - evasive monotone graph property , then it has a complete @xmath182-skeleton .
4 . give examples of simplicial complexes that are contractible , but not collapsible .
( the `` dunce hat '' is a key word for a search in the literature ) 5 .
assume that when testing some unknown set @xmath357 with respect to a set system @xmath183 , you always get the answer yes ( unless you have already proved that the answer is no , in which case you would nt ask ) .
* show that with this type of answers you _ always _ need @xmath369 questions for _ any _ algorithm ( and thus @xmath183 is evasive ) if and only if @xmath183 satisfies the following property : * * for any @xmath718 there is some @xmath719 such that @xmath720 .
* show that for @xmath721 , the family @xmath183 of edge sets of planar graphs satisfies property ( * ) . * give other examples of graph properties that satisfy ( * ) , and are thus evasive .
+ ( this is the `` simple strategy '' of milner & welsh @xcite ; see bollobs @xcite . )
let @xmath34 be a vertex - homogeneous simplicial complex with @xmath81 vertices and euler characteristic @xmath722 .
suppose that @xmath723 is the prime factorization and let @xmath724 .
prove that @xmath725 7 .
let @xmath726 be the set of all words of length @xmath81 in the alphabet @xmath727 , @xmath728 .
for subsets @xmath729 , let @xmath479 be the least number of inspections of single letters ( or rather , positions ) that the best algorithm needs in the worst case @xmath730 in order to decide whether `` @xmath731 '' + define the polynomial @xmath732 where @xmath733 for @xmath734 .
+ show that @xmath735 e. c. milner and d. j. a. welsh . on the computational complexity of graph theoretical properties . in c. s. j. a. nash - williams and j. sheehan , editors ,
fifth british combinatorial conference , aberdeen 1975 _ , utilitas math . , pages 471487 , winnipeg , 1976 . | these lecture notes were written about 15 years ago , with a history that goes back nearly 30 years for some parts . they can be regarded as a `` prequel '' to the book @xcite , and one day they may become a part of a more extensive book project .
they are not particularly polished , but we decided to make them public in the hope that they might be useful .
we refer to @xcite for notation and terminology not explained here . |
four decades after the first pioneering neutrino radiation - hydrodynamic calculations of stellar collapse ( @xcite ) , the details of the core - collapse supernova explosion mechanism remain obscure
. however , certain essentials are clear .
the collapse of the evolved stellar core to a protoneutron star ( pns ) and its evolution to a compact cold neutron star provides a gigantic reservoir of gravitational energy , @xmath23@xmath310@xmath4 erg , a mass - energy equivalent of @xmath20.17 @xmath5 .
any core - collapse supernova mechanism must tap this energy and convert the fraction needed to match type - ii supernova observations ( @xmath210@xmath6 erg @xmath7 1 bethe [ b ] ) into kinetic and internal energy of the exploding stellar envelope .
there is general agreement that the prompt hydrodynamic explosion mechanism does not work and that the bounce shock always stalls , falling short of blowing up the star ( e.g. , @xcite ) , and must be re - energized to lead to a supernova .
however , there is no agreement on the detailed mechanism that revives and endows the shock with sufficient energy to make a canonical @xmath21-bethe supernova . for decades ,
the `` neutrino - driven '' mechanism , first proposed in its direct form by @xcite , and in its delayed form by @xcite and @xcite , seemed compelling .
it relies on a subtle imbalance of neutrino heating and cooling that leads to a net energy deposition behind the stalled shock , sufficient to revive it and drive the explosion on a timescale of hundreds of milliseconds .
while appealing , it has been shown to fail for regular massive stars in spherical symmetry ( 1d ) when the best neutrino physics and transport are used @xcite . yet , weak explosions may be obtained in 1d for the lowest mass progenitors , o - ne - mg cores ( @xcite ) .
it is now almost certain that the canonical explosion mechanism must be multi - dimensional ( 2d/3d ) in nature .
the multi - d dynamics associated with convective overturn in the postshock region ( e.g. , @xcite ) and the recently identified standing accretion shock instability ( sasi , e.g. , @xcite ) lead to a dwell time of accreting outer core material in the postshock region that is larger on average than in the 1d case .
this results in a greater neutrino energy deposition efficiency behind the shock and , thus , creates more favorable conditions for explosion @xcite .
the first generation of multi - dimensional supernova calculations , still employing gray flux - limited diffusion ( or yet simpler schemes ) for neutrino transport , indeed found that neutrino - driven convective overturn in the region between the stalled shock and the pns sufficiently increased the neutrino energy deposition rate to lead to a delayed explosion ( @xcite ) .
the more sophisticated studies that followed changed this picture .
recent long - term axisymmetric ( 2d ) supernova calculations with multi - group , multi - species neutrino physics and transport find it difficult to explode garden - variety massive stars via the neutrino mechanism .
@xcite report explosion only for the low - mass ( 11.2 @xmath5 ) progenitor of @xcite , while @xcite report the onset of explosion in a 15-@xmath5 model of @xcite , given moderately fast rotation and the use of the lattimer - swesty equation of state ( eos ; @xcite ) with a nuclear compressibility modulus @xmath8 of 180 mev , which is significantly softer than the current best experimental values ( @xmath9 mev ; @xcite ) .
on the other hand , @xcite obtain explosions for 11-@xmath5 and 15-@xmath5 progenitors from @xcite only when they take silicon and oxygen burning into account and due to a synergy between nuclear burning , the sasi , and neutrino heating .
@xcite do not obtain neutrino - driven explosions ( except in the case of o - ne - mg cores and accretion - induced collapse ; @xcite ) , but observe the excitation of pns core @xmath10-modes . in their calculations ,
the pns core oscillations reach non - linear amplitudes and damp via the emission of strong sound waves that propagate through the postshock region and efficiently deposit energy into the shock , eventually leading to late explosions at @xmath21 second after bounce . this _
acoustic _ mechanism appears to be robust enough to blow up even the most massive and extended progenitors @xcite , but remains controversial and needs to be confirmed by other groups ( see , e.g. , @xcite ) . in the context of rapid progenitor rotation , @xcite , @xcite , and @xcite ( the latter for the accretion - induced collapse scenario )
have shown that energetic mhd - driven explosions may be obtained if field - amplification by the magneto - rotational instability @xcite is as efficient in the core - collapse context as suggested @xcite .
whether rotation alone and without strong magnetic fields favors or disfavors a neutrino - driven explosion remains to be seen @xcite , but rapid rotation has been shown to damp convection @xcite and weaken the sasi @xcite .
neutrinos , their creation , propagation , and interactions with supernova matter , are of paramount importance to the core - collapse supernova problem .
they carry away @xmath299% of the final neutron star s gravitational binding energy and @xmath21% of this energy would be sufficient to blow up the star . depending on progenitor characteristics that set the postbounce rate of mass accretion onto the pns ,
a successful supernova explosion should occur within @xmath211.5 s after bounce to match observational and theoretical neutron star upper mass limits around @xmath222.5 @xmath5 ( @xcite , and references therein ) .
consequently , the explosion mechanism must deliver canonical 1-b explosions on this timescale and , if the explosion is neutrino - driven , the neutrino heating efficiency and @xmath11 neutrinos and their anti - particles do not contribute much to the heating .
] must be on the order of 10% to yield an explosion that achieves an energy of 1 b within @xmath21 s. the neutrinos travelling through the postshock region in a postbounce supernova core are not in thermal equilibrium with the baryonic matter .
they should ideally be treated with full kinetic theory , describing the neutrino distributions and their temporal distribution with the boltzmann equation ( @xcite ) .
boltzmann transport is in its most general form a 7-dimensional problem .
the 6d neutrino phase space ( usually split up into 3d spatial coordinates , neutrino energy , and 2 angular degrees of freedom ) and time .
in addition , there are up to 6 neutrino types ( 3 particle species , and their anti - particles ) to deal with .
spherically - symmetric boltzmann transport schemes have been devised and implemented in the core - collapse context @xcite , but general boltzmann transport in multiple spatial dimensions is computationally challenging and will remain so in the intermediate term .
hence , approximations must be made in devising computationally tractable neutrino transport schemes for multi - d simulations .
a highly sophisticated approximation that arguably comes close to full boltzmann transport in the case of quasi - spherical configurations in 2d is that presented in @xcite , and based on earlier work by @xcite .
these authors solve equations for the zeroth and first angular moments of spherically - symmetric radiation fields along multiple radial rays ( ray - by - ray approach ; @xcite ) and perform a variable eddington factor closure @xcite via a single spherically symmetric boltzmann solution on an averaged 1d profile of the 2d hydrodynamics data .
neighboring rays are coupled to provide for limited treatment of latitudinal transport .
their multi - group ( multi - energy and multi - neutrino species ) scheme includes inelastic neutrino - electron scattering , aberration , gravitational redshift , and frame effects to @xmath12 .
@xcite implemented a genuinely 2d direct solution of a reduced boltzmann equation via the method of discrete ordinates ( s@xmath13 : see , e.g. , @xcite , and references therein ) in the code vulcan/2d , neglecting energy redistribution and fluid - velocity dependence . a common , more approximate way to handle neutrino transport that has a long pedigree in 1d core - collapse studies is multi - group ( energy / neutrino species ) non - equilibrium flux - limited diffusion ( mgfld ; @xcite , @xcite , @xcite , @xcite , @xcite ) .
fld schemes solve a diffusion equation for the mean radiation intensity , the zeroth angular moment of the specific radiation intensity .
hence , they drop all local angular dependence of the radiation field , while , in the mgfld case , retaining the spectral neutrino distribution .
mgfld accurately describes the radiation field at high optical depth where the diffusion approximation is exact . in the free - streaming limit
, the flux must be limited to maintain causality and an interpolation must be performed between diffusion and free - streaming regimes by an ad - hoc prescription ( using a flux limiter ) .
2d fld schemes were pioneered in the core - collapse context by @xcite and modern mgfld implementations can be found in @xcite and in @xcite .
it is not a priori clear whether mgfld is an accurate enough prescription to yield postbounce supernova dynamics in qualitative and quantitative agreement with a more accurate multi - angle treatment . since net energy deposition by neutrinos
is favored only in the semi - transparent gain layer , the quality of a mgfld scheme may sensitively depend on the flux limiter chosen @xcite .
the fact that 2d gray fld schemes have in the past led to neutrino - driven explosions @xcite , while mgfld schemes appear not to @xcite , emphasizes the importance of a spectral treatment of neutrino transport . in 1d ,
mgfld and boltzmann neutrino transport were compared on static hydrodynamic postbounce backgrounds by @xcite , @xcite , @xcite , and @xcite . also in 1d ,
@xcite compared boltzmann transport and mgfld evolutions in the collapse phase , while @xcite performed the only comparison to date of 1d long - term boltzmann and mgfld supernova evolutions .
the static studies all agree that boltzmann transport yields larger instantaneous neutrino heating rates in the gain region , mostly because of a more slowly decreasing inverse flux factor ( @xmath14 over the ratio of flux to neutrino energy density ) , a quantity that can be related to the rate of energy absorption . on the other hand ,
@xcite find no significant dynamical differences between mgfld and boltzmann transport evolutions in their long - term comparison study with the 13-@xmath5 progenitor model of @xcite . in this paper , we present 2d _ multi - angle _ , multi - group neutrino transport supernova calculations using the newtonian axisymmetric vulcan/2d code @xcite . comparing multi - d boltzmann and mgfld treatments ,
we perform postbounce simulations with vulcan/2d and compare 2d steady - state snapshots , as well as fully - coupled dynamical 2d radiation - hydrodynamics evolutions , for non- and rapidly - rotating 20-@xmath5 models whose precollapse profiles are taken from @xcite .
we analyze our angle - dependent neutrino radiation fields and provide for the first time local 2d map projections of the specific intensity @xmath15 . in
[ section : methods ] , we describe our hydrodynamic and radiation - transport schemes and the microphysics that we use in this postbounce core - collapse supernova study . in [ section : initialmodels ] , we introduce the presupernova models and the postbounce configurations , the setup , and the methodology of our boltzmann - transport mgfld comparisons . in [ section : snapshots ] , we present results of snapshot boltzmann transport calculations and compare them with their mgfld counterparts . in [ section : evolution ] ,
we then discuss time - dependent calculations , the dynamical differences between boltzmann transport and mgfld runs , and the consequences for postbounce supernova model evolution .
we wrap up in [ section : summary ] with a summary and critical discussion of the work presented in this paper .
we employ the arbitrary lagrangian - eulerian ( ale , with second - order total - variation - diminishing [ tvd ] remap ) radiation - hydrodynamics code vulcan/2d .
the hydrodynamics module was first described by @xcite .
the 2d time - explicit hydrodynamics scheme is second - order accurate ( in smooth parts of the flow ) , unsplit , and implements a finite - difference representation of the newtonian euler equations with artificial viscosity on arbitrarily structured grids and in cylindrical coordinates .
the computational grid employed here is set up to resemble a spherical - polar grid at radii greater than 20 km and gradually transitions to a cartesian structure at smaller radii @xcite .
this ( a ) avoids hydrodynamic timestep restrictions due to focussing of angular grid lines and ( b ) liberates the pns core , thus allowing mass motion along the axis of symmetry .
self - gravity is implemented via direct grid - based solution of the newtonian poisson equation , as described in @xcite , and we employ the finite - temperature nuclear equation of state of @xcite .
the calculations are run with 230 logarithmically - spaced radial and 120 angular zones ( including the inner , quasi - cartesian region ) .
the grid encompasses a radial extent of 4000 km and the full 180@xmath16 of the axisymmetric domain .
vulcan/2d contains two multi - group , multi - species neutrino radiation - transport options .
as we discuss below ( [ section : initialmodels ] ) , both modules are used in this study . the module implementing 2d transport in the mgfld approximation , evolving the zeroth moment of the radiation field ,
is discussed in @xcite . the angle - dependent transport module that evolves the specific neutrino radiation intensity , @xmath17 , via the method of discrete ordinates ( s@xmath13 ) ,
was first discussed by @xcite ( see also @xcite ) . for convenience and future reference
, we define the zeroth , first , and second moments of the radiation field , @xmath18 note the vector and tensor natures of @xmath19 and @xmath20 , respectively .
@xmath21 is the radiation field unit vector whose coordinate - dependent components are given in @xcite for various common coordinate systems .
here we employ cylindrical coordinates ( see fig .
[ fig : coord ] ) .
the radiation - pressure tensor @xmath22 obeys the trace condition @xmath23 $ ] @xcite .
the spectral neutrino flux is defined as @xmath24 . as explained in @xcite ,
the time - implicit s@xmath13 solver in vulcan/2d updates the specific intensity in the laboratory frame via the boltzmann transport equation ( @xcite ) without fluid - velocity dependence , @xmath25 where we have dropped the neutrino group index @xmath26 .
@xmath27 , where @xmath28 is the inverse absorption mean - free path and @xmath29 is the inverse scattering mean - free path ( both equivalent to the corresponding cross section multiplied by the number density ) .
we assume scattering to be isotropic and employ the transport cross section @xmath30 instead of the total scattering cross section @xmath31 . this approach has been shown to work well in spherically - symmetric core - collapse supernova calculations @xcite . the right - hand side source term @xmath32 equals @xmath33 , where @xmath34 is the emissivity .
the transport grid is identical to the hydrodynamics grid .
the specific intensity and its moments are defined at cell centers , facilitating spatially - consistent coupling with the scalar hydrodynamics variables , as discussed in @xcite .
radiation stress at cell corners is computed via linear interpolation employing cell - centered values of the radiation flux . as a consequence of the neglect of @xmath35 terms in our transport formulation , neutrino advection ,
doppler shifts and aberration effects are not considered .
this greatly limits the computational complexity of the problem , but its impact on the transport solution depends on the particular choice of reference frame and was examined in @xcite .
it is clear that around core bounce and neutrino breakout , during the non - linear phase of the sasi hundreds of milliseconds after bounce , and in the case of rapid rotation , including @xmath35 terms is advisable .
we leave them out here in order to make long - term multi - angle radiation - hydrodynamics simulations feasible and allow direct comparison with the mgfld variant of vulcan/2d . full @xmath35 boltzmann transport with energy redistribution
will be addressed using the code bethe currently under development by a subset of our group ( @xcite ) .
transport scheme implemented in vulcan/2d .
the radiation direction vector @xmath21 is defined in terms of @xmath36 and @xmath37 .
@xmath36 is the angle with respect to the coordinate - grid @xmath38-axis at all spatial positions ( @xmath38,@xmath39 ) . at each ( @xmath38,@xmath39 ) ,
the local momentum - space unit sphere is covered by @xmath40 zones in @xmath36 and at each @xmath36 location by a number @xmath41 of @xmath37-zones , so that each zone in @xmath42 covers roughly the same solid angle .
[ fig : coord],width=264 ] we discretize the angular radiation distribution evenly in @xmath43 from -1 to 1 and in @xmath37 evenly from 0 to @xmath44 ( treating only one hemisphere because of axial symmetry ) .
we make the number of @xmath37-bins a function of @xmath43 to tile the hemisphere more or less uniformly in solid angle . in our time - dependent runs
we employ 8 @xmath45 bins , resulting in a total of 40 angular zones .
steady - state radiation fields are computed with 8 @xmath45 bins , 12 @xmath45 bins ( 92 total angular zones ) and 16 @xmath45 bins ( 162 total angular zones ) at each spatial grid point .
the standard set of neutrino - matter interactions listed in @xcite is included and all computations are performed with 16 discrete neutrino energy bins , approximately logarithmically spaced from 2.5 mev to 220 mev .
electron neutrinos ( @xmath46 ) and electron - antineutrinos ( @xmath47 ) are treated independently while we lump together the heavy - lepton @xmath48 , @xmath49 , @xmath11 , and @xmath50 neutrinos into one group ( `` @xmath51 '' ) .
the code is very efficiently parallelized via mpi in energy groups and species . as an additional simplification
, we do not include energy redistribution by inelastic neutrino - electron scattering .
such energy redistribution and scattering are of modest ( @xmath52 ) relevance for the trapped electron fraction ( @xmath53 ) and entropy of the inner core at core bounce , but otherwise arguably quite subdominant @xcite .
the time - implicit s@xmath13 scheme in vulcan/2d is iterative and suffers convergence problems in regions where the transport problem is scattering - dominated and the optical depth is high @xmath54 . as a consequence , @xcite limited the timestep at postbounce times to @xmath20.10.3 @xmath48s to ensure accuracy and stability . in the present study
, we take a different approach and introduce a hybrid s@xmath13mgfld transport scheme that treats the quasi - isotropic transport problem in the optically - thick pns interior in the diffusion approximation and transitions to full multi - angle s@xmath13 transport in a region of moderate optical depth ( @xmath55 ) , but that is still significantly interior to the neutrinospheres ( @xmath56 ) where the neutrinos decouple from matter and begin to stream .
we chose a radius of 20 km in our calculations for the transition from mgfld to s@xmath13 .
this is a sensible choice , ( 1 ) because the neutrinosphere radii of all groups ( energies / species ) remain larger than 20 km throughout the postbounce period our simulations cover and , ( 2 ) because 20 km also marks the radius at which the transition from the inner irregular quasi - cartesian grid to the outer regular grid is complete .
this boundary is smooth and the s@xmath13mgfld transition does not suffer from cartesian cornerstone effects .
the transition is implemented by setting up for each energy group and species an approximate specific intensity @xmath15 at the centers of the zones below the s@xmath13mgfld interface using the information available from mgfld .
this approximate @xmath15 is obtained via its angular expansion to first order in @xmath21 ( the eddington approximation ) : @xmath57 here , @xmath58 and @xmath59 , where @xmath60 is the flux , and @xmath61 is the first moment of @xmath15 . in mgfld , @xmath60 is computed via @xmath62 \vec{\nabla } j\,\,,\ ] ] where @xmath63 with bruenn s flux limiter @xcite , @xmath64 = \frac{d}{1+d|\vec{\nabla } j|/j}\,\,.\ ] ] the first angular moment of eq .
( [ eq : approxi ] ) , @xmath65 , is then equal and the fact that we are not using gaussian - quadrature - type angular zoning , the integrals of @xmath66 are only accurate to @xmath25% when 8 @xmath36-zones are used and accurate to @xmath21% when 12 and more @xmath36-zones are employed . to ensure conservation of energy in the s@xmath13mgfld matching ,
we employ purely geometrical and temporally constant correction factors to enforce @xmath67 at the interface . ] to @xmath60 and the s@xmath13mgfld matching is consistent and provides a representation of the specific intensity @xmath66 that is accurate to first order in @xmath21 .
given the essentially isotropic neutrino radiation field deep inside the pns , this approximation yields excellent results .
we note that the scheme makes the implicit assumption that the radial gradient of the mean intensity at the transition radius is always negative or zero .
this condition is generally fulfilled in pnss .
we employ the spherically - symmetric solar - metallicity @xmath68 ( at zams ) model s20.0 from the stellar evolutionary study of @xcite , who evolved it to the onset of core collapse . at that moment ,
its iron core mass , at the outer edge of the iron core where @[email protected] . ] is @xmath21.46 @xmath5 and its central density has reached @xmath28.4@xmath310@xmath69 g @xmath70 .
a graph of the progenitor s precollapse density stratification as a function of enclosed mass can be found in fig . 1 of @xcite .
note that in the study of @xcite , iron core mass and extent vary non - monotonically in the 1020 @xmath5 zams mass range and that their solar - metallicity 20-@xmath5 model has , in fact , a more compact central configuration than the corresponding 15-@xmath5 model .
stellar evolution theory of massive stars has yet to converge and studies by different groups do not presently yield the same presupernova structures .
we set up two initial models in vulcan/2d : s20.nr and s20.@xmath44 .
both models are mapped from 1d onto our 2d hydrodynamic grid under the assumption of spherical symmetry .
model s20.nr is kept nonrotating , while we impose an initial angular velocity profile in model s20.@xmath44 according to the rotation law , @xmath71 where @xmath39 is the distance from the rotation axis and @xmath72 is a parameter governing precollapse differential rotation .
this rotation law enforces constant angular velocity on cylindrical shells and , for sensible choices of @xmath72 , reproduces qualitatively ( @xcite ) predictions from presupernova models that include rotation in a 1d fashion @xcite .
since the computational complexity of this study inhibits us from performing a sweep of the @xmath73@xmath72 parameter space , we chose @xmath74 km and @xmath75 rad s@xmath76 .
hence , the initial central period is 2 s this is an identical rotational setup to the fiducial model in @xcite . as discussed in @xcite , 2
s is rather short , leads to a rapidly rotating postbounce configuration with a millisecond - period pns , and , unless significant postbounce spin - down ( e.g. via mhd torques ) occurs , is inconsistent with average pulsar birth spin estimates .
we chose such rapid rotation simply because we wish to study a postbounce supernova core with significant rotationally - induced asymmetry .
key model parameters and characteristics are summarized in table [ table : models ] .
we collapse both models with the mgfld variant of vulcan/2d and evolve them to @xmath2160 ms after core bounce .
then , we transition to @xmath0 boltzmann transport and solve for the stationary neutrino radiation field based on the artificially frozen hydrodynamics data at this postbounce time .
once we have obtained a converged angle - dependent radiation field , we activate neutrino - matter coupling and hydrodynamics and evolve in time the coupled radiation - hydrodynamics equations . for direct comparison , we also continue the mgfld simulations to later times .
all steady - state snapshots are computed in three momentum - space angular resolutions , s@xmath77 , s@xmath78 , and s@xmath79 , while the long - term evolution calculations could only be performed with s@xmath79 , due to computational constraints . in fig .
[ fig : intro2d ] , we show entropy colormaps of both models at 160 ms after bounce .
fluid velocity vectors are superposed , providing a snapshot of the flow . by 160
ms after bounce , in the nonrotating model s20.nr convection in the high - entropy ( @xmath80(10 ) @xmath81/baryon ) gain layer has developed fully .
the shock sits at @xmath2175 km and is slightly deformed by the onset of the sasi .
not visible on the scale of this figure is the lepton - gradient - driven convective region deep inside the pns , which was extensively discussed in @xcite .
the pns in the rapidly rotating model s20.@xmath44 is rotationally flattened , with unshocked low - entropy inner - core pole / equator asymmetry ratios below @xmath20.5 .
the shock is slightly prolate and has attained an average radius of @xmath2230 km .
the moment - of - inertia - weighted mean period of the unshocked ( specific entropy @xmath82 3 k@xmath83 ) inner core is @xmath22.0 ms .
differential rotation between @xmath220 and 200 km is very large , with the angular velocity @xmath84 dropping from @xmath21600 rad s@xmath76 to a mere @xmath215 rad s@xmath76 over this radial equatorial interval .
yet , the specific angular momentum @xmath85 is still monotonically and rapidly increasing .
it flattens , but does not decrease , only at radii greater than @xmath2100 km .
this positive gradient in @xmath85 stabilizes the postbounce core against convective instability at low latitudes @xcite , confining overturn to the polar regions and large equatorial radii where the @xmath85 gradient is less steep .
llcllll s20.nr&s20.0&0.0&&179.2&160.0&500.0 + s20.@xmath44&s20.0&@xmath44&1000&193.7&160.0&550.0 $ ] in model s20.nr at 160 ms after core bounce , at selected equatorial radii , and for @xmath46 neutrinos at @xmath86 mev .
at each radius , we normalize the specific intensity by its local maximum .
shown is the variation with @xmath37 at fixed @xmath87 .
the graphs are based on a s@xmath77 calculation . at @xmath88 30
km , the radiation field is practically isotropic , but is already appreciably forward - peaked at the neutrinosphere ( @xmath89 km ; optical depth @xmath90 ) and thereafter smoothly transitions over @xmath2200300 km to the free - streaming limit .
[ fig : s20nr_psi1],width=321 ]
in this section , we present our @xmath0 multi - angle transport results for steady - state model snapshots at 160 ms after core bounce .
we diagnose the angle - dependent neutrino radiation fields and carry out a comparison between multi - angle and mgfld transport results based on local and global radiation - field variables .
the quintessential problem in treating neutrino radiation transport in core - collapse supernova cores is the fact that the neutrino transport mean - free path , the average distance a neutrino can travel without experiencing scattering or absorption , changes by orders of magnitude from inside to outside .
moreover , the neutrino transport mean - free - path @xmath91 varies locally strongly with neutrino energy ( @xmath92 ) and matter density . as a consequence ,
gray transport schemes are problematic , since neutrino - energy averages can be defined only locally and the mean neutrino energy varies significantly throughout the supernova core . from a more geometric point of view
, the radiation field in momentum space goes from being completely isotropic ( net flux @xmath2zero ) to being focussed into the radial direction ( `` forward - peaked '' ) in the free - streaming regime . in the mgfld approximation , the mean intensity @xmath93
is evolved in time and the angular information , in particular the information on the degree of forward - peaking , is captured only by computing spatial gradients in @xmath93 and employing a flux limiter to interpolate between diffusion and free streaming .
the @xmath0 boltzmann solver in vulcan/2d is able to self - consistently handle the transition from isotropic to forward - peaked radiation .
figure [ fig : s20nr_psi1 ] depicts the angular distribution in the azimuthal angle @xmath37 ( see fig .
[ fig : coord ] ) of the normalized specific spectral neutrino intensity @xmath15 for electron neutrinos at 12.6 mev . in fig .
[ fig : s20nr_psi1 ] , the polar angle @xmath36 is set equal to @xmath94 and the @xmath37-distribution is given at various radii in the equatorial plane of model s20.nr .
at 30 km from the center , the radiation field of @xmath46 s at @xmath95 = 12.6 mev is nearly isotropic , which corresponds to a circle in fig .
[ fig : s20nr_psi1 ] . with increasing radius ( and , of course , decreasing matter density ) the transport
mean - free path at fixed @xmath95 increases and the radiation field gradually departs from isotropy and becomes more and more forward - peaked .
we define the neutrinosphere as the surface at which the optical depth @xmath96 , given by @xmath97 is equal to 2/3 . at around this @xmath96 , the neutrinos
decouple from matter and begin to stream freely . at 160 ms after bounce in model s20.nr , the 12.6 mev @xmath46 neutrinosphere is located at @xmath9855 km . as is obvious from fig .
[ fig : s20nr_psi1 ] , the radiation field at the neutrinosphere is not yet dramatically forward - peaked , but becomes so with increasing radius .
however , complete forward - peaking only obtains at radii @xmath99 250300 km , beyond which the angular resolution of our @xmath0 scheme becomes suboptimal , even with @xmath100 .
however , calculations with varying number of @xmath36 ( and , hence , @xmath37 ) angles reveal that the transition from isotropy to moderate and large anisotropy is adequately reproduced at small and intermediate radii ( out to @xmath2200 km ) even in the case of s@xmath79 . for the purpose of displaying and studying the local neutrino radiation field , we provide equal - area hammer - type map projections @xcite .
such map projections are new to the field of neutrino radiation transport and beautifully reveal the multi - d angular - dependence of the radiation field . in fig .
[ fig : hammer1 ] , we present such hammer projections on the equator ( spatial @xmath101 ) of model s20.nr at radii of 60 , 120 , and 240 km for the three neutrino species included in our simulations at @xmath102 mev . in each plot , we normalize the specific intensity to the mean intensity to set a common scale . the colormap is logarithmic and chosen to have regions on the sphere with high intensity appear red and regions of low intensity appear black . for neutrinos on the equator ,
the momentum - space forward direction is @xmath103 .
electron neutrinos generally have the shortest transport mean - free path of all species in the core - collapse context and decouple from matter at the lowest densities .
the hammer projection in the top - left corner of fig .
[ fig : hammer1 ] of the @xmath95 = 12.6 mev equatorial radiation field at 60 km corresponds roughly to the blue line graph in fig .
[ fig : s20nr_psi1 ] , which portrays only its variation with @xmath37 . at fixed neutrino energy group @xmath95 , electron anti - neutrinos and `` @xmath104 '' neutrinos decouple at smaller radii .
hence , as fig .
[ fig : hammer1 ] shows , at 60 km , they already manifest greater local anisotropy than the @xmath46s .
this trend continues at all considered radii in fig .
[ fig : hammer1 ] . in fig .
[ fig : hammer2 ] , we again present hammer projections of the normalized specific intensity , but this time consider only @xmath46s , keep the radius fixed at 150 km , and vary the neutrino energy and the angular position on the grid . the bottom row of fig .
[ fig : hammer2 ] shows the normalized @xmath15 at the equator ( @xmath105 ) and for the 12.6-mev and 35.7-mev @xmath46 energy groups .
the center and top rows show the same groups at @xmath106 and at @xmath107 , respectively . from the discussion of fig .
[ fig : hammer1 ] , we are already familiar with the overall radiation field geometry .
the transport mean - free path scales roughly inversely with @xmath108 .
hence , at any given position in the postbounce supernova core , more energetic neutrinos should be locally more isotropically distributed in momentum space than less energetic ones .
the less forward - peaked angular @xmath15 distribution of the higher - energy neutrinos reflects this .
the degree of forward - peaking in @xmath36 and @xmath37 of the radiation field in the quasi - spherically symmetric nonrotating model s20.nr is essentially independent of the angular position on the grid and the radiation fields at any given radius can be transformed into one another by simple rotation . because of the aspherical and oblate distribution of matter in the rotating model s20.@xmath44 , the forward - peaking is also a function of polar angle .
due to its pns s oblateness ( see fig . [
fig : intro2d ] ) , the neutrinos generally decouple at significantly smaller radii near the pole than near the equator , in turn leading to more strongly forward - peaked radiation fields in the polar than in the equatorial regions @xcite .
the radiation - pressure tensor @xmath20 , also known as the eddington tensor , represents the second angular moment of the specific intensity and is defined by eq .
( [ eq : eddy ] ) . in the following ,
we use its normalized variant @xmath109 . in spherical symmetry , @xmath110 is diagonal and has a single independent component , the eddington factor @xmath111 . for isotropic radiation , @xmath112 and @xmath113 , while in the streaming regime , @xmath114 and @xmath115 . in the transition from isotropy to free streaming
, @xmath111 generally varies from @xmath116 to @xmath117 , but in special cases , e.g. , enhanced radiation perpendicular to the radial direction , may assume values below @xmath116 .
note that one of the common assumptions of mgfld is the eddington closure , setting @xmath118 everywhere .
in axisymmetry and ignoring velocity - dependent terms , the eddington tensor has four independent components whose individual meaning depends on the coordinates chosen .
we assume and transform to spherical coordinates for our discussion , since they make the interpretation of the components most straightforward . in fig .
[ fig : eddy ] , we present radial profiles of normalized eddington tensor components at selected electron - neutrino energies @xmath95 in models s20.nr and s20.@xmath44 . the nonrotating model can be considered nearly spherically symmetric , and , hence , should and does exhibit the expected eddington - factor systematics . at small radii and high densities , where neutrinos and matter are in equilibrium , @xmath119 and with increasing radius , @xmath120 and @xmath121 . as expected from the basic decoupling hierarchy , the value of the eddington tensor components is a strong function of @xmath95 .
lower-@xmath95 neutrinos decouple at higher densities , and , hence , have eddington tensor components which depart from @xmath122 at smaller radii than @xmath46s of higher energy .
this systematics applies , of course , to @xmath47s and `` @xmath104 ' 's as well .
the off - diagonal component @xmath123 is zero in the isotropic region , does not exhibit clear systematics , and stays an order - of - magnitude smaller than the diagonal components for all @xmath95 and species .
the rotating model s20.@xmath44 has a postshock configuration that is far from spherically symmetric ( fig .
[ fig : intro2d ] ) .
we present in fig .
[ fig : eddy ] separate plots for its eddington tensor components in regions near the pole and near the equator . in the polar regions and at small radii ( @xmath124100 km ) , the eddington tensor components show the same qualitative behavior as in model s20.nr . at larger radii
, however , the systematics are reversed and lower-@xmath95 electron neutrinos have more isotropic radiation fields ( smaller @xmath125 ) than their higher-@xmath95 counterparts . analyzing their radiation fields and
matter coupling in detail , we find that this surprising feature is a consequence of electron capture and the polar compactness ( large density gradient due to rotation ) of the supernova core . electron capture near the shock leads to isotropic neutrino emission that can locally isotropize the radiation field in semi - transparent regions . with decreasing density and temperature , the mean energy of neutrinos emitted by capture processes shifts to lower @xmath95 .
this leads to greater local isotropization of lower-@xmath95 neutrinos , which in turn is reflected in the more slowly increasing @xmath125 of these neutrinos .
this interpretation is confirmed by the fact that we do not find any such feature in the eddington tensor components of the `` @xmath104 '' neutrinos that are not produced in capture processes .
we also do not observe significant isotropization in the @xmath47 radiation fields , since the emission of @xmath47s by positron capture on neutrons is weaker due to the lower positron abundance . in regions of model s20.@xmath44 near the equator
where the pns is most extended , the neutrino radiation fields stay isotropic to large radii and decouple from matter only slowly with radius . since the matter densities in the equatorial plane stay roughly a factor of four larger than in the polar regions , the cross - over feature in @xmath126 does not appear and these components follow the standard decoupling hierarchy .
interestingly , and different from in the nonrotating model , @xmath127 and @xmath128 show quantitatively distinct variation with @xmath95 , the latter exhibiting significantly less variation with @xmath95 at any given radius .
the interpretation of this observation is not straightforward , but we suggest that it can be attributed to the fact that in model s20.@xmath44 the radiation field at any given point on the equator of the rotationally - flattened core and for any @xmath95 and neutrino species varies locally less in the @xmath36 direction than in the @xmath37 direction .
this , in combination with the fact that on the equator the radiation field asymptotically peaks into the ( @xmath129,@xmath130 ) direction , results on average in smaller @xmath128 with less spread in energy than exhibited by @xmath127 . the off - diagonal component @xmath123 ( not shown in fig .
[ fig : eddy ] ) vanishes in @xmath131 regions , but can become relatively large at greater radii ( up to @xmath20.2 in magnitude ; increasing with @xmath95 and radius ) and flips sign at the equator .
the interpretation of @xmath123 is not straightforward , since its magnitude depends on the choice of coordinates .
we do not attempt to study it , nor its implications for neutrino shear viscosity , in any detail .
mgfld comparison for the nonrotating model s20.nr at 160 ms after bounce .
all s@xmath13 results where obtained with a 16 @xmath36-angle calculation .
see text for details and discussion .
top panel : neutrino luminosity as a function of radius and broken down into the three neutrino species considered .
the `` @xmath104 '' neutrinos dominate in luminosity and their luminosity profiles are scaled by a factor of 1/4 to preserve the overall scale of the plot .
center panel : angle - averaged energy - mean inverse neutrino flux factor profiles .
bottom panel : rms neutrino energy profiles .
[ fig : s20nr_line],width=321 ] so far we have studied aspects of neutrino transport inaccessible to mgfld .
we now go on to discuss radiation field diagnostics that facilitate a @xmath0mgfld comparison . for further reference and comparison with previous studies @xcite
, we define the neutrino luminosity per species @xmath132 at spherical radius @xmath133 , @xmath134 where @xmath135 is the spectral radial neutrino flux in species @xmath136 at energy @xmath95 .
@xmath137 is the spatial solid - angle element , @xmath138 in axisymmetry .
furthermore , we define the mean inverse flux factor @xmath139 , @xmath140 where @xmath141 is the spectral neutrino energy density , and the neutrino rms energies are @xmath142 the above three quantities are particularly useful diagnostics , since the @xmath95-averaged energy deposition rate by charged - current absorption of @xmath46 and @xmath47 on neutrons and protons scales linearly with their product ( @xcite ) . in fig . [
fig : s20nr_line ] , we plot neutrino luminosities @xmath132 , angle - averaged mean inverse flux factors , and the angle - averaged @xmath143 for the postbounce snapshot at 160 ms of the nonrotating model s20.nr .
the asymptotic total luminosity at this time is @xmath2150 b s@xmath76 and is already dominated by the thermally - produced `` @xmath104 ' 's that cool the pns , but contribute little to the heating in the gain region , since they can not take part in charged - current absorption processes . in this quasi - spherically symmetric model , we define a spherical gain radius @xmath144 as the radial position beyond which net neutrino energy deposition occurs . at 160
ms after bounce , @xmath145 90 km and the gain region extends almost out to the shock at @xmath2175 km .
the mgfld luminosities in fig .
[ fig : s20nr_line ] are systematically lower by @xmath25% for @xmath46s , @xmath23.5% for @xmath47s , and @xmath24% for `` @xmath104 ' 's , but qualitatively resemble the @xmath0luminosity profiles in the gain region . at around the shock position
, all mgfld luminosities increase by @xmath25% .
this is a due to the combination of the artificially spread - out shock ( over @xmath245 zones ) , the rapid change of the inverse neutrino mean - free path in the spread - out shock and the implementation of the flux limiter in vulcan/2d . since this mgfld artefact occurs right at the shock , it can have only little influence on the heating in the gain region , but leads to somewhat overestimated asymptotic luminosities in the mgfld case . the center panel of fig .
[ fig : s20nr_line ] shows the @xmath95-averaged inverse flux factors for the three neutrino species in the mgfld and @xmath0 steady - state calculations of model s20.nr
. for isotropic radiation @xmath146 tends to infinity , while it approaches one when the radiation field becomes forward - peaked at low optical depth .
focussing on the gain region between @xmath144 and the shock position , we find that mgfld yields mean inverse flux factors that are up to @xmath25% larger for @xmath47s ( less for the other species ) in the inner gain region . at radii @xmath147150 km ,
the mgfld @xmath139 quickly drops to 1 ( free streaming ) , becoming up to 8% lower than the @xmath0 values in the outer gain region .
we note that `` @xmath104 '' interact only via neutral - current weak interactions , hence , decouple from matter at higher densities and temperatures .
next in the decoupling hierarchy are electron anti - neutrinos followed by electron neutrinos .
both @xmath0and mgfld realize this hierarchy at radii below @xmath2150 km , beyond which mgfld rapidly transitions to free streaming irrespective of neutrino species . , @xmath47 , and `` @xmath104 '' neutrinos at 160 ms after bounce in model s20.nr .
solid lines correspond to s@xmath13 results , while dashed lines are obtained using mgfld .
the spectra have the canonical shape and the quantitative behavior found in nonrotating intermediate - time postbounce supernova calculations ( e.g. , @xcite ) with the `` @xmath104''-neutrinos peaking at the highest energies , since they decouple from the fluid at the smallest radii .
mgfld and s@xmath13 spectra agree closely in shape , but mgfld is overestimating slightly the total asymptotic luminosity ( cf .
fig [ fig : s20nr_line ] ) .
[ fig : s20nr_spec160],width=321 ] calculations with 8 , 12 , and 16 @xmath36-angles , corresponding to a total number of angular zones of 40 , 92 , and 162 .
the gain region extends from @xmath290 km to the shock position at @xmath2175 km .
the three different @xmath0 resolutions yield net gain profiles that agree very well ( relative differences below 1% even for s@xmath79 ) .
the mgfld calculation underestimates the total net gain in the outer gain region by at most 10% locally and @xmath1485% on average .
[ fig : s20nr_gainprof],width=321 ] is shown on the right .
the differences between s@xmath13 and mgfld are marginal at this time in this model and are practically indiscernible by eye . as a consequence of convection in the gain region and the onset of the sasi ,
even this nonrotating model exhibits significant angular and radial variations in the neutrino energy deposition not captured by the average profiles in fig .
[ fig : s20nr_gainprof ] .
[ fig : s20nr_net_gain160],width=332 ] ) as seen by observers near the pole ( solid lines ) and near the equator ( dashed lines ) in model s20.@xmath44 at 160 ms after bounce .
red graphs correspond to @xmath0 results , black graphs depict mgfld results .
top , center , and bottom panels show @xmath149 for @xmath46 , @xmath47 , and `` @xmath104 , '' respectively .
all @xmath0 results were obtained with @xmath150 , but for comparison we also plot in the top panel polar profiles that were obtained with s@xmath151 and s@xmath78 and find that both s@xmath77 and s@xmath78 are very well converged , while s@xmath151 has troubles at radii greater than @xmath2200 km .
however , it agrees very well at smaller radii with the higher - resolution @xmath0 calculations .
[ fig : s20pi_lum160],width=321 ] the behavior we observe with radius of the luminosity and mean - inverse flux factor agrees with the general findings of @xcite .
in particular , we agree with their assessment that the artificially accelerated transition to free streaming in mgfld occurs not at the neutrinospheres ( which are generally below the gain region ) , but at relatively large radii within which most of the neutrino source is enclosed . in the bottom panel of fig .
[ fig : s20nr_line ] we present profiles of the rms neutrino energy for all species in mgfld and @xmath0 snapshots of model s20.nr .
the corresponding luminosity spectra ( extracted at 500 km ) are shown in fig .
[ fig : s20nr_spec160 ] .
both mgfld and @xmath0 capture the energy systematics that is set essentially by the matter temperature in the decoupling region .
neutrino species that decouple at smaller radii ( higher densities and temperatures ) have higher rms energies and harder spectra than neutrinos decoupling at larger radii . quantitative differences in rms energies and in the spectra between mgfld and @xmath0
are small , the slightly higher mgfld spectral luminosities being mostly a result of the artificially enhanced mgfld luminosities near and beyond the shock .
we now conclude our discussion of the 160 ms - postbounce snapshot of model s20.nr by considering the instantaneous neutrino energy deposition rates .
figure [ fig : s20nr_gainprof ] depicts angle - averaged radial profiles of the specific neutrino heating / cooling rates in units of erg ( g s)@xmath76 .
the region of net gain extends from @xmath290 km to the shock radius and the chief contribution to the heating comes from charged - current @xmath47-capture processes on protons , exceeding the corresponding @xmath46-capture on neutrons by a factor of two and more in the narrow radial interval from 145 to 175 km .
mgfld underestimates the specific net gain in the angle - averaged radial profile by at most 10% locally and by @xmath25% on average at radii greater than @xmath2110 km .
the integral total net gain predicted by s@xmath77 is 2.13 b s@xmath76 .
this is only 3% larger than the mgfld value of 2.07 b s@xmath76 .
we note in passing that s@xmath79 overestimates the integrated gain rate by at most @xmath21.6% while s@xmath78 agrees with s@xmath77 to better than @xmath20.3% .
figure [ fig : s20nr_net_gain160 ] depicts the 2d distribution of neutrino heating and cooling in the snapshot of model s20.nr considered here .
regions of net gain range from green to red , cooling regions are blue to black .
the colormap demonstrates the somewhat misleading character of angle - averaged profiles . while we find that there is little spatial angular variation in the neutrino radiation field , the neutrino
matter coupling depends strongly on angular position , and energy deposition is generally greatest in regions of high entropy ( cf .
fig [ fig : intro2d ] ) . as we discussed in the context of the eddington tensor in [ section : eddy ] and as may be guessed from the significant rotational deformation of the core in model s20.@xmath44 ( fig .
[ fig : intro2d ] ) , the radiation field in this model exhibits a strong rotationally - induced asymmetry between pole and equator . in fig .
[ fig : s20pi_fluxes ] , we present 2d colormaps of the radial spectral flux component ( in erg s@xmath76 @xmath152 mev@xmath76 ) and isoenergy - density contours ( @xmath153 in erg @xmath70 mev@xmath76 ) at a representative @xmath95 of 12.6 mev and for all species
. numbers for both @xmath0 and mgfld are compared side by side .
the global radiation - field anisotropy systematics are qualitatively similar to what was found in the previous mgfld rotating core - collapse study of @xcite . at small radii
, the radiation field ( energy density ) follows the density distribution and is oblate , but in the snapshot at 160 ms after bounce shown in fig .
[ fig : s20pi_fluxes ] has a pole
equator ratio of only 1:2 .
this ratio increases as the pns cools and contracts .
the polar compactness of the pns core leads to a decoupling of matter and neutrinos at smaller radii in regions near the pole , resulting there in greater spectral fluxes at higher neutrino energies and in a prolate distribution of neutrino fluxes and isoenergy - density contours . ) in @xmath0 ( red ) and mgfld ( black ) variants of model s20.@xmath44 as seen by observers near the pole .
@xmath46 spectra have solid lines , @xmath47 spectra are shown in dashed lines , and `` @xmath104 ' 's have dashed - dotted spectra .
the spectra are taken from a s@xmath77 calculation at a radius of 300 km at 160 ms after core bounce .
* bottom : * `` luminosity '' spectra seen by equatorial observers .
[ fig : s20pi_spec160],width=321 ] the most striking difference between the @xmath0 and mgfld radiation fields presented in fig .
[ fig : s20pi_fluxes ] is the former s much greater prolateness at large radii for all species ( and all energies , though we show only @xmath154 12.6 mev ) . with the mgfld prescription ,
the prolateness of the flux is muted and does not extend to large radii .
though the radiation fields are smoothed out at radii @xmath147150 km by mgfld , the @xmath0 fluxes and energy densities remain prolate through the entire postshock region and beyond . at radii outside @xmath2200 km ,
the typical striping pattern of @xmath0 ( @xcite ) becomes visible , though not yet dominant . in fig .
[ fig : s20pi_lum160 ] , we plot line profiles of the polar and equatorial `` luminosities '' ( @xmath155 ) of each neutrino species . profiles obtained with @xmath0 and mgfld are shown .
the asymptotic `` luminosities '' obtained with s@xmath13 have pole - to - equator ratios of 2.2 ( @xmath46 ) , 1.8 ( @xmath156 ) , and 2.4 ( `` @xmath104 '' ) .
mgfld smoothes out these large asymmetries , yielding higher equatorial and significantly lower polar `` luminosities '' at radii greater than @xmath2100 km .
this is consistent with the more qualitative findings based on fig .
[ fig : s20pi_fluxes ] .
we note that the mgfld variant of vulcan/2d still conserves total flux and energy . for the @xmath0calculation
, we find total asymptotic luminosities of 21.1 b s@xmath76 for @xmath46 neutrinos ( mgfld : 20.4 b s@xmath76 ) , 22.7 b s@xmath76 for @xmath47 neutrinos ( mgfld : 22.6 b s@xmath76 ) , and 53.0 b s@xmath76 for `` @xmath104 '' neutrinos ( mgfld : 52.3 b s@xmath76 ) . hence , @xmath0 and mgfld total luminosities per species agree very well ( and differ at most by @xmath23.5% in the @xmath46 case ) , while their flux distributions disagree significantly .
figure [ fig : s20pi_spec160 ] , depicting polar and equatorial `` luminosity '' spectra ( @xmath157 ) , reveals that in the @xmath0 calculation ( polar : black graphs , equatorial : red graphs ) the neutrino radiation emerging from the pns and postshock environments through the polar region not only has greater fluence , but also a significantly different and in the @xmath46 case a significantly harder spectrum . @xmath46 neutrinos decouple at the largest radii .
their `` luminosity '' spectrum observed by a polar observer peaks at @xmath1589.5 mev , while for an observer near the equator it peaks at @xmath27.5 mev .
both @xmath47 and `` @xmath104 '' neutrinos ( which decouple further in ) exhibit a smaller variation in peak energy from pole to equator .
the mgfld calculation , on the other hand , shows much smaller variations in neutrino energy and flux between pole and equator ( green and blue graphs , respectively ) .
we note in passing that the emerging neutrino spectra of model s20.@xmath44 are systematically softer by up to @xmath210% in each species than those of the nonrotating model s20.nr presented in fig .
[ fig : s20nr_spec160 ] .
this is a direct consequence of the rotationally - induced lower overall compactness of the pns in model s20.@xmath44 .
the rms neutrino energies in model s20.@xmath44 show the same overall qualitative behavior and decoupling hierarchy discussed in the context of model s20.nr .
hence , we do not show them here , but rather state quantitative results .
they do , of course , trace the strong pole
equator asymmetry that we observe in the radiation field .
the rms energies in the 160 ms @xmath0 snapshot are 11.5 mev ( pole ) and 10.5 mev ( equator ) for @xmath46 , 16.7 mev ( pole ) and 15.2 mev ( equator ) for @xmath47 , and 25.8 mev ( pole ) and 24.8 mev ( equator ) for `` @xmath104 '' neutrinos .
the mgfld values converge at pole and equator to 10.5 mev ( @xmath46 ) , 15.2 mev ( @xmath47 ) , and 25.0 mev ( `` @xmath104 '' ) . at 160
ms after bounce in polar regions ( solid lines ) and equatorial regions ( dashed lines ) . shown
are profiles for @xmath46 neutrinos obtained with @xmath0 ( red ) and mgfld ( black ) , as well as profiles for @xmath47 neutrinos ( @xmath0 blue , mgfld green ) .
@xmath0 and mgfld graphs agree well inside @xmath250 km at the pole and inside @xmath280 km in equatorial regions . for @xmath46 neutrinos
, @xmath0 yields systematically larger mean inverse flux factors in polar and equatorial regions . for @xmath47 , however , @xmath0 predicts larger mean inverse flux factors in polar regions , yet transitions slightly faster than mgfld to free streaming in equatorial regions.[fig : s20pi_ff160],width=321 ] at 160 ms after core bounce ( in units of erg s@xmath76 g@xmath76 ) .
the left section of the plot depicts the mgfld result and the right shows the result of the @xmath0calculation . note the distinctively enlarged polar gain regions and greater specific gain of the s@xmath13 result compared to the mgfld calculation .
this is in part a consequence of the larger polar neutrino fluxes and overall greater flux asymmetry in the s@xmath13 model ( see fig . [
fig : s20pi_fluxes ] ) .
a feature prevalent in both s@xmath13 and mgfld versions of this rapidly rotating core is an extended loss region between the shock and the small gain region at low latitudes ( cf .
[ fig : s20pi_netgain2 ] ) .
the material in the loss region is still proton rich ( @xmath1590.4 ) and efficiently captures electrons as it advects in , radiating away a significant flux of neutrinos ( see , e.g. , the increase in the equatorial `` luminosity '' between 120 and 150 km in the @xmath0 variant of this model , visible in the top panel of fig .
[ fig : s20pi_lum160 ] ) . note that both mgfld and @xmath0 exhibit a very small artefact ( lower gain / loss ) at the symmetry axis associated with imperfect numerics / regularization .
[ fig : s20pi_netgain],width=321 ] at 160 ms after core bounce . shown
are results from the s@xmath13 ( red ) and mgfld ( black ) calculations .
both polar and equatorial radial profiles are obtained by averaging over 20@xmath16 wedges .
as is already clear from fig .
[ fig : s20pi_netgain ] , s@xmath13 yields significantly greater polar specific neutrino energy gain than mgfld .
the s@xmath13 gain region extends further in by @xmath210 km and the gain is more than a factor of two larger in the interval from @xmath290 to 200 km .
given the larger flux asymmetry in the s@xmath13 calculation ( fig .
[ fig : s20pi_fluxes ] ) , less neutrino flux is going through regions of low latitude , resulting in the _ lower _ specific gain at low latitudes predicted by s@xmath13 . * bottom * : neutrino gain density ( density - weighted specific gain ) . due to rapid rotation higher densities
obtain out to larger radii at low latitudes .
this results in a partial reversal of the picture presented by the upper panel ; weighted by density , the neutrino gain ( now per unit volume ) in the equatorial wedge becomes comparable to that near the poles .
furthermore , equatorial regions , since they subtend the largest solid angles , contribute most to the volume integral .
the integral numbers for the net gain in the polar wedge ( counting both poles ) for s@xmath13 ( mgfld ) are 0.17 b s@xmath76 ( 0.047 b s@xmath76 ) and in the equatorial wedge are 0.35 b s@xmath76 ( 0.47 b s@xmath76 ) .
the total integrated net gain is 1.603 b s@xmath76 and 1.637 b s@xmath76 for s@xmath13 and mgfld , respectively .
these numbers are surprisingly close given the large qualitative and quantitative _ local _ differences in the neutrino gain distribution .
[ fig : s20pi_netgain2],title="fig:",width=321 ] at 160 ms after core bounce . shown
are results from the s@xmath13 ( red ) and mgfld ( black ) calculations .
both polar and equatorial radial profiles are obtained by averaging over 20@xmath16 wedges .
as is already clear from fig .
[ fig : s20pi_netgain ] , s@xmath13 yields significantly greater polar specific neutrino energy gain than mgfld .
the s@xmath13 gain region extends further in by @xmath210 km and the gain is more than a factor of two larger in the interval from @xmath290 to 200 km .
given the larger flux asymmetry in the s@xmath13 calculation ( fig .
[ fig : s20pi_fluxes ] ) , less neutrino flux is going through regions of low latitude , resulting in the _ lower _ specific gain at low latitudes predicted by s@xmath13 . * bottom * : neutrino gain density ( density - weighted specific gain ) . due to rapid rotation higher densities
obtain out to larger radii at low latitudes .
this results in a partial reversal of the picture presented by the upper panel ; weighted by density , the neutrino gain ( now per unit volume ) in the equatorial wedge becomes comparable to that near the poles .
furthermore , equatorial regions , since they subtend the largest solid angles , contribute most to the volume integral .
the integral numbers for the net gain in the polar wedge ( counting both poles ) for s@xmath13 ( mgfld ) are 0.17 b s@xmath76 ( 0.047 b s@xmath76 ) and in the equatorial wedge are 0.35 b s@xmath76 ( 0.47 b s@xmath76 ) .
the total integrated net gain is 1.603 b s@xmath76 and 1.637 b s@xmath76 for s@xmath13 and mgfld , respectively .
these numbers are surprisingly close given the large qualitative and quantitative _ local _ differences in the neutrino gain distribution .
[ fig : s20pi_netgain2],title="fig:",width=321 ] in fig .
[ fig : s20pi_ff160 ] , we plot polar and equatorial mean inverse flux factor profiles for @xmath46 and @xmath47 neutrinos in our steady - state snapshot for model s20.@xmath44
. results from mgfld and s@xmath77 runs are shown .
a free - streaming radiation field has an inverse flux factor of one . due to the steeper density gradient in polar regions ,
neutrinos decouple from matter at smaller radii than at the equator .
while mgfld must handle the decoupling and increased forward - peaking of the radiation field via the flux limiter , @xmath0can track it self - consistently . for @xmath46 neutrinos and along the poles ,
@xmath0 predicts significantly greater mean inverse flux factors with shallower slopes than mgfld , indicating a more gradual transition to free streaming than predicted by the flux limiter . in the radial interval of @xmath260100 km ,
the relative difference is @xmath21219% , decreasing to @xmath2612% out to 200 km . in equatorial regions ,
the @xmath46 radiation field is somewhat more forward - peaked in the @xmath0 calculation at radii below @xmath2120 km , beyond which mgfld transitions quickly to free streaming while @xmath0approaches it more gradually , exhibiting @xmath268% larger mean inverse flux factors in the outer postshock region . for @xmath47 neutrinos , the behavior of the mean inverse flux factors in polar regions essentially mirrors that observed for the @xmath46s . in equatorial regions ,
the @xmath0 mean inverse flux factor of the @xmath47s stays below that using mgfld out to 165 km , beyond which the mgfld @xmath47 radiation field rapidly transitions to free streaming . at 180 km ,
the mgfld @xmath156 mean inverse flux factor is @xmath21% smaller than that predicted by @xmath0 . at 220 km ,
this difference has grown to @xmath25% .
having established the overall neutrino radiation - field characteristics in the 160-ms postbounce snapshot of model s20.@xmath44 , we now turn our focus to the neutrino cooling and heating rates in this model .
we have found little difference in the net neutrino heating between @xmath0 and mgfld variants in the 160-ms postbounce snapshot of the nonrotating model s20.nr .
however , based on the differences between @xmath0 and mgfld in neutrino fluxes , rms energies , and flux factors we have highlighted in this section , we may expect to find significant differences in the neutrino heating rates for model s20.@xmath44 .
figure [ fig : s20pi_netgain ] depicts 2d colormaps of the neutrino energy gain and loss rate per unit mass ( accounting for all energies and species ) , computed for the 160-ms postbounce snapshot of model s20.@xmath44 using @xmath0 ( left panel ) and mgfld ( right panel ) . at low latitudes near the equator , @xmath0 and mgfld agree very well to the eye . in regions near the pole , both mgfld and @xmath0
show a pronounced region of net loss at @xmath38-coordinates between @xmath240 and @xmath280 km , beyond which a region of net gain ( colors light blue and green to red ) prevails out to the shock position at @xmath2230 km . while the gain region has roughly the same physical extent in mgfld and @xmath0 , the latter yields significantly higher energy deposition rates .
this is particularly the case in the lower gain region at polar angles below @xmath220@xmath16 and at radii between @xmath280 and 150 km , where the @xmath0 gain rate is larger by a factor of two and more .
the top panel in fig .
[ fig : s20pi_netgain2 ] provides a more quantitative comparison of @xmath0 and mgfld gain / loss rates , since it contrasts average specific gain / loss profiles obtained from polar and equatorial 20@xmath16 wedges . in the polar region ,
the @xmath0 gain region begins at a radius of @xmath280 km ( mgfld : @xmath288 km ) and the @xmath0 specific gain rate magnitude exceeds the mgfld numbers by a factor of 2.6 at 100 km , increasing to 3.2 at 200 km . near the equator
, net energy deposition occurs only in a small radial interval of @xmath290120 km and the mgfld specific gain rate is larger by 80% at 95 km , 41% at 100 km , and 26% at 110 km . the net energy loss between @xmath2120210 km ( captured by both @xmath0 and mgfld ) results from strong electron capture that dominates energy deposition by neutrino absorption .
the observed local differences in neutrino energy deposition between @xmath0 and mgfld are due primarily to the vastly different degree to which the two schemes capture the global pole - equator asymmetry of the radiation field in the rapidly - rotating postbounce supernova core of model s20.@xmath44 . @xmath0 yields much larger fluxes in the polar direction than mgfld , but predicts lower neutrino fluxes in equatorial regions ( cf .
[ fig : s20pi_lum160 ] ) .
differences in the radial mean - inverse flux factors and rms energies are much smaller , and , hence , are of only secondary importance .
the @xmath0 steady - state snapshot yields an integrated gain rate of 1.603 b s@xmath76 while mgfld predicts 1.637 b s@xmath76 for the s20.@xmath44 snapshot under consideration .
this corresponds to @xmath22.1% _ more _ energy deposition per unit time in the mgfld calculation .
given the above discussion , the reader may be surprised by these numbers .
the explanation consists of two factors .
owing to rotation , the amount of mass per unit volume ( i.e. , the rest - mass density ) is higher at any given equatorial radius than at the same radius in the polar direction . plotting the neutrino gain / loss rate per unit volume instead of per unit gram , the bottom panel of fig .
[ fig : s20pi_netgain2 ] clearly shows the rotation - induced enhancement of the energy deposition ( per unit volume ) near the equator and the larger gain rate per unit volume predicted by mgfld at small to intermediate radii .
the second factor is the simple fact that the volume of the equatorial gain regions is much larger than that of the polar gain regions . as we shall discuss in the following section , the large local differences in neutrino heating between the @xmath0 and mgfld snapshots have a dynamical consequence for the rapidly rotating model and lead to a significant polar expansion of the shock in the @xmath0 postbounce evolution calculation
in order to study differences between @xmath0 and mgfld in a time - dependent postbounce setting , we follow our relaxed 160-ms @xmath0 models in fully coupled radiation - hydrodynamics fashion for @xmath2340 ms ( model s20.nr ) and 390 ms ( model s20.@xmath44 ) of postbounce time . in parallel with the @xmath0 runs , we continue their mgfld counterparts for the same time span .
since we begin the mgfld and @xmath0 calculations from an identical hydrodynamic configuration at 160 ms after bounce , any qualitative or quantitative differences in their evolutions must ultimately be due to differences in the neutrino heating and cooling between @xmath0 and mgfld . in the left panel of fig .
[ fig : s20_gain ] , we display the time evolution of the integral neutrino energy deposition ( net gain ) in the gain region of model s20.nr .
the net gain systematically declines at early postbounce times , due ( a ) to the declining neutrino luminosity and ( b ) to the rapid settling of accreting material into the net loss region near the pns core ( cf .
fig . 7 of @xcite ) . at later times ,
sasi - modulated convection increases the dwell time of accreting outer core material in the gain layer and the slope of the net gain evolution flattens .
both @xmath0 and mgfld track these systematics without qualitative difference .
the @xmath0calculation predicts on average @xmath2510% higher net gain in the postbounce interval from @xmath2160 ms to @xmath2220 ms . between @xmath2220
ms and @xmath2280 ms , mgfld and @xmath0 net gain rates agree to within a few percent . towards the end of this interval ,
the net gain of the @xmath0 calculation grows and settles at values that are on average 2030% higher than those of the mgfld run .
this trend is confirmed by the right panel of fig .
[ fig : s20_gain ] , which portrays the heating efficiency , defined as the ratio of net gain to the sum of @xmath46 and @xmath47 luminosities . the left panel of fig .
[ fig : s20_lum_rms ] depicts the temporal evolution of the @xmath46 `` luminosities '' ( @xmath155 ) as seen by observers situated at 250 km along the north pole and south pole as well as in the equatorial plane of models s20.nr and s20.@xmath44 .
here we focus on model s20.nr and note for the @xmath0 variant that north pole ( thin solid black lines ) and south pole ( thin solid green lines ) `` luminosities '' agree ( on average ) in magnitude , but exhibit oscillations about their temporal average that are roughly out of phase by half a cycle .
the mgfld calculation ( thin dashed lines ) , on the other hand , does exhibit some short - period `` luminosity '' variations , yet shows no appreciable difference between poles and equator .
the time at which @xmath0 begins to yield systematically larger neutrino heating rates ( fig .
[ fig : s20_gain ] ) coincides with the growth of the sasi - related shock excursions to large amplitudes ( fig .
[ fig : shockrad ] ) .
this suggests that the increased heating is related at least in part to the @xmath0variant s ability to better capture radiation field asymmetries ( see also the discussion in [ section : s20pi ] ) , induced at late times by the rapidly varying shock and postshock hydrodynamics in this model .
other factors that contribute to the increased heating in the @xmath0 calculation are the higher rms neutrino energies ( by @xmath25% ; shown in the right panel of fig .
[ fig : s20_lum_rms ] ) and the more gradual transition of the @xmath0 neutrino radiation field to free streaming in the postshock region ( see [ section : s20nr ] ) .
figure [ fig : s20nr_evol2d ] contrasts @xmath0 and mgfld simulations of model s20.nr by means of colormaps depicting the specific entropy distributions in the two variants . to visualize the hydrodynamic flow
, we superpose fluid velocity vectors .
each panel of this figure corresponds to a specific postbounce time and each panel s left - hand - side depicts the state of the mgfld calculation , while the right - hand - side depicts the corresponding @xmath0 calculation .
the figure covers a postbounce interval from 160 ms ( top left ) to 500 ms ( bottom right ) . at the beginning of the runs ,
the sasi - driven deviation from sphericity of the stalled shock is mild , but grows with time , showing @xmath160 excursions now generally recognized as characteristic of the sasi=1 dominance still obtains , yet reaches smaller relative amplitudes , since not only higher @xmath161 modes , but also @xmath162 modes , may now contain power .
however , @xcite , who performed a perturbative study without symmetry constraints , argued that in the 3d case with rotation , a dominant @xmath163 ( @xmath164 ) mode is likely to emerge in the case of slow ( rapid ) rotation . ]
@xcite .
as expected from the discussion of the s20.nr 160-ms postbounce steady - state snapshot in [ section : snapshots ] , @xmath0 and mgfld variants of this model do not differ significantly in the early sasi phase .
however , at later sasi stages , in particular at postbounce times @xmath99 300350 ms , the simulations diverge , showing different local qualitative and quantitative behavior within the overall sasi theme .
this is also reflected in fig .
[ fig : shockrad ] , which depicts the evolution of the average shock radius , as well as the shock radii along north pole and south pole .
the shock positions in the @xmath0 and mgfld simulations remain close and the sasi stays practically in phase ( right panel of fig .
[ fig : shockrad ] ) until @xmath2 350 ms after bounce .
only then do they begin to show significant departures from each other .
the sasi in the @xmath0calculation appears more pronounced at later times , exhibiting larger local ( in time ) shock excursions .
yet , quite surprisingly , given the significant increase in neutrino energy deposition , the @xmath0calculation does not exhibit any increase in the average shock radius , nor does it appear to be any closer to explosion than its mgfld counterpart .
the diagnosis of the radiation - hydrodynamic evolution of the rapidly spinning model s20.@xmath44 is less straightforward than for the nonrotating model s20.nr . as discussed in
[ section : s20pi ] , rotation creates a global pole - equator asymmetry in the hydrodynamics of this model .
mgfld and @xmath0 track the effect of globally asymmetric matter distributions on the neutrino radiation field to different degrees . in the steady - state snapshot at 160
ms , @xmath0 predicts stronger neutrino heating in polar regions , yet weaker heating in the higher - density , larger - volume equatorial regions .
the polar , equatorial , and angle - averaged shock positions portrayed by fig . [ fig : shockrad ] show that the hydrodynamics responds immediately to the increased polar heating in the @xmath0 calculation by a pronounced expansion of the shock along the poles .
this expansion lasts for @xmath240 ms , after which the shock has expanded by @xmath220% from @xmath2230 km to @xmath2275 km on both poles .
it stagnates at this radius and subsequently contracts again when feedback of the hydrodynamics to the neutrino microphysics leads to increased cooling ( cf .
the increased polar neutrino emission shown in fig .
[ fig : s20_lum_rms ] ) .
the increased postshock volume also results in a larger gain region and increased ( compared to mgfld ) total neutrino energy deposition and heating efficiency .
however , this increased heating is not able to sustain the large postshock volume .
the shock slowly recontracts in the postbounce interval from @xmath2250 ms to @xmath2380 ms and eventually settles at radii similar to those obtained by the mgfld shock . in fig . 21 , we present a sequence of 2d entropy colormaps with superposed velocity vectors , portraying the postbounce evolution of model s20.@xmath44 from 160 ms on .
the rapid rotation in this model not only partially stabilizes convection , but also weakens and delays the growth of the characteristic @xmath160 sasi and @xmath164 sasi - related spiral structure . ] .
since larger shock radii are associated with an increased growth rate of the sasi ( e.g. , @xcite ) , the @xmath0 variant begins to develop periodic shock excursions along the symmetry axis at much earlier times than the mgfld simulation ( fig .
[ fig : shockrad ] ) .
however , at times later than @xmath2400 ms , the mgfld model picks up the large - amplitude sasi as well and both calculations exhibit large - scale radial shock excursions beyond @xmath2400 km along the pole ( fig .
[ fig : shockrad ] ) .
the average shock radius increases in both calculations in this late postbounce phase .
we observe neither such large shock excursions nor a systematic late - time increase of the average shock radius in the nonrotating model .
the observed behavior is most likely due to the rapid rotation and the resulting rarefaction of the polar regions that reduces , in particular at late times , the ram pressure of accretion and allows for the more pronounced sasi . in the left panel of fig .
[ fig : s20_lum_rms ] , we contrast the @xmath46 `` luminosities '' ( @xmath155 ) seen by observers located at a radius of 250 km above the north pole , the south pole , and in the equatorial plane of model s20.@xmath44 .
the mgfld variant predicts a pole - equator flux asymmetry of @xmath16510% that is roughly constant with time .
the @xmath0 calculation yields a very different picture .
polar and equatorial `` luminosities '' at 250 km ( i.e. , near the shock ) are vastly different ( cf .
[ section : s20pi ] ) . over time , the equatorial `` luminosity '' decreases while the `` luminosity '' along the poles is enhanced . at @xmath2200
ms , polar and equatorial `` luminosities '' differ by a factor of @xmath23 .
by @xmath2500 ms , this factor has grown to 4 .
in addition , the @xmath0 simulation shows sasi - induced variations in north and south - pole `` luminosities '' that grow to @xmath235% at late times and are not tracked in the mgfld variant .
these variations are akin those reported for the nonrotating model s20.nr , yet have longer periods , since the large shock excursions in model s20.@xmath44 occur on longer timescales .
as in the nonrotating model , we also find in model s20.@xmath44 that @xmath0yields systematically higher rms neutrino energies for all species and at all times .
however , as shown in the right panel of fig .
[ fig : s20_lum_rms ] , the angle - averaged rms energies do not exhibit a significant increase in the time interval covered by our simulations .
this , again , is due to rapid rotation which slows down the pns s contraction . not shown in fig . [
fig : s20_lum_rms ] , but present in the @xmath0 variant throughout its postbounce evolution , are @xmath21020% ( roughly constant in time and independent of species ) higher rms energies for neutrinos emitted from polar regions compared to those emitted from the pns equator .
this is consistent with our analysis of the neutrino spectra and rms neutrino energies for the 160 ms postbounce steady - state snapshot presented in
[ section : s20pi ] .
we end our postbounce simulations of model s20.@xmath44 with @xmath0 and mgfld at 550 ms after bounce .
though within roughly the same qualitative picture , the two approaches to neutrino transport yield appreciable differences in the postbounce radiation - hydrodynamics evolutions .
importantly , and in contrast to our findings for the nonrotating model , @xmath0 in model s20.@xmath44 does not lead to systematically higher integral neutrino energy deposition , and at late postbounce times , shows a volume - integrated heating rate that is even @xmath230% lower ( on average ) than in its mgfld counterpart .
using the code vulcan/2d @xcite , we perform long - term full-2d multi - angle , multi - group neutrino radiation - hydrodynamic calculations in the core - collapse supernova context .
based on postbounce hydrodynamic configurations from mgfld simulations , we first compute 2d angle - dependent ( @xmath0 ) steady - state solutions for models without precollapse rotation and with rapid rotation ( @xmath166 rad s@xmath76 ) . from these snapshots , we numerically follow the radiation - hydrodynamics evolution with @xmath0neutrino transport , tracking the nonrotating model to 500 ms and the rotating model to 550 ms after bounce
. done for the first time in 2d , we investigate in detail the angle - dependent specific intensities and neutrino radiation fields .
we compute angular moments of the specific intensity , including the eddington tensor , and introduce hammer - type map projections to visualize the angle dependence of the specific intensity .
these we employ to demonstrate the decoupling systematics of the neutrinos and the gradual transition to free - streaming of the radiation fields with decreasing optical depth .
we compare our @xmath0 simulations with mgfld counterparts .
we find for both models and at all times that the @xmath0 specific intensity distributions transition less rapidly from isotropy to free - streaming in the semi - transparent outer postshock regions .
yields mean inverse flux factors and rms neutrino energies in these regions that are @xmath210% larger than those obtained with mgfld . in the context of the neutrino mechanism of core - collapse supernova explosions , differences in the net neutrino energy deposition rates between mgfld and multi - angle @xmath0 transport
are of greatest interest . in the quasi - spherical early postbounce phase of the nonrotating model , we find that @xmath0 predicts a 510% greater neutrino energy deposition rate than mgfld . at later times ,
when the sasi has reached large amplitudes and globally deforms the postshock region , we find that @xmath0 yields consistently larger ( up to 30% on average ) energy depositions and leads to significantly larger temporary shock excursions around average shock radii that do not depart much from those in the mgfld calculation .
convection on small and intermediate scales and sasi on large scales , are the key agents of the breaking of spherical symmetry in nonrotating ( or slowly rotating ) core - collapse supernovae . while we observe no large qualitative differences in the growth and dynamical evolutions of convection and sasi between nonrotating @xmath0 and mgfld models , we find that the imprint of the asymmetric hydrodynamics on the neutrino radiation fields is captured with greater detail by the multi - angle transport scheme . for the late - time ,
heavily sasi - distorted postbounce core , @xmath0 predicts asymptotic neutrino fluxes that have variations with time and angle of 510% in magnitude .
mgfld is able to capture the temporal variations of the neutrino luminosity , but smoothes out the angular flux variations at large radii / low optical depths .
rapid rotation leads to large deviations from spherical symmetry and a rotationally - deformed pns emits , by von zeipel s law of gravity darkening , a greater neutrino flux along its rotational axis than through its equatorial regions @xcite .
we find that both 2d mgfld and @xmath0 yield similar radiation fields and pole - equator flux ratios at radii smaller than @xmath2100 km . at larger radii ,
the mgfld radiation fields sphericize and show little pole - equator asymmetry in their asymptotic variables .
@xmath0 , on the other hand , captures large pole - equator flux ratios of up to 4:1 at late times and predicts polar neutrino spectra that are harder in peak energy ( rms energy ) than on the equator by up to 30% ( 1015% ) for @xmath46 neutrinos , and somewhat less for the other species .
all this results in a neutrino energy deposition rate per unit mass in polar regions that is locally up to @xmath22.53 times higher when multi - angle transport is used .
this increased polar neutrino heating has a dynamical effect on the postbounce evolution , leading to rapid shock expansion in the polar regions and an earlier onset of the ( initially ) rotationally - weakened sasi .
however , at late times , the sasi in the mgfld calculation catches up and yields shock excursions of a similar magnitude . in summary ,
our results show that 2d multi - angle neutrino transport manifests interesting differences with 2d mgfld when addressing local and global radiation field asymmetries associated with rapid rotation and the non - linear sasi at late postbounce times .
in addition , multi - angle transport results in enhanced neutrino energy deposition .
the latter is most significant in the polar regions of rapidly rotating postbounce configurations and affects dynamically the postbounce evolution , including the growth of the sasi .
however , in the large postbounce interval covered by our simulations , the local and global differences between multi - angle transport and mgfld calculations do not appear large enough to alter the overall simulation outcome .
importantly , the multi - angle models do not appear to be closer to explosion than their mgfld counterparts .
although we neglect velocity - dependent transport terms and coupling of neutrino energy bins , we do not expect our conclusions to be altered by their inclusion , since they are not likely to affect significantly the differences between multi - angle transport and mgfld .
further significant limitations of our present study are the neglect of general relativistic and mhd effects , the restriction to only one finite - temperature nuclear eos , the limited resolution in momentum - space imposed by the computational cost of multi - angle calculations , and the use of two spatial dimensions , plus rotation . in the future
, we will investigate the dependence of our results ( e.g. , heating rates , radiation - field asymmetries etc . ) on the choice of flux limiter and will consider different progenitor models .
the core - collapse supernova problem is one of many feedbacks .
larger heating rates and heating efficiencies than found in our models appear to be necessary to break the feedback cycle between neutrino radiation fields and hydrodynamics , revive the stalled shock , and unbind the supernova envelope if the neutrino mechanism is to obtain in the way presently envisioned .
future work will have to go beyond the limitation of axisymmetry and must address in detail the entire ensemble of possible factors relevant in the supernova problem , including , but not limited to , 3d dynamics , multi - angle neutrino transport with velocity dependence and inelastic @xmath46-@xmath167 scattering , progenitor structure , rotational configuration , magnetohydrodynamics , convection , the sasi , pns g - modes , general relativity , the nuclear eos , and neutrino - matter interactions .
we acknowledge helpful discussions with and input from jeremiah murphy , ivan hubeny , casey meakin , jim lattimer , alan calder , stan woosley , ed seidel , harry dimmelmeier , h .- thomas janka , kei kotake , thierry foglizzo , ewald mller , bernhard mller , martin obergaulinger , benjamin d. oppenheimer , thomas marquart , and erik schnetter .
this work was partially supported by the scientific discovery through advanced computing ( scidac ) program of the us department of energy under grant numbers de - fc02 - 01er41184 and de - fc02 - 06er41452 .
c.d.o . acknowledges support through a joint institute for nuclear astrophysics postdoctoral fellowship , sub - award no .
61 - 5292ua of nfs award no .
86 - 6004791 .
acknowledges support by the israel science foundation ( grant 805/04 ) .
the computations were performed at the local arizona beowulf cluster , on the columbia sgi altix machine at the ames center of the nasa high end computing program , at the national center for supercomputing applications ( ncsa ) under teragrid computer time grant tg - mca02n014 , at the center for computation and technology at louisiana state university , and at the national energy research scientific computing center ( nersc ) , which is supported by the office of science of the us department of energy under contract de - ac03 - 76sf00098 . | we perform axisymmetric ( 2d ) multi - angle , multi - group neutrino radiation - hydrodynamic calculations of the postbounce phase of core - collapse supernovae using a genuinely 2d discrete - ordinate ( @xmath0 ) method .
we follow the long - term postbounce evolution of the cores of one nonrotating and one rapidly - rotating 20-m@xmath1 stellar model for @xmath2400 milliseconds from 160 ms to @xmath2550 ms after bounce .
we present a multi - d analysis of the multi - angle neutrino radiation fields and compare in detail with counterpart simulations carried out in the 2d multi - group flux - limited diffusion ( mgfld ) approximation to neutrino transport .
we find that 2d multi - angle transport is superior in capturing the global and local radiation - field variations associated with rotation - induced and sasi - induced aspherical hydrodynamic configurations . in the rotating model ,
multi - angle transport predicts much larger asymptotic neutrino flux asymmetries with pole to equator ratios of up to @xmath22.5 , while mgfld tends to sphericize the radiation fields already in the optically semi - transparent postshock regions . along the poles ,
the multi - angle calculation predicts a dramatic enhancement of the neutrino heating by up to a factor of 3 , which alters the postbounce evolution and results in greater polar shock radii and an earlier onset of the initially rotationally weakened sasi . in the nonrotating model ,
differences between multi - angle and mgfld calculations remain small at early times when the postshock region does not depart significantly from spherical symmetry . at later times
, however , the growing sasi leads to large - scale asymmetries and the multi - angle calculation predicts up to 30% higher average integral neutrino energy deposition rates than mgfld . |
for those about to embark on the study of this paper , let us start with a warning : this article is not a report on ground - breaking results . neither will we present spanking new circuits , ripe for patent . rather in spirit of sm@xmath0acd s explicit policy to be `` a forum for the discussion of new ideas and methodologies ''
, we would like to advocate the introduction of computer algebra into the design process of analog circuits . in order to do this , we present some examples , where computer algebra ( together with numerics ) was used in the design , especially the sizing of the given circuits .
these examples do not represent a systematic approach , but they should be considered as tentative steps towards getting to know the capabilities ( and some of the shortcomings ) of computer algebra systems ( cas ) , and making these tools bring forth fruit in the daily work of electrical engineers .
one of the principal problems in analog circuit design is to determine element values in a given circuit in such a way that the resulting network function is in accordance with some previously given performance specifications . in more mathematical terms
: suppose the network function is demanded to be a rational function of form @xmath1 with the numbers @xmath2 @xmath3 @xmath4 determined by the initial specifications .
then we need to find an analog circuit @xmath5 built from elements , which themselves are described by parameters @xmath6 , such that its network function @xmath7 where the terms @xmath8 and @xmath9 are polynomials in @xmath10 satisfies @xmath11 in other words : at the end of the day , many design problems result in a system of nonlinear equations
@xmath12 @xmath13 and @xmath14 for variables @xmath15 if indeed the terms @xmath16 are polynomials with coefficients in some number field , the set of solutions , otherwise also known as the zero set of ( [ designproblem ] ) , forms what is called an _
affine algebraic variety _ , an object which belongs to the mathematical subdiscipline of _ algebraic geometry .
_ thus when trying to solve these equations , we could profit both from the centuries of experience , mathematicians have gained in this particular field , and from the more recent results and algorithmic instruments developed the concept of _ grbner bases , _ as well as the _ buchberger algorithm _ and its `` relatives '' which are realized in nearly all of the current computer algebra systems .
the result would be exact ( `` symbolic '' ) descriptions of the sought after solution sets , from which we could deduce numeric results afterwards .
now , when we take a look at the advances in symbolic methods applied to circuit design , we can make a peculiar observation : nowadays it has become nearly routine to calculate any given network function as a multivariable rational function of the complex frequency @xmath17 and the symbolic circuit elements ( even though there is still the problem of the combinatorial explosion for large circuits ) .
however , when it comes to finding solutions for the nonlinear equations arising from the task of sizing a particular circuit , even if it is relatively small , researchers , as well as designers , fall back upon numerical means and various approximative methods with their various drawbacks .
in section 6.3.3 . of @xcite the following filter circuit was presented : the aim there was to construct a fourth order butterworth filter with this topology .
using one of the available symbolic analysis programs , we see that the circuits symbolic transfer function is given by @xmath18 with @xmath19 being a polynomial , and -2em @xmath20 in order to specify element values in such a way that @xmath21 becomes the transfer function of a fourth order butterworth filter , we can proceed from these data by matching the coefficients with those numerical values , which can be found in any table for filter design , and then try to solve the resulting equations by numerical means . since we only have four equations for nine unknowns ,
there is the need to reduce the existing degrees of freedom by more or less arbitrarily attaching numerical values to some of the element values .
in fact , that is what was done in @xcite , where @xmath22 was set equal to @xmath23 and all the @xmath24 were set to @xmath25 ( all element values in this example are considered to be normalized ) ; afterwards some variant of the newton - raphson method was used , to deduce one single solution .
the disadvantages of such an approach are obvious : using rounded values ( from precalculated tables ) introduces errors right from the start .
both , the assignment of ( quite arbitrary ) numbers to arbitrarily chosen parameters , and the choice of a starting vector for the newton - raphson method might hinder from finding a solution , or even worse , might make a solution impossible furthermore , questions like `` do alternatives exist ?
how many are there ? '' , can not be answered either , and changing any of the numerical parameters forces us to redo most of the calculations . when we tried to solve this design task , first of all we started with a slightly different approach , using the feldtkeller equation @xmath26 for the special case of the fourth order butterworth filter .
thus we got rid of the precalculated approximate values for the @xmath27 by using instead the defining algebraic equations given by matching the coefficients from @xmath28 with @xmath29 consequently we enlarged the set of equations ( [ 4filterglg ] ) by adding @xmath30 to it , and regarding @xmath31 as further unknowns .
even though in this paper we propose the usage of the algebraic capabilities of a cas , at this point we used mathematica ( version 6.0.1.0 ) as a numerical engine not in order to approximate one solution , but to find approximates for * all * solutions .
moreover as was done in @xcite , we also fix @xmath32 and @xmath33 for @xmath34 since obviously the enlargement of the set of equations results in a bigger set of solutions , we were left with the final task of `` post - processing ''
this zero set , by singling out those particular solutions for which @xmath35 determine a polynomial @xmath36 , all of whose zeroes have negative real part (= filter stability ) , and such that all the @xmath37 are _ non - negative _ (= realizability by classic elements ) .
indeed , we observe that from all the possible solutions calculated , only nine satisfy the restrictions concerning the @xmath27 and again only one of these , @xmath38 results in the @xmath37 all being positive .
so , where does computer algebra enter this picture ? first of all mathematica s nsolve - procedure , which we used in the above task , at its core heavily depends on algebraic algorithms : the fact that it is able to calculate _ all _ solutions stems from the algebraic variety , which was defined by ( [ 4filterglg ] ) and ( [ bw ] ) , being _
zero dimensional_. that is , it consists of finitely many points .
again we can be sure of this fact , only because there is an algorithm the _ buchberger algorithm ) . ] _ which is used by mathematica as a subprocedure of nsolve and which for any system of algebraic equations produces particularly well - behaved equivalent systems of equations
so - called _ grbner bases _ which allow much more insight into the properties of the zero set , and may be used for a subsequent elimination of variables .
thus e.g. , when we apply mathematica s groebnerbasis - command to ( [ 4filterglg ] ) and ( [ bw ] ) , with the values of @xmath39 fixed as above , and the variables ordered lexicographically by @xmath40 ( which amounts to saying that first @xmath41 is to be eliminated , then @xmath42 etc . ) , then the resulting grbner basis contains the polynomial @xmath43 @xmath44 @xmath45 @xmath46 @xmath47 @xmath48 @xmath49 @xmath50 in @xmath51 of degree @xmath52 which implies that there are only finitely many possible values for @xmath53 furthermore this grbner basis contains polynomials @xmath54 @xmath55 and @xmath56 each of which being linear in the respective @xmath57 and otherwise consisting only of a high degree polynomial in @xmath53 since these polynomials as part of a grbner basis have to be zero , too , consequently each zero of @xmath58 determines precisely one solution vector @xmath59 .
those solutions , which result in the additional specifications for the @xmath60 to be fulfilled , originate from a factor of the polynomial @xmath58 of degree @xmath61 which can be calculated using mathematica s factor - procedure , when we allow the coefficients to be sums of integers and integer multiples of @xmath62 this factor polynomial is -1em @xmath63 substituting @xmath64 we see that its structure is effectively determined by a polynomial of degree @xmath65 which results in the number of admissable solutions being nine .
to finish this section , let us remark that by modifying the feldtkeller equation to @xmath66 where @xmath67 denotes the fourth order chebyshev polynomial , and slightly changing the initial values for the @xmath68 after some trial - and - error we were able to calculate the following set of parameter values , with which we can make the initially given circuit into a chebyshev filter @xmath69 the question remains for each filter characteristic : which set of prescribed values for @xmath70 imply solutions for @xmath71 which make sense in an electrical engineering point of view , i.e. , for which positive @xmath70 do we get positive @xmath72 ?
from @xcite - a students textbook in analog electronics - we take our next example : a rudimentary bjt amplifier circuit with only a few additional classic elements for biasing and sizing purposes ( see figure ( [ naundorfpicture ] ) ) . -.4
cm -1em -0.4 cm [ naundorfpicture ] -2em the biasing resistances are given by @xmath73 and @xmath74 the biased transistor is modelled by concrete h - parameters @xmath75 and its capacitances @xmath76 , and @xmath77 . with denoting its transfer function by @xmath78 a full symbolic analysis with sapwin 3.0 .
gives @xmath79 and @xmath80 with @xmath81 with regard to the concretely given transistor parameters at the bias point , the nominator and denominator of @xmath21 become @xmath82 and @xmath83 respectively .
thus , e.g. setting @xmath84 @xmath85 and @xmath86 we get @xmath87 the poles being ( approximately ) @xmath88 this results in the bode plot shown in figure [ naundorf1 ] . -0.4 cm
-0.4 cm we will now use this example as a problem in circuit design .
our task shall be : with the bias point remaining unchanged , find parameters @xmath89 and @xmath90 in such a way that a new transfer function @xmath91 is achieved , which has the same zeroes as @xmath92 and possesses two ( new ) poles @xmath93 and @xmath94 . introducing a new parameter @xmath95 ( since due to the simplicity of our design topology
, we can not specify the maximal attenuation of @xmath96 as we will see below ) , we thus are asked to determine the parameters @xmath97 and @xmath90 in such a way that @xmath98 this is equivalent to @xmath99 factoring out common terms on both sides of the @xmath100-sign , we are left with a third order polynomial in @xmath17 with coefficients being integer polynomials in the variables @xmath101 to solve ( [ naundorfdesign ] ) we need to set these coefficients to zero , which consequently results in the following system of four algebraic equations since we consider @xmath93 and @xmath94 to be fixed . ] ( the @xmath102-th equation resulting from the coefficient of @xmath103 ) : -1em @xmath104 alhough this system is nonlinear , using its procedure solve , mathematica is able to calculate four symbolic solutions for @xmath97 and @xmath90 as functions of @xmath93 and @xmath105 voicing the explicit caveat : `` equations may not give solutions for all `` solve '' variables . '' .
nevertheless , indeed there are only four different solutions .
the first is irrelevant for the concrete problem at hand from an electrical engineering point of view , since it is given by @xmath106 and thus results in a trivial transfer function . the second one
, though it leads to nontrivial terms for @xmath107 and @xmath108 furthermore implies @xmath109 which again has to be considered absurd from an engineering point of view .
finally , there are two solutions left which lead to two different valid sets of design parameters , both of them having in common that @xmath110 here we will not present the explicit formulas for @xmath111 but we will give a numerical example , and we will further show , how the formulas for @xmath112 ( and beyond this our claim , that there are only four solutions ) can be derived with the help of grbner bases . first , for the numerical example let @xmath113 and @xmath114 then the two remaining approximate solutions of the above set of equations are @xmath115 and @xmath116 indeed , the corresponding bode plots for the respective circuits can be seen in figure 4 . ) ( dashed ) , and ( [ drei ] ) ( thick),title="fig : " ] -0.4
cm [ bode3 ] finally , applying mathematica s groebnerbasis - command to the right - hand sides of equations ( [ naundorfglg ] ) , with the order of variables chosen to be @xmath117 results in a grbner basis consisting of @xmath118 polynomials with exactly one of them containing only the variables @xmath119 this particular polynomial is of degree four in the variable @xmath120 keeping in mind that the zero set of the original polynomials constituting ( [ naundorfglg ] ) is equal to the zero set of its grbner basis , we thus see that there truly are only four possibilities for the values of @xmath120 factoring this particular polynomial ( again with the help of mathematica ) we get the following factors : @xmath121 further analysis shows that @xmath122 results in @xmath123 and the solution of the linear ( with respect to @xmath90 ) equation implies the negative value of @xmath112 according to ( [ mies ] ) . thus the two solutions , whose existence we claimed above , have their origin in the quadratic equation ( [ qq ] ) .
symbolic methods always suffer from one major deficiency the `` combinatorial explosion '' of terms which most of the time results in both , a massive , nearly insatiable demand for space , and an extraordinary high amount of computing time .
this observation also applies to all grbner basis methods : it is known that by necessity the problem of finding a grbner basis in the general case needs an exponential amount of space . on the other hand , in recent years
there have been major improvements with regard to the run - time behaviour of the original buchberger algorithm , which have resulted in faster and thus more powerful algorithms to compute grbner bases .
as we could see in the above examples , even moderately sized small circuits already give rise to mid - sized equations , which again by way of the methods from computer algebra result in even bigger symbolic expressions for the sought - after design parameters .
thus the `` new toolbox '' computer algebra , the usage of which i try to recommend with this article , seems to be doomed from the start .
still , bad behaviour _ in general _ does not mean that the instruments from computer algebra _
always _ behave there might be classes of problems for which results are computed fast , or can be written down very economically . as electrical engineers know , there are close ( but in the author s opinion by far not completely understood ) connections between a circuit s topology and the algebraic structure of its network function , which may be taken advantage of with regard to design tasks .
take on the other hand researchers in symbolic methods , who are accustomed with the phenomenon of swelling expressions arising in the analysis of circuits . as a workaround a number of them have proposed so - called `` sequences of expressions '' .
it is interesting to note that even electrical engineers without experience in methods of symbolic analysis may already be well versed in this `` sequential approach '' , if only they knew classic filter theory , in particular the cauer canonical forms of @xmath124-circuits . in form of an example , let us show , what we mean .
suppose we are given an @xmath124-ladder network , consisting of three inductances and three capacitances , as shown in figure 5 .
-.4 cm -1 cm as is well known its impedance @xmath125 can be described by a continued fraction @xmath126 that , in fact , is just short hand for a sequence of expressions .
this can be seen by deducing ( [ frak ] ) with a process of alternatively combining impedances in series and in parallel .
thus we have @xmath127 where @xmath128 denotes the impedance of the ladder circuit , with @xmath129 having been removed .
further : @xmath130 where @xmath131 is the impedance of the circuit , that is left , when @xmath129 and @xmath132 have been removed .
continuing like this , taking away one more circuit element in each step , this process can be repeated , until we get to one last impedance @xmath133 ( which would just consist of @xmath134 ) . reinserting the various @xmath135 we get @xmath136 and thus are able to calculate @xmath137
when conversely one wants to realize an lc transimpedance function as an lc ladder , classic theory tells us to expand the given rational function into a continued fraction ( an algorithm which in case of rational functions always comes to an end after a finite number of steps ) and read off the element values directly from this expression , which according to theorems in classic filter theory all are positive .
if we want to bring together the capabilities of a cas for solving algebraic equations and the sequence of expression approach in our example , we first observe that we can easily calculate the form of the intermediate impedances @xmath138 i.e. we have @xmath139 with @xmath140 and the @xmath60 denoting polynomials in the circuit parameters @xmath141 a sensible design task would now be to specify an impedance function @xmath142 with real numbers @xmath143 and to demand @xmath144 analogously to the above example of the bjt amplifier , by matching coefficients , this results in six nonlinear equations , which have to be solved for the variables @xmath145 and @xmath146 here we only present the ( easy ) solution ( determined by mathematica again ) for the first step , which is -1.5em @xmath147 to calculate further element values , this procedure needs to be repeated for the other @xmath148 thus one has to solve next the equations resulting from @xmath149 where @xmath128 should still be described in terms of the coefficients @xmath27 and @xmath131 is given by the ansatz @xmath150 with new unknowns @xmath151 this has to be continued until we have equations for each set of coefficients of @xmath135 in terms of the coefficients appearing in @xmath152 plus additional equations for the element values . starting from those equations , successive
reinserting leads to the desired expressions for the element parameters .
one has to concede that compared to the traditional continued fraction expansion
this procedure looks very cumbersome .
but in fact , if one looks a little bit closer , it is equivalent to repeatedly doing polynomial division as part of a continued fraction expansion
. furthermore our approach might be generalized to other impedance functions which need not come from ladders , but still ultimately result from a sequence of series , respectively parallel reductions applied to the initially given circuit , i.e. the process of replacing two impedances in series , or in parallel by a single equivalent impedance in series to a single resistance @xmath153 from the resulting equation @xmath154 when only given the value of @xmath153 we can not deduce uniquely the values of the @xmath155 . beyond this
there is a further generalization to handle delta - wye and wye - delta transformations . on the other hand by results from epifanov and truemper
, it is known that every planar two terminal graph can be reduced to a single edge by these four kinds of transformations ( plus removing loops and pendant edges ) .
moreover there is an algorithm , developed by feo and provan , which calculates the sequence of necessary graph transformations . since to each of these transformations there corresponds an equation , which gives the value of the resulting element(s ) in terms of the values of previously existing elements
, there always is a _ sequence of equations _ which leads from the initially given element values to the driving point impedance of a planar two terminal circuit .
circuit design for ( midsized to large ) planar two terminal circuits , which consist of impedances only , thus becomes the task of `` reversing '' these equations , in an analogous manner to what we did with lc ladders .
the author of this paper firmly believes , that computer algebra systems will be a valuable tool in tackling this particular problem .
the challenge remains to find classes of circuits for which the _ sequence of reversed equations _ can be effectively calculated , or to prove what might be also possible , and would be an interesting result in itself that , besides the classic , continued- or partial - fraction - topologies attached to the names of cauer and foster , no such further classes exist .
10 georges gielen and willy sansen , , kluwer academic , boston , 1991 .
tran chi hieu , , ph.d .
thesis , technische universitt braunschweig , 2002 .
university of florence .
department of electronics and telecommunications , `` sapwin 3.0 - symbolic analysis program for window , '' http://cirlab.det.unifi.it/sapwin/ , viewed september 1 , 2008 .
david cox , john little , and donal oshea , , springer , new york , @xmath01997 .
david cox , john little , and donal oshea , , springer , new york , @xmath02004 .
uwe naundorf , , hthig , heidelberg , @xmath1562001 .
joachim von zur athen and jrgen gerhard , , cambridge university press , cambridge , @xmath1561999 .
jean - charles faugre , `` a new efficient algorithm for computing grbner bases @xmath157 , '' , vol .
139 , pp . 6188 , 1999 .
jean - charles faugre , `` a new efficient algorithm for computing grbner bases without reduction to zero @xmath158 , '' mora , teo ( ed . ) , issac 2002 .
proceedings of the 2002 international symposium on symbolic and algebraic computation , lille , france , july 0710 , 2002 .
pp . 75 - 83 ,
acm press , new york , ny . , 2002 .
benedykt s. rodanski and marwan hassoun , `` symbolic analysis , '' in _ the circuits and filters handbook _ , wai - kai chen , ed . , chapter 24 , pp .
crc press , boca raton , florida , 2nd edition , 2002 .
louis weinberg , , mcgraw - hill , new york , 1962 .
frank garvan , , chapman & hall/ crc , boca raton , fl , 2001 .
thomas a. feo and j. scott provan , `` delta - wye transformations and the efficient reduction of two - terminal planar graphs , '' , vol .
41 , pp . 572582 , 1993 .
sven domann , `` implementierung des feo - provan - algorithmus zur stern - dreieck - reduktion planarer graphen , '' studienarbeit , technische universitt braunschweig , 2003 .
in this electronic document keywords , acm and msc classifications have been added .
( march 14th ( pi day ) , 2009 ) [ ] eberhard h .- a .
gerbracht received a dipl .- math .
degree in mathematics , a dipl .- inform .
degree in computer science , and a ph.d .
( dr . rer.nat . )
degree in mathematics from the technical university braunschweig , germany , in 1990 , 1993 , and 1998 , respectively . from 1992 to 1997 , he worked as research fellow and teaching assistant at the institute for geometry at the tu braunschweig . from 1997 to 2003 he was an assistant professor in the department of electrical engineering and information technology at the tu braunschweig .
after a two - year stint as a mathematics and computer science teacher at a grammar school in braunschweig and a vocational school in gifhorn , germany , he is currently working as tutor , and independent researcher in various areas of mathematics .
his research interests include combinatorial algebra , c*-algebras , the history of mathematics in the 19th and early 20th century and applications of computer algebra and dynamical geometry to graph theory , calculus , and electrical engineering .
dr . gerbracht is a member of the german mathematical society ( dmv ) , the society of computer science teachers in lower saxony within the gesellschaft fr informatik ( gi - nill ) , and founding member of the society web portal : history in braunschweig - http://www.gibs.info[www.gibs.info ] . | * in this paper , by way of three examples a fourth order low pass active rc filter , a rudimentary bjt amplifier , and an lc ladder we show , how the algebraic capabilities of modern computer algebra systems can , or in the last example , might be brought to use in the task of designing analog circuits .
* * acm classification : * i.1 symbolic and algebraic manipulation ; j.2 physical sciences and engineering ; g.2.2 graph theory * mathematics subject classification ( 2000 ) : * primary 94c05 ; secondary 94c15 , 68w30 , 13p10 , 05c85 * keywords : * analog circuits , filter design , polynomial equations , computer algebra , delta - wye transformation . |
the phenomenon of high - temperature superconductivity in strongly correlated materials such as the cuprates , still represents one of the most challenging topic in condensed matter physics twenty years after the discovery .
a reason for the difficulty in dealing with the cuprates is certainly the richness and complexity of their phenomenology , which gives rise to many competitive explanations , making it extremely hard to identify the relevant physical processes underlying the outstanding properties of these compounds . in this light
, very important information would be inferred extracting from the whole body of experimental observations , the data measuring general and fundamental properties ( e.g. , thermodynamic quantities and sum rules ) , focusing as much as possible on ubiquitous ( material independent ) aspects . one important example , in the context of the infrared spectroscopy experiments , is represented by the analysis of the optical sumrule ( sr)@xcite .
a big effort has been devoted in the last few years to the evaluation of the frequency integrals of the optical conductivity @xmath0 in several cuprates ( e.g. , @xmath1 ( bscco)@xcite , @xmath2 ( lsco)@xcite , @xmath3 ( ybco)@xcite , and more recently also @xmath4-based cuprates@xcite ) , and to the study of their behavior as a function of temperature and doping .
one of the most general results emerging from these studies is the strong temperature dependence of the optical spectral weight ( or partial optical sum rule ) defined as @xmath5 where @xmath6 is an upper cut - off , whose role will be discussed in the following .
more in detail , leaving aside the behavior of the superconducting phase which establishes below the critical temperature @xmath7 , where even the sign of the temperature variation of @xmath8 depends on doping@xcite , the main results of the infrared estimate of the partial sr in the normal phase ( @xmath9 ) of the cuprates are essentially two : ( i ) a strong enhancement of the @xmath10 is observed when @xmath11 is reduced .
such enhancement is generally quadratic in temperature and is not strongly doping dependent ; ( ii ) the extrapolated spectral weight at @xmath12 , @xmath13 displays , instead , a clear doping dependence , increasing monotonically with @xmath14 .
the qualitative results do not depend significantly on the value of the cut - off @xmath6 and they hold in particular when it reaches the plasma frequency ( around 1 ev ) . in this work we present a detailed analysis of the behavior of the optical sumrule based on dynamical mean - field theory , a non perturbative many - body approach which allows for an accurate treatment of strong correlation effects , which turn out to be crucial to account for the experimental observations .
we base our analysis on the previous work of ref . , which we extend by including realistic two - dimensional bandstructures with van hove singularities .
the paper is organized as follows : in sec .
ii we discuss previous results on non interacting models , and on the qualitative effect of correlations .
iii contains a brief introduction to dmft and our solution .
iv is devoted to the discussion of our dmft results , while a simple interpretation of the same results is presented in sec . v. sec .
vi is dedicated to concluding remarks .
most of the interest in the optical sr comes from its relation with the kinetic energy@xcite .
the most straightforward identification is actually limited to tight - binding ( tb ) models with nearest - neighbor hopping , where , both in the presence and in the absence of electron - electron interactions we have @xmath15 where @xmath16 is the lattice constant , @xmath17 the cell volume , and @xmath18 the kinetic energy of the carriers .
the relation holds if the cutoff @xmath6 is chosen large enough to contain the whole optical spectrum . for a non interacting system
this would correspond to integrate up to the `` plasma frequency '' of the model .
we will see briefly that this identification can be significantly modified by the inclusion of next - nearest ( and further ) neighbor hopping .
( [ eq : tb ] ) , connecting the sr to the kinetic energy , leads to a potential trivial explanation of point ( i ) .
the @xmath19 behavior can indeed be recovered through a simple sommerfeld expansion of ( [ eq : tb ] ) , which gives indeed @xmath20 , with @xmath21 and @xmath22 ( for a flat dos of bandwidth @xmath23 , e.g. , @xmath24 , see ref . ) . on the other hand , as pointed out in refs .
, this one - parameter model can not account simultaneously for the experimentally measured values of @xmath25 and of @xmath26 .
in particular , realistic hopping parameters are not compatible with the size of the observed temperature variation of @xmath8 .
the experimental value of @xmath26 could be recovered only invoking a value of @xmath27 mev smaller _ by more than one order of magnitude _ than the values determined either theoretically by means of bandstructure calculations@xcite , or experimentally through photoemission measurements@xcite .
the simple nearest - neighbor non interacting model appears even poorer when considering the behavior of the zero temperature sumrule ( ii ) .
first , the above mentioned value @xmath28 mev is totally incompatible with the experimental values of @xmath13 ( which ranges from @xmath29 mev ) .
second , the simple model predicts @xmath13 decreasing with the doping level , just opposite to the observations . a first natural step to heal the inadequacy of
the above nearest - neighbor model is to include a more realistic bandstructure . in particular as we already mentioned , eq .
( [ eq : tb ] ) is valid only for pure nearest - neighbor hopping . when releasing this restriction , the generalization of eq .
( [ eq : tb ] ) reads @xmath30 where the sum is performed over all the @xmath31 momenta @xmath32 of the first brillouin zone , @xmath33 is the dispersion and @xmath34 the occupation number for a given @xmath32 state . it has been pointed out in ref . that in the non - interacting case the inclusion of the next - nearest neighbor hopping term @xmath35 can determine remarkable changes in the above picture .
more specifically , for realistic values of the hopping parameters , remarkable differences between the kinetic energy and the optical spectral weight behavior ( defined in eq .
( [ eq : tb2 ] ) ) appear , mainly for dopings close to the two - dimensional van hove singularity ( vhs ) : while the temperature dependence of @xmath18 is only weakly affected by the doping level , @xmath36 is more sensitive to the vhs , whose proximity determines a stronger temperature dependence .
although a stronger @xmath11-dependence of the spectral weight goes certainly in the direction of the experimental evidence , many inconsistencies remain : first of all the enhancement of the @xmath11-dependence of @xmath36 is still not enough to account for the experiments ; secondly for some ( and realistic ) values of the hopping parameters the results can even show a change of sign in the temperature variation of @xmath36 , which has never been observed experimentally .
finally -as already noted in ref .
- the doping dependence of @xmath37 remains opposite to the experimental data .
the failure of these simple noninteracting models to capture the experimental behavior can not be surprising in light of the unquestioned role of electronic correlations in the cuprates . in ref .
we have shown that the inclusion of correlations determines indeed a huge step ahead in the understanding of experimental data .
a strong suggestion to proceed in this direction comes also from point ( ii ) ( see introduction ) , because the monotonically increasing values of @xmath13 would find a very natural explanation in a strongly correlated scenario , where the electronic mobility is minimal at half - filling and it increases with doping . in ref .
it is demonstrated how the presence of strong interactions can actually determine a _ separation of the energy scales _ controlling @xmath13 and its temperature dependence .
more specifically , if one performs the frequency integral of @xmath0 up to a cut - off which includes both the drude and the mid infrared ( mir ) contribution , mimicking the experimental situation , a strong @xmath11-dependence of @xmath8 , and a qualitatively correct behavior of @xmath13 are obtained . on the other hand , some discrepancy with the experimental observations
is present also in the data of ref . :
the dmft data , which are computed for a hubbard model with a semicircular dispersion which has no vhs , predict a stronger temperature dependence of @xmath10 at small doping than for the overdoped compounds , as a result of a larger distance from the mott transition , in contrast with experiments , where this effect is not seen , and even an opposite behavior occurs in bscco@xcite .
one of the reasons for such a discrepancy is that dmft neglects spatial correlations .
the effect of this neglect is expected to be stronger for smaller dopings@xcite .
a consequence of the lack of spatial correlation is the vanishing quasiparticle renormalization factor @xmath38 ( renormalization of the coherent electronic bandwidth ) when @xmath39 in contrast with photoemission data , in which @xmath38 is finite for any doping .
thus , a crucial step for a proper analysis of the small doping region is to consider cluster extensions of dmft able to capture at least short - range correlations ( such as cellular dmft@xcite and dca@xcite ) , where the mott transition can take place with a finite @xmath38 .
present cluster - dmft studies have been mainly dedicated to the superconducting phase , which is not addressed in this paper@xcite . in this work
we do not address the cluster extensions of dmft , while we try to supplement our previous single - site dmft analysis by including the effects of a more realistic two - dimensional bandstructure displaying a vhs , which , as we commented above , can introduce remarkable effects for non interacting systems .
keeping in mind the limitations of single - site dmft , we will not consider extremely small dopings , where non local correlations will become crucial .
we notice in passing that the @xmath19 behavior of the sr is not limited to noninteracting systems , but it is rather characteristic of all fermi - liquid systems , either weakly or strongly correlated .
the actual coefficient is generally a function of @xmath38 and of the imaginary part of the self - energy ( proportional to @xmath19 in a fermi - liquid ) .
what correlation effects and interaction with bosonic modes can change are on one hand the actual values of the @xmath12 value and of the coefficient @xmath26 , as we have already found , and on the other hand they can introduce coherence scales which set the limit under which the system behaves like a fermi liquid . beyond that scale
the temperature behavior can appear different from quadratic , but such a deviation is not necessarily the consequence of deep changes in the groundstate properties .
in this paper we consider the two - dimensional hubbard model , i.e. , @xmath40 where @xmath41 are annihilation(creation ) operators for fermions of spin @xmath42 on site i , @xmath43 , and the sum in the first term includes nearest - neighbors ( nn ) and next - to - nearest neighbor ( nnn ) hopping processes , whose amplitudes are given by @xmath44 and @xmath45 respectively .
our choice of the parameter appearing in the hamiltonian ( [ hubbard ] ) , namely @xmath46 aims to the closest contact with the cuprate properties . specifically , we concentrate on two of the most studied cuprates , i.e. , bscco and lsco , for which a certain agreement about the estimates of the hopping parameters has been reached both on the theoretical ( density - functional theory calculations ) and the experimental side ( photoemission ) . in particular ,
for both bscco and lsco the nn hopping parameter @xmath47 is estimated around @xmath48 , while the different crystal structure of their unit cells reflects in different nnn hopping term : lda calculations predicts a @xmath49 for lsco , whereas larger values of @xmath45 are estimated for bscco @xmath50 .
let us just note that , in the non - interacting case , these values correspond a vhs located at lower energy with respect to the half - filling chemical potential : in the case of @xmath51 ( lsco ) , the vhs would be crossed at a doping level slightly below @xmath52 , while for @xmath53 ( bscco ) the vhs would not be reached even for the higher doping levels relevant for the cuprates , being located roughly at @xmath54 .
the choice of the value of the repulsive term @xmath55 is certainly less obvious : although the relevance of electron - electron correlations is widely , if not universally , accepted as a key element in the properties of the cuprates , it is difficult to estimate precisely its value . in this paper
, we have chosen @xmath56 , consistently with ref .
, which allows for a direct comparison with the results for the simpler semicircular dos used in that paper , and it allows for a clear separation between the mir and the hubbard contributions in the optical sum rule ( see next section ) .
more generally , the choice of @xmath56 is also guided by experimental evidence from the neutron scattering data in the cuprates , which estimate the antiferromagnetic exchange @xmath57 mev .
a recent dmft analysis suggests however that a slightly smaller value @xmath58 is more appropriate to describe the zero temperature optical spectra and their doping dependence.@xcite .
as we anticipated , we use dmft to solve the hamiltonian ( [ hubbard ] ) .
dmft is a nonperturbative method that maps a lattice model onto an effective local model , in which the effect of the neighboring sites on a given site is expressed through a `` dynamical weiss field '' .
the local effective model can be parameterized by an anderson impurity model ( aim ) , in which an interacting site is hybridized with a non interacting bath describing the weiss field . for more details
we refer to ref . .
the mean - field is enforced by a self - consistency condition which relates the green s function of the effective model to the local component of the lattice green s function of the original model .
namely @xmath59 where @xmath60 and @xmath61 are the green s function and the self - energy of the effective model and @xmath62 is the bare dispersion on the chosen lattice ( from now on , we are setting @xmath63 ) . the self - consistency condition requires to solve iteratively the aim until eq . ( [ selfcons ] )
is obeyed . in this work
we consider the two - dimensional dispersion @xmath64 , while in ref .
we considered a simple semicircular density of states characteristic of an infinite coordination bethe lattice .
we emphasize that the lattice structure enters the dmft only in the self - consistency condition .
we use exact diagonalization ( ed ) to solve the aim . in this method
the model can be solved at zero temperature by discretizing the bath function into a small number of levels , which here will be @xmath65 .
the main limitation of the approach is that the spectral properties are those of a finite system , hence the fine details can not be resolved with great accuracy .
this choice is particularly useful for the subject of this work , considering that we are interested in integrated optical spectra , rather than in their details , and that the temperatures of interest are very small with respect to the energy scales of the hamiltonian ( [ hubbard ] ) ( the range of temperature @xmath66 corresponds to @xmath67 ) .
in particular , the very low temperature range allows us to exploit the lanczos algorithm at finite @xmath11 , which has been developed for dmft in ref . .
in this scheme we avoid to compute the whole spectrum of the hamiltonian , limiting to the relevant low - lying states , allowing for a faster calculation than the full diagonalization of the matrix .
ad discussed in ref . and in several precedent works , the interpretation of the optical spectra in dmft can be greatly helped by an inspection of the single - particle spectral function .
in particular the @xmath68integrated spectral function , namely the interacting density of states , can be easily computed through the knowledge of the retarded self - energy @xmath69 on the real axis @xmath70 in fig .
[ fig : dos ] we plot @xmath71 for @xmath72 , the smallest temperature we considered for two doping levels and for parameters corresponding to the two different materials .
quite generally , the evolution of the density of states is analogous to the bethe lattice case and contains three features : ( i ) a strongly renormalized quasiparticle peak ( qp ) at the fermi level ( of width @xmath73 , where @xmath74 is the quasiparticle residue ) , which is basically attached to ( ii ) the lower hubbard band ( of width roughly equal to the bare bandwidth @xmath23 ) , and well separated from the upper hubbard band , whose center is located at @xmath75 , again with width close to @xmath23 .
the weight of the qp clearly increases with the doping level , while the spectral gap between the qp and the hubbard bands appears rather stable in the range of doping considered ( being of order of @xmath76 ) .
the effect of @xmath45 is hardly visible in @xmath71 and seems to affect mainly the shape of the qp , as can be seen by comparing the results for the two values of @xmath45 corresponding to bscco and lsco . interacting density of states @xmath71 computed via dmft at low temperature ( @xmath77 ) for the case of @xmath78 ( @xmath79 ) and @xmath80 ( @xmath81 ) for two different values of doping .
note the presence of a narrow qp at the fermi level , very close to the lower hubbard band , while the upper hubbard band is clearly visible at @xmath82 .
, width=302,height=264 ] information about the spectral function are particularly useful to understand optical spectra due to the simplifications introduced by dmft in this regard .
the locality of @xmath61 and of the two - particle irreducible vertices in dmft , together with the odd symmetry of the current operator for @xmath83 determine in fact the vanishing of all the vertex corrections to the current - current paramagnetic kernel@xcite . as a consequence @xmath0
can be computed via the simple `` bubble '' @xmath84 where @xmath85 , @xmath86 is the retarded green function , and finally @xmath87 the fermi function . for sake of simplicity , the normalization of eq .
[ eq : sigma ] has been chosen so that @xmath8 is approaching directly the value of @xmath88 for @xmath89 .
examples of the dmft results for @xmath0 are shown in fig .
[ fig : sig ] for the same set of parameters of the dos of fig .
[ fig : dos ] .
one can immediately recognize how the main features of @xmath71 discussed above reflect in @xmath0 .
similarly to the results of ref . , the optical conductivity displays ( i ) a clear drude peak at low frequencies ( @xmath90 ) , determined by optical transitions occurring at energies within the qp width , then ( ii ) a mid infrared ( mir ) bump at @xmath91 , related to transitions between the qp and the lower hubbard band , and finally a high energy contribution at @xmath92 , which is related to transitions involving the upper hubbard band . the rather neat separation between the low energy features ( drude and mir ) and the higher energy hubbard contribution suggests a value of @xmath93 as a natural cut - off to compare ( see next section ) our calculations with the experimental data .
this choice has the same spirit of choosing the plasma frequency as a cut - off in the experiments in order to select only the contribution from the lowest absorption band . yet , for our choice of parameters , the cut - off we use is @xmath94 times larger than the plasma frequency of bscco and lsco , but using the experimental values in our calculations would result in including only a part of the mir in the integral , making the analysis less significant , at least in our view .
optical conductivity @xmath0 computed via dmft at low temperature ( @xmath77 ) for the same parameters of fig .
[ fig : dos ] .
one can immediately notice the narrow drude peak at low - frequency , a broader mir contribution at @xmath95 and a small hubbard term at higher frequencies.,width=302,height=264 ]
in this section we analyze the results for the optical spectral weight @xmath8 of the hubbard model through direct numerical integration of the optical conductivity using a cut - off @xmath23 in eq .
( [ eq : defw ] ) . from a theoretical point of view , it is also interesting to compare @xmath8 with the total @xmath96 ( including in this case also the hubbard contributions ) , given by eq .
( [ eq : tb2 ] ) , where @xmath97 .
even if our focus is the dependence on temperature of these two quantities , a brief analysis of the @xmath12 behavior is necessary before moving to finite temperature . in fig .
[ fig : w0tpri ] we report the values of the extrapolated @xmath12 values for @xmath8 and sr for three different doping levels for the set of hopping parameters which refer to bscco and lsco .
both quantities increase as a function of doping , analogously to the bethe lattice case . in this respect , as one could have expected , considering a more realistic bandstructure dispersion reflects at most in minor corrections of the dmft results of ref , and in minor differences between the two compounds .
this clearly shows that the strong correlation effects play the dominant role in determining the values of both @xmath25 and @xmath98 , regardless the shape of the non interacting bands .
we notice that , apart from the natural difference ( which ranges from @xmath99 mev ) due to the contribution of the hubbard bands , the doping behavior of @xmath25 roughly tracks that of @xmath98 , since for the doping considered the magnitude of the frequency integral in the sr is mostly determined by the mir and the drude contributions .
optical spectral weight including drude and mir contributions , and total sr computed in dmft at @xmath12 as a function of doping for the case of @xmath78 and @xmath80.,width=302,height=264 ] we turn now to the temperature dependence of the spectral weight .
the results of the dmft calculations are reported in fig .
[ fig : delwtpri ] , where the ratio between the value of @xmath100 ( and the sr ) and its corresponding @xmath12 extrapolation for the case of bscco and lsco at @xmath101 are shown as a function of temperature .
we see immediately that for both bscco and lsco the relative variation @xmath102 between t=0 and room temperature ( in our units @xmath103 corresponds to @xmath104 ) is larger for the underdoped compounds and constantly decreasing with doping . for the case of bscco
such trend is more pronounced than in the bethe lattice case@xcite and it is only partly due to the doping dependence of @xmath105 : a more quantitative analysis through a quadratic fit ( @xmath106 , solid lines in fig . [
fig : delwtpri ] ) of the data at low @xmath11 clearly demonstrates that the main contribution to this results stems from a remarkable reduction of the coefficient @xmath26 , which controls the low-@xmath11 behavior .
for instance , in the case of bscco @xmath107 at @xmath108 , and it decreases down to @xmath109 at @xmath110 . in the case of lsco ,
the temperature variation at @xmath111 is very close to that at @xmath110 because @xmath26 has a minimum for doping slightly below the vhs .
a second important observation is in order about the temperature dependence of the sum rule . with the only exception of overdoped lsco , in the parameter region considered here ,
the temperature variation of @xmath112 is always bigger than that of @xmath113 ( for bscco by more than a factor @xmath114 ) : this results is therefore not determined simply by the small difference between @xmath25 and @xmath98 : the coefficient of the quadratic fit is always lower in the case of the @xmath96 indicating that the temperature dependence of the high - energy hubbard contribution has always the opposite sign with respect to that of @xmath8 .
this should be taken into account when comparing theoretical prediction on the spectral weight computed in presence of strong interactions with the experimental data .
this problem can be faced , e.g .
, in cluster dmft , where the difficulty to evaluate the vertex corrections may suggest to resort to the calculation of the total sr . relative variation with respect to @xmath12 of @xmath8 ( left column ) and of the total @xmath96 ( right column ) as a function of @xmath11 for bscco ( first row ) and lsco ( second row ) .
the solid lines are the results of the quadratic fit at low temperatures ( see text).,width=302,height=264 ] the main outcome of these dmft results is that in presence of a strong correlation the effects of the details of more realistic bandstructure are not very evident in the behavior of the integrated optical spectral weight . as one clearly sees in fig .
[ fig : w0tpri ] and [ fig : delwtpri ] , the behavior of @xmath25 and of its dependence on temperature when the doping level is enhanced from @xmath108 to @xmath110 is dominated by the quasiparticle renormalization factor @xmath38 which obviously increases with the distance from the mott transition . in this respect ,
our dmft results show that most of the peculiar trends determined by the two - dimensional density of states in a simple non - interacting scheme@xcite are washed out by strong interaction effects , so that the main outcomes of ref . remain unaltered .
nonetheless we observe some effects of the more realistic bandstructure also in the presence of strong correlation , and the outcome is not completely obvious . in particular , one could have expected that , when the chemical potential moves close to a vhs , the largest temperature variation of @xmath8 would have be achieved , because the proximity to the vhs and the strong interaction could have cooperated summed up somehow their effects in making more pronounced the @xmath11-dependence of @xmath8 .
this is apparently not the case , since the vhs is reached at a doping level slightly below @xmath52 for @xmath80 and larger than @xmath110 for bscco , and no trace of any enhancement in the @xmath11-dependence of the spectral weight is found there .
quite remarkably , instead , one can observe that in bscco the weakening of the @xmath11-dependence of @xmath8 with increasing doping is more evident than in the bethe lattice , where no vhs is present , case@xcite , and that in lsco the change of the temperature variation for doping levels below the vhs ( i.e. , @xmath111 and @xmath110 ) is smaller than expected .
this clearly suggest that the proximity to a vhs and the strong interaction effects are partially competitive .
it is worth underlying that no evidence of the change of sign of the temperature variation when the chemical potential crosses the vhs is found in the lsco data for @xmath111 and @xmath115 as opposed to the non interacting case studied in ref . .
an extremely simple scheme to understand the origin of these partly unexpected dmft results will be discussed in the next section .
in this section we will present a simple explanation of the surprising interplay between vhs and strong correlations displayed in dmft . as we discuss below , the main idea behind
our explanation is to introduce a simple cartoon for the correlated system , where the sr is given by the value obtained for a non interacting case with bands renormalized by the factor @xmath38 obtained in dmft .
the starting point of our arguments is the temperature dependence of the total @xmath96 at @xmath116 , which was already analyzed in ref . , and it is shown in fig .
[ fig : somm ] for the bandstructure parameter considered here and different doping levels .
it is worth noticing that , as it was pointed out in ref .
, the low temperature behavior ( which can be obtained also directly via the sommerfeld expansion ) displays remarkable changes for density close to the vhs .
in particular , defining @xmath117 as the doping corresponding to the chemical potential at the vhs ( i.e. , @xmath118 ) , when @xmath14 approaches @xmath117 from below one observes a strong - enhancement of the slope of the @xmath119 at low @xmath11 , while as soon as @xmath120 ( e.g. , @xmath121 for bscco , and @xmath52 for lsco ) the _ sign _ of the slope changes abruptlyto positive , in contrast to natural expectation ( and experimental evidence ) , before changing again sign for higher doping levels .
one should also note that , as a consequence of the big and rapid changes occurring close to the vhs , the temperature behavior of @xmath96 can deviate rather remarkably from the standard @xmath19 behavior _ la _ sommerfeld . relative variation with respect to @xmath12 of the sr for a non interacting system with parameters of bscco and lsco.,width=302,height=207 ] as we mentioned in the previous section
, however , most of these vhs effects , clearly visible in the non - interacting case , are quite weakened ( and , in some cases , they even disappear ) in our dmft results , as a consequence of the strong correlation .
to understand the reason for this effect we observe that in the non interacting systems the anomalous behavior of the sr due to the vhs is always limited to a small temperature range , whose size becomes smaller and smaller as the vhs is approached .
if we consider that in presence of the strong interaction the low - temperature physics should be mainly controlled by the strongly renormalized qp at the fermi level ( see again the dos reported in fig .
[ fig : dos ] ) , one can reasonably expect that correlations determine a reduction ( by a factor @xmath38 ) of the size of the low - temperature region which is controlled by the vhs .
in particular , if the qp is strongly renormalized , as in the case we have considered here ( the @xmath38 evaluated in dmft ranges between @xmath122 for bscco at @xmath108 and @xmath123 for lsco at @xmath110 ) , the effects of the vhs should be limited to temperatures so low to become hardly visible in the experiments and in the temperature range we considered in our dmft results .
we can try to test this generic idea by mimicking the strongly interacting fermi liquid physics by simply rescaling by a factor @xmath38 the energy scale of the non interacting system .
more precisely , the renormalization of the qp peak determines a renormalization of the coefficient @xmath26 of the @xmath19 term ( @xmath124 because @xmath26 has the dimensions of the inverse of an energy ) , this means that the scaled results can be simply obtained by replotting the data of fig .
[ fig : somm ] rescaling the @xmath11 dependence of the sr for a given doping level , with the square root of @xmath38 extracted from dmft ( i.e. , @xmath125 ) .
relative variation with respect to @xmath12 of the sr , after having rescaled the @xmath11-dependence of the non - interacting model.,width=302,height=207 ] the results of such rescaling are shown in fig .
[ fig : somm_resc ] . even at first glance
, one can notice that the peculiar features associated to the vhs are much less evident now , and that the rescaled data resemble in many respects the outcome of the full dmft calculation . more specifically
, the slope of sr(t ) does not change sign but in a very low temperature regime , with the exception of the bscco at @xmath121 , which is however above the doping regime of our interest .
the same consideration applies to the increasing of the slope , which is observed for @xmath126 in the non interacting case , and which is now limited to a very tiny low - temperature regime .
these results explain why , in the presence of a strong correlation , the effect of the vhs is not only small , but even _ opposite _ to what one can expect from a sommerfeld expansion at @xmath116 : the deviation from the low - t `` sommerfeld '' regime occurs really at very low t , partly reducing the effect of the qp renormalization on the overall variation of @xmath96 between @xmath12 and @xmath127 . as a second point , one can note that , apart small deviations , the low temperature behavior of the sum rule can be considered to first approximation quadratic , in agreement with dmft calculations and with most of the experimental measurements .
one should also remark that , after the rescaling , the size of the temperature variation of the sr becomes much closer ( some percent going from @xmath12 to @xmath128 ) to that of the dmft data and the experimental observations , while even at the doping of the vhs the overall relative variation of sr does not exceed @xmath129 . although one can not expect this oversimplified analysis to reproduce exactly all the dmft results ( e.g.
, it would predict a slightly smaller temperature variation with respect to dmft for lsco ) , it sheds light on the essence of the physics which determines the sum rule behavior in the presence of strong correlation .
more precisely it indicates rather clearly that the main features of the integrated optical spectra are determined by the physics of a strongly renormalized fermi liquid , and the effects of the two - dimensional bandstructure are generally small .
in this manuscript we presented a dmft study of the optical sumrules of the two - dimensional hubbard model , aiming to understand the interplay of the strong electron - electron correlations and the properties of the two - dimensional density of states , namely the van hove singularity .
even if both correlations@xcite and the vhs singularity@xcite have been invoked to explain the anomalously large temperature dependence of the optical spectral weight , the two effects are found not to be cooperative .
our dmft analysis shows that correlations actually reduce the effect of the vhs on the optical sumrules .
the mechanism underlying this reduced effect is a shrinking of the temperature scale below which the effects of the vhs are appreciable . when correlations are strong ( namely in doped cuprates ) this effective temperature scale becomes lower than the temperatures accessed in the experiments .
this means that the inclusion of a more realistic bandstructure can not alterate the conclusions of ref .
about the necessity of considering strong correlations to get the correct order of magnitude of observed t - dependence of the optical spectral weight in the cuprates ( as well as the doping dependence of its @xmath12 extrapolation ) . at the same time
, our results imply that the almost negligible doping dependence of the temperature variation of @xmath100 which has been clearly observed in several experiments , can not be explained within a pure dmft calculation , even including the effects of a more realistic bandstructure . in this respect ,
non - local corrections beyond the dmft level certainly play a role , since they can introduce different renormalizations for the hopping parameters and the qp weight , and the physics of the mott transition can become richer , including for example pseudogap features at low temperatures .
the inclusion of such non local effects certainly represent an interesting challenge for future studies .
it is a pleasure to thank g. sangiovanni , c. castellani , l. benfatto , a.j .
millis , p. hansmann , k. held , l. de medici , m. ortolani , p. calvani for useful discussions .
m.c . acknowledges financial support of miur prin 2005 , prot .
200522492 .
a. f. santander - syro , r. p. s. m. lobo , n. bontemps , z. konstantinovic , z. li , and h. raffy , phys .
lett . * 88 * , 097005 ( 2002 ) ; a. f. santander - syro , r. p. s. m. lobo , and n. bontemps , europhys . lett . * 62 * , 568 ( 2003 ) .
f. carbone , a. b. kuzmenko , h. j. a. molegraaf , e. van heumen , v. lukovac , f. marsiglio , d. van der marel , k. haule , g. kotliar , h. berger .
s. couriault , p.h .
kes , and m. li , phys .
b * 74 * , 064510 ( 2006 ) . | we theoretically analyze some of the anomalies of the optical sumrules in the high - temperature superconductors .
in particular we address the particularly strong dependence on temperature of the sumrule in the normal state . both electron - electron correlations and the presence of a van - hove singularity
have been shown to enhance such a dependence . here
we consider both effects simultaneously by means of dynamical mean - field theory for a two dimensional hubbard model with realistic parameters for different cuprates , and we find that the two effects are not cooperative , as they appear to compete one another in the region of parameters relevant for the experiments . . |
since the advent of gauge theory , many new construction techniques , such as rational blowdowns , generalized fiber sum , knot surgery and luttinger surgery , have been introduced and successfully employed to produce symplectic @xmath5-manifolds homeomorphic but not diffeomorphic to rational surfaces @xmath1 , for @xmath8 .
( e.g. @xcite . )
all these @xmath5-manifolds should admit symplectic _ lefschetz pencils _ ( which , by definition , always have base points ) by the ground - breaking work of donaldson @xcite . despite this a priori existence result ,
no explicit lefschetz pencils on these _ small _ ( as in _ small _ second homology ) exotic @xmath5-manifolds were known up to date .
the main goal of this article is to introduce novel ways of constructing _ small _ positive factorizations ( as in _ small _ number of positive dehn twists ) of the boundary multi - twist in the mapping class group @xmath9 of a compact genus-@xmath7 surface with @xmath10 boundary components , which correspond to monodromy factorizations of genus-@xmath7 lefschetz pencils on small symplectic @xmath5-manifolds .
a first application yields the first explicit examples of pencils on exotic rational surfaces : [ main1 ] there are symplectic genus-@xmath0 lefschetz pencils @xmath11 ) whose total spaces are homeomorphic but not diffeomorphic to @xmath12 for @xmath13 .
they realize the smallest possible genera pencils in these homeomorphism classes .
a key result leading to theorem [ main1 ] is that @xmath14 , for @xmath15 , does not admit a genus @xmath16 lefschetz fibration or a genus-@xmath7 lefschetz pencil with @xmath17 base points ( lemma [ notrational ] ) , which provides a new criterion to argue that the total spaces of such pencils / fibrations like the ones in theorem [ main1 ] are exotic .
we moreover observe that @xmath18 is the smallest possible genus for a lefschetz pencil on these exotic @xmath5-manifolds ( remark [ smallest ] ) .
we hope that the additional information on these exotic symplectic @xmath5-manifolds we construct , namely the existence of genus-@xmath0 pencils on them , will help with understanding whether there are distinct minimal symplectic @xmath5-manifolds in the homeomorphism classes @xmath14 , for @xmath19 ( a question which is still open as of today ; see remark [ distinctsymp ] ) . to produce the positive factorizations for the pencils in theorem [ main1 ]
, we employ a new way of deriving a positive factorization @xmath20 in @xmath9 , @xmath21 from a collection of positive factorizations @xmath22 of _ boundary twists _ in @xmath23 , with @xmath24 and @xmath25 . for @xmath26 the lefschetz pencil / fibration corresponding to @xmath20 and @xmath27 to @xmath28
, we say @xmath26 is obtained by _ breeding _
this involves embedding the latter into @xmath9 as factorizations for _ achiral _ lefschetz pencils / fibrations , and then canceling matching pairs of positive and negative dehn twists . the idea for this construction scheme comes from the smallest hyperelliptic genus-@xmath0 lefschetz _ fibration _ mustafa korkmaz and the author produced in @xcite . a careful variation of our construction of @xmath27 yield to our next theorem : [ main2 ] for any pair of non - negative integers @xmath30 , there exists a symplectic genus-@xmath0 lefschetz pencil @xmath31 with @xmath32 , , and @xmath33 , for each @xmath34 .
infinitely many of these pencils have total spaces homotopy inequivalent to a complex surface .
in contrast , i know of only one explicit example of a non - holomorphic genus-@xmath0 lefschetz _ pencil _ in the literature , due to ivan smith .
( @xcite[theorem 1.3 ] ; see also remark [ nonholompencil ] . )
our construction technique yielding the pencils in theorem [ main2 ] can be easily generalized to obtain many other non - holomorphic pencils .
the second part of our article will deal with new constructions of symplectic calabi - yau surfaces .
recall that a symplectic @xmath5-manifold @xmath35 is called a _
symplectic calabi - yau surface _
( scy ) if it has trivial canonical class @xmath36 , in obvious analogy with complex calabi - yau surfaces .
as shown by tian - jun li , any _ minimal _ symplectic @xmath5-manifold with kodaira dimension zero is a symplectic calabi - yau surface or has trivial canonical class @xcite . up to date , the only known examples of symplectic calabi - yau surfaces that are not diffeomorphic to complex calabi - yau surfaces are torus bundles over tori ( at least _
virtually_)-bundles over @xmath37-bundles over @xmath38 , although not known to be torus bundles themselves , are finitely covered by them @xcite .
] , leading to the question @xcite : [ scyclassification ] is every symplectic calabi - yau surface with @xmath39 diffeomorphic to an oriented torus bundle over a torus
? works of tian - jun li and stefan bauer independently established that any symplectic calabi - yau surface with @xmath40 has the same rational homology as a torus bundle ( and that of @xmath41 and the enriques surface when @xmath42 ) @xcite .
all torus bundles over tori have _ solvmanifold fundamental groups _
@xcite , and notably , it was shown by stefan friedl and stefano vidussi that the _ homeomorphism type _ of a symplectic calabi - yau surface with a solvmanifold fundamental group is uniquely determined @xcite . as stated by tian - jun li @xcite ,
what seems to commonly provide a posteriori reasoning for a positive answer to the question [ scyclassification ] is the lack of any new constructions of symplectic @xmath5-manifolds of kodaira dimension zero .
surgical operations widely used to construct most interesting symplectic @xmath5-manifolds in the past , such as luttinger surgery , generalized fiber sums , knot surgery , or simplest rational blow - down operations , do not produce any new scys @xcite .
we will present a new construction of symplectic calabi - yau surfaces with _ via lefschetz pencils _ , where the genus of the pencil @xmath26 versus @xmath43 , building on the work of cliff taubes , will determine the kodaira dimension of @xmath44 ( lemma [ scypencil ] ; also @xcite ) .
once again , we will focus on generating the smallest genus examples : genus-@xmath0 pencils . to produce our examples
, we will use a version of the aforementioned breeding technique for producing new positive factorizations in @xmath45 , which we will call _ inbreeding _ , where this time copies of the same positive factorization in @xmath46 will be embedded into @xmath45 by identifying some of the boundary components of the fiber @xmath47 and introducing extra @xmath48-generators this way .
the result is a family of positive factorizations @xmath49 in @xmath50 , where @xmath51 is parametrized by pairs of mapping classes @xmath52 , and the curves @xmath53 are as shown in figure [ scycurves ] . here
we inbreed the monodromy factorizations of matsumoto s well - known genus-@xmath54 lefschetz fibration @xcite , and using further lifts of it obtained by noriyuki hamada @xcite , we can even derive an explicit lift of @xmath55 , which is a positive factorization of the boundary multi - twist @xmath56 in @xmath57 . letting @xmath58
denote the corresponding symplectic genus-@xmath0 lefschetz pencil , we obtain ( theorem [ scyfamily ] ) : [ main3 ] positive factorizations @xmath55 prescribe symplectic genus-@xmath0 lefschetz pencils on a family of symplectic calabi - yau surfaces @xmath59 in all rational homology classes of torus bundles over tori .
this provides a first step towards analyzing the monodromies of pencils on a rich family of scys , as proposed by simon donaldson in @xcite[problem 5 ] .
the first question that arises here is whether or not all @xmath59 are diffeomorphic to torus bundles over tori , in reference to question [ scyclassification ] .
torus bundles over tori _ with sections
_ do admit symplectic genus-@xmath0 pencils @xcite ; however the uncanny freedom we have in the choice of @xmath60 appears to exceed that of a torus bundle with a section ( see the discussion in remark [ comparesmith ] ) .
the explicit nature of our construction allows us to derive explicit presentations for @xmath61 , which for instance yields an @xmath44 ( when @xmath60 is trivial ) that is at least homeomorphic to the @xmath5-torus .
we will discuss this example in complete detail in section [ sec:4torus ] .
it is not clear to us at this point whether or not _ all _ @xmath61 are @xmath5-dimensional solvmanifold groups ( some of them certainly are ) , which we hope to understand in future work . on the other hand
, we observe that _ all _ @xmath59 are obtained from @xmath44 , our scy homeomorphic to the @xmath5-torus , via _ fibered _ luttinger surgeries along lagrangian tori or klein bottles ( remark [ compareholi ] ) .
so if our family of @xmath59 is equal to the family of torus bundles over tori , it immediately confirms an improved version of an interesting conjecture by tian - jun li and chung - i ho @xcite[conjecture 4.9 ] : _ any smooth oriented torus bundle @xmath44 possesses a symplectic structure @xmath62 such that @xmath35 can be obtained by applying luttinger surgeries to @xmath63 _ ( remark [ compareholi ] ) .
finally , theorem [ main3 ] implies that we have symplectic genus-@xmath0 pencils @xmath64 with @xmath65 for any @xmath66 .
this yields further ( infinitely many ) examples of non - holomorphic genus-@xmath0 pencils , in the same fashion as in our earlier examples in theorem [ main2 ] . the techniques we employed to produce the genus-@xmath0 pencils in theorems [ main1 ] , and [ main3 ] can be easily adapted to produce higher genera pencils on similar @xmath5-manifolds ; we can in fact derive many of these by repeatedly breeding / inbreeding the same collection of positive factorizations used in this paper .
this direction will be explored elsewhere .
we will however discuss one immediate generalization of our construction of a genus-@xmath0 pencil to higher genera .
it will allow us to address a problem raised by mustafa korkmaz @xcite[problem 2.7 ] , which we re - express here for pencils ( which are even more constrained than fibrations ) : is @xmath67 for any non - trivial genus-@xmath7 lefschetz pencil @xmath26 ? is there an upper bound on @xmath68 linear in @xmath7 , and sharper than @xmath69 ? in section [ sec : bound ] , generalizing our construction of a genus-@xmath0 pencil on @xmath44 homeomorphic to the @xmath5-torus , we will produce genus-@xmath7 lefschetz pencils violating the above bound for every odd @xmath70 . though not much ; we will indeed revamp korkmaz s question by raising the upper bound to @xmath71 ( question [ b1refinedquestion ] ) . _ acknowledgements .
_ we would like to thank weimin chen , bob gompf , jonathan hillman , tian - jun li , and stefano vidussi for helpful conversations related to our constructions in this paper .
thanks to noriyuki hamada for telling us about his preprint on the sections of the matsumoto fibration . our results on small symplectic calabi - yau surfaces via lefschetz fibrations with exceptional sections
were first presented at the great lakes geometry conference in ann arbor in march 2015 ; we would like to thank the organizers for motivating discussions .
last but not least , we are grateful to mustafa korkmaz for his interest in our work and numerous stimulating discussions .
the author was partially supported by the nsf grant dms-@xmath72 .
here we quickly review the definitions and basic properties of lefschetz pencils and fibrations , dehn twist factorizations in mapping class groups of surfaces , and symplectic calabi - yau surfaces .
the reader can turn to @xcite for more details . a _ lefschetz pencil _ on a closed , smooth
, oriented @xmath5-manifold @xmath44 is a smooth surjective map @xmath73 , defined on the complement of a non - empty finite collection of points @xmath74 , such that around every _ base point _ @xmath75 and _ critical point _ @xmath76 there are local complex coordinates ( compatible with the orientations on @xmath44 and @xmath77 ) with respect to which the map @xmath78 takes the form @xmath79 and @xmath80 , respectively .
lefschetz fibration _ is defined similarly for @xmath81 .
blowing - up all the base points @xmath75 in a pencil @xmath26 , one obtains a lefschetz fibration @xmath82 with disjoint @xmath83-sphere sections @xmath84 corresponding to @xmath75 , and vice versa .
we say @xmath26 is a _
genus @xmath7 lefschetz pencil _ or _ fibration _ for @xmath7 the genus of a _ regular fiber _
@xmath85 of @xmath78 .
the fiber containing the critical point @xmath76 has a nodal singularity at @xmath76 , which locally arises from shrinking a simple loop @xmath86 on @xmath85 , called the _ vanishing cycle_. a singular fiber of @xmath26 is called _ reducible _ if @xmath86 is separating . when @xmath86 is null - homotopic on @xmath85 , one of the fiber components becomes an _ exceptional sphere _ , an embedded @xmath54-sphere of self - intersection @xmath87 , which we can blow down without changing the rest of the fibration .
in this paper we use the term lefschetz fibration only when the set of critical points @xmath88 is non - empty , i.e. when the lefschetz fibration is _ non - trivial_. we moreover assume that the fibration is _ relatively minimal _ , i.e. it there are no exceptional spheres contained in the fibers , and also that the points @xmath76 lie in distinct _ singular fibers _ , which can be always achieved after a small perturbation .
a widely used way of constructing lefschetz fibrations is the _ fiber sum _ operation : if @xmath27 are genus-@xmath7 lefschetz fibrations with regular fiber @xmath89 for @xmath90 , then their _ fiber sum _ is a genus-@xmath7 lefschetz fibration @xmath26 , where @xmath44 is obtained by removing a fibered tubular neighborhood of each @xmath89 from @xmath91 and then identifying the resulting boundaries via a fiber - preserving , orientation - reversing self - diffeomorphism , and where @xmath78 restricts to @xmath92 on each @xmath93 .
a lefschetz fibration @xmath26 that can be expressed in this way is called _ decomposable _ , and each @xmath27 are called its _
summands_. allowing local models @xmath94 around the critical points @xmath76 , called _ negative _ critical points , we can extend all of the above to _ achiral _ lefschetz pencils and fibrations .
let @xmath95 denote a compact , connected , oriented surface genus @xmath7 with @xmath96 boundary components .
we denote by @xmath9 its _ mapping class group _ ; the group composed of orientation - preserving self homeomorphisms of @xmath95 which restrict to the identity along @xmath97 , modulo isotopies that also restrict to the identity along @xmath97 .
we write @xmath98 , and @xmath99 for simplicity .
denote by @xmath100 the positive ( right - handed ) dehn twist along the simple closed curve @xmath101 .
the inverse @xmath102 denotes the negative ( left - handed ) dehn twist along @xmath103 .
let @xmath104 be a _
non - empty _ collection of simple closed curves on @xmath95 , which do not become null - homotopic when @xmath97 is capped off by disks , and let @xmath105 be a collection of curves parallel to distinct boundary components of @xmath95 . if the relation @xmath106 holds in @xmath9 , we call the word @xmath107 on the left - hand side a _ positive factorization _ of the boundary multi - twist @xmath108 in @xmath109 .
( we will also use @xmath110 instead of @xmath111 to denote boundary components at times when there are several surfaces with boundaries involved in our discussion . ) capping off @xmath97 induces a homomorphism @xmath112 , under which @xmath107 maps to a similar positive factorization of the identity element @xmath113 .
the positive factorization in ( [ factorization ] ) gives rise to a genus-@xmath7 lefschetz fibration @xmath82 with @xmath96 disjoint @xmath83-sections @xmath84 , and equivalently a genus-@xmath7 lefschetz pencil @xmath26 with @xmath96 base points .
identifying the regular fiber @xmath85 with @xmath114 , we can view the vanishing cycles as @xmath86 .
in fact , every lefschetz fibration or pencil can be described by a positive factorization @xmath107 as in ( [ factorization ] ) @xcite , which is called the _ monodromy factorization _ of @xmath115 or @xmath78 .
let @xmath107 be a positive factorization of the form @xmath116 in @xmath9 , where @xmath117 are products of positive dehn twists along curves which do not become null - homotopic when @xmath97 is capped off by disks .
if @xmath118 , as a mapping class , commutes with some @xmath119 , we can produce a new positive factorization @xmath120 , where @xmath121 denotes the conjugate factorization @xmath122 in this case , we say @xmath55 is obtained from @xmath107 by a _ partial conjugation _
@xmath60 along the factor @xmath123 .
when @xmath124 are positive factorizations , @xmath125 is another positive factorization , lefschetz fibration corresponding to which is a fiber sum of the fibrations corresponding to @xmath126 and @xmath127 . allowing _ negative _ dehn twists , which correspond to negative critical points
, we can more generally work with _ factorizations _ for achiral lefschetz fibrations and pencils .
all of the above extend to this general case .
it was shown by donaldson that every symplectic @xmath5-manifold @xmath35 admits a _
lefschetz pencil whose fibers are symplectic with respect to @xmath62 @xcite .
conversely , generalizing a construction of thurston , gompf showed that the total space of a ( nontrivial ) lefschetz fibration , and in particular blow - up of any pencil , always admits a symplectic form @xmath62 with respect to which all regular fibers and any preselected collection of disjoint sections are symplectic @xcite .
whenever we take a symplectic form @xmath62 on a lefschetz pencil or fibration @xmath26 we will assume that it is of thurston - gompf type , with respect to which any explicitly discussed sections of @xmath78 will always be assumed to be symplectic as well .
the kodaira dimension for projective surfaces can be extended to symplectic @xmath5-manifolds as follows : let @xmath128 be the canonical class of a minimal model @xmath129 of @xmath35 .
the _ symplectic kodaira dimension _ of @xmath35 , denoted by @xmath130 is defined as @xmath131<0 \mbox { or } k_{x_{\text{min}}}^{2}<0 \\ 0 & \mbox{\ \ if } k_{x_{\text{min}}}\cdot[\omega_{\text{min}}]= k_{x_{\text{min}}}^{2}=0\\ 1 & \mbox{\ \ if } k_{x_{\text{min}}}\cdot[\omega_{\text{min}}]>0\mbox { and } k_{x_{\text{min}}}^{2}=0\\2 & \mbox{\ \ if } k_{x_{\text{min}}}\cdot[\omega_{\text{min}}]>0\mbox { and } k_{x_{\text{min}}}^{2}>0\end{array}\right .\ ] ] remarkably , @xmath132 is not only independent of the minimal model @xmath129 but also of the chosen symplectic form @xmath62 on @xmath44 : it is a smooth invariant of the @xmath5-manifold @xmath44 @xcite .
symplectic @xmath5-manifolds with @xmath133 are classified up to symplectomorphisms , which are precisely the rational and ruled surfaces @xcite .
symplectic @xmath5-manifolds with @xmath134 are the analogues of the calabi - yau surfaces that have torsion canonical class @xcite
. it has been shown by tian - jun li , and independently by stefan bauer @xcite , that the rational homology type of any _ minimal _ symplectic calabi - yau surface is that of either a torus bundle over a torus , the @xmath41 surface , or the enriques surface .
here we construct explicit positive factorizations for symplectic genus-@xmath0 lefschetz pencils , whose total spaces are homeomorphic but not diffeomorphic to the complex rational surfaces . in @xcite , we have obtained the following relation in @xmath135 @xmath136 which can be rewritten as @xmath137 where @xmath138 .
embed this relation into @xmath139 so that @xmath140 is mapped to @xmath141 and the remaining curves are as in figure [ exoticcurves ] , denoted by the same letters .
( so , along with the embedding of @xmath142 , the original curves on @xmath135 can be all seen in the obvious subsurface of @xmath139 in figure [ exoticcurves ] . )
we thus get the following relation in @xmath139 : @xmath143 or @xmath144 , where we set @xmath145 .
note that @xmath146 and @xmath147 all commute with each other .
of the first embedding are given on the left .
curves @xmath148 of the second embedding ( along with the same @xmath149 ) and @xmath150 of the third embedding are on the right .
, width=453 ] our second building block will be a positive factorization of the boundary multi - twist in @xmath151 , which is a
_ lift _ of the monodromy factorization of matsumoto s well - known genus-@xmath54 lefschetz fibration @xcite .
such a lift to @xmath135 was first found by ozbagci and stipsicz in @xcite and a further lift to @xmath151 by korkmaz in @xcite .
we will however work with another lift discovered by noriyuki hamada , which yields the following relation in @xmath151 @xcite : @xmath152 where @xmath111 are the boundary parallel curves , and the curves @xmath153 and @xmath149 are as shown on the left - hand side of figure [ matsumoto ] .
we can rewrite this relation in @xmath151 as @xmath154 where each @xmath155 for @xmath156 ( figure [ matsumoto ] ) .
let us first view this relation in @xmath135 by capping off one of the boundary components , say @xmath157 .
embed the resulting relation into @xmath139 , as induced by the same embedding of @xmath135 into @xmath139 we used above so that the boundary is mapped to @xmath141 , and all the other curves are as shown in figure [ exoticcurves ] , once again denoted by the same letters for simplicity .
so the following holds in @xmath139 : @xmath158 or @xmath159 , where @xmath160 . here
@xmath161 and @xmath147 all commute with each other .
finally , we will embed the lat relation we had above in @xmath151 into @xmath139 using an embedding of @xmath162 into @xmath163 such that @xmath157 is mapped to @xmath103 , @xmath164 is mapped to @xmath165 , and the remaining curves are as shown in figure [ exoticcurves ] , this time their labels are decorated by prime .
this gives a third relation in @xmath139 : @xmath166 or @xmath167 , where @xmath168 .
similar to the above , @xmath169 and @xmath147 all commute with each other .
now for @xmath170 ( where @xmath171 are as in figure [ pi1generators ] ) , and in fact for any mapping class in @xmath139 that keeps @xmath149 and @xmath172 fixed , we have @xmath173 where the first equality follows from the commutativity relations noted above , and the second equality follows from the relations @xmath174@xmath175 .
so @xmath176 is a positive factorization of @xmath177 in @xmath139 .
similarly , we get two more positive factorizations @xmath178 and @xmath179 of @xmath177 in @xmath139 .
each @xmath28 prescribes a symplectic genus-@xmath0 lefschetz fibration @xmath180 with a @xmath83-section , which can be blown - down to arrive at a symplectic genus-@xmath0 lefschetz pencil @xmath27 with one base point .
they are all obtained by _ breeding _ the genus-@xmath54 lefschetz pencils whose monodromies we have lifted to @xmath139 .
[ homeo ] we will prove the following : [ lemhomeo ] @xmath91 is homeomorphic to @xmath181 , for @xmath34 . the proof will follow from freedman s celebrated work @xcite , for which we will determine the euler characteristic , the signature , and the spin type of @xmath91 , while showing that @xmath91 is simply - connected for all @xmath34 .
the euler characteristic of @xmath91 is given by @xmath182 where @xmath183 is the number of dehn twists in the positive factorization @xmath28 . since @xmath184 ,
we have @xmath185 , and in turn , @xmath186 . the signature of @xmath91 can be calculated as @xmath187 using ozbagci s algorithm @xcite .
here is a quick argument for this calculation : we have obtained the monodromy factorization of @xmath188 after canceling a pair of @xmath189 and @xmath190 , and two pairs of @xmath147 and @xmath191 in the factorization @xmath192 in @xmath50 ( where we capped off the boundary ) .
the latter is a factorization of a genus-@xmath0 _ achiral _ lefschetz fibration , which is a fiber sum of three achiral lefschetz fibrations prescribed by @xmath193 .
@xmath193 , and @xmath194 .
each one of these three fibrations is obtained by a _ section sum _ of an achiral genus-@xmath54 lefschetz fibration and a trivial torus fibration along a section of self - intersection zero , so by novikov additivity , its signature is equal to that of the achiral genus-@xmath54 fibration .
the hyperelliptic signature formula for genus-@xmath54 fibrations @xcite allows us to easily calculate these as @xmath195 , @xmath195 , and @xmath196 , which add up to give signature @xmath197 for the original genus-@xmath0 achiral lefschetz fibration .
the signature of @xmath188 is the same as the signature of this achiral fibration , since canceling a pair of separating dehn twists with opposite signs amounts to taking out a pair of @xmath198 contribution to the signature .
hence , we get @xmath199 . by identical arguments ,
we get @xmath200 and @xmath201 .
clearly each @xmath91 has an odd intersection form : this can be easily seen from the presence of the reducible fiber component split off by the separating vanishing cycle @xmath149 and not hit by the @xmath83-sphere .
( alternatively , we can use rokhlin s theorem after showing that @xmath202 . )
lastly , we will show that @xmath203 , by calculating the latter as the quotient of @xmath204 by the subgroup normally generated by the vanishing cycles of @xmath180 , for each @xmath34 .
it suffices to find enough relations that kill the generators @xmath205 of @xmath204 ( which are as shown in figure [ pi1generators ] ) , so we do not need to work with the full presentation .
each positive factorization @xmath28 contains the factor @xmath206 .
so the following relations hold for the finite presentations we have for each @xmath207 : @xmath208[a_2,b_2][a_3,b_3]=1,\label{eqn:0 } \\ & \phantom{o } & [ a_1,b_1]=1 , \label{eqn : c } \\ & \phantom{o } & a_2a_3=1 , \label{eqn : b'0 } \\ & \phantom{o } & a_2 \bar{b}_2 a_3 \bar{b}_3 = 1 , \label{eqn : b'1 } \\ & \phantom{o } & b_3 b_2 = 1 , \label{eqn : b'2 } \end{aligned}\ ] ] where the relators @xmath209@xmath210 come from the vanishing cycles @xmath211 , respectively .
we have @xmath212 from @xmath213 and @xmath214 from @xmath210 .
together with @xmath215 , these imply @xmath216=[a_3,b_3]=1 $ ] .
together with @xmath209 , we conclude @xmath217=1 $ ] for every @xmath218 ( and @xmath219 becomes a trivial relation ) . from the factor @xmath220
, we get the following relations ( among many others ) @xmath221 b_1 a_2 = 1 , \label{eqn : x3 } \\ & \phantom{o } & b_2 b_1 [ b_3 , a_3 ] = 1 , \label{eqn : b2}\end{aligned}\ ] ] induced by the vanishing cycles @xmath222 and @xmath223 , respectively .
adding these to the previous relators from the factor @xmath224 , we immediately see that @xmath225=1 $ ] and @xmath226 imply @xmath227 .
so @xmath228 implies that @xmath229 can be generated by @xmath230 and @xmath231 . from the factor
@xmath232 we get the following relations ( again , among many others ) @xmath233 = 1 , \label{eqn : b1 } \\ & \phantom{o } & b_2 b_1 [ b_3 , a_3 ] = 1 , \nonumber\end{aligned}\ ] ] induced by the vanishing cycles @xmath234 and @xmath223 , respectively .
we get @xmath235 , and together with the relators from @xmath224 we once again get @xmath227 , since @xmath236=1 $ ] .
since the positive factorization @xmath237 contains the factor @xmath238 and @xmath239 both contain the factor @xmath240 , the above discussion shows that every @xmath207 is a factor of the abelian group @xmath241 generated by @xmath230 and @xmath242 .
( in fact no additional relators come from the remaining twists in @xmath243 or @xmath224 . ) for the remaining relators coming from the conjugated factors @xmath244 or @xmath245 , it therefore suffices to look at the relators they induce in @xmath246 .
moreover , this allows us to simply look at the homology classes of the vanishing cycles in these conjugated factors .
( we havent bothered to give explicitly for this very reason . ) without the conjugated factor , in each case we have the abelianized relations @xmath247 and depending on whether @xmath28 contains the factor @xmath220 or @xmath232 , either @xmath248 where we used @xmath249 to simplify . for our conjugation @xmath170 , we easily check using the picard - lefschetz formula that we get the additional relator @xmath250 from both @xmath244 and @xmath245 whereas , depending on whether it is @xmath244or @xmath245 we get either @xmath251 therefore , @xmath249 and @xmath252 , which hold for all @xmath207 , imply @xmath253 .
the relators in @xmath254 , @xmath255 and @xmath256 , @xmath257 involved in a given @xmath28 then easily give @xmath258 , for each @xmath34 .
hence in all cases we have got @xmath203 . by freedman
, we have @xmath91 homeomorphic to @xmath181 , for @xmath34 . a quick amendment to the title
: we will determine the `` non - diffeomorphism type '' of @xmath27 , namely , we will prove that @xmath91 are not diffeomorphic to rational surfaces .
this will follow from the next lemma .
[ notrational ] the rational surface @xmath259 , for @xmath15 , does not admit a genus @xmath16 lefschetz fibration or a genus-@xmath7 lefschetz pencil with @xmath17 base points .
it suffices to show this for @xmath260 , since we can always blow - up on the fibers to produce a genus-@xmath7 lefschetz fibration ( resp .
pencil ) on @xmath44 from a given genus-@xmath7 lefschetz fibration ( resp .
pencil ) on @xmath259 , for any @xmath19 .
assume that @xmath44 is equipped with a genus @xmath16 lefschetz fibration or a pencil with @xmath17 base points .
for our arguments to follow , it will be convenient to allow @xmath96 to be non - negative so that @xmath261 marks the fibration case .
let @xmath262 be the fiber class , where @xmath263 is generated by @xmath264 and @xmath265 , with @xmath266 , @xmath267 , and @xmath268 .
since @xmath269 , we have @xmath270 we can equip @xmath44 with a thurston - gompf symplectic form @xmath62 which makes the fibers symplectic .
moreover , we can choose an @xmath62-compatible almost complex structure @xmath271 , even a generic one in the sense of taubes ( see e.g @xcite ) , with respect to which @xmath78 is @xmath271-holomorphic ( for a suitable choice of almost complex structure on the base @xmath54-sphere ) .
it was shown by li and liu @xcite that for a generic @xmath62-compatible @xmath271 , the class @xmath264 in the rational surface @xmath44 has a @xmath271-holomorphic representative .
hence , @xmath85 and @xmath264 both have @xmath271-holomorphic representatives , which implies that @xmath272 .
there is a unique symplectic structure on @xmath44 up to deformation and symplectomorphisms @xcite , so we can apply the adjunction formula to get @xmath273 since @xmath274 , and @xmath16 , from the above equalities we have @xmath275 where the last inequality is by cauchy - schwartz , and in turn @xmath276 implying @xmath277 .
the contradiction shows that there exists no such symplectic surface @xmath85 . in turn
, there is no such fibration or pencil .
in fact , if @xmath44 is a symplectic @xmath5-manifold in the homeomorphism class of a rational surface @xmath259 with @xmath15 , @xmath44 is exotic _ if and only if _ it admits a genus @xmath16 pencil with @xmath17 base points .
this follows from lemma [ notrational ] and donaldson s result on the existence of lefschetz pencils on symplectic @xmath5-manifolds ( together with the fact there is a unique genus-@xmath278 pencil , which is on @xmath279 ) . combining the lemmas [ lemhomeo ] and [ notrational ] we conclude : [ exoticrational ] @xmath280 are symplectic genus-@xmath0 lefschetz pencils whose total spaces are homeomorphic but not diffeomorphic to @xmath12 for @xmath281 .
[ smallest ] these are the smallest genera examples one can find .
any lefschetz pencil of genus @xmath282 or @xmath278 has a total space diffeomorphic to ( a blow - up of ) @xmath283 , @xmath284 or @xmath285 @xcite . after blowing - up all the base points of a genus-@xmath54 lefschetz pencil we arrive at a genus-@xmath54 lefschetz fibration on a non - minimal symplectic @xmath82 .
now if @xmath286 is an exotic rational surface , it can not have the rational homology type of a ( blow - up of a ) symplectic calabi - yau surface either .
thus there could be at most one base point , and as observed by sato , @xmath82 has at most one reducible fiber ; see theorem 5 - 5(iii ) and theorem 5 - 12(iii ) in @xcite . by lemmas 4 and 5 in @xcite ,
there is no genus-@xmath54 lefschetz fibration @xmath82 with at most one reducible fiber and say @xmath287 .
in fact , in @xcite , we have constructed _ minimal _ genus-@xmath54 lefschetz fibrations whose total spaces are homeomorphic to rational surfaces @xmath14 , for @xmath288 @xcite .
these fibrations decompose as fiber sums , which implies their minimality @xcite , and in turn why they are not diffeomorphic to blown - up rational surfaces .
the above observation generalizes to any genus @xmath289 lefschetz pencil @xmath26 whose total space is an exotic @xmath259 , for @xmath290 , allowing us to conclude that @xmath78 has at most @xmath291 base points .
it follows from the following more general observation : a collection of exceptional classes in the corresponding lefschetz fibration @xmath292 can be represented by disjoint multisections @xmath84 , so each one intersects the regular fiber @xmath85 positively at least once .
however , @xmath293 implies that @xmath294 by the work of sato @xcite . since @xmath44 is not rational ( nor can possibly be ruled ) , we already know that @xmath295 .
further , @xmath296 or @xmath278 implies that the minimal model of @xmath44 has @xmath297 , which is impossible for any @xmath44 , as @xmath298 and blow - down increases @xmath299 .
hence @xmath300 and has at most @xmath291 disjoint exceptional spheres .
in particular we see that each @xmath91 in theorem [ exoticrational ] is either minimal or it is only once blown - up of a minimal symplectic @xmath5-manifold . using a variation of our construction with @xmath0 different embeddings of @xmath220 into @xmath50
, we can get a genus-@xmath0 lefschetz _ fibration _ @xmath301 where @xmath302 is an exotic symplectic @xmath303 .
i expect this monodromy lifts to @xmath139 , and gives another pencil @xmath304 whose total space is an exotic @xmath305 .
[ distinctsymp ] the existence of exotic symplectic @xmath5-manifolds in the homeomorphism classes in theorem [ exoticrational ] was already established .
the first example of an exotic symplectic @xmath306 , for @xmath307 , was the complex dolgachev surface @xmath308 , as shown by donaldson @xcite . for @xmath309 , the first example was barlow s complex surface , as proved by kotschick @xcite . for @xmath310 ,
the first examples were constructed via generalized rational blowdowns by jongil park @xcite , which are symplectic by the work of symington @xcite .
infinitely many distinct smooth structures in these homeomorphism classes were constructed using logarithmic transforms , knot surgeries and luttinger surgeries @xcite ( all of which are indeed instances of surgeries along tori @xcite . )
notably , only for @xmath307 it is known that there is an _
infinite _ family of pairwise non - diffeomorphic _ symplectic _ @xmath5-manifolds in the homeomorphism class of @xmath306 ( e.g. @xcite )
. moreover , it was observed by stipsicz and szabo that seiberg - witten invariants can not distinguish any two _ minimal _
symplectic @xmath5-manifolds homeomorphic to @xmath311 for @xmath312 @xcite[corollary 4.4 ] .
it remains an open question whether there are two distinct symplectic @xmath5-manifolds in these homeomorphism classes @xcite[problem 11 ] .
it is thus desirable to have examples with more structure in order to address this intriguing question .
for instance , one can use other @xmath60 to obtain simply - connected symplectic genus-@xmath0 pencils in the same homeomorphism classes , possibly not isomorphic to the ones in theorem [ exoticrational ] . by varying our construction of the positive factorizations
@xmath28 , we can also produce symplectic genus-@xmath0 lefschetz pencils with any prescribed abelian group of rank at most @xmath54 .
let @xmath315 , where @xmath316 and @xmath317 are nonnegative integers .
for @xmath318 let @xmath319 be the lefschetz fibration prescribed by the positive factorization @xmath28 with @xmath320 , and @xmath321 be the corresponding pencil .
note that we still get a positive factorization @xmath28 since @xmath60 keeps @xmath149 and @xmath172 fixed .
we claim that @xmath322 . to verify it ,
let s go back to our fundamental group calculation via the finite presentation which has the generators @xmath323 of @xmath204 , and relators induced by the dehn twist curves in @xmath28 .
the arguments we gave in section [ homeo ] for the relators coming from the factors @xmath243 and @xmath224 appearing in @xmath28 apply here as well .
in particular , @xmath324 is also a quotient of the abelian group @xmath241 generated by @xmath230 and @xmath242 , and the following abelianized relations hold : @xmath325 importantly , there are no other relations coming from the factors @xmath243 and @xmath224 : this is easy to see by abelianizing the relators @xmath326@xmath327 , which include all the relators induced by the curves @xmath328 . missing are the relators induced by the separating curves @xmath329 from @xmath220 , the curves @xmath330 from @xmath220 , and the curves @xmath331 from @xmath224 .
the first two are trivial in homology , so they have no contribution to the list of relators we already have . on the other hand , for each @xmath156 , @xmath332 is homologous to @xmath333 , because @xmath334= [ t_c(b_j)]=[b_j ] + ( c \cdot b_j ) [ c]$ ] by the picard - lefschetz formula , where @xmath149 is a separating cycle .
similarly each @xmath335 is homologous to @xmath336
. therefore the abelianized relations they induce are identical to those we already had from @xmath337 . for our conjugation @xmath338 , we get the additional relators @xmath339 similar to before , @xmath340 and @xmath341 , which hold for all @xmath342 , imply @xmath343 .
the remaining relators involved in a given @xmath28 then easily give @xmath344 , for each @xmath34 .
hence , @xmath345 , as claimed .
since the monodromy of each @xmath346 is equivalent to that of @xmath180 by a partial conjugation with @xmath347 , they have the same euler characteristic and signature .
so @xmath32 and @xmath348 , for @xmath34 .
for each @xmath349 , we clearly have an infinite family of symplectic genus-@xmath0 pencils among @xmath350 with pairwise homotopy inequivalent total spaces easily distinguished by @xmath351 . by parshin and arakelov s proofs of the geometric shafarevich conjecture , there are finitely many holomorphic fibrations with fixed fiber genus @xmath352 and degeneracy @xcite .
this implies that all but finitely many of these symplectic pencils are non - holomorphic .
furthermore , for @xmath353 with @xmath354 , we get an infinite subfamily of pencils , whose total spaces can not be complex surfaces , since there is no complex surface @xmath355 with @xmath356 and @xmath357 ( e.g. cf . @xcite[lemma 2 ] ) .
( note that the same argument applies to the infinite family of symplectic calabi - yau surfaces we will construct in the next section . )
we have proved : [ nonholom ] there exists a symplectic genus-@xmath0 lefschetz pencil @xmath31 with @xmath32 , @xmath348 , and @xmath33 , for any pair of non - negative integers @xmath30 , and for each @xmath34 .
infinitely many of these pencils have total spaces homotopy inequivalent to a complex surface .
[ nonholompencil ] there is only one example of a non - holomorphic genus-@xmath0 lefschetz _ pencil _ in the literature , which was given by smith in @xcite[theorem 1.3 ] without appealing to donaldson s theorem .- sections which could be blown - down to produce pencils @xcite . ] ( we learn from naoyuki monden that in a joint work with hamada and kobayashi they produce two more examples . )
theorem [ nonholom ] improves the situation , at least quantitatively .
[ pi1genusinvariant ] following the works of donaldson @xcite and gompf @xcite , every finitely presented group is the fundamental group of a symplectic lefschetz pencil .
( also see @xcite and @xcite . )
an invariant of finitely presented group @xmath358 is the @xmath359 , the minimal genus of a genus-@xmath7 pencil @xmath26 with @xmath360 , which is not as easy to calculate in general .
however , theorem [ nonholom ] shows that for @xmath361 , @xmath362 .
well - known examples of genus-@xmath278 and genus-@xmath54 pencils show that @xmath363 for @xmath364 ( the pencil on @xmath365 ) , and @xmath366 for @xmath367 ( the pencil on the enriques surface ) or @xmath368 ( e.g. matsumoto s pencil on @xmath369 ) .
it seems likely that for all the other @xmath358 as above , @xmath370 .
here we will give a new construction of an infinite family of symplectic calabi - yau surfaces with @xmath372 and @xmath0 , along with a symplectic calabi - yau surface with @xmath4 , homeomorphic to the @xmath5-torus .
this will follow from our construction of new positive factorization of boundary multi - twists in @xmath57 corresponding to symplectic genus-@xmath0 lefschetz pencils .
we find it more convenient for presentation purposes to discuss a special example first , and discuss the most general construction afterward .
these will correspond to our examples with @xmath4 , and then the others with @xmath372 and @xmath0 .
our main building block will be the following relation in @xmath151 @xcite : @xmath152 where @xmath111 are the boundary parallel curves , and the curves @xmath153 and @xmath149 are as shown on the left - hand side of figure [ matsumoto ] . and
@xmath373 in hamada s lift of matsumoto s fibration .
on the left are the curves of the positive factorization in @xmath151 , along with the curves @xmath332 we get after hurwitz moves . on the right are the curves of the further lift in @xmath374 .
, width=377 ] the relation @xmath375 can be rewritten as @xmath154 where each @xmath155 for @xmath156 are given on the left - hand side figure [ matsumoto ] ) .
we can embed this relation into @xmath50 by gluing along the two boundary components @xmath157 and @xmath164 so that @xmath376 is mapped to @xmath172 and the curves @xmath377 are as shown in figure [ scycurves ] , denoted by the same letters for simplicity .
thus the following holds in @xmath50 : @xmath378 or @xmath379 , where @xmath380 .
clearly @xmath381 and @xmath147 all commute with each other .
a similar embedding into @xmath50 can be given by mapping @xmath376 to @xmath149 instead , where the curves @xmath377 are now as shown in figure [ scycurves ] , all decorated by a prime notation this time ( so @xmath149 indeed maps to @xmath172 above ) .
we thus get another relation in @xmath50 of the form : @xmath382 or @xmath383 , where @xmath384 .
similarly , @xmath385 and @xmath147 all commute with each other . combining the above , we get a positive factorization @xmath386 in @xmath50 , since @xmath387 let @xmath388 be the genus-@xmath0 lefschetz fibration prescribed by @xmath107 . we can equip @xmath292 with a gompf - thurston symplectic form @xmath389 , in particular @xmath286 is a symplectic @xmath5-manifold . we will show that @xmath390 is the @xmath5 times blow - up of a symplectic calabi - yau surface @xmath44 homeomorphic to the @xmath5-torus
. we will indeed show that it is a blow - up of a symplectic genus-@xmath0 pencil @xmath26 , which we will first argue implicitly below , and then explicitly in the next section .
[ scypencil ] let @xmath82 be a genus @xmath16 lefschetz fibration and @xmath391 . if @xmath286 is not a rational or ruled surface , then @xmath82 is a blow - up of a symplectic lefschetz pencil @xmath26 , where @xmath44 is a ( minimal ) symplectic calabi - yau surface
. we can equip @xmath82 with a gompf - thurston symplectic form @xmath389 . by the work of taubes @xcite ,
any _ minimal _ symplectic @xmath5-manifold has non - negative @xmath299 .
so @xmath392 , where @xmath393 is a minimal model for @xmath286 and @xmath277 , which means that @xmath286 contains at least @xmath394 disjoint exceptional spheres @xmath84 , @xmath395 .
there is an @xmath389-compatible almost complex structure @xmath271 with respect to which both a regular fiber @xmath85 of @xmath115 and all @xmath84 are @xmath271-holomorphic @xcite , which implies that @xmath396 . by the work of sato @xcite
( also see @xcite ) , this is only possible if the symplectic kodaira dimension @xmath397 . as we assumed
@xmath286 is not a rational or a ruled surface , we have @xmath398 @xcite , leaving the only possibility as @xmath399 and also that there are no other disjoint exceptional spheres than @xmath84 in @xmath286 .
thus @xmath400 , so one can blow - down these exceptional spheres ( which can be assumed to be symplectic after deforming @xmath389 if needed @xcite ) to arrive at a symplectic genus-@xmath0 lefschetz pencil @xmath26 , where @xmath401 is a minimal symplectic calabi - yau surface .
the signature of @xmath286 can be calculated as @xmath404 using the local signatures and ozbagci s algorithm @xcite .
since @xmath292 is not hyperelliptic , we do not have the luxury of employing endo s closed formula for a direct calculation , however the following observation allows us to deduce the signature from a calculation in @xmath405 , where any factorization is hyperelliptic : we have obtained @xmath107 from the factorization which had two pairs of @xmath189 and @xmath147 with opposite signs .
this is an _ achiral _
genus-@xmath0 lefschetz fibration , which is a fiber sum of two achiral fibrations prescribed by the factorizations @xmath406 and @xmath407 .
each one is a _ section sum _ of an achiral genus-@xmath54 lefschetz fibration and a trivial torus fibration along a section of self - intersection zero , which , by novikov additivity , has signature equals to that of the achiral genus-@xmath54 lefschetz fibration . by the hyperelliptic signature formula for genus-@xmath54 fibrations @xcite ,
the signature of the latter is @xmath195 . invoking novikov additivity again
, we conclude that the decomposable achiral genus-@xmath0 fibration has signature @xmath408 , the sum of the signature of its summands .
lastly , we claim that canceling the pairs @xmath409 and @xmath410 does not change the signature . this can be easily seen by splitting the fibration as @xmath411 ( or @xmath412 ) and the rest .
the piece with monodromy factorization @xmath411 has zero signature , which is easily calculated from the induced handle decomposition , or by observing that it can be further split into two identical fibrations whose total spaces have opposite orientations ( and thus opposite signatures ) . by novikov
additivity once again , the rest of the fibration has the same signature as the total signature of the achiral fibration .
so @xmath404 .
it follows that @xmath413 .
assuming @xmath414 for the moment , let us complete our arguments regarding the homeomorphism type of the minimal model @xmath44 of @xmath286 .
by lemma [ scypencil ] @xmath292 is a blow - up of a symplectic genus-@xmath0 lefschetz pencil @xmath26 , where @xmath44 is a symplectic calabi - yau surface .
since @xmath44 is a symplectic calabi - yau surface with @xmath415 , it has the rational homology type of a torus bundle over a torus @xcite .
moreover , since @xmath416 is a virtually poly-@xmath417 group , the borel conjecture holds in this case by the work of farrell and jones @xcite , and as observed by friedl and vidussi , this implies that a symplectic calabi yau surface with @xmath418 is unique up to homeomorphism @xcite .
hence @xmath44 is homeomorphic to the well - known symplectic calabi - yau surface , the @xmath5-torus .
we now calculate @xmath419 , which will constitute the rest of this section . using the standard handlebody decomposition for @xmath292 , we can obtain a finite presentation for @xmath420 , where @xmath421 denotes the subgroup of @xmath204 generated normally by the vanishing cycles of @xmath115 . for @xmath422 standard generators of @xmath423 as given in figure [ pi1generators ] , we get @xmath424 [ a_2 , b_2 ] [ a_3 , b_3 ] , r_1 , \ldots , r_{12 } \ , \rangle \ , , \ ] ] where each @xmath425 is a relation obtained by expressing a corresponding vanishing cycle ( oriented arbitrarily ) in @xmath422 . recall that @xmath292 has a section ( in fact @xmath5 @xmath83-sections ) , so the above presentation is complete , there is no other relation .
we have the following relations , the ones after the first one coming from @xmath426 all induced by the vanishing cycles @xmath427 , @xmath428 , @xmath223 , @xmath429 , @xmath430 , @xmath431 , @xmath432 , @xmath433 , @xmath434 , @xmath435 , @xmath436 , @xmath437 ( which are all given in figure [ scycurves ] , in the given order : @xmath438[a_2 , b_2][a_3,b_3]=1 , \label{0 } \\ & \phantom{o } & a_1 a_3=1 , \label{b0 } \\ & \phantom{o } & a_1 \bar{b}_1 a_2 b_2 \bar{a}_2 a_3 \bar{b}_3=1 , \label{b1 } \\ & \phantom{o } & \bar{b}_1 a_2 b_2 \bar{a}_2 \bar{b}_3 = 1 , \label{b2 } \\ & \phantom{o } & a_1 [ b_3 , a_3 ] \ , b_2 a_3 \bar{b}_2 \ , [ a_3 , b_3]=1 , \label{a0 } \\ & \phantom{o } & a_3 \bar{b}_3 \bar{b}_2 [ a_3 , b_3 ] \ , a^2_1 \bar{b}_1 \bar{a}_1 [ b_3 , a_3 ] \ , b_2 [ b_3 , a_3 ] \ , b_2=1 , \label{a1 } \\
& \phantom{o } & a_1 \bar{b}_1 \bar{a}_1 [ b_3 , a_3 ] \ , b_2 [ b_3 , a_3 ] \ , b_2 \bar{b}_3 \bar{b}_2 [ a_3 , b_3 ] = 1 , \label{a2 } \\ &
\phantom{o } & \bar{a}_2 a_1 a_2 a_3=1 , \label{b'0 } \\ & \phantom{o } & a_1 \bar{b}_1 a_2 a^2_3 \
, \bar{b}_3 \bar{a}_3 b_2 \bar{a}_2 = 1 , \label{b'1 } \\ & \phantom{o } & b_1 a_2 \bar{b}_2 a_3 b_3 \bar{a}_3 \bar{a}_2 = 1 , \label{b'2 } \\ & \phantom{o } & a_1 a_2 \bar{b}_2 a_3 b_2 \bar{a}_2=1 , \label{a'0 } \\
& \phantom{o } & a_1 \bar{b}_1 a_2 \bar{b}_2 a^2_3 \ , \bar{b}_3 \bar{a}_3 b^2_2 \ ,
\bar{a}_2=1 , \label{a'1 } \\ & \phantom{o } & \bar{b}_1 a_2 \bar{b}_2 a_3 \bar{b}_3 \bar{a}_3 b^2_2 \
, \bar{a}_2 = 1 , \label{a'2 } \end{aligned}\ ] ] here @xmath439 denotes the inverse of the element @xmath440 in the group . of the genus-@xmath0 fibration @xmath26 .
on the left are the curves coming from the factorization @xmath123 and on the right are those coming from the factorization @xmath441 , which correspond to the two different embeddings of the factorization in @xmath151 into @xmath50 .
( dotted lines drawn on the surfaces on the top are the images of @xmath376 under these embeddings.),height=699 ] when abelianized , the relations coming from each one of the triples @xmath442 , @xmath443 , @xmath444 , @xmath445 give the same three relations @xmath446 where we identified the images of the generators with the same letters .
any @xmath54 of these relations imply the other .
so we can eliminate @xmath447 and @xmath448 , and get a free abelian group of rank @xmath5 generated by @xmath449 and @xmath450 .
now , going back to the presentation we have above for @xmath451 , we easily see that it is also generated by @xmath452 , since @xmath453 by @xmath454 and @xmath455 by @xmath456 .
to prove @xmath414 , it therefore suffices to show that @xmath452 all commute with each other .
replacing @xmath229 with @xmath457 in @xmath458 gives @xmath459=1 $ ] .
from @xmath456 we have @xmath460 . substituting this in @xmath461 , and replacing @xmath229 with @xmath457 , we get @xmath236=1 $ ] . with @xmath462 and @xmath236=1 $ ] , the relation @xmath463 simplifies to @xmath464=1 $ ] .
so @xmath465 commute with each one of @xmath466 .
we can rewrite @xmath467 as @xmath468 , since @xmath462 and @xmath465 commutes with @xmath230 and @xmath469 .
so @xmath216=1 $ ] .
it implies that @xmath470 .
note that @xmath236=1 $ ] means @xmath471 .
using these last two identities and @xmath472 we derive @xmath473 .
so @xmath474 = [ a_2 , \bar{b}_2 b_3]=1 $ ] .
these commuting relations , together with @xmath216=1 $ ] , give us the last two commuting relations we need : @xmath475=[a_2 , b_3]=1 $ ] .
hence @xmath476 , generated by @xmath452 .
this completes our claim that @xmath26 is a symplectic genus-@xmath0 lefschetz pencil , where @xmath44 is a symplectic calabi - yau surface homeomorphic to the @xmath5-torus .
[ [ symplectic - calabi - yau - surfaces - with - b_12-and-3-via - pencils ] ] symplectic calabi - yau surfaces with @xmath372 and @xmath0 via pencils ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ we will now generalize the construction in the previous section , and construct an infinite family of examples of genus-@xmath0 lefschetz pencils on symplectic calabi - yau surfaces with @xmath372 and @xmath0 .
we will produce these pencils explicitly via positive factorizations of the boundary multi - twist @xmath477 in @xmath374 , where @xmath110 are boundary parallel curves for distinct boundary components of @xmath478 .
for this , we will use a further lift of the monodromy factorization of matsumoto s genus-@xmath54 fibration to @xmath374 given by hamada , which will allow us to obtain a nice and symmetric presentation at the end .
the factorization @xmath375 has a further lift to @xmath374 @xcite of the form : @xmath479 where again @xmath111 are the boundary parallel curves , and the curves @xmath480 are given on the right - hand side of figure [ matsumoto ] .
when the two boundary components @xmath481 , drawn in the middle of @xmath482 in figure [ matsumoto ] , are capped off , the curves @xmath483 descend to @xmath333 and @xmath373 to @xmath149 for each @xmath156 and @xmath90 .
the relation @xmath485 can be rewritten as the following relation in @xmath374 : @xmath486 where each @xmath487 for @xmath156 . embed this configuration of curves on @xmath482 into @xmath478 by gluing a cylinder with @xmath54 holes along two of the boundary components of @xmath478 .
choose the embedding so that @xmath157 maps to @xmath488 , @xmath164 to @xmath489 , @xmath490 to @xmath491 and @xmath492 to @xmath493 .
( see figure [ scygluing ] . )
we then map the interior of @xmath482 so that the curves @xmath494 and @xmath495 all map to the curves @xmath333 and @xmath332 in figure [ scycurves ] when the boundary components @xmath496 and @xmath497 are capped off .
( that is , we map the interior so that when @xmath484 is capped off , we get the first embedding we had in the previous section . ) thus the following holds in @xmath57 : @xmath498 or @xmath499 , where @xmath500
. a similar embedding into @xmath57 can be given by mapping @xmath157 to @xmath501 , @xmath502 , @xmath503 and @xmath504 ( see figure [ scygluing ] , where the interior is mapped in a similar fashion as above so as to get the curves @xmath336 and @xmath335 of figure [ scycurves ] .
we get another relation in @xmath57 of the form : @xmath505 or @xmath506 , where @xmath507 .
now let @xmath60 be any mapping class in @xmath57 which fix the curves @xmath508 .
then the product of @xmath509 and @xmath510 then give : @xmath511 in the first equality we used the commutativity of disjoint dehn twists @xmath512 , @xmath513 , @xmath514 , @xmath515 and that they all commute with @xmath516 ( and @xmath510 ) .
the second equality comes from our choice of @xmath60 so that @xmath60 commutes with dehn twists along @xmath517 , @xmath501 , @xmath489 , @xmath488 .
therefore @xmath518 is a positive factorization of the boundary multi - twist @xmath519 in @xmath57 for any @xmath60 in @xmath57 fixing the curves @xmath508 on @xmath478 .
[ scyfamily ] each @xmath58 is a symplectic genus-@xmath0 lefschetz pencil on a minimal symplectic calabi - yau surface , where @xmath44 is homeomorphic to the @xmath5-torus for @xmath520 .
the family of genus-@xmath0 lefschetz pencils @xmath521 contains a subfamily @xmath522 we will show that neither one of these ruled surfaces contains an embedded symplectic surface @xmath85 of genus @xmath0 and self - intersection @xmath5 , whereas the fiber of a pencil as in the statement of lemma would give such an @xmath85 . for @xmath524 ,
@xmath525 is generated by @xmath526 and @xmath527 , where @xmath528 , @xmath529 , and @xmath530 .
now let @xmath531 be a symplectic curve of genus @xmath0 with @xmath532 .
so @xmath533 , which implies @xmath534 .
since there is a unique symplectic structure on a minimal ruled surface up to deformations and symplectomorphisms @xcite , we can apply the adjunction formula and derive @xmath535 and get @xmath536 .
it however contradicts with @xmath534 above , so there is no such @xmath85 . for @xmath537 ,
@xmath525 is generated by the fiber @xmath538 and section @xmath539 of the degree-@xmath278 ruling on @xmath44 , where @xmath528 , @xmath540 , and @xmath530 .
let @xmath531 be a symplectic curve of genus @xmath0 with @xmath532 .
so @xmath541 , which implies @xmath542 .
by the same argument as above , we can apply the adjunction to derive @xmath543 and get @xmath544 , which contradicts with @xmath542 .
so there is no such @xmath85 in this case either . for @xmath520
we get the pencil @xmath26 constructed in the previous section , now with an explicit lift of its monodromy to @xmath57 as a positive factorization @xmath545 . since any @xmath58 can be derived from @xmath26 by a partial conjugation along the first factor @xmath516 in this factorization , @xmath546 , @xmath547 , and @xmath548 . by lemma [ scypencil ] , we conclude that @xmath59 is a minimal symplectic calabi - yau surface , provided it is not rational or ruled . from @xmath549 and @xmath550
we see that @xmath551 only if @xmath552 .
so the only rational or ruled surface with these characteristic numbers can be @xmath369 or @xmath523 .
however , by the previous lemma , neither one of these surfaces admit a genus-@xmath0 pencil , so any @xmath59 is a symplectic calabi - yau surface . to prove the second part of the theorem , consider genus-@xmath0 pencils @xmath58 with @xmath553 ( where @xmath6 and @xmath465 are as in figure [ pi1generators ] ) .
note that @xmath6 and @xmath465 are disjoint from @xmath508 , so @xmath60 fixes all .
for @xmath554 any pair of non - negative integers , we denote this pencil by @xmath555 , and its monodromy factorization by @xmath556 , @xmath553 . as we observed in the previous section , every triple of vanishing cycles @xmath442 , @xmath443 , @xmath444 , @xmath445
give the same three relations in @xmath557 : @xmath558 where the first and the third relations imply the second .
we can draw two conclusions : first , the vanishing cycles coming from the non - conjugated factor @xmath510 induce exactly these relations in @xmath559 .
second , the vanishing cycles coming from the conjugated factor @xmath509 induce the following relations we can easily calculate using the picard - lefschetz formula : @xmath560 we easily see that @xmath561 and @xmath562 imply @xmath563 , and @xmath564 and @xmath565 imply @xmath566 , whereas the others are already implied by these relations . from @xmath567 and @xmath568 , we conclude that @xmath559 is generated by @xmath569 with @xmath563 and @xmath570 .
hence @xmath571 , as promised .
[ comparesmith ] the very first question here is whether or not every @xmath59 is a torus bundle over a torus , as they are commonly conjectured to exhaust the list of all symplectic calabi - yau surfaces with @xmath40 .
if for any @xmath60 , @xmath61 is not a @xmath5-dimensional solvmanifold group @xcite , this would imply that @xmath59 is _ not _ a torus bundle over a torus , and it is a new symplectic calabi - yau surface .
as our arguments in the proof of theorem [ scyfamily ] show , if any partial conjugation along any hurwitz equivalent factorization to @xmath55 results in a pencil with a fundamental group which is not a solvmanifold group , we can arrive at a similar conclusion as well .
i do not know for the moment if either one of these can be realized or ruled out for all possible @xmath60 .
it is worth comparing our pencils with those on torus bundles over tori . in @xcite ivan
smith constructs genus-@xmath0 pencils on torus bundles _ admitting sections _
( not all do ) , by generalizing the algebraic geometric construction of holomorphic genus-@xmath0 pencils on abelian surfaces .
a natural question is whether or not our examples overlap with smith s .
( well , if they do , the explicit factorizations @xmath55 can be seen to prescribe very nice handle decompositions for such torus bundles . )
note that that any torus bundle over a torus with a section @xmath538 would admit a second disjoint section as well ; for any section of a surface bundle over a torus has self - intersection zero @xcite and can be pushed - off of itself .
so any genus-@xmath0 lefschetz fibration of smith would be determined by a pair @xmath52 _ subject to _
@xmath572=1 $ ] .
now for a comparison , let us note that the stabilizer subgroup of @xmath57 which fix @xmath517 , @xmath501 , @xmath489 , @xmath488 , is generated by dehn twists along curves which do not intersect the stabilized collection , and the obvious involution swapping the genus one surfaces bounded by @xmath573 and by @xmath574 .
( this involution maps each @xmath575 to itself ! ) so any @xmath60 in theorem [ scyfamily ] can be expressed as @xmath576 , where each @xmath577 , for @xmath578 , is either contained in the subsurface @xmath579 in @xmath57 bounded by @xmath517 and @xmath488 , or @xmath580 bounded by @xmath501 and @xmath489 .
clearly these have disjoint supports , so we can rearrange the indices so that @xmath581 , where each @xmath582 is compactly supported on @xmath89
. hence we see that @xmath58 is determined by a pair of mapping classes @xmath52 , _ but with no relation to each other whatsoever!_. [ compareholi ] as observed in the previous remark , for each @xmath60 in theorem [ scyfamily ] , we have factorization @xmath576 , where @xmath577 are disjoint from @xmath508 . so @xmath583 is obtained by a _ sequence _ of partial conjugations by @xmath584 , which amounts to performing fibered luttinger surgeries along lagrangian tori or klein bottles that are swept off by @xmath577 on the fibers over a loop on the base @xcite .
the reason we can break down the conjugation of @xmath516 by @xmath60 to a sequence of partial conjugations by dehn twists is that at each step the conjugated mapping class @xmath585 indeed commutes with @xmath586 .
this is easy to see once we recall that as a mapping class @xmath587 and that each @xmath577 is disjoint from @xmath149 and @xmath172 .
the mapping class @xmath588 certainly commutes with @xmath586 .
moreover , each @xmath577 is supported away from @xmath589 , so these surgeries can be viewed in the minimal models @xmath59 . since luttinger surgery do not change @xmath62 away from these lagrangians and since the canonical class @xmath590 can be supported away from them @xcite , @xmath591 or @xmath592 $ ] do not change under these surgeries . thus the symplectic kodaira dimension does not change under these surgeries .
this gives an alternative proof ( without using lemma [ ruledno ] ) of all @xmath58 being symplectic calabi - yau surfaces , since all are obtained from the symplectic calabi - yau @xmath26 we constructed in the previous section by partial conjugations . with this
said , we observe that if our family @xmath593 coincides with torus bundles over tori , we immediately get a proof of an improved version of a conjecture by chung - i ho and tian - jun li : that every symplectic torus bundle over a torus can be obtained from the @xmath5-torus via luttinger surgeries along tori @xcite , or klein bottles , as we add .
( this conjecture is easier to verify for _ lagrangian torus bundles _ over tori , which however only constitute a subfamily of torus bundles with @xmath594 . )
all our examples are clearly obtained from @xmath44 , our symplectic calabi - yau surface homeomorphic to the @xmath5-torus , via luttinger surgeries .
[ [ the - upper - bound - on - the - first - betti - number - of - a - genus - g - fibration ] ] the upper bound on the first betti number of a genus-@xmath7 fibration ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [ sec : bound ] in @xcite , korkmaz asked if it is true that @xmath67 for any non - trivial genus-@xmath7 lefschetz fibration @xmath26 , noting that this is the case for all the examples he knows of with a single counter - example provided by smith s genus-@xmath0 fibration on @xmath595 . here
we will sketch a generalizing of our construction of a genus-@xmath0 lefschetz pencil @xmath26 with @xmath4 in section [ sec:4torus ] to higher genera , which provides a negative answer to korkmaz s question , for all odd @xmath70 . our input will be the generalization of matsumoto s genus-@xmath54 fibration constructed by korkmaz himself @xcite ( and also by cadavid @xcite ) : for every @xmath597
, there is a genus-@xmath596 lefschetz fibration on the ruled surface @xmath598 .
as shown by hamada , the monodromy factorization of these can be lifted to @xmath599 to obtain : @xmath600 where the curves @xmath601 and the boundary parallel curves @xmath602 are as shown in figure [ generalizedmatsumoto ] .
embed this relation into @xmath603 , for @xmath604 , by gluing along the two boundary components @xmath157 and @xmath164 of @xmath605 .
we can also embed it by rotating the resulting surface by @xmath606 degrees as before .
the product of the two factorizations we get in @xmath607 through these two embeddings prescribes a fiber sum decomposable _ achiral _ genus-@xmath7 lefschetz fibration .
removing pairs of matching dehn twists with opposite signs ( coming from the two embeddings of @xmath608 and @xmath609 ) gives us a genus-@xmath7 lefschetz fibration @xmath610 with @xmath611 , which can be seen from a straightforward calculation as before .
moreover , similar to our refined construction in theorem [ scyfamily ] , we could the further lifts of the cadavid - korkmaz fibration in @xmath612 to obtain a pencil @xmath613 with @xmath5 base points , blowing - up which yields @xmath610 .
we have @xmath614 .
[ b1refinedquestion ] is the first betti number @xmath68 of any non - trivial genus-@xmath7 lefschetz pencil @xmath26 bounded above by @xmath71 ?
what is the supremum for @xmath68 of all genus-@xmath7 lefschetz pencils @xmath26 for a given @xmath7 ?
it has been known for a while by now that @xmath616 for any non - trivial genus-@xmath7 lefschetz
pencil / fibration @xmath26 @xcite .
the first part of question [ b1refinedquestion ] asks if @xmath71 can be a sharper bound , still linear in @xmath7 .
the second part of question [ b1refinedquestion ] , which is formulated in the most general way , can already be answered for low genera pencils : for @xmath617 this supremum is clearly @xmath282 , and it is @xmath54 for @xmath618 , which can be easily shown using the kodaira dimension ( invoking @xcite , as in remark [ smallest ] ) . for @xmath18
, we conjecture that the supremum is indeed @xmath619 .
note that , the analogous question for the supremum for the second betti number @xmath620 of _ relatively minimal _
genus-@xmath7 _ pencils _ @xmath26 is now fully answered .
it is @xmath621 for @xmath617 , realized by the pencil on @xmath622 ; @xmath623 for @xmath618 , following from smith s analysis of genus-@xmath54 pencils in @xcite ; whereas the supremum does _ not _ exist when @xmath289 , which is implied by the arbitrarily long positive factorizations of the boundary multi - twist produced in @xcite .
however , constructions of fixed genus pencils with large @xmath242 is rather antagonistic to that of fixed genus pencils with large @xmath6 : larger @xmath242 means more vanishing cycles ( typically ) killing @xmath6 .
d. auroux , _ mapping class group factorizations and symplectic @xmath5-manifolds : some open problems , _ problems on mapping class groups and related topics , 123132 , proc . sympos .
pure math .
74 , amer .
soc . , providence , ri , 2006 . f. t. farrell and l. e. jones , _ classical aspherical manifolds , _ cbms regional conference series in mathematics , 75 .
published for the conference board of the mathematical sciences , washington , dc ; by the american mathematical society , providence , ri ( 1990 ) .
y. matsumoto , _ lefschetz fibrations of genus two - a topological approach _ - , proceedings of the 37th taniguchi symposium on topology and teichmller spaces , ( s. kojima , et .
al . , eds . ) , world scientific , 1996 , 123148 .
r. stern , _ will we ever classify simply - connected smooth 4-manifolds ? _ , floer homology , gauge theory , and low - dimensional topology , 225239 , clay math .
proc . , 5 , amer .
soc . , providence , ri , 2006 . | we explicitly produce symplectic genus-@xmath0 lefschetz pencils ( with base points ) , whose total spaces are homeomorphic but not diffeomorphic to rational surfaces @xmath1 for @xmath2 .
we then give a new construction of an infinite family of symplectic calabi - yau surfaces with first betti number @xmath3 , along with a surface with @xmath4 homeomorphic to the @xmath5-torus .
these are presented as the total spaces of symplectic genus-@xmath0 lefschetz pencils we construct via new positive factorizations in the mapping class group of a genus-@xmath0 surface .
our techniques in addition allow us to answer in the negative a question of korkmaz regarding the upper bound on @xmath6 of a genus-@xmath7 fibration . |
since the discovery of quasi periodic oscillations ( qpos ) in the lightcurves of giant flares from soft gamma repeaters ( sgr ) ( israel et al . , 2005 ; strohmayer & watts , 2005 ; watts & strohmayer , 2006 ; barat et al . , 1983 ) there has been considerable interest in their physical origin .
one of the appealing explanations is that the qpos are driven by torsional oscillations of the neutron stars whose magnetic energy powers the flares ( duncan 1998 ) .
this opens a unique possibility to perform an asteroseismological analysis of neutron stars , and possibly obtain a new observational window to study the neutron - star interiors .
many authors have considered torsional modes to be confined to the magnetar crust , and have shown that seismological information about such modes would strongly constrain the physics of the crust ( piro 2005 , watts & strohmayer 2006 , watts & reddy 2007 , samuelsson & andersson 2007 , steiner & watts 2009 ) .
however , it was quickly understood that the theoretical analysis of magnetar oscillations is complicated by the presence of an ultra strong magnetic field ( @xmath1 @xmath2 ) that is frozen into the neutron star and penetrates both the crust and the core .
the field provides a channel for an intense hydro - magnetic interaction between the motion of the crust and the core , which becomes effective on the timescale of @xmath3 second ( levin 2006 ) .
since the qpos are observed for hundreds of seconds after the flare , it is clear that the coupled motion of the crust and the core must be considered . in recent years
, significant theoretical effort has gone into the study of this problem ( e.g. , glampedakis et al .
2006 , levin 2007 , gruzinov 2008b , lee 2008 ) .
this paper s analysis is based , in part , on an approach of levin ( 2007 , l07 ) . to make progress in computing the coupled crust - core motion ,
l07 has studied the time evolution of an axisymmetric toroidal displacement of a star with axisymmetric poloidal magnetic field . in that case
the alfven - type motions on different flux surfaces decouple from each other , a well - known fact from previous mhd studies ( for a review see goedbloed & poedts 2004 , hereafter gp ) .
one can then formulate the full dynamics of the system in terms of discrete modes of the crust which are coupled to a continuum of alfven modes in the core .
l07 demonstrated that ( i ) the global modes with frequencies inside the continuum are strongly damped via a phenomenon known in mhd as _ resonant absorption _ ( see gp ) , and ( ii ) in many cases , asymptotically the system tends to oscillate with the frequencies close to the continuum edges .
this result was later confirmed by gruzinov 2008b , who has used a powerful analytical technique to solve the l07 s normal - mode problem ( gruzinov noted that the resonant absorption is mathematically equivalent to landau damping ) .
oscillations near the continuum edge frequencies were also observed in a number of numerical general - relativistic mhd simulations of purely fluid stars ( sotani et al .
2008 , colaiuda et al .
2009 , cerda - duran et al .
2009 ) .
apart from finding qpos near the continuum edges , l07 s dynamical simulations identified transient qpos with drifting frequencies ; these were transiently amplified near the crustal frequencies .
no explanation for the origin of the drifts was given .
in this paper , we extend the previous analyses of the hydro - magnetic crust - core coupling in an essential way . in section 2 ,
we re - analyse l07 s toy model of a single oscillator coupled to a continuum , and we show that this system generically contains the _ edge normal modes _ with frequencies near the continuum edges .
we show that these modes dominate the late - time dynamics of the system , and develop a formalism which allows one to predict analytically the edge mode s amplitude from the initial data .
we then explore the effect of viscosity on the system ( introduced as a friction between the neighbouring continuum oscillators ) , and show that the edge mode is longer lived than all other motions of the system .
we also provide a non - trivial analytical formula for the time dependence of the overall energy dissipation . in section 3 ,
we describe how transient qpos , not associated with the normal modes of the system , are obtained when parts of the core spectrum consist of densely and regularly spaced discrete modes ( and in section 5 we show that such an array of discrete modes is expected when the magnetic field in the core is not perfectly axisymmetric but has some degree of tangling ) . as a by - product of our analysis , we explain the origin of the qpo frequency drifts seen in l07 simulations .
we provide simple analytical fits to the drifts , and show that when the regularity of the continuum sampling is removed ( e.g , when the frequencies are sampled as random numbers picked from the continuum range ) , the drifts disappear . in section 4 , we set up models with a more realistic hydro - magnetic structure of the neutron - star core .
we show how to find the continuum modes and their coupling to the crust for an arbitrary axisymmetric poloidal field , with an arbitrary density profile on the core ( l07s calculations , for simplicity and concreteness , were restricted to constant - density core and homogeneous magnetic field ) .
we treat a more complicated case of a mixed axisymmetric toroidal - poloidal field , with radial stratification , in the appendix b. we demonstrate that for realistic field configurations , the alfven continuum of modes coupled to the crust may show a number of gaps .
if a crustal mode frequency belongs to one of these gaps , a strong global discrete mode arises which dominates the late - time dynamics and whose frequency also belongs to the gap .
the frequency of the gap global mode does not generally coincide with , but is often close to that of the crust .
we suggest that it was these gap modes that appeared in lee s ( 2008 ) calculations as well - defined discrete global modes .
so far , only axisymmetric magnetic fields have been considered in the magnetar - qpo literature , with the alfven continuum modes occupying the flux surfaces of the field . in section 5
we argue that if the field is not axisymmetric but instead is highly tangled , then the alfven continuum modes become localized within small regions of individual field lines , and therefore become dynamically unimportant .
instead , a set of discrete alfven modes appears , with the spacing between the modes strongly dependent on the degree of field tangling .
we devise a phenomenological prescription which allows us to parametrize the field tangling for computing the dynamically important modes , and introduce an easily solvable `` square box '' model suitable for exploring the parameter range . finally ,
in section 6 , we use the suite of models built in the previous sections to explore their connection to the qpo phenomenology .
we find that(a ) within the standard magnetar model , it is possible to produce strong long - lived or transient qpos with frequencies in the range of around @xmath4@xmath5hz , but only if the neutrons are decoupled from the alfven - like motion of the core ; this implies that at least one of the baryonic components of the core is a quantum fluid.(b ) our models could not produce the high - frequency @xmath0hz qpo within the standard paradigm of a magnetar core composition .
in this section , we study the motion of a harmonic oscillator ( which we hereafter call the large oscillator ) which is coupled to a continuum of modes .. this model was introduced in l07 and it provides a qualitative insight into the behaviour of crustal modes ( represented by the large oscillator ) coupled to a continuum of alfven modes in the core of a magnetar .
l07 found that if the large oscillator s proper frequency was within the range of the continuum frequencies , then the late - time behaviour of the system was dominated by oscillatory motion near the edges of the continuum interval . here
, we give an explanation of this phenomenon in terms of the _ edge modes_. our analysis allows us to use initial data and predict the displacement amplitudes and frequencies of the system at late times .
the model consists of the large mechanical oscillator with mass @xmath6 and proper frequency @xmath7 , representing a crustal elastic shear mode .
attached to the large oscillator is a set of @xmath8 smaller oscillators of mass @xmath9 and proper frequency @xmath10 constituting a quasi - continuum of frequencies @xmath10 ( where @xmath11 ) .
the continuum is achieved when @xmath12 while the total small - oscillator mass @xmath13 remains finite .
the convenient pictorial representation is through suspended pendulae , as shown in fig .
[ fig1 ] ( see also fig . 2 of l07 ) . of small pendulae , representing the ( quasi- ) continuum , are coupled to one large pendulum , representing the crust.,width=288 ] the equations of motion are obtained as follows .
each small oscillator is driven by the the motion of the large oscillator : @xmath14 where @xmath15 is the displacement of the @xmath16 small oscillator in the frame of reference of the large oscillator , @xmath17 is the displacement of the large oscillator in the inertial frame of reference , and the right - hand side represents the non - inertial force acting on the small oscillator due to the acceleration of the large one .
the large oscillator experiences the combined pull of the small ones : + @xmath18 here @xmath19 is the frequency of the big pendulum corrected for the mass loading by the small pendulae , i.e. @xmath20 . in this subsection
we explore the behavior of this system by direct numerical simulations .
we found this to be helpful in the building of our intuition .
we defer a semi - analytical normal - mode analysis to the next subsection .
we follow l07 and for concreteness concentrate on a specific example ; it will be clear that the conclusions we reach are general .
we choose @xmath21rad / second and mass @xmath22 .
we choose a total number of 1000 small pendulae with frequencies @xmath23rad / second and masses @xmath24 , to mimic the continuum frequency range between @xmath25rad / second and @xmath26rad / second .
the simulation is initiated by displacing the large oscillator while keeping the small pendulae relaxed ( this mimics the stresses in the crust ) , and then releasing .
the subsequent motion of the system is then followed numerically by using a second order leapfrog integration scheme which conserves the energy with high precision .
the resulting motion of the large pendulum can be decomposed into three stages ( see fig .
[ fig2a ] and fig .
[ fig2b ] ) : .
the blue horizontal lines denote the theoretically predicted amplitude of the dominating upper edge - mode ( see section 2.3).,width=336 ] \(1 ) during the first 50 - 60 seconds , there is a rapid exponential decay of the large oscillator s motion , during which most of the energy is transferred to the multitude ( i.e. , the continuum ) of small oscillators .
this is the so - called phenomenon of `` resonant absorption '' , which has been studied for decades in the mhd and plasma physics community ( e.g. , ionson 1978 , hollweg 1987 , goedbloed & poedts 2004 , l07 , gruzinov 2008b ) . in this first stage ,
the amplitude of the big pendulum motions drops by a factor of @xmath27 .
\(2 ) after @xmath28 seconds , the exponential decay stops abruptly as the large oscillator now reacts to the collective pull of the small ones .
this second stage is characterized by a slow algebraic decay of the amplitude of the big pendulum displacement .
gruzinov ( 2008b ) explains this as being due to the branch cut in the oscillator s response function .
\(3 ) the motion of the large oscillator stabilizes at a constant level ( l07 missed this stage in his simulations , which he stopped too early ) .
fourier transform reveals 2 qpos at the frequencies close to the continuum edges , @xmath29 and @xmath30 ; the same qpo frequencies can be observed in the previous stage ( 2 ) as well .
what is the origin of the qpos , and how is this eventual stability established ? in fig .
[ fig5a ] and [ fig5aa ] , we show how the amplitude of the small oscillators evolves with time .
.,width=336 ] . at later times energy
is transferred to the oscillators near the edge of the continuum.,width=336 ] after the initial resonant absorption phase , the amplitude is distributed as a lorentzian centered on the frequency around @xmath31 ; this is because the small oscillators in resonance with the large one are the ones which gain the most energy .
however , in subsequent times we see that the energy exchange occurs between the small oscillators , and that the net result of this exchange is the energy flow towards the oscillators whose frequencies are near the edges . by the time the third stage begins , the amplitudes of the oscillators near the edge stabilize and their phases become locked .
they are pulling and pushing the large oscillator in unison . in the next subsection , we show that this behavior is due to the presence of the _ edge normal modes _ , and we shall derive their frequencies and amplitudes . in this section
we deal with the system of coupled harmonic oscillators , and one should be able to find its normal modes using the standard techniques ( landau and lifshitz mechanics , 23 ) . however , the fact that all small oscillators are attached to the large one , and there is no direct coupling between the small oscillators , allows us a significant shortcut ( in appendix a , we treat a more general problem of _ several _ large oscillators coupled to a multitude of the core modes ) .
we proceed as follows : suppose that we impose on the large oscillator a periodic motion with angular frequency @xmath32 , by driving it externally with the force @xmath33 .
this motion in turn drives the small oscillators according to eq .
( [ vier ] ) : @xmath34 which has the steady state solution : @xmath35 where we have omitted the time dependent factor @xmath36 on both sides .
the combined force @xmath37 of the small oscillators acting back on the large one ( see eq .
( [ vijf ] ) ) is given by @xmath38 according to newton s second law , @xmath39 if @xmath32 corresponds to the normal - mode frequency , then @xmath40 .
hence by substituting eq .
( [ no8 ] ) into eq .
( [ secondlaw ] ) we get the following eigenvalue equation for @xmath32 : @xmath41 in the continuum limit @xmath12 , the above equation becomes @xmath42 where @xmath43 is the mass per unit frequency of the continuum modes . in the discrete case , the solutions of eq .
( [ discretemodes ] ) are @xmath44 frequencies @xmath45 that are within the quasi - continuum ( @xmath46 , for @xmath47 ; quasi - continuum modes ) , and @xmath48 modes with frequencies @xmath49 and @xmath50 ( the edge - modes ) that are near the edges , but outside , of the continuum ( i.e. @xmath51 and @xmath52 ) .
it can be shown from eq .
( [ discretemodes ] ) that in the limit @xmath53 and @xmath54 , the contribution of the small oscillator to the @xmath55-th quasi - continuum mode is completely dominated by the pendulae that are in close resonance with the mode . more precisely , one can show that as the number of oscillators @xmath8 increases and @xmath9 decreases , the number of small oscillators contributing to the mode energy remains constant .
however , for the two edge modes there is no such singular behavior in the limit of large @xmath8 , and consequently they play a special role in the dynamics of the system .
this last point is clearly seen in the continuum case represented by eq .
( [ contmodes ] ) .
the eigenvalue equation has no real solutions in the range of small - oscillator continuum @xmath56 , since the response function @xmath57 is ill - defined in this interval is performed along the contour chosen according to the landau rule .
one then obtains a `` resonantly absorbed '' or `` landau - damped '' mode ( gruzinov 2008b , l07 ) , which exactly represents the exponential decay of stage ( 1 ) in our numerical experiment of the previous subsection . ] .
however , the edge modes on both sides of the continuum interval remain , and their frequencies can be found by numerically evaluating the zero points of @xmath57 in eq .
( [ contmodes ] ) . for the numerical of the previous subsection
, one finds @xmath58 and @xmath59 .
analytically , one can find the following scaling for the distance @xmath60 between the mode frequency and the nearest edge @xmath61 of the continuum range : @xmath62 where @xmath63 is a constant of order unity .
the larger is the density of continuum modes at the edge @xmath64 , the further is the edge mode pushed away from the continuum range .
it is particularly interesting to consider the case when the continuum interval is limited by a turning point ( l07 ) with the divergent density of states near the edge , @xmath65 , where @xmath66 is a constant . in this case
the distance from the edge - mode frequency to the nearest continuum edge is given by @xmath67 the quadratic dependence in eq .
( [ distance1 ] ) vs. the exponential dependence in eq .
( [ distance ] ) implies that the continua with turning points typically feature much more pronounced edge modes and stronger qpos than the ones with linear edges . in the next section ,
we show how to calculate the edge - mode amplitudes and qpo strengths from the initial data . in the continuum limit ,
the only modes with real oscillatory frequency are the edge modes .
thus , as we demonstrate explicitly below , they dominate the late - time dynamics of the system when the number @xmath8 of small oscillators becomes large .
our analysis proceeds as follows : lets define a new set of variables , expressed as a vector @xmath68 with components @xmath69 and @xmath70 for @xmath71 . with these new variables ,
the kinetic energy of the system is a trivial quadratic expression @xmath72 where the inner product of two vectors @xmath73 and @xmath74 is defined as @xmath75 .
the potential energy is a positive - definite quadratic form , whose exact form is unimportant here . the mutually orthogonal eigenmodes @xmath76 can be found via a procedure outlined in the previous section .
their eigenfreguencies @xmath45 are identified by finding zeros of @xmath57 in eq .
( [ discretemodes ] ) , and the corresponding eigenvector components are given by @xmath77 , lets assume that we initiate our simulation by displacing the large oscillator by an amount @xmath78 while keeping the small oscillators relaxed @xmath79 and all initial velocities at zero .
in the new variables , the initial state of the system is given by the vector @xmath80 , where @xmath81 and @xmath82 .
the time evolution of the system is given by : @xmath83 substituting the initial conditions , and the expression in eq . ( [ eigen1 ] ) for the eigenvector components ,
we get @xmath84 the coordinate of the large oscillator is simply given by @xmath85 . for the continuum of small modes , the above expansion breaks down , since the eigenvalue equation has no real solutions inside the continuum range
. however , the edge modes are well defined , and they determine the dynamics at late times .
therefore , for the continuum case we can still write down the analogous expression which is valid only at late times : @xmath86 the sums of eq .
( [ no10 ] ) are replaced with the corresponding integrals , and we have the following expression for the displacement of the large oscillator at late times : @xmath87 this expression is in excellent agreement with the numerical simulations . in the numerical example of subsection @xmath88 ,
the upper edge mode dominates the late - time behavior of the system , and its calculated contribution is plotted in fig .
[ fig2b ] , together with the numerically simulated motion .
we now add an extra degree of realism by introducing viscous friction into the system . in mhd
, continuum modes are spatially localized , and the effect of viscosity is to frictionally couple the neighboring modes ( see , e.g. , hollweg 1987 ) . in our simple model we introduce viscosity by adding frictional forces between the neighboring oscillators , @xmath89 where @xmath90 is the force from the @xmath91th oscillator acting on the @xmath92th .
we now calculate how the system dissipates energy as a function of time
. we will show that it occurs in two stages ( see fig .
[ fig4 ] ) : ( 1 ) initially , the small oscillators are strongly and simultaneously excited by the `` landau - damped '' large oscillator , then they become dephased , with the average relative motion between the neighboring oscillators growing linearly in time .
this leads to a very rapid dissipation of the bulk of the initial energy .
( 2 ) the edge modes persist , since the participating small oscillators move in phase and the energy dissipation is small .
the energy of the modes is damped exponentially on a timescale much longer than that of the first stage . ) , width=336 ] the dissipated energy is given by @xmath93 in the continuum limit , the small oscillators are labeled not by a discrete index @xmath91 , but by a continuous variable @xmath94 . the expression for the dissipated energy is then @xmath95 where @xmath96 is the viscous coefficient .
after the initial exponential damping of the large oscillator and the excitation of the small oscillators , the latter initially move independently , with @xmath97 , \label{xlambda}\ ] ] where @xmath98 is the amplitude of the @xmath94th oscillator . from the above equation ,
we then obtain @xmath99 ^ 2+\omega_\lambda^2 \tilde{x}_\lambda^2(d\omega_\lambda / d\lambda)^2t^2\right ] , \label{dlambda}\ ] ] where the @xmath100 stands for time - averaging over many oscillation periods . for times
@xmath101 the second term on the right - hand side of eq .
( [ dlambda ] ) dominates . for a simple model with @xmath102 and @xmath103 , @xmath104 where @xmath105 is the total energy of the system and @xmath106
the analytical solution for the energy and dissipated power , @xmath107 agrees very well with numerical simulations , see fig .
while the equations above were derived for restrictive assumptions ( @xmath102 and @xmath103 ) , we found that the analytical formulae in eq .
( [ ediss ] ) provide a good fit for a large variety of simulations .
this is because it is the small oscillators with the frequencies near that of the large oscillator which carry most of the energy , and in that narrow band our approximations hold .
after the energy dissipation due to dephasing is over , only the edge modes remain .
this is illustrated in fig .
[ fig5b ] , where we show how the energies of the small oscillators evolve with time . at late times , only the oscillators taking part in the edge modes move substantially ; this is because they remain in phase and do not dissipate much . at this stage
the energy is drained by slow exponential decay of the edge modes . )
, this figure shows the amplitudes of the small oscillators at different times @xmath108 .
the energy of most oscillators is drained due to viscous dissipation . at late times , only the oscillators near the edges of the continuum have substantial energy.,width=336 ]
finite - size mhd systems feature a mix of continuum and discrete modes ( see poedts et al . , 1985 and gp ) .
for axisymmetric field configurations the continuum modes occupy the whole flux surfaces and play an important role in the oscillatory dynamics ; this was the motivation for l07 and our study of the previous section .
we will argue in section 5 that if the core field is highly tangled , the continuum modes become localized in space and discrete core modes will play a more important role .
thus it is important to study the case when the the crustal modes are coupled to a set of discrete core modes . in this section we show that if the frequencies of the discrete modes are regularly spaced in some frequency intervals , then the system displays _ transient qpos _ that are entirely missed by its normal - mode analysis .
this is interesting from the observational point of view , since many of the observed magnetar qpo features are transient .
suppose that a set of discrete modes are located in the interval @xmath109 around frequency @xmath7 and are separated by a regular frequency interval @xmath110 , and assume the following hierarchy : @xmath111 after the modes are excited , they are initially in phase but will de - phase rapidly on the timescale @xmath112 . however , at times @xmath113 the modes come into phase again and pull coherently on the large oscillator .
therefore , a transient qpo feature should appear around these times at a frequency close to @xmath7 . in fig .
[ figx ] and fig .
[ figy ] we show the dynamical spectrum from a simulation where the conditions for transient qpos were engineered for two frequencies .
the transient qpos agree well with the expectations .
as is seen from the figures , the strongest transient qpos are those whose frequencies are the closest to that of the large oscillator ; this is because the response of the large oscillator is the strongest around its proper frequency . and
@xmath114 ( the colored scale denotes @xmath115 ) .
the green horizontal line denotes the frequency of the large oscillator ( @xmath116 ) .
, width=336 ] .
we have shifted the frequency of the large oscillator , denoted by the green horizontal line , to @xmath117 . by comparison wit fig .
[ figx ] it is clear that the drifting qpo s at @xmath114 are now much stronger as they are closer to the large oscillator frequency . note that the edge mode at @xmath118 is clearly visible.,width=336 ] one can now easily understand the frequency drifts in fig .
10 of l07 ( fig .
[ figw ] in this paper ) . in the simulations of that paper ,
the core continuum was sampled with a set of densely and regularly - placed alfven modes by slicing the field into finite - width flux shells .
the spacing @xmath110 between the modes was not constant but a function of the alfven frequency @xmath119 .
in that case , the qpo drifts with the qpo frequency @xmath120 given by the inverse relation @xmath121 with this relation we are able to fit all of l07 drifting qpos , as shown in fig .
[ figw ] and [ figz ] .
note that multiple qpos correspond to different branches of the alfven continuum .
as expected , the drifting qpos amplified near the crustal frequencies , since there the response of the crust to the core modes pull is the strongest .
.,width=336 ] ) to fit the drifting qpo s from figure [ figw ] .
the red curves are @xmath122 drifts , green curves are @xmath123 and blue curves are @xmath124 .
the higher frequency drifts originate from alfven overtones.,width=336 ]
in this section we extend the constant magnetic field and constant - density magnetar model from l07 to include more realistic pressure and density profiles and more general ( but still axisymmetric ) magnetic field configurations .
our aim is to use this model to : 1 ) calculate numerically alfven eigenmodes and their eigenfrequencies on different flux surfaces inside the star , in order to determine the continuous spectrum of the fluid core , and 2 ) use these modes to simulate the dynamics of a realistic magnetar . in order to calculate the alfven eigenmodes and eigenfrequencies for a realistic magnetar model
, we employ the linearized equations of motion for an axisymmetric magnetized , self - gravitating plasma .
the general equations , which are derived in detail in poedts et al .
( 1985 , hereafter p85 ) and given in their equations ( 53 ) and ( 54 ) , constitute a fourth order system of coupled ordinary differential equations in the case of a mixed poloidal and toroidal magnetic field .
the formalism of p85 is briefly summarized in the appendix b. in the case of a purely poloidal magnetic field , the system simplifies to two uncoupled second order differential equations ( p85 , equations ( 70 ) and ( 71 ) ) .
+ we assume our star is non - rotating and neglect its deformation due to the magnetic pressure , which is expected to be small .
therefore , we consider a spherically symmetric background model that is a solution of the tolman - oppenheimer - volkoff equation ( tov equation ) .
the hydrostatic equilibrium is calculated using a sly equation of state ( douchin & haensel 2001 ; haensel & potekhin , 2004 ; haensel , potekhin & yakovlev 2007 ) , which can be found in tabulated form on the website _
http://www.ioffe.ru / astro / nsg / nseos/_. the integration of the tov equation is performed using a 4th order runge - kutta scheme , integrating from the center of the star outward until we reach a mass density @xmath125 @xmath126 , which is consistent with the crust - core interface in the equation of state from douchin & haensel ( 2001 ) .
the resulting model has a central mass density @xmath127 @xmath126 , a total mass of @xmath128 @xmath129 and a radius of @xmath130 @xmath131 . to this spherical model
we add a poloidal magnetic field , which we generate by placing a circular current loop of radius @xmath132 and current @xmath133 around the center of the star .
the field is singular near the current loop , however all the field lines which connect to the crust ( and thus are physically related to observable oscillations ) carry finite field values .
this particular field configuration is chosen as an example ; there is an infinite number of ways to generate poloidal field configurations . in the appendix b
we will add to this field a toroidal component and calculate the corresponding alfven continuum of the core . in order to find the equations of motion for the magnetized material in the neutron star core
, we would need to add self - gravity to the ideal magnetohydrodynamic equations .
this problem has been solved by p85 in a tour the force mathematical approach . in that paper
the authors assume a self - gravitating axisymmetric equilibrium with a field geometry consisting of mixed poloidal and toroidal field components , and they derive linearized equations of motion .
for this field geometry it is convenient to work with so - called flux - coordinates @xmath134 .
the basic concept behind this curvilinear coordinate system is the magnetic flux - surface , which is defined as the surface perpendicular to the lorentz force @xmath135 . from this definition
it is clear that the magnetic field lines lie in flux surfaces . if one considers a closed loop on a flux surface which makes one revolution around the axis of symmetry , then the magnetic flux @xmath136 through the loop depends on the flux surface only and is the same for all of the loops .
therefore @xmath136 is chosen as the coordinate labeling the flux surfaces . in each flux - surface
we can denote a position by its azimuthal angle @xmath137 and its poloidal coordinate @xmath138 , which is defined as the length along @xmath139 line .
in p85 , it is shown that the equations of motion allow for a class of oscillatory solutions that are located on singular flux surfaces , constituting a continuum of eigenmodes and eigenfrequencies . in the case of a purely poloidal field ( @xmath140 ) ,
the continuum solutions are degenerate and polarized either parallel ( @xmath141 ) or perpendicular ( @xmath142 ) to the magnetic field lines .
in the latter case the displacement is @xmath137-independent .
it is clear that in contrast to the @xmath138-polarized modes , the @xmath137-polarized modes are purely horizontal and are therefore unaffected by gravity .
this latter case is considered here .
the equation of motion is in this case simply the alfven wave equation : @xmath143 , \label{poedts}\ ] ] where the operator @xmath144 is given by @xmath145= \frac{b}{4 \pi x \rho } \frac{\partial}{\partial \chi } \left [ x^2 b \frac{\partial}{\partial \chi } \left ( \frac{\xi_{\phi}(\psi , \chi)}{x } \right ) \right ] .
\label{poedts1}\ ] ] here @xmath146 is the distance to the magnetic axis of symmetry .
although in the presence of a mixed poloidal and toroidal field the equations still give rise to a continuous set of solutions , the calculations are significantly complicated as the continuum modes are affected by the toroidal component of the field , by gravity , and by compressibility . for the sake of simplicity we will ignore toroidal fields in our dynamic simulations .
we will however , calculate the continuum frequencies for a mixed poloidal and toroidal field in the appendix b. for determining the spectrum of the core continuum , the appropriate boundary conditions are @xmath147 , where @xmath148 marks the location of the crust - core interface . with this boundary condition , equation ( [ poedts ] )
constitutes a sturm - liouville problem on each separate flux surface @xmath136 . using the stellar structure model and magnetic field configuration from section 4.1
, we can calculate the eigenfunctions and eigenfrequencies for each flux surface @xmath136 .
the reflection symmetry of the stellar model and the magnetic field with respect to the equatorial plane assures that the eigenfunctions of eq .
( [ poedts ] ) are either symmetric or anti - symmetric with respect to the equatorial plane .
we can therefore determine the eigenfunctions by integrating eq .
( [ poedts ] ) along the magnetic field lines from the equatorial plane @xmath149 to the crust - core interface @xmath150 .
let us consider the odd modes here for which @xmath151 , and solve eq .
( [ poedts ] ) with the boundary condition @xmath152 at the crust - core interface ; for even modes , the boundary condition is @xmath153 .
we find the eigenfunctions by means of a shooting method ; using fourth order runge - kutta integration we integrate from @xmath154 to @xmath155 .
the correct eigenvalues @xmath156 and eigenfunctions @xmath157 are found by changing the value of @xmath158 until the boundary condition at @xmath159 is satisfied . in this way we gradually increase the value of @xmath158 until the desired number of harmonics is obtained . in figure [ core_cont1 ]
we show a typical resulting core - continuum . as a function of the angle @xmath160 , the polar angle at which the flux - surface @xmath136 intersects the crust .
since we are only considering odd crustal modes , the only alfven modes that couple to the motion of the star are the ones with an odd harmonic number @xmath91 .
this particular continuum was calculated using a poloidal field with an average surface value @xmath161 , generated by a circular ring current of radius @xmath162.,width=336 ] , gaps in the continuum become visible around @xmath163 and @xmath164.,width=336 ] according to sturm - liouville theory the normalized eigenfunctions @xmath159 of eq .
( [ poedts ] ) form an orthonormal basis with respect to the following inner product : @xmath165 where @xmath166 is the kronecker delta and @xmath167 is the weight function .
we have checked that the solutions we find satisfy the orthogonality relations .
we are now ready to compute the coupled crust - core motion . here
we follow l07 and assume that the crust is an infinitely thin elastic shell .
we label the lattitudinal location by the flux surface @xmath136 intersecting the crust , and consider the crustal axisymmetric displacements @xmath168 . in the mhd approximation
, the magnetic stresses enforce a no - slip boundary condition at the crust - core interface , such that @xmath169 instead of @xmath170 .
it is useful to make the following substitution @xmath171 where we choose the function @xmath172 so that ( a ) it corresponds to the static displacement in the core and hence satisfies @xmath173 , and ( 2 ) @xmath174 .
therefore the new quantity satisfies the boundary condition @xmath175 and can be expanded into the alfven normal modes @xmath159 which satisfy the same boundary conditions .
we now proceed by substituting eq .
( [ substit0 ] ) into eq .
( [ poedts ] ) thus obtaining a simple equation of motion for @xmath176 @xmath177 from the definition of the operator @xmath144 it follows that for the odd modes @xmath178 here the constant @xmath179 is chosen such that @xmath180 , in order that @xmath181 on both boundaries .
we expand @xmath176 and @xmath182 into a series of @xmath159 s : @xmath183 eq .
( [ substit1 ] ) reduces to the following equations of motion for the eigenmode amplitudes @xmath184 @xmath185 these equations show how the core alfven modes are driven by the motion of the crust . to close the system
, we must address the motion of the crust driven by the hydromagnetic pull from the core .
the equation of motion for the crust is given by @xmath186 where the acceleration due to elastic stresses @xmath187 is @xmath188 \label{eqb}\end{aligned}\ ] ] where @xmath189 is the polar angle ( cf .
the acceleration @xmath190 due to the magnetic stresses between the crust and the core can be expressed as @xmath191 where @xmath146 is the distance to the axis of the star , @xmath192 is column mass - density of the crust and @xmath193 is the angle between the magnetic field line and the normal vector of the crust .
it is convenient to express the crustal displacement @xmath194 as a fourier series , being a sum normal modes of the free - crust problem .
using eq .
( [ eqb ] ) is straightforward to show analytically that the eigenfunctions @xmath195 of the free - crust problem ( eq .
( [ eqa ] ) with @xmath196 ) are @xmath197 with eigenfrequencies @xmath198 here @xmath199 is the @xmath200 spherical harmonic of degree @xmath201 .
the normalized functions @xmath195 form an orthonormal basis , so that @xmath202 where @xmath203 is again the kronecker delta . the crustal displacement can then be expressed in terms of @xmath195 @xmath204 substituting eq .
( [ eqc ] ) into eq .
( [ eqa ] ) we obtain the equations of motion for the crustal mode amplitudes @xmath205 @xmath206 we can express @xmath190 as @xmath207_{\chi = \chi_c } \label{substit6}\end{aligned}\ ] ] up to this point the derived equations of motion for the crust and the fluid core are exact .
we are now ready to discretize the continuum by converting the integral of eq .
( [ substit5 ] ) into a sum over @xmath8 points @xmath208 . in order to avoid the effect of phase coherence
( see section 3 ) which caused drifts in the results from l07 , we sample the continuum randomly over the @xmath189-interval @xmath209 $ ] . in the following
, functional dependence of the coordinate @xmath136 or @xmath210 is substituted by the discrete index @xmath55 which denotes the @xmath55-th flux surface .
@xmath211_{\chi = \chi_c } \label{substit7}\end{aligned}\ ] ] @xmath212 these are the equations that fully describe dynamics of our magnetar model . as with the toy model from section 2 we integrate them using a second order leap - frog scheme which conserves the total energy to high precision . as a test we keep track of the total energy of the system during the simulations .
further we have checked our results by integrating equations ( [ substit7 ] ) and ( [ subs3 ] ) with the fourth - order runge - kutta scheme and found good agreement with leap - frog integration .
based on our section-2 results , we have a good idea of what type of dynamical behavior should occur in our more realistic magnetar model .
first , we expect that crustal modes with frequencies inside the alfven continuum will be damped quickly by resonant absorption ( `` landau - damping '' in the terminology of gruzinov 2008b ) .
second , as with our previous model we expect the late time behavior of the system to show qpo s near the edges of the continuum , or edge modes .
the third important feature of our model is that the continuum may possibly contain gaps , as is shown in fig .
[ core_cont2 ] . in this case
there is the possibility that crustal frequencies fall inside the gaps and remain undamped . in all of our simulations
these expectations have come true .
we will now show the results from a simulation which illustrate the above mentioned effects .
the basic freedom of choice that we have for our model is the strength and geometry of the equilibrium magnetic field .
we choose here a purely poloidal magnetic field with an average strength at the surface of @xmath213 , induced by a circular current loop of radius @xmath214 .
this field gives us a gap in the continuum at frequencies @xmath215 .
we consider the lowest degree odd crustal modes with frequencies @xmath216 and @xmath217 , which we couple to 5000 continuum oscillators ( the alfven continuum ) .
we sample the continuum at 1000 randomly chosen flux surfaces , and at each flux surface we consider 5 alfven overtones . like with our toy model from section 2
, we initiate the simulation by displacing the crust ( @xmath218 ) while keeping the continuum oscillators ( the alfven modes ) relaxed ( @xmath219 ) . in figures
[ fig9 ] and [ fig10 ] we show the resulting power spectrum for 2 different models . in the first one ,
the crustal frequencies are located inside the core continuum range , and the peaks due to the edge modes appear .
by contrast , in the second case one of the crustal frequencies belongs to the gap , and a peak representing the global gap mode strongly stands our above the background .
we note that the gap - mode s frequency lies close to but does not coincide with the crustal - mode frequency ; we found this to be a generic feature of our models , with the difference of @xmath220% for the typical model parameters .
the gap modes are particularly interesting because they have relatively large amplitudes , and are not as strongly damped by viscosity as the edge modes .
it must be emphasized that for all persistent modes in the system , the position in the frequency space of the core alfven continuum plays the key role in setting the global - mode frequency and in determining its longevity .
we note that lee ( 2008 ) has used a different method to identify discrete modes in a magnetar with similar magnetic configuration to ours .
these modes were not associated with crustal frequencies , and we strongly suspect that they were located in the gaps of the continuum spectrum and could be identified with the edge or gap modes presented in this work .
our preceding discussion of the continuum was predicated on the foliation of the axisymmetric magnetic field into the flux surfaces , with each of the singular continuum mode localized on the flux surfaces .
these modes are `` large''-they are coherent over the spacial extent comparable to the size of the system , and thus they play an important role in the overall dynamics - they are responsible for the resonant absorption of the crust oscillations , and contribute to generating the edge and gap modes . but what happens if the field can not be foliated into the flux surfaces , but is instead tangled in a complicated way ?
one can argue that the continuum part of the spectrum still persists , as follows : consider an arbitrary field line anchored at the crust - core interface at both ends , and choose a tube of field lines of infinitesimal radius which is centered on the original field line ( see fig .
[ fig7 ] ) .
it is clear that a twisting alfven mode exists in the tube : it is obtained by the circular rotation of the fluid around the central field line , propagating along the central field line with the local alfven velocity . since there is a continuum of the field - line lengths ,
there is also a continuum of alfven modes .
however , the modes we constructed are highly localized in space and and have a small leverage in the overall dynamics .
we conjecture that the more tangled the fields are , the less role do the singular continuum modes play in the overall dynamics . whilst we can not rigorously prove this conjecture , we can motivate it as follows : consider an area element @xmath221 of random orientation with the normal @xmath222 inside the star , and consider a shearing motion along the element .
this shearing motion will be resisted by the @xmath223 component of the magnetic field , with the effective shear modulus of order @xmath224 where @xmath225 stands for averaging over the area element . for ordered field
, it is possible to choose the orientation of the area element so that @xmath226 ; the presence of such an orientation makes a fundamental difference between mhd and elasticity theory and is responsible for the presence of continuous spectrum in mhd .
however , if the linear size of the @xmath221 is greater than the characteristic length on which the field is tangled , then @xmath227 is non - zero for all orientations of @xmath222
. therefore , for highly - tangled fields there can be no large - scale singular continuum modes , and their existence is restricted to the small scales .
hence our assertion that for strongly tangled fields continuum modes play a secondary dynamical role .
one is then faced with the problem when crustal modes are coupled to a set of discrete core alfven modes . in appendix
a we show how to find the eigen - solution of such a problem , provided that all of the coupling coefficients are known .
how does one quantify the degree to which the fields are tangled ?
some insight comes from the numerical simulations of braithwaite and colleagues , who have studied what type of fossil fields remain in a stratified star after an initial period of fast relaxation .
consider a stable fossil field field configuration , such as the one obtained in the simulations of braithwaite and spruit ( 2004 ) and braithwaite and nordlund ( 2006 ) [ see also gruzinov ( 2008a ) for analytical considerations ] .
there , the final field is nearly , but not perfectly axisymmetric and has a small - scale random component . for a less - centrally concentrated initial field , braithwaite ( 2008 )
shows that the final fossil field is in general non - axisymmetric and can have a complicated topology . as a starting point
, we shall consider the nearly axisymmetric field with a small random component .
the latter acts like a small extra shear modulus @xmath227 and dynamically couples the flux surfaces of the axisymmetric component .
we then quantify the degree of tangling by the relative value of @xmath227 and @xmath228 .
to study this idea further , we specify a very simple model of a neutron star , motivated by the one considered in levin ( 2006 , hereafter l06 ) see fig .
[ boxje ] that never - the - less captures the essential physics .
-direction and the displacements of the plates ( crust ) are in the @xmath229-direction .
the magnetic field is directed along @xmath230-axis and its strength varies as a function of @xmath146.,width=288 ] consider a perfectly conducting homogeneous fluid of density @xmath231 contained in a box with width @xmath232 , length @xmath233 and depth @xmath234 . the magnetic field in this box
is everywhere aligned with the @xmath230-axis and its strength is a function of @xmath146 only .
we assume that gravity is zero and consider a lagrangian displacement @xmath235 of the fluid along the @xmath229-direction ; we specify periodic boundary conditions in this direction ( one should think of the @xmath229 direction as azimuthal ) .
we now add to this model a small effective shear modulus @xmath227 due to the field tangling .
the fluid equation of motion is : @xmath236 here @xmath237 is the alfven velocity and @xmath238 is the @xmath227-generated shear velocity . if we assume a small shear speed , i.e. @xmath239 , eq .
( [ waveeq ] ) reduces to + @xmath240 we now find the core alfven eigenmodes . after adapting the no - slip boundary conditions + @xmath241
the problem can be easily solved by separation of variables @xmath242 \right\ } x \left ( x \right)$ ] , where @xmath243 . plugging this in eq .
( [ waveeqred ] ) we find for the the @xmath244dependent part of the solution : + @xmath245x . \label{solution}\end{aligned}\ ] ] here @xmath246 can be interpreted as the frequency of the @xmath247-th alfven overtone at @xmath146 . from the above expression it is clear that in the limit of very small @xmath248 , the solution for @xmath249 must be close to zero everywhere except in a very small neighborhood of @xmath250 .
it is in this limit that the solutions are located on singular flux surfaces .
however , in the presence of the non - vanishing shear velocity @xmath248 , the eigenmodes spread out on neighboring field lines , effectively coupling the motion on different flux surfaces .
the continuum of alfven frequencies @xmath251 will in this case be no longer solutions of the system .
instead , the coupling term gives rise to a discrete set of solutions rather than a continuum .
( [ solution ] ) is the mathematical equivalent of schrdingers equation , which can in general cases be solved numerically .
however , for many special case exact solutions exist .
let us consider , for the sake of simplicity , a field configuration in our box such that : @xmath252 we can rewrite eq .
( [ solution ] ) as follows : @xmath253 this differential equation is the mathematical equivalent of the quantum harmonic oscillator problem for which the exact solution is well known .
the eigenfrequencies are given by @xmath254 here @xmath91 @xmath255 is the quantum number labeling the harmonic - oscillator wavefunctions .
we see that instead of a continuum , we obtain a densely packed discrete set of frequencies with the frequency spacing @xmath256 . with the no - slip boundary conditions on the left and right sides @xmath257
, the eigenvalue equation must be solved numerically .
an example of such calculation is shown in fig .
[ fig11 ] .
there , the spacing between the discrete alfven modes is shown to increase as one increases the effect of the field tangling characterized by the @xmath227 .
we now introduce the crustal modes into the problem by making the top and bottom of the box elastic and mobile .
we allow their displacement @xmath258 in the z - direction , and impose the boundary conditions on the sides : @xmath259 here the subscripts `` @xmath108 '' and `` @xmath260 '' stand for the top and bottom of the box , respectively .
the top and bottom are assumed to be thin and have mass @xmath261 and surface density @xmath262 .
the crustal equation of motion is given by @xmath263 where @xmath264 is the shear velocity in the crust .
the crustal angular frequencies are given by @xmath265 with the corresponding crustal mode eigenfunctions @xmath266\}$ ] , where @xmath267 is roughly equivalent to @xmath201 in the spherical case .
the symmetry of the problem allows one to consider either symmetric @xmath268 or antisymmetric @xmath269 crustal modes .
this will couple to the symmetric ( @xmath270 ) or antisymmetric ( @xmath271 ) alfven modes of the core . just as in section 4
, it is now convenient to define a new variable @xmath272 for the core displacement : @xmath273 where @xmath274 the new variable observes the regular boundary condition @xmath275 on all the box edge , and satisfies the following inhomogeneous partial differential equation : @xmath276 where @xmath277 the advantage of the new variable is that it satisfies the regular boundary condition @xmath275 on all the boundaries of the box .
it can therefore be expanded as a series consisting of eigenfunctions @xmath278 of the right - hand side of equation ( [ waveeqred ] ) : @xmath279 the rest of the procedure is very similar to that in section 4 .
we expand the crustal displacement into a series consisting of the eigenmode wavefunctions @xmath280 : @xmath281 where @xmath282 and @xmath283 are real numbers .
the magnetar deformation is now fully represented by a set of generalized coordinates @xmath284 $ ] .
the coupled equations of motion are derived by following the procedure specified in section 4 .
we obtain the following system of equations : @xmath285 \alpha^{(p)}_{(mn)j } \nonumber\\ & & -\sigma_j\left[\ddot{q}_j+c_s^2\left({j\pi\over l_x}\right)^2q_j\right ] \alpha^{(q)}_{(mn)j},\label{couple1}\end{aligned}\ ] ] and @xmath286 where @xmath287 ^ 2 dx dy}\nonumber\\ \alpha^{(q)}_{(mn)j}&=&{\int \left({1\over 2}-{y\over l_y}\right ) \xi_{mn}(x , y)\bar{\xi}_j(x ) dx dy\over \int [ \xi_{mn}(x , y)]^2 dx dy}\label{couple4}\end{aligned}\ ] ] and @xmath288 ^ 2 dx}. \label{couple5}\ ] ] thus we have obtained a system of linear second - order differential equations , which describes the time evolution of the square - box magnetar .
these equations are solved by trancating all the series [ i.e. , rescricting the range of indices @xmath289 and then by either solving the eigenvalue problem in order to find the normal modes , or by integrating the equations numerically .
one then checks that the series trancation does not introduce errors in the magnetar s motion within the frequency range of our interest .
so far we have worked in the approximation of the thin crust , i.e. we have effectively included the crustal modes which have no radial nodes in their wavefunction . however , several observed high - frequency qpos , and in particular the strong qpo at @xmath0hz ( watts & strohmayer 2006 ) necessitate introduction of higher radial - order modes into our model . in the square - box model
we do this phenomenologically , as follows .
we assume that higher radial - order crustal modes have amplitudes @xmath290 and @xmath291 and natural eigenfrequencies @xmath292 , with @xmath293 being the number of radial nodes , and assume that they cause displacement at the crust - core interface given by @xmath294 .
this mirrors realistic spherically - symmetric case where the functional form of the crust - core displacement due to the torsional @xmath295 mode of the @xmath91th radial order is a very weak function @xmath91 .
the amplitudes @xmath290 and @xmath291 are then introduced on into the equations of motions ( [ couple1 ] ) and ( [ couple2 ] ) in the same way as the other @xmath296 and @xmath297 amplitudes , with the same @xmath298-dependent coupling coefficients but with @xmath299 instead of @xmath300 on the left - hand side of eq .
( [ couple2 ] ) .
we now have the basic ingredients of building a phenomenological modes with tangled fields . to sum up , we ( 1 ) quantify tangling using the effective shear modulus , ( 2 ) find discrete core eigenmodes and evaluate their coupling to the crustal model , and
( 3 ) either find eigenfrequencies of the total star by diagonalizing the potential energy of the system , or simulate the time - dependent behavior directly .
an example of a resulting powerspectrum is shown in fig [ fig12 ] for the model described in this section .
in this paper we have developed a formalism which allows one to build a magnetar model with a variety of the spectral features of the core alfven waves , including continua with gaps and edges , and the large - scale discrete modes generated by the field tangling .
we have constructed a number of magnetar models and explored the resulting qpos , both for the case of axisymmetric magnetar with core alfven continuum , and for the `` square '' magnetar models with the tangled fields ( see the previous section )
. the full range of model parameters , and detailed comparison with the data will be the subject of a separate study .
for now , we have restricted ourselves to the standard magnetar model , in which the core is a perfect conductor , the field of @xmath301 g penetrates both the core and the crust , and the proton fraction in the star is the one tabulated in haensel , potekin , and yakovlev ( 2007 ) .
our models give us the following robust conclusions , as compared against qpo observations : ( 1 ) .
a number of strong qpos have been observed in the 1998 and 2004 giant flares , with frequencies ranging between 18 hz and 150 hz ( israel et al .
2005 , strohmayer and watts 2006 , watts and strohmayer 2006 ) .
these qpos can be qualitatively explained as gap and/or edge modes of sections 4 and 2 , or even transient qpos of section 3 .
however , this was only possible if the neutrons were decoupled from the alfven waves in the core . if the neutrons took part in the alfven motion , then the effective mass of the alfven modes shifted up by a factor of @xmath302 and their frequencies shifted down by a factor @xmath303 ( easson & pethick 1979 , alpar et al .
1984 , van hoven & levin 2008 , andersson et al . 2009 ) . as a result ,
all modes at frequencies above @xmath304hz were strongly damped ( see fig .
[ figspec1 ] ) . increasing
the magnetic - field tension by a factor of 3 did not affect this conclusion ( fig [ figspec3 ] ) .
for the spherical magnetar models of section 4 we obtain similar results if couple the neutrons to the alfven motion in the core . the key point that we would like the reader to appreciate
is that alfven modes in the core are key to determining the frequency and strength of the observable qpos , and thus qpos are very sensitive probe of the core interior . but with neutron mass - loading .
due to the mass - loading the frequencies have shifted down by a factor of @xmath305 .
note that there is no significant power above the lower edge - mode frequency of @xmath306 hz.,width=336 ] , but in addition to the neutron mass - loading , we have increased the magnetic field strength by a factor of 3 .
all frequencies above @xmath307 hz are significantly damped.,width=336 ] \(2 ) a number of the high - frequency qpos have been measured in the 2004 giant flare by watts and strohmayer ( 2006 ) , the strongest among them being the qpo at @xmath0 hz .
this qpo is particularly strong and long - lived in the hard x - rays , reaching the amplitude of @xmath308% over the time interval of @xmath27 seconds ( i.e. , it persists for almost @xmath309 oscillation periods ! ) .
watts and strohmayer ( 2006 ) argued that this frequency corresponds to the crustal shear mode with a single radial node ( see also piro 2005 ) ; this interpretation , if correct , would strongly constrain the thickness of the crust and rule out the fluid strange stars as magnetar candidates ( watts & reddy , 2007 ) . to investigate this suggestion ,
we have introduced several high - frequency low - j crustal modes into our square - box simulations . however , as is demonstrated in figs .
[ fig9 ] and [ fig10 ] , the high - frequency modes are strongly damped and at no time during the simulations do we observe any significant power at those frequencies .
this is to be expected .
no natural axisymmetric model has gaps in the alfven continuum at such high frequencies , so global modes are strongly absorbed .
one could expect that if the alfven modes are discrete in the core due to the field tangling , this problem would not arise .
however , even in the discrete case the frequency spacing between the modes is around @xmath4 hz , which is much smaller than @xmath310 hz .
thus the grid of alfven waves is so dense that it is effectively seen as the absorbing continuum by the modes around @xmath310 hz . our detailed simulations , of the type shown in figs .
[ fig9 ] and [ fig10 ] , fully confirm this qualitative picture .
the concern about the viability of high - frequency qpos as being due to the physical oscillations of standard - model magnetars has been raised in the original l06 paper on the basis of rather simplistic calculations .
as our work here shows , more detailed calculations partially alleviate the l06 concern for the frequencies below @xmath311hz , but only if the neutrons are decoupled from the alfven motion in the core , i.e. if at least one baryonic superfluid ( protons or neutrons ) are present in the neutron - star core .
our analysis sustains l06 concern for the high - frequency qpos , in particular for the strong long - lived qpo at 625hz .
its explanation seems to require either qpo production in the magnetosphere , or a somewhat radical revision of the magnetar model .
just how radical this revision has to be will be explored in a separate study .
our work presented here has several shortcomings .
we have limited ourselves to the linear approximation , and a non - linear regime may bring surprises .
direct non - linear simulations of axisymmetric oscillations of a magnetised fluid star has been carried out recently by cerda - duran , stergioulas , & font ( 2009 ) . at this stage
it is difficult to say whether non - linearities introduce significantly new qpo features to their model ; their results have largely been in agreement with the linear simulations of colaiuda , beyer , & kokkotas ( 2009 ) . however , the computational techniques seem promising and we do not exclude that large - amplitude simulations of stars with the crust will show qualitatively new features .
another limitation of our work is that we have assumed that once the flare sets the magnetar into motion , the magnetar s oscillations are not driven externally
. this may not be the case in real flares : some energy stored in the pre - flare magnetar may be released gradually , and this release could be extended in time into the flare s tail . the latter consideration is straightforward to incorporate phenomenologically into our model , and we plan to address it in our future work .
we thank andrei beloborodov , anna watts , peter goldreich , chris thompson , and especially andrei gruzinov for useful discussions .
this research has been supported by leiden observatory and lorentz institute for theoretical physics through internal grants .
alpar , m. a. , langer , s. a. , & sauls , j. a. 1984 , apj , 282 , 533 + andersson , n. , glampedakis , k. , & samuelsson , l. 2009 , mnras , 396 , 894 + barat c. et al . , 1983 , a & a , 126 , 400 + braithwaite , j. 2008 , mnras , 386 , 1947 + braithwaite , j. , & spruit , h. c. 2004 , nature , 431 , 819 + braithwaite , j. , & nordlund , a. 2006 , a&a , 450 , 1077 + cerda - duran , p. , stergioulas , n. , & font , j. 2009 , mnras , 397 , 1607 + colaiuda , a. , beyer , h. , & kokkotas , k.d . 2009 , mnras , 396 , 1441 + duncan , r. c. 1998 , apj , 498 , l45 + easson , i. , & pethick , c. j. 1979 , 227 , 995 + glampedakis k. , samuelsson l. , & andersson n. , 2006 , mnras , 371 , l74 + goedbloed , j. p. h. , & poedts , s. 2004 , principles of megnetohydrodynamics ( gp in the text ; cambridge university press ) + gruzinov , a. , 2008a , arxiv : 0801.4032 [ astro - ph ] + gruzinov , a. , 2008b , arxiv : 0812.1570 [ astro - ph ] + gruzinov , a. , 2009 , arxiv : 0905.0911 [ astro - ph ] + haensel p. , & potekhin a. y. , 2004 , a&a , 428 , 191 + http://www.ioffe.ru/astro/nsg/nseos/ + haensel p. , potekhin a. y. , & yakovlev d. g. , 2007 .
neutron stars 1 : equation of state and structure ( new york : springer ) + hollweg , j. v. 1987 , apj , 312 , 880 + israel g. l. et al . , 2005 , apj , 628 , l53 + ionson , j. a. 1978 , apj , 226 , 650 + landau , l. d. , & lifshits , e. m. 1976 , mechanics , chapter v ( pergamon press ) + lee , u. 2008 , mnras , 385 , 2069 + levin y. , 2006 , mnras letters , 368 , 35 ( l06 in the text ) + levin y. , 2007 , mnras , 377 , 159 ( l07 in the text ) + piro , a. 2005 , apj , 634 , l153 + poedts , s. , hermans , d. , & goossens , m. 1985 , a&a , 151 , 16 ( p85 in the text ) + samuelsson , l. , & andersson , n. 2007 , 374 , 256 + sotani , h. , kokkotas , k. d. , & stergioulas , n. 2008 , mnras letters , 385 , 5 + steiner w. , & watts a. l. , 2009 , phys .
letters , 103r1101s + strohmayer , t. e. & watts , a. l. , 2005 , apj , 632 , l111 + van hoven , m. , & levin , y. 2008 , mnras , 391 , 283 + watts , a. l. & strohmayer t. e. , 2006 , apj , 637 , l117 + watts , a. l. , & reddy , s. 2007 , mnras letters , 379 , 63 +
in this appendix we generalize the normal - mode treatment of section 2.2 , and write down the general prescription of how to find the eigenmodes when _ several _ `` large '' crustal shear modes are coupled to a multitude of small core alfven modes , provided the coupling coefficients are known . in this paper ,
the coupling coefficients are worked out in simple models of sections 4 and 5 ; we shall postpone the discussion of how the coefficients are computed in more general case to the future paper .
let us denote the displacement of the crustal and core modes by @xmath312 and @xmath313 respectively .
since both the crustal and the core modes are not directly coupled to themselves ( i.e. , @xmath249 s are only coupled to @xmath146 s ) , most general equations of motion take the form @xmath314 where @xmath315 and @xmath316 are the proper frequencies of the crustal and core modes , and @xmath193 s and @xmath317 s are the coupling coefficients .
we look for an oscillatory solutions of the above equations with angular frequency @xmath32 .
one can trivially re - write these equation as a matrix eigen - equation with @xmath318 as an eigenvalue , and solve it using standard methods . however , if the number of crustal modes is not too large , it is convenient to make a shortcut . using the second of eq .
( [ multimode1 ] ) to express @xmath313 s through @xmath312 s , and substituting into the first one , we get the following equation : @xmath319 where the elements of the matrix @xmath320 are given by @xmath321 one obtains the eigenfrequencies by finding numerically the zeros of @xmath322 .
in this appendix we will calculate the continuum of alfven frequencies in a magnetar core in the case of a axisymmetric magnetic field with mixed toroidal and poloidal components .
the general mhd equations of motion for spherically symmetric , self - gravitating equilibrium with an axisymmetric field , are derived in detail in p85 .
in contrast to the special case of a purely poloidal field ( see section 4.2 ) which leads to two uncoupled differential equations , the continuum for a mixed toroidal - poloidal field is described by a system of fourth order coupled odes . due to this coupling
, the solutions are complicated as they are no longer polarized in the directions parallel ( so - called `` cusp solutions '' ) and perpendicular ( alfven solutions ) to the magnetic field lines , but rather have a mixed character .
strictly speaking , one can only speak of an `` alfven continuum '' in the limit that the variations in @xmath231 , @xmath323 and @xmath324 are small in the magnetic flux - surfaces .
the general equations of motion are given in eqs .
( 53 ) and ( 54 ) of p85 .
we note however , that in magnetars the speed of sound @xmath325 , and therefore we consider poedts et al.s equations ( 53 ) and ( 54 ) in the incompressible limit ( p85 , eqs .
( 73 ) and ( 74 ) ) . for completeness
we give the equations here , the variables @xmath328 and @xmath329 are components of the fluid displacement perpendicular and parallel to the magnetic field lines , the operator @xmath330 is a differential operator along the field lines , @xmath331 can be thought of as a brunt - visl frequency for displacements along the field lines . according to gauss law for magnetism ,
the toroidal component of the magnetic field is of the form @xmath332 , where @xmath333 is the distance to the polar axis , and @xmath334 is an arbitrary function of @xmath136 . in the following calculation we adopt a toroidal field component of the form here @xmath160 is the polar angle at which the flux surface @xmath136 intersects the stellar crust .
clearly this choice for @xmath336 is completely arbitrary and one could in principle try many different toroidal geometries . + as with our calculation of the alfven continuum in the case of a purely poloidal field ( section 4.2 ) , we adopt the zero - displacement boundary conditions at the crust , and use the fact that our equilibrium model is ( point- ) symmetric with respect to the equatorial plane .
this enforces the existence of classes of symmetric and anti - symmetric eigenfunctions , @xmath337 and @xmath338 .
we consider only the odd modes , and calculate the eigenfunctions by means of the shooting method ; we use a fourth order runge - kutta scheme to integrate eqs .
( [ b1 ] ) and ( [ b2 ] ) .
starting with @xmath339 and @xmath340 at the equator , we integrate outward until we reach the crust at @xmath155 .
we find the eigenfrequencies by changing the value of @xmath158 until we match the boundary conditions at the crust .
a resulting continuum is plotted in figure [ pt_freqs ] . as a function of the angle @xmath160 , the polar angle at which the flux - surface @xmath136 intersects the crust . in the presence of a toroidal field ,
the degeneracy between the cusp - solutions and the alfven solutions is broken and we find two separate solutions for each wave number @xmath91 ; waves with primarily alfven character ( red curves ) and waves with primarily cusp character ( blue curves ) .
this particular continuum was calculated using a poloidal field with an average surface value @xmath341 ( again generated by a circular ring current of radius @xmath162 ) and a toroidal field strength at the equator and the crust - core interface of @xmath342 g ( see eq .
( [ btor])).,width=384 ] | quasi - periodic oscillations ( qpos ) observed at the tail end of soft gamma repeaters giant flares are commonly interpreted as the torsional oscillations of magnetars . from a theoretical perspective
, the oscillatory motion is influenced by the strong interaction between the shear modes of the crust and magnetohydrodynamic alfven - like modes in the core .
we study the dynamics which arises through this interaction , and present several new results : ( 1 ) we show that discrete _ edge modes _ frequently reside near the edges of the core alfven continuum , and explain using simple models why these are generic and long - lived .
( 2 ) we compute the magnetar s oscillatory motion for realistic axisymmetric magnetic field configurations and core density profiles , but with a simplified model of the elastic crust . we show that one may generically get multiple gaps in the alfven continuum .
one obtains strong discrete _ gap modes _ if the crustal frequencies belong to the gaps ; the resulting frequencies do not coincide with , but are in some cases close to the crustal frequencies .
( 3 ) we deal with the issue of tangled magnetic fields in the core by developing a phenomenological model to quantify the tangling .
we show that field tangling enhances the role of the core discrete alfven modes and reduces the role of the core alfven continuum in the overall oscillatory dynamics of the magnetar .
( 4 ) we demonstrate that the system displays transient qpos when parts of the spectrum of the core alfven modes contain discrete modes which are densely and regularly spaced in frequency .
the transient qpos are the strongest when they are located near the frequencies of the crustal modes .
( 5 ) we show that if the neutrons are coupled into the core alfven motion , then the post - flare crustal motion is strongly damped and has a very weak amplitude .
we thus argue that magnetar qpos give evidence that the proton and neutron components in the core are dynamically decoupled and that at least one of them is a quantum fluid .
( 6 ) we show that it is difficult to identify the high - frequency @xmath0hz qpo as being due to the physical oscillatory mode of the magnetar , if the latter s fluid core consists of the standard proton - neutron - electron mixture and is magnetised to the same extent as the crust . |
the general theory of the quantum measurement of relative - phase is summarized in [ 1 ] wherein it is also shown to be equivalent to the quantum measurement of an angle about an axis ( complementary to the angular momentum along that axis ) when the relative - phase is taken to be between the `` up''and `` down '' oscillators of schwinger s harmonic oscillator model of angular momentum [ 2 ] .
the quanta of schwinger s oscillators are presumably unphysical ( as they are spin-1/2 primitives which behave like bosons ) yet that model has proved useful for a variety of visualizations and calculations including the analysis of beam splitters and interferometers in quantum optics [ 3 ] .
a slight modification of schwinger s harmonic oscillator model of angular momentum permits the inclusion of photons [ 4 ] wherein the primitives become the physically realizable quanta of the right and left circular polarization modes of an electromagnetic plane wave . therein the quantum angle measurement ( equivalent to the relative - phase measurement between the two circular polarization modes ) provides novel insight on optical polarization ; including the fact that for x - polarized photon number states such angle will _ never _ correspond to the y - axis for an odd number of photons , but for an even number of photons it _ always _ can [ 1 ] , as demonstrated in fig . 1 for the cases of one , two and three x - polarized photon number states .
owing to a factor of two between schwinger s model and its photonic variation [ 1,4 ] the analogy of this surprizing effect , for fermions and ordinary ( i.e. , non - photonic ) bosons , is that `` spin up can point down '' for bosons but it never does for fermions ( as demonstrated in fig . 2 for spin-1/2 , spin-1 and spin-3/2 particles ) ; wherein by `` spin up along @xmath0 '' we mean @xmath1 where @xmath2 is the eigenvalue of @xmath3 for a particle of total angular momentum labeled by @xmath4 ; and by `` point down '' we mean the angle about the z - axis , @xmath5 , is @xmath6 ( where @xmath7 corresponds to the x - axis ) .
note in passing that fig .
2 also demonstrates the phenomenon of `` spin up really does point up '' [ 4 ] i.e. , not only can @xmath7 occur for such states , but that is also its most likely value in contrast to a vector model which ( prior to the establishment of the quantum angle formalism ) would attempt to _ infer _ @xmath5 from the measurement of @xmath8 and @xmath9 ( rather than directly describing the _ measurement _ of @xmath5 ) .
returning to the `` spin up can point down '' for bosons phenomenon : note that the angular distributions in fig .
2 are also the relative - phase distributions between the two oscillators of schwinger s model . since such oscillators can also model the upward and downward paths in an interferometer the following question naturally arises .
namely , can the surprizing behavior at @xmath10 in fig . 2 ( analogous to the surprizing behavior at @xmath11 in fig .
1 ) be observed in a standard su(2 ) interferometer ?
the answer turns out to be no . the quantum phase representation of an su(2 )
interferometer discussed herein shows that there is no inconistancy and illucidates the relationship between these two quantum measurements .
it will also fascilitate the calculation of some of our results as well as provide insight on the range and sensitivity that one can expect from a quantum phase algorithm which can surpass the `` heisenberg limit '' by orders of magnitude while utilizing the simple apparatus of an su(2 ) interferometer !
the transformation between linear and circular polarization basis annihilation operators follows that of their associated classical vectors : @xmath12 thus one x - polalrized photon @xmath13 is a supperposition of one right - handed ( @xmath14 ) circularly polarized photon and one left - handed ( @xmath15 ) circularly polarized photon , which have been created `` in phase '' ( which we define to mean that there is no phase shift between the creation operators which create the two modes ) i.e. , @xmath16 throughout this paper we adhere to the above definiiton of two modes being `` in phase , '' which is not to be confused with the probability of the outcome of zero when their relative - phase is measured . to form the quantum angle representation of such a state we use the photonic primitives [ 1,4 ] ( instead of schwinger s [ 2 ] primitives ) which reflect the fact that a photon is a `` spin-1 particle with @xmath17 missing '' [ 5 ] wherein @xmath18 in general , as summarized in [ 1 ] , we can perform the quantum angle measurement via a `` snapshot '' in absolute time ( a conditional probability operator measure , which involves the addition of amplitudes associated with different @xmath4 ) or via a `` time average '' ( a marginal probability operator measure , which involves the addition of probabilities associated with different @xmath4 ) . for states with a unique value of @xmath4 however ( such as the x - polarized photon number states )
both of these procedures reduce to a simple fourier transform of the complementary wavefunction @xmath19 , the magnitude square of which yeilds the quantum angle distribution , viz . , @xmath20 the quantum angle distribution of a single x - polarized photon utilizes ( 3 ) so that @xmath21 increments by two and we obtain @xmath22 ^ 2.\ ] ] similarly , we have @xmath23 & & % \end{equation}\end{aligned}\ ] ] so that the quantum angle distribution of a two - photon x - polarized number state is @xmath24 ^ 2.\ ] ] in general the x - polarized number states yield @xmath25 ^ 2\ ] ] where @xmath26 is always an integer and @xmath21 so that the sum is from 1 when @xmath27 is odd ( in which case @xmath28 ) or from 2 when @xmath27 is even .
the sum is to an upper limit of @xmath29 in increments of two ; and the coefficients are generated recursively from @xmath30 for @xmath31 with an initial condition of @xmath32 and @xmath33 for even @xmath4 ( otherwise it vanishes ) .
we note that @xmath34 and we will later show how these coefficients are related to the wigner d matricies .
note the generalization of `` spin-1 with @xmath17 missing '' is that @xmath35 is odd ( even ) when @xmath27 is odd ( even ) .
therefore ( 8) reveals that x - polarized photon number states of an odd number have @xmath36 so that the angle will _ never _ correspond to the y - axis for such states . yet
( 8) also shows that x - polarized photon number states of an even number have @xmath37 so that the angle _ always _ can correspond to the y - axis for such states .
a subtle point which now bears mentioning is that the optical polarization basis transformation in ( 1 ) belongs to the u(2 ) group whereas rotations are isomorphic to su(2 ) ( where the s stands for special in that its unitary matricies also have a determinant equal to one ) .
the beam splitters of section ii also belong to su(2 ) but notice we restricted our attention to only x - polarized photons in the above ( leaving the y - polarization in the vacuum state )
. thus we will be able to map to analogous behavior by also restricting our attention to only single - port input excitations of the su(2 ) interferometer ( leaving the other input port in the vacuum state ) .
formally , the quantum angle measurement for a state with a single value of @xmath4 , e.g. , that of a spin-@xmath4 particle , will require an additional system such as an electromagnetic field to provide a hilbert space large enough to permit wavefunction collapse ( or equivalently a desription of the measurement in terms of sets of commuting observables ) [ 1 ] .
if however the field is in ( for example ) the vacuum state prior to the measurement then the single-@xmath4 , so simply presented in ( 4 ) , will manefest .
to maintain an analogy with the x - polarized photon number states we now consider a spin-@xmath4 particle which is `` spin up along the x - axis , '' i.e. , a particle in the state @xmath38 where @xmath2 is the eigenvalue of @xmath3 .
of course we need not employ any oscillators , but in schwinger s model these states would correspond to up / down oscillators for @xmath8 ( instead of @xmath39 ) of photon number @xmath40 and @xmath41 ( analogous to a single - port input excitation ) .
we transform the @xmath42 representation into the @xmath43 representation via a @xmath44 rotation about the y - axis @xmath45 where we inserted a resolution of the identity operator , @xmath46 , on the left of the rotation operator , @xmath47 for kets without a subscript , the z - axis is implied ( as it is also implied for @xmath48 and @xmath5 throughout this paper ) . for simplicity we supressed the @xmath4 label in the kets ( and bras ) since our attention is currently on the single-@xmath4 statistics and the @xmath49 are from the well - known wigner d matrix elements : @xmath50 which can be calculated in a variety of ways [ 7 ] .
the `` spin up along x '' kets have @xmath51 so the coefficients in their @xmath43 representation correspond to the first column of the wigner d matricies ( of @xmath52 ) .
an explicit example will mitigate the risk of any notational confusion so let s consider spin - one ( i.e. , @xmath53 ) in the following .
we have @xmath54 so that ( 4 ) yields @xmath55 ^ 2.\ ] ] notice that the quantum angle distribution of spin-1 up along @xmath0 is related to that of the two - photon x - polarized number state as follows : @xmath56 which exemplifies how the surprizing behavior at @xmath11 illustrated in fig .
1 turns into the surprizing behavior at @xmath10 illustrated in fig .
2 . of course
, the wigner d matricies can be calculated via schwinger s harmonic oscillator model [ 7 ] and we could have performed an analysis similar to that which led to ( 8) via schwinger s primitives instead of right and left circular polarized photons . moreover , schwinger s up / down oscillators can also model the upward / downward paths inside an su(2 ) interferometer[8 ] and the angle distribution in ( 4 ) is also the relative - phase distribution [ 1 ] between these two modes .
in the quantum theory of an interferometer [ 3 ] there are many possible choices of beam splitter reference planes which correspond to many possible choices of axies of rotation .
we choose the y - axis so that the wigner d matrix elements will occur in what follows .
thus , we can choose to have either the @xmath42 representation `` outside '' the beam splitters and the @xmath43 representation `` inside the interferometer '' or vice - versa .
we choose the former in order to maintain similarity to the above , as well as to simplify the notation so that we wo nt need a subscript on @xmath5 since the z - axis will be implied .
we also choose the first beam splitter to correspond to a rotation of @xmath57 and the second one to be of @xmath58 so that spin up will always correspond to the upward path .
the transformation of the first beam splitter s input annihilation operators ( denoted with an `` a '' ) into its output annihilation operators ( denoted with a `` b '' ) follows that of @xmath59 : @xmath60 thus @xmath61 ( analogous to @xmath62 ) splits the single - port input into two `` in phase '' modes inside the interferometer ; and we create the @xmath43 representation of the input number states ( of @xmath42 representation ) in parallel with how we created the circular polarization representation of the linear polarized number states ; but this time , instead of ( 3 ) we have @xmath63 and @xmath64 , i.e. , schwinger s primitives , so that each quanta is associated with a z - component angular momentum of 1/2 .
notice that in ( 14 ) we changed @xmath4 ( and @xmath5 , because @xmath48 changed ) but the coefficients did not change . since we could also do that calculation via the wigner d matricies , the @xmath65 are an alternate way of calculating the same numbers as the first column of these matricies and we have @xmath66
this also proves that the @xmath67 must be symmetric in @xmath48 , so their fourier transform results in terms only of the form @xmath68 and the interpretation of the surprizing effect at @xmath10 is that two modes which have been created `` in phase '' ( via an _ even _ number of creation operations ) can have a relative - phase of @xmath6 when it is measured ; yet those created via an _ odd _ number of creation operations can not . note in passing that a mode input to the @xmath69 port of this beam splitter would be split into two - modes which are created to be 180 degrees out of phase but herein we do not excite that input port , just as we did not bring the y - polarization out of the vacuum state in section ii .
so we also need not concern ourselves with the difference between u(2 ) and su(2 ) , e.g. , the absence of an @xmath70 in @xmath71 as opposed to @xmath72 since we never used either of these equations ; and restricted our attention instead to @xmath61 and the analogous @xmath62 . moreover , if we excited some y - polarization ( or an input to the 180 degree splitter port ) then we would not be surprized to observe an optical polarization angle at @xmath44 ( or a relative - phase of @xmath6 ) . in the next section
we derive the quantum phase representation of an su(2 ) interferometer without any restriction on its input state .
after which , we will then return to the analysis of the single - port input statistics .
although they correspond to quantum measurements which are realizable ( in principle ) no one , to date , has identified an apparatus which can realize the quantum phase or quantum angle measurements .
interferometers are typically used to @xmath73 phase , so the question arises as to wheather or not such `` out of phase type behavior '' can be demonstrated in an su(2 ) interferometer ?
the answer clearly is no , since if we phase shift the upper path by some amount @xmath74 and recombine the two via a beam splitter of reference planes associated with a rotation about @xmath75 by @xmath58 then at @xmath76 the interferometer reduces to the indentity operator reproducing the input state at its output @xmath77 input states @xmath78 .
thus , spin up at the input can never result in any photons going into the `` down port '' photodetector at the output when @xmath76 independent of the value of @xmath79 .
this is not to say that such an effect can not be observed via some @xmath69 quantum measurement for which we presently have a physical realization ( e.g. , perhaps it could be observed in heterodyne or homodyne detection , etc . ) .
it is instructive to explicitly demonstrate the relationship between the quantum measurement of phase and the quantum measurement realized in an su(2 ) interferometer : not only to demonstrate that there is no inconsistancy in their predictions and provide clues as to how one might modify the interferometer s apparatus in order to observe the surprizing effects above ; but also the quantum phase representation of an su(2 ) interferometer will provide novel insight as to how an interferometer works ( as did the analogies to rotations [ 3 ] ) .
most importantly , it will facilitate proofs regarding the range and sensitivity one can anticipate in quantum phase statistical inference ( qpsi ) algorithms which can yeild accuracies that are orders of magnitude better than the `` heisenberg limit , '' while utilizing the simple apparatus of an su(2 ) interferometer ( and classical signal processing that has been armed with quantum knowledge of the possible outcomes ) [ 8,9 ] . of : m = j = 1/2 ( solid , blue ) ; m = j = 1 ( dashed , red ) ; and m = j = 3/2 ( dot - dashed , green ) .
concentric gray circles deliniate steps of 20 db . ]
first , there is a subtle point that we wish to clarify so that the notation is transparent to an audience wider than just the quantum optics community .
recall we chose to have the @xmath42 representation `` outside '' the beam splitters and the @xmath43 representation `` inside the interferometer . '' but it would be wrong to say , for example , that with an input state of @xmath80 the state of the first beam splitter s output is @xmath81 and also utilize ( 10 ) , since that would be tantamount to performing the rotation assosicated with the input beam splitter , @xmath82 , twice .
the beam splitter of course @xmath83 perform an operation on these states ( in accordance with the transformation of annihitaion operators in the heisenberg picture , as in ( 15 ) for our choice ) .
the use of ( 10 ) makes it look as if we chose to use two different representations of the @xmath84 state with its @xmath42 representation on the left of @xmath82 and its @xmath42 representation on the right of @xmath82 .
the subtlety is that those are different _ spatial _ modes ( as clarified in ( 15 ) by the use of @xmath85 versus @xmath86 ) . in other words ,
the upward / downward paths on the left are different spatial modes than the upward / downward paths on the right .
perhaps the safest notation is to let @xmath78 represent the state at the output of the first beam splitter ( and let @xmath82 be thought of as part of the state preperation step examples of which in the heisenberg and schrodinger pictures being given in the previous sections ) .
thus , the interferometer s measurement statistics can be written as @xmath88 where is the underlying wavefunction in measurement outcome @xmath90 .
note that @xmath91 ( the eigenvalue of @xmath3 ) is 1/2 the difference of the number of photodetection events in the up / down photodetectors ( within some observation time @xmath92 ) when one imposes a differential phase shift @xmath74 between the upward and downward paths inside the interferometer analogous to the rotation @xmath93 . for arbitrary states ( rather than those comprised of a single value of @xmath4 , such as the single - port photon number states )
we could argue that the measurement outcomes from different @xmath4 are distinguishable in principle ( wheather or not we record @xmath4 ) so that in a standard interferometer we would add probabilities : @xmath94 [ 10 ] . inserting a resolution of the identity operator in terms of the angle - kets , @xmath100 , on the left of the rotation operator in ( 17 )
then yields @xmath101 now , utilizing the adjoint of ( 10 ) we have @xmath102 so that the quantum phase representation of an su(2 ) interferometer is @xmath103\ ] ] where the functions @xmath104 are the fourier transforms of the m - th row of the wigner d matricies ( at @xmath105 ) .
the @xmath104 functions represent the action of the output beam splitter and its two photodetectors ; as they project the angle - kets onto the kets of what the interferometer measures .
these state - independent functions are known ( in as much as the @xmath106 are ) and the only dependence of the interferometer s measurement statistics on the quantum state of the field is via their convolution with its angle representation ( which reduces the action of the rotation operator to a simple translation of a functional argument ) .
it is easy to show that the @xmath104 form a set of orthogonal functions ( for any @xmath4 ) @xmath107 so , from the orthogonality of the @xmath108 @xmath109 where the sum is over the columns of `` row m '' times `` row n '' ( and we supressed @xmath110 ) .
but the n - th row of @xmath111 is also the n - th column of its inverse , hence @xmath112 ( the kronecker delta ) as asserted .
when a single - port number state is input to the interferometer ( 16 ) shows that @xmath113 itself becomes @xmath114 so @xmath115 \nonumber \\ & \rightarrow & \int_{-\pi}^{\pi}\frac{d\phi}{2\pi } \ s^{(j)}_{m = j}(-\phi ) \
s^{(j)}_{m=- j}(\phi ) % \nonumber \\ % \textit{as } \phi & \rightarrow & \pi\end{aligned}\ ] ] as @xmath116 , where we used @xmath117 under a rotation by @xmath6 .
orthogonality then shows that this vanishes , independent of the value of @xmath118 .
thus there is no inconsistency .
the state - independent behavior of the interferometer at the extremes of @xmath76 and @xmath119 can of course been seen more readily by considering it as one equivalent rotation , but now we have clues as to how we might modify the apparatus in order to reveal the behavior at @xmath10 .
it is this squaring of the wigner d matrix elements which prevents the observation of such effects and we make the conjecture that instead of `` beating the signal with itself '' ( so to speak ) we should beat it with a seperate local oscillator ( as in heterodyne detection ) .
an analgous conjecture would be that these effects might be observed in scattering experiments on trapped ions ( e.g. , perhaps as in raman heterodyne spectroscopy ) .
we now turn our attention to the use of this representation to provide insight on the range and sensitivity that one can expect from quantum phase statistical inference algorithms [ 8,9 ] .
the squares of the @xmath120 turn out to be proportional to the binomial coefficients , thus we can prove that for the single - port number states we have @xmath121^{4j}$ ] . similarly we can show that @xmath122^{4 j -2 } \left [ \text{sin}(\phi /2)\right]^{2}\ ] ] etc . , lowering in the same fashion down to @xmath123^{4j}$ ] .
closed - form expressions are useful but so are geometric interpretations .
3 presents the @xmath124 versus @xmath74 for @xmath125 .
the behavior of @xmath126 stems from ( 23 ) as it is the phase representation of the state convolved with itself .
let @xmath127 for the moment . as @xmath48 decreases towards zero the higher frequency components in @xmath104 increase in amplitude ( because the @xmath128 get larger in absolute value ) so that their convolution with the phase representation will displace the resulting peaks further away from @xmath129 .
the behavior for @xmath130 follows from a rotation by @xmath131 when we consider e.g. , a simultaneous lms fitting of these functions to measured interferometer statistics ( obtained during a fixed and unknown value of @xmath74 ) we see that the @xmath132 monotonicity of the @xmath133 on [ 0 , @xmath6 ] will resolve which side of the peaks we must be on for the other statistics ; which in turn enable decent sensitivity ( slope ) throughout this range .
4 illustrates similar behavior for @xmath135 wherein we also see the increased sensitivity resulting from the use of more photons ; as well as the anticipation of good performance closer to zero and @xmath6 ( where all slopes must vanish due to symmetry ) . as designers of quantum measurements
however one could envision a `` stroboscopic interferometer '' in which we would add the amplitudes for outcomes @xmath91 of different @xmath4 .
indeed , such a measurement would be complementary to the `` snapshot '' angle measurement [ 1 ] . | we show that the quantum angle measurement for x - polarized photon number states results in an angle which will _ never _ correspond to the y - axis for an odd number of photons ; yet for an even number of photons it _ always _ can .
the analogy of this surprising effect for other particles or ions ( qubits in their spin up state ) is presented and we show why such an effect can not be observed in a standard su(2 ) interferometer .
the quantum phase representation of such an interferometer provides clues as to how its apparatus might be modified in order to demonstrate these phenomena . the same representation is then used to provide insight on the range and sensitivity that one can expect from quantum phase statistical inference algorithms which can surpass the `` heisenberg limit . ''
pacs numbers : : 42.50.st , 42.50.dv , 42.50.ex , 42.50.lc |
in this section we describe how the hamiltonian is diagonalized , subsequently how one obtains the fermionic correlation functions and from them the spin correlations functions that are analyzed in the main manuscript . despite the presence of three - body interactions the hamiltonian ( [ hamiltonian ] )
can be mapped into a delocalized fermionic form via a generalization of the approach described in ref .
@xcite . to obtain this map
we introduce the jordan - wigner transformation @xcite that connects the spin operators to the fermionic operator via @xmath94 where @xmath95 . applying this transformation to the hamiltonian ( [ hamiltonian ] ) we obtain @xmath96 this hamiltonian corresponds to a so - called free fermion case in which non - interacting fermions can jump between nearest and next to nearest neighbors of the chain via open boundary conditions , i.e. @xmath12 .
this hamiltonian can be brought in a diagonal form @xmath97 using the following linear transformation @xmath98 where both vectors @xmath99 and @xmath100 are a set of @xmath2 vectors with @xmath2 components that are solutions of the coupled equations @xmath101 here the matrix @xmath102 ( symmetric ) and @xmath103 ( anti - symmetric ) are , respectively , @xmath104 and @xmath105 and @xmath106 . either @xmath100 or @xmath99 can be eliminated from the equations ( [ psipsh1 ] ) resulting in @xmath107 for @xmath108 we can choose one of these two equations that gives us one set of vectors and with the help of the euations .
( [ psipsh1 ] ) one obtains the second set of vectors . for @xmath109
one obtains @xmath100 and @xmath99 from the equations ( [ psipsh1 ] ) .
the set of vectors @xmath100 and @xmath99 plays an important role for the evaluation of the fermionic correlations functions . to see that let us introduce two new fermionic operators @xmath110 the advantage to use this set of operators , @xmath111 and @xmath112 , with respect to the set of operators @xmath113 and @xmath114 comes from the fact that , taking into account the eqs .
( [ eta ] ) , we obtain particular simplifications computing the expectation values of the product of two operators , in particular : @xmath115 in the finite size case the @xmath116 depends on the choice of the site @xmath16 and @xmath93 and , therefore , also the spin correlation functions depend on the position of the spins in the chain .
however , if we evaluate the correlation functions between three central spins of a chains with an odd @xmath2 , the mirror symmetry plays the role of the invariance under spatial translation , imposing that @xmath117 on the contrary , in the thermodynamic limit the invariance under spatial translation is restored by the fact that the end of the chain is infinitely far away . in this case , imposing the invariance under spatial translation , we can solve eqs .
( [ psipsh1 ] ) considering that each components of @xmath99 and @xmath100 vectors obey to the following rule @xmath118 with a long but straightforward evaluation we obtain the fermionic correlation functions in the thermodynamic limit given in eq .
( [ gfunction ] ) and eq .
( [ lambda ] ) , respectively .
having introduced a method to evaluate the fermionic correlation functions we can compute the spin correlation functions from which we obtain any reduced density matrix .
we can divide the spin correlation functions in two families .
the first is consists of correlation functions from operators that do not break the symmetry of the hamiltonian . in this case transforming the spin operators into the fermionic operators @xmath111 and @xmath112 one always obtains an operator constructed via an equal number of @xmath111 and @xmath112 . applying the wick s theorem we
then obtain the expression of the correlation functions in terms of the @xmath52 .
let us also point out that the dependence of the spin correlation functions on the fermionic correlation functions are independent of the model .
a list of these spin correlation functions given in terms of fermionic correlation functions can be founded in ref .
@xcite .
the situation is much more involved in the case the symmetry of the hamiltonian is broken . in this second case ,
if we try to apply the same strategy as described above , we immediately observe that the elements due to parity immediately vanish .
in fact , re - writing the operators in terms of @xmath111 and @xmath112 , we obtain an expression with different numbers of fermionic operators , each one referring to distinct sites , and , in agreement with eq .
( [ correlations ] ) , this implies that the correlation function must be zero .
this result holds for all ground states that preserve the symmetry of the hamiltonian , however , in the thermodynamic limit and in the anti - ferromagnetic phase one has also ground states that break the @xmath10 symmetry of the hamiltonian . for these ground states some of these correlation functions
are expected to be different from zero . to evaluate these correlation functions
we adopt the approach of barouch and mccoy @xcite . given an operator @xmath119 defined on a set of spins @xmath120 that does not commute with the parity operator along the @xmath121 axis one can define the correlation function @xmath122 via @xmath123 for analyzing the entanglement properties of three adjacent spins one has to compute all the one , two and three body correlation functions that do not commute with the parity operators along the @xmath121 axis .
all the correlation functions of this type are vanishing in the limit of @xmath124 except @xmath73 and @xmath74 . in fig .
[ plotsupp1 ] we have reported , for some values of @xmath1 , the values of @xmath73 and @xmath74 .
we observe that the farer we are away from the quantum critical point the faster is the convergence of the correlation functions to the asymptotic values .
moreover , for all @xmath1 we observe that @xmath125 for large enough @xmath39 ( we have tested this relations for more that @xmath126 different value of @xmath127 ) . the fact that this equality holds impose that the genuine tripartite entanglement between three adjacent spins does not depends on the particular ground state selected as emphasized in the main body of the manuscript . | we evaluate and analyze the exact value of a measure for local genuine tripartite entanglement in the one - dimensional cluster - ising model .
this model is attractive since cluster states are considered to be relevant sources for applying quantum algorithms and the ising interaction is an expected perturbation . whereas bipartite entanglement is identically vanishing , we find that genuine tripartite entanglement is non zero in the anti - ferromagnetic phase and also in the cluster phase well before the critical point .
we prove that the measure of local genuine tripartite entanglement captures all the properties of the topological phase transition .
remarkably , we find that the amount of genuine tripartite entanglement is independent of whether the considered ground states satisfy or break the symmetries of the hamiltonian .
we provide also strong evidences that for this experimentally feasible model local genuine tripartite entanglement represents the unique non vanishing genuine multipartite entanglement among any spins .
quantum many - body systems usually posses a highly entangled ground state that exhibits collective quantum phenomena .
therefore , the analysis of the entanglement properties of a ground state becomes an important resource that flanks the standard analysis based on the grinzburg - landau paradigm @xcite that is based on spontaneous symmetry breaking and a non - vanishing order parameter , e.g. , `` magnetization '' @xcite .
although the presence and the importance of entanglement is recognized in many different physical contexts and physical systems , entanglement detection and quantification is a highly involved task ( for mixed states ) . except for the easiest case , bipartite spin-@xmath0 systems ,
no necessary and sufficient general method is known .
the problem becomes much more pronounced for multipartite entanglement which has been found to play an essential role in many phenomena and may provide a high - capacity resource for quantum computing . among multipartite entanglement
_ genuine _ multipartite entanglement is the most interesting one since all subsystems truly contribute to the entanglement .
for example genuine multipartite entanglement is shown to be the necessary property for protecting a secret shared among many parties against eavesdropping or unfaithfully parties @xcite . in this letter
we relate two rich phenomena in modern physics , i.e. the quantum phenomenon entanglement and ( topological ) phase transitions . in condensed matter systems phase transitions
are at the heart of our understanding of critical phenomena .
previous works have shown that phase transitions can be characterized in terms of bipartite entanglement @xcite or in terms of the von neumann entropy of bipartite spin - block entanglement @xcite , via global geometric entanglement @xcite or via analyzing the existence of points in which the entanglement is completely absent @xcite .
obviously , such methods fail for condensed matter systems that do not exhibit bipartite entanglement .
recent works @xcite have shown the presence of genuine multipartite entanglement in subsystems of typical condensed matter systems by computing criteria capable to detect different types of genuine multipartite entanglement .
this contribution shows that the exact amount of genuine tripartite entanglement can be computed exploiting a measure for genuine multipartite entanglement and that this measure characterizes fully the phase transition .
in particular , our model exhibits a topological phase ; those phases are known to play a prominent role e.g. in the quantum hall effect @xcite , for superconductivity @xcite , for the confinement problem in qcd or in string theory .
surprisingly , we find for all ground states , those preserving and those breaking the symmetries of the hamiltonian , the same amount of genuine tripartite entanglement .
the condensed matter model that we investigate has in one limit a pure ising interaction and in the other limit a cluster state as a ground state .
cluster states are a special type of multi - qubit states of graph states which are conjectured to be important resources for quantum algorithms ( see , e.g. , ref.@xcite ) .
this cluster - ising model can be put to reality , e.g. , for cold atoms in a triangular optical lattice @xcite . in the thermodynamic limit
the anti - ferromagnetic phase is characterized by a standard staggered local order parameter whereas in the cluster phase one finds a gapped energy spectrum @xcite with diverging localizable entanglement due to a non - vanishing string order parameter @xcite . in summary , this system is interesting since it can be experimentally put to reality , it is fully analytically solvable , has a quantum phase transition from an anti - ferromagnetic phase to a symmetry protected topological phase @xcite and provides a physical platform for quantum computation @xcite . plotted for the three central spins in the chain as function of the parameter @xmath1 for different lengths @xmath2 of the spin chain .
the different curves correspond to : red dotted line @xmath3 ; blue dashed line @xmath4 ; black dot - dashed line @xmath5 ; green solid line @xmath6 .
the vertical grid line indicate the critical value @xmath7 . in the inset
we plot the behavior of @xmath8 , i.e. the value @xmath1 for which the measures becomes nonzero , as function of the the length of the chain.,width=283 ] we start our analysis by introducing the hamiltonian and the definition of genuine multipartite entanglement .
we find that the reduced density matrix of three adjacent spins obtained from the symmetry preserving ground state of the cluster - ising model holds the property that it can be turn always in the @xmath9-form @xcite .
the @xmath9-form allows to compute analytically the exact amount of the _ genuine multipartite entanglement _
@xcite which is in this case identically to the general criteria for detecting different types of multipartite entanglement introduced in refs .
@xcite .
we then prove that a non vanishing value of such a measure of genuine tripartite entanglement characterizes as well the anti - ferromagnetic phase as the cluster phase close to the quantum phase transition ( except for the factorization point @xcite ) .
in particular , we prove that this measure captures all critical properties known as the universality property , i.e. the behaviour at the critical point is independent of the finite size scaling . in a second step
we extend our analysis to ground states breaking the @xmath10 symmetry of the hamiltonian in the anti - ferromagnetic phase .
we find that even in this case the reduced density matrix posses the same value of genuine tripartite entanglement ,
i.e. symmetry breaking does not affect the amount of entanglement .
last but not least we consider different subsets and analyze them according to their genuine multipartite entanglement content .
our results suggest that the genuine tripartite entanglement between three adjacent spins is the only source of genuine multipartite entanglement in this model . to set the stage for our analysis let us introduce the one dimension cluster - ising model , that is characterized by the interplay between a three - body cluster like interactions and a two - body anti - ferromagnetic ising term @xmath11 the number of spins in the chain are
@xmath12 and @xmath1 is the relative weight between the two different interactions and may assume any values in the interval @xmath13 $ ] .
@xmath14 are the standard spin-@xmath15 operators acting on the @xmath16-th site .
we point out that all the evaluation made for finite size are done assuming open boundary conditions and @xmath2 equal to an integer multiple of @xmath17 .
regardless its apparent complexity the cluster - ising model can be analytically solved making use of the jordan - wigner transformation that brings spin operators in fermionic ones @xcite . with this transformation
the spin model is mapped into a bilinear fermionic problem that can be analytically solved both for a finite size system and in thermodynamic limit ( see appendix [ suppmaterial ] ) . thanks to the existence of such analytical solutions , the phase diagram of the one dimensional cluster - ising model in the thermodynamic limit can be computed @xcite . for @xmath18
the system is in an anti - ferromagnetic phase characterized by a twofold degenerate ground state with a non vanishing staggered local order parameter along the @xmath19 axis @xmath20 . on the contrary when @xmath21 the system is in the so called _ cluster phase _ with a four fold degenerate ground state that is characterized by a highly non local string order parameter .
, according to the procedure described in ref
. @xcite .
the different colors represent different data that goes from @xmath22 to @xmath23 spins .
the critical exponent is taken equal to @xmath24 . in the inset
we plot the behavior of @xmath25 close to @xmath26 for chains of different lengths as function of @xmath1 : red dot - dot - dashed line @xmath3 ; blue dashed line @xmath4 ; black dotted line @xmath27 ; green dot - dashed line @xmath28 , orange solid line @xmath6.,width=283 ] the first remarkable result concerns the local entanglement properties in the cluster - ising chain @xcite . for any two spins taken out of the chain the entanglement is zero for all @xmath1 ( e.g. , computed via hill - wootters concurrence @xcite which is a necessary and sufficient measure of entanglement of bipartite spin @xmath0 systems ) .
clearly , the vanishing of any pairwise entanglement disables the characterizing of the phase transition or the critical properties of the complex system .
this result , together with the fact that the residual tangle is always fixed to its maximum value for the ground state that satisfies the symmetry of the hamiltonian and never drops to zero also for the symmetry broken states @xcite ( except for @xmath29 ) , suggest that , in the cluster - ising model an important role must be played by the genuine multipartite entanglement .
let us now define _ genuine _ multipartite entanglement .
a state is called @xmath30-separable if and only if it can be written in the form @xmath31 with @xmath32 and @xmath33 and @xmath34 , where @xmath35 and each @xmath36 ( @xmath37 ) is living on a different and non - overlapped hilbert subspace .
note that biseparable states do not need to be separable via a specific partition since one sums up over different configurations .
states which are not biseparable ( @xmath38 ) are dubbed genuine multipartite entangled .
computing the reduced density for three spins with equal distances @xmath39 , @xmath40 ( @xmath41 ) , we found that for all values of @xmath1 these reduced density matrices are in the so called @xmath9-matrix form @xcite , i.e. only diagonal or off - diagonal elements are non - zero : @xmath42 in such a case it is possible to evaluate the genuine multipartite concurrence ( @xmath43 ) @xcite that is a faithful measure of the genuine @xmath44-partite tangle ( @xmath45 ) . in the presence of a density matrix in @xmath9 form the @xmath43 is given by @xcite @xmath46 let us first consider the state @xmath47 , a subset of three central spins @xmath48 from the thermal ground state that preserves all the symmetries of the hamiltonian , @xmath49\end{aligned}\ ] ] where @xmath50 , @xmath51 and @xmath52 are the fermionic correlation functions ( see appendix [ suppmaterial ] ) . in the case of a finite size system
the fermionic correlation functions are evaluated numerically using an algorithm that generalize the approach of ref .
@xcite . on the contrary in the thermodynamic limit they are given by @xcite @xmath53 where @xmath54^{\frac{1}{2}}\;.\ ]
] regardless the value of the fermionic correlation functions , applying the unitary operator @xmath55 turns @xmath47 into the @xmath9-form with the additional symmetry that @xmath56 with @xmath57 ( corresponding to the vanishing of @xmath58 ) . from eq .
( [ rho3 t ] ) it is straightforward to obtain the value of the genuine tripartite entanglement as @xmath59 which behavior as function of @xmath1 is plotted both in thermodynamic limit and for finite size systems in fig . [ plotgmcfuncalpha ] . independently , of the number of spins @xmath2 the behavior of the measure @xmath43 is similar and converges with increasing number fast to a nonzero value for the anti - ferromagnetic phase ( except the factorization point @xmath29 ) .
remarkably , it becomes non - zero already well before the critical point @xmath60 and also independently of @xmath2 .
in the inset of fig .
[ plotgmcfuncalpha ] we have plotted the dependence of the value of @xmath1 when it becomes non - zero in dependence of @xmath2 . for the thermodynamic limit we obtain @xmath61 , a value clearly below @xmath7 .
this rising of the genuine tripartite measure before the quantum phase transition plays the roles of a precursor of the quantum phase transition .
we can go one step further by considering the derivative @xmath62 and exploring its non trivial behavior close to the quantum critical point .
for the finite size case we find a maximum before the critical point that converges with increasing number of sizes @xmath2 to a maximum at the critical point ( see inset of fig . [ plotgmcmaxderalpha ] ) .
being more precise the maximum value of @xmath63 diverges logarithmically with @xmath2 as @xmath64 and in the thermodynamic limit the derivative @xmath65 diverges approaching the critical value @xmath66 :
@xmath67 according to the scaling ansatz @xcite , which validity is tested in fig .
[ plotgmcmaxderalpha ] , in the case of logarithmic singularities , the ratio between the two prefactors of the logarithm in eq .
( [ finitesizescaling1 ] ) and eq .
( [ finitesizescaling2 ] ) is the exponent @xmath68 that governs the divergence of the correlation length @xmath69 .
we find @xmath24 that is consistent with the characterization obtained from the analysis of the behavior of the correlation function . in the cluster phase ,
regardless the huge degeneracy of the ground space , there is no ground state that breaks the symmetry of the hamiltonian , in contrast to the anti - ferromagnetic phase .
let us superpose two ground states that each preserves in the anti - ferromagnetic phase all the symmetries of the hamiltonian , then we obtain a maximally symmetry broken ground state which reduced density matrix of three adjacent spins is given by @xmath70\ ] ] where @xmath71 and @xmath72 are the expectation values of @xmath73 and @xmath74 , respectively .
the subscript @xmath75 indicates that the average is evaluated on states breaking maximally the symmetry .
since the additional terms are only proportional to @xmath76 the @xmath9-form can be recovered using the same local unitary transformation as defined in eq .
( [ unitary ] ) .
moreover , as shown in the supplementary materials , @xmath77 holds and , consequently , the same value of the genuine tripartite measure is obtained .
this also holds for any ground state that breaks the @xmath10 symmetry .
hence , we conclude that @xmath43 does not depend on the choice of the ground state and , therefore , captures all the relevant properties of the system . detecting genuine multipartite entanglement @xcite if greater than zero is plotted as function of @xmath1 for the subsets of three spins @xmath78 ( blue dashed lines ) and the subsets of four spins @xmath79 ( red solid lines ) .
the inset shows @xmath80 for @xmath81 adjacent spins for @xmath82 .
no genuine multipartite entanglement is detected . ]
let us now investigate the entanglement properties of different subsets .
increasing the distance between the spins but still preserving the symmetry of the subset , i.e. considering the subset @xmath83 with @xmath84 we find that the reduced density matrices can be brought into the @xmath9-form but the measure vanishes .
if we increase the subset , e.g. considering the four spin subset @xmath79 the reduced density matrix can no longer be brought into the @xmath9-form .
this is due to the presence of non vanishing two body correlation function along the @xmath85 direction @xmath86 . in fig .
[ ploti2 ] we plotted the function @xmath80 introduced in refs .
@xcite , which is a lower bound on genuine multipartite entanglement , for three and four adjacent spins .
the heart of this approach are convex functions of the density matrix @xmath87 and permutations operators @xmath88 on subsets @xmath89 defined on copies of @xmath87 which are symmetries of a certain type of multipartite entanglement , i.e. @xmath90=0\longrightarrow \rho^{\otimes 2}=\mathcal{p}_\beta\rho^{\otimes 2}\mathcal{p}_\beta$ ] . if the symmetry is satisfied the following convex function is bounded by @xmath91 for @xmath92 some arbitrary states with dimension of @xmath87 and @xmath93 denotes a certain partition in the subset @xmath89 and @xmath30 defines the number of partitions one is interested in .
the states satisfying the above inequality for a given partition @xmath30 are just the @xmath30-separable states defined in eq .
( [ k - sep ] ) , hence , any positive values detects states that are not @xmath30-separable . fig .
[ ploti2 ] shows that no genuine multipartite entanglement is detected for any value of @xmath1 , moreover , the value is rather far away from the bound zero . in the inset of fig .
[ ploti2 ] we plotted the value of @xmath80 at the critical point for four to six adjacent spins .
we observe only a very small in crease of the value .
these results strongly suggest that there is no other type of genuine multipartite entanglement present than the genuine tripartite one of three adjacent spins . summarizing , in the present letter
we have provided the first evaluation of the genuine multipartite entanglement in the ground state of a complex quantum systems for which bipartite entanglement is zero .
we have determined analytically the genuine tripartite entanglement for a subset made by three adjacent spins in dependence on weight @xmath1 between the cluster and the anti - ferromagnetic interactions in a finite size system and in the thermodynamic limit .
we find for any size of the system that local genuine tripartite entanglement is non - zero before the critical point @xmath82 , i.e. genuine tripartite entanglement acts as a precursor of the phase transition .
moreover , we have shown that the non - trivial behaviour of the genuine tripartite entanglement measure characterizes the critical and scaling properties fully .
we have provided strong results that the genuine tripartite entanglement seems to be the only manifestation of genuine multipartite entanglement in this particular condensed matter system . in future perspective
, it would be desirable to extend such analysis to other condensed matter systems and to try to connect the presence of genuine multipartite entanglement both with the behavior of the entanglement spectrum @xcite and with scaling property of the quantum frustration @xcite .
_ acknowledgments _ - smg and bch acknowledge gratefully the austrian science fund ( fwf - p23627-n16 ) and fruitful discussions with m. huber and smg with o. ghne . 99 l. amico , r. fazio , a. osterloh , and v. vedral , rev .
mod .
phys . * 80 * , 517 ( 2008 ) .
m. n. barber in _ phase transitions and critical phenomena _ * vol . 8 * ( eds domb , c. & leibovitz , j. l. ) 146 ( academic , london , 1983 ) .
st .
schauer , m. huber , and b.c .
hiesmayr , phys .
rev .
a * 82 * , 062311 ( 2010 ) .
a. osterloh , l. amico , g. falci and r. fazio , nature * 416 * , 608 ( 2002 ) .
t. j. osborne and m.a .
nielsen , phys .
rev .
a * 66 * , 032110 ( 2002 ) .
g. vidal , j. i. latorre , e. rico , and a. kitaev , phys .
rev .
lett . * 90 * , 227902 ( 2003 ) .
j. i. latorre , e. rico , and g. vidal , quantum inf .
comput . * 4 * , 48 ( 2004 ) .
x. chen , z .- c . gu and x .- g .
wen , phys rev .
b * 82 * , 155138 ( 2010 ) .
t .- c wei , d. das , s. mukhopadyay , s. vishveshwara , and p. m. goldbart , phys .
rev .
a * 71 * , 060305(r ) ( 2005 ) .
s. m. giampaolo , g. adesso and f. illuminati , phys .
rev .
lett . * 100 * , 197201 ( 2008 ) .
s. m. giampaolo , g.adesso and f. illuminati , phys .
rev .
b * 79 * , 224434 ( 2009 ) .
s. m. giampaolo , g. adesso and f. illuminati , phys .
rev .
lett . * 104 * , 207202 ( 2010 ) .
s. m. giampaolo and b. c. hiesmayr , phys .
rev .
a * 88 * , 052305 ( 2013 ) .
m. hofmann , a. osterloh and o. ghne , `` _
_ scaling of genuine multiparticle entanglement at a quantum phase transition _ _ '' , arxiv:1309.2217 .
c.l . kane and e.j .
mele , phys .
rev .
lett * 95 * , 146802 ( 2005 ) .
g. e. volovik , `` _ the universe in a helium droplet _ '' , ( clarendon oxford ) ( 2003 ) .
d. bruss and c. macchiavello , phys .
rev .
a * 83 * , 052313 ( 2011 ) .
j. k. pachos and m. b. plenio , phys .
rev .
lett . * 93 * , 056402 ( 2004 ) . c. becker , p. soltan - panahi , j. kronjger , s. drscher , k. bongs and k. sengstock , new j. phys .
* 12 * , 065025 ( 2010 ) .
m. popp , f. verstraete , m. a. martn - delgado and j. i. cirac , phys .
rev .
a * 71 * , 042306 ( 2005 ) .
l.c .
venuti and m. roncaglia , phys .
rev .
lett . * 94 * , 207207 ( 2005 ) .
s .- p .
kou and x .- g .
wen , phys .
rev .
b * 80 * , 224406 ( 2009 )
. w. son , l. amico , r. fazio , a. hamma , s. pascazio and v. vedral , europhys . lett . * 95 * , 50001 ( 2011 ) .
p. smacchia , l. amico , p. facchi , r. fazio , g. florio , s. pascazio , and v. vedral , phys .
rev .
a * 84 * , 022304 ( 2011 ) .
h. j. briegel and r. raussendorf , phys .
rev .
lett . * 86 * , 910 ( 2001 ) . a. miyake , phys . rev
. lett . * 105 * , 040501 ( 2010 ) .
j. m. renes , a. miyake , g. k. brennen and s. d. bartlett , new j. phys . * 15 * , 025020 ( 2013 ) .
j .- y .
wu , m. rossi , h. kampermann , s. severini , l. chuan kwek , c. macchiavello and d. bruss , `` _ _ randomized graph states and their entanglement properties _ _ '' , arxiv:1403.3828 .
t. yu and j.h .
eberly , quantum inf .
comput .
* 7 * , 459 ( 2007 ) .
b.c .
hiesmayr and m. huber , phys .
rev .
a * 78 * , 012342 ( 2008 ) .
z .- h .
ma , z .- h .
chen , j .- l .
chen , ch .
spengler , a. gabriel and m. huber , phys .
rev .
a * 83 * , 062325 ( 2011 ) .
s. m. hashemi rafsanjani , m. huber , c. j. broadbent , and j. h. eberly , phys .
rev .
a * 86 * , 062303 ( 2012 ) .
m. huber , f. mintert , a. gabriel , b.c .
hiesmayr , phys .
rev .
lett . * 104 * , 210501 ( 2010 ) .
a. gabriel , b. c. hiesmayr and m. huber , quantum information and computation * 10 * , 829 ( 2010 ) .
p. jordan and e. wigner , z. phys . * 47 * , 631 ( 1928 ) .
s. hill and w. k. wootters , phys .
rev .
lett . * 78 * , 5022 ( 1997 ) .
w. k. wootters , phys .
rev .
lett . * 80 * 2245 ( 1998 ) . v. coffman , j. kundu and w. k. wootters , phys .
rev .
a * 61 * , 052306 ( 2000 ) .
t.j .
osborne and f. verstraete , phys .
rev .
lett . * 96 * , 220503 ( 2006 ) .
e. lieb , t. schultz and d. mattis , annals of physics * 16 * , 407 , ( 1961 ) .
s. m. giampaolo , s. montangero , f. dellanno , s. de siena and f. illuminati , phys . rev .
b * 88 * , 125142 ( 2013 ) .
s. m. giampaolo , g. gualdi , a. monras , f. illuminati , phys .
rev .
lett . * 107 * , 260602 ( 2011 ) .
u. marzolino , s. m. giampaolo , and f. illuminati , phys .
rev .
a * 88 * , 020301(r ) ( 2013 ) .
g. c. wick , phys . rev . *
80 * , 268 ( 1950 ) .
e. barouch and b. mccoy , phys .
rev .
a. * 3 * , 786 ( 1971 ) . |
it is known that properties of quantum spin systems tend to approach those of their corresponding classical spin systems as the spin magnitude increases . however , this is not necessarily the case for a sequence of models in which the number of multipolar - interaction terms increases as the spin magnitude increases .
consequently the higher - spin systems may have stronger quantum effects in this case .
in other words , such additional terms can break a classical correspondence down even in high dimensions .
these `` large - spin - magnitude '' systems mentioned above include systems in which one unit has more than two degrees of freedom , such as orbitally degenerate systems@xcite . in such systems ,
many coupling constants appear in general .
however , experimental information about multipolar couplings is limited . as a starting point to explore such systems , understanding of one of the extreme limits
must be useful .
therefore in this paper , we investigate an su(@xmath0 ) symmetric case @xcite in three dimensions , where @xmath0 is the total number of internal degrees of freedom , by means of high - temperature series expansion ( htse ) .
we consider a simple cubic lattice .
let each site take one of the @xmath0 colors denoted by @xmath8 with @xmath9 . using the hubbard operator @xmath10 , the exchange operator is expressed as @xmath11 colors of sites @xmath12 and
@xmath13 are exchanged when @xmath14 is applied . then , an su(@xmath0 ) symmetric hamiltonian reads @xmath15 where the summation is taken over all the nearest neighbor pairs .
we consider the antiferromagnetic case , @xmath16 .
let us show the relations with spin operators explicitly for some of the special cases below .
\(i ) when @xmath6 , this model is reduced to the ordinary heisenberg model with @xmath17 by relation @xmath18 .
\(ii ) the su(3 ) case corresponds to @xmath19 , competing quadratic and biquadratic exchange interaction,@xcite @xmath20 \(iii ) the su(4 ) heisenberg model is related to spin 3/2 systems but more often discussed in the context of orbital- and spin - degenerate systems@xcite . in particular , the model in three dimensions has been used as an effective model of ceb@xmath21 to explain magnetic - field dependence of the transition temperature of an antiferro - orbital ordering .
@xcite the four local states can be represented by @xmath22 $ ] , @xmath23 $ ] , @xmath24 $ ] , @xmath25 $ ] , where @xmath26 and @xmath27 $ ] represent an orbital state and a spin state , respectively .
the pseudo - spin operators are defined by @xmath28 , @xmath29 , @xmath30=\pm\frac12|\pm]$ ] , @xmath31=|\pm]$ ] , and the exchange operator for @xmath32 is rewritten as @xmath33 the pauli matrices @xmath34 and @xmath35 may simplify this expression , _
i.e. _ , @xmath36 note that there is a different representation of su(@xmath0 ) heisenberg model studied in detail using quantum monte carlo method ( qmc)@xcite .
however , the qmc for the hamiltonian ( [ hamil ] ) suffers from minus sign problems in more than one dimensions @xcite
if the number of competing order parameters is large and frustration exists , the transition temperature can greatly be reduced from the mean - field value even in three dimensions
. that will be the case with the hamiltonian ( [ hamil ] ) .
first , it is isotropic with respect to @xmath37 independent interacting components .
furthermore , it contains frustration as most clearly seen in eq .
( [ paulirep ] ) of the su(4 ) case .
each of the 15 components , @xmath38 , @xmath39 , @xmath40 , attempts an antiparallel alignment .
however , it can not be attained simultaneously , _
e.g. _ , simultaneous nel states of @xmath41 and @xmath42 produce a ferromagnetic alignment of the product @xmath43 .
such frustration is not so explicit in the su(3 ) case in eq .
( [ eq : su3rep ] ) as in the su(4 ) case , yet appearance of the minus - sign problem in the qmc indicates that similar frustration lies in it.@xcite this frustration should become stronger as @xmath0 increases .
one can see in one dimension how the system finds a compromise against this frustration .
the su(@xmath0 ) heisenberg model in one dimension can be exactly solved.@xcite an important point is that the ground state has a quasi@xmath0-site periodicity .
the su(4 ) case is numerically studied in more detail ; ground - state properties by the density matrix renormalization group@xcite , and thermodynamic properties by the qmc@xcite and the htse@xcite .
the real - space spin - spin correlation function as a function of spin - spin distance has a positive value every four sites .
its fourier transform has a outstanding cusp at @xmath44 and a small cusp at @xmath45 .
regarding the temperature dependence , as temperature decreases correlation with two - site periodicity develops , and then at lower temperature another correlation occurs with @xmath0-site periodicity .
@xcite such peculiar correlation in one dimension could suggest an exotic ordering in higher dimensions .
however , not much is known about the antiferromagnetic su(@xmath0 ) heisenberg model in three dimensions .
in fact , its ferromagnetic variant is studied in ref . by the htse for the uniform susceptibility that needs less effort to be calculated than susceptibility of other wave - numbers .
to our knowledge , our calculation in this paper is the first htse aiming at antiferromagnetic exchange of this model . in this study , we investigate the model systematically by changing parameter @xmath0 of su(@xmath0 ) , and report some unique features of spatial correlation at finite temperature .
namely , in sec . [ sec : corfunc ] we show that the behavior of correlation functions is similar to that in one dimension , and in sec .
[ sec : neeltem ] that the nel temperature disappears as @xmath0 increases from @xmath6 .
what should be calculated here is @xmath46 , which is related to a correlation function .
for example , when @xmath32 , fifteen components in eq .
( [ paulirep ] ) contribute equally , and thus @xmath47 . for general @xmath0
, we can define correlation function between sites @xmath12 and @xmath13 as @xmath48 for @xmath49 , which does not depend on @xmath50 nor @xmath51 because of the su(@xmath0 ) symmetry .
its fourier transform is denoted by @xmath1 .
the high - temperature expansion is performed by expanding the boltzmann factor @xmath52 in @xmath51 . in practice ,
the series coefficients in the thermodynamic limit are exactly obtained by a linked - cluster expansion.@xcite to obtain the series expansion of @xmath53 up to @xmath54 $ ] , we need to calculate @xmath55 $ ] and @xmath56 $ ] for @xmath57 , where @xmath58 is the hamiltonian in a linked cluster . in calculating the traces
, we use a property of the permutation operator , and calculation of the traces is reduced to a combinatorial problem to count the number of circular permutations in a product of permutations .
@xcite here , an important point of our analysis is that the series coefficients are obtained as polynomials of @xmath0 , for example , @xmath59 namely , the order of the series for every @xmath0 is the same .
we have obtained the series for @xmath60 up to @xmath61 $ ] for arbitrary @xmath62 , and consequently @xmath1 up to @xmath63 $ ] for arbitrary @xmath64 . as a special interest , @xmath5 is obtained up to @xmath65 $ ] .
the series are extrapolated using the pad approximation ( pa ) .
first of all , in order to see the temperature - dependent nature of the spatial correlation , we analyze the _ next_-nearest - neighbor correlation function @xmath66 as we have done in one dimension in ref . .
figure [ fig : nncor ] shows the results .
both the axes are scaled so that the high - temperature limit of every @xmath0 matches . here ,
we have simultaneously plotted extrapolation from several different choices of the pa , and the difference between them approximately represents an error of the extrapolation .
the lowest order of the series of @xmath60 has a nel - order - type correlation in any @xmath0 .
that is , the series of @xmath67 starts with @xmath68 $ ] with sign @xmath69 .
therefore , @xmath70 at high temperature .
however , as antiferromagnetic correlation of each interacting component becomes larger , each short - range order disturbs another because of the frustration .
the change of the sign of @xmath66 suggests that the correlation acquire a longer period at low temperature . for @xmath71 ,
it is clear that @xmath66 changes the sign at a low temperature that increases with @xmath0 in this scale . for smaller @xmath0
, the relevant temperature range is below the converged region , and it is difficult to conclude from this data .
such a change of correlation at low temperature also appears in the fourier transform of the correlation function . a naive extrapolation of the series for @xmath1 shows bad convergence for several @xmath64 .
this is probably because information at position @xmath72 is lacking when @xmath73 . in order to avoid it
, we extrapolate the series of a complex function @xmath74 , and after that take the real part of the extrapolated function . here
, the summation is taken for @xmath75 , @xmath76 , @xmath77 , and @xmath78 is two to the power of the number of nonzeros in @xmath79 .
this method also makes use of its imaginary part , and the convergence becomes better than the naive extrapolation . as we have seen in the next - nearest - neighbor correlation function , the analysis becomes easier as @xmath0 increases .
therefore , we show @xmath1 of the su(16 ) case in fig .
[ fig : sq ] for a diagonal direction in the @xmath64-space , @xmath80 . here , its high temperature limit , which does not depend on @xmath64 , is subtracted . as temperature decreases
, antiferromagnetic correlation develops and @xmath5 increases as the curve at @xmath81 shows .
however , with further decrease of temperature , @xmath5 starts decreasing and @xmath1 with other @xmath64 increases .
it is not very clear from fig .
[ fig : sq ] if the maximum of @xmath1 starts moving from @xmath82 .
hence , let us show another quantity . note that if the second derivative of @xmath1 at @xmath4 changes sign , the position of the maximum clearly moves .
its temperature dependence for the su(16 ) model is plotted in fig .
[ fig : ddsqn16 ] .
it shows a change of the sign around @xmath83 .
in fact , this temperature of changing sign seems to hardly depend on @xmath0 of su(@xmath0 ) .
the result above suggests that the nel order disappear at least in the su(@xmath0 ) model with large @xmath0 .
namely , order with wave number @xmath84 , or disorder , should appear .
then , the next question is , at which @xmath0 the nel order disappears .
it is known that the su(2 ) heisenberg model has the nel order at low temperature . therefore in this section
, we gradually increases @xmath0 of su(@xmath0 ) from @xmath6 . as a preparation for that
, we find a reliable way to estimate a transition temperature first , using the su(2 ) model .
the nel temperature @xmath85 can be characterized by divergence of @xmath5 , namely , by a singularity of @xmath5 as a function of @xmath3 . in order to analyze such a singularity
, we use so - called d - log - pad approximation ( dlpa ) , _ i.e. _ , the pa for the logarithmic derivative . in using the dlpa , transformation of the expansion variable may improve the convergence of extrapolation . we choose a transformation@xcite @xmath86 where @xmath87 is an adjustable parameter that improves convergence . in fact , the singularity closest to the origin in the original series is near the imaginary axis , while @xmath85 corresponds to a singularity on the real axis . with this transformation ,
singularities on the real axis approach to the origin and those near the imaginary axis go away . since the dlpa can estimate a position of the nearest singularity most accurately
, errors of the dlpa can become smaller by this transformation . in order to find an optimal @xmath87
, we calculate @xmath85 as a function of @xmath87 for a couple of choices of the dlpa , and we adopt @xmath87 at which difference among different dlpas is the smallest .
figure [ fig : valtran ] shows @xmath85 as a function of @xmath87 for three different choices of the dlpa . here ,
@xmath88 $ ] denotes the dlpa with a polynomial of order @xmath89 over a polynomial of order @xmath0 . since [ 4/4 ]
is from the series one - order lower than the others , we choose @xmath87 at which [ 5/4 ] and [ 4/5 ] are the closest , namely , @xmath90 .
then , @xmath91 obtained here is 1.898 , which is close to those in the literature , @xmath92 by the qmc@xcite and @xmath93 by the htse@xcite .
in addition , we have obtained a critical exponent simultaneously .
we assume that the critical exponent does not depend on the spin magnitude , and compare it with that of the _ classical _ antiferromagnetic heisenberg model , in which @xmath5 is identical to the staggered susceptibility equivalent to the ` uniform susceptibility of the ferromagnetic model ' .
hence we can compare the critical exponent @xmath94 .
the estimation from our calculation above is @xmath95 , which is close to 1.396 by a monte carlo method@xcite , 1.406 by htse@xcite , and 1.388 by a field theory@xcite , of the classical o(3 ) model .
therefore , we trust this way of analyzing @xmath85 , and use it also for the su(@xmath0 ) model with @xmath7 . also for @xmath7 , the variable transformation ( [ eq : trans ] ) should work because the singularity closest to the origin in the original series is near the imaginary axis .
originally @xmath0 is an integer because it is the number of internal degrees of freedom .
however , after obtaining the series as an analytic function of @xmath0 , we can regard @xmath0 as a continuous variable that has a physical meaning only when it happens to take integer values .
in fact , as @xmath0 changes continuously , the properties of the series change also continuously . therefore , we gradually increases @xmath0 from @xmath6 and see @xmath0-dependence of @xmath85 .
figure [ fig : ndep ] shows the results . here
, we fix @xmath90 , and we plot [ 5/4 ] and [ 4/5 ] together .
the optimal @xmath87 may depend on @xmath0 .
however , since the difference between the two curves is very small , we regard it as an optimal @xmath87 . as @xmath0 increases
, @xmath85 decreases almost linearly , and at @xmath96 the singularity corresponding to @xmath85 runs away from the real axis to a complex value with a finite imaginary part . for larger @xmath0
, we do not find any singularity on the antiferromagnetic side of the real axis .
therefore , we conclude that the value of @xmath0 at which nel order disappears lies in the range @xmath97 . in addition , we have also analyzed @xmath1 with @xmath84 .
however , we have not found any symptom of ordering with @xmath84 in the temperature region that we can reach by the present order of the series .
in summary , we have performed high - temperature series expansions for the su(@xmath0 ) heisenberg model in three dimensions with arbitrary @xmath0 .
first of all , we have calculated a _
next_-nearest - neighbor correlation function .
at least at large @xmath0 , it changes the sign at low temperature , which suggests that the correlation should not be like nel order , but have a longer period .
analysis of the fourier transform of the correlation function also supports that the ground state of the large-@xmath0 su(@xmath0 ) heisenberg model does not have the nel - order correlation .
then , we have turned to an approach from @xmath6 .
since the su(2 ) heisenberg model has the nel order , it should disappear at a certain @xmath0 .
we have first found a reliable way to estimate a transition temperature by analyzing the divergence of the @xmath82 component of the correlation function .
next , we generalize @xmath0 to a continuous variable and increase @xmath0 gradually from @xmath6 .
we have concluded that the nel ordering disappears for @xmath7 .
the author would like to thank y. kuramoto for directing him to this topic and for stimulating discussions .
he also appreciates useful comments on the manuscript from a. honecker .
this work was supported in part by the deutsche forschungsgemeinschaft ( dfg ) .
early stages of this work were supported by the japan society for the promotion of science and by the max - planck institute for the physics of complex systems .
different stages of this work were supported by the dfg and the technical university braunschweig .
parts of the numerical calculations were performed on the cfgauss at the computing centers of the tu braunschweig .
uimin , jetp lett .
* 12 * , 225 ( 1970 ) .
b. sutherland , phys .
b * 12 * , 3795 ( 1975 ) .
a. fujii , a. klmper , nucl .
b546 * , 751 ( 1999 ) .
i. affleck , nucl . phys . *
b265 * , 409 ( 1986 ) .
n. kawakami , phys .
b * 46 * , 3191 ( 1992 )
. m. t. batchelor , x .- w .
guan , n. oelkers , k. sakai , z. tsuboi , a. foerster , phys .
lett . * 91 * , 217202 ( 2003 ) .
y. yamashita , n. shibata and k. ueda , phys .
b * 58 * , 9114 ( 1998 )
. b. frischmuth , f. mila and m. troyer , phys .
* 82 * , 835 ( 1999 ) . n. fukushima and y. kuramoto , j. phys .
. jpn . * 71 * , 1238 ( 2002 ) .
n. fukushima , j. stat .
phys . * 111 * , 1049 ( 2003 ) .
y. q. li , m. ma , d. n. shi , and f. c. zhang , phys .
* 81 * , 3527 ( 1998 ) .
m. van den bossche , p. azaria , p. lecheminant , f. mila , phys .
86 * , 4124 ( 2001 ) .
m. van den bossche , f
.- c . zhang , and f. mila , eur .
j. b * 17 * , 367 - 370 ( 2000 ) .
a. mishra , m. ma , and f. c. zhang , phys .
b * 65 * , 214411 ( 2002 ) . s. q. shen , phys .
b * 66 * , 214516 ( 2002 ) .
k. penc , m. mambrini , p. fazekas , f. mila , phys .
b * 68 * , 12408 ( 2003 ) .
f. j. ohkawa : j. phys .
54 * , 3909 ( 1985 ) .
r. shiina , h. shiba and p. thalmeier , j. phys .
* 66 * , 1741 ( 1997 ) .
h. h. chen and r. k. joseph , j. math .
* 13 * , 725 ( 1972 ) .
pan , phys .
b * 59 * , 1168 ( 1999 ) .
a. w. sandvik , phys .
lett . * 80 * , 5196 ( 1998 ) .
j. oitmaa and w. zheng , preprint ( cond - mat/0409041 ) .
m. campostrini , m. hasenbusch , a. pelissetto , p. rossi , e. vicari , phys .
b * 65 * , 144520 ( 2002 ) . p. butera and m. comi , phys .
b * 56 * , 8212 ( 1997 ) .
f. jasch and h. kleinert , j. math . phys .
* 42 * , 52 ( 2001 ) . | the su(@xmath0 ) heisenberg model represented by exchange operators is studied by means of high - temperature series expansion in three dimensions , where @xmath0 is an arbitrary positive integer . the spin - spin correlation function and its fourier transform @xmath1
is derived up to @xmath2 $ ] with @xmath3 being the nearest - neighbor antiferromagnetic exchange in units of temperature .
the temperature dependence of @xmath1 and next - nearest - neighbor spin - spin correlation in the large @xmath0 cases show that dominant correlation deviates from @xmath4 at low temperature , which is qualitatively similar to that of this model in one dimension .
the nel temperature of su(2 ) case is precisely estimated by analyzing the divergence of @xmath5 .
then , we generalize @xmath0 of su(@xmath0 ) to a continuous variable and gradually increases from @xmath6 .
we conclude that the nel ordering disappears for @xmath7 . |
elastic x - ray scattering confirmed @xcite that the distortion in the i phase is incommensurate .
the deviation from dimerization @xmath16 increases with increasing magnetic field .
this can be illustrated in the unfrustrated xy model which corresponds via the jordan - wigner transformation to free fermions .
the susceptibility towards distortion becomes maximum at @xmath17 since @xmath17 allows to create particle - hole pairs of vanishing energy .
a distortion with @xmath18 is formed where @xmath19 is the magnetization .
since this distortion couples the degenerate states at @xmath20 and at @xmath21 a gap appears there .
let us now gradually increase the interaction corresponding to the @xmath22 terms at fixed particle number . for continuity no state can cross the gap such that the picture of a distortion at @xmath23 remains valid @xcite .
so we have @xmath24 given by ( [ qm - bedg ] ) @xcite ) @xmath25 { \bf s}_i { \bf s}_{i+1 } \ .\ ] ] the distribution of the local magnetizations @xmath26 is found from nmr experiments @xcite . with some success ,
the experimental data were compared to a continuum theory @xcite .
this theory , however , is based on a hartree - fock treatment where all hartree and fock terms are spatially constant . in ref .
@xcite it is shown that this is a too crude approximation reducing the physics to the one of a xy chain .
the antiferromagnetic correlations found in this way are much smaller than those of an isotropic xyz chain .
this conclusion is corroborated by several works @xcite .
but the spin isotropy of cuprates can hardly be questioned . to account for the smaller amplitudes it is proposed @xcite that experimentally only an effective magnetization @xmath27
is seen which is an average @xmath28 the results for @xmath29 agree well with experiment @xcite .
the microscopic origin of the average is discussed in sect .
the reason for strong local magnetizations around the zeros of the modulation ( cf . fig . [ inkomm ] ) is found in the localization of a spinon .
each zero binds exactly one spinon @xcite .
summing the @xmath30 around a magnetization maximum yields 1/2 .
the order of the transition d @xmath31 i can be determined by investigating the ground state energy @xmath32 as function of the average magnetization @xmath19 @xcite . by means of a legendre transformation @xmath33 ,
one obtains the dependence of ground state energy @xmath34 on the magnetic field @xmath35 .
it is found that a discontinuous jump for @xmath36 occurs .
this implies that the transition d @xmath31
i for fixed sinusoidal modulation is of first order .
the mean square of @xmath37 jumps discontinuously from 1 to 1/2 if @xmath38 deviates infinitesimaly from @xmath39 @xcite since the @xmath40 are summed over integer values only .
experimentally , however , the observed first order jumps are much lower than those found for fixed sinusoidal modulation @xcite .
since it was stated above that sinusoidal modulation alone does not account for the weak first order d @xmath31 i transition we turn to the full minimization of the ground state energy of ( [ hamilton ] ) .
derivation with respect to @xmath5 yields @xmath41 where @xmath42 stands for the expectation value and the average along the chain .
the double - bracketed term accounts for the constraint of the vanishing average of the @xmath5 .
the minimization is done iteratively @xcite .
the generic result is depicted in fig .
[ inkomm ] .
the local magnetizations do not display major differences to the results for sinusoidal modulation in @xcite because the spinon localization is in essence determined only by the slope with which the modulation vanishes .
note that the envelope of @xmath30 is proportional to the probability of finding a spinon at that site @xcite .
hence , the magnetic part of the soliton displays localization as for sinusoidal modulation . for the distortions
the relation ( [ bondmin ] ) implies that the deviations from constantly alternating dimerization are also localized .
the distortion belonging to an isolated soliton is a kink , i.e. the distortion between two solitons resembles the one in the d phase .
this implies a crucial advantage over the sinusoidal modulation
. for kink - like solitons the reduction of the mean square distortion is proportional to the soliton number .
hence , a low soliton concentration leads only to a small change of the energy such that @xmath32 is continuous ( in the sense of lipschitz ) on @xmath36 @xcite . the investigations in ref .
@xcite of the model ( [ hamilton ] ) yielded a continuous phase transition even though the magnetization grows very quickly above the critical field @xmath43 .
most of the results of the continuum theories comply also with a phase transition of second order @xcite .
solely horovitz mentions the possibility of soliton attraction in an early work @xcite implying a first order transition .
_ expect a first order transition at @xmath44 @xcite .
in fact , the details of the models matter .
cross @xcite argued already that an elastic energy _ with _
dispersion @xmath45 being minimum at @xmath46 leads to a first order transition .
the positive curvature of @xmath45 around @xmath46 suppresses higher harmonics in the distortion .
hence , sinusoidal modulation is favoured .
the concomitant concavity in @xmath32 at low magnetization @xmath19 implies phase separation via the maxwell construction .
so the phase transition is first order @xcite .
this finding is in accordance with the conclusion from a phenomenological ginzburg - landau description @xcite that the d @xmath31 i transition is generically of first order . at the transition
the distance between two solitons is rather large so that the difference between sinusoidal and adaptive modulation matters most . at higher soliton concentrations
the adaptive modulation becomes more and more sinusoidal but with a concentration dependent amplitude .
the mean square distortion could be determined from the elastic lattice constants .
these experimental results agree well with the predictions based on the model ( [ hamilton ] ) , see @xcite .
the continuum theories applying to the isotropic heisenberg chain @xcite provide the following results ( details in ref .
@xcite ; @xmath47 : elliptic jacobi functions ) @xmath48 the fits in fig .
[ inkomm ] are based on eqs .
( [ fit1a],[fit1b ] ) . identity ( [ zus1 ] ) is _ not _ complied with , see sect .
3 . otherwise no agreement would be obtained .
relation ( [ zus2 ] ) is imposed on the fits whereas eq .
( [ zus3 ] ) serves as check .
it is fulfilled within 4% .
the fact that @xmath49 is considerably above unity complies nicely with the experimental findings .
elastic x - ray scattering @xcite found @xmath50 while the nmr investigations provide @xmath51 @xcite close to the transition .
there is also another way to introduce solitons in a spin - peierls system than the application of magnetic field .
doping non - magnetic impurities in a spin - peierls systems cuts the infinite chains into finite chain segments @xcite . in a number of works
@xcite it has been shown that each impurity frees one spinon which is situated either before or after the impurity on the chain . assuming a fixed dimerization it is easy to see that the spinon is bound to its generating impurity @xcite in accordance with experimental results @xcite .
but in a spin - peierls system the change of the modulation has to be taken into account , too .
this is done by introducing @xmath52\ , \ ] ] such that the impurity is at site -1 . the important amendment compared to @xmath53 in eq .
( [ hamilton ] ) is the last term . if the spinon moves away from the impurity the distortion pattern is changed between impurity and spinon . due to
an _ elastic _ interchain interaction ( parametrized by @xmath54 ) a coherent distortion pattern throughout the whole three - dimensional system is preferred .
a deviation from this pattern is energetically unfavourable .
so one is led to include the last term in eq .
( [ stoerkopp ] ) assuming that the adjacent chains are dimerized as in the unperturbed d phase .
the relation @xmath55 ensures the consistency of the distortion amplitude with its bulk value .
[ impur ] displays the results for various elastic interchain interactions .
the soliton is _ not _ localized at the impurity but at a certain distance . for lower @xmath54 values the soliton resembles very much the ones in the i phase ( cf .
[ inkomm ] ) . on increasing @xmath54 the soliton is squeezed more and more towards the impurity .
analogous results can be found by qmc @xcite , too .
sure enough , the soliton is bound to the impurity confirming previous ideas @xcite .
moreover , the first excitations for the same distortions and in the same spin sector are found below the singlet - triplet gap at 52% of @xmath14 for @xmath56 and at 64% for @xmath57 .
this matters for the spectroscopic analysis .
most remarkable is the difference between the distortive soliton width @xmath2 and the magnetic soliton width @xmath3 in the numerical results ( cf . fig .
[ inkomm ] ) .
it amounts to 30% at @xmath58 depending mainly on the frustration for low soliton concentrations @xcite .
the challenge is to extend the existing continuum description to account for this fact .
let us revisit the semiclassical treatment of the bosonized description of the spin - peierls problem @xcite . minimizing the total energy the variation of the distortion @xmath59 leads to @xmath60 where @xmath61 denotes the renormalizing fluctuations of the local bosonic field @xmath62 about the classical field @xmath63 .
a soliton corresponds to a solution where @xmath63 increases by @xmath39 in a kink - like fashion . if @xmath64 is assumed to take the spatially constant value that it has in the ground state @xcite one obtains @xmath65 wherein @xmath66 is given by the ratio @xmath67 of the spin wave velocity and the gap .
the alternating component of the magnetization @xmath68 is proportional to @xmath69 .
hence , one has @xmath70 .
but the presence of the soliton induces a deviation @xmath71 from its ground state value .
[ locren ] displays a generic result for this deviation .
it is calculated on top of the solution of nakano and fukuyama @xcite .
the alternating component @xmath72 is thus indeed narrower than before due to the spatial dependence of the renormalization factor .
the result in fig .
[ locren ] is only a first step since the influence of the altered magnetic behaviour is not included . yet
it is clear that the renormalization is a local quantity and that this is the origin of the difference between @xmath2 and @xmath3 .
in sect . 1 the discrepancy between theoretical and experimental amplitudes of the alternating local magnetizations was solved by an averaging procedure ( [ mittel ] ) . passing to adaptive modulations does not change the amplitudes much @xcite .
hence , one still has to find the microscopic origin for the averaging .
non - adiabatic effects are in fact responsible .
the soliton lattice oscillates about its equilibrium positions .
these oscillations are best understood in a continuum description which is well justified if the typical length @xmath66 is noticeably larger than the lattice constant .
in such a continuum description the modulation @xmath59 can be shifted along the chains without energy cost .
this continuous translational invariance is spontaneously broken by the soliton lattice .
hence , there are massless goldstone modes , the so - called phasons .
they are analogous to the phonons of a crystal lattice except that they do not have three branches but only with one .
while the atoms of a crystal lattice can be shifted in all three spatial directions the solitons can be shifted only along the chains .
albeit the ideal spin - peierls system is magnetically one - dimensional the distortions on different chains are elastically coupled .
thus , the phasons are governed by a 3d , though anisotropic , dispersion .
the dispersion parameters are determined from the anisotropy of the correlation lengths assuming a ginzburg - landau description @xcite .
the corresponding @xmath73 term in the specific heat has been measured @xcite and theory and experiment agree astonishingly well @xcite . the zero point motion and the excited motion of phasons lead to the averaging ( [ mittel ] ) .
let us denote the adiabatic result for the local magnetizations by @xmath74 where @xmath40 is the component along the chains , @xmath75 the alternating component and @xmath76 the uniform component of the magnetizations .
a local shift can be implemented by replacing @xmath77 where @xmath78 denotes a phase shift operator . on the long time scales of a nmr measurement one
measures @xmath79 with @xmath80 .
so there is an amplitude reduction engendered by the local fluctuations .
the factor @xmath81 is comparable to a debye - waller factor . using the values fixed previously @xcite yields @xmath82 with @xmath83k @xcite .
( [ mittel ] ) is retrieved by estimating @xmath84 and @xmath85 from the discrete values @xmath30 by @xmath86 and by @xmath87 . inserting these formulae into eq .
( [ mi - measured ] ) yields eq .
( [ mittel ] ) with @xmath88 . at @xmath89
, @xmath90 takes the value @xmath91 in accordance with experiment @xcite .
in this report the modulated phases of spin - peierls systems were discussed .
such modulations are induced either by magnetic field or by impurities . in both ways
the singlet pairing in the d phase is broken and spinons are freed .
the lattice distortion adapts to the spinon by forming a zero to which the spinon is bound .
this new entity constitutes the spin - peierls soliton . the order of the transition d
@xmath31 i phase on increasing field depends on model details .
imposed sinusoidal modulation leads to a pronounced first order transition .
allowing the system to choose an optimum modulation makes the transition continuous _ if _ the elastic energy is wave vector independent
. if the elastic energy itself pins the modulation to @xmath39 a weak first order transition is found .
doping induced solitons are bound to their generating impurity . the distortion pattern between impurity and soliton is not coherent with the bulk pattern .
this costs energy which acts as a confining potential .
binding occurs for which experimental evidence exists @xcite . the difference between magnetic and distortive soliton width could be traced back to the so far neglected spatial dependence of the renormalizing local fluctuations .
a fully self consistent analysis is in progress .
the reduction of the alternating magnetic amplitude due to phasons provides striking evidence for the importance of the lattice dynamics .
the inclusion of non - adiabatic effects on top of an otherwise adiabatic calculation might still be unsatisfactory .
so other approaches to non - adiabatic behaviour should be extended to the i phase @xcite .
i thank c. berthier , j.p .
boucher , b. bchner , t. lorenz , m. horvati , th .
nattermann for fruitful discussions , f. schnfeld for reliable numerical work , e. mller - hartmann for generous support and g. gntherodt s group for intense collaboration
. financial support of the dfg by the sfb 341 is acknowledged .
j. w. bray , l. v. interrante , i. c. jacobs , j. c. bonner , in _ extended linear chain compounds _ , by j. s. miller ( plenum press , new york , 1983 ) , vol . 3 , p. 353 ; j. p. boucher , l. p. regnault , j. phys .
i france * 6 * , 1939 ( 1996 ) .
g. b. martins , e. dagotto , j. a. riera , phys .
rev . b*54 * , 16032 ( 1996 ) ; g. b. martins , m. laukamp , e. dagotto , j. a. riera , phys .
. lett . * 78 * , 3563 ( 1997 ) ; m. laukamp _
et al . _ ,
rev . b*57 * , 10755 ( 1998 ) . | lattice modulations in the high magnetic field phase and close to impurities in spin - peierls systems are considered and compared to experiment .
necessary extensions of existing theories are proposed .
the influence of zero - point fluctuations on magnetic amplitudes is shown . the discovery of the first inorganic spin - peierls substance cugeo@xmath0 by hase _ et al . _
@xcite renewed the interest in the phases of spin - peierls substances ( for reviews , see @xcite ) .
the constituting element of these substances are quasi one - dimensional spin chains coupled to the lattice degrees of freedom .
the incommensurably modulated ( i ) phase is particularly interesting since this phase is the most complex one from the theoretical and from the experimental point of view .
already 20 years ago much work was devoted to it ( e.g. [ 3 - 9 ] ) . yet , detailed experimental investigations of the nature of this phase were not possible at that time .
this has changed for cugeo@xmath0 .
x - ray experiments in the i phase permitted to detect the incommensurability of the lattice in @xmath1-space @xcite .
the structure of the lattice modulation was investigated by measuring the intensity of the third harmonic @xcite .
the distribution of local magnetizations in cugeo@xmath0 was measured by nmr @xcite .
recently , the shape and the amplitude of the magnetic part of a soliton were deduced @xcite . here
we will review several aspects of the theoretical description of the i phase .
special emphasis is put on the discrepancies between the so far existing theories and the experimental findings and to resolving these discrepancies @xcite .
the term `` soliton '' will be used for the combination of a zero in the modulated distortion _ and _ the concomitant localized spinon @xcite .
the distortive soliton width @xmath2 is the width of the kink - like modulation .
the magnetic soliton width @xmath3 is the spatial width of the local magnetizations @xcite .
the incommensurate modulation in the i phase is viewed as an equidistant array ( lattice ) of solitons . the schematic phase diagram in fig . [ phasdia
] shall serve as a guide for the following discussion . for the sake of concreteness we consider the hamiltonian @xmath4 + \frac{k}{2 } \delta_i^2\ ] ] which neglects the phonon dynamics ( adiabatic treatment ) since the distortions @xmath5 are numbers of which the average is zero .
the nearest neighbour coupling @xmath6 is modulated by the @xmath5 .
frustrating next - nearest neighbour coupling @xmath7 is also considered . the spring constant @xmath8 measures the elastic energy of a given distortion . without magnetic field
the spin chains are very susceptible towards dimerization @xmath9 .
the energy lowering due to dimerization is anomalously large @xmath10 overcompensating the energy loss in elastic energy @xmath11 @xcite .
applying a magnetic field to the dimerized ( d ) phase of a spin - peierls system ( @xmath12 small ) leaves the d phase unchanged until the critical field @xmath13 is reached .
the stability stems from the singlet - triplet gap @xmath14 .
hence an upper bound for @xmath13 is @xmath15 .
in fact , the critical field @xmath13 is smaller .
its zeeman energy represents only 80% of the gap @xmath14 because the modulation also changes above the critical field .
the system enters the incommensurably modulated phase . |
in this note we present two injective proofs of a strengthening of a conjecture of simion @xcite . to describe the result ,
let @xmath0 be the ferrers diagram of a partition viewed as a set of squares in english notation .
( see any of the texts @xcite for definitions of terms that we do not define here . ) the shape @xmath1 will be fixed for the rest of this paper .
consider a grid with the vertices labeled @xmath2 for @xmath3 as in figure [ rgrid ] .
place @xmath1 in the northwest corner of this array so that its squares coincide with those of the grid .
a _ northeastern lattice path _ is a lattice path on the grid in which each step goes one unit to the north or one unit to the east . let @xmath4 be the number of northeastern lattice paths from @xmath5 to @xmath6 that _ do not go inside @xmath1 _ ( although they may touch its boundary ) , and let @xmath7 be the set of such paths . in particular , @xmath8 if either the starting or ending point is inside @xmath1 .
simion @xcite conjectured that for all @xmath9 the sequence @xmath10 is unimodal .
lattice path techniques for proving unimodality were investigated by sagan @xcite , but the conjecture remained open at that point .
recently , hildebrand @xcite proved the stronger result that this sequence is actually log concave by mostly algebraic means . shortly thereafter , wang @xcite simplified hildebrand s proof using results about polya frequency sequences . in the present work , we will give two injective proofs of the stong version of simion s conjecture .
the one in sections [ seceasy ] and [ sechard ] employs ideas from hildebrand s proof while the one in section [ direct ] is more direct .
our injections come from a method of lindstrm @xcite , later popularized by gessel and viennot @xcite , that can be used to prove total positivity results for matrices . for an exposition ,
see sagan s book @xcite .
bna @xcite has used related ideas to prove the log concavity of a sequence counting @xmath11-stack sortable permutations .
we end this section by reiterating the statement of the main theorem for easy reference .
notice that when @xmath12 it specializes to the well - known result that the rows of pascal s triangle are log concave .
[ main ] let @xmath1 be the ferrers diagram of a partition and let @xmath4 be the number of northeastern lattice paths in the grid from @xmath5 to @xmath6 which do not intersect the interior of @xmath1 .
then for all @xmath9 the sequence @xmath10 is log concave .
this preliminary part of the first proof is from @xcite . we include it so that our exposition will be self contained .
we need to prove that for all @xmath13 we have @xmath14 to prove this , it suffices to show that @xmath15 because then , by symmetry , we also have @xmath16 now multiplying the last two equations together and simplifying gives the first .
the second inequality can be proved by demonstrating another pair of equations , namely @xmath17 and @xmath18 multiplying these two equations together and cancelling gives the desired result .
in this section we prove that ( [ easy ] ) holds by constructing an injection @xmath19 consider a path pair @xmath20 . then @xmath21 and @xmath22 must intersect .
let @xmath23 be their first ( most southwestern ) intersection point .
say that @xmath23 splits @xmath21 into parts @xmath24 and @xmath25 , and splits @xmath22 into parts @xmath26 and @xmath27 .
then the concatenation of @xmath24 and @xmath27 is a path in @xmath7 , and the concatenation of @xmath26 and @xmath25 is a path in @xmath28 .
so define @xmath29 .
it is easy to see that the image of @xmath30 is exactly all @xmath31 such that @xmath32 and @xmath33 intersect .
it is also simple to verify that if @xmath34 , then applying the same algorithm to @xmath35 recovers @xmath36 .
so @xmath30 is injective .
see figure [ easyf ] for an example .
in this section we construct an injection @xmath37 thus proving ( [ hard ] ) .
let @xmath38 .
if @xmath39 and @xmath40 are vertices of @xmath21 and @xmath22 , respectively , with the same first coordinate , define the _ vertical distance from @xmath21 to @xmath22 at @xmath41 and @xmath42 _ to be @xmath43 .
the vertical distance from @xmath21 to @xmath22 starts at 2 for their initial vertices and ends at 0 for their final ones .
since vertical distance can change by at most one with a step of a path , there must be some vertical distance equal to 1 .
let @xmath41 and @xmath42 be the first ( most southwest ) pair of points with vertical distance one .
it follows from our choice of vertices that @xmath21 must enter @xmath41 with an east step and @xmath22 must enter @xmath42 with a north step .
let @xmath24 and @xmath25 be the portions of @xmath21 before and after @xmath41 , respectively , and similarly for @xmath22 .
now let @xmath44 where @xmath45 is @xmath24 moved south one unit and @xmath46 is @xmath26 moved north one unit . since @xmath41 and @xmath42 are the first pair of points at vertical distance one , @xmath46 will not intersect @xmath47 and the concatenations are valid paths in @xmath48 .
in fact , the image of @xmath49 is exactly all path pairs @xmath50 that have a pair of points at vertical distance -1 .
applying the same procedure to the first such pair inverts the map and so @xmath49 is injective .
see figure [ newphi ] for an example .
this completes the first proof of theorem [ main ] .
the reader may wonder if we can do away with splitting our problem into two parts , that is , equations ( [ easy ] ) and ( [ hard ] ) .
the answer is yes , and the necessary injection is just a modification @xmath51 of the map @xmath49 .
this will give us a second , completely combinatorial , proof of our main theorem take a path pair @xmath52 .
notice that @xmath21 and @xmath22 must intersect .
so before the first intersection there must be a first pair of points @xmath53 ( on @xmath54 respectively ) at vertical distance 1 .
similarly , after the last intersection there must be a last pair of points @xmath55 at horizontal distance 1 , where horizontal distance is defined analogously .
let @xmath41 and @xmath56 divide @xmath21 into subpaths @xmath57 and use the same notation for @xmath22 . then define @xmath58 where @xmath45 is @xmath24 moved south one unit , @xmath59 is @xmath60 moved west one unit , and @xmath61 are defined in the analogous way but moving in the opposite directions .
it is a simple job to verify that @xmath51 is well - defined and injective just as we did with @xmath49 .
we have two final remarks .
first of all , it is clear from the geometry of the situation that if @xmath47 is self - conjugate then the sequence in theorem [ main ] is also symmetric , but this does not hold in general .
one might also wonder if this sequence has the stronger property that the associated polynomial generating function has only real zeros .
this is not always true as can be seen by taking @xmath62 and @xmath63 . in this case
the associated polynomial is @xmath64 which has two complex roots .
it might be interesting to determine for which shapes the real zero property holds . | simion @xcite conjectured the unimodality of a sequence counting lattice paths in a grid with a ferrers diagram removed from the northwest corner .
recently , hildebrand @xcite and then wang @xcite proved the stronger result that this sequence is actually log concave .
both proofs were mainly algebraic in nature .
we give two combinatorial proofs of this theorem . |
this paper seeks to investigate the appearance of periodic and non - periodic cycles in the time series of stock market returns .
the analysis of the existence of cycles and trends in stock indexes provides us with an essential contribution to the understanding of market efficiency and the distribution of stock indexes returns .
cycles in the economic data have been studied extensively @xcite , resulting in a number of stylized facts that characterize some cyclical or seasonal effects to financial time series @xcite .
the study of cycles in economic data dates back to the early 1930s @xcite .
various techniques to measure seasonality have been widely applied , combining ideas from mathematics , physics , economics and social sciences .
these efforts have resulted in research findings of , among other , intraday trading effects @xcite , weekend and/or three - day effects @xcite , intramonth effects @xcite , quarterly and annual cycles @xcite , and various multi - year cyclical variations in stock market index returns @xcite . a consensus of opinion on the nature or the character of cyclic effects , however , has not been reached .
financial markets belong to a class of human - made systems exhibiting complex organization and dynamics , and similarity in behavior @xcite .
complex systems have a large number of mutually interacting parts , are often open to their environment , and self - organize their internal structure and dynamics , which produce various forms of a large - scale collective behaviors .
they operate simultaneously at different scales .
the outputs of such systems , time series of records of their activity , display co - existence of collectivity and noise @xcite ; the complexity of systems is reflected in time series that exhibit a wealth of dynamic features , including trends and cycles on various scales @xcite . the tools to study such systems
thus can not be analytical , but rather must be adapted to enable accurate quantification of their long - range order . in this sense , we have chosen to contribute to the debate about the existence and types of cycles in stock market data in two ways : by way of applying wavelet spectral analysis @xcite to study market returns data , and through the use of hurst exponent formalism @xcite to study local behavior around market cycles and trends .
firstly , we utilized wavelets to study cyclical consistency in time series of stock market indexes ( smis ) .
wavelet analysis is appropriate for such a task ; it was originally introduced to study complex signals @xcite .
we use wavelet - based spectral analysis , which estimates the spectral characteristics of a time - series as a function of time @xcite , revealing how the different periodic components of a particular time - series evolve over time .
it enables us to compare stock market index time series wavelet spectra from different economies , to examine the similarities in contributions of cycles at various characteristic frequencies to the total energy spectrum , to observe when exactly these contributions happen in time for different economies , and to compare if the ups and downs of each cycle occur simultaneously in different smi series . with this tool we can attempt to address the question of whether the complexity of a financial market is specifically limited to the statistical behavior of each smi time series or parts of an smis series complexity
can be attributed to the overall world market @xcite .
we use the hurst exponent formalism to attempt to address the question of what type , or types , of cyclical behavior smi data possess . in recent years
, the application of the hurst - exponent - based analyses has led many researchers to conclude that financial time series possess multi - scaling properties @xcite .
in addition , these methods have allowed for the examination of local scaling around a given instance of time , so that the complex dynamical properties of various time series can be analyzed locally rather than globally @xcite . in this paper
, we use the time dependent hurst exponent approach to test the local character of cycles at various characteristic frequencies of smi time series from different economies . in doing
so we aim to compare the behavior of each cycle across stock markets and to find ways to classify various markets according their cyclical behavior .
we use the wavelet and the hurst exponent approach to analyze three types of smi time series : data belonging to stock markets of developed economies , emerging economies , and of the underdeveloped or transitional economies .
previous and recent work by our group and others has demonstrated that smi series exhibit scaling properties connected to the level of growth and/or maturity of the economy the stock market is embedded in @xcite , and that all the smi series exhibit the effects of various periodic or non - periodic cycles that are visible in their wavelet spectra @xcite .
it has been demonstrated that in emerging or transitional markets stock indexes do not fully represent the underlying economies @xcite , therefore we wanted to tailor our smi study with this in mind and differentiate between underdeveloped ( transitional ) economies , emerging economies , and developed economies .
our study is structured as follows . in sec .
2 . we give a brief overview of the smi dataset and the variables that we study . in sec .
we describe the general framework of the wavelet transform ( wt ) spectral analysis and the specificities of the use of the morlet wavelet basis for the transformation we apply on our dataset .
we apply the wt framework to study the appearance and consistency of cycles across stock markets , and we present our results within this section . in addition , within sec .
we have also investigated the statistical effects of the observed seasonal behavior on the spectral behavior of our smi data . in sec .
we give an introduction to the technique of time - dependent hurts exponent analysis and present the specificities of the detrended moving average ( dma ) method and its time - dependent variation ( tddma ) .
we then list the results of the use of tddma on our smi data , and develop a quantitative indicator ( that we have dubbed the development index ) , which may help classify the level of development of a particular market according to the markets local cyclical behavior .
we end our paper with a list of conclusions and a few suggestions for future work in sec .
in this paper , we investigate data from the following stock markets : the new york stock exchange nyse index , the standard & poor s 500 ( s&p500 ) index , the uk ftse 100 index , the tokyo stock exchange nikkei 225 index , the french cac 40 index , and the german stock market dax index , which we consider developed economies ; the shanghai stock exchange sse composite index , the brazil stock market bovespa index , the johannesburg stock exchange jse index , the turkey stock market xu 100 index , the budapest stock exchange bux index , and the croatian crobex index , which we consider emerging economies ; the tehran tepix index , the egyptian stock market egx 30 index , and the indexes of the developing economies in the western balkans - the belgrade stock exchange belexline index , the montenegrin montex 20 index , the sasx 100 index of the market of bosnia and herzegovina and the birs index of bosnian entity republic of srpska , representing markets of underdeveloped economies . table 1 .
lists general characteristics of the smi time series which we have analyzed ; depending mainly on the market development level , they are of varying duration .
the variables studied in our paper are the daily price logarithmic returns that are defined as @xmath0 + where @xmath1 is the closure price of the stock market index at day @xmath2 , and the lag period @xmath3 is a time interval of recording of index values @xmath1 .
all of the analyzed time series of prices on the stock markets @xmath1 are publicly available ( from the official web - sites of the markets in question or from the yahoo finance database ) , and are given in local currencies .
the values of the smi data are listed only for trading days that is , they are recorded according to the market calendar , with all weekends and holidays removed from datasets . ]
the wavelet transform ( wt ) was introduced @xcite in order to circumvent the heisenberg uncertainty principle problem in classical signal analysis , and achieve good signal localization in both time and frequency that a classical fourier transform approach lacks .
namely , in wt the window of examination length is adjusted to the frequency analyzed : slow events are examined with a long window , whilst a shorter window is used for fast events . in this way an adequate time resolution for high frequencies and a good frequency resolution for low frequencies
is achieved in a single transform @xcite .
the continuous wavelet transform @xcite of a discrete sequence @xmath4 is defined as the convolution of @xmath4 with wavelet functions @xmath5 in the following way : @xmath6 with @xmath7 and @xmath8 being the scale and translation - in - time ( coordinate ) parameters , and @xmath9 the total length of the smi series studied .
the wavelet functions @xmath5 , used in eq.(2 ) , are related to the scaled and translated version of the mother - wavelet @xmath10 , through @xmath11 + in order to examine the existence of cycles in smi data , we used the wavelet scalegrams ( mean wavelet power spectra ) @xmath12 , that are defined by @xmath13 + the scalegram @xmath12 can be related @xcite to the corresponding fourier power spectrum @xmath14 via the formula @xmath15 + this formula implies that if the two spectra , @xmath12 and @xmath14 , exhibit power - law behavior , then they should have the same power - law exponent @xmath16 .
the meaning of the wavelet scalegram is the same as that of the classical spectrum - it gives a contribution to the signal energy at a specific scale parameter @xmath7 .
we are thus able to view and estimate the peaks of wavelet spectra in the same way as one would approach this problem in a classical fourier approach . in this paper , we find it convenient to follow the choice of many researchers before us and use the standard set of morlet wavelet functions as a wavelet basis for our analysis .
the choice of the wavelet is an important aspect of any wt analysis , as the wavelet coefficients @xmath17 contain combined information on both the function we analyze and the analyzing wavelet @xcite . in applying the wt method
it is important to choose a mother - wavelet that is well localized in both time and frequency , and is at least of zero mean ( of course , these properties remain valid for the corresponding generated wavelet functions @xmath18 ) .
the morlet wavelet @xcite has proven to possess this optimal joint time - frequency localization @xcite , and can thus be used for detecting locations and spatial distribution of singularities in time series @xcite .
the morlet set is defined by @xmath19 with the corresponding fourier transform @xmath20 where @xmath21 , @xmath22 is the heaviside step function , @xmath23 is the fourier circular frequency , and @xmath24 is the characteristic nondimensional parameter ( a real number ) . for the morlet wavelet ,
the relation between the scale and the frequency is given by @xmath25 where @xmath26 is a real number that characterize the particular wavelet .
in our study we use @xmath27 .
we have calculated wt power spectra for all our smi series , and for all the periods ( durations ) these data series were available to us .
we took into consideration only the values of the wt spectra between the minimum time scale of @xmath28 and the statistically meaningful maximum time scale @xcite of @xmath29 , and searched for characteristic peaks ( local maxima ) within those limits .
figure [ fig1 ] . depicts the typical findings we got for all our smi series , represented by the wt spectra of the dax returns series ( as a primer of a developed market ) , xu 100 returns series ( representing emerging markets ) , and tepix returns series ( representing the underdeveloped economies ) . from figure [ fig1 ] characteristic peaks , or rather characteristic cycle periods around characteristic peaks , can be recognized on different time scales .
these include : \1 . the working week cycle , or the peak around 5 days .
this peak is visible in all our smi series , with the exception of the tepix series , where it is lacking altogether .
the exact positions of the peak are not the same in all the series , but all the peaks are embedded in the period ( equivalent to the frequency band in a classical fourier case ) of 2 days to 6 days ( see figure [ fig1 ] , where the borders of all periods are drawn around the corresponding peaks ) .
the peak around 7 days , which probably belongs to a calendar - week cycle .
this peak is not visible in xu 100 and sse series , while it is visible in all other smi series . in all the series where it is visible
, this peak is embedded in the period of 6 to 10 days .
the peak around 14 days , probably tracing a two - week cycle .
this peak is visible in all out smi series .
the period that surrounds this peak is from 10 days to 25 days .
in addition to this peak , some series show a very near one at around 20 days ( probably equivalent to a three - week cycle ) ; this additional peak appears in ftse 100 , nyse , cac 40 , dax and sse markets .
since the 20-day peak appears in only few ( developed ) economies , we have decided not to list it as a separate cycle , just to mention it in the text .
it is also depicted in figure [ fig1 ] .
, where the border around it is drawn with dotted vertical line .
+ 4 . the monthly cycle , or the peak around 30 days .
it is present in all our smi series , and is embedded in the period between 25 days and 60 days . at this point in time , again , an additional peak appears in some of our smi series . where it exists , this peak is positioned at around 45 days , and it appears in ftse 100 , s&p 500 , cac 40 , dax , xu 100 , bovespa and egx 30 markets .
for the same reasons as above , we have chosen just to mention this cycle here ( and have shown it s border with dotted vertical line in figure [ fig1 ] ) .
+ 5 . the quarterly cycle , or the 90-day peak .
this peak was found in all our smi series , and it is positioned in the region between 60 days and 110 days .
+ 6 . the 4 - 5 month cycle , or the 150-day peak .
this peak was found in all our smi series , and is positioned in the region between 110 days and 190 days .
the 6 - 7 month peak , or the semi - annual cycle .
this peak is not visible in all our underdeveloped smi series , and in the nyse and nikkei 225 series .
where it appears , the peak is positioned in the period between 190 and 250 days .
the annual cycle , or the peak around 360 days .
this peaks was not found in ftse 100 , sse and egx 30 series . in all other cases ,
it is embedded in the region between 250 days and 450 days .
+ 9 . the bi - annual cycle , or the 600-day peak .
this multi - year cycle was found in all our smi series .
it is embedded in a time interval between 450 days and 950 days .
other multi - year cycles have been found in the series of developed economies .
these data series contain sufficient data points that enable the calculation of wt spectra on higher scales .
due to our inability to investigate the existence of these peaks in all our data , we will just mention them here , but will not label them in the way we labelled other peaks and peak regions . in order to be able to compare and characterize the obtained wavelet spectra of our stock market data , we have calculated relative energy content and relative amplitude of all the regions ( periods , bands ) under characteristic peaks in all our data series .
the relative energy content of the @xmath30-th peak in a wt power spectrum is defined @xcite as : @xmath31 where @xmath32 represents the average energy content of the period surrounding the @xmath30-th peak : @xmath33 and @xmath34 is the total energy content of the wt spectrum of the stock market series analyzed .
the energy content is a physical quantity behind a wt power spectrum , so it represents it s natural characteristic .
similarly , the relative amplitude of the spectral band under the @xmath30-th peak is defined as : @xmath35 with @xmath36 its average amplitude , and @xmath37 the total amplitude of the wt power spectrum of the stock market series of interest .
the amplitude of the wt power spectrum depends @xcite on the variability of the frequency ( scale ) band analyzed - the more constant the frequency , the higher the amplitude .
we calculated the relative energy contents @xmath38 and the relative amplitudes @xmath39 for all the obtained peaks in all the analyzed wt spectra .
we then performed statistical analysis of three groups of data sets - those belonging to the developed economies , the emerging markets , and the underdeveloped economies .
we first performed the shapiro - wilk test for normality of distributions within these three data groups .
if normality of distributions existed within our datasets , we performed the one - way anova test to compare our sample means , with the significance level of @xmath40 . if the anova test confirmed the existence of differences of means , the average means for all three groups of data
was compared using the bonferroni method .
if , however , the shapiro - wilk test did not confirm the existence of normality of distributions within our dataset , we performed the kruskal - wallis anova test to compare the means , with the significance level of @xmath40 .
if the kruskal - wallis anova test confirmed the existence of differences in the groups means , the comparison of average means for all three groups of data was done using the wilcoxon mann - witney method .
lists the calculated average values of relative energy content @xmath38 and the relative amplitudes @xmath39 of all the peaks for the three smi groups .
the statistically significantly different values between the groups for each of the peaks are marked in bold - if only one value is bolded , then it differs from the other two market groups in a peak group ; if two values are bolded they differ mutually ; and if all three values have been bolded then all the three market groups values differ from each other . and b ) the relative amplitudes @xmath39 for all three market groups .
results are depicted for the three peak regions - a small scale region surrounding the peak at 5 days , a mid - scale region surrounding the peak at 150 days , and a large scale region surrounding the peak at 600 days .
squares enclose the @xmath41 of the values within the smi group , while the error bars depict the maximum and the minimum value within the same group.[fig2 ] ] our results are also illustrated in figure [ fig2 ] , where average values of the relative energy content @xmath38 and the relative amplitudes @xmath39 for all three market groups , and in three peak regions - a small scale region surrounding the peak at 5 days , a mid - scale region surrounding the peak at 150 days , and a large scale region surrounding the peak at 600 days , are depicted .
figure [ fig2 ] and table 2 .
show that in the small scales region ( peaks of up to 90 days ) the values of both the relative energy contents @xmath38 and the relative amplitudes @xmath39 under the spectral peaks for the underdeveloped markets are smaller than the values of the two other groups in a clear , statistically significant manner .
even more so , for the values of the relative energy content @xmath38 for the two small scale peaks ( peaks at 5 days and at 14 days ) , we can statistically differentiate all three market groups .
for the peaks at lager scales ( peaks at 150 days and more ) , the behavior of underdeveloped markets data does not differ from the other two groups , except in the case of a large scale region of the peak at 600 days .
these results can be understood in the light of values of slopes of the wavelet spectra of our entire dataset .
it has been shown that the smi series from the underdeveloped economies have wt spectra that show highly correlated long - range behavior , with the exponent @xmath42 @xcite .
the difference of values of @xmath38 and @xmath39 for the peaks in the small scales region could explain the smaller contribution of small scale peaks to the overall spectral behavior in underdeveloped markets data , which bring about greater contribution of higher scale peaks and consequently higher slopes of wavelet spectra for these data .
further , the slopes of wavelet spectra for the other two groups of market s data ( for developed and emerging economies ) are either close or equal to @xmath43 , which means that contributions of different peaks in wavelet spectra have to be approximately equal .
it seems , therefore , that the transitional markets do not follow the same behavioral pattern as the markets of emerging or developed economies at time scales of days , weeks , and even several months .
our results thus show that measures like @xmath38 and @xmath39 for the peaks in the small scales region could be used for partial differentiation between market economies .
in order to gain another insight into the local complexity of our smi data , and the possibility of finding means to further quantitatively distinguish the three groups of smi data we use , we have applied the time - dependent detrended moving average ( dma ) algorithm to all our smi series .
firstly , in order to study the general statistics of our smi data , we have employed the centered detrended moving average ( cdma ) technique .
the variation of a standard cdma method that we use here @xcite is performed in three consecutive steps . in the first step , for each sequence of @xmath44 , we construct the sequence of cumulative sums @xmath45 . with @xmath9
we assigned the total number of recorded values for a given series ( the smi number @xmath2 plays the role of time ) . in the next step ,
the entire series of @xmath46 is divided into a set of overlapping segments of the length @xmath47 , and a moving average for each segment is calculated . in the next step a new series , the so - called series of detrended walk @xmath48 is calculated , which we get by removing the moving average from the @xmath49 .
@xmath50 where @xmath47 is the window size .
hence , the moving average function considers @xmath51 data points in the past and same number of points in the future .
finally , one has to calculate the variance about the moving average for each segment and determine the average of these variances over all segments , which brings about the detrended moving average function @xcite @xmath52 + by increasing the segment length @xmath47 the function @xmath53 increases as well .
when the analyzed time series follows a scaling law , the cdma function is of a power - law type , that is , @xmath54 , with @xmath55 . scaling exponent @xmath56
is usually called the hurst exponent of the series @xcite . in the case of short - range data correlations ( or no correlations at all )
@xmath53 behaves as @xmath57 . for data with power - law long - range autocorrelations
one may expect that @xmath58 , while in the long - range negative autocorrelation case we have @xmath59 .
when scaling exists , the exponent @xmath56 can be related to the fourier and wt power spectrum exponent @xmath16 through the scaling relation @xcite @xmath60 .
we applied the time - dependent dma algorithm ( tddma ) @xcite to the subset of data in the intersection of the smi signal and a sliding window of size @xmath61 , which moves along the series with step @xmath62 .
the scaling exponent @xmath56 is calculated for each subset and a sequence of local , time - dependent hurst exponent values is obtained .
the minimum size of each subset @xmath63 is defined by the condition that the scaling law @xmath54 holds in the subset , while the accuracy of the technique is achieved with appropriate choice of @xmath63 and @xmath64 @xcite .
we have chosen windows of up to @xmath65 , with the step @xmath66 for our tddma algorithm , while the scaling features are studied in the nine regions that enclose smi spectral peaks ( listed above and depicted in figure [ fig1 ] . in figure [ fig3 ]
we give an example of the calculated tddma values for the three randomly selected representatives of smi market groups , given for a time interval from year 2008 to year 2011 . ] in order to be able to quantify the local behavior of smi data , we have built the hurst vectors @xmath67 .
each coordinate of stock market vector @xmath67 corresponds to the mean value of local hurst exponents from a selected interval that includes and borders each peak .
our analysis was performed on nine intervals that separate nine market peaks ( listed in the text above , and defined through the use of wt analysis ) , marked by index @xmath30 ( @xmath68 ) , while @xmath69 counts the smi series . from all these values
we have built the hurst reference smi vector @xmath70 , composed of the mean values of @xmath67 for each coordinate ( peak ) @xmath30 and for all the smis that we have analyzed .
we have done that in order to be able to compare efficiency of stock markets on each of the separate nine time intervals .
the hurst reference smi vector is thus defined as : @xmath71 for @xmath72 different smi indexes in our dataset . in the case of our dataset
, the values of the reference vector @xmath73 are changing with the addition of new smi data ( markets ) , but for @xmath74 this change becomes insignificantly small . from the two vectors @xmath67 and @xmath75
we have calculated the relative smi hurst unit vectors @xmath76 that we have defined as : @xmath77 the unit vectors @xmath76 give us the information of the direction of the representative point for each market in relation to the hurst reference point that , to a certain accuracy , marks the overall financial status of the markets dataset that we use .
however , in the case of our dataset , the distance of representative points @xmath76 from the hurst reference point did not provide us with any relevant additional information about the market development or efficiency .
this can be demonstrated through the use of the so - called cosine similarity , which is the scalar euclidean product of two @xmath76 vectors , and which can quantify the level of correlation between vectors @xmath76 .
scalar products of @xmath76 are defined as : @xmath78 + where @xmath69 and @xmath16 count smi series ( @xmath79 ) , while @xmath80 numbers peaks ( peak regions ) .
we have arranged values of these scalar products so as to form a correlogram of market development @xmath81 .
the markets correlogram is depicted in figure [ fig4 ] ; it consists of two block matrices that differentiate strong correlations within the group of underdeveloped markets ( upper left corner ) and within the group of developed markets ( lower right corner ) .
the correlogram depicted in figure [ fig4 ] displays the existence of market group that does not belong neither to developed nor to underdeveloped type .
members of this third group - the emerging markets - mainly correlate weakly with other two groups and within the group , and show random unpredictable strong correlations with some members ( markets ) in the developed or underdeveloped market group .
this inability to look alike clearly differentiates emerging markets in correlogram @xmath81 and in figure [ fig4 ] . .
positive correlations of market s hurst unit vectors are given in shades of blue ( for @xmath82 ) , while negative correlations are depicted in shades of red ( @xmath83 ) .
horizontal and vertical white lines mark , from left to right , borders between groups of underdeveloped , emerging , and developed markets . [ fig4 ] ] market development correlogram depicted in figure [ fig4 ] prompted us to try to find a unique hurst indicator of market development , based on the projection of market s hurst unit vectors @xmath84 on the direction of development ( or maturity ) of markets dataset analyzed . in order to calculate this indicator ( we dubbed it the development index ) ,
we have firstly defined the direction of development in markets indexes hurst space .
we have made an assumption that the unit vector of development has to be able to link representative points of ideal markets , that is , markets with hurst vectors @xmath85 made of identical values for all the peak intervals @xmath30 .
this could be defined in a following simple way in the general hurst coordinate space : @xmath86 where @xmath87 stands for the @xmath88 vectors made of all unit components . in the next step , in order to project hurst reference vectors to the direction of development , it is necessary to define the representation of the unit vector of development in the hurst space that originates at the hurst reference vector @xmath70 .
the relative ( in relation to the market s hurst reference vector @xmath70 ) direction of development would then be : @xmath89 in the case of our dataset , the values of this new vector s components have not significantly changed with the addition of new smi data to dataset for @xmath74 ( @xmath47 being the number of markets in the dataset analyzed ) . the relations in eqs .
17 - 19 led us to the value of @xmath90 for our dataset of @xmath72 stock market indexes : @xmath91 with the error for each component @xmath30 being @xmath92 .
we have defined the development index ( di ) as a projection of hurst unit vectors onto this direction of development : @xmath93 graphical illustration of these projections is given in figure [ fig5 ] .
, while the direction of the main axes is given by the unit vector of the direction of developemnt @xmath90 .
the development index @xmath94 is calculated as a projection of hurst unit vectors @xmath84 onto the @xmath90 , which is directed to a portion of hurst space where representative points of developed markets are grouped .
[ fig5 ] ] values of di for markets in our dataset are given in table [ tab_4 ] .
it is visible from table [ tab_4 ] that the three market categories ( underdeveloped , emerging , and developed markets ) can be differentiated by this order parameter .
we have decided to define the borders that separate our three market categories following the phenomenological arguments we have set in section 4.1 .
namely , the values of the hurst vectors and a correlogram that we have calculated all point to the existence of two distinct groups that are well clustered ( underdeveloped and developed markets ) , divided by a claster of smi time series that transitions between these two groups ( the emerging markets ) .
this is how , in the case of our dataset , we have decided to use the symmetry principle and define the border between the group of developed and emerging markets at @xmath95 , while the underdeveloped and emerging markets are , in our case , separated at @xmath96 ( for our dataset , @xmath97 ) . based on the criterion that we have produced , the egyptian stock market index egx30 would be classified as an emerging market , rather than an underdeveloped market as we initially assumed , while the hungarian bux index would classify as developed rather than the emerging market smi .
[ tab_4 ] with this procedure we can examine the stock market time series in groups or individually , for any given smi time series .
in this paper we have analyzed spectral properties of time series of stock market indexes ( smis ) of developed , emerging , and underdeveloped ( or transitional ) market economies , in order to examine differences and similarities in their seasonal behavior , and to try to re - classify markets in our dataset according to the character of their cyclical behavior .
we have used two different well established techniques of data analysis to obtain and verify our findings : the wavelet transformation ( wt ) spectral analysis and the time - dependent hurst exponent approach ( tddma ) . we have found multiple peaks in wavelet spectra of all our smi time series .
moreover , we have found all the peaks positioned at roughly the same times ( or time intervals ) in all our data , which points to the similarity in seasonal behavior across different market economies in our dataset .
we have identified what can be termed a working - week cycle ( or a 5-day peak ) , a one - week cycle ( or a 7-day peak ) , a two - week cycle ( or a 14-day peak ) , a monthly cycle ( or a 30-day peak ) , a quarterly cycle ( or a 90-day peak ) , a 4- to 5-month cycle ( or a 150-day peak ) , a semi - annual cycle ( or a 6- to 7-month peak ) , an annual cycle ( or a 360-day peak ) , and a bi - annual ( or a 600-days ) multi - year cycle in our dataset .
the dissimilarities between smi records from the different economies that we have observed occur only in the lack of a spectral peak in some of the analyzed markets ( in the case of a one - week cycle , a semi - annual cycle , and an annual cycle ) , or a slight lack of synchronization at a particular peak interval ( peaks are not positioned at exactly the same time instances in all the smi series analyzed ) .
this prompted us to conclude that the seasonal behavior in different markets is probably a reflection of universality in market behavior , rather than a local characteristic of a particular economy .
given that financial markets are human - made complex systems , it is plausible to believe that our findings can be explained by the fact that business cycles are a reflection of common human working habits and behavior .
some authors find this commonality even desirable for the optimal functioning of a stock market , as was , for example , shown for the euro monetary area @xcite .
some researchers , on the contrary , claim that these effects are not significant for the effectiveness of a stock market @xcite . in order to examine whether the observed seasonal adjustments to the behavior of stock markets could be used as an indicator of the level of development or strength of the economy that underlies the specific stock market ,
we have performed a statistical analysis of the properties of wavelet spectra that characterize particular peak behaviors .
we have statistically compared the relative energy content and the relative amplitude of each peak between the three groups of smi series that we have analyzed - those belonging to developed economies , emerging economies and economically underdeveloped ( or transitional ) countries .
we have found that the underdeveloped markets do not follow the same behavioral pattern as emerging or developed economies at time scales of days , weeks , and even several months .
namely , their wt spectra show , in a statistically significant manner , less pronounced effects of fast ( small time scale ) cycles on the overall spectral behavior than in the wt spectra of the smi series from developed and emerging economies .
in contrast , developed economies appear to even out all the cyclical ( peak ) effects in their wt spectra , or even to show a larger influence of the fast ( small time scale ) peak regions on their overall spectral behavior .
the emerging markets spectra behave somewhere in the middle of these two cases .
these observed differences could contribute to the variations in scaling behavior of markets , which has been reported previously @xcite .
it was proven that the economies of underdeveloped countries have wt spectra that show highly correlated long - range behavior , with the exponent @xmath42 ( @xmath58 ) , opposite to emerging and developed economies , which show uncorrelated or even slightly anti - correlated spectral behavior , with @xmath98 ( @xmath99 ) .
the observed sensitivity of scaling exponents to the level of development of economies could be related to the findings we present here - to the relative influence of the small scale spectral peaks on the overall smi spectral behavior .
finally , in this paper we propose a way to quantify the level of development of a stock market , based on the relative influence ( or , in some cases , existence ) of wt spectral peak intervals on the overall scaling behavior of smi time series . in order to do that we have used the hurst exponent approach to calculate what we named the development index , which proved , at least in the case of our dataset , to be suitable to rank the smi series that we have analyzed in three distinct groups .
further verification of this method remains open for future studies by us , or by other groups . | in this paper we have analyzed scaling properties and cyclical behavior of the three types of stock market indexes ( smi ) time series : data belonging to stock markets of developed economies , emerging economies , and of the underdeveloped or transitional economies .
we have used two techniques of data analysis to obtain and verify our findings : the wavelet spectral analysis to study smi returns data , and the hurst exponent formalism to study local behavior around market cycles and trends .
we have found cyclical behavior in all smi data sets that we have analyzed .
moreover , the positions and the boundaries of cyclical intervals that we have found seam to be common for all markets in our dataset .
we list and illustrate the presence of nine such periods in our smi data .
we also report on the possibilities to differentiate between the level of growth of the analyzed markets by way of statistical analysis of the properties of wavelet spectra that characterize particular peak behaviors .
our results show that measures like the relative wt energy content and the relative wt amplitude for the peaks in the small scales region could be used for partial differentiation between market economies .
finally , we propose a way to quantify the level of development of a stock market based on the hurst scaling exponent approach . from the local scaling exponents calculated for our nine peak regions
we have defined what we named the development index , which proved , at least in the case of our dataset , to be suitable to rank the smi series that we have analyzed in three distinct groups .
stock market returns , wavelet analysis , hurst exponent formalism , cycles and trends , development index 82c41 , 82c80 , 91b84 |
the luminosity distance - redshift relation obtained using type ia supernova provided the first clear evidence for an accelerating universe ( @xcite ) .
further evidence , based on different types of observational data , has since become available ( @xcite ) .
the acceleration of the expansion rate of the universe rapidly became one of the most intriguing problems in modern cosmology .
since then , there have been many solutions proposed and some have survived the increasingly tight constraints imposed by observational data . among those solutions , which include changes to the left - hand side ( or the geometry / gravity part ) of einstein s field equations , as in modified gravity theories ( @xcite ) , or the introduction of extra spatial dimensions , as in braneworld models ( @xcite ) , one of the most popular assumes the existence of an exotic form of energy in the universe , characterized by a negative pressure , known as dark energy .
if it exists , then present - day observations suggest it accounts for approximately @xmath2 per cent of the universe s total energy density , and its pressure - to - density ratio ( known as the equation of state , @xmath3 ) is nearly constant with time and close to @xmath4 ( @xcite ; see @xcite for more details ) .
if future observations continue to point towards a value of @xmath3 near @xmath4 , then the hypothesis of the dark energy being in the form of a cosmological constant , @xmath5 , will gain strength , demanding for more insight on the theoretical problems raised by such a scenario ( see @xcite for a review ) .
but it is also conceivable that observational data will become more easily reproducible with a value for @xmath3 close to , but not exactly equal to , @xmath4 , opening the door for alternative hypothesis , such as quintessence theories . in quintessence theories ,
the dark energy is assumed to be the result of a scalar field , @xmath6 , minimally coupled to ordinary matter through gravity .
the field evolves in a potential @xmath7 with a canonical kinetic term in its lagrangian , which results in a dark energy equation of state @xmath8 , not necessarily constant .
these models have already been extensively studied ( @xcite ) and , usually , it is assumed that the evolution of the scalar field is monotonic in a potential that satisfies the typical inflationary slow roll conditions ( see @xcite for a review on inflation ) , as given by : @xmath9 and @xmath10 under these conditions , the dark energy equation of state will always be close to @xmath4 , and the motion of the field becomes greatly simplified ( @xcite ) , allowing for an analytical expression for @xmath3 as a function of the scale factor @xmath11 .
the motion of the field will also depend on the present - day values of @xmath3 and of the dark energy density , @xmath12 , where @xmath12 is the ratio of the dark energy density , @xmath13 , to the universe s critical density , @xmath14 . however , the inflationary slow - roll conditions are sufficient but not strictly necessary in quintessence models .
relaxing one of the conditions opens up the possibility of complex scalar field dynamics , allowing for a more dynamical evolution of the dark energy equation of state , while still being on average close to @xmath4 . in this paper
we will be particularly interested in scalar field models where late - time oscillations in the expansion rate of the universe can arise due to non - standard scalar field dynamics .
in particular , we will focus on two scalar field models for which such behavior is possible .
model i assumes a quadratic potential , where an extra degree of freedom , related to the curvature of the potential at the extremum , is introduced on the evolution of the scalar field ( @xcite ) .
more specifically , we consider the case where the curvature of the potential is such that it enables the field to oscillate around the stable extremum , a model already confronted with supernovae data ( @xcite ) . here
we determine the evolution of its equation - of - state and study its impact on structure formation , forecasting the viability of the model against future euclid data . in the case of model ii , we consider rapid oscillations in the equation - of - state , corresponding to an effectively constant equation of state when averaged over the oscillation period ( see @xcite , @xcite , and references therein ) .
an oscillating equation - of - state is a way to solve the coincidence problem , removing the special character of the current accelerating phase , and of including the crossing of the phantom barrier @xmath15 .
various phenomenological parameterizations of the amplitude , frequency and phase of the oscillations have been proposed ( @xcite ) , not necessarily arising from oscillations in quintessence models , and have been constrained mostly with background observables , except for @xcite where the impact on structure formation have been investigated . in our paper , the rapid oscillations are produced by a scalar field in a power - law potential where the minimum is zero .
the field starts evolving monotonically , and only close to today it starts oscillating around the minimum , guaranteeing that the observed cosmic history is not significantly affected ( @xcite ) , and producing rapid oscillations in the equation - of - state . in section 2
, we review the dynamics of quintessence and present the models on which we focus our attention . in section 3 , we present the method and cosmological probes we will use to obtain the euclid forecasts presented in section 4 . we conclude in section 5 . throughout the paper
we shall assume the metric signature @xmath16}}$ ] and natural units of @xmath17 .
the action for quintessence ( see @xcite for a thorough review ) is given by @xmath18 where @xmath19 is the ricci scalar , @xmath20 and @xmath21 are , respectively , the matter and scalar field lagrangians .
the field s lagrangian takes the form @xmath22 where @xmath7 is the field s potential .
we will be considering the line element for a flat , homogeneous and isotropic universe as given by the friedmann - roberston - walker ( frw ) metric .
varying the field s individual action with respect to the metric elements @xmath23 , one can obtain the field s energy - momentum tensor elements , @xmath24 . from the latter , the quintessence equation of state , defined as the ratio between the dark energy pressure , @xmath25 , and energy density , @xmath13 ,
is given by @xmath26 this equation shows that , in order to have a dark energy equation of state close to @xmath4 , the field s evolution has to be potential dominated , such that @xmath27 .
this is why , generically , quintessence models are associated to scalar fields slowly evolving in the respective potentials .
this can be ensured , for instance , by the slow - roll conditions given by equations ( [ s1 ] ) and ( [ s2 ] ) .
the scalar field equation of motion is given by the euler - lagrange equation @xmath28 through this equation we can see that the field will evolve in the potential @xmath7 rolling towards a minimum in the quintessence potential , while its motion is damped by the presence of the hubble parameter , @xmath29 .
considering a flat universe , with the frw metric , the hubble parameter , as a function of the scale factor @xmath11 , is given by @xmath30 , where @xmath31 is the universe s total energy density .
we will be considering a universe consisting of pressureless matter and a scalar field playing the role of dark energy .
this implies that @xmath32 . as a function of the scale factor , @xmath33 , where @xmath34 is the present - day value of the matter density ; the field s energy density will be given by @xmath35 . in a flat universe ,
the present - day value of the matter energy density is determined by @xmath36 , where @xmath37 and @xmath38 are the ratios of the present - day values of the matter and dark energy densities to the critical density , @xmath14 .
the first model we consider is that of a scalar field rolling close to a stable non - zero minimum of its potential .
the field is assumed to evolve in a potential that satisfies the slow - roll condition of equation ( [ s1 ] ) , while the other slow - roll condition is somewhat relaxed since @xmath39 can be large , namely as a result of the curvature at the potential s minimum .
this model was extensively studied by @xcite , where a set of solutions for the evolution of the field was derived , as well as the respective equation of state , @xmath40 , as a function of the scale factor @xmath11 .
these solutions are applicable to a wide array of potentials with non - zero minima , and depend on the present - day values of @xmath3 and @xmath12 , and on the curvature of the potential at the minimum , controlled by the parameter @xmath41 .
the latter is given by equation @xmath42 in @xcite , which we reproduce here : @xmath43 where @xmath44 , each prime representing a derivative with respect to the scalar field , and @xmath45 are , respectively , the curvature and potential values at the minimum of the potential , @xmath46 .
the latter corresponds to the present - day value of the dark energy density , @xmath47 . for this model
we have @xmath48 , which means that @xmath49 . in the case of @xmath50 , the behavior of @xmath40 is oscillatory , and can be analytically approximated through a combination of sinusoidal functions ( @xcite ) . for this to happen , we need to have @xmath51 , requiring the potential to be significantly curved at its minimum . in our analysis
we do not use the analytical approximation for @xmath40 derived by @xcite , but instead calculate @xmath3 directly from equation ( [ eqstate ] ) by numerically solving equation ( [ eqmotion ] ) as a function of @xmath11 .
the solution depends on three parameters : @xmath38 , @xmath41 and the initial value for the field , @xmath52 .
the field s initial velocity , @xmath53 is assumed to be zero , which determines the initial value of the equation of state to be @xmath4 .
we assume a quadratic potential , such that @xmath54 where @xmath55 is the curvature at the minimum , determined by @xmath56 . for models with large negative values of the curvature parameter @xmath41 ,
the curvature of the potential at the minimum is larger , and the field is able to significantly overcome the hubble friction term and cross the potential s minimum , oscillating around it with decreasing amplitude .
this produces a more dynamical behavior for @xmath3 , as shown in fig .
[ w1 ] , exhibiting damped oscillations with local maxima located between points where the field s motion comes to a halt .
for lower absolute values of @xmath41 , @xmath3 exhibits a less dynamical behaviour , evolving monotonically in the case @xmath57 .
we note that if @xmath58,no oscillations occur , regardless the values of the other parameters , and the field effectively behaves as a cosmological constant .
the second model we consider is that of a scalar field oscillating around zero , with a potential of the form ( @xcite ) @xmath59}}.\ ] ] this potential is very flat for small field values and acquires a quadratic form for large field values .
the scale @xmath60 determines the curvature of the potential at the minimum , @xmath61 , whereas the scale @xmath62 determines where the potential changes shape and enters the quadratic region , where the field will eventually oscillate .
this model allows , if certain conditions are met , to have a field slowly - rolling until recently and then present some oscillatory behavior close to the present .
it has been show that for a power - law potential , @xmath63 , one obtains @xmath64 when averaged over the oscillation period , @xmath65 , as long as it is much smaller than the hubble time , i.e. , @xmath66 ( @xcite ) . therefore ,
in the oscillatory region ( @xmath67 ) of the potential being considered , @xmath3 should average to zero , because we then effectively have @xmath68 , i.e. @xmath69 .
this means the scalar field would behave like pressureless matter , which goes against the need for having a negative pressure component to drive the cosmic acceleration .
however , this model can be fine - tuned so that the oscillatory region of the potential is reached close enough to the present in order for the predicted large - scale dynamics of the universe not deviate much from that expected in a model where it is the presence of a cosmological constant that makes the universe accelerate . this
fine - tuning can be achieved if the parameters of the model being considered take values around @xmath70 , @xmath71 , with @xmath72 being the present - day value of the hubble parameter , @xmath73 and @xmath74 , where @xmath75 is the initial value of the field ( @xcite ) . in this particular realization of model ii ,
the oscillatory behavior only starts around @xmath76 , avoiding the appearance of the gravitational instabilities that @xcite have shown to be a problem for early rapidly oscillating models with a negative averaged equation of state .
we add that the frequency of oscillations in @xmath40 is much larger in model ii than in model i , as shown in fig .
we note that a model with a higher @xmath62 enters the quadratic region of the potential sooner and thus the oscillations would start earlier , while an increase of @xmath52 has the opposite effect .
however , in fig . [ w2 ] the model with higher @xmath62 is the one where the oscillations start later .
this steems from the fact that we keep the ratio @xmath77 fixed in our analysis , and shows that the effect of @xmath52 is the dominating one .
euclid is a space mission scheduled to be launched by the european space agency in @xmath78 .
its main objective is to help better constrain the large - scale geometry of the universe and nature of the dark energy and dark matter components of the universe . in particular
, euclid aims to measure the dark energy parameters @xmath79 and @xmath80 with @xmath81 and @xmath82 accuracy , respectively ( @xcite ) , and will be able to constrain a large variety of dark energy models ( @xcite ) . to achieve its core scientific objectives , euclid will use the information contained in the baryonic acoustic oscillations ( bao ) present in the matter power spectrum and in the weak lensing ( wl ) features of the large - scale distribution of matter , acquiring spectroscopic and photometric data .
the formation of acoustic waves in the primordial photon - baryon plasma , in the early stages of our universe , is imprinted in the form of a series of peaks in the cosmic microwave background radiation ( cmb ) and in the large - scale distribution of matter ( bao ) .
it presents a characteristic scale , the sound horizon at recombination , @xmath83 , which can be accurately measured using present cmb data , such as that acquired by wmap ( @xcite ) .
its identification , both in the transverse , @xmath84 , and radial , @xmath85 , directions using the galaxy power spectrum , then allows the comoving distance , @xmath86 , and the hubble parameter , @xmath87 , as a function of redshift , to be constrained , given that ( @xcite ) as @xmath88 and @xmath89 in order to constrain the two quintessence models with euclid s spectroscopic survey , we will use the icosmo software package ( @xcite ) .
we first compute @xmath84 and @xmath85 given the models equations - of - state and then compute the fisher matrix .
the inverse of the fisher matrix contains lower limits for the errors of the estimated values for the models parameters , according to the cramer - rao theorem ( @xcite ) .
the bao fisher matrix combines the information on @xmath84 and @xmath85 and is given by ( @xcite ) @xmath90 here the sums run over the observational bins at different redshifts @xmath91 , @xmath92 represents the model parameter with respect to which the partial derivative is taken , while @xmath93 and @xmath94 represent the relative errors associated to the measurements of the transverse and radial scales , respectively .
we note that icosmo uses the analytical approximation of @xcite to evaluate the expected accuracy of an experiment s measurement of the bao scales .
we assumed the following euclid s spectroscopic wide survey specifications ( @xcite ) : the galaxy redshift distribution of @xcite , for a limiting flux of @xmath95 erg s@xmath96@xmath97 ; a median redshift of @xmath98 ; a redshift measurement error of @xmath99 ; and a redshift range of @xmath100 $ ] distributed over @xmath101 redshift bins .
weak lensing is a subtle gravitational lensing effect , caused by the slight deflection of the light emitted by distant galaxies due to the existing matter distribution along their line of sight .
the mapping between source , @xmath102 , and image , @xmath103 , planes can be approximated to first order through a simple matrix transformation ( @xcite ) @xmath104 where @xmath105 and @xmath106 are the components of the anisotropic shear and @xmath107 is the isotropic convergence .
the magnitude of the lensing signal , measured in shear correlation functions , depends both on the amount of matter along the line of sight , and on the distances between the observer , the lens and the source .
this makes weak lensing ideal for measuring the universe s mass distribution and geometry and , therefore , for constraining cosmological parameters , such as those that can be used to characterize the dark energy .
we use the icosmo software to forecast the wl constraints on our models .
we evaluate two - point correlation functions of the shear field on separate photometric redshift bins . the fisher matrix associated with this so - called weak lensing tomography
is given by @xmath108}}^{-1}_{jk}\frac{\partial c_{km}(l)}{\partial p^{\beta}},\ ] ] where @xmath109 is the cross - power spectrum between bins @xmath110 and @xmath111 for the multipole @xmath112 .
here @xmath92 represents the model parameter with respect to which the partial derivative is calculated , @xmath113 is the fraction of the sky covered by the survey , and @xmath114}}_{ij}$ ] is the covariance matrix for a given @xmath115 mode and the @xmath116 bin pair ( @xcite ) , which depend on the specifications of the euclid survey .
following @xcite , we use a number density of galaxies of @xmath117 per arcmin@xmath118 , a sky coverage of @xmath119 deg@xmath120 and a galaxy redshift distribution given by the analytical formula provided by @xcite , with a median redshift of @xmath121 and parameters @xmath122 and @xmath123 ( @xcite ) .
the photometric redshift measurement error is assumed to be @xmath124 , while the random mean - square intrinsic shear per component , @xmath125 , was taken to be @xmath126 ( @xcite ) .
assuming that the universe contains only photons , baryons , cold dark matter and a quintessence scalar field , as long as this last component does not dominate the total energy density of the universe at early times , namely before recombination , the transfer function of @xcite can be used to determine the power spectrum of the initial matter density perturbations .
we use thus this transfer function and evaluate the linear growth function for the quintessence equations - of - state of our models . in order to infer the non - linear corrections to the evolving linear power spectrum
, we used the halofit model ( @xcite ) , which may lead to some loss of precision when applied to dark energy models with evolving @xmath3 ( @xcite ) .
we compute auto and cross power spectra for @xmath127 redshift bins between @xmath128 and for a multipole range limited to @xmath129 to discard effects of baryonic feedback on the power spectrum ( @xcite ) .
in this section , we present the results for euclid s expected constraints on the parameters of the quintessence models under consideration .
given that there are nt a priori theoretically preferred values for these parameters , we will set their fiducial values to those that maximize the joint posterior probability distribution of the parameters given the most recent data available from the supernova cosmology project ( scp ) .
supernovae type ia are standard candles , since they reach essentially the same luminosity at their peak brightness .
the ratio of this constant peak luminosity to the measured flux is proportional to the luminosity distance squared .
if we also know the redshift at which a supernova occurs , we can then determine the relation between co - moving distance and redshift , which carries information on all the cosmological parameters that affect it .
namely , those that characterize the quintessence models we are considering . assuming that , in each of the quintessence models being considered , all possible values for the free parameters have equal prior probability , then the joint posterior probability distribution of the parameters for each model is proportional to the likelihood of the most recent scp data .
this is provided as @xmath130 where @xmath91 , @xmath60 and @xmath62 are , respectively , the redshift , ( measured ) apparent and ( assumed ) absolute magnitudes of the observed supernovae .
the quantity @xmath131 is related to the cosmology - dependent luminosity distance , @xmath132 ( in parsecs ) , through @xmath133 given that the uncertainty associated with the scp estimates of @xmath131 is assumed to be well characterized through a gaussian probability distribution , the likelihood of the most recent scp data is then proportional to the exponential of @xmath134 , where ( @xcite ) @xmath135{\ensuremath{\left[e^{-1}\right]}}_{ij}[f_j(p ) - \mu_j(z_i)].\ ] ] with the @xmath136 and @xmath137 representing , respectively , the scp data and the inverse of its covariance matrix , which includes systematic uncertainties .
the @xmath138 are the theoretically expected values for the @xmath139 , assuming the measured redshift @xmath140 and the set of parameters @xmath141 for the model @xmath142 to be true . for the model i ,
we have allowed @xmath38 , @xmath52 and @xmath41 to vary freely within the intervals @xmath143 $ ] , @xmath144 $ ] and @xmath145 $ ] , respectively . in the case of model ii , we have allowed @xmath38 and @xmath62 to vary freely within the intervals @xmath146 $ ] and @xmath147 $ ] , respectively .
the interval allowed for @xmath62 , as well as the conditions @xmath148 and @xmath149 , were set to force the scalar field to only start oscillating around @xmath76 , rendering the model viable as discussed before .
the present - day value of the hubble parameter was always assumed to be @xmath150 . for each model , the combination of parameters that maximizes the likelihood of the most recent scp data , and
hence the joint posterior given our assumptions of flat priors for the parameters , is * model i : @xmath151 , @xmath152 , @xmath153 * model ii : @xmath154 , @xmath155 the @xmath156 values of these two models that best - fit the scp data are @xmath157 and @xmath158 .
for comparison , we fitted the same data with @xmath159 models with similar number of free parameters , and found a value @xmath160 for its best - fit model .
the values are comparable , with the best - fit of model i fitting slghtly better the data than any @xmath159 model . in the subsequent analysis
, we take these best - fit values as our fiducial model parameter sets .
we will also always assume to be working on a spatially - flat universe , thus @xmath161 , where the power spectrum of primordial adiabatic gaussian matter density perturbations is approximately scale - invariant @xmath162 , the abundance of baryons is taken to be @xmath163 and the present - day amplitude of the matter density perturbations smoothed on a sphere with a radius of @xmath164 is equal to @xmath165 , motivated by the wmap-9 results ( @xcite ) . in fig .
[ h ] we show the evolution of the hubble parameter , @xmath166 , for the two fiducial models , normalized by @xmath166 in @xmath159 .
the deviation evolves with redshift and is within @xmath167 .
applying the fisher matrix formalism , we compute the marginalized constraints on the model parameters for the euclid forecasts .
two - dimensional @xmath168 and @xmath169 confidence regions , alongside with the one dimensional distributions are shown in fig .
[ constraints1 ] and fig .
[ constraints2 ] for wl and bao , respectively .
there is a striking difference between the wl and bao results , with cosmic shear being an order of magnitude more constraining , and dominating the joint constraints .
note , however , that the full information contained in the spectroscopically - derived matter power spectrum , @xmath170 , such as its full shape , amplitude and redshift distortions , that will be available from euclid ( @xcite ) , was not used in the analysis , and thus galaxy clustering data may still prvide complementary constraints for the models considered .
we find that the core model parameters , i.e , k , which determines the potential of model i , and m , which sets the starting redshift of the oscillations in model ii , will be constrained by euclid with around @xmath171 @xmath0 precision .
@xmath172{constraints1.png } \\
\includegraphics[scale = 0.42]{constraints2.png } \end{array}$ ] in the case of model i , we observe an anti - correlation between @xmath41 and @xmath173 .
this is expected , since a more negative @xmath41 would lead to earlier oscillations in the quintessence field , which could be prevented with higher values of @xmath173 . on the other hand , wl and bao
are complementary in the @xmath174 plane , showing a correlation and an anti - correlation , respectively .
@xmath172{constraints3.png } \\
\includegraphics[scale = 0.42]{constraints4.png } \end{array}$ ] in the case of model ii , there is a positive correlation between the two parameters . in general , the oscillations in the field start sooner both for larger @xmath62 and @xmath38 .
however , fixing the ratio @xmath77 , to keep the start of the oscillatory period roughly constant , produces the opposite behaviour as discussed earlier on .
hence , an increase of @xmath62 ( and the corresponding decrease of @xmath52 ) are compensated by a larger @xmath38 producing thus a positive correlation .
tables [ table1 ] and [ table2 ] summarize the @xmath0 constraints for models i and ii , respectively .
.[table1 ] the @xmath0 marginalized values for the first model parameters obtained by applying the weak lensing and bao methods , individually and jointly , to euclid s wide survey [ cols="^,^,^,^",options="header " , ]
in this paper we considered two physically - motivated quintessence models and forecasted their parameter constraints for the future euclid space mission .
the first model has a non - monotonical evolution of the dark energy equation - of - state , while in the second one the equation - of - state oscillates rapidly in the late universe , in contrast to the current paradigm of @xmath159 for which the equation of state takes a constant value of @xmath4 throughout cosmic evolution . in order to obtain plausible fiducial values for the parameters of the models
, we have used the most recent data available from the supernova cosmology project . for both cases we obtained fiducial values
consistent with an overall expansion history of the universe very similar to that of @xmath159 , although the recent evolution of the equation - of - state for both models is very different than the @xmath159 one .
moreover , the fiducial oscillating models were found to be as good a fit to the data as the concordance @xmath159 model , in agreement with the results of @xcite , where significant statistical evidence was found for a variety of phenomenological oscillating models ( depending on the information criterium used ) .
this encouraging result motivated us to pursue the analysis and test our models using probes of structure .
hence , using a fisher matrix approach , we considered euclid core probes , weak lensing and galaxy clustering ( bao only ) .
the impact of the dark energy oscillations on the cosmic shear signal is subtle , even with tomography , since oscillations as function of redshift are smeared out by integrations over @xmath175 .
nevertheless , we forecast euclid s cosmic shear data to have the power to discriminate them from the the concordance @xmath159 model .
this is in agreement with @xcite , where analysing phenomenological models , weak lensing was considered the best approach to discriminate dark energy oscillations .
we would like to thank henk hoekstra and martin kunz for their comments on this manuscript .
nal , ptpv and it acknowledge financial support from fundao para a cincia e a tecnologia ( fct ) , respectively , through grant sfrh / bd/85164/2012 , project ptdc / fis/111725/2009 , and grant sfrh / bpd/65122/2009 .
it acknowledges further fct support from project cern / fp/123618/2011 and pest - oe / fis / ui2751/2014 .
99 amin m.a . , zukin p. , bertschinger e. , 2012 , phys .
d , 85 , 103510 amendola l. , polarski d. , gannouji r. , tsujikawa , s. , 2007 , phys . rev .
d , 75 , 083504 amendola l. et al .
, 2013 , living rev .
relativity 16 , 6 basset b.a .
, hlozek r. , 2010 , in dark energy , pp .
246 - 278 , ed .
ruiz - lapuente p. , cambridge university press belloso a.b .
, garca - bellido j. , sapone d. , 2011 , jcap , 10 , 10 bennett c.l .
et al . , 2013 ,
apjs , 208 , 20 blake c. et al .
, 2006 , mnras , 365 , 255 caldwell r.r .
, dave r. , steinhardt p.j .
, 1998 , phys .
lett . , 80 , 1582 camera s. , carbone c. , moscardini l. , 2012 , jcap , 3 , 39 chiba t. , 2009 , phys .
d , 79 , 083517 clifton t. , ferreira p.g . , padilla a. , skordis c. , 2012 , physics reports , 513 , 1 dutta s. , scherrer r.j . , 2008a , phys . rev .
d 78 , 083512 dutta s. , scherrer r.j . , 2008b , phys .
d , 78 , 123525 dutta s. , saridakis e.n .
, scherrer r.j . , 2009 ,
d , 79 , 103005 eisenstein d.j .
et al . , 2005 , apj , 633 , 560 feng b. , li m. , piao y.s .
, zhang x. , 2006 , phys lett b , 634 , 101 geach j.e .
et al . , 2010 ,
mnras , 402 , 1330 gregory p. , bayesian logical data analysis for the physical sciences , cambridge university press hinshaw g. et al . , 2013 , apjs , 208 , 19 hoekstra h. , jain b. , 2008 , ann .
nucl . part .
sci . , 58 , 99 hu w. , eisenstein d.j . , 1998 , apj , 496 , 605 johnson m.c . , kamionkowski m. , 2008 , phys .
d , 78 , 063010 kessler r. et al .
, 2009 , apjs , 185 , 32 kowalski m. et al . , 2008 , apj , 686 , 749 lahav o. , liddle a.r .
, 2013 , arxiv:1401.1389 laureijs r. et al . , 2011 , euclid definition study report ( red book ) , esa / sre(2011)12 lazkoz r. , salzano v. , sendra i. , 2010 , phys lett b , 694 , 198 linder e.v . , 2003 , phys .
d , 90 , 091301 linder e.v .
, 2006 , astrop phys , 25 , 167 maartens r. , 2004 , living rev .
relativity , 7 , 7 mazumdar a. , rocher j. , 2011 , phys .
rept . , 497 , 85 mcdonald p. , trac h. , contaldi c. , 2006 , mnras , 366 , 547 pace f. , fedeli c. , moscardini l. , bartelmann m. , 2012 , mnras , 422 , 1186 padmanabhan t. , 2003 , phys .
rept . , 380 , 235 parkinson d. et al . , 2007 , mnras , 377 , 185 perlmutter s. et al . , 1999 , apj , 517 , 565 ratra b. , peebles p.j.e . , 1988 ,
d , 37 , 3406 refregier a. , 2003 , ara&a , 41 , 645 refregier a. , amara a. , kitching t.d .
, rassat a. , 2011 , a&a , 528 , a33 riess a.g .
et al . , 1998 ,
aj , 116 , 1009 scaramella r. et al , 2009 , in proceedings of wide field astronomy & technology for the square kilometre array ( skads 2009 ) , 15 , proceedings of science , sissa scherrer r.j . , sen a.a . , 2008 ,
d , 77 , 083515 semboloni e. , tereno i. , van waerbeke l. , heymans , c. , 2009 , mnras , 397 , 608 semboloni e. et al .
, 2011 , mnras , 417 , 2020 smail i. et al . , 1994 , mnras , 270 , 245 smith r.e .
et al . , 2003 ,
mnras , 341 , 1311 tegmark m. , taylor a.n .
, heavens a.f , 1997 , apj , 480 , 22 tsujikawa s. , 2013 , class . quant .
grav . , 30
, 214003 turner m.s . , 1983 ,
d , 28 , 1243 turner m.s . ,
white m. , 1997 , phys . rev .
d , 56 , 4439 weinberg d. h. et al .
, 2013 , phys .
rept . , 530 , 87 zlatev i. , wang l. , steinhardt p.j . , 1999 , phys . rev .
lett . , 82 , 896 | euclid is a future space - based mission that will constrain dark energy with unprecedented accuracy .
its photometric component is optimized for weak lensing studies , while the spectroscopic component is designed for baryon acoustic oscillations ( bao ) analysis .
we use the fisher matrix formalism to make forecasts on two quintessence dark energy models with a dynamical equation of state that leads to late - time oscillations in the expansion rate of the universe .
we find that weak lensing will place much stronger constraints than the bao , being able to discriminate between oscillating models by measuring the relevant parameters to @xmath0 precisions of 5 to @xmath1 .
the tight constraints suggest that euclid data could identify even quite small late - time oscillations in the expansion rate of the universe .
[ firstpage ] dark energy
quintessence
observational cosmology . |
so far the standard approach to baryon spectroscopy is the constituent quark model where the hamiltonian contains a spin independent part formed of the kinetic plus the confinement energies and a spin dependent part given by a hyperfine interaction .
the latter can be either due to one gluon exchange or to goldstone boson exchange between quarks , or it can be an instanton induced interaction . the results are naturally model dependent .
it is therefore very important to develop model independent methods that can help in alternatively understanding baryon spectroscopy and that can support quark model assumptions .
large @xmath1 qcd offers such a method . in 1974
t hooft proposed to generalize qcd from su(3 ) to su(@xmath1 ) @xcite where @xmath1 is an arbitrary number of colors and suggested a perturbative expansion in the parameter @xmath0 , applicable to all qcd regimes .
witten has generalized the approach to baryons @xcite and this has lead to a powerful @xmath0 expansion method to study static properties of baryons , as for example , the masses , the magnetic moments , the axial currents , etc .
the method is systematic and predictive .
it is based on the discovery that , in the limit @xmath2 , qcd possesses an exact contracted su(2@xmath3 ) symmetry @xcite where @xmath3 is the number of flavors .
this symmetry is only approximate for finite @xmath1 so that corrections have to be added in powers of @xmath0 .
the @xmath0 expansion method has extensively and successfully been applied to ground state baryons @xcite ( for recent developments see ref .
its applicability to excited states is a subject of current investigations . in this case
the symmetry under consideration is assumed to be su(2@xmath3 ) @xmath4 o(3 ) where su(2@xmath3 ) is related qcd , as introduced above .
however o(3 ) is not related to qcd but it brings an additional degree of freedom .
it is of common practice to introduce it in order to construct orbitally excited states .
the direct product su(2@xmath3 ) @xmath4 o(3 ) is also used in quark models to classify three quark states , but there su(2@xmath3 ) is not an intrinsic symmetry .
thus the two approaches have formally the same symmetry in common which does not imply common dynamical assumptions .
the only common feature is that the excited states are stable in a first approximation .
the purpose of the present study is to see whether or not there is a compatibility between the two methods .
if such a compatibility exists , an important support to the constituent quark model can be provided by the model independent @xmath0 expansion method , and a better understanding of the physical content of large @xmath1 mass formulas can be gained . in the language of quark models
, the baryon states can roughly be classified into excitation bands with @xmath5 for the ground state band and @xmath6 for excited states , where @xmath7 represents units of excitation , like in a harmonic oscillator picture .
the key tool of this comparative study is that one can analyze both the @xmath0 expansion results and the quark model basic ingredients in terms of @xmath7 which makes the comparison between the two methods possible and very convenient .
the paper is organized as follows .
the next section introduces the mass formula used in the @xmath0 expansion method .
section iii gives a mass formula obtained from a hamiltonian quark model where the kinetic energy is relativistic , the confinement is an y - junction flux tubes and the hyperfine interaction is of an one - gluon exchange nature .
section iv is devoted to the comparison between terms of the mass formula which are common in the two approaches .
the last section is devoted to conclusions .
the mass formula for the ground state up to order @xmath0 is simple because one can replace @xmath37 by @xmath36 , due to an identity which holds for symmetric @xmath42 $ ] states @xcite .
as there is no orbital excitation , the mass formula ( [ massoperator ] ) takes the following simple form @xmath43 which means that for @xmath5 only the operators @xmath44 and @xmath45 contribute to the mass .
thus the fit gives quantitative information only for @xmath46 and @xmath47 .
for @xmath48 , @xmath49 mev for @xmath50 , and @xmath51 mev for @xmath52 , one gets @xmath53 among the excited states , those belonging to the @xmath54 band , or equivalently to the @xmath30 $ ] multiplet , have been most extensively studied , either for @xmath55 @xcite or for @xmath56 @xcite . in the latter case , first order corrections in su(3 ) symmetry breaking were also included .
the @xmath57 band contains the @xmath58 $ ] , @xmath59 $ ] , @xmath60 $ ] ( @xmath61 = 0 , 2 ) and @xmath62 $ ] multiplets
. there are no physical resonances associated to @xmath62 $ ] .
the few studies related to the @xmath57 band concern the @xmath58 $ ] for @xmath3 = 2 @xcite , @xmath59 $ ] for @xmath56 @xcite , and @xmath60 $ ] for @xmath55 @xcite , later extended to @xmath56 @xcite .
the method has also been applied @xcite to highly excited non - strange and strange baryons belonging to @xmath63 $ ] , the lowest multiplet of the @xmath64 band @xcite .
the group theoretical similarity of excited symmetric states to the ground state makes the analysis of these states simple @xcite . for mixed symmetric states ,
the situation is more complex .
there is a standard procedure which reduces the study of mixed symmetric states to that of symmetric states .
this is achieved by the decoupling of the baryon into an excited quark and a symmetric core of @xmath65 quarks .
this procedure has been applied to the @xmath30 $ ] multiplet @xcite and to the @xmath60 $ ] ( @xmath61 = 0 , 2 ) multiplet @xcite .
in fact the decoupling is not necessary , provided one knows the matrix elements of the su(2@xmath3 ) generators between mixed symmetric states .
the case of su(4 ) has been presented in ref .
@xcite . in section iv
, we collect the values of @xmath46 , @xmath66 and @xmath47 obtained in the above studies in order to make a comparison between those values and their analogs resulting from the quark model described below .
in the framework of potential models , it is generally assumed that a baryon , viewed as a bound state of three quarks , can be described in a first approximation by the following spinless salpeter hamiltonian @xmath67 where @xmath68 is the current mass of the quark @xmath25 , and @xmath69 the confining interaction potential .
the nonperturbative part of the gluon exchanges , responsible for the confinement , can be successfully described in the flux tube model @xcite . in this framework , each quark is assumed to generate a string , or a flux tube , characterized by its energy density ( string tension ) .
recent developments in lattice qcd tend to confirm the y - junction as the correct configuration for the flux tubes in baryons @xcite . in this picture ,
a flux tube starts from each quark and the tubes meet at the toricelli point of the triangle formed by the three quarks .
this point , denoted by @xmath70 , is such that it minimizes the sum of the flux tube lengths , and its position is a complicated function of the quark coordinates @xmath71 .
moreover , the energy density of the tubes appears to be equal for mesons and baryons .
the y - junction potential reads @xmath72 in ref .
@xcite , it has been shown that this complicated potential is successfully approximated by the more easily computable expression @xmath73,\ ] ] where @xmath74 is the position of the center of mass . if @xmath75 , eq .
( [ pot1 ] ) is a simplified y - junction , where the toricelli point is replaced by the center of mass .
if @xmath76 , this interaction reduces to a @xmath77-type potential .
results of ref .
@xcite , obtained in the framework of a potential model , show that @xmath75 gives a better description than @xmath76 , and that the y - junction is approximated at best by @xmath78 close to @xmath79 .
let us now introduce auxiliary fields , in order to get rid of the square roots appearing in the hamiltonian ( [ ssh ] ) .
we get @xmath80\nonumber\\ & & + \alpha\sum^3_{j=1}\left[\frac { a^2 ( \vec{x}_j-\vec{r})^2}{2\nu_j}+ \frac{\nu_j}{2}\right]+\frac{(1-\alpha)}{2}\sum_{j < k}\left[\frac{a^2 ( \vec{x}_j-\vec{x}_k)^2}{2\lambda_{jk}}+\frac{\lambda_{jk}}{2}\right].\end{aligned}\ ] ] the auxiliary fields , denoted as @xmath81 , and @xmath82 are , strictly speaking , operators . although being formally simpler , @xmath83 is equivalent to @xmath84 up to the elimination of the auxiliary fields thanks to the constraints [ elim ] @xmath85 it is worth mentioning that @xmath86 can be seen as a dynamical mass of a quark of current mass @xmath68 , while @xmath87 is , in this case , the static energy of the straight string linking the quark @xmath25 to the toricelli point @xcite .
similarly , @xmath88 can be interpreted as the static energy of a straight string joining the quarks @xmath25 and @xmath89 .
although the auxiliary fields are operators , the calculations are considerably simplified if one considers them as real numbers .
they are then finally eliminated by a minimization of the masses with respect to them @xcite .
the extremal values of @xmath90 , @xmath91 , and @xmath82 , considered as numbers , are logically close to the values of @xmath86 , @xmath92 , and @xmath93 given by relations ( [ elim ] ) .
this procedure leads to a spectrum which is an upper bound of the true spectrum " ( computed without auxiliary fields ) @xcite : it can be shown that , the more auxiliary fields are introduced , the higher are the masses compared to those without auxiliary fields @xcite .
let us finally mention that , for @xmath75 , the hamiltonian ( [ ham3b ] ) can be related to the rotating string model for a baryon ( see for example ref .
@xcite ) . in ref .
@xcite , it has been shown that the eigenvalues of a hamiltonian of the form ( [ ham3b ] ) can be analytically found by making an appropriate change of variables , the quark coordinates @xmath94 being replaced by new coordinates @xmath95 .
the center of mass is defined as @xmath96 with @xmath97 and @xmath98 being the two relative coordinates . from ref .
@xcite , it can be immediately found that the mass spectrum of bound states of three massless particles ( @xmath99 for the @xmath100 and @xmath101 quarks ) is given by @xmath102 with @xmath103},\ ] ] @xmath104 and @xmath105 .
an obvious symmetry argument helps us to make the identification @xmath106 , @xmath107 , and @xmath108 . in this symmetric case ,
properties of the equilateral triangle together with the relations ( [ elim ] ) allow to make the following ansatz @xmath109 defining @xmath110 and @xmath111 we find @xmath112 formula ( [ mass3_s ] ) is clearly symmetric in @xmath113 and @xmath114 .
that means that we can set @xmath115 .
this equality can be viewed as a sort of virial theorem .
then we have @xmath116 one can easily find that the relation @xmath117 implies @xmath118^{1/2},\ ] ] and @xmath119 , as observed in ref .
writing explicitly the square mass , we see that the model of ref .
@xcite also predicts regge trajectories , which are in agreement with the experimental data for light baryons @xmath120 the regge slope is here given by @xmath121 .
however , from experiment we know that the regge slope for light baryons and light mesons are approximately equal . for light mesons ,
the exact value in the relativistic flux tube model is @xmath122 , a lower value than the one obtained from formula ( [ mass3q ] ) .
this is due to the auxiliary fields method : the more auxiliary fields we introduce , the more the masses are overestimated @xcite .
what can be done to cure this problem is to rescale @xmath123 : let us define @xmath124 such that @xmath125 .
then , formula ( [ mass3q ] ) is able to reproduce the light baryon regge slope for a physical value @xmath126 gev@xmath127 .
note that the best value for @xmath78 is @xmath79 .
consequently , the best value for @xmath128 is @xmath129 .
it is worth mentioning that such a rescaling of the string tension has already given good results in the study of hybrid mesons @xcite . although including only the confining energy is sufficient to understand the regge trajectories of light baryons , it is well - known that the absolute value of the masses which are obtained are too high with respect to the experimental data .
other contributions are needed to decrease these masses and we shall estimate their effect perturbatively .
the most widely used is a coulomb interaction term of the form @xmath130 arising from one gluon exchange processes , where @xmath131 is the strong coupling constant , usually assumed to be around @xmath132 for light hadrons @xcite .
lattice qcd calculations also support this value @xcite . assuming that @xmath133 , and using symmetry arguments , relations ( [ elim ] ) lead to @xmath134 @xmath135 another interesting contribution to the mass , which can be added perturbatively , is the quark - self energy .
recently , it was shown that the quark self - energy , which is created by the color magnetic moment of a quark propagating through the vacuum background field , adds a negative constant to the hadron masses @xcite .
its negative sign is due to the paramagnetic nature of the particular mechanism at work in this case .
the quark self - energy contribution for three massless quarks is given by @xcite @xmath136 the factor @xmath137 has been computed in lattice qcd studies .
first quenched calculations gave @xmath138 @xcite .
a more recent unquenched work @xcite gives @xmath139 . since
its value is still a matter of research , we will only assume that @xmath140 $ ] . with the unperturbed baryon mass @xmath141 , given by eq .
( [ mass3q ] ) , the total mass is given by the sum @xmath142 .
then , in the first order of perturbation and for @xmath143 , it is straightforward to obtain the following mass formula for baryons @xmath144 where the scaling @xmath125 has been used .
the effects of the one gluon exchange term and of the quark self - energy are thus to shift the square mass spectrum by a global negative amount .
let us note that the symbol @xmath7 defined by eq .
( [ n ] ) and the quantity @xmath7 used to classify baryon states and used to plot results from the @xmath0 expansion are the same .
this common @xmath7 will be used in the next section to perform a comparison between the results obtained in both approaches .
the mass formula ( [ massf ] ) does not take into account spin relativistic contributions , as the spin - spin or spin - orbit forces . within the auxiliary field formalism , all these corrections to the static potential
are expanded in powers of @xmath145 where @xmath113 is the constituent quark mass @xcite .
all the spin corrections to the mass formula ( [ massf ] ) must depend both on the matrix elements of the interaction and on the coefficient @xmath145 .
in the following , we shall consider that the dominant dynamical effect is due to the constituent mass , while the matrix elements remain roughly constant with @xmath7 , as presented in the next section .
in the @xmath0 expansion method , the first term @xmath146 in the mass formula of eq .
( [ massoperator ] ) contains the main spin - independent contribution to the baryon mass , which in a quark model language , represents the confinement and the kinetic energy .
so , it is natural to identify this term with the mass given by the formula ( [ massf ] ) .
then , for @xmath48 , we assume the relation @xmath147 which gives @xmath148 fig .
[ fig1 ] shows a comparison between the values of @xmath149 obtained in the @xmath0 expansion method and those derived from the eq .
( [ c1qm ] ) for various values of @xmath7 . from this comparison
one can see that the results of large @xmath1 qcd are entirely compatible with the formula ( [ c1qm ] ) . from a fit
, one has @xmath150
gev@xmath127 , a rather low but still acceptable value according to usual potential models , and @xmath151 gev@xmath127 . to reproduce @xmath152 , we can set @xmath153 , @xmath154 : these are very standard values . computed in the @xmath0 expansion ( full circles ) from a fit to experimental data ( eq . ( [ c2c4 ] ) for @xmath5 , refs .
@xcite for @xmath54 , ref .
@xcite for @xmath57 and ref .
@xcite for @xmath64 ) , compared with results from a fit ( see text ) of the formula ( [ c1qm ] ) ( empty circles and dotted line to guide the eyes ) .
no data is available for @xmath155 in large @xmath1 studies .
values of @xmath149 as predicted by formula ( [ c1qm ] ) for @xmath156 $ ] gev@xmath127 are located in the shaded area.,width=340 ] in most of the quark models however , the string tension is generally assumed to lie in the range @xmath157 $ ] gev@xmath127 . if the value of @xmath124 is chosen in this interval , the corresponding values for @xmath149 , given by eq .
( [ c1qm ] ) , are located in the shaded area of fig .
although the agreement with large @xmath1 data is not so good than in the optimal case , where @xmath158 gev@xmath127 , it remains satisfactory if we choose @xmath159 for @xmath160 gev@xmath127 , together with @xmath153 .
these values are larger than what is expected .
it could be argued that other mechanisms than the quark self - energy are present , their contribution decreasing the total mass @xmath161 . in mesons for example ,
retardation effects due to the finite interaction speed were shown to be also proportional to @xmath162 , like the quark self - energy @xcite .
it is possible that , when retardation effects are included , @xmath137 can again be chosen in the interval @xmath163 $ ] with a standard value of @xmath124 .
but , no model for retardation effects in baryon has been proposed yet .
computed in the @xmath0 expansion ( full circles ) from a fit to experimental data ( refs .
@xcite for @xmath54 , ref .
@xcite for @xmath57 and ref .
@xcite for @xmath64 ) , compared with results from formula ( [ c2c4qm ] ) ( empty circles and dotted line to guide the eyes ) .
no data is available for @xmath155 in large @xmath1 studies.,width=340 ] computed in the @xmath0 expansion ( full circles ) from a fit to experimental data ( eq . ( [ c2c4 ] ) for @xmath5 , refs .
@xcite for @xmath54 , ref .
@xcite for @xmath57 and ref .
@xcite for @xmath64 ) , compared with results from formula ( [ c2c4qm ] ) ( empty circles and dotted line to guide the eyes ) .
no data is available for @xmath155 in large @xmath1 studies.,width=340 ] within the auxiliary field formalism , we can expect that @xmath66 and @xmath164 , and thus @xmath165 we see that this behavior is coherent with the large @xmath1 results in figs .
[ fig2 ] and [ fig3 ] .
we chose @xmath166 mev so that the point with @xmath167 , for which the uncertainty is minimal , is exactly reproduced .
let us note that the spin - orbit term is vanishing for @xmath168 , so no large @xmath1 result is available in this case . to compute the parameter @xmath169 a fit
is performed on all the large @xmath1 data .
we obtain then @xmath170 mev .
note that @xmath171 .
this shows that the spin - spin contribution is much larger than the spin - orbit contribution , which justifies the neglect of the spin - orbit one in quark model studies .
this study supports the quark model basic assumptions by the compatibility of its mass formula with the mass formula derived from the model independent @xmath0 expansion .
these assumptions are : relativistic kinetic energy for light quarks , y - junction confining interaction , negligible spin - orbit interaction , hyperfine interaction dominated by a spin - spin term .
a recent analysis shows that a flux tube model and a feeble spin - orbit interaction give a succesful account of hadron spectroscopy @xcite .
in addition this study suggests that a good description of the bulk content of the baryon mass can be obtained with a spin independent energy eigenvalue of the form @xmath172 where @xmath173 is the number of excitation units , as in the harmonic oscillator .
it also shows that the spin - orbit and spin - spin interactions vanish with the excitation energy .
moreover this comparative study gives a better insight into the large @xmath1 mass operator where the coefficients @xmath19 encode the qcd dynamics .
g. t hooft , nucl . phys . *
b72 * , 461 ( 1974 ) .
e. witten , nucl .
* b160 * , 57 ( 1979 ) . j. l. gervais and b. sakita , phys .
lett . * 52 * , 87 ( 1984 ) ; phys .
d * 30 * , 1795 ( 1984 ) .
r. dashen and a. v. manohar , phys .
b * 315 * , 425 ( 1993 ) ; ibid , 438 ( 1993 ) .
y. koma , e. m. ilgenfritz , t. suzuki , and h. toki , phys .
d * 64 * , 014015 ( 2001 ) ; t. t. takahashi , h. matsufuru , y. nemoto , and h. suganuma , phys .
86 * , 18 ( 2001 ) .
b. silvestre - brac , c. semay , i. m. narodetskii , and a. i. veselov , eur .
j. c * 32 * , 385 ( 2004 ) .
w. lucha and f. f. schberl , phys .
a * 51 * , 4419 ( 1995 ) ; f. buisseret and c. semay , phys .
d * 70 * , 077501 ( 2004 ) . c. semay , b. silvestre - brac , and i. m. narodetskii , phys .
d * 69 * , 014003 ( 2004 ) .
w. lucha and f. f. schberl , phys .
a * 54 * , 3790 ( 1996 ) ; f. buisseret and c. semay , phys .
e * 71 * , 026705 ( 2005 ) .
f. buisseret and v. mathieu , eur .
j. a * 29 * , 343 ( 2006 ) .
a. simonov , phys .
* 66 * , 338 ( 2003 ) .
f. buisseret and c. semay , phys .
d * 73 * , 114011 ( 2006 ) .
a. b. kaidalov and yu .
a. simonov , phys .
b * 636 * , 101 ( 2006 ) .
f. buisseret and c. semay , phys .
d * 74 * , 114018 ( 2006 ) .
a. m. badalian and v. l. morgunov , phys .
d * 60 * , 116008 ( 1999 ) ; a. m. badalian and b. l. g. bakker , phys . rev .
d * 66 * , 034025 ( 2002 ) .
g. s. bali , phys .
rep . * 343 * , 1 ( 2001 ) . yu .
a. simonov , phys .
b * 515 * , 137 ( 2001 ) ; f. buisseret and c. semay , phys . rev .
d * 71 * , 034019 ( 2005 ) . a. di giacomo , h. panagopoulos , phys . lett .
b * 285 * , 133 ( 1992 ) .
a. di giacomo , yu . a. simonov , phys .
b * 595 * , 368 ( 2004 ) . v. l. morgunov , a. v. nefediev , and yu .
a. simonov , phys .
b * 459 * , 653 ( 1999 ) .
f. buisseret and c. semay , phys .
d * 72 * , 114004 ( 2005 ) .
a. selem and f. wilczek , [ hep - ph/0602128 ] . | a connection is establisehd between the results of a potential quark model with dynamical masses and the @xmath0 expansion mass formula used in the description of baryon resonances .
it is shown that a remarkable compatibility exists between the two methods . |
great progress has been made over the past few years in our observational understanding of galaxies at the end of the first billion years of the universe .
much of this progress has come at @xmath9 and has been due in large part to the substantial gains in surveying efficiency made in the optical by the advanced camera for surveys ( acs ) on hst .
this instrument provides us with a nearly @xmath10 increase in optical imaging efficiency over what was available with wfpc2 , higher resolution imaging , and a reasonably efficient , red @xmath11-band filter . putting together these capabilities with the well - established dropout technique ( e.g. , [ 27 ] ) , it became possible to select literally hundreds to thousands of galaxies at redshifts of @xmath9 @xcite .
experience over the past two years has shown that a simple @xmath12 color selection is remarkably effective at isolating galaxies at @xmath0 @xcite .
though a small fraction of the selected sources are found to be low mass stars and low - redshift galaxies , most of these contaminants can be eliminated by making some requirement on their stellarity ( i.e. , how pointlike they are ) or their flux in a bluer band ( e.g. , the @xmath13-band ) .
these results have now been verified spectroscopically down to @xmath14 @xcite . with this increasing understanding of the @xmath15-dropout selection , different studies attempted to construct rest - frame continuum @xmath16 ( @xmath17 ) luminosity functions ( lfs ) at @xmath0 @xcite . despite small differences in detail ,
each of these studies find that there were significantly fewer galaxies at @xmath0 ( per volume element ) than at @xmath2 .
a recent determination of this lf by our team is shown in figure [ fig : ilf]a and incorporates the results from some @xmath18 @xmath15-dropouts selected from the two goods fields ( enhanced to include the sne search data to be goods v2.0 ) , two acs parallel fields to the udf , and the udf @xcite .
for comparison , we have also plotted the @xmath2 lf @xcite , and it is amazing to note that the @xmath0 lf extends fainter than at @xmath2 , demonstrating the remarkable surveying efficiency of current optical instrumentation .
it is also evident that our @xmath0 lf shows a much larger deficit at the bright end than at the faint end , suggesting that higher redshift galaxies are of much lower luminosity ( on average ) .
this explains ( at least in part ) why early searches down to brighter limiting magnitudes @xcite found more substantial evolution than similar searches down to fainter magnitudes @xcite .
we can look at this evolution more quantitatively . in figure
[ fig : ilf]b , we plot the likelihood contours for the @xmath0 lf and contrast it with the equivalent values at @xmath2 ( thick black cross ) .
the red shaded regions on this plot indicate those values of @xmath19 and @xmath20 which are ruled out at 95% confidence using the current hst data . though the constraints on the faint - end slope @xmath19 at @xmath0 are still somewhat modest , the preferred values for the characteristic luminosity @xmath20 are much fainter than at @xmath2 , suggesting that there has been evolution in the characteristic scales on which star formation is occurring .
this provides us with one of our first , most direct evidences for the hierarchical buildup of galaxies early in the history of the universe ( see also [ 14 ] and [ 29 ] ) .
further refinements to the @xmath0 lf should be forthcoming over the next year due to the availability of two additional deep fields to be taken with acs ( go-10632 ) .
extending the dropout search beyond @xmath0 requires the detection of objects in the infrared .
this has been difficult due to the well - known limitations of infrared technology and because @xmath7 objects are likely to be very faint . at @xmath0 ,
the characteristic luminosity is nearly @xmath21 mags fainter than at @xmath2 .
extrapolated to even higher redshift , these luminosities should be even fainter still , perhaps 0.3 @xmath22 . in the @xmath23-band
, this corresponds to a magnitude of 27.3 , which can only be reached in deep nicmos studies and very deep ground - based studies around massive lensing clusters .
the deep nicmos imaging over the optical udf provided us with one of our first opportunities to find @xmath7 galaxies .
the @xmath24 limiting depths in these data were 27.6 in the @xmath25-band and 27.4 in the @xmath26-band .
moreover , the optical data for this field were more than sufficient to set strong constraints on the @xmath27-band fluxes . using a relatively aggressive set of detection criteria
, we carried out a @xmath27-dropout selection on these data and found 5 @xmath7 candidates .
successive tests on our selection including scattering experiments and selection on the negative images suggested that most of our 5 candidates were likely at @xmath7 and there was only one contaminant .
we therefore adopted as our likely sample 4 fiducial candidates . to assess the implications of this first statistical sample of @xmath7 objects
, we generated no - evolution predictions based upon a lower redshift @xmath28 @xmath29-dropout sample@xcite using our well - established cloning machinery @xcite .
an important consideration in projecting these lower redshift samples to high redshifts was the observed evolution in size @xcite and uv color @xcite .
running through these simulations , we estimated that 14 objects would be found ( if there was no - evolution from @xmath28 ) .
we compared this prediction with our 4 fiducial @xmath11-dropout candidates , given the expected small but non - zero contamination .
this suggested that the rest - frame uv ( @xmath30 ) luminosity density at @xmath7 was just @xmath31 that at @xmath28 ( number weighted ) or @xmath32 the @xmath28 value ( if we use a luminosity weighting ) .
though a first estimate of the rest - frame @xmath16 luminosity density at @xmath7 , our determination still suffered from some substantial uncertainties , notably the poissonian errors ( @xmath33% ) , cosmic variance ( factor of 2 ) , as well as the overall contamination level ( @xmath34 objects )
. this situation should improve substantially over the next year using data from two hst programs : ( 1 ) a deep @xmath27-dropout search in the field ( go-10632 ) and ( 2 ) a similar search around seven massive lensing clusters ( go-10504 and go-10699 ) .
both should yield @xmath35 @xmath7 candidates , substantially reducing uncertainties from our previously quoted estimates based on the hudf nicmos footprint .
simultaneously , searches with large ground - based telescopes are ongoing and have yielded a sizeable number of candidates , particularly around lensing clusters @xcite .
( @xmath17 ) lf estimated from the udf , the udf - ps , and the goods fields , shown in terms of the best - fit stepwise parameterizations ( _ red circles _ with @xmath36 errors ) and schechter function ( _ solid black line_)@xcite .
the @xmath2 lf is shown for comparison ( _ thick green line _ ) @xcite .
our @xmath0 lf shows a clear turnover at the bright end relative to the @xmath2 lf and suggests that there has been a shift in the characteristic luminosity from @xmath0 to @xmath2 .
_ ( right panel ) . _ the allowed regions of parameter space ( 95% confidence ) for the @xmath0 lf using current hst data @xcite .
the equivalent @xmath2 values are shown with the large black cross .
though only modest constraints are possible on the @xmath0 faint - end slope @xmath19 , the characteristic luminosity at @xmath0 appears to be @xmath37 mags fainter than at @xmath2.[fig : ilf],title="fig : " ] ( @xmath17 ) lf estimated from the udf , the udf - ps , and the goods fields , shown in terms of the best - fit stepwise parameterizations ( _ red circles _ with @xmath36 errors ) and schechter function ( _ solid black line_)@xcite .
the @xmath2 lf is shown for comparison ( _ thick green line _ ) @xcite .
our @xmath0 lf shows a clear turnover at the bright end relative to the @xmath2 lf and suggests that there has been a shift in the characteristic luminosity from @xmath0 to @xmath2 .
_ ( right panel ) . _ the allowed regions of parameter space ( 95% confidence ) for the @xmath0 lf using current hst data @xcite .
the equivalent @xmath2 values are shown with the large black cross .
though only modest constraints are possible on the @xmath0 faint - end slope @xmath19 , the characteristic luminosity at @xmath0 appears to be @xmath37 mags fainter than at @xmath2.[fig : ilf],title="fig : " ]
the detection and confirmation of galaxies at @xmath5 appears to be significantly more difficult than even at @xmath7 .
though the additional distance plays a small role , by far the biggest challenge is their luminosity : @xmath5 galaxies are expected to have very low luminosities , several times lower than at @xmath0 or @xmath7 @xcite . even assuming no evolution in luminosity from @xmath7 to @xmath5 , typical magnitudes for these objects would be @xmath3828 in the @xmath23-band , suggesting that one would need to probe to very faint magnitudes indeed .
one way of reaching such magnitudes is to use the very deep ir imaging capabilities of the hst nicmos camera .
unfortunately , even with this instrument , searches for @xmath5 objects are still extremely expensive , e.g. , @xmath39 orbits are needed to detect just one @xmath5 object , and this assumes that the uv lf does not evolve from @xmath5 to @xmath0 .
it is thus no surprise that there has been a lack of dedicated searches .
nevertheless , it is possible to take advantage of the deep @xmath25 and @xmath26 parallel data associated with several hst deep fields to conduct a search .
such data is available for the hdf - s ( 150 orbits ) and the udf ( @xmath38340 orbits ) .
there is also deep nicmos j+h imaging over the wfpc2 hdf - n ( 176 orbits ) and a portion of the acs udf ( 144 orbits ) .
the @xmath24 limiting magnitudes of these data range from @xmath40 to @xmath41 , which is sufficient to identify @xmath5 galaxies down to @xmath42 . to see what we could learn from these data
, we carried out a search for @xmath43 @xmath44-dropouts and found 11 objects @xcite .
we eliminated those objects with @xmath45 detections in the optical bands or with @xmath46 colors substantially redder than the typical starburst type object ( i.e. , uv continuum slopes @xmath47 ) .
this left us with 3 candidates , all of which are in the nicmos parallel fields to the udf . since we did not have comparably deep images blueward or redward of the @xmath48 passbands , it was difficult for us to be very sure about the redshifts of our 3 candidates . nevertheless , we could still compare these 3 candidates with the numbers expected assuming no - evolution from @xmath0 . repeating the simulations described in
3 , we estimated that 4.8 @xmath25-dropouts would be found , only somewhat higher than our 3 candidates .
if we assume that all three candidates were at @xmath5 , the normalization we would obtain for the @xmath5 lf is just @xmath49 that at @xmath0 . on the other hand , assuming that none of these candidates were at these redshifts , the normalization we would obtain is just @xmath50 the @xmath0 value ( @xmath36 ) .
remarkably , we found that these results were not very sensitive to cosmic variance ( @xmath51% rms ) .
this is due to the extremely large comoving distances probed in @xmath44-dropout searches ( @xmath52 mpc ) .
the current round of hst proposals should strengthen these constraints significantly . not only
will we increase the total nicmos observing time relevant to these searches by @xmath53% ( improving the statistics ) , but we will obtain some very deep acs observations ( @xmath54 mag at @xmath24 for apertures that match the nicmos data ) over the deep nicmos parallels containing our 3 @xmath5 candidates .
independent searches are also ongoing around lensing clusters ( go-10380 , go-10504 , gto-10699 , and [ 22 ] ) , and we note that there are claims by some teams to have a very good set of candidates .
current observations are now of sufficient quality to robustly determine the rest - frame @xmath16 ( @xmath55 ) luminosity function at @xmath0 .
these determinations extend to nearly three magnitudes below @xmath1 , fainter than has been possible at @xmath2 , demonstrating the efficiency of current optical technology .
substantial changes are evident in the lf relative to @xmath2 , suggesting that typical galaxies have more than doubled their star formation rates over this interval .
the observed evolution is suggestive of that expected from popular hierarchical models , and would seem to indicate that we are literally witnessing the buildup of galaxies over this range .
progress with higher redshift dropout samples has been less dramatic , but is still forthcoming .
our sample of @xmath7 @xmath27-dropouts in the hudf provided our first statistical sample of galaxies in this epoch and thus allowed an initial estimate of the luminosity density@xcite .
searches for higher redshift @xmath5 @xmath44-dropouts have resulted in some candidates as well as yielding some useful upper limits on the bright end of the @xmath16 lf @xcite .
galaxies at both times stand to undergo substantial improvements as the result of future and ongoing programs .
of course , for truly substantial gains in this area , we will need to wait for the availability of high resolution , high sensitivity space - based imagers , such as wfc3 or nircam ( jwst ) . | hst acs and nicmos data are now of sufficient depth and areal coverage to place strong constraints on the formation and evolution of galaxies during the first 1 - 2 gyrs of the universe . of particular interest are galaxies at @xmath0 since they represent the earliest epoch accessible to current high - efficiency optical instrumentation . to this end , we have constructed a sample of 506 @xmath0 objects from all the deepest wide - area hst data ( udf , udf - parallel , and goods fields ) .
they have been used to construct an optimal determination of the rest - frame continuum uv lf at @xmath0 .
our lf extends to over 3 magnitudes below @xmath1 , fainter than has been done at @xmath2 . over the interval @xmath0 to @xmath2
, we find strong evidence for evolution in the uv lf .
though constraints on the faint - end slope remain modest ( and are consistent with no - evolution ) , the characteristic luminosity appears to have approximately doubled over the interval @xmath0 to @xmath2 , consistent with hierarchical expectations .
remarkably , this shift to lower luminosities extends to even higher redshifts . using all deep @xmath3 nicmos observations ( 800 orbits in total ) ,
we have been able to demonstrate that the bright end of the lf ( @xmath4 ) is at least 5 times lower at @xmath5 than at @xmath6 , with a similar deficit being established from our recent detections and first statistical sample of @xmath7 galaxies using our udf nicmos data . in these proceedings
, we discuss what is known about the uv lf and uv luminosity density at @xmath8 from current data and its evolution relative to @xmath2 .
we also describe several exciting prospects for advance in this area over the next year . |
the exploration of physics with @xmath1-flavoured hadrons offers a very fertile testing ground for the standard model ( sm ) description of electroweak interactions .
one of the key problems to be studied is the phenomenon of cp violation , which , although already discovered in 1964 by christenson , cronin , fitch and turlay in the neutral kaon system @xcite , is still one of the experimentally least constrained phenomena .
another main topic is the study of rare @xmath1 decays induced by flavour changing neutral current ( fcnc ) transitions @xmath2 , which are loop - suppressed in the sm and thus very sensitive to new - physics effects . during the last few years
, @xmath0 physics has received a lot of attention , both from theorists and experimentalists , and we are presently at the beginning of the @xmath0-factory era in particle physics .
the babar ( slac ) , belle ( kek ) and hera - b ( desy ) detectors have already seen their first events , and cleo - iii ( cornell ) , cdf - ii and d0-ii ( fermilab ) will start taking data in the near future ( see @xcite for a recent experimental overview ) .
although the physics potential of these experiments is very promising , it may well be that the `` definite '' answer in the search for new physics in @xmath0 decays will be left for second - generation @xmath0 experiments at hadron machines . in the following
, we will give an overview of the @xmath0-physics potential of the lhc experiments atlas , cms and lhcb , with the main focus on sm physics . among the most interesting aspects and unsolved mysteries[p :
mystery ] of modern particle physics is the violation of cp symmetry .
studies of cp violation are particularly exciting , as they may open a window to the physics beyond the sm .
there are many interesting ways to explore cp violation , for instance through certain rare @xmath3- or @xmath4-meson decays ( a very recent comprehensive description of all aspects of cp symmetry and its violation can be found in ref .
. however , for testing the sm description of cp violation in a quantitative way , the @xmath0 system appears to be most promising @xcite . within the framework of the sm
, cp violation is closely related to the cabibbo
maskawa ( ckm ) matrix @xcite , connecting the electroweak eigenstates @xmath5 of the down , strange and bottom quarks with their mass eigenstates @xmath6 through the following unitary transformation : @xmath7 the elements of the ckm matrix describe charged - current couplings , as can be seen easily by expressing the nonleptonic charged - current interaction lagrangian in terms of the electroweak eigenstates ( [ ckm ] ) : @xmath8 where the gauge coupling @xmath9 is related to the gauge group su@xmath10(2 ) and the @xmath11 fields describe the charged @xmath12-bosons . in the case of three generations , three generalized cabibbo - type angles @xcite and a single _ complex phase _
@xcite are needed in order to parametrize the ckm matrix .
this complex phase allows one to accommodate cp violation in the sm , as was pointed out by kobayashi and maskawa in 1973 @xcite .
a closer look shows that cp - violating observables are proportional to the following combination of ckm matrix elements @xcite : @xmath13 which represents a measure of the `` strength '' of cp violation in the sm . since @xmath14
, cp violation is a small effect .
however , in scenarios of new physics @xcite , typically several additional complex couplings are present , leading to new sources of cp violation .
as far as phenomenological applications are concerned , the following parametrization of the ckm matrix , the `` wolfenstein parametrization '' @xcite , which corresponds to a phenomenological expansion in powers of the small quantity @xmath15 , turns out to be very useful : @xmath16 the terms of @xmath17 can be taken into account systematically @xcite , and will play an important rle below .
concerning tests of the ckm picture of cp violation , the central targets are the _ unitarity triangles _ of the ckm matrix .
the unitarity of the ckm matrix , which is described by @xmath18 leads to a set of 12 equations , consisting of 6 normalization and 6 orthogonality relations .
the latter can be represented as 6 triangles in the complex plane , which all have the same area @xcite .
however , in only two of them , all three sides are of comparable magnitude @xmath19 , while in the remaining ones , one side is suppressed relative to the others by @xmath20 or @xmath17 .
the orthogonality relations describing the non - squashed triangles are given as follows : @xmath21 the two non - squashed triangles agree at leading order in the wolfenstein expansion , i.e. at @xmath19 , so that we actually have to deal with a single triangle at this order , which is usually referred to as `` the '' unitarity triangle of the ckm matrix @xcite .
however , in the lhc era , the experimental accuracy will be so tremendous that we will have to take into account the next - to - leading order terms of the wolfenstein expansion , and distinguish between the unitarity triangles described by ( [ ut1 ] ) and ( [ ut2 ] ) , which are illustrated in fig .
[ fig : ut ] . here , @xmath22 and @xmath23 are related to the wolfenstein parameters @xmath24 and @xmath25 through @xcite @xmath26 note the angles of the triangles , in particular those designated by @xmath27 , @xmath28 , @xmath29 and @xmath30 .
these will be referred to frequently throughout this report .
the sides @xmath31 and @xmath32 of the unitarity triangle shown in fig .
[ fig : ut](a ) are given as follows : @xmath33 and will also appear in the following discussion . [ cols= " < , > " , ] among the rare @xmath34 decay processes that have been discussed in the literature
are the radiative decays @xmath35 and @xmath36 , induced at the quark level by the @xmath37 and @xmath38 transitions , respectively @xcite .
the interest for the former decay mode is related to the possibility of studying the @xmath39 electromagnetic penguin transition , which in the charm mesons is overwhelmed by long - distance contributions . in the case of @xmath34 , long- and short - distance contributions
have been estimated to be of comparable size , and the branching fraction @xmath40 is predicted , in the sm , at the level of @xmath41 .
@xmath34 decays can give information about cp violation and the weak ckm phases .
promising channels are @xmath42 , in particular the one where the charmonium state is a @xmath43 , whose decay mode to @xmath44 can be easily identified . in this case , cp violation is due to the difference between the weak phases of the tree and penguin diagrams contributing to the decay . the cp asymmetry @xmath45 has been estimated : @xmath46 @xcite .
interesting channels are also those having a light meson in the final state , e.g. @xmath47 and @xmath48 . however , in this case the sizeable rle played by the annihilation mechanism makes it difficult to predict the decay rates and the cp asymmetries .
decay modes such as @xmath49 can also be considered , although considerable difficulties would be met in the experimental detection of @xmath50 and in the removal of the background from @xmath51 decays . finally , the decay @xmath52 has been proposed as an interesting source of flavour - tagged @xmath53 mesons for the study of mixing and cp violation in the @xmath53 sector @xcite .
atlas have studied the reconstruction of @xmath34 mesons using the decays @xmath54 and @xmath55 , with @xmath56 ( see @xcite ) . for this study ,
the following branching ratios have been assumed : @xmath57 and @xmath58 .
it is estimated that after 3 years of running at low luminosity , it will be possible to fully reconstruct 12000 @xmath59 events and 3@xmath6010@xmath61 events in the @xmath62 channel .
the statistics would allow a very precise determination of the @xmath34 mass and lifetime .
the @xmath34 meson is of particular interest as a unique case to study the impact of qcd dynamics on weak decays .
applications in flavour physics ( ckm parameters , rare decays , @xmath53 flavour tagging ) have also been considered in the literature .
important theoretical questions that need further attention are the issues of quark - hadron duality for inclusive decays and , for exclusive modes , the importance of corrections to the heavy - quark and nonrelativistic limits , as well as corrections to the factorization approximation .
the experimental feasibility for various observables needs likewise to be assessed in more detail .
the aim of the present section has been to give a flavour of the special opportunities that exist , from a theoretical perspective , in studying the physics of the @xmath34 .
some of these are realizable at the lhc , where it will be possible to investigate also the production , spectrum , lifetimes and decays of baryons containing two heavy quarks @xcite .
it is to be hoped that the results summarized in this section will trigger more detailed experimental studies .
the studies presented at and initiated by the workshop have clearly shown that the lhc is very well equipped and prepared to pursue a rigorous @xmath1 physics programme .
the main emphasis in the studies presented here has been on exploring lhc s potential for measuring cp violating phenomena and , on the theory side , a meaningful extraction of information on the underlying mechanism on cp violation in the sm .
most of the presented `` strategies '' aim at extracting the three angles of the unitarity triangle , @xmath27 , @xmath28 and @xmath29 , as well as @xmath63 , in as many different ways as possible ; any significant discrepancy between the extracted values or with the known lengths of the sides of the triangle would constitute evidence for new physics .
apart from detailed studies of the @xmath64@xmath65 b factory `` benchmark modes '' , also the hadron collider `` gold - plated '' mode @xmath66 has been studied , and new strategies for measuring @xmath28 and @xmath29 from @xmath53 decays , which can not be accessed at @xmath64@xmath65 b factories , have been developed .
we conclude that the three experiments are well prepared to solve the `` mystery of cp violation '' ( p. ) .
another important goal to be pursued is the measurement of b mixing parameters , and the studies summarized i m this report make clear that all 3 lhc experiments have excellent potential .
there is sensitivity in one year s operation to a mass difference in the @xmath53 system far beyond the sm expectation , and similarly good prospects for a rapid measurement of the width difference .
the second focus of the workshop was the assessment of lhc s reach in rare decays .
the discussion centered on decay modes with the favourable experimental signature of two muons or one photon in the final state .
it has been demonstrated that the decay @xmath67 with a sm branching ratio of @xmath68 can be seen within one year s running .
it has also been shown that decay spectra of semimuonic rare decays like @xmath69 are accessible , which opens the possibility to extract information on short - distance ( new ) physics in a theoretically controlled way .
lhc s full potential for rare decays has , however , not yet been fully plumbed , and further studies , in particular about the feasibility of inclusive measurements , are ongoing . of the many other possible @xmath1-physics topics ,
only a few could be marked out , and we have reported some recent developments in the theoretical description of nonleptonic decays and discussed a few issues in @xmath34 physics . the exploration of other exciting topics , such as physics with @xmath1 baryons or ( non - rare ) semileptonic decays , to name only a few , has to await a second round of workshops . for reviews , see , for instance , y. grossman , y. nir and r. rattazzi , hep ph/9701231 ; + m. gronau and d. london , phys . rev . * d55 * , 2845 ( 1997 ) ; + y. nir and h.r .
quinn , annu .
nucl . part .
sci . * 42 * , 211 ( 1992 ) ; + r. fleischer , hep ph/9709291 ; + l. wolfenstein , phys . rev .
* d57 * , 6857 ( 1998 ) . the atlas coll .
, _ atlas detector and physics performance technical design report , _ cern / lhcc/99 - 14 and cern / lhcc/99 - 15 . in particular , the _ inner detector _
is described in vol .
i , chapter 3 , pp .
5398 and the b - physics performance in vol .
ii , chapter 17 , pp .
561618 .
the atlas coll .
, _ inner detector technical design report , _ cern / lhcc/97 - 16 and cern / lhcc/97 - 17 ; _ pixel detector technical design report , _ cern / lhcc/98 - 13 . the cms coll . , _ the tracker system project technical design report , _ cern / lhcc/98 - 6 . the atlas coll . , _
calorimeter performance technical design report , _ cern / lhcc/96 - 40 ; _ liquid argon calorimeter technical design report , _ cern / lhcc/96 - 41 ; _ tile calorimeter technical design report , _ cern / lhcc/96 - 42 . the cms coll . , _ the hadronic calorimeter technical design report , _ cern / lhcc/97 - 31 ; _ the electromagnetic calorimeter technical design report , _ cern / lhcc/97 - 33 . the atlas coll . , _ muon spectrometer technical design report , _ cern / lhcc/97 - 22 . the cms coll . , _ muon technical design report , _ cern / lhcc/97 - 32
. t. sjstrand , comp .
* 82 * , 74 ( 1994 ) .
see the chapter on b production in this book , hep
ph/0003142 .
the atlas coll .
, _ trigger performance status report , _ cern / lhcc/98 - 15 . g. wrochna , cms note cms - cr-1996 - 002 . the atlas coll . , _ level-1 trigger technical design report , _ cern / lhcc/98 - 14
. j. pilszka and g. wrochna , cms note cms - note-1998 - 075 .
the atlas coll . , _
daq , event filter , level-2 and dcs technical progress report , _ cern / lhcc/98 - 16 .
y. coadou et al . , atlas note atl - phys-99 - 022 .
k. ackerstaff et al .
( opal coll . ) , eur .
j. * c5 * , 1998 ( 379 ) ; + f. abe et al .
( cdf coll . ) , phys .
81 * , 1998 ( 5513 ) ( for an updated analysis , see preprint cdf / pub / bottom / cdf/4855 ) ; + r. forty et al .
( aleph coll . ) , preprint aleph 99 - 099 .
m. gronau , phys . lett .
* b300 * , 163 ( 1993 ) ; + j.p . silva and l. wolfenstein , phys .
* d49 * , r1151 ( 1994 ) ; + r. aleksan et al . ,
* b356 * , 95 ( 1995 ) ; + a.j .
buras and r. fleischer , phys .
b360 * , 138 ( 1995 ) ; + f. dejongh and p. sphicas , phys .
rev . * d53 * , 4930 ( 1996 ) ; + m. ciuchini et al . ,
nucl . phys .
* b501 * , 271 ( 1997 ) ; + p.s .
marrocchesi and n. paver , int .
a13 * , 251 ( 1998 ) ; + a. ali , g. kramer and c.d .
l , phys .
rev . * d59 * , 014005 ( 1999 ) .
j. schwinger , phys .
* 12 * , 630 ( 1964 ) ; + r.p .
feynman , in `` symmetries in particle physics '' , ed .
a. zichichi ( acad . press 1965 ) ; + o. haan and b. stech , nucl .
* b22 * , 448 ( 1970 ) ; + d. fakirov and b. stech , nucl .
* b133 * , 315 ( 1978 ) ; + l.l .
chau , phys .
rept . * 95 * , 1 ( 1983 ) .
l. wolfenstein , phys .
rev . * d52 * , 537 ( 1995 ) ; + j .-
grard and j. weyers , eur .
j. * c7 * , 1 ( 1999 ) ; + a.f .
falk et al . , phys . rev . * d57 * , 4290 ( 1998 ) ; + d. atwood and a. soni , phys .
rev . * d58 * , 036005 ( 1998 ) .
r. fleischer and t. mannel , hep
ph/9706261 ; + r. fleischer , in ref .
@xcite ; m. neubert , in ref .
@xcite ; + d. choudhury , b. dutta and a. kundu , phys . lett .
* b456 * , 185 ( 1999 ) ; + x.g .
he , c.l .
hsueh and j.q .
shi , phys .
lett . * 84 * , 18 ( 2000 ) .
d. melikhov , n. nikitin and s. simula , phys .
* d57 * , 6814 ( 1998 ) ; phys . lett .
* b430 * , 332 ( 1998 ) ; phys .
* b442 * , 381 ( 1998 ) ; + t.m .
aliev , c.s .
kim and y.g .
kim , hep
ph/9910501 ; + c.s .
kim , y.g .
kim and c.d .
ph/0001151 .
s. godfrey and n. isgur , phys .
* d32 * , 189 ( 1985 ) ; + p. colangelo , g. nardulli and m. pietroni , phys .
* d43 * , 3002 ( 1991 ) ; + e. eichten and c. quigg , phys .
* d49 * , 5845 ( 1994 ) ; + s.s .
gershtein et al . , phys .
* d51 * , 3613 ( 1995 ) ; + l.p .
fulcher , phys .
* d60 * , 074006 ( 1999 ) .
e. bagan et al .
* b287 * , 176 ( 1992 ) ; phys .
* b301 * , 243 ( 1993 ) ; + d.b .
lichtenberg , r. roncaglia and e. predazzi , phys .
* d52 * , 1722 ( 1995 ) ; + s.s .
gershtein et al . , heavy ion phys .
* 9 * , 133 ( 1999 ) ; + v.v .
kiselev , a.k .
likhoded and a.i .
onishchenko , hep ph/9901224 . | we review the prospects for @xmath0-decay physics at the lhc as discussed in the 1999 workshop on standard model physics at the lhc .
+ + cern
th/2000101 + |
entanglement is one of the most profound features of quantum mechanics and has been considered as an important resource in quantum information processing ( qip ) including teleportation,@xmath0}$ ] cryptography and secure direct communication,@xmath1}$ ] and quantum communication and computation.@xmath2}$ ] in order to realize quantum information process , great effort has been devoted to the generation of entanglement in linear optics,@xmath3}$ ] cavity qed @xmath4}$ ] and ion trap@xmath5}$ ] schemes . in recent years
much attention has been focused on the spin system with heisenberg exchange interaction , which is a typical quantum system , and the entanglement properties have been extensively investigated in spin systems , such as ising model@xmath6}$ ] and isotropic and anisotropic heisenberg models.@xmath7}$ ] however ; as far as we know , most discussions mentioned above merely focused on the models with nearest neighbour interaction of exchange couplings , while next - nearest neighbour interaction has not been taken into account .
dimerized systems and frustrated systems with next - nearest - neighbour ( nnn ) interactions@xmath8}$ ] play an important role in condensed matter theory , and entanglement in dimerized and frustrated systems has been considered.@xmath9}$ ] sun et al@xmath10}$ ] investigate entanglement properties in dimerized and frustrated spin - one models by applying the concept of negativity . in ref.[20 ] , entanglement is studied in an open alternating chain of nuclear spins @xmath11 with spin - spin couplings in an external magnetic field under the thermodynamic equilibrium conditions .
gu et al@xmath12}$ ] have investigated entanglement in frustrated spin - half heisenberg chains .
chen yan et al@xmath13}$ ] analyzed sublattice entanglement and quantum phase transitions in antiferromagnetic spin chains , chen shu et al@xmath14}$ ] study the fidelity and quantum phase transition for the heisenberg chain with next - nearest - neighbouring interaction .
in fact , some one - dimensional and two - dimensional compounds ( @xmath15 ) have manifested such interactions .
therefore , it is worthwhile to include next - nearest neighbouring interaction in the studies of spin chain entanglement .
+ the 1d heisenberg model is a simple but realistic and extensively studied in solid state system .
recently , osterloh et al @xmath16}$ ] examined the entanglement between two spins of position i and j in the spin chains for the pure case , huang et al,@xmath17}$ ] osenda etal@xmath18}$ ] and we@xmath19}$ ] have demonstrated that entanglement can be controlled by introducing impurities into the systems .
meanwhile , the entanglement shares many features with quantum phase transition ( qpt ) for a many - body system.@xmath20}$ ] qpt , which occurs at absolute zero temperature and is purely driven by quantum fluctuation , is the structural change in the properties of the ground state .
the associated level crossings lead to the presence of non - analyticities in the energy spectrum .
therefore , the knowledge about the ground - state entanglement , the nonlocal correlation in quantum systems , is considered as a key to further understand qpt . in this paper
, we study the pairwise ground state entanglement between the nearest - neighbor sites and that of the next - nearest neighbouring sites in one - dimensional @xmath21 heisenberg xy spin chain with dimerised exchange couplings and next - nearest neighbour coupling , to our knowledge , which has not been reported before .
the present study will help us to further understand the behavior of the entanglement and qpt in one - dimensional alternating heisenberg xy model with next - nearest neighbour interactions , and this model can display a variety of interesting physical phenomena .
more interestingly , we can control or manipulate the nne and nnne in qpt point with the help of dimerised exchange couplings , dimerised external magnetic fields and frustrated exchange interaction .
we consider one - dimensional alternating heisenberg xy model of n spin-@xmath22 particles with nearest - neighbour interactions and next - nearest neighbour interactions .
one - dimensional hamiltonian can be written as @xmath23 where @xmath24 and @xmath25 are alternating exchange interaction and nearest - neighbour interactions respectively , @xmath26 is the strength of external magnetic field on site i , @xmath27 are the pauli matrices , @xmath28 is a dimensionless parameter characterizing the anisotropy of the model and n is the total number of sites .
furthermore , the periodic boundary conditions satisfy @xmath29 . + the spectrum of this hamiltonian can be determined exactly by a straightforward application of the standard methods .
the first step in the procedure is to perform a jordan - wigner transformation@xmath30}$ ] by introducing fermi operators @xmath31 , @xmath32 , as a result , the hamiltonian ( 1 ) is mapped in the free fermion hamiltonian @xmath33 - 2\sum_{i=1}^{n}h_{i}(c_{i}^{+}c_{i}-\frac { 1}{2})\nonumber\\&-&\sum_{i=1}^{n}j_{i , i+2}[(c_{i}^{+}c_{i+2}+\gamma c_{i}^{+}c_{i+2}^{+})+h.c]\end{aligned}\ ] ] in this paper , the exchange interaction and external magnetic field have the form @xmath34 , @xmath35 , @xmath36 , @xmath37 respectively , where @xmath38 introduces the dimerised parameter , the ratio of next - nearest ( @xmath39 ) to nearest - neighbor interaction coefficients @xmath40 is called the frustration parameter . @xmath41 and @xmath42 correspond to the antiferromagnetic and the ferromagnetic cases , respectively . for @xmath43 , eq .
( 2 ) reduces to one - dimensional alternating(dimerised ) heisenberg model , for @xmath44 , it reduces to one - dimensional frustrated heisenberg model .
+ the density - matrix renormalization - group ( dmrg)method@xmath45}$]is applied to obtain the spin correlation function . the two - point reduced density matrix obtained by tracing the full density matrix of the system over all sites except the pair @xmath46 has the form @xmath47 in this part , we choose the concurrence defined by wootters @xmath48}$ ] as a measurement of the pairwise entanglement . for a pure or mixed state of two qubits described by the density matrix @xmath49
, the concurrence c may be calculated explicitly as @xmath50 where @xmath51 , @xmath52 , @xmath53 , @xmath54 are the eigenvalues in descending order of the spin - flipped density operator r , which is defined by @xmath55 , where @xmath56 , @xmath57 denotes the complex conjugate of @xmath49 , @xmath58 is the usual pauli matrix .
all the matrix elements in the density matrix can be calculated from the different spin - spin correlation functions .
in this paper , we focus our discussion on the transverse ising model with @xmath59 and introduce the dimensionless parameter @xmath60 .
the goal of our present study is to find dynamic characteristics of the entanglement in varying the dimerised parameter of exchange couplings , external magnetic fields and nnn interaction .
figure 1 displays the diagram of the @xmath61 model which can be considered either as a linear chain(a ) or a zigzag chain(b ) .
the alternating nearest - neighbouring interaction is @xmath62 and @xmath63 respectively , while the next - nearest - neighbouring interaction is parameterized by @xmath64 .
we consider three different physical models , i.e , dimerized ferromagnetic spin chain , frustrated ferromagnetic - ferromagnetic(f - f ) spin chain , frustrated ferromagnetic - antiferromagnetic(f - af ) spin chain .
first , we examine the dimerized heisenberg chain , this model is characterized by an alternation of strong and weak bonds between two nearest - neighbouring spins .
figure 2 shows the nearest neighbouring concurrence c(1,2 ) and c(2,3 ) , which correspond to two nearest - neighbouring spins coupled by bonds @xmath62 and @xmath63 respectively , as a function of the reduced coupling constant @xmath65 at different values of the dimerised parameter @xmath66 of exchange couplings , parameter @xmath67 of external magnetic fields with the system size n = 59 . numerical results in figs.2(a ) and 2(c )
show that the concurrence c(1,2 ) decreases when the dimerised parameter @xmath66 of exchange couplings increases , while for the concurrence c(2,3 ) the situation is opposite , and the pair of qubits ( 1,2 ) and ( 2,3 ) have the same pairwise entanglement for the homogeneous chain @xmath68 .
in fact , in the case of weak alternation of the nearest - neighbor exchange coupling , we have here the dimerised spin chain which can be considered qualitatively as a set of non - interacting spin pairs . in the limit of strong dimerization ,
the concurrence reaches a maximum value .
moreover , spin 2 can be entangled both with spin 1 and with spin 3 . since spin 2
is strongly entangled with spin 1 , the entanglement of spins 2 and 3 is weaker .
obviously , the increase of exchange interaction @xmath63 suppresses the entanglement of spins coupled by @xmath62 .
the increase of @xmath63 can enhance the entanglement between spins coupled by @xmath63 .
this explains the behavior of the concurrence displayed in fig.2 .
the effect of the alternation of external magnetic field @xmath67 is also shown in figs.2(b ) and 2(d ) .
however , different from the effect of the exchange couplings , the concurrence ( 1,2 ) and ( 2,3 ) have similar behavior and increase firstly and then decrease with increasing the value of the parameter @xmath67 , suggesting a critical alternation of external magnetic field @xmath69 , where the maximum concurrence occurs , must be exist . + from fig.2 , we can see that the alternate interaction plays an important role in enhancing the nearest neighbouring concurrence .
the maximum value of the concurrence between neighboring sites does not occur at the critical point .
the reasons are based on the properties of shared entanglement to expect that this maximum should occur away from the critical point .
entanglement sharing is relevant to the quantum phase transition in the transverse ising model as it provides a fundamental bound on the amount of entanglement that may be distributed among the other sites , which means that as the overall entanglement in the lattice is increased , some sites become disentangled . however , as we all know , the maximum value of the concurrence between the next - nearest - neighbouring sites does occur at the critical point @xmath70 , so it is necessary to discuss in this subsection the effect on the next - nearest - neighbouring entanglement(nnne ) .
figure 3(a ) and 3(b ) show the change of concurrence c(1,3 ) as a function of @xmath65 for different values of alternate interactions @xmath66 and @xmath67 with @xmath71 , i.e. in the absence of nnn interaction .
we can see that there is no entanglement at @xmath72 and @xmath73 .
the weak alternate interactions @xmath66 and @xmath67 suppress and finally completely destroy the nnn concurrence . the dashed line in fig.3(a )
depicts at the critical point , @xmath70 , of the transverse ising model(@xmath68 ) , which maximum next - nearest - neighbouring entanglement occurs , there is a fundamental transition in the structure of the ground state .
this is consistent with the former result .
as @xmath66 increases the nnn concurrence tends to increase faster , the peak value induced by the nn interaction increases and the @xmath74 , where concurrence approaches a maximum , shifts to left very rapidly .
the dimerized interaction does assist the entanglement of formation because of the inside alternation .
therefore , we can further understand the relation between the entanglement and quantum transition . in figs.3(c ) and 3(d ) , we plotted our numerical results for the threshold nn alternate interactions @xmath66 and @xmath67 for the case of next - nearest - neighboring concurrence . only above the critical @xmath66 and @xmath67 , nnn
concurrence appears .
+ for the frustrated model , there is a competition between the nn and nnn interactions .
so the model displays some special features of entanglement . in fig.4 , we give the results of the nn and nnn concurrence as a function of the parameter @xmath65 for different frustrated interaction @xmath75 .
the nnn interaction has a different frustration effect on c(1,2 ) . for frustrated ferromagnetic - ferromagnetic(f - f )
spin chain @xmath76 , by the comparison among the different curves in fig.4(a ) , it is interesting to find that the entanglement peak between the nearest neighbours in the dotted line increase to a value larger than those in the solid line . with the increasing of @xmath75 , in the dash - dotted line , the concurrence decreases .
we can imagine there must be a critical frustration strength ( @xmath77 ) , below @xmath77 , nnn interaction enhances entanglement , while above @xmath77 , nnn interaction shrinks entanglement .
this behaviour is due to the energy level cross at the point @xmath78 , as seen clearly from fig.4(a ) . before and after the critical point ,
the entanglements display distinct behaviours .
however , for frustrated ferromagnetic - antiferromagnetic(f - af ) spin chain @xmath79 , the situation becomes different due to the frustration , with the increase in absolute value of @xmath75 and weak parameter @xmath65 , antiferromagnetic nnn interaction always shrinks entanglement , which is different from the results obtained from fig.4(a ) .
another important character revealed by fig.4(b ) is that , as parameter @xmath65 increases , the concurrence increases and tends to move to infinity(@xmath80 ) by increasing the value of the parameter @xmath75 , the strong @xmath75 is helpful to keeping the better entanglement .
it is the frustrated interaction that leads to considerable different evolutions of the entanglement , hence the entanglement is rather sensitive to any change with the nnn interaction . from figs.4(c ) and 4(d ) , we can see that the height of the peak increases with the increase of antiferromagnetic @xmath75 , which is a characteristic feature of one dimensional heisenberg antiferromagnetic chain .
this suggests that the antiferromagnetic component becomes significant for larger @xmath75 .
meanwhile , the antiferromagnetic nnn interaction results to a larger value of nnne in comparison to the case when the nnn interaction is ferromagnetic . in order to explain these results it is necessary to take into account the influence of the competing roles played by nn and nnn interaction on the entangled states .
thus , by adjusting nnn interaction constant , one can control or manipulate the nn and nnn entanglement .
+ in summary , from the above analysis , it is clearly noted that dimerised exchange couplings , dimerised external magnetic fields and frustrated exchange interaction , which play the competing roles in enhancing quantum entanglement , have a notable influence on nn and nnn concurrence in one - dimensional @xmath21 xy spin system .
the nn and nnn concurrence exhibits some interesting phenomena . for dimerized ferromagnetic spin chain
, nnne appears only above the critical dimerized interaction , which effects quantum phase transition point and increases nnne . for a ferromagnetic - antiferromagnetic frustrated spin chain ,
the nnn interaction is predominant in the competing role and can enhance the entanglement to a steady value .
the ferromagnetic nnn interaction increases and shrinks nne below and above critical frustrated interaction respectively , while the antiferromagnetic nnn interaction always decreases nne .
meanwhile , the antiferromagnetic nnn interaction can generate a larger value of nnne in comparison to the case when the nnn interaction is ferromagnetic .
so we can employ nnn interaction strength as well as dimerized interaction to realize quantum entanglement control .
as for the case of @xmath81(xy model ) , we will present further reports in the future .
+ 100 bennett c h , brassard g , crpeau c , jozsa r , peres a and wootters w k 1993 _ phys .
lett . _ * 70 * 1895 + shan c j , man z x , xia y j , liu t k 2007 _ int . j. quantum information _ * 5 * 359 deutsch d , ekert a , jozsa r , macchiavello c , popescu s and sanpera a 1996 _ phys .
lett . _ * 77 * 2818 + shan c j , man z x , xia y j , liu t k 2007 _ int . j. quantum information _ * 5 * 335 man z x and xia y j 2007 _ chin . phys .
* 1197 wei h , deng z j , zhang x l and feng m 2007 _ phys . rev .
a _ * 76 * 054304 + wei h , fang r r , liu j b , zhou f , yang w l and deng z j 2008 _ j. phys .
b _ * 41 * 085506 kim m s , lee j , ahn d and knight p l 2002 _ phys .
rev . a _ * 65
* 040101 + zhang l h , dong p and cao z l 2007 _ chin .
_ 16 640 zheng s b and guo g c 2000 _ phys .
* 85 * , 2392 hu y h , fang m f and liao x p 2007 _ chin .
_ 16 1344 shan c j , xia y j 2006 _ acta .
_ * 55 * 1585 + liu t k , cheng w w , shan c j , gao y f and wang j s 2007 _ chin
. phys . _ * 16 * 3697 wang c z and fang m f 2003 _ chin .
phys . _ * 12 * 287 turchette q a , wood c s , king b e , myatt c j , leibfried d , itano w m , monroe c and wineland d j 1998 _ phys
* 81 * 3631 zhang y , liu d and long g l 2007 _ chin . phys . _
16 324 wang x g 2001 _ phys . rev .
a _ * 64 * 012313 + wang x g 2002 _ phys .
a _ * 66 * 044305 + wang x g 2002 _ phys . rev
a _ * 66 * 034302 zhu y , zhu s q and hao x 2007 _ chin . phys .
* 2229 zhou l , song h s , guo y q and li c 2003 _ phys .
a _ * * 68**024301 +
zhang g f and li s s 2005 _ phys .
rev . a _ * 72 * 034302 cheng w w , huang y x , liu t k and li h 2007 _ physica e _ * 39 * 150 zvyagin a a and skorobagatko g a 2006 _ phys . rev .
b _ * 73 * 024427 beccaria m , campostrini m and feo a 2006 _ phys . rev .
b _ * 73 * 052402 mkhitaryan v v and sedrakyan a g 2008 _ phys .
b _ * 77 * 035111 sun z , wang x g and li y q 2005 _ new j. phys . _
* 7 * 83 doronin s i , pyrkov a n and feldman e b 2007 _ jetp letters _ * 85 * 519 gu s j , li h b , li y q and lin h q 2004 _ phys
. rev . a _ * 70 * 052302 chen y , zanardi p , wang z d and zhang f c 2006 _ new j. phys . _
* 8 * 97 chen s , wang l , gu s j and wang y p 2007 _ phys . rev .
e _ * 76 * 061108 osterloh a , amico l , falci g and fazio r 2002 _ nature_(london ) * 416 * 608 huang z , osenda o , kais s 2004 _ phys . lett . a _ * 322 * 137 osenda o , huang z , kais s 2003 _ phys . rev .
a _ * 67 * 062321 shan c j , cheng w w , liu t k , huang y x and li h 2008 _ chin .
* 17 * 0794 + shan c j , cheng w w , liu t k , huang y x and li h 2008 _ chin .
lett . _ * 25 * 817 vidal g , latorre j i , rico e and kitaev a 2003 _ phys .
* 90 * 227902 roscilde t , verrucchi p , fubini a , haas s and tognetti v 2005 _ phys .
lett . _ * 94 * 147208 jordan p and wigner e 1928 _ z. phys . _ * 47 * 631 white s r 1993 _ phys .
b _ * 48 * 10345 + schollwck u 2005 _ rev .
phys . _ * 77 * 259 ren j and zhu s q 2008 _ phys
. rev . a _ * 77 * 034303 wooters w k 1998 _ phys
lett . _ * 80 * 2245 model which can be considered either as a linear chain(a ) or a zigzag chain(b ) .
the alternating nearest - neighbouring interaction is @xmath62 and @xmath63 respectively , while the next - nearest - neighbouring interaction is parameterized by @xmath64,title="fig:",scaledwidth=100.0% ] + and @xmath82 as a function of the reduced coupling constant @xmath65 for various alternating values of the nearest - neighbour interaction and external magnetic field , nnn interaction @xmath71.,title="fig:",scaledwidth=100.0% ] + as a function of the reduced coupling constant @xmath65 for various alternating values of the nearest - neighbour interaction and external magnetic field , nnn interaction @xmath71 .
, title="fig:",scaledwidth=100.0% ] + and @xmath83 as a function of the reduced coupling constant @xmath65 for various frustrated values of next - nearest - neighbouring interaction , nn interaction @xmath84.,title="fig:",scaledwidth=100.0% ] + | by using the method of density - matrix renormalization - group to solve the different spin - spin correlation functions , the nearest - neighbouring entanglement(nne ) and next - nearest - neighbouring entanglement(nnne ) of one - dimensional alternating heisenberg xy spin chain is investigated in the presence of alternating nearest neighbour interactions of exchange couplings , external magnetic fields and next - nearest neighbouring interactions . for dimerized ferromagnetic spin chain
, nnne appears only above the critical dimerized interaction , meanwhile , the dimerized interaction effects quantum phase transition point and improves nnne to a large value .
we also study the effect of ferromagnetic or antiferromagnetic next - nearest neighboring ( nnn ) interactions on the dynamics of nne and nnne . the ferromagnetic nnn interaction increases and shrinks nne below and above critical frustrated interaction respectively , while the antiferromagnetic nnn interaction always decreases nne .
the antiferromagnetic nnn interaction results to a larger value of nnne in comparison to the case when the nnn interaction is ferromagnetic . |
the multi - component microscopic nature of superconductivity has been established in a number of materials ( mgb@xmath1 @xcite , cuprates @xcite , strontium rutenates @xcite , iron - arsenic compounds @xcite , nb - doped srtio@xmath2 @xcite , nbse@xmath1 @xcite , v@xmath2si @xcite etc . ) .
the first relevant theoretical scenarios have been discussed in the literature long time ago @xcite and later on they have been developed for various systems ( see e.g. refs . @xcite and references therein ) .
the peculiar properties of multi - component superconductors appear due to the involvement of coupled electron subsystems ( orbitals , different fermi surface sheets or other elements of electron structure ) into the formation of superconducting ordering varying substantially the physical nature of the phenomenon and opening the possibilities for novel effects . as a result of the presence of inter - component pairing the certain quantities (
e.g. coherence lengths and the relaxation times of superconductivity fluctuations ) can not be attributed to the initially independent subsystems and they describe the joint features of the whole two - component condensate @xcite . in a two - orbital superconductor with inter - orbital pairing , similarly to the results of refs .
@xcite , one of two solutions for coherence lengths diverges near the phase transition point as a function of temperature , while the other one is non - critical @xcite .
we note also that the existence of critical and non - critical length scales in the coherency associates with the conception of critical and non - critical fluctuations @xcite in a two - component superconductor damping in different time - scales . in the present contribution
we continue to examine the temperature behaviour of characteristic lengths of a two - orbital superconductor described by negative-@xmath0 hubbard model @xcite with inter - orbital pair - transfer interaction in dependence on the number of electrons .
our preliminary study has been done focusing on band filling effects in the spatial coherency including van hove singularity @xcite . however , in that work @xcite the orbital energy differences and charge fluctuations were neglected . in the present paper
we clarify their role in the scaling of critical and non - critical coherence lengths and magnetic field penetration depth by band fillings .
the numerical calculations have been carried out for a two - dimensional square lattice .
two - orbital superconductivity is modeled by the hamiltonian of the following form ( see e.g. @xcite ) : @xmath3a_{i\alpha\sigma}^{+}a_{j\alpha\sigma } \nonumber \\ & + & \frac{1}{2}\sum_{\alpha}\sum_{i}\sum_{\sigma}u^{\alpha\alpha } n_{i\alpha\sigma}n_{i\alpha-\sigma}\nonumber \\ & + & \frac{1}{2}\sum_{\alpha,\alpha'}^{,}\sum_{i}\sum_{\sigma}u^{\alpha\alpha'}a_{i\alpha\sigma}^{+}a_{i\alpha'\sigma } a_{i\alpha-\sigma}^{+}a_{i\alpha'-\sigma } \
, , \nonumber \\\end{aligned}\ ] ] where @xmath4 ( @xmath5 ) is the electron creation ( destruction ) operator in the orbital @xmath6 localized at the site @xmath7 ; @xmath8 is the spin index ; @xmath9 is the hopping integral ; @xmath10 is the orbital energy ; @xmath11 is the chemical potential ; @xmath12 is the intra - orbital attraction energy ; @xmath13 is the particle number operator ; @xmath14 with @xmath15 is the inter - orbital interaction energy .
note that the both intra- and interorbital interaction channels involved lead to superconducting ordering . in the reciprocal space
the hamiltonian ( [ 1 ] ) can be represented as @xmath16a_{\alpha\mathbf{k}\sigma}^{+}a_{\alpha\mathbf{k}\sigma } \nonumber \\ & + & \frac{1}{2n}\sum_{\alpha,\alpha'}\sum_{\mathbf{k},\mathbf{k}'}\sum_{\mathbf{q}}\sum_{\sigma}u^{\alpha\alpha ' } a_{\alpha\mathbf{k}\sigma}^{+ } a_{\alpha(-\mathbf{k}+\mathbf{q})-\sigma}^{+ } \nonumber \\ & \times & a_{\alpha'(-\mathbf{k}'+\mathbf{q})-\sigma } a_{\alpha'\mathbf{k}'\sigma } \ , .\end{aligned}\ ] ] where @xmath17 is the electron band energy associated with the orbital @xmath18 and @xmath19 is the number of lattice sites ( number of atoms ) .
the bulk superconducting state is described on the basis of the hamiltonian ( [ 2 ] ) by means of the hartree - fock - gorkov self - consistent equations @xmath20 @xmath21 \ , , \nonumber \\\end{aligned}\ ] ] @xmath22 here @xmath23 , @xmath24 , @xmath25 are the occupation numbers of orbitals ( bands ) per lattice site , and @xmath26 is the total number of electrons per site .
( [ 3 ] ) determine the superconducting gaps of orbitals ( bands ) , the eq .
( [ 4 ] ) gives us the orbital occupation numbers and the eq .
( [ 4a ] ) , the chemical potential @xmath11 in dependence on @xmath26 and other relevant quantities .
the temperature of superconducting phase transition @xmath27 satisfies the equation @xmath28 where @xmath29 in the general case of spatial inhomogeneity and non - zero magnetic field we have the following ginzburg - landau equations @xmath30\delta_{\alpha ' } ( \mathbf{r } ) \ , .\end{aligned}\ ] ] @xmath31 here @xmath32_{\delta_{\alpha}=0 } \ , , \end{aligned}\ ] ] @xmath33 with @xmath34\right\}_{\mathbf{q}=0 } \ , , \end{aligned}\ ] ] @xmath35 is the vector potential , @xmath36 is the density of supercurrent , @xmath37 is the magnetic flux quantum and @xmath38 is the volume of unit cell . according to the eq .
( [ 10 ] ) we restrict ourselves to the isotropic electron energy spectrum .
by considering the solutions of eqs .
( [ 7 ] ) which correspond to the small deviations of gaps from their bulk values , @xmath39 , one can find two length scales characterizing the spatial behaviour of superconductivity in a two - orbital system : @xmath40 where @xmath41 \nonumber\\ & -&\left[1+u^{11}\tilde{g}_{1}(t ) \right]u^{22}\beta_{2 } \nonumber\\ & -&\left[1+u^{22}\tilde{g}_{2}(t)\right]u^{11}\beta_{1 } \ , , \end{aligned}\ ] ] @xmath42\left[1+u^{22}\tilde{g}_{2}(t)\right ] \nonumber\\ & -&\left(u^{12}\right)^{2}\tilde{g}_{1}(t)\tilde{g}_{2}(t ) \,\end{aligned}\ ] ] with @xmath43 , and @xmath44\beta_{1}\beta_{2 } \ , .\end{aligned}\ ] ] the soft or critical coherence length @xmath45 diverges at the phase transition point @xmath46 , while the rigid or non - critical coherence length @xmath47 remains finite . on the basis of eq .
( [ 8 ] ) one obtains in a standard way the magnetic field penetration depth for a two - orbital superconductor , @xmath48 where @xmath49 is the magnetic permeability of free space .
below we present the results of numerical calculations for two - dimensional square lattice with hopping integrals between nearest neighbours @xmath50 and electron band energies associated with @xmath51-orbitals @xmath52 $ ] , @xmath53 is the lattice constant , and @xmath54 .
the intra- and interorbital interactions are chosen as @xmath55 , @xmath56 , @xmath57 , and the site energy @xmath58 is fixed at @xmath59 . in all calculations
we have chosen @xmath60 .
using eqs .
( [ 5 ] ) and ( [ 6 ] ) we calculated the superconducting critical temperature @xmath61 .
1 shows the dependence of @xmath61 on the band filling @xmath62 .
one can see that the curve @xmath27 vs @xmath26 is most symmetric as @xmath63 ( see fig1 , curve ( a ) ) . in this case , the small shift of the maximum of @xmath27 towards the smaller values of occupation number @xmath26 is caused by effective hartree corrections @xmath64 .
first of all , these corrections are different for various orbitals as the intra - orbital interactions @xmath65 are different . consequently , the distances between the chemical potential @xmath11 position and the lower edges of orbital bands are slightly different
. therefore the position of the maximum of @xmath61 is moved reflecting the above mentioned difference which appears here due to the choice of intra - orbital pairing potentials @xmath66 . by increasing the difference between the bare site energies
@xmath67 we observe that the asymmetry of the function @xmath68 increases . in the cases ( a ) and ( b ) in fig .
1 , the site energies of orbitals are closer and the influence of the hartree terms does not lead to dramatically different distances between the chemical potential and lower band edges . as a result
the single peak in @xmath61 vs @xmath26 is preserved . however , with the further increase of the difference between of bare site energies @xmath69 the effect of the hartree renormalization @xmath70 becomes stronger and we observe the splitting of the single peak of @xmath61 into two maxima ( see fig1 , curves ( c ) and ( d ) ) caused by the circumstance that the chemical potential passes in the region @xmath71 two van hove singularities related to the bands @xmath6 .
it should be also noted that the larger values of @xmath67 mean the larger difference between the band occupation numbers @xmath72 and @xmath73 which corresponds to the stronger hartree renormalization . and
@xmath74 , panels ( a)(d ) , respectively . , title="fig:",width=147 ] and @xmath74 , panels ( a)(d ) , respectively . ,
title="fig:",width=147 ] and @xmath74 , panels ( a)(d ) , respectively . ,
title="fig:",width=147 ] and @xmath74 , panels ( a)(d ) , respectively .
, title="fig:",width=147 ] next we solve the gap equations ( eq .
( [ 3 ] ) ) below @xmath61 . in fig . 2
one can see the plotted superconducting gaps @xmath75 vs temperature for different band fillings .
note that the increase of @xmath26 changes the order of @xmath76 and @xmath77 in the energy scale for fixed temperature , c.f . figs . 2a , 2b and 2c , 2d .
with that for the intermediate values on @xmath26 the gaps @xmath76 and @xmath77 are relatively close ( figs .
2b and 2c ) while for @xmath78 and @xmath79 one of the gaps strongly dominates ( figs .
2a and 2d ) .
the latter peculiarity is caused by the chemical potential vicinity to the van hove singularity which supports superconductivity in the corresponding band .
the weaker superconductivity in another band is related , at least partially , to the interband proximity effect . at the same time the band fillings @xmath78 and @xmath79 correspond approximately to the maxima of @xmath80 in fig .
1 . thus , the formation of two peaks of @xmath27 as a function of @xmath26 in fig . 1 reflects the redistribution of the driving roles of bands in the appearance of superconductivity . in a single - orbital ( single - band ) system
the ginzburg - landau parameter @xmath81 defines the type - i ( @xmath82 ) and type - ii ( @xmath83 ) superconductivity .
the coherence lengths ( see eq .
( [ 12 ] ) ) and magnetic field penetration depth ( see eq .
( [ 16 ] ) ) for two - orbital model are plotted in fig . 3 for the band fillings discussed above .
apart from the divergence of the soft length scale @xmath84 at @xmath27 , we observe also a maximum for the rigid length scale @xmath85 slightly below @xmath27 in figs . 3b and 3c .
this is due to the closeness of the superconductivity gaps for these band fillings , see figs .
2b and 2c .
another interesting observation is related to the crossing point of @xmath86 and @xmath87 . in figs .
3b - d this crossing appears at the temperatures very close to @xmath61 while in fig .
3a the crossing point is located remarkably lower in the temperature scale . thus , there exists always a small temperature region in figs 3a - d where @xmath88 . to make the hartree - fock - gorkov approximation plausible .
in this situation we found that the penetration depth in our calculations is always smaller than the critical coherence length .
the negative - u hubbard model for larger @xmath89 has to be treated in a different way @xcite .
] one can treat the fulfilment of the inequality @xmath90 as necessary but not sufficient condition for type-1.5 superconductivity @xcite , see ref .
@xcite . to shed more light on the problem one has to study the character of the interaction between vortices @xcite .
the larger extent of the temperature region in fig .
3a where @xmath88 may be related to the interband proximity effect which is more pronounced in this case ( see fig .
it was shown in @xcite that type-1.5 superconductivity can arise due to the proximity effect in a two - band system .
we presented the results of band filling effect on two - orbital superconductor .
the dependencies of critical and non - critical coherence lengths on temperature were compared to the penetration depth for different electron numbers .
it was shown that the band structure including site energy and position of van hove singularity is important for superconducting gaps temperature behaviour as well as in the relationship between the magnitudes of critical and non - critical coherence lengths and penetration depth .
we have identified a temperature region where @xmath88 .
its larger extent could be associated with the presence of interband proximity effect in a two - orbital superconducting system .
this research was supported by the european union through the european regional development fund ( centre of excellence `` mesosystems : theory and applications '' , tk114 ) .
we acknowledge the support by the estonian science foundation , grant no 8991 .
gl kindly acknowledges a financial support by the 7th framework programme fp7-regpot-2009 - 1 , under grant agreement no .
. 00 s. tsuda , t. yokoya , y. takano , h. kito , a. matsushita , f. yin , j. itoh , h. harima , s. shin , phys .
rev . lett . * 91 * , 127001 ( 2003 ) .
r. khasanov , a. shengelaya , a. maisuradze , f. la mattina , a. bussmann - holder , h. keller , k. a. mller , phys .
lett . * 98 * , 057007 ( 2007 ) .
y. maeno , t. m. rice , and m. sigrist , phys . today 54 , 42 ( 2001 ) .
h. ding , p. richard , k. nakayama , k. sugawara , t. arakane , y. sekiba , a. takayama , s. souma , t. sato , t. takahashi , z. wang , x. dai , z. fang , g. f. chen , j. l. luo , n. l. wang , epl * 83 * , 47001 ( 2008 ) . g. binning , a. bartoff , h. e. hoenig , j. g. bednorz , phys . rev . lett .
* 45 * , 1352 ( 1980 ) .
t. yokoya , t. kiss , a. chainani , s. shin , m. nohara , h. takagi , science * 294 * , 2518 ( 2001 ) .
a. nefyodov , a. m. shuvaev , m. r. trunin , epl * 72 * , 638 ( 2005 ) .
h. suhl , b. t. matthias and l. p. walker , phys .
lett . * 3 * , 552 ( 1959 ) .
v. a. moskalenko , fiz . met . metalloved . *
8 * , 503 ( 1959 ) .
j. kondo , progr .
* 29 * , 1 ( 1963 ) .
v. z. kresin , s. a. wolf , phys .
b * 46 * , 6458 ( 1992 ) .
n. kristoffel , p. konsin , t. rd , riv .
nuovo cimento * 17*(9 ) , 1 ( 1994 ) .
a. bianconi , j. superconductivity * 18 * , 625 ( 2005 ) .
r. caivano , m. fratini , n. poccia , a. ricci , a. puri , z .- a .
l . dong , j. yang , w. lu , z .- x .
zhao , l. barba , a. bianconi , supercond .
* 22 * , 014004 ( 2009 ) .
annett , g. litak , b.l .
gyorffy , k.i .
wysokiski , phys .
b * 66 * , 134514 ( 2002 ) .
j. carlstrm , e. babaev , m. speight , phys .
b * 83 * , 174509 ( 2011 ) .
t. rd , k. rgo , a. vargunin , j. supercond .
novel magn .
* 25 * , 1351 ( 2012 ) .
g. litak , t. rd , k. rgo , a. vargunin , acta phys . pol .
a * 121 * 247 ( 2012 ) .
t. rd , k. rgo , a. vargunin , j. supercond .
novel magn . ,
* 22 * , 85 ( 2009 ) .
m. silaev , e. babaev , phys .
b * 84 * , 094515 ( 2011 ) .
m. silaev , e. babaev , phys .
b * 85 * , 134514 ( 2012 ) t. rd , k. rgo , a. vargunin , in : physical properties of nanosystems , ed . j. bona , s. kruchinin , nato science for peace and security series b : physics and biophysics , p.177 , springer , dordrecht ( 2011 ) .
r. micnas , j. ranninger , and s. robaszkiewicz , rev .
phys . * 62 * , 113 ( 1990 ) .
a. ciechan , k.i .
wysokiski , phys .
b * 80 * , 224523 ( 2009 ) .
r. s. markiewicz , j. phys .
solids * 58 * 1179 ( 1997 ) .
v. moschalkov , m. menghini , t. nishio , q. h. chen , a. v. silhanek , v. h. dao , l. f. chibotaru , n. d. zhigadlo , j. karpinski , phys .
lett . * 102 * , 117001 ( 2009 ) .
e. babaev , j. carlstrm m. speight , phys .
* 105 * , 067003 ( 2010 ) . | the two - orbital superconducting state is modeled by on - site intra - orbital negative-@xmath0 hubbard correlations together with inter - orbital pair - transfer interactions .
the critical temperature is mainly governed by intra - orbital attractive interactions and it can pass through an additional maximum as a function of band filling . for the certain number of electrons the clear interband proximity effect is observable in the superconducting state of the band with a smaller gap .
the influence of band fillings and orbital site energies on the temperature dependencies of two - component superconductivity coherence lengths and magnetic field penetration depth is analyzed .
the presence of proximity effect is probably reflected in the relative temperature behaviour of characteristic lengths . two - orbital superconductor , coherence lengths , magnetic field penetration depth , hubbard model , interband proximity effect |
flavor - changing neutral current ( fcnc ) processes of quarks are highly sensitive to new physics at or above the electroweak scale .
the highly successful physics program of the b factories has revealed that these fcncs are dominantly governed by the cabibbo - kobayashi - maskawa ( ckm ) mechanism@xcite of the standard model .
this is an important constraint on the new , still undiscovered theory of tera scale physics .
models whose only source of flavor violation is the ckm matrix are termed _ _ minimally flavor violating ( mfv)__@xcite .
still , it is easy to construct models in which non mfv are naturally suppressed while still leading to measurable effects .
an example are grand unified theories ( guts ) in which mfv is implemented far above the gut scale , but subsequently altered by renormalisation group effects@xcite .
particularly well - suited for the search of corrections to the ckm mechanism are cp violating observables in @xmath5 transitions : ckm driven cp violation in @xmath5 transitions is small , just as in the @xmath6 transitions probed in kaon physics .
interestingly , in supersymmetric gut models the large atmospheric neutrino mixing angle can influence @xmath5 transitions@xcite . clearly , the `` holy grail '' of @xmath5 fcnc physics is the @xmath0 mixing amplitude @xmath7 .
new physics will affect its magnitude and phase , and already small contributions to @xmath7 can lift the small ( and in many cases unobservably tiny ) cp asymmetries to sizable values . in this talk
i summarize the avenues to ( over)constrain the @xmath0 mixing complex .
then a new , more precise , theory prediction for the element @xmath1 of the decay matrix is presented .
@xmath1 enters the formulae for the width difference @xmath2 of the two mass eigenstates in the @xmath8 system .
our new result further decreases the theoretical uncertainty in the extraction of the @xmath0 mixing phase from the cp asymmetry in flavor
specific @xmath8 decays .
finally , constraints from d data on @xmath7 are derived with the help of the new result for @xmath1 .
the presented results are from ref .
@xmath0 oscillations are governed by a schrdinger equation @xmath9 with the mass matrix @xmath10 and the decay matrix @xmath11 .
the physical eigenstates @xmath12 and @xmath13 with the masses @xmath14 and the decay rates @xmath15 are obtained by diagonalizing @xmath16 .
there are three physical quantities in @xmath0 mixing : @xmath17 the phase @xmath4 is responsible for _ cp violation in mixing_. the mass and width differences between @xmath18 and @xmath19 are @xmath20 here and in the following i neglect numerically irrelevant corrections of order @xmath21 .
the average width of the two eigenstates is denoted by @xmath22 .
the precise measurements from the d and cdf experiments@xcite @xmath23 determine @xmath24 sharply .
new physics affects magnitude and phase of @xmath7 , but can barely change @xmath1 , which dominantly stems from the cabibbo - favored tree level @xmath25 decays . in the standard model
@xmath4 is tiny , so that @xmath26 and new physics can only decrease @xmath2 in eq .
( [ dmdg])@xcite .
the success of the standard model suggests that new physics enters low energy observables at the loop level .
a different viewpoint has recently been taken in ref .
@xcite , where potentially huge contributions to @xmath1 from leptoquarks with fcnc couplings have been claimed .
the authors of ref .
@xcite exploit the plethora of free parameters in leptoquark models to place their effect into the yet unmeasured decay @xmath27 , whose branching ratio could then be enhanced to up to 18% .
the effect of some new decay mode on the ratio @xmath28 can be as large as twice its branching ratio , so that ref .
@xcite finds an enhancement of @xmath28 from its standard model value around 0.15 to up to 0.51 .
however , the enhanced @xmath29 coupling invoked in this model would also lead to sizable new @xmath30 decays modes of @xmath31 and @xmath32 mesons . while there are no precise data on final states with two @xmath33 s yet
, the extra decay modes would sizably alter well - measured and well - calculated inclusive quantities such as the semileptonic branching ratio @xmath34@xcite , the inclusive branching ratio @xmath35@xcite and even the lifetimes of all b mesons .
thus the idea of ref .
@xcite is not viable and it is safe to assume that @xmath1 is unaffected by new physics . while the pristine measurement in eq .
( [ dmexp ] ) already gives a powerful constraint on new physics * * ? ? ?
* * , there are two reasons to seek further experimental information from other observables : first , @xmath36 only constrains @xmath24 but not @xmath4 and second , the translation of eq .
( [ dmexp ] ) into constraints on fundamental parameters involves a hadronic parameter , which is difficult to compute and inflicts a theoretical uncertainty of order 30% onto the analysis .
new physics entering @xmath7 can be parameterized as @xmath37 thus every measurement related to @xmath0 mixing gives a constraint on the complex @xmath38 plane .
the standard model corresponds to @xmath39 . while @xmath40 is unphysical and depends on phase conventions , @xmath41 is a physical cp phase .
the mixing phase @xmath42 in eq .
( [ pq ] ) can be written as @xmath43 where @xmath44 is the standard model prediction for @xmath4 .
@xmath44 is negligible , see eq .
( [ finphi ] ) below .
the relationship of @xmath38 to the parameters used in @xcite is @xmath45 we find it more transparent to plot @xmath46 vs. @xmath47 than to plot @xmath48 vs. @xmath49 .
next we list the key measurements which ( over)constrain @xmath38 : * @xmath36 in eq .
( [ dmexp ] ) determines @xmath50 . *
measuring the lifetime in an untagged @xmath51 decay @xmath52 , where @xmath53 is a cp eigenstate , determines @xmath54@xcite .
the time - dependent decay rate reads @xmath55 & \propto & \frac{1 \pm \cos(\phi_s^\delta -2 \beta_s ) } { 2 } e^{-\gamma_l t } \ ; + \ : \frac{1\mp \cos(\phi_s^\delta -2 \beta_s ) } { 2 } e^{-\gamma_h t } { \nonumber \\ } & = & e^{-\gamma_s t } { \left } [ \cosh \frac{{\ensuremath{\delta \gamma}}_s \ , t}{2 } \mp \cos(\phi_s^\delta -2 \beta_s ) \sinh
\frac{{\ensuremath{\delta \gamma}}_s \ , t}{2 } { \right } ] .
\label{twoexp}\end{aligned}\ ] ] here the sign convention of @xmath56 is that of ref .
@xcite . currently this measurement is applied to @xmath57 . here
the cp quantum number of the final state depends on the orbital angular momentum , the p wave state is @xmath58 and the s wave and d wave components correspond to @xmath59 in eq .
( [ twoexp ] ) . neglecting @xmath60 , @xmath44 and expanding to first order in @xmath2 one
verifies from eq .
( [ twoexp ] ) that the lifetime measurement determines@xcite @xmath61 * the angular analysis of an untagged @xmath62 sample not only determines @xmath63 as discussed in item 2 , but also contains information on @xmath64 through a cp - odd interference term . * the cp asymmetry in _ flavor - specific _
@xmath65 decays is @xmath66 @xmath67 is typically measured by counting the number of positively and negatively charged leptons in semileptonic @xmath68 decays .
observing further the time evolution of these untagged @xmath69 decays , @xmath70 \ , -\
, \gamma [ { \raisebox{7.7pt}{$\scriptscriptstyle(\hspace*{8.5pt})$ } \hspace*{-10.7pt}\!{\,\overline{\rm b}}_{s}}\to x^+ \ell^- { \overline{\nu}}_\ell , t ] } { \gamma [ { \raisebox{7.7pt}{$\scriptscriptstyle(\hspace*{8.5pt})$ } \hspace*{-10.7pt}\!{\,\overline{\rm b}}_{s}}\to x^- \ell^+\nu_\ell , t ] \ , + \ , \gamma [ { \raisebox{7.7pt}{$\scriptscriptstyle(\hspace*{8.5pt})$ } \hspace*{-10.7pt}\!{\,\overline{\rm b}}_{s}}\to x^+ \ell^- { \overline{\nu}}_\ell , t ] } & = & \frac{a_{\rm fs}^s}{2 } { \left } [ 1- \frac{\cos({\ensuremath{\delta m}}_s \ , t)}{\cosh{({\ensuremath{\delta \gamma}}_s\ , t/2 ) } } { \right } ] , \label{afst}\end{aligned}\ ] ] may help to control systematic experimental effects@xcite
. often the average lifetime of the two @xmath8 eigenstates is included into global experimental analyses of @xmath2 .
heavy quark physics implies that the average widths @xmath71 and
@xmath72 in the @xmath32 and @xmath8 systems are equal up to corrections of order 1% . the average @xmath8 lifetime will then exceed the @xmath32 lifetime by a term which is quadratic in @xmath73@xcite .
for realistic values of @xmath2 this term is too small to determine @xmath2 .
we close this section with an important remark : up to now there is _ no _ experimental information on the sign of @xmath2 available ! from eq .
( [ dmdg ] ) one verifies immediately that @xmath74 and ( see eq .
( [ gc2 ] ) ) the untagged analysis described in item 2 is only sensitive to @xmath75 .
also the cp
odd quantities described in items 3 and 4 only determine @xmath76 , which comes with a two fold ambiguity for @xmath77 : the two solutions correspond to different signs of @xmath78 and thereby different signs of @xmath2 .
the determination of @xmath79 is discussed in ref .
it is easy to check that the formula for the angular distribution in @xmath57@xcite is unchanged if one simultaneously flips the sign of @xmath2 ( i.e. interchanges @xmath80 and @xmath81 ) and the sign of @xmath82 .
thus current experimental results should be quoted for @xmath75 rather than @xmath2 .
the predictions of @xmath7 and @xmath83 involve hadronic matrix elements of four quark operators .
these matrix elements are computed with the help of lattice gauge theory and dominate the theoretical uncertainty . in the standard model prediction for @xmath7 one only encounters the operator @xmath84 where @xmath85 and @xmath86 are color indices .
the matrix element is usually parameterized as @xmath87 where @xmath88 and @xmath89 are mass and decay constant of the @xmath8 , respectively .
@xmath90 is called a bag factor .
then@xcite @xmath91 the number in eq .
( [ dmsnum ] ) is found from eq .
( [ dms ] ) with @xmath92 and @xmath93@xcite .
the situation with @xmath1 is more complicated : its prediction requires the expansion in two parameters , @xmath94 and @xmath95 . here
@xmath96 is the relevant hadronic scale and @xmath97 is the strong coupling constant@xcite . in the first step one
finds that three operators contribute to @xmath1 at leading order in @xmath94 : @xmath98 defined in eq .
( [ defq ] ) , @xmath99 we parameterize the new matrix elements with bag factors @xmath100 and @xmath101 .
subsequently one trades one of the operators for @xmath102 here @xmath103 are qcd correction factors@xcite .
this is done , because the matrix element of @xmath104 is suppressed by @xmath94 , so that it belongs to the subleading order@xcite . in refs .
@xcite eq .
( [ defr0 ] ) has been used to eliminate @xmath105 from the operator basis .
this results in the prediction @xmath106 \ , \mbox{ps}^{-1 } , \label{oldr } \end{aligned}\ ] ] where @xmath107 and @xmath108 are the bag factors of @xmath104 and another subleading operator , @xmath109 , and the uncertainties of the coefficients are not shown .
the other sub - leading operators come with smaller coefficients and are accounted for with a common bag factor @xmath110 in eq .
( [ oldr ] ) .
this result is pathological in several respects : the @xmath94 corrections exceed their natural size of 20% and are negative , the next to leading order qcd corrections of refs .
@xcite ( which are contained in the numbers 0.002 and 0.094 ) are also large and decrease the result further and finally @xmath111 is dominated by @xmath100 , so that the cancellation of hadronic physics from the ratio @xmath112 is imperfect .
the starting point of the improvement in ref .
@xcite is the observation that the matrix element of @xmath113 is small@xcite . keeping in mind that the matrix element of @xmath104 is power suppressed and therefore also small , eq .
( [ defr0 ] ) encodes a strong numerical correlation between @xmath90 and @xmath100 .
( [ defr0 ] ) implies for the bag parameters : @xmath114 hence trading @xmath101 for a linear combination of @xmath90 , @xmath100 and @xmath107 expresses a small number in terms of the difference of two big numbers : @xmath115 .
so one tends to introduce a theoretical uncertainty into the problem , which is not inherent to the calculated quantity .
the most straightforward way to take care of this is to keep @xmath98 and @xmath113 in the basis and to abandon @xmath116 instead .
this results in@xcite @xmath117 \!\big ] \ , \mbox{ps}^{-1}. \quad \label{finaldg } \end{aligned}\ ] ] the quoted result further includes the resummation of logarithms of the charm mass to all orders in perturbation theory .
now all pathologies have disappeared .
the smallness of the coefficient of @xmath107 compared to eq .
( [ oldr ] ) can be understood with the help of the @xmath118 expansion .
( @xmath119 is the number of colors . )
the coefficients of @xmath98 and @xmath113 are leading in @xmath118 , while the coefficient of @xmath116 is color suppressed .
if one eliminates @xmath116 in terms of @xmath104 , the coefficient of @xmath116 becomes the coefficient of @xmath104 , so that the number multiplying @xmath120 is small . in the old result in eq .
( [ oldr ] ) , however , the coefficient of @xmath120 stems from the color - favored coefficient of @xmath113 and is large . on the other hand ,
the ( equally welcome ) reduction of the nlo qcd correction , which is related to the qcd factors @xmath121 , appears accidental . from our new result for @xmath1
we also get a new prediction for @xmath67 and the cp phase @xmath4 in the standard model .
our predictions are @xmath122 for the precise values of the input parameters i refer to ref .
@xcite .
the prediction of the ratio @xmath123 in eq .
( [ dgdmr ] ) got much sharper , because most of the hadronic uncertainies now cancel , since @xmath2 in eq .
( [ finaldg ] ) is dominated by the term involving @xmath90 . with eq .
( [ dmexp ] ) one finds from eq .
( [ dgdmr ] ) : @xmath124 any future measurement of @xmath2 outside this range will signal new physics in @xmath36 or @xmath2 . for predictions of the corresponding quantities in the @xmath125 mixing complex
i refer to refs .
the formulae relating @xmath36 , @xmath2 and @xmath67 to @xmath38 defined in eq .
( [ defdel ] ) are @xmath126 for our analysis we use the the cdf data on @xmath36 in eq .
( [ dmexp ] ) and d data on the angular distribution in @xmath127@xcite , the semileptonic cp asymmetry @xmath128@xcite and on the same sign di muon asymmetry @xmath129@xcite , which is related to @xmath130 and @xmath131 as ( updated from ref .
@xcite ) @xmath132 the angular analysis of @xmath57@xcite involves two strong phases @xmath133 and @xmath134 .
they can be determined from the data , albeit with discrete ambiguities which imply a four fold ambiguity in @xmath4 .
the cp conserving piece of the angular distribution depends on @xmath135 , so that the experimental error on @xmath135 is smaller than the error on @xmath136 and @xmath137 , which appear in the cp violating piece proportional to @xmath138 .
the d result for @xmath139@xcite is in good agreement with theoretical model calculations predicting @xmath140 and @xmath141@xcite . in our analysis
we have fixed @xmath142 and @xmath143 , which reduces the four fold ambiguity in @xmath77 to a two fold one .
d finds@xcite @xmath144 \mbox{or}\qquad \delta \gamma_s & = & - 0.17 \pm 0.09{}_{\mbox{\scriptsize ( stat ) } } \pm 0.03{}_{\mbox{\scriptsize ( syst ) } } \ , \mbox{ps}^{-1 } \ , { \nonumber \\ } & & \mbox{and } \quad \phi_s^\delta-2\beta_s \ ; = \ ; -0.79 \pm 0.56{}_{\mbox{\scriptsize ( stat ) } } \pm 0.01 { } _ { \mbox{\scriptsize ( syst ) } } \ ; + \ ; \pi \ , . \label{eqdgexp4}\end{aligned}\ ] ] the second solution deviates from the standard model by several standard deviations . in order to assess the compatibility of the measurement with the standard model we only need to consider the first solution in eq
( [ eqdgexp5 ] ) . the same sign di muon asymmetry @xmath145 in eq .
( [ asllk ] ) involves both @xmath130 and @xmath131 . in order to translate the d measurement@xcite of @xmath146 into a number for @xmath130 we have used our theory prediction for @xmath131@xcite .
combining the result with the measurement of @xmath130@xcite gives @xmath147 fig .
[ boundbandreal ] shows the combination of all measurements .
it indicates a deviation from the standard model value @xmath39 by @xmath148 . for this conclusion
it is crucial that we use the theoretical value for @xmath149@xcite , which is much more precise than the current measurement of this quantity ( see ref . [ 14 ] of ref .
@xcite ) : @xmath150{$\scriptstyle+1.0$\\$\scriptstyle-1.2 $ } } } { \right } ) \ , \cdot 10^{-4 } , \qquad \quad a_{\rm fs}^{d,\rm exp } \ ; = \ ; { \left } ( -47 \pm 46{\right } ) \cdot 10^{-4 } .
\label{afsnum } \end{aligned}\ ] ] this assumes , of course , that no new physics enters @xmath149 .
another combined analysis of the d data of refs .
@xcite has been performed in ref .
this analysis has used the experimental value @xmath151 in eq .
( [ asllk ] ) , which leads to a significantly weaker constraint from the same sign di muon asymmetry . also no theory input on @xmath152 has been used in the analysis of ref .
the final result for @xmath4 is therefore more conservative than ours and implies a deviation from the standard model by only @xmath153 .
i presented an improved theoretical prediction of @xmath1 , which permits more accurate predictions of the width difference @xmath2 and of the cp asymmetry in flavor - specific decays , @xmath67 , in scenarios of physics beyond the standard model . applying the new formulae to d data we find that the @xmath0 mixing phase deviates from the standard model value by @xmath148 .
we conclude that current experiments are reaching the sensitivity to probe new physics in the @xmath0 mixing phase .
i acknowledge the warm hospitality of the organizers of the ctp07 workshop . this work was supported in part by the dfg grant no .
ni 1105/11 , by the eu marie - curie grant mirg
ct2005029152 , by the bmbf grant 05 ht6vkb and by the eu contract no . mrtn - ct-2006 - 035482 , flavianet .
n. cabibbo , phys .
* 10 * , 531 ( 1963 ) ; m. kobayashi and t. maskawa , prog . theor . phys .
* 49 * , 652 ( 1973 ) .
g. dambrosio , g. f. giudice , g. isidori and a. strumia , nucl .
b * 645 * ( 2002 ) 155 [ arxiv : hep - ph/0207036 ] .
effects of the model in ref .
@xcite on @xmath0 mixing have been studied in : s. jger and u. nierste , eur .
j. c * 33 * , s256 ( 2004 ) [ arxiv : hep - ph/0312145 ] ; s. jger and u. nierste , in _ proceedings of the 12th international conference on supersymmetry and unification of fundamental interactions ( susy 04 ) _ , 17 - 23 jun 2004 , tsukuba , japan , p. 675 - 678 , ed .
k. hagiwara , j. kanzaki , n. okada [ hep - ph/0410360 ] ; s. jger , hep - ph/0505243 , _ proceedings of the xlth rencontres de moriond , electroweak interactions and unified theories _
, 5 - 12 mar 2005 , la thuile , aosta valley , italy , ed .
j. tran thanh van .
a. lenz and u. nierste , jhep * 0706 * ( 2007 )
072 [ arxiv : hep - ph/0612167 ] .
talks by d. glenzinski ( plenary ) , t. moulik and s. giagu at _
33rd international conference on high energy physics ( ichep 06 ) _ ,
26 jul 2 aug 2006 , moscow , russia . v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 97 * ( 2006 ) 021802 [ arxiv : hep - ex/0603029 ] .
a. abulencia _ et al .
_ [ cdf collaboration ] , phys . rev .
* 97 * ( 2006 ) 242003 [ arxiv : hep - ex/0609040 ] . y. grossman , phys . lett .
b * 380 * ( 1996 ) 99 [ arxiv : hep - ph/9603244 ] .
i. dunietz , r. fleischer and u. nierste , phys .
d * 63 * ( 2001 ) 114015 [ arxiv : hep - ph/0012219 ] .
a. dighe , a. kundu and s. nandi , arxiv:0705.4547 [ hep - ph ] .
e. bagan , p. ball , v. m. braun and p. gosdzinsky , phys .
b * 342 * ( 1995 ) 362 [ erratum - ibid .
b * 374 * ( 1996 ) 363 ] [ arxiv : hep - ph/9409440 ] .
a. lenz , u. nierste and g. ostermaier , phys .
d * 56 * ( 1997 ) 7228 [ arxiv : hep - ph/9706501 ] ; phys .
d * 59 * ( 1999 ) 034008 [ arxiv : hep - ph/9802202 ] . c. greub and p. liniger , phys .
d * 63 * ( 2001 ) 054025 [ arxiv : hep - ph/0009144 ] .
y. grossman , y. nir and g. raz , phys .
* 97 * ( 2006 ) 151801 [ arxiv : hep - ph/0605028 ] .
updated result of j. charles _ et al . _ [ ckmfitter group ] , eur .
j. c * 41 * ( 2005 ) 1 [ arxiv : hep - ph/0406184 ] for the summer conferences 2006 , see http://www.slac.stanford.edu/xorg/ckmfitter/ckm@[email protected] ; updated result of m. bona _ et al . _
[ utfit collaboration ] , arxiv : hep - ph/0606167 for the summer conferences 2006 , see http://utfit.roma1.infn.it/ ; updated result of e. barberio _ et al . _
[ heavy flavor averaging group ( hfag ) ] , arxiv : hep - ex/0603003 for the summer conferences 2006 , see http://www.slac.stanford.edu / xorg / hfag/. k. anikeev _ et al .
_ , _ @xmath90 physics at the tevatron : run ii and beyond _ , [ hep - ph/0201071 ] , chapters 1.3 and 8.3 .
u. nierste , [ hep - ph/0406300 ] , in : _ proceedings of the xxxixth rencontres de moriond , electroweak interactions and unified theories _ , 21 - 28 mar 2004 , la thuile , aosta valley , italy , ed
. j. tran thanh van .
a. s. dighe , i. dunietz , h. j. lipkin and j. l. rosner , phys .
b * 369 * ( 1996 ) 144 [ arxiv : hep - ph/9511363 ] .
a. s. dighe , i. dunietz and r. fleischer , eur .
j. c * 6 * ( 1999 ) 647 [ arxiv : hep - ph/9804253 ] .
buras , m. jamin and p.h .
weisz , nucl .
* b347 * , 491 ( 1990 ) .
s. hashimoto and t. onogi , arxiv : hep - ph/0407221 ; s. hashimoto , int .
j. mod .
phys . a * 20 * ( 2005 ) 5133 [ arxiv : hep - ph/0411126 ] ; m. okamoto , pos * lat2005 * ( 2006 ) 013 [ arxiv : hep - lat/0510113 ] .
a. ali khan _ et al .
_ [ cp - pacs collaboration ] , phys .
d * 64 * ( 2001 ) 054504 [ arxiv : hep - lat/0103020 ] ; a. ali khan _ et al . _
[ cp - pacs collaboration ] , phys .
d * 64 * ( 2001 ) 034505 [ arxiv : hep - lat/0010009 ] ; c. bernard _ et al .
_ [ milc collaboration ] , phys . rev .
d * 66 * ( 2002 ) 094501 [ arxiv : hep - lat/0206016 ] ; s. collins , c. t. h. davies , u. m. heller , a. ali khan , j. shigemitsu , j. h. sloan and c. morningstar , phys . rev .
d * 60 * ( 1999 ) 074504 [ arxiv : hep - lat/9901001 ] .
s. aoki _ et al . _
[ jlqcd collaboration ] , phys .
* 91 * ( 2003 ) 212001 [ arxiv : hep - ph/0307039 ] .
v. gimenez and j. reyes , nucl .
proc . suppl .
* 94 * ( 2001 ) 350
[ arxiv : hep - lat/0010048 ] . n. yamada _
[ jlqcd collaboration ] , nucl .
* 106 * ( 2002 ) 397 [ arxiv : hep - lat/0110087 ] .
m. wingate , c. t. h. davies , a. gray , g. p. lepage and j. shigemitsu , phys .
* 92 * ( 2004 ) 162001 [ arxiv : hep - ph/0311130 ] ; a.
gray _ et al . _
[ hpqcd collaboration ] , phys .
* 95 * ( 2005 ) 212001 [ arxiv : hep - lat/0507015 ] .
e. dalgic _ et al .
_ , arxiv : hep - lat/0610104 ; j. shigemitsu for hpqcd collaboration , talk at lattice 2006 , http://www.physics.utah.edu/lat06/ abstracts/sessions/weak/s1//[email protected] .
m. jamin and b. o. lange , phys .
d * 65 * ( 2002 ) 056005 [ arxiv : hep - ph/0108135 ] .
j. g. korner , a. i. onishchenko , a. a. petrov and a. a. pivovarov , phys .
* 91 * ( 2003 ) 192002 [ arxiv : hep - ph/0306032 ] .
u. nierste , _ @xmath32 and @xmath8 mixing : mass and width differences and cp violation _ , talk at _ beauty 2006 _ , 25 - 29 sep 2006 , oxford , england , uk .
arxiv : hep - ph/0612310 . v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 98 * ( 2007 ) 121801 [ arxiv : hep - ex/0701012 ] .
v. m. abazov _ et al .
_ [ d0 collaboration ] , arxiv : hep - ex/0701007 . v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
d * 74 * ( 2006 ) 092001 [ arxiv : hep - ex/0609014 ] .
m. wirbel , b. stech and m. bauer , z. phys .
c * 29 * ( 1985 ) 637 .
m. bauer , b. stech and m. wirbel , z. phys .
c * 34 * ( 1987 ) 103 .
j. m. soares , phys .
d * 53 * ( 1996 ) 241 [ arxiv : hep - ph/9503285 ] .
h. y. cheng , z. phys .
c * 69 * ( 1996 ) 647 [ arxiv : hep - ph/9503219 ] . v. m. abazov _ et al . _
[ d0 collaboration ] , arxiv : hep - ex/0702030 . | i summarize the observables constraining the @xmath0 mixing complex and present a new calculation of the element @xmath1 of the decay matrix .
@xmath1 enters the prediction of the width difference @xmath2 , for which we obtain @xmath3 , if no new physics enters @xmath0 mixing .
applying our formulae to tevatron data we find a deviation of the @xmath0 mixing phase @xmath4 from its standard model value by 2 standard deviations .
i stress that present data do not give any information on the sign of @xmath2 . |
physics of matter under high baryon density is one of the most challenging problems in quantum chromodynamics ( qcd ) both from technical and physical point of view .
so far there have been proposed various novel phases which include the @xmath9 neutron superfluidity @xcite , the pion @xcite and kaon @xcite condensations , the deconfined quark matter @xcite , the strange matter @xcite , the color superconductivity @xcite , the ferromagnetic quark ordering @xcite and so on .
unfortunately , analyzing these phases from the first principle lattice simulations is retarded due to the complex fermion determinant for 3-color qcd at finite baryon chemical potential with quarks in the fundamental representation .
the situation is however different for 2-color qcd in which the fermion determinant can be made real and the monte - carlo simulations are attainable as was first demonstrated in @xcite .
because of this reason , 2-color qcd provides an unique opportunity to compare various ideas at finite chemical potential with the results from lattice simulations .
one of the major differences between 2-color qcd and 3-color qcd lies in the fact that the color - singlet baryons are bosons in the former .
this implies that the ground state of the 2-color system at finite baryon density in the color - confined phase is an interacting boson system , i.e. a bose liquid , although the quark fermi liquid may be realized at high baryon density in the color - deconfined phase .
how this bose liquid changes its character as a function of the temperature @xmath1 , the quark chemical potential @xmath2 , and the quark mass @xmath3 is an interesting question by itself and may also give a hint to understand physics of the color superconductivity in 3-color qcd in which the crossover from the bose - einstein condensate of tightly bound quark pairs to the bcs type condensate of loosely bound cooper pairs may take place @xcite .
in the present paper , we revisit the thermodynamics of the strong coupling limit of 2-color lattice qcd with staggered fermions with chiral and diquark condensates which was originally studied in @xcite .
our main purpose is to analyze its phase structure and the interplay among the chiral condensate @xmath6 , the diquark condensate @xmath7 , and the baryon density @xmath8 as functions of @xmath1 , @xmath2 , and @xmath3 .
this would give not only a useful guide to the actively pursued lattice qcd simulations of the same system in the weak coupling @xcite but also give us physical insight into the bose liquid together with the other knowledge from the instanton liquid model @xcite , the random matrix model @xcite , the chiral perturbation theory @xcite , and the renormalization group @xcite .
this paper is organized as follows . in sec .
[ sec : formulation ] , we introduce 2-color qcd with one - component staggered quarks ( which correspond to quarks with 4 flavors in the continuum limit ) . by introducing the auxiliary fields @xmath10 and @xmath11 corresponding to the chiral and diquark condensates , we derive an effective free energy @xmath12 $ ] in the @xmath4 expansion ( @xmath5 being the number of spatial dimensions ) and the mean field approximation .
the resultant free energy is enough simple so that one can make analytic studies at least in the chiral limit @xmath13 with finite @xmath1 and @xmath2 , and in the zero temperature limit @xmath14 with finite @xmath2 and @xmath3 .
[ sec : analytic ] is devoted to such studies and useful relations of the critical temperature and chemical potential are derived . in sec .
[ sec : numerical ] , we make numerical analyses on the chiral and diquark condensates as well as the quark number density as functions of @xmath1 , @xmath2 , and @xmath3 .
the phase structure of 2-color qcd at strong coupling is clarified in the three dimensional @xmath15 space . in sec .
[ sec : summary ] , summary and concluding remarks are given . in appendices
[ app : summation ] and [ app : proof ] , we give some technical details in deriving and analyzing the free energy .
in this section , we derive an effective action for meson and diquark fields starting from the lattice action with the staggered fermion .
the readers who are interested in the results in advance can skip the derivation and directly refer to the expression of the free energy given in eq .
( [ eq : free_energy ] ) with eq .
( [ eq : quasi_energy ] ) .
the action on the lattice is given by @xmath16=s_{\mathrm{g}}[u ] + s_{\mathrm{f}}[u,\chi,\bar{\chi}]\ , , \label{eq : action}\end{aligned}\ ] ] which consists of the gluonic part , @xmath17=\frac{2n_{\mathrm{c}}}{g^2}\sum_{x,\mu,\nu } \left\{1-\frac{1}{n_{\mathrm{c}}}\mathrm{re}\mathrm{tr}\ , u_{\mathrm{\mu\nu}}(x)\right\}\ , , \qquad u_{\mu\nu}(x)=u_\nu^\dagger(x)u_\mu^\dagger(x+\hat{\nu } ) u_\nu(x+\hat{\mu})u_\mu(x)\ , , \label{eq : action_gluon}\end{aligned}\ ] ] and the fermionic part with finite chemical potential @xcite , @xmath18=m\sum_x\bar{\chi}(x)\chi(x ) & + \frac{1}{2}\sum_x\sum_{j=1}^d\eta_j(x ) \left\{\bar{\chi}(x)u_j(x)\chi(x+\hat{j } ) -\bar{\chi}(x+\hat{j})u_j^\dagger(x)\chi(x)\right\}\\ & + \frac{1}{2}\sum_x\eta_0(x)\left\{\bar{\chi}(x)\mathrm{e}^\mu u_0(x)\chi(x+\hat{0})-\bar{\chi}(x+\hat{0})u_0^\dagger(x ) \mathrm{e}^{-\mu}\chi(x)\right\}\ , .
\end{split } \label{eq : action_quark}\end{aligned}\ ] ] @xmath19 stands for the quark field in the fundamental representation of the color @xmath20 group and @xmath21 is the @xmath20 valued gauge link variable .
@xmath22 is the gauge coupling constant and @xmath5 represents the number of spatial directions which takes 3 in reality .
later we sometimes use a notation @xmath23 in which @xmath24 ( @xmath25 ) represents the temporal ( spatial ) coordinate .
@xmath26 and @xmath27 inherent in the staggered formalism are defined as @xmath28 @xmath2 in eq .
( [ eq : action_quark ] ) is the quark chemical potential , while the temperature @xmath29 with @xmath30 being the lattice spacing and @xmath31 being the number of temporal sites .
we will write all the dimensionful quantities in unit of @xmath30 and will not write @xmath30 explicitly . since we are interested in 2-color qcd in this paper , we take @xmath0 in the following .
it is worth mentioning here on the symmetry breaking pattern of 2-color qcd formulated on the lattice .
the staggered fermion with a single component , which corresponds to @xmath32 in the continuum limit , has the global @xmath33 symmetry for @xmath34 @xcite . here
@xmath35 ( @xmath36 ) corresponds to the conservation of the baryon number ( axial charge ) . for @xmath0 , @xmath33
is graded to a larger symmetry u(2 ) @xcite .
this is because the color su(2 ) group is pseudo - real and the action possesses pauli - grsey s fermion@xmath37anti - fermion symmetry @xcite .
introduction of a finite chemical potential @xmath2 explicitly breaks the u(2 ) symmetry down to the original @xmath33 .
further introduction of a finite quark mass @xmath3 retains only the @xmath35 symmetry .
listed in table [ tab : symmetry ] are the symmetries realized in various circumstances and their breaking patterns .
.symmetry realized in the single component staggered - fermion action for @xmath0 .
possible symmetry breaking patterns and the number of nambu - goldstone ( ng ) modes are also listed .
[ cols="^,^,^",options="header " , ] before addressing the computational details to derive the effective action for the meson and diquark system from the original action eq .
( [ eq : action ] ) , we shall summarize the actual procedure which is similar to that of @xcite : [ step 1 ] strong coupling limit @xmath38 is taken .
then the gluonic part of the action @xmath39 $ ] in eq .
( [ eq : action_gluon ] ) vanishes , because it is inversely proportional to @xmath40 .
consequently the gauge field remains only in the fermionic part ( [ eq : action_quark ] ) .
[ step 2 ] large dimensional ( @xmath4 ) expansion is employed in the spatial directions in order to facilitate the integration over the spatial link variable @xmath41 .
the temporal link variable @xmath42 is left untouched at this step and will be exactly integrated out later in step 4 . [ step 3 ] bosonization is performed by introducing the auxiliary fields @xmath10 for @xmath43 and @xmath11 for @xmath44 . then the mean field approximation is adopted for the auxiliary fields .
namely , @xmath10 and @xmath11 are regarded as spatially uniform condensates .
[ step 4 ] integration with respect to @xmath19 , @xmath45 , and @xmath42 are accomplished exactly to result in an effective free energy written in terms of @xmath10 ( chiral condensate ) and @xmath11 ( diquark condensate ) .
some comments are in order here to emphasize the importance of the exact integration over @xmath42 without using the @xmath4 expansion in step 4 .
suppose we performed the @xmath4 expansion not only in the spatial directions but also in the temporal direction to perform the @xmath41 and @xmath42 integrations .
then , the quarks would be totally replaced by the mesons and diquarks , and the temperature would not enter into the free energy within the mean field approximation .
in other words , one needs to consider meson and diquark fluctuations to obtain thermal effects in such an approach . in our procedure , on the other hand , the thermal effect can be incorporated even in the mean field level by exempting the @xmath4 expansion in making @xmath42 integration .
similar observation has been made also in an exactly solvable many fermion system @xcite , in which it was shown that the approach similar to ours has better convergence to the exact result .
let us now describe each procedure in some details with a particular emphasis on the importance of the interplay between the chiral and diquark condensates .
after taking the strong coupling limit @xmath46 , the partition function is written as @xmath47\,\mathcal{d}[u_0]\ , \mathcal{d}[u_j]\,\mathrm{e}^{-s_{\mathrm{f}}[u,\chi,\bar{\chi}]}\,.\end{aligned}\ ] ] because @xmath19 and @xmath45 are fermion fields , the taylor expansion of @xmath48}$ ] generates at most @xmath49 terms on each site @xmath50 .
thus the integration with respect to the link variable could be performed exactly in principle .
instead , we truncate such an expansion here up to the leading order in the @xmath4 expansion . as explained soon below , the lowest order of the taylor expansion
gives the leading contribution in the @xmath4 expansion .
then we can integrate each term with respect to @xmath41 at each site @xmath50 .
as we stated in step 2 of the above summary , @xmath42 is left untouched at this stage : @xmath51\,\exp\left[-\frac{1}{2 } \sum_{j=1}^d \eta_j(x)\left\{\bar{\chi}(x)u_j(x)\chi(x+\hat{j } ) -\bar{\chi}(x+\hat{j})u_j^\dagger(x)\chi(x)\right\}\right]\notag\\ & = 1+\frac{1}{8}\sum_{j=1}^d\bar{\chi}^a(x)\chi^a(x)\bar{\chi}^b ( x+\hat{j})\chi^b(x+\hat{j})-\frac{1}{16}\sum_{j=1}^d\epsilon_{ab } \bar{\chi}^a(x)\bar{\chi}^b(x)\epsilon_{cd}\chi^c(x+\hat{j } ) \chi^d(x+\hat{j})+\text{h.\,c.}+\mathrm{o}(1/d)\notag\\ & = \exp\left[\sum_y m(x)v_{\mathrm{m}}(x , y)m(y)+\sum_y \bar{b}(x)v_{\mathrm{b}}(x , y)b(y)\right]+\mathrm{o}(1/d ) , \label{eq : action_mb}\end{aligned}\ ] ] where the latin indices @xmath52 are summed over in the color space . in integrating the link variable ,
we have utilized the formulae for the su(2 ) group integration , @xmath53=1\,,\quad \int\mathrm{d}[u]\,u_{ab}u^\dagger_{cd}=\frac{1}{2 } \delta_{ad}\delta_{bc}\,,\quad \int\mathrm{d}[u]\,u_{ab}u_{cd}=\frac{1}{2}\epsilon_{ac } \epsilon_{bd}\,.\end{aligned}\ ] ] the mesonic composite @xmath54 and the baryonic composite @xmath55 are defined respectively as @xmath56 and their propagators in the spatial directions are given by @xmath57 the reason why the bilinear forms of @xmath54 and @xmath55 correspond to the leading order of the @xmath4 expansion can be understood by changing the normalization as @xmath58 and @xmath59 so that the mesonic propagator @xmath60 is o(1 ) in the large @xmath5 limit . as a result ,
the more @xmath54 is contained in the higher order terms in the taylor expansion , the more @xmath61 is associated with it .
the same argument holds exactly for the baryonic composite @xmath55 with @xmath0 , where @xmath55 is composed of two quarks . for @xmath62 , however , the baryonic contribution is of higher order in the @xmath4 expansion as compared with the mesonic one . for further details on the @xmath4 expansion for general @xmath63 ,
see also the discussion in @xcite .
the resultant action ( [ eq : action_mb ] ) describes the nearest neighbor interaction between the mesonic composites and between the baryonic composites . since they are the four - fermion interactions , one may linearize them with a standard gaussian technique with the auxiliary fields @xmath10 and
@xmath11 ; @xmath64 = \int\mathcal{d}[\sigma]\,\exp\left[-\sum_{x , y}\left\ { \sigma(x)v_{\mathrm{m}}(x , y)\sigma(y ) + 2\sigma(x)v_{\mathrm{m}}(x , y)m(y)\right\}\right]\ , , \intertext{and } \begin{split } & \exp\left[\sum_{x , y}\bar b(x)v_{\mathrm{b}}(x , y)b(y)\right]\\ & \qquad\quad=\int\mathcal d[\delta]\,\exp\left [ -\sum_{x , y}\left\{\delta^\ast(x)v_{\mathrm{b}}(x , y)\delta(y ) -\delta^\ast(x)v_\mathrm{b}(x , y)b(y)-\bar b(x)v_{\mathrm{b}}(x , y ) \delta(y)\right\}\right]\ , .
\end{split}\end{aligned}\ ] ] from the above transformations , it is easy to show the relation , @xmath65 where we have intentionally chosen the definition of the field @xmath10 so that the sign of @xmath66 becomes positive for @xmath67 .
now we take a mean field approximation .
namely , we replace the auxiliary fields @xmath68 and @xmath69 by the constant condensates @xmath10 and @xmath11 and ignore any fluctuations around the condensates .
it is obvious from eq .
( [ eq : condensate ] ) that @xmath10 should be identified with the chiral condensate and @xmath11 be the diquark condensate . then we can write the partition function as @xmath70\,\mathcal{d}[\chi,\bar{\chi}]\ , \mathrm{e}^{-s'[u_0,\chi,\bar{\chi};\sigma,\delta]}\ , , \label{eq : partition } \intertext{with } \begin{split } s'[u_0,\chi,\bar{\chi};\sigma,\delta ] & = \sum_x\left[m\bar\chi(x)\chi(x)+\frac{1}{2}\left\{\bar\chi(x ) \mathrm{e}^\mu u_0(x)\chi(x+\hat0)-\bar{\chi}(x+\hat0 ) u_0^\dagger(x)\mathrm{e}^{-\mu}\chi(x)\right\}\right.\\ & \qquad\qquad\qquad\left.+\frac{d}{2}\sigma^2+\frac{d}{2}|\delta|^2 + d\,\sigma m(x)-\frac{d}{2}\delta^\ast b(x ) -\frac{d}{2}\delta\bar{b}(x)\right]\ , .
\end{split } \label{eq : priact}\end{aligned}\ ] ] since @xmath71 $ ] is in a bilinear form with respect to the quark fields @xmath19 and @xmath45 , we can integrate out them immediately .
the integration with respect to @xmath42 , which seems to be tough at first glance , turns out to be feasible as demonstrated in the next subsection . in order to complete the remaining integrals
, we adopt a particular gauge in which @xmath72 is diagonal and independent of @xmath24 ( often called the polyakov gauge ) , @xmath73 also we make a partial fourier transformation for the quark fields ; @xmath74 the anti - periodic condition in the temporal direction , @xmath75 , is satisfied owing to the fermionic matsubara frequency @xmath76 . substituting eqs .
( [ eq : gauge_choice ] ) and ( [ eq : fourier ] ) into the action ( [ eq : priact ] ) and taking the summation over @xmath24 , we reach the action in a nambu - gorkov representation ; @xmath77 = \sum_x\left(\frac{d}{2}\sigma^2+\frac{d}{2}|\delta|^2\right ) -\frac{1}{2}\sum_{\vec{x}}\sum_{m , n}\,\ !
[ x^a(m,\vec{x})]^{\mathrm{t}}\,g^{-1}_{ab}(m , n;\theta(\vec{x}))\ , x^b(n,\vec{x})\,,\end{aligned}\ ] ] where @xmath78 and @xmath79\delta_{mn}\delta_{ab}\\ -\left[m+\mathrm{i}\sin\left(k_m+\frac{\theta_a(\vec{x } ) } { n_\tau}-\mathrm{i}\mu\right)\right]\delta_{mn}\delta_{ab } & -\frac{d}{2}\delta\,\delta_{m , n_\tau - n+1}\epsilon_{ab } \end{pmatrix}\,.\end{aligned}\ ] ] the indices @xmath30 and @xmath80 run from 1 through 2 in the color space . @xmath81 denotes the dynamical quark mass ( to be distinguished from the mesonic composite field @xmath54 ) defined by @xmath82 we can perform the grassmann integration over @xmath19 and @xmath45 as @xcite @xmath83\,\mathrm{e}^{x^{\mathrm{t}}g^{-1}x } = \prod_{\vec{x}}\sqrt{\mathrm{det}\left [ g^{-1}_{ab}(m , n;\theta(\vec{x}))\right]}\ , .
\label{eq : grassmann_integral}\end{aligned}\ ] ] det stands for the determinant with respect to the color indices and the matsubara frequencies . the square root of the determinant may be simplified as @xmath84}=\prod_{m=1}^{n_\tau } \left[\left(\frac{d}{2}\right)^2|\delta|^2+\left\{m+\mathrm{i}\sin \left(k_m+\theta / n_\tau-\mathrm{i}\mu\right)\right\}\cdot\left\{m -\mathrm{i}\sin\left(k_m+\theta / n_\tau+\mathrm{i}\mu\right)\right\ } \right]\ , , \label{eq : pre - det}\end{aligned}\ ] ] where @xmath85 is substituted . the product with respect to @xmath3 can be performed using a technique similar to that in the calculation of the free energy in finite - temperature field theory in the continuum .
the details of the calculation is given in appendix [ app : summation ] .
the result turns out to be a rather simple form , @xmath86 } = \left(\cos\theta+\cosh n_\tau e_-\right)\cdot \left(\cos\theta+\cosh n_\tau e_+\right)\ , .
\label{eq : sqrtdet}\end{aligned}\ ] ] here @xmath87 is the excitation energy of quasi - quarks , @xmath88 with the dynamical quark mass @xmath81 defined in eq .
( [ eq : d_mass ] ) .
finally , all we have to do is to integrate this resultant determinant with respect to @xmath42 , or @xmath89 , to derive the effective action ; @xmath90=-\log z = \sum_x\left(\frac{d}{2}\sigma^2+\frac{d}{2}|\delta|^2\right ) -\sum_{\vec{x}}\log\left\{\int\frac{\mathrm{d}\theta(\vec{x})}{2\pi } \sin^2\theta(\vec{x})\sqrt{\mathrm{det } \left[g^{-1}_{ab}(m , n;\theta(\vec{x}))\right]}\right\}\ , , \label{eq : effact}\end{aligned}\ ] ] where we have used the su(2 ) haar measure in the polyakov gauge ( [ eq : gauge_choice ] ) , @xmath91=\left.\prod_{\vec{x}}\int_{-\pi}^\pi \frac{\mathrm{d}\theta(\vec{x})}{2\pi}\sin^2\theta(\vec{x } ) \right|_{\theta(\vec{x})=\theta_1(\vec{x})=-\theta_2(\vec{x})}\ , .
\label{eq : measure}\end{aligned}\ ] ] the @xmath42 integration projects out the color singlet pairing of quarks and anti - quarks among arbitrary excitations in the determinant .
thus what excites thermally is no longer single quarks , but color singlet mesons or diquarks . in this sense ,
the strong coupling limit in 2-color qcd inevitably leads to the boson system .
( [ eq : sqrtdet ] ) , ( [ eq : effact ] ) , and ( [ eq : measure ] ) immediately yield the effective free energy ; @xmath92 = s_{\mathrm{eff}}/({\textstyle \sum_x})=\frac{d}{2}\sigma^2 + \frac{d}{2}|\delta|^2-t\log\left\{1 + 4\cosh\left(e_+/t\right)\cdot \cosh\left(e_-/t\right)\right\}\ , , \label{eq : free_energy}\end{aligned}\ ] ] where we have rewritten @xmath31 in terms of the temperature @xmath93 .
although the quasi - quark energy @xmath94 given in eq .
( [ eq : quasi_energy ] ) takes a complicated form , @xmath94 is reduced to a simple expression in the _ naive _ continuum @xmath95 with @xmath30 being the lattice spacing .
assuming @xmath96 , @xmath97 , @xmath98 in this limit , @xmath94 amounts to a familiar form in the continuous space - time , @xmath99 the quasi - quarks are static in this framework because of the mean field approximation .
in the chiral limit @xmath13 with zero chemical potential @xmath100 , the free energy in the mean field approximation given in eq .
( [ eq : free_energy ] ) is a function only in terms of @xmath101 . as a result ,
the free energy is invariant under the transformation mixing the chiral condensate with the diquark one .
this corresponds to a subgroup of the u(2 ) symmetry of the original action at @xmath34 in table [ tab : symmetry ] . because of this symmetry , the chiral condensate is indistinguishable from the diquark condensate for @xmath34 , so that a state with finite @xmath10 can be arbitrarily transformed to a state with finite @xmath11 .
finite @xmath2 would act on the free energy as an external field tending to make the diquark condensation favored .
consequently even an infinitesimal introduction of @xmath2 leads to the diquark condensation phase with zero chiral condensate in the chiral limit .
this is a peculiar feature of 2-color qcd and is in a sharp contrast to the 3-color qcd .
in fact , we can prove that the minimizing condition for our @xmath12 $ ] does not allow non - zero @xmath10 at any @xmath1 and @xmath2 in the chiral limit .
see appendix [ app : proof ] for the proof .
taking the fact that @xmath10 vanishes in the chiral limit for granted , the free energy is simply written as @xmath102 = \frac{d}{2}|\delta|^2-t\log\left\ { 1 + 4\cosh^2\left(e_0/t\right)\right\}\,,\qquad e_0=\mathrm{arccosh}\left(\sqrt{(d/2)^2|\delta|^2+\cosh^2\mu } \right)\ , .
\label{eq : free_energy_zero}\end{aligned}\ ] ] assuming that the finite @xmath1 phase transition with fixed @xmath2 is of second order , which will be confirmed numerically later , the critical temperature @xmath103 may be determined by expanding @xmath104 $ ] in terms of @xmath105 and extracting the point where the coefficient of @xmath105 changes its sign .
since the expansion reads @xmath106 = -t\log\left\{3 + 2\cosh\left(2\mu / t\right)\right\ } + \left\{\frac{d}{2}-\frac{d^2}{3 + 2\cosh\left(2\mu / t\right ) } \frac{\sinh\left(2\mu / t\right)}{\sinh2\mu}\right\}|\delta|^2 + \mathrm{o}\left(|\delta|^4\right)\,,\end{aligned}\ ] ] one finds @xmath107 the values of @xmath108 for two typical cases are @xmath109 for @xmath110 .
@xmath111 is a monotonically decreasing function of @xmath2 connecting the above two points .
this will be discussed later in sec .
[ sec : numerical ] together with the case at finite @xmath3 .
it is interesting to calculate the quark number density @xmath112 which is defined as @xmath113 where @xmath114 is the quasi - quark mass in the chiral limit defined in eq .
( [ eq : free_energy_zero ] ) .
although the above expression seems to be a little complicated , it can be simplified by using the gap equation ; @xmath115 in general the gap equation may have two solutions , @xmath116 and @xmath117 .
the former ( the latter ) is the solution to minimize the free energy for @xmath118 ( @xmath119 ) .
it can be alternatively said that the former ( the latter ) is the solution for @xmath120 ( @xmath121 ) with the critical chemical potential @xmath122 as a solution of eq .
( [ eq : critical_temp ] ) in terms of @xmath2 . by eliminating most of the complicated part of eq .
( [ eq : density ] ) by means of the gap equation ( [ eq : gap_eq ] ) , we have the following expressions ; @xmath123 at zero temperature , eq . ( [ eq : quark_density ] )
is reduced to @xmath124 with @xmath125 at which the quark number density @xmath112 gets to be saturated to two .
note that quarks on each lattice site have only two degrees of freedom in the color space because the spin and flavor degrees of freedom are dispersed on the staggered lattice sites .
therefore `` two '' is the maximum number of quarks placed on each lattice site due to the pauli exclusion principle .
even a small quark mass could modify the phase structure substantially from that given in sec .
[ sec : analytic - a ] .
this is because the pion , that is the nambu - goldstone mode associated with the chiral @xmath36 symmetry breaking , comes to have a finite mass , @xmath126 .
as a result , as far as the quark chemical potential is smaller than a threshold value , the vacuum is empty and the diquark condensate vanishes ; @xmath127 and @xmath128 . to clarify such threshold effects in more detail ,
let us now focus our attention on the @xmath14 system with finite @xmath3 .
we derive analytic formulae for lower - critical chemical potential @xmath129 at which @xmath11 starts to be non - vanishing and the upper - critical chemical potential @xmath130 at which @xmath11 ceases to be non - vanishing . the free energy at @xmath14 with finite @xmath3 takes a simple form ; @xmath92 = \frac{d}{2}\sigma^2+\frac{d}{2}|\delta|^2-\left(e_++e_-\right)\,.\end{aligned}\ ] ] the expansion of @xmath12 $ ] in terms of @xmath105 near the threshold gives a condition to determine the critical chemical potential @xmath122 , @xmath131 where @xmath132 and @xmath133 as defined in eq . ( [ eq : d_mass ] ) . in eq .
( [ eq : sta_chem ] ) the dynamical mass @xmath81 , or @xmath10 , is determined by the condition to minimize the free energy at the threshold .
we can reduce that free energy much more by putting @xmath116 , @xmath134 = \frac{d}{2}\sigma^2 -\left(e+\mu+|e-\mu|\right)\ , .
\label{eq : free_energy_muc } \end{aligned}\ ] ] the stationary condition of the free energy @xmath135 $ ] with respect to @xmath10 yields the following chiral gap equations ; @xmath136 combining eq . (
[ eq : sta_chem ] ) and eq .
( [ eq : gap_eq_chiral ] ) , we find @xmath137 where we define @xmath138 as the solution of the equation , @xmath139 with finite @xmath3 at @xmath14 , the empty vacuum gives @xmath140 and @xmath116 as long as @xmath141 . on the other hand ,
the non - vanishing value of @xmath11 is possible for @xmath142 . for @xmath143 ,
the saturation of the quark number density occurs leading to @xmath144 and @xmath116 .
these behaviors will be confirmed numerically in sec .
[ sec : numerical ] .
note that we have not assumed the quark mass @xmath3 to be small , which is in contrast to the approaches based on chiral perturbation theory @xcite .
therefore , eq . ( [ eq : muc - threshold ] ) can relate the critical chemical potentials to arbitrary values of @xmath3 . for sufficiently small @xmath3
, we have a relation ; @xmath145 the former may be rewritten as @xmath146 with @xmath147 obtained from the excitation spectrum in the vacuum @xcite up to the leading order of the @xmath4 expansion .
this observation is consistent with the discussion given in @xcite .
the latter relation is simply rewritten as @xmath148 with @xmath149 defined in eq.([eq : muc0 ] ) . for sufficiently large @xmath3 , we find asymptotically , @xmath150 this is because @xmath151 for large @xmath3 as is evident from eqs .
( [ eq : gap_eq_chiral ] ) and ( [ eq : gap_eq_k ] ) .
in this section , we determine the chiral condensate @xmath10 and the diquark condensate @xmath11 numerically by minimizing the effective free energy in eq .
( [ eq : free_energy ] ) .
the quark number density @xmath112 is also calculated numerically .
the results are shown in figs .
[ fig : massless ] and [ fig : massive ] for @xmath13 and @xmath152 , respectively .
the phase diagrams of the system in the @xmath1-@xmath2 plane , in the @xmath2-@xmath3 plane , and in the three dimensional @xmath1-@xmath2-@xmath3 space are also shown in figs .
[ fig : phase_diagram ] , [ fig : phase_diagram - mum ] , and [ fig : phase_diagram-3d ] , respectively . , diquark condensate @xmath11 , and the quark number density @xmath112 for @xmath13 with @xmath110 . in the left panels
they are plotted as functions of the chemical potential @xmath2 for two typical values of temperature . in the right panels , they are plotted as functions of temperature @xmath1 for two typical values of @xmath2 .
all the dimensionful quantities are in unit of the lattice spacing @xmath30 , which is implicitly understood in all the figures in this section.,width=604 ] let us first consider the diquark condensate and the quark number density as functions of @xmath2 and @xmath1 shown in the four panels of fig .
[ fig : massless ] .
( note that the chiral condensate is always zero in the chiral limit as we have proved in appendix [ app : proof ] . ) at @xmath14 ( the upper left panel ) , the diquark condensate @xmath11 decreases monotonously as a function of @xmath2 and shows a second order transition when @xmath2 becomes of order unity . on the other hand ,
the quark number density @xmath112 increases linearly for small @xmath2 and grows more rapidly for large @xmath2 until the saturation point where quarks occupy the maximally allowed configurations by the fermi statistics .
( see the analytic formula given in eq .
( [ eq : quark_density_t0 ] ) . ) those behaviors of @xmath11 and @xmath112 are also observed in the recent monte - carlo simulations of 2-color qcd @xcite .
note that the rapid increase of @xmath112 near the upper - critical chemical potential @xmath130 takes place even in the strong coupling limit as shown here .
namely it does not necessarily be an indication of the existence of free quarks at high density unlike the suggestion given in the last reference in @xcite . as @xmath1 increases
, the magnitude of the diquark condensate decreases by the thermal excitations of quark and anti - quark pairs in the last term of eq .
( [ eq : free_energy ] ) , which is shown in the lower left panel of fig .
[ fig : massless ] .
it is worth mentioning here that the diquark condensate disappears even before the complete saturation ( @xmath144 ) takes place .
next we consider the diquark condensate as a functions of @xmath1 for two typical values of the chemical potential in the right panels of fig .
[ fig : massless ] .
the diquark condensate shows a second order transition at @xmath103 given analytically by eq .
( [ eq : critical_temp ] ) . , diquark condensate @xmath11 , and quark number density @xmath112 for @xmath152 with @xmath110 .
all the definitions are the same with those of fig .
[ fig : massless ] except that the dotted line indicates a total magnitude of the condensates @xmath154.,width=604 ] in the upper left panel of fig .
[ fig : massive ] , the chiral condensate , the diquark condensate , and the quark number density are shown as functions of @xmath2 for small quark mass @xmath152 at @xmath14 . as we have discussed in sec .
[ t0mn0 ] , there exists a lower - critical chemical potential @xmath129 given by eq .
( [ eq : muc - threshold ] ) . both @xmath11 and @xmath112
start to take finite values only for @xmath155 at @xmath14 .
one can view the behavior of chiral and diquark condensates with finite quark mass as the manifestation of two different mechanisms : one is a continuous `` rotation '' from the chiral direction to the diquark direction near @xmath156 or @xmath157 with @xmath154 ( shown by the dotted line ) varying smoothly .
the other is the saturation effect which forces the diquark condensate to decrease and disappear for large @xmath2 as seen in the previous case of @xmath13 .
the `` rotation '' can be understood as follows : the free energy in the mean field approximation at small @xmath3 and @xmath2 has an approximate symmetry which mixes the chiral condensate with the diquark one as we have discussed in sec .
[ sec : analytic - a ] .
the effect of @xmath3 ( @xmath2 ) is to break this symmetry in the direction of the chiral ( diquark ) condensation favored .
therefore , a relatively large chiral condensate predominantly appears for small @xmath2 region .
, thus the chiral condensate is always non - vanishing although it is suppressed in magnitude at high @xmath1 and @xmath2 . ] just above the lower - critical chemical potential @xmath129 , the chiral condensate decreases , while the diquark condensate increases because the effect of @xmath2 surpasses that of @xmath3 .
as @xmath2 becomes large , the diquark condensate begins to decrease in turn by the effect of the saturation and eventually disappears when @xmath2 exceeds the upper - critical value @xmath130 ( order of unity for @xmath14 ) .
similar `` rotation '' and the saturation effect are also seen at finite @xmath1 as shown in the lower left panel of fig .
[ fig : massive ] . at finite @xmath1
, both the chiral and diquark condensates are suppressed due to the effect of temperature and the diquark condensate disappears _ before _ the complete saturation occurs .
next we consider the chiral and diquark condensates as functions of @xmath1 ( the right panels of fig .
[ fig : massive ] ) . at low @xmath1 ,
both the chiral and diquark condensates have finite values for @xmath158 and @xmath159 .
the diquark condensate decreases monotonously as @xmath1 increases and shows a second order transition . on the other hand , the chiral condensate increases as the diquark condensate decreases so that the total condensate @xmath154 is a smoothly varying function of @xmath1 .
the understanding based on the chiral - diquark mixing symmetry is thus valid .
an interesting observation is that the chiral condensate , although it is a continuous function of @xmath1 , has a cusp shape associated with the phase transition of the diquark condensate .
finally let us compare the @xmath13 case in fig .
[ fig : massless ] and the @xmath152 case in fig .
[ fig : massive ] . looking into two figures at the same temperature or chemical potential , we find that the diquark condensate @xmath11 for @xmath13 and the total condensate @xmath154 for @xmath152 have almost the same behavior .
this indicates that although the current quark mass suppresses the diquark condensate , the price to pay is to increase the chiral condensate so as to make the total condensate insensitive to the presence of small quark mass . restating this by use of the radial and the angle variables defined by @xmath160 and @xmath161 as in eq .
( [ eq : radial - angle ] ) , a small @xmath3 hardly changes the behavior of @xmath162 but shifts @xmath163 from zero .
-@xmath3 plane at @xmath14 .
two solid lines separate the region where @xmath116 from the region where @xmath164 .
[ fig : phase_diagram - mum],width=302 ] -@xmath3 plane at @xmath14 .
two solid lines separate the region where @xmath116 from the region where @xmath164 .
[ fig : phase_diagram - mum],width=302 ] now we show the phase diagram of the strong coupling 2-color qcd in the @xmath1-@xmath2 plane in fig . [
fig : phase_diagram ] .
the solid line denotes a critical line separating diquark superfluid phase @xmath164 and the normal phase @xmath116 in the chiral limit @xmath13 , which is determined analytically by eq .
( [ eq : critical_temp ] ) .
the dashed line denotes the critical line determined numerically for @xmath152 .
the phase transition is of second order on these critical lines .
the chiral condensate @xmath10 is everywhere zero for @xmath13 , while it is everywhere finite for @xmath153 . in the latter case
, however , @xmath10 is particularly large in the region between the solid line and the dashed line . shown in fig .
[ fig : phase_diagram - mum ] is a phase diagram in @xmath2-@xmath3 plane at @xmath14 .
the lower right of the figure corresponds to the vacuum with no baryon number present , @xmath140 . on the other hand
, the upper left of the figure corresponds to the saturated system , @xmath144 , in which every lattice site is occupied by two quarks .
there is a region with @xmath165 and @xmath164 bounded by the above two limiting cases , which is separated by two critical lines given in eq.([eq : muc - threshold ] ) .
finally , bringing all the discussions together , the phase structure in the three dimensional @xmath1-@xmath2-@xmath3 space is shown in fig .
[ fig : phase_diagram-3d ] .
the diquark condensate has a none - vanishing value inside the critical surface and the phase transition is of second order everywhere on this critical surface .
the second order phase transition is consistent with other analyses employing the mean field approximation ; the random matrix model and the nambu@xmath37jona - lasinio model @xcite , and also the chiral perturbation theory at @xmath14 @xcite . on the other hand , monte - carlo simulations of 2-color qcd @xcite show indication of a tricritical point in the @xmath1-@xmath2 plane at which the property of the critical line changes from the second order to the first order as @xmath2 increases
this is also supported by the chiral perturbation theory beyond the mean field approximation at finite @xmath1 @xcite . aside from the fact that we are working in the strong coupling limit ( @xmath166 ) , though the lattice simulations are aiming at the weak coupling limit ( @xmath167 ) , it would be of great interests to go beyond the mean field approximation in our analysis and study corrections to the phase structure given in fig .
[ fig : phase_diagram-3d ] . -@xmath2-@xmath3 space .
the surface represents the critical surface for the diquark condensation , which separates the region where @xmath116 from the region where @xmath164 .
[ fig : phase_diagram-3d],width=377 ]
in this paper , we have studied the phase structure of 2-color qcd in the strong coupling limit formulated on the lattice .
we have employed the @xmath4 expansion ( only in the spatial direction ) and the mean field approximation to derive the free energy written in terms of the chiral condensate @xmath10 and the diquark condensate @xmath11 at finite @xmath1 , @xmath2 and @xmath3 .
a major advantage of our approach is that we can derive rather simple formulae in an analytic way for the critical temperature ( @xmath103 ) , the critical chemical potentials ( @xmath129 and @xmath130 ) , and the quark number density ( @xmath112 ) , which are useful to have physical insight into the problem .
although our results are in principle limited in the strong coupling , the behavior of @xmath10 , @xmath11 , and @xmath112 in the @xmath1-@xmath2-@xmath3 space has remarkable qualitative agreement with the recent lattice data .
since we do not have to rely on any assumption of small @xmath2 nor small @xmath3 in our approach , the strong coupling analysis presented in this paper is complementary to the results of the chiral perturbation theory and directly provides a useful guide to the lattice 2color - qcd simulations .
there are several directions worth to be explored starting from the present work . among others ,
the derivation of an effective action written not only with the chiral and diquark fields but also with the polyakov loop , @xmath168 , is interesting because this will allow us a full comparison for all the physical quantities calculated in our formalism and lattice simulations @xcite .
this is a straightforward generalization of the works in @xcite , which allows us to analyze @xmath169 under the influence of chiral and diquark condensates .
also calculation of the meson and diquark spectra including pseudo - scalar and pseudo - diquark channels will give us deeper understanding about the non - static feature of the present system .
another direction is to study the response of the bose liquid discussed in this paper under external fields . although mesons and diquarks are color neutral in 2-color qcd , they can have electric charge so that the external electromagnetic field leads to a non - trivial change of the phase structure if the field intensity is enough strong .
generalization of the whole machinery in this paper to strong coupling 3-color qcd is also one of the challenging problems to be studied in the future .
authors are grateful to k. iida , m. tachibana , s. sasaki , h. abuki , and k. itakura for continuous and stimulating discussions on 2-color qcd .
thanks g. baym for discussions on the bose - einstein condensation of tightly bound diquarks .
this work is partially supported by the grants - in - aid of the japanese ministry of education , science and culture ( no .
15540254 ) .
k. f. is supported by the japan society for the promotion of science for young scientists .
let us first take the logarithm of eq.([eq : pre - det ] ) ; @xmath170 } & = \sum_{m=1}^{n_\tau}\log\left[-\cos^2\left(k_m+\theta / n_\tau\right ) + 2y\cos\left(k_m+\theta / n_\tau\right)+x^2\right]\,.\end{aligned}\ ] ] here @xmath171 and @xmath172 are defined as @xmath173 by differentiating @xmath174}$ ] with respect to @xmath171 , we obtain @xmath175 } = \frac x{\sqrt{x^2+y^2}}\sum_{m=1}^{n_\tau } \left[\frac1{\cos\left(k_m+\theta / n_\tau\right)-y+\sqrt{x^2+y^2 } } -\frac1{\cos\left(k_m+\theta / n_\tau\right)-y-\sqrt{x^2+y^2}}\right]\,.\end{aligned}\ ] ] because @xmath176 is invariant under the shift @xmath177 , we can make the summation of @xmath3 over the range @xmath178 to @xmath179 with an appropriate degeneracy factor @xmath180 . then the residue theorem enables us to replace the summation by a complex integral as @xmath181 } = \frac{x}{\sqrt{x^2+y^2}}\frac{1}{\omega } & \left[\oint\frac{\mathrm{d}z}{2\pi\mathrm{i } } \frac{1}{\cos\left(z+\theta / n_\tau\right)+k_- } \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{\mathrm{i}n_\tau z } } -\sum_{\bar{z}}\frac{1}{-\sin\bar{z } } \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{\mathrm{i}n_\tau\bar{z } } } \right.\\ & \left.-\oint\frac{\mathrm{d}w}{2\pi\mathrm{i } } \frac{1}{\cos\left(w+\theta / n_\tau\right)-k_+ } \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{\mathrm{i}n_\tau w } } + \sum_{\bar{w}}\frac{1}{-\sin\bar{w } } \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{\mathrm{i}n_\tau\bar{w } } } \right]\ , , \end{split } \label{eq : problem}\end{aligned}\ ] ] where @xmath182 owing to the infinite range of the summation over @xmath3 , we can choose the closed contour at infinity for the complex integrals with respect to @xmath183 and @xmath184 , and thus such complex integrals go to zero .
@xmath185 and @xmath186 are the residues satisfying @xmath187 solving these equations , we obtain @xmath188 where @xmath189 substituting eq .
( [ eq : solution ] ) into eq .
( [ eq : problem ] ) , we obtain @xmath190 } \notag\\ & = \frac{x}{\sqrt{x^2+y^2}}\frac{1}{\omega}\sum_{n=-\infty}^\infty \left[\frac{1}{-\mathrm{i}\sqrt{k_-^2 - 1}}\left ( \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{-n_\tau e_--\mathrm{i}\theta + \mathrm{i}n_\tau\pi}}- \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{n_\tau e_--\mathrm{i}\theta + \mathrm{i}n_\tau\pi}}\right)\right . \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\left .
-\frac{1}{\mathrm{i}\sqrt{k_+^2 - 1}}\left ( \frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{-n_\tau e_+-\mathrm{i}\theta } } -\frac{-\mathrm{i}n_\tau}{1+\mathrm{e}^{n_\tau e_+-\mathrm{i}\theta } } \right)\right]\notag\\ & = \frac{\partial e_-}{\partial x } \left\{\frac{n_\tau}{1+\mathrm{e}^{-n_\tau e_--\mathrm{i}\theta } } -\frac{n_\tau}{1+\mathrm{e}^{n_\tau e_--\mathrm{i}\theta}}\right\ } + \frac{\partial e_+}{\partial x } \left\{\frac{n_\tau}{1+\mathrm{e}^{-n_\tau e_+-\mathrm{i}\theta } } -\frac{n_\tau}{1+\mathrm{e}^{n_\tau e_+-\mathrm{i}\theta}}\right\ } \notag\\ & = \frac{\partial}{\partial x } \left[\log\left\{\mathrm{e}^{n_\tau e_- } + \mathrm{e}^{-\mathrm{i}\theta}\right\ } + \log\left\{\mathrm{e}^{-n_\tau e_- } + \mathrm{e}^{-\mathrm{i}\theta}\right\ } + \log\left\{\mathrm{e}^{n_\tau e_+ } + \mathrm{e}^{-\mathrm{i}\theta}\right\ } + \log\left\{\mathrm{e}^{-n_\tau e_+ } + \mathrm{e}^{-\mathrm{i}\theta}\right\}\right]\notag\\ & = \frac{\partial}{\partial x } \left[-2\mathrm i\theta+2\log2 + \log\left\{\cos\theta+\cosh n_\tau e_-\right\ } + \log\left\{\cos\theta+\cosh n_\tau e_+\right\}\right]\,.\end{aligned}\ ] ] we note that the degeneracy factor @xmath180 is just canceled by the infinite degeneracy of the summation on @xmath191 .
also we have used the fact that @xmath31 must be an even integer for the staggered fermion . after the integration with respect to @xmath171 , we find that @xmath192}$ ] is expressed in a rather simple form up to irrelevant factors , @xmath86 } = \left(\cos\theta+\cosh n_\tau e_-\right)\cdot \left(\cos\theta+\cosh n_\tau e_+\right)\ , .
\label{eq : sqrtdet2}\end{aligned}\ ] ] this is the final form shown in eq .
( [ eq : sqrtdet ] ) in the text .
for @xmath13 , it is useful to rewrite the free energy eq .
( [ eq : free_energy ] ) in terms of the radial and angle variables , @xmath194 then the free energy is @xmath195=\frac{d}{2}\lambda^2-t\log\left\ { 1 + 4\cosh\left(e_+/t\right)\cdot\cosh\left(e_-/t\right)\right\}\,,\end{aligned}\ ] ] with the quasi - quark energy given by @xmath196 what we are going to prove here is that @xmath197 $ ] has the global minimum at @xmath198 for arbitrary @xmath1 , @xmath2 , and @xmath162 .
actually we can prove the following statement in a more abstract expression ; @xmath199 we can apply this corollary to our problem to prove that @xmath198 is the global minimum of the free energy .
obviously @xmath198 turns out to be the global maximum of @xmath200 once we substitute , @xmath201\,.\end{aligned}\ ] ] thus we have proven that @xmath197 $ ] is globally minimized at @xmath198 , in other words , the chiral condensate vanishes in the chiral limit .
r. tamagaki , prog .
phys . * 44 * , 905 ( 1970 ) .
+ m. hoffberg , et al .
* 24 * , 775 ( 1970 ) .
+ for further references , see the review : + t. kunihiro , t. muto , t. takatsuka , r. tamagaki , and t. tatsumi , prog .
suppl . * 112 * , 1 ( 1993 ) . j.c .
collins and m.j .
perry , phys .
* 34 * , 1353 ( 1975 ) .
+ g. baym and s.a .
chin , phys . lett .
b * 62 * , 241 ( 1976 ) . + for further references , see the review : + h. heiselberg and m. hjorth - jensen , phys .
rept . * 328 * , 237 ( 2000 ) .
n. itoh , prog .
44 * , 291 ( 1970 ) .
bodmer , phys .
d * 4 * , 1601 ( 1971 ) .
+ e. witten , phys .
d * 30 * , 272 ( 1984 ) . + for further references , see the review : + j. madsen , e - print : .
d. bailin and a. love , phys .
rept . * 107 * , 325 ( 1984 ) .
+ m. iwasaki and t. iwado , phys .
b * 350 * , 163 ( 1995 ) .
+ r. rapp , t. schfer , e.v .
shuryak , and m. velkovsky , phys .
lett . * 81 * , 53 ( 1998 )
. + m.g .
alford , k. rajagopal , and f. wilczek , phys .
b * 422 * , 247 ( 1998 ) . + for further references , see the reviews : + m.g .
alford , ann .
nucl . part .
sci . * 51 * , 131 ( 2001 ) .
+ k. rajagopal and f. wilczek , in _ at the frontier of particle physics : handbook of qcd _
m. shifman ( world scientific singapore , 2001 ) , e - print : . | 2-color qcd ( quantum chromodynamics with @xmath0 ) at finite temperature @xmath1 and chemical potential @xmath2 is revisited in the strong coupling limit on the lattice with staggered fermions .
the phase structure in the space of @xmath1 , @xmath2 , and the quark mass @xmath3 is elucidated with the use of the mean field approximation and the @xmath4 expansion ( @xmath5 being the number of spatial dimensions ) .
we put special emphasis on the interplay among the chiral condensate @xmath6 , the diquark condensate @xmath7 , and the quark density @xmath8 in the @xmath1-@xmath2-@xmath3 space .
simple analytic formulae are also derived without assuming @xmath2 nor @xmath3 being small .
qualitative comparisons are made between our results and those of recent monte - carlo simulations in 2-color qcd . |
the following problem appears in the field of quantum communication and in quantum statistics : a collection of statistical operators with some a priori probabilities ( initial ensemble ) describes the possible initial states of a quantum system and an observer wants to decide in which of these states the system is by means of a quantum measurement on the system itself .
the quantity of information given by the measurement is the classical mutual information @xmath0 of the input / output joint distribution ( shannon information ) .
interesting upper and lower bounds for @xmath0 , due to the quantum nature of the measurement , are given in the literature @xcite , where the measurement is described by a _
generalized observable _ or _ positive operator valued _ ( pov ) _ measure _ ; an exception is the paper @xcite , which considers also the information left in the post - measurement states .
with respect to a pov measure , a more detailed level of description of the quantum measurement is given by an _ instrument _
@xcite : given a quantum state ( the preparation ) as input , the instrument gives as output not only the probabilities of the outcomes but also the state after the measurement , conditioned on the observed outcome ( the a posteriori state ) .
we can think the instrument to be a channel : from a quantum state ( the pre - measurement state ) to a quantum / classical state ( a posteriori state plus probabilities ) .
the mathematical formalization of the idea that an instrument _ is _ a channel is given in section [ instrsec ] , together with a new construction of the a posteriori states . in section [ ment+bounds ] , by using the identification of the instrument with a channel and the notion of quantum mutual entropy , we are able to give a unified approach to various bounds for @xmath0 and for related quantities , which can be thought to quantify the informational performances of the instrument .
one of the most interesting inequality is the strengthening ( [ sww ] ) of holevo s bound ( [ holevos_bound ] ) ; in the finite case it has been obtained in ref .
@xcite where the authors introduce a specific model of the measuring process ( without speaking explicitly of intruments ) and use the strong subadditivity of the von neumann entropy .
the introduction of the general notion of instrument , the association to it of a channel and the use of uhlmann s monotonicity theorem allows us to obtain the same result in a more direct way and to extend it to a more general set up . in section [ hall ] a new upper bound ( [ newbound ] ) for the classical mutual information @xmath0 is obtained by combining an idea by hall @xcite and inequality ( [ sww ] ) .
we already gave some results in @xcite , mainly in the discrete case .
here we give the general results , which are based on the theory of relative entropy on von neumann algebras @xcite .
continuous parameters appear naturally in quantum statistical problems , but also in the quantum communication set up infinite dimensional hilbert spaces and general initial ensembles are needed @xcite .
some of the informational quantities presented here have been studied in @xcite in the case of instruments describing continual measurements .
we denote by @xmath1 the space of bounded linear operators from @xmath2 to @xmath3 , where @xmath4 are banach spaces ; moreover we set @xmath5 .
let @xmath6 be a separable complex hilbert space ; a normal state on @xmath7 is identified with a statistical operator , @xmath8 and @xmath9 are the trace - class and the space of the statistical operators on @xmath6 , respectively , and @xmath10 , @xmath11 , @xmath12 .
more generally , if @xmath13 belongs to a @xmath14-algebra and @xmath15 to its dual @xmath16 or predual @xmath17 , the functional @xmath15 applied to @xmath13 is denoted by @xmath18 .
let @xmath19 be a measure space , where @xmath20 is a @xmath21-finite measure
. by theorem 1.22.13 of @xcite , the @xmath14-algebra @xmath22 ( @xmath14-tensor product ) is naturally isomorphic to the @xmath14-algebra @xmath23 of all the @xmath7-valued @xmath20-essentially bounded weakly@xmath24 measurable functions on @xmath25 . moreover
( @xcite , proposition 1.22.12 ) , the predual of this @xmath14-algebra is @xmath26 , the banach space of all the @xmath8-valued bochner @xmath20-integrable functions on @xmath25 , and this predual is naturally isomorphic to @xmath27 ( tensor product with respect to the greatest cross norm @xcite , pp . 45 , 58 , 59 , 67 , 68 ) .
let us note that a normal state @xmath28 on @xmath23 is a measurable function @xmath29 , @xmath30 , such that @xmath31 is a probability density with respect to @xmath20 .
a _ channel _ @xmath32 ( @xcite p. 137 ) , or dynamical map , or stochastic map is a completely positive linear map , which transforms states into states ; usually the definition is given for its adjoint @xmath33 .
the channels are usually introduced to describe noisy quantum evolutions , but we shall see that also quantum measurements can be identified with channels .
[ channdfn ] let @xmath34 and @xmath35 be two @xmath14-algebras .
a linear map @xmath33 from @xmath35 to @xmath34 is said to be a _ channel _ if it is completely positive , unital ( i.e. identity preserving ) and normal ( or , equivalently , weakly@xmath24 continuous ) . due to the equivalence of w@xmath24-continuity and existence of a preadjoint @xmath32 @xcite , definition [ channdfn ] is equivalent to : @xmath32 is a completely positive linear map from the predual @xmath36 to the predual @xmath37 , normalized in the sense that @xmath38 , { \mathds1}_2 \rangle_2= \langle \rho,{\mathds1}_1\rangle_1 $ ] , @xmath39 .
let us note also that @xmath32 maps normal states on @xmath34 into normal states on @xmath35 .
note that the composition of channels gives again a channel .
if we have three channels @xmath40 , @xmath41 , @xmath42 and such that @xmath43 , following @xcite we say that @xmath44 is a _ coarse graining _ of @xmath45 or that @xmath45 is a _ refinement _ of @xmath44 .
the general definition of the relative entropy @xmath46 for two states @xmath28 and @xmath47 is given in @xcite ; here we give only some particular cases of the general definition .
given a separable hilbert space @xmath6 and two states @xmath48 the quantum relative entropy of @xmath49 with respect to @xmath50 is defined by @xmath51 given two normal states @xmath52 on @xmath53 , i.e. two probability measures such that @xmath54 , the classical relative entropy of @xmath55 with respect to @xmath56 , or kullback - leibler divergence , is @xmath57 given two normal states @xmath58 on @xmath23 , the relative entropy of @xmath59 with respect to @xmath60 is @xmath61 let us define the two probabilities @xmath62 and the two measurable families of density operators @xmath63 ( these definitions hold where the denominators do not vanish and are completed arbitrarily where the denominators vanish ) .
then , eq.([cqs1 ] ) gives immediately @xmath64 finally , let us denote by @xmath65 _ the von neumann entropy _
, i.e. @xmath66 all the relative entropies and entropies take values in @xmath67 $ ] .
note that we have used a subscript `` c '' for classical quantities , a subscript `` q '' for purely quantum ones and no subscript for general quantities , eventually of a mixed character .
a key result which follows from the convexity properties of the relative entropy is _
uhlmann s monotonicity theorem _
( @xcite , theor . 1.5
p. 21 ) , which implies that channels decrease the relative entropy .
[ uhltheo ] if @xmath28 and @xmath47 are two normal states on @xmath34 and @xmath33 is a channel from @xmath68 , then @xmath69| \lambda[\pi])$ ] .
[ restr ] note also that the operation of restricting the states to some subalgebra is a channel ; so , if @xmath70 and @xmath71 are two normal states on @xmath72 and @xmath73 and @xmath74 are their restrictions to @xmath75 , then @xmath76 , @xmath77 .
the classical notion of mutual entropy can be immediately generalized to states on von neumann algebras .
let @xmath70 be a normal state on @xmath72 and let us denote by @xmath78 and @xmath79 its _ marginals _
, i.e. its restrictions to @xmath34 and @xmath35 , respectively .
the _ mutual entropy _ of @xmath80 is by definition the relative entropy @xmath81 of the state with respect to the tensor product of its marginals .
we shall use the following results on mutual entropies .
let @xmath82 be a normal state on @xmath83 and denote all its possible marginals by @xmath84 ( @xmath85 with @xmath86 and @xmath87 ) , @xmath88 ( @xmath89 ) . from corollary 5.20 of ref .
@xcite we obtain the chain rules @xmath90 and from remark [ restr ] we obtain @xmath91 and the similar inequalities given by permutation of the indices .
the notion of instrument is central in quantum measurement theory ; an instrument gives the probabilities and the state changes @xcite .
[ instrdfn ] let @xmath92 be two separable complex hilbert spaces and @xmath93 be a measurable space .
an _ instrument _
@xmath94 is a map valued measure such that ( iii ) @xmath95 , @xmath96 is completely positive , @xmath97 , ( normalization ) @xmath98\right\}={\operatorname{tr}}_{{\mathcal{h}}_1}\left\ { \rho\right\}$ ] , @xmath99 , ( @xmath21-additivity ) for every countable family @xmath100 of pairwise disjoint sets in @xmath101 @xmath102,\ , a \right\rangle_2= \big\langle { \mathcal{i}}\big(\bigcup_i f_i\big)[\rho],\ , a \big\rangle_2\ , , \qquad\forall \rho\in { \mathcal{t}}({\mathcal{h}}_1 ) , \quad\forall a\in { \mathcal{l}}({\mathcal{h}}_2).\ ] ] unlike the usual definitions of instrument we have introduced two hilbert spaces , an initial one @xmath103 and a final one @xmath104 ; we allow the hilbert space where the quantum system lives to be changed by the measurement , which is the standard set up when quantum channels are considered @xcite and which is usefull when we shall construct something similar to the compound state of ohya @xcite .
the map @xmath105 $ ] turns out to be a positive operator valued ( pov ) measure on @xmath103 ( the observable associated with the instrument @xmath94 ) . for every @xmath106
the map @xmath107 , with @xmath108\right\rangle_1 \equiv { \operatorname{tr}}_{{\mathcal{h}}_2 } \ { { \mathcal{i}}(f)[\rho]\},\ ] ] is a probability measure : @xmath109 is the probability that the result of the measurement be in @xmath110 when the pre - measurement state is @xmath15 .
moreover , given the result @xmath110 , the post - measurement state is @xmath111 $ ] . [ abscont ] it is easy to show that all the measures @xmath112 , @xmath113 , are absolutely continuous with respect to @xmath114 , where @xmath115 is any faithful normal state on @xmath116 .
so , we can fix also a @xmath21-finite measure @xmath20 on @xmath117 such that all the probabilities measures @xmath112 are absolutely continuous with respect to @xmath20 .
moreover we complete @xmath118 and extend the instrument to the extended @xmath21-algebra in the same way as ordinary measures are extended ( @xcite problem 3.10 , p. 49 ) : for any set @xmath119 in the extended @xmath21-algebra , there exist @xmath120 such that @xmath121 ( @xmath122 is the symmetric difference ) with @xmath123 and we define @xmath124 . for the extended objects we use the same symbols as for the original ones .
it is always possible to take for @xmath20 a probability measure , but it is convenient to leave more freedom ; for instance , in the case of a discrete @xmath25 one takes for @xmath20 the counting measure or in the case of a measurement of position and/or momentum one takes for @xmath20 the lebesque measure . from now on @xmath125 are two separable complex hilbert spaces , @xmath19 is a complete @xmath21-finite measure space , @xmath94 is an instrument as in definition [ instrdfn ] and the associated probabilities ( [ probi ] ) are such that @xmath126 then , we introduce the @xmath14-algebras @xmath127 [ theochann ] let us set @xmath128\rangle_1 : = \int_\omega f(\omega ) \langle { \mathcal{i}}({\mathrm{d}}\omega)[\rho],a\rangle_2\ , , \quad \forall \rho \in { \mathcal{t}}({\mathcal{h}}_1 ) , \
\forall a \in { \mathcal{m}}_2\ , , \ \forall
f\in { \mathcal{m}}_3\,;\ ] ] by linearity and continuity the map @xmath129 can be extended to a channel @xmath130 viceversa , the instrument @xmath94 is uniquely determined by the channel .
let us note that by approximating @xmath131 with simple functions we get from ( [ itochann ] ) @xmath132\rangle_1 \leq { \left\vert\rho\right\vert}_{{\mathcal{t}}({\mathcal{h}}_1 ) } { \left\verta\right\vert}_{{\mathcal{l}}({\mathcal{h}}_2 ) } { \left\vertf\right\vert}_{l^\infty}$ ] ; then , the direct statement follows by standard arguments .
viceversa , given a channel @xmath129 , an instrument @xmath94 is defined by : @xmath97 @xmath133,a\rangle_2:=\langle \rho , \
, \lambda_{\mathcal{i}}^ * [ a\otimes 1_f]\rangle_1\ , , \qquad \forall \rho \in { \mathcal{t}}({\mathcal{h}}_1 ) , \quad \forall a \in { \mathcal{m}}_2\,.\ ] ] the @xmath21-additivity follows from the weak@xmath24 continuity of the channel ; all the other properties are more or less evident .
now , let us consider the preadjoint of the channel we have constructed @xmath134 the quantity @xmath135 $ ] is an equivalence class of bochner integrable @xmath136-valued functions of @xmath137 ; let @xmath138(\omega)$ ] be a representative .
if @xmath139 , then @xmath135(\omega)\geq 0 $ ] , @xmath20-a.s . , and in this case we take the representative to be positive everywhere ; we asked the completeness of @xmath20 just to have the freedom of making modifications inside null sets without having to take care of measurability .
moreover , if @xmath15 is normalized , also @xmath135 $ ] is normalized .
so , we have @xmath140 @xmath141(\omega)\geq 0\ , , \quad \forall \omega\in \omega\ , , \qquad \int_\omega { \operatorname{tr}}_{{\mathcal{h}}_2 } \left\ { \lambda_{\mathcal{i}}[\rho](\omega)\right\ } q({\mathrm{d}}\omega)=1\ , , \\ \label{rnder } \frac{p_\rho({\mathrm{d}}\omega ) } { q({\mathrm{d}}\omega ) } = { \operatorname{tr}}_{{\mathcal{h}}_2 } \left\ { \lambda_{\mathcal{i}}[\rho](\omega)\right\ } \qquad \text{(radon - nikodim derivative ) } , \\
\int_f\lambda_{\mathcal{i}}[\rho](\omega ) q({\mathrm{d}}\omega)={\mathcal{i}}(f)[\rho]\ , , \quad \forall f\in { \mathcal{f}}\ , , \qquad \text{(bochner integral)}.\end{gathered}\ ] ] let us normalize the positive trace - class operators @xmath135(\omega)$ ] by setting @xmath142(\omega ) \right\ } \right)^{-1 } \lambda_{\mathcal{i}}[\rho](\omega ) & \text{if } { \operatorname{tr}}_{{\mathcal{h}}_2}\left\{\lambda_{\mathcal{i}}[\rho](\omega ) \right\}>0 \\
\tilde \rho \quad \big(\tilde \rho\in { \mathcal{s}}({\mathcal{h}}_2)\text { , fixed } \big ) & \text{if } { \operatorname{tr}}_{{\mathcal{h}}_2}\left\{\lambda_{\mathcal{i}}[\rho](\omega ) \right\}=0 \end{cases}\ ] ] by eqs .
( [ rnder])([apstates ] ) we have @xmath143\ , , \quad \forall f\in { \mathcal{f}}\ , , \qquad \text{(bochner integral)}.\ ] ] this construction gives directly the result by ozawa on the _ existence _ of a family _ of a posteriori states _
@xcite , with the small generalization of the use of two hilbert spaces .
[ propapost ] let @xmath144 be as above
. for any @xmath106 there exists a @xmath112-a.s .
unique family of _ a posteriori states _
@xmath145 for @xmath146 , which means that the function @xmath147 is measurable and that eq .
( [ intapstates ] ) holds .
theorem [ theochann ] and proposition [ propapost ] generalize immediately to the case of @xmath148 substituted by von neumann algebras with separable predual ; the separability is needed in the results quoted in subsection [ qcalgebra ] and taken from @xcite and which are at the bases of the whole construction .
in quantum statistics , the following problem of identification of states is a natural one .
there is a parametric family of quantum states @xmath149 ( the subscript `` i '' stays for `` initial '' ) , where @xmath150 belongs to some parameter space @xmath119 and it is distributed with some a priori probability @xmath151 .
the experimenter has to make inferences on @xmath150 by using the result of some measurement on the quantum system . in quantum communication theory ,
the problem of the transmission of a message through a quantum channel is similar .
a message is transmitted by encoding the letters in some quantum states , which are possibly corrupted by a quantum noisy channel ; at the end of the channel the receiver attempts to decode the message by performing measurements on the quantum system .
so , one has an alphabet @xmath119 and the letters @xmath152 are transmitted with some a priori probabilities @xmath151 .
each letter @xmath150 is encoded in a quantum state and we denote by @xmath149 the state associated to the letter @xmath150 as it arrives to the receiver , after the passage through the transmission channel .
let us give the formalization of both problems ; we use the language of the quantum communication set up .
first of all , we have a @xmath21-finite measure space @xmath153 ; @xmath119 is the alphabet and the a priori probabilities for the letters are given by @xmath154 , where @xmath155 is a suitable probability density with respect to @xmath156 .
the _ letter states _ are @xmath157 with @xmath158 measurable and the mixture @xmath159 can be called the _ initial a priori state_. one calls @xmath160 the _ initial ensemble_. it would be possible to take @xmath151 as @xmath156 ; then , @xmath161 . however , it is convenient to distinguish @xmath151 and @xmath156 , mainly for the cases when one has more initial ensembles .
note that @xmath162 is nothing but a random variable in the probability space @xmath163 with value in @xmath164 .
let the decoding measurement be represented by the instrument @xmath94 of the previous section with the associated pov measure @xmath165 . by using the notations of section [ instrsec ] and , in particular , the radon - nikodim derivative ( [ rnder ] )
, we can construct the following probabilities , conditional probabilities and densities : @xmath97 , @xmath166 @xmath167(\omega)\ } , \\
p_{{\mathrm{f}}}(f):= \int_a p_{{\mathrm{f}}|{\mathrm{i}}}(f|\alpha)\ , p_{\mathrm{i}}({\mathrm{d}}\alpha)= p_{\eta_{\mathrm{i}}}(f ) , \qquad q_{{\mathrm{f}}}(\omega):= \frac{p_{{\mathrm{f}}}({\mathrm{d}}\omega)}{q({\mathrm{d}}\omega)}={\operatorname{tr}}_{{\mathcal{h}}_2}\{\lambda_{\mathcal{i}}[\eta_{\mathrm{i}}](\omega)\ } , \\ p_{{\mathrm{i}}{\mathrm{f}}}({\mathrm{d}}\alpha\times { \mathrm{d}}\omega):= p_{{\mathrm{f}}|{\mathrm{i}}}({\mathrm{d}}\omega|\alpha)\,p_{\mathrm{i}}({\mathrm{d}}\alpha ) , \qquad q_{{\mathrm{i}}{\mathrm{f}}}(\alpha,\omega):= \frac{p_{{\mathrm{i}}{\mathrm{f}}}({\mathrm{d}}\alpha \times { \mathrm{d}}\omega)}{\nu({\mathrm{d}}\alpha)\,q({\mathrm{d}}\omega)}=q_{{\mathrm{f}}|{\mathrm{i}}}(\omega|\alpha)q_{\mathrm{i}}(\alpha ) , \\ \label{probz } p_{{\mathrm{i}}|{\mathrm{f}}}(b|\omega):=\frac{p_{{\mathrm{i}}{\mathrm{f}}}(b\times{\mathrm{d}}\omega)}{p_{\mathrm{f}}({\mathrm{d}}\omega)}\ , , \qquad \qquad q_{{\mathrm{i}}|{\mathrm{f}}}(\alpha|\omega):= \frac { p_{{\mathrm{i}}|{\mathrm{f}}}({\mathrm{d}}\alpha|\omega)}{\nu({\mathrm{d}}\alpha ) } = \frac{q_{{\mathrm{i}}{\mathrm{f}}}(\alpha,\omega)}{q_{{\mathrm{f}}}(\omega)};\end{gathered}\ ] ] the subscript `` f '' stays for `` final '' .
if we apply the measurement , but we do not do any selection on the system , we obtain the _ post - measurement a priori states _
@xmath168,\qquad \eta_{\mathrm{f}}:= { \mathcal{i}}(\omega)[\eta_{\mathrm{i}}]=\int_a p_{\mathrm{i}}({\mathrm{d}}\alpha)\,\eta_{{\mathrm{f}}}^{\alpha}.\ ] ] by applying the definition ( [ apstates ] ) we can introduce two families of a posteriori states : @xmath169 by using eqs .
( [ intapstates ] ) for @xmath170 , ( [ iapriori])([2apost ] ) , one obtains @xmath171 here and in the following integrals on states are in the bochner sense .
let us stress that the states @xmath149 , @xmath172 are uniquely defined @xmath151-almost surely , @xmath173 @xmath174-a.s . and @xmath175 @xmath176-a.s .
with respect to the algebras given in ( [ alg123 ] ) we have one more von neumann algebra , @xmath177 ; then , we set @xmath178 in particular , we have the identification @xmath179 the states are represented by densities with respect to @xmath180 , @xmath181 , @xmath182 , @xmath183 it is easy to see that the initial ensemble @xmath160 can be seen as a normal state on @xmath184 .
by using a superscript which indicates the algebras on which a state is acting , we can write @xmath185 for the initial state and its marginals . we already constructed the channel @xmath186 ; by dilating it with the identity we obtain the _ measurement channel _ @xmath187 by applying the measurement channel to the initial state we obtain the final state @xmath188= \{q_{\mathrm{i}}(\alpha ) \lambda_{\mathcal{i}}[\rho_{\mathrm{i}}(\alpha)](\omega)\ } = \{q_{{\mathrm{i}}{\mathrm{f}}}(\alpha,\omega ) \rho_{{\mathrm{f}}}^{\alpha}(\omega)\},\ ] ] whose marginals are @xmath189 let us note that @xmath190=\sigma^{0}_{\mathrm{f}}\otimes \sigma^{23}_{\mathrm{f}}\,.\ ] ] holevo s bound ( [ holevos_bound ] ) involves a mean quantum relative entropy , which is often called _
holevo s chi - quantity _ , given by @xmath192 in general , given a probability space @xmath193 and a measurable family @xmath194 of statistical operators on some hilbert space @xmath6 , the @xmath191-quantity of the ensemble @xmath195 is defined by @xmath196 in this definition the set @xmath197 could be @xmath198 itself , see @xcite pp .
by using the definition ( [ relqentropy ] ) of the quantum relative entropy and the definition of von neumann entropy , when @xmath199 , one has @xmath200 the expressions of the mutual entropies we shall need will contain the @xmath191-quantities @xmath201 , @xmath202 , @xmath203 , @xmath204 and the mean @xmath191-quantities @xmath205 @xmath206 the mixtures appearing in these @xmath191-quantities are given by eqs .
( [ iapriori ] ) , ( [ post - m ] ) , ( [ somemixtures ] ) . by using the definitions above and property ( [ cqs2 ] ) , it is easy to compute all the mutual entropies involving the initial and the final state
. first of all we get that holevo s @xmath191-quantity is the initial mutual entropy @xmath207 and that the mutual entropy involving only the classical part of the final state is the shannon input / output classical mutual entropy , i.e. the classical information on the input extracted by the measurement : @xmath208 then , the remaining mutual entropies turn out to be @xmath209 @xmath210 by the chain rules ( [ chain ] ) we get @xmath211 which gives the expression of the `` tripartite '' mutual entropy @xmath212 and the identities @xmath213 uhlmann s monotonicity theorem ( see theorem [ uhltheo ] ) and eqs .
( [ defsigmaf ] ) , ( [ tensf ] ) give us the inequality @xmath214|\lambda[\sigma_{\mathrm{i}}^0\otimes \sigma_{\mathrm{i}}^1])= s(\sigma_{\mathrm{f}}^{023}|\sigma_{\mathrm{f}}^0\otimes \sigma_{\mathrm{f}}^{23});\ ] ] by eqs .
( [ initmentr ] ) , ( [ 3mentr ] ) this inequality becomes @xmath215 in @xcite this inequality was found in the discrete case ; in @xcite it was derived , again in the discrete case , by using relative entropies as here and the general case was announced .
roughly , eq . ( [ sww ] ) says that the quantum information contained in the initial ensemble @xmath216 is greater than the classical information extracted in the measurement plus the mean quantum information left in the a posteriori states .
inequality ( [ sww ] ) can be seen also as giving some kind of information - disturbance trade - off , a subject to which the paper @xcite , which contains a somewhat related inequality , is devoted .
holevo s bound @xcite , generalized to the continuous case in @xcite , is @xmath217 or , in terms of mutual entropies , @xmath218 the derivation of holevo s bound given in @xcite is based on a measurement channel involving only the pov measure , not the whole instrument ; the fact that inequality ( [ sww ] ) is stronger than holevo s bound ( [ holevos_bound ] ) is a consequence of the fact that our channel @xmath219 is a refinement of the channel used in @xcite ( see the discussion given in @xcite ) . by using one of the identities ( [ idts ] ) ,
the inequality ( [ sww ] ) can be rewritten in an equivalent form , which is slightly more symmetric : @xmath220 by restriction of the states ( see remark [ restr ] ) we get the inequality @xmath221 by eqs .
( [ 2mentr ] ) and ( [ 3mentr ] ) we get @xcite @xmath222 which says that the classical information extracted in the measurement plus the mean quantum information left in the a posteriori states is greater than the quantum information left in the post - measurement a priori states . all the other inequalities which can be obtained from the final state are also consequences of inequality ( [ lb ] ) and identities ( [ idts ] ) .
given an instrument @xmath223 and a statistical operator @xmath224 , an interesting quantity , which can be called the _
quantum information gain _ , is @xmath225 this is nothing but the quantum entropy of the pre - measurement state minus the mean entropy of the a posteriori states .
it is a measure of the gain in purity ( or loss , if negative ) in passing from the pre - measurement state to the post - measurement a posteriori states and it gives no information on the ability of the measurement in identifying the pre - measurement state , ability which is contained in @xmath0 . by using the expression of a @xmath191-quantity in terms of entropies and
mean entropies , as in ( [ altchi ] ) , one can see that , when @xmath226 inequality ( [ sww ] ) is equivalent to @xmath227 here the state @xmath228 is given and @xmath160 has to be thought as any demixture of @xmath228 .
an interesting question is when the quantum information gain is positive .
groenewold has conjectured @xcite and lindblad @xcite has proved that the quantum information gain is non negative for an instrument of the von neumann - lders type . the general case has been settled down by ozawa , who in @xcite has proved the following theorem in the case @xmath229 .
a shorter proof with respect to ozawa s one is based on inequality ( [ iqineq ] ) @xcite .
[ theooza ] let @xmath92 be two separable complex hilbert spaces , @xmath93 be a measurable space and @xmath94 a completely positive instrument as in definition [ instrdfn ] .
then , the instrument @xmath223 sends any pure input state into almost surely pure a posteriori states if and only if @xmath230 , for all statistical operators @xmath224 for which @xmath231 .
\(b ) @xmath232 ( a ) is trivial : put a pure state @xmath228 into the definition and you get @xmath233 is pure @xmath174-a.s . to see ( a ) @xmath232 ( b ) , we take a demixture of @xmath228 into pure states ; then , by ( a ) also the states @xmath175 are pure and @xmath234 ; then , eq .
( [ iqineq ] ) gives @xmath235 .
in @xcite ohya introduced a notion of compound states which involves the input and output states of a quantum channel .
taking inspiration from this idea , we are able to produce some inequalities which strengthen a lower bound on @xmath236 given by scutaru in @xcite . first of all we need some new families of statistical operators and the relationships among them : @xmath237 the state ( [ scustate ] ) has been introduced by scutaru @xcite and the state ( [ ohyastate ] ) is similar to the compound state introduced by ohya @xcite for quantum channels .
now , let us construct a first compound state on @xmath238 and let us give some of its marginals : @xmath239 for this state we have @xmath240 and remark [ restr ] gives the inequalities @xmath241 which give @xmath242 @xmath243 is scutaru s bound .
let us give also a second compound state and some of its marginals : @xmath244 as before we get the inequalities @xmath245 @xmath246 it is possible to obtain these inequalities also by constructing suitable channels and by using the idea of the refinement of a channel @xcite .
in @xcite hall exhibits a transformation on the initial ensemble and on the pov measure which leaves invariant @xmath0 but not the initial @xmath191-quantity and in this way produces a new upper bound on the classical information . inspired by hall s transformation , a new instrument can be constructed in such a way that the analogous of inequality ( [ sww ] ) produces an upper bound on @xmath0 stronger than both hall s and holevo s ones .
let us set @xmath249 : = m(\alpha ) \tau m(\alpha)^*\ , , \quad \forall \tau \in { \mathcal{t}}({\mathcal{h}}_1);\ ] ] by eq .
( [ iapriori ] ) the operators @xmath250 satisfy the normalization condition @xmath251 then , the position @xmath252 defines an instrument from @xmath253 into @xmath253 with value space @xmath254 .
the instrument @xmath248 has been constructed by using only the old initial ensemble @xmath255 .
the associated pov measure is @xmath256 now , we can construct the associated channel and a posteriori states , as in section [ instrsec ] . by looking at eq .
( [ itochann ] ) one has immediately @xmath257(\alpha)={\mathcal{g}}(\alpha)[\tau]=m(\alpha ) \tau m(\alpha)^ * , \qquad \forall \tau\in { \mathcal{t}}({\mathcal{h}}_1)\ ] ] and by looking at eq .
( [ apstates ] ) one has that , for @xmath258 , @xmath259 is a family of a posteriori states for @xmath260 . let us stress that @xmath248 sends pure states into a.s .
pure a posteriori states ; therefore , by theorem [ theooza ] one has @xmath261 let @xmath262 be a c.o.n.s . of eigenvectors of @xmath228 , so that we can write @xmath263 , with @xmath264 and @xmath265 . as in remark [ abscont ] one can show that the complex measures @xmath266 are absolutely continuous with respect to @xmath267 ; therefore the radon - nikodim derivatives @xmath268 exist and the position @xmath269 defines a family of statistical operators ; in an abbreviated way we write @xmath270 now we consider @xmath271 as initial ensemble for @xmath248 ; note that one gets @xmath272 let us consider now holevo s bound for the new set up : @xmath273 the pov measure @xmath274 and the states @xmath275 have been constructed just in order to have @xmath276 as it is easy to verify ; this implies immediately @xmath277 therefore , we have @xmath278 which is the `` continuous '' version of hall s bound @xmath279eq . ( 19 ) of @xcite@xmath280 .
this bound , in the discrete case , is discussed also in refs .
@xcite .
having defined a new instrument and not only a pov measure , we obtain from ( [ sww ] ) the inequality @xmath281 which gives a stronger bound than hall s one ( [ halldualbound ] ) . in order to render more explicit this bound , it is convenient to start from the equivalent form ( [ iqineq ] ) , which now reads @xmath282 by eqs . ( [ malpha ] ) and ( [ apostalpha ] )
we obtain @xmath283 ; together with eqs .
( [ newiq ] ) , ( [ newe ] ) , ( [ altchi ] ) , this gives @xmath284 therefore , eq .
( [ nnn ] ) gives the new bound @xmath285 let us stress that @xmath286 because of eq .
( [ newiq ] ) . more explicitly , by eqs .
( [ newe ] ) , ( [ newinistat ] ) , ( [ newiq ] ) , we have @xmath287 where @xmath275 is given by ( [ newinistat ] ) and , by eqs . ( [ malpha ] ) , ( [ apostalpha ] ) , ( [ newinistat ] ) , @xmath288 this last quantity is defined similarly to ( [ newinistatrig ] ) , by starting from the diagonalization of @xmath149 .
let us stress that the upper bound in ( [ newbound ] ) involves the initial ensemble @xmath160 and the pov measure @xmath165 , not the full instrument @xmath94 , while the bound ( [ sww ] ) involves @xmath160 , @xmath165 and also the a posteriori states of @xmath94 .
both bounds ( [ sww ] ) and ( [ newbound ] ) are stronger than holevo s bound ( [ holevos_bound ] ) .
99 , _ entropy and information gain in quantum continual measurements _ , in : quantum communication , computing , and measurement 3 , p. tombesi and o. hirota ( eds . ) , kluwer , new york , 2001 , 4957 ; arxiv : quant - ph/0012115 . , _ instrumental processes , entropies , information in quantum continual measurements _ , in : quantum information , statistics , probability , o. hirota ( ed . ) , rinton , princeton , 2004 , 3043 ; arxiv : quant - ph/0401114 . ,
_ instruments and channels in quantum information theory _ , arxiv : quant - ph/0409019 v1 3 sep 2004 .
, _ probability and measure _ , wiley , new york , 1995 .
, _ on the heisenberg principle , namely on the information - disturbance trade - off in a quantum measurement _ ,
fortschr .
* 51 * ( 2003 ) , 318330 ; * doi * 10.1002/prop.200310045 . , _ quantum theory of open systems _ , academic press , london , 1976 . , _ an operational approach to quantum probability _ , commun .
* 17 * ( 1970 ) , 239260 . ,
_ les algbres doprateurs dans lespace hilbertien _
, gauthier - villars , paris , 1957 . , _ a problem of information gain by quantal measurements _ , int.j .
phys . * 4 * ( 1971 ) , 327338 . ,
_ quantum information and correlation bounds _
a * 55 * ( 1997 ) , 100113 . ,
_ techniques for bounding quantum correlations _ , in : quantum communication ,
computing , and measurement , o. hirota , a. s. holevo and c. m. caves ( eds . ) , plenum , new york , 1997 , 5361 .
, _ some estimates for the amount of information transmittable by a quantum communication channel _ , probl .
* 9 * , no . 3 ( 1973 ) , 177183 ( engl .
transl . : 1975 ) . , _ continuous ensembles and the @xmath191-capacity of infinite - dimensional channels _ , arxiv : quant - ph/0408176 v1 30 aug 2004 .
, _ capacity of quantum channels using product measurements _ , j. math . phys .
* 42 * ( 2001 ) , 8798 ; arxiv : quant - ph/0004062 . , _ an entropy inequality for quantum measurements _ , commun .
math.phys .
* 28 * ( 1972 ) , 245249 . , _ efficient measurements , purification , and bounds on the mutual information _ ,
rev . a * 68 * art .
054302 ( 2003 ) ; arxiv : quant - ph/0306039 . ,
_ on compound state and mutual information in quantum information theory _ , ieee trans .
theory * it-29 * ( 1983 ) , 770774 . ,
_ quantum entropy and its use _ , springer , berlin , 1993 . , _ quantum measuring processes of continuous observables _
, j. math.phys .
* 25 * ( 1984 ) , 7987 .
, _ conditional probability and a posteriori states in quantum mechanics _ , publ .
r.i.m.s .
kyoto univ .
* 21 * ( 1985 ) , 279295 .
, _ concepts of conditional expectations in quantum theory _ , j. math.phys .
* 25 * ( 1985 ) , 19481955 .
, _ on information gain by quantum measurements of continuous observables
_ , j. math .
* 27 * ( 1986 ) , 759763 .
, _ inequalities for quantum entropy : a review for conditions for equality _ , j. math .
* 43 * ( 2002 ) , 43584375 ; arxiv : quant - ph/0205064 . , _ @xmath289-algebras and @xmath14-algebras
_ , springer , berlin , 1971 . , _ limitation on the amount of accessible information in a quantum channel _ , phys .
* 76 * ( 1996 ) , 34523455 .
, _ lower bound for mutual information of a quantum channel _ , phys.rev .
* 75 * ( 1995 ) , 773776 .
, _ quantum information theory , the entropy bound , and mathematical rigor in physics _ , in : quantum communication , computing , and measurement , o. hirota , a. s.holevo and c. m. caves ( eds . ) , plenum , new york , 1997 , 1723 . ,
_ ultimate information carrying limit of quantum systems _ , phys .
* 70 * ( 1993 ) , 363366 . | general quantum measurements are represented by instruments . in this paper the mathematical formalization is given of the idea that an instrument is a channel which accepts a quantum state as input and produces a probability and an a posteriori state as output .
then , by using mutual entropies on von neumann algebras and the identification of instruments and channels , many old and new informational inequalities are obtained in a unified manner .
such inequalities involve various quantities which characterize the performances of the instrument under study ; in particular , these inequalities include and generalize the famous holevo s bound . |
a supersymmetry ( susy ) between fermions and bosons provides a natural explanation of the higgs electroweak symmetry breaking ( ewsb ) mechanism in the framework of a grand unified theory ( gut ) .
the evolution of renormalization group equations ( rges ) @xcite with the particle content of the minimal supersymmetric standard model ( mssm ) @xcite is consistent with a grand unified scale at @xmath15 gev and an effective susy mass scale in the range @xmath16 tev . with a large top quark yukawa coupling to a higgs boson at the gut scale , needed for @xmath17 yukawa unification @xcite
, radiative corrections can drive the corresponding higgs boson mass squared parameter negative , spontaneously breaking the electroweak symmetry and naturally explaining the origin of the electroweak scale . in the minimal supersymmetric gut with a large top quark yukawa coupling ( @xmath18 ) , there is an infrared fixed point ( irfp ) @xcite at low @xmath19 , and the top quark mass is correspondingly predicted to be @xmath20 @xcite . at high @xmath19 ,
another quasi - irfp solution ( @xmath21 ) exists at @xmath22 @xcite . for @xmath23 gev ,
the two irfps of top quark yukawa coupling appear at @xmath2 and @xmath3 @xcite .
supersymmetry extends the poincar symmetry to new dimensions of space and time and local gauge invariance with susy includes gravity . in supergravity ( sugra )
models @xcite , it is assumed that supersymmetry is broken in a hidden sector with susy breaking communicated to the observable sector through gravitational interactions , leading naturally but not necessarily @xcite to a common scalar mass ( @xmath24 ) , a common gaugino mass ( @xmath25 ) , a common trilinear coupling ( @xmath26 ) and a common bilinear coupling ( @xmath27 ) at the gut scale . through minimization of the higgs potential , the @xmath28 parameter and magnitude of the superpotential higgs mixing parameter @xmath29
are related to the ratio of vacuum expectation values ( vevs ) of higgs fields ( @xmath1 ) , and the mass of the @xmath9 boson ( @xmath30 ) .
the susy particle masses and couplings at the weak scale can be predicted by the evolution of rges from the unification scale @xcite . in our analysis
, we evaluate susy mass spectra and couplings in the minimal supergravity model with four parameters : @xmath24 , @xmath25 , @xmath26 and @xmath19 ; and the sign of the higgs mixing parameter @xmath29 .
the neutralino relic density is only slightly affected by a change in the @xmath26 parameter since @xmath26 mainly affects the masses of third generation sfermions .
therefore , for simplicity we take @xmath31 in most of our calculations . the mass matrix of the neutralinos in the weak eigenstates ( @xmath32 , @xmath33 , @xmath34 , @xmath35 ) has the following form @xcite @xmath36 this matrix is symmetric and can be diagonalized by a single matrix to obtain the mass eigenvalues and eigenvectors . the form of eq .
( [ eq : xino ] ) establishes our sign convention for @xmath29 .
recent measurements of the @xmath37 decay rate by the cleo @xcite and lep collaborations @xcite place constraints on the parameter space of the minimal supergravity model @xcite .
it was found that @xmath37 excludes most of the minimal supergravity ( msugra ) parameter space when @xmath19 is large and @xmath38 @xcite . in this article , we present results for both @xmath38 and @xmath39 .
we calculate the masses and couplings in the higgs sector with one loop corrections from both the top and the bottom yukawa interactions in the rge - improved one - loop effective potential @xcite at the scale @xmath40 . with this scale choice
, the numerical value of the cp - odd higgs boson mass ( @xmath11 ) at large @xmath19 @xcite is relatively insensitive to the exact scale and the loop corrections to @xmath41 are small compared to the tree level contribution .
in addition , when this high scale is used , the rge improved one - loop corrections approximately reproduce the dominant two loop perturvative calculation of the mass of the lighter cp - even higgs scalar ( @xmath42 ) .
our numerical values of @xmath42 are very close to the results of ref .
@xcite where somewhat different scales higher than @xmath30 have been adopted in evaluating the effective potential .
the matter density of the universe @xmath43 is described in terms of a relative density @xmath44 with @xmath45 the critical density to close the universe . here , @xmath46 is the hubble constant , @xmath47 , and @xmath48 is newton s gravitational constant .
there is compelling evidence from astronomical observations for the existence of dark matter @xcite .
studies on clusters of galaxies and large scale structure suggest that the matter density in the universe should be at least @xmath49 of the critical density ( @xmath50 ) @xcite , but the big - bang nucleosynthesis and the measured primordial abundance of helium , deuterium and lithium constrain the baryonic density to @xmath51 @xcite .
the anisotropy in the cosmic microwave background radiation measured by the cosmic background explorer ( cobe ) suggests that at least @xmath52 of the dark matter should be cold ( nonrelativistic ) @xcite . in supersymmetric theories with a conserved @xmath53-parity for the sm particles and higgs bosons and @xmath54 for their superpartners .
] , the lightest supersymmetric particle ( lsp ) can not decay into normal particles and the lsp is an attractive candidate for cosmological dark matter @xcite . in most of the supergravity parameter space , the lightest neutralino ( @xmath7 )
is the lsp @xcite .
inflationary models usually require @xmath55 for the universe @xcite , although models with @xmath56 have recently been proposed @xcite .
recent measurements on the hubble constant are converging to @xmath57 @xcite .
therefore , we conservatively consider the cosmologically interesting region for @xmath58 to be @xmath59 following the classic papers of zeldovich , chiu , and lee and weinberg @xcite , many studies of the neutralino relic density in supergravity models have been made with universal @xcite and nonuniversal @xcite soft breaking masses .
recent calculations have taken into account threshold effects and integration over breit - wigner poles @xcite .
previous analyses have focused on the small to moderate @xmath19 range , @xmath60 , but large @xmath19 is equally of interest .
some theoretical models of fermion masses require a large value @xmath19 @xcite . in this article
, we make more complete calculations of the neutralino relic density over the full range of @xmath19 and sugra parameter space with universal gut scale boundary conditions . we employ the calculational methods of ref .
@xcite , with some significant improvements .
the neutralino annihilation cross section is evaluated with helicity amplitude techniques to eliminate any uncertainty from a neutralino velocity expansion .
the neutralino relic density is calculated with relativistic boltzmann averaging , neutralino annihilation threshold effects and breit - wigner poles .
we introduce a transformation to improve the numerical integration over breit - wigner poles and use appropriate extension of the rges for large @xmath19 solutions .
we determine the constraints on @xmath25 and @xmath19 implied by eq .
( [ eq : ohs ] ) .
we also impose the constraints from the chargino search at lep 2 @xcite .
the effects of the common trilinear coupling ( @xmath26 ) on @xmath61 , @xmath62 and the neutralino relic density @xmath0 are also studied .
the present mass density of the lightest neutralino is inversely proportional to the annihilation cross section of @xmath7 pairs . when kinematically allowed , the neutralino pairs annihilate into fermion pairs ( @xmath63 ) , gauge boson pairs ( @xmath64 and @xmath65 ) , higgs boson pairs ( lighter ) and @xmath6 ( heavier ) , one cp - odd @xmath5 , and a pair of singly charged @xmath66 . ]
( @xmath67 , @xmath68 , @xmath69 , @xmath70 @xmath71 , @xmath72 and @xmath73 ) and associated pairs of gauge and higgs bosons ( @xmath74 , @xmath75 , @xmath76 , and @xmath77 ) through s , t , and u channel diagrams . the most convenient approach to calculate the annihilation cross section including interferences among all diagrams is to calculate the amplitude of each diagram with the helicity amplitude formalism then numerically evaluate the matrix element squared with a computer .
therefore , following ref .
@xcite , we employ the helas package @xcite to calculate the helicity amplitudes .
the annihilation cross section of @xmath7 pairs depends on its mass ( @xmath78 ) which we find can be empirically expressed in gev units as a function of the gut scale gaugino mass @xmath25 and @xmath79 as @xmath80 for 100 gev @xmath81 1000 gev and all @xmath19 for which perturbative rge solutions exist .
these neutralino mass formulas hold to an accuracy of @xmath82 . in the regions of breit - wigner resonance poles
the near - singular amplitudes make the integration unstable . to greatly improve the efficiency and accuracy in calculating the annihilation cross section for processes with a breit - wigner resonance ,
we introduce a transformation @xmath83 such that @xcite @xmath84 and scale @xmath85 to the range ( 0,1 ) .
the time evolution of the number density @xmath86 of weakly interacting mass particles is described by the boltzmann equation @xcite @xmath87 % \ ] ] where @xmath88 is the hubble expansion rate with @xmath89 the effective number of relativistic degrees of freedom , @xmath90 is the planck mass , @xmath91 is the thermally averaged cross section times velocity , @xmath92 is the relative velocity and @xmath93 is the annihilation cross section , and @xmath94 is the number density at thermal equilibrium . in the early universe , when the temperature @xmath95 , the lsp existed abundantly in thermal equilibrium with the lsp annihilations into lighter particles balanced by pair production .
deviation from the thermal equilibrium began when the temperature reached the freeze - out temperature ( @xmath96 ) .
after the temperature dropped well below @xmath97 , the annihilation rate became equal to the expansion rate and @xmath98 .
the resulting relic density can be expressed in terms of the critical density by @xcite @xmath99 where @xmath100 is the number density of @xmath7 .
the thermally averaged cross section times velocity is @xmath101 where @xmath102 ( @xmath103 ) and @xmath104 ( @xmath105 ) are the momentum and energy of the two colliding particles in the cosmic , co - moving frame of reference , and @xmath106 is the temperature .
this expression has been simplified to a one - dimensional integral @xcite of the form @xmath107 where @xmath108 , @xmath109 , @xmath110 is the subprocess energy , and the @xmath111 are modified bessel functions of order @xmath112 . to calculate the neutralino relic density ,
it is necessary to first determine the scaled freeze - out temperature @xmath113 .
the standard procedure is to iteratively solve the freeze out relation @xmath114,\end{aligned}\ ] ] starting with @xmath115 . the relic density at the present temperature ( @xmath116 )
is then obtained from @xmath117 where @xmath118 the ratio @xmath119 is the reheating factor @xcite of the neutralino temperature ( @xmath120 ) compared to the microwave background temperature ( @xmath121 ) .
the integration in eq .
( [ eq : rho ] ) is evaluated with the modified bessel functions expanded as power series in @xmath122 , and then integrated over @xmath122 .
the result is @xmath123 where @xcite @xmath124 \right\ } \nonumber \\ & & + { 2\over 3 } \left ( { 285\over 32}-{45\over 32a}-{15\over 128a^2 } \right)\times \nonumber \\ & & \left\ { e^{-{{a-2}\over x_f}}\left [ x_f^{3\over 2}-2(a-2)\sqrt{x_f}\right ] + 2\sqrt{\pi } ( a-2)^{3\over 2 } \left [ 1-erf\left(\sqrt{{a-2}\over x_f } \right ) \right ] \right\}.\end{aligned}\ ] ] essentially , all of the contribution to the integral in eq . (
[ eq : sigmav ] ) comes from @xmath125 . to generate the susy particle mass spectrum and couplings for the neutralino relic density calculation
, we run the gauge coupling rges from the weak scale up to a high energy scale is large @xcite . ] to fix the unification scale @xmath126 , and the unified gauge coupling @xmath127 .
then we run all the rges from @xmath126 to the weak scale of @xmath30 .
the susy mass scale ( @xmath128 ) is chosen to be 2 tev ; closely similar results are obtained with a 1 tev mass scale . in figure 1 , we present masses , in the case follows that of ref .
@xcite and is opposite to that of ref .
@xcite ; see eq .
( [ eq : xino ] ) . ] @xmath38 , for the lightest neutralino ( @xmath7 ) , the lighter top squark ( @xmath129 ) , the lighter tau slepton ( @xmath130 ) , and two neutral higgs bosons : the lighter cp - even ( @xmath8 ) and the cp - odd ( @xmath5 ) . to a good approximation
, the mass of the lightest chargino ( @xmath131 ) is about twice @xmath78 and the mass of the heavier cp - even higgs boson ( @xmath6 ) is close to @xmath11 @xcite . for @xmath132 gev and @xmath133 gev ,
the mass of @xmath130 can become smaller than @xmath78 so such regions are theoretically excluded .
also shown in fig .
1 are the regions that do not satisfy the following theoretical requirements : electroweak symmetry breaking ( ewsb ) , tachyon free , and the lightest neutralino as the lsp .
the region excluded by the @xmath134 gev limit from the chargino search @xcite at lep 2 is indicated .
masses for @xmath39 are shown in fig .
there are several interesting aspects to note in figs . 1 and 2 : + ( i ) for @xmath135
, @xmath136 generates heavier neutralinos and charginos but lighter @xmath8 than @xmath39 for the same @xmath25 and @xmath24 .
+ ( ii ) an increase in @xmath19 leads to a larger @xmath42 but a reduction in @xmath78 , @xmath137 , @xmath11 and @xmath12 . + ( iii ) for @xmath138 , @xmath139 can become comparable to @xmath78 .
+ ( iv ) increasing @xmath24 raises @xmath11 , @xmath12 and the masses of other scalars significantly .
the relic density of the neutralino dark matter ( @xmath0 ) for the case @xmath38 is presented in fig .
3 versus @xmath25 for several values of @xmath24 and @xmath19 .
figure 4 shows similar contours for @xmath39 .
when @xmath13 is close to the mass of @xmath9 , @xmath8 , @xmath6 , or @xmath5 , the cross section is significantly enhanced by the breit - wigner resonance pole , and a corresponding dip appears in the relic density distribution . for @xmath136 , @xmath140 and @xmath141 gev , there are two dips , the first one occurs at the @xmath8 pole ( @xmath142 gev , @xmath143 gev and @xmath144 gev ) , and the second one appears at the @xmath9 pole ( @xmath145 gev and @xmath146 gev @xmath147 ) .
an increase in @xmath24 raises the mass of higgs bosons but does not affect the mass of @xmath7 .
therefore , comparing the @xmath141 gev results to those of @xmath148 gev , the @xmath9-pole dips occur at about the same @xmath25 as above , but the @xmath149-pole dips appear at larger @xmath25 and @xmath78 . increasing
@xmath19 slightly raises @xmath42 , slightly lowers @xmath78 , and greatly reduces @xmath11 and @xmath12 . for @xmath150 ,
extra dips appear in figs .
3(c ) and 4(c ) .
these resonance dips are associated with the @xmath5 and @xmath6 masses and occur where @xmath151 . for @xmath152 and @xmath153 gev ,
the annihilation cross section is too small for a cosmologically interesting @xmath0 , while for @xmath150 , @xmath24 must be close to 1000 gev for @xmath154 gev and be larger than 400 gev for @xmath155 gev to obtain an interesting relic density . for @xmath156 , @xmath157 , and there are no @xmath5 and @xmath6 resonant contributions ; an acceptable @xmath158 is found for @xmath159 gev and 400 gev @xmath81 800 gev .
to show the constraints of eq .
( [ eq : ohs ] ) on the sugra parameter space , we present contours of @xmath160 and 0.5 in the ( @xmath161 ) plane in fig . 5 for @xmath38 .
also shown are the regions that do not satisfy the theoretical requirements ( electroweak symmetry breaking , tachyon free , and lightest neutralino as the lsp ) and the region excluded by the chargino search ( @xmath162 gev ) at lep 2 @xcite .
figure 6 shows similar contours for @xmath39 .
we summarize below the central features of these results .
+ if @xmath24 is close to 100 gev , * most of the ( @xmath161 ) plane with @xmath10 is excluded by the above mentioned theoretical requirements . *
the chargino search at lep 2 excludes the region where @xmath163 gev for @xmath38 and @xmath164 120 gev for @xmath39 . *
the cosmologically interesting region is 100 gev @xmath81 400 gev and @xmath60 for either sign of @xmath29 .
if @xmath24 is close to 500 gev , * most of the ( @xmath161 ) plane is theoretically acceptable for @xmath165 and @xmath166 gev . * the lep 2 chargino search excludes ( i ) @xmath167 gev for @xmath168 , and ( ii ) @xmath169 gev ( @xmath38 ) or @xmath170 gev ( @xmath39 ) for @xmath135 . *
the cosmologically interesting regions lie in two narrow bands .
contours of @xmath160 and 0.5 in the ( @xmath171 ) plane are presented in fig . 7 for @xmath38 with @xmath140 and @xmath172 .
also shown are the theoretically excluded regions .
figure 8 shows similar contours for @xmath39 .
salient features of these results are summarized in the following .
+ if @xmath19 is close to 1.8 , * most of the ( @xmath171 ) parameter space is theoretically acceptable . *
the chargino search at lep 2 excludes the region where @xmath169 gev for @xmath38 and @xmath164 110 gev for @xmath39 .
* most of the cosmologically interesting region is 80 gev @xmath81 450 gev and @xmath173 gev .
if @xmath19 is close to 50 , * the theoretically acceptable region in the ( @xmath171 ) plane is constrained to have @xmath174 gev and @xmath175 gev for @xmath138 . *
the lep 2 chargino search excludes ( i ) the region with @xmath176 gev for @xmath38 or ( ii ) the region with @xmath177 gev for @xmath39 , which is already inside the theoretically excluded region . *
the cosmologically interesting region lies in a band with ( i ) 475 gev @xmath81 800 gev for @xmath38 , or ( ii ) 500 gev @xmath81 840 gev for @xmath39 , and @xmath178 gev . to correlate the acceptable values of neutralino relic density to the masses of susy particles at the weak scale , we present @xmath0 and representative susy mass spectra versus @xmath25 in fig .
9 , for @xmath38 with ( a ) @xmath140 , @xmath179 gev and ( b ) @xmath180 , @xmath181 gev .
figure 10 shows similar contours for @xmath39 .
also shown are the theoretically excluded regions and the excluded region from the chargino search at lep 2 .
the neutralino relic density and representative susy mass spectrum versus @xmath24 are presented in fig .
11 , for @xmath38 with ( a ) @xmath140 , @xmath182 gev and ( b ) @xmath180 , @xmath183 gev .
figure 12 shows similar contours for @xmath39 .
also shown are the theoretically excluded regions .
the foregoing results were based on the gut scale trilinear coupling choice @xmath31 . to examine the sensitivity of the results to @xmath26 , we present the neutralino relic density versus @xmath26 in fig . 13 for various values of @xmath184 and @xmath19 . for @xmath4
, @xmath0 is almost independent of @xmath26 . for @xmath185 ,
the relic density is reduced by a positive @xmath26 but enhanced by a negative @xmath26 .
significant impact of @xmath26 on @xmath0 occurs only when both @xmath24 and @xmath25 are small and @xmath19 is large . for @xmath186 200 gev and @xmath187
the theoretically allowed range of @xmath26 is @xmath188 .
the value of @xmath0 with @xmath189 590 gev ( @xmath190380 gev ) is about 0.22 ( 2.4 ) times the value with @xmath191 . for @xmath186 800 gev and @xmath187
the value of @xmath0 with @xmath189 1000 gev ( @xmath1901000 gev ) is about 0.58 ( 1.7 ) times the value with @xmath191 .
the dependence of @xmath0 on @xmath26 occurs because the masses of top squarks are sensitive to @xmath61 , and the t - channel diagrams involving the stops make important contribution to neutralino annihilation cross section . in figures 14 ( @xmath38 ) and 15 ( @xmath39 ) ,
we show the mass of the lighter top squark ( @xmath62 ) and the trilinear coupling @xmath61 versus @xmath26 for various values of @xmath184 and @xmath19 .
the trilinear coupling @xmath61 is always negative and it is nearly a linear function of @xmath26 . for @xmath19 close to the top yukawa infrared fixed point , @xmath2 , @xmath61 approaches an infrared fixed point vaule as well , @xmath192 , to about @xmath193 accuracy , independent of the value of @xmath26 @xcite .
the existence of dark matter in the universe provides a potentially important bridge between particle physics and cosmology . within sugra guts ,
the lightest neutralino is the most likely dark matter candidate .
we evaluated the neutralino relic density using helicity amplitude methods including all relevant diagrams .
we applied a transformation to improve the efficiency and accuracy of annihilation cross section calculations in regions with breit - wigner resonance poles . requiring that the neutralino relic density should be in the cosmologically interesting region
, we were able to place tight constraints on the sugra parameter space , especially in the plane of @xmath25 versus @xmath19 , since the mass of the lightest neutralino depends mainly on these two parameters .
we derived empirical formulas relating @xmath78 to @xmath25 and @xmath194 .
the cosmologically interesting regions of the parameter space with @xmath19 close to the top yukawa infrared fixed points are found to be @xmath195 where the high @xmath19 result is based on @xmath191 .
both regions are nearly independent of the sign of @xmath29 .
we presented expectations for the susy particle mass spectra corresponding to these regions of parameter space .
for the irfp at @xmath140 , the lighter scalar tau ( @xmath130 ) and the lightest scalar neutrino ( @xmath196 ) can be relatively light ( @xmath197 gev ) in the cosmologically interesting region . for @xmath4 ,
the neutralino relic density @xmath0 is almost independent of the trilinear couplings @xmath26 . for @xmath185 ,
the relic density is reduced by a positive @xmath26 while enhanced by a negative @xmath26 .
the value of @xmath26 significantly affects @xmath0 only when @xmath19 is large and both @xmath24 and @xmath25 are small .
supernovae @xcite , cosmic microwave background @xcite and other astronomical observations will soon provide precise information on the hubble parameter @xmath149 and the cold dark matter component of @xmath198 .
this information will pinpoint the sugra gut parameters , which in turn predict the susy particle masses in these models .
we are grateful to howie baer and michal brhlik for providing us with their computer code , and to mike berger , gary steigman and dieter zeppenfeld for valuable discussions .
we thank manuel drees for beneficial comments regarding the choice of scale for the one - loop effective potential .
this research was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 95er40896 , and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation .
k. inoue , a. kakuto , h. komatsu and h. takeshita , prog .
. phys . * 68 * , 927 ( 1982 ) and * 71 * , 413 ( 1984 ) .
nilles , phys .
* 110 * , 1 ( 1984 ) ; h. haber and g. kane , phys . rep . * 117 * , 75 ( 1985 ) . for recent reviews ,
see x. tata , in _ the standard model and beyond _ , edited by j.e .
kim , ( world scientific , singapore , 1991 ) ; tasi-95 lectures , university of hawaii report no .
uh-511 - 833 - 95 ( 1995 ) .
chanowitz , j. ellis and m.k .
gaillard , nucl .
phys . * b128 * , 506 ( 1977 ) .
v. barger , m.s .
berger , p. ohmann , phys .
d * 47 * , 1093 ( 1993 ) ; phys .
d * 49 * , 4908 ( 1994 ) ; v. barger , m.s .
berger , p. ohmann and r.j.n .
phillips , phys . lett .
b * 314 * , 351 ( 1993 ) .
b. pendleton and g.g .
ross , phys .
b * 98 * , 291 ( 1981 ) ; c.t .
hill , phys .
d * 24 * , 691 ( 1981 ) .
froggatt , r.g .
moorhouse and i.g .
knowles , phys . lett .
b * 298 * , 356 ( 1993 ) .
j. bagger , s. dimopoulos and e. masso , phys .
* 55 * , 920 ( 1985 ) ; h. arason , _ et al .
_ , phys . rev .
lett . * 67 * 2933 ( 1991 ) ; and phys .
d * 46 * , 3945 ( 1992 ) ; p.
langacker , n. polonsky , phys .
d * 50 * , 2199 ( 1994 ) ; w.a .
bardeen , m. carena , s. pokorski and c.e.m .
wagner , phys .
b * 320 * , 110 ( 1994 ) ; m. carena , m. olechowsk , s. pokorski and c.e.m .
wagner nucl .
b * 419 * , 213 ( 1994 ) ; m. carena and c.e.m .
wagner , nucl .
b * 452 * , 45 ( 1995 ) ; b. schremp , phys . lett .
b * 344 * ( 1995 ) 193 ; b. schrempp and m. wimmer , desy-96 - 109 ( 1996 ) , hep - ph/9606386 , and references therein .
chamseddine , r. arnowitt and p. nath , phys .
lett . * 49 * , 970 ( 1982 ) ; l. ibaez and g. ross , phys . lett . *
b110 * , 215 ( 1982 ) ; l. ibaez , phys . lett . * b118 * , 73 ( 1982 ) ; j. ellis , d. nanopoulos and k. tamvakis , phys . lett . *
b121 * , 123 ( 1983 ) ; l. alvarez - gaum , j. polchinski and m. wise , nucl . phys . *
b121 * , 495 ( 1983 ) .
v. berezinskii , _ et al .
_ , astropart . phys . * 5 * , 1 ( 1996 ) ; p. nath and r. arnowitt , northeastern report no .
nub - th-3151 - 97 , ( 1997 ) , hep - ph/9701301 .
j. ellis and f. zwirner , nucl .
b338 * , 317 ( 1990 ) ; g. ross and r.g .
roberts , nucl . phys .
* b377 * , 571 ( 1992 ) ; r. arnowitt and p. nath , phys . rev . lett . *
69 * , 725 ( 1992 ) ; m. drees and m.m .
nojiri , nucl .
phys . * b369 * , 54 ( 1993 ) ; s. kelley _ et .
nucl . phys . *
b398 * , 3 ( 1993 ) ; m. olechowski and s. pokorski , nucl . phys . * b404 * , 590 ( 1993 ) ; g. kane , c. kolda , l. roszkowski and j. wells , phys . rev . *
d49 * , 6173 ( 1994 ) ; d.j .
castao , e. piard and p. ramond , phys . rev . *
d49 * , 4882 ( 1994 ) ; w. de boer , r. ehret and d. kazakov , z. phys . * 67 * , 647 ( 1995 ) ; h. baer , m. drees , c. kao , m. nojiri and x. tata , phys . rev .
d * 50 * , 2148 ( 1994 ) ; h. baer , c .- h .
chen , r. munroe , f. paige and x. tata , phys .
d * 51 * , 1046 ( 1995 ) . m.s .
alam et al . , ( cleo collaboration ) , phys .
74 * , 2885 ( 1995 )
colrain and m.i .
williams , talk presented at the international europhysics conference on high energy physics , jerusalem , israel , august 1997 .
p. nath and r. arnowitt , phys .
b * 336 * , 395 ( 1994 ) ; phys .
lett . * 74 * , 4592 ( 1995 ) ; phys .
d * 54 * , 2374 ( 1996 ) ; f. borzumati , m. drees and m. nojiri , phys .
d * 51 * , 341 ( 1995 ) ; h. baer and m. brhlik , phys .
d * 55 * , 3201 ( 1997 ) . and references therein .
h. haber and r. hempfling , phys .
* 66 * ( 1991 ) 1815 ; j. ellis , g. ridolfi and f. zwirner , phys .
b * 257 * , 83 ( 1991 ) ; t. okada , h. yamaguchi and t. tanagida , prog .
* 85 * , 1 ( 1991 ) ; we use the calculations of m. bisset , ph.d .
thesis , university of hawaii ( 1994 ) .
h. baer , c .- h .
chen , m. drees , f. paige and x. tata , florida state university report no .
fsu - hep-970501 , hep - ph/9704457 .
m. carena , j.r .
espinosa , m. quiros , and c.e.m .
wagner , phys .
b * 355 * , 209 ( 1995 ) ; m. carena , m. quiros , c.e.m .
wagner , nucl .
b461 * , 407 ( 1996 ) ; h. haber , r. hempfling and a. hoang , cern - th/95 - 216 ( 1996 ) , hep - ph/9609331 . for reviews on dark matter and its detection ,
e. g. _ , v. trimble , annu .
. astrophys . * 25 * , 425 ( 1987 ) ; j.r .
primack , b. sadoulet and d. seckel , annu .
nucl . part .
sci . * 38 * , 751 ( 1988 ) ; p.f .
smith and j.d .
lewin , phys .
rep . * 187 * , 203 ( 1990 ) .
a. dekel and m.j .
rees , _ astrophys .
j. _ * 422 * , l1 ( 1994 ) ; a. dekel , annu .
. astrophys . * 32 * , 371 ( 1994 ) ; m. strauss and j. willick , phys . rep . * 261 * , 271 ( 1995 ) ; n. bahcall , l.m . lubin and v. dorman ,
j. * 447 * , l81 ( 1995 ) ; r.g .
carlberg , h.k.c .
yee , and e. ellingson , astro - ph/9512087 .
for a review and recent developments , see a. dekel , d. burstein , and s. white , talk presented at conference on critical dialogs in cosmology , princeton , nj , june ( 1996 ) , astro - ph/9611108 .
walker , g. steigman , d.n .
schramm , k.a .
olive and h - s .
kang , astrophys .
j * 376 * ( 1991 ) 51 ; m.s .
smith , l.h .
kawano and r.a .
malaney , astrophys .
* 85 * ( 1993 ) 219 ; c.j .
copi , d.n . schramm and m.s .
turner , science * 267 * , 192 ( 1995 ) .
m. white , d. scott and j. silk , annu .
astron . astrophys . *
32 * , 319 ( 1994 ) .
kolb and m.s .
turner , _ the early universe _ , ( addison - wesley , redwood city , ca , 1989 ) . for reviews , see g. jungman , m. kamionkowski and k. griest , phys .
* 267 * ( 1996 ) 195 ; p.
nath and r. arnowitt , presented at the international workshop on aspects of dark matter in astrophysics and particle physics , heidelberg , germany , sep 1996 , hep - ph/9610460 ; m. drees , invited talk at 12th dae symposium on high - energy physics , guahati , india , dec 1996 , hep - ph/9703260 ; and references therein .
m. bucher , a.s .
goldhaber , and n. turok , phys .
d * 52 * , 3314 ( 1995 ) ; a.d .
linde and a. mezhlumian , _ ibid _ * 52 * , 6789 ( 1995 ) ; j.r .
gott , nature , * 295 * , 304 ( 1992 ) .
riess , r.p .
krishner , and w.h . press , astrophys .
j. * 438 * , l17 ( 1995 ) ; astrophys .
j. * 473 * , 88 ( 1996 ) ; d. branch , a. fisher , e. baron , p. nugent , astrophys .
j. 470 , l7 ( 1996 ) ; w. freedman , talk presented at conference on critical dialogs in cosmology , princeton , nj , june ( 1996 ) , astro - ph/9612024 ; r. giovanelli _ et al _ , astro - ph/9612072 .
zeldovich , adv .
astrophys , * 3 * , 241 ( 1965 ) ; h .- y .
chiu , phys .
* 17 * , 712 ( 1966 ) ; b. lee and s. weinberg , phys .
* 39 * , 165 ( 1977 ) .
r. barbieri , m. frigeni and g.f .
giudice , nucl .
b * 313 * , 725 ( 1989 ) ; j.l .
lopez , d.v .
nanopoulos and k .- j .
yuan , phys .
b * 267 * , 219 ( 1991 ) and nucl .
b370 * , 445 ( 1992 ) ; j. ellis and l. roszkowski , phys .
b * 283 * , 252 ( 1992 ) ; l. roszkowski and r. roberts , phys .
b * 309 * , 329 ( 1993 ) ; m. kawasaki and s. mizuta , phys .
d * 46 * , 1634 ( 1992 ) ; g.l .
kane , c. kolda , l. roszkowski and j.d .
wells , phys .
d * 49 * , 6173 ( 1994 ) ; e. diehl , g.l .
kane , c. kolda and j.d .
wells , phys .
d * 52 * , 4223 ( 1995 ) .
m. drees and m.m .
nojiri , phys .
d * 47 * , 376 ( 1993 ) .
r. arnowitt and p. nath , phys .
b * 299 * , 58 ( 1993 ) ; b * 307 * , 403(e ) ( 1993 ) ; phys . rev
. lett . * 70 * , 3696 ( 1993 ) ; phys .
d * 54 * , 2374 ( 1996 ) ; j.l .
lopez , d.v .
nanopoulos and k. yuan , phys .
d * 48 * , 2766 ( 1993 ) ; m. drees and a. yamada , phys .
d * 53 * , 1586 ( 1996 ) ; j. ellis , t. falk , k.a .
olive and m. schmitt , phys .
b * 388 * , 97 ( 1996 ) ; cern report no .
cern - th-97 - 105 , hep - ph/9705444 .
h. baer and m. brhlik , phys .
d * 53 * , 597 ( 1996 ) .
d. matalliotakis and h.p .
nilles , nucl .
b * 435 * , 115 ( 1995 ) ; m. olechowski and s. pokorski , phys .
b * 344 * , 201 ( 1995 ) ; v. berezinskii , a. bottino , j. ellis ( cern ) , n. fornengo , g. mignola , s. scopel , astropart .
phys . * 5 * , 1 ( 1996 ) ; p.
nath and r. arnowitt , phys . rev . * d56 * , 2820 ( 1997 )
. k. griest and d. seckel , phys .
d * 43 * , 3191 ( 1991 ) .
p. gondolo and g. gelmini , nucl
b360 * , 145 ( 1991 ) .
g. anderson , s. raby , s. dimopoulos , and l.j .
hall , phys .
d * 49 * , 3660 ( 1994 ) ; s. dimopoulos , l.j .
hall and s. raby , phys .
d * 45 * , 4192 ( 1992 ) ; phys .
lett . * 68 * , 1984 ( 1992 ) .
aleph collaboration , talk presented at cern by g. cowan , february ( 1997 ) .
helas : helicity amplitude subroutines for feynman diagram evaluations , h. murayama , i. watanabe and k. hagiwara , kek-91 - 11 ( 1992 ) . v. barger and r. phillips , _ collider physics , updated edition _ , ( addison - wesley publishing company , redwood city , ca , 1997 ) . j. ellis , j. hagelin , d. nanopoulos , k. olive and m. srednicki , nucl . phys . *
b238 * , 453 ( 1984 ) .
t. blazek and s. raby , phys .
b * 392 * , 371 ( 1997 ) ; t. blazek , m. carena , s. raby and c.e.m .
ohio state report no . ohstpy - hep - t-96 - 026 , hep - ph/9611217 .
m. carena , m. olechowski , s. pokorski and c.e.m .
wagner , nucl .
b * 419 * , 213 ( 1994 ) .
s. perlmutter , _ et al .
_ , lbl-39291 , ( 1996 ) , astro - ph/9608192 .
lineweaver , d. barbosa , a. blanchard and j.g .
bartlett , astro - ph/9610133 ; m. zaldarriaga , d.n .
spergel and u. seljak , astro - ph/9702157 . | we calculate the relic density ( @xmath0 ) of neutralino dark matter in supergravity models with radiative electroweak symmetry breaking , for values of the higgs sector parameter @xmath1 between the two infrared fixed points of the top quark yukawa coupling , @xmath2 and @xmath3 . for @xmath4 ,
the cp - odd higgs pseudoscalar ( @xmath5 ) and the heavier cp - even higgs scalar ( @xmath6 ) are much heavier than the lightest neutralino ( @xmath7 ) , and the annihilation cross section of @xmath7 pairs is significantly enhanced only in the vicinity of the lighter cp - even higgs boson ( @xmath8 ) and the @xmath9 boson poles . for @xmath10 , @xmath11 and @xmath12
become comparable to @xmath13 , and the neutralino annihilation cross section can be significantly enhanced by the poles of the @xmath6 and the @xmath5 as well . for a cosmologically interesting relic density , @xmath14 , we find that the supergravity parameter space is tightly constrained .
the mass spectra for supersymmetric particles are presented for unification mass parameters in the cosmologically interesting region .
@=11 |
the recently claimed detection , at high confidence , of b - modes in cosmic microwave background ( cmb ) polarization that can not be attributed to lensed e - modes by the bicep2 experiment @xcite was greeted with both enthusiasm and caution .
if all or part of this b - mode signal were confirmed to be primordial , it would constitute the first detection of a smoking - gun for inflation , a direct probe of yet - unknown physics such as quantum gravity , and thus one of the most important discoveries in astrophysics , cosmology , and high - energy physics in the past several decades .
however , a joint analysis of bicep2 and _ planck _ data @xcite showed that most , if not all , of the bicep2 signal was the result of the most severe , ever - present contaminant in the studies of cmb polarization : polarized emission from interstellar dust .
this development does not slow down the search for primordial b - modes ; but it does highlight the need for a much - enhanced ability to understand and subtract foregrounds . _
planck _ has now established that even in sky regions away from the galactic plane , where the dust emission is minimum , the primordial b - mode signal is below that of dust @xcite . in order for a future b - mode experiment to detect a primordial signal ,
the polarized emission from interstellar dust and its polarization pattern will have to be subtracted from the cmb sky at high accuracy .
any systematic errors entering this correction will need to be carefully considered : even if they are small compared to the level of the dust signal , they may be imporatnt compared to the ( currently unknown ) level of the primordial signal .
it is such a systematic effect that we describe here .
currenty the subtraction of the dust signal is based on measuring polarized emission from dust at frequencies where it is dominant ( e.g. , @xmath0350 ghz ) and extrapolating the dust polarization pattern to frequencies dominated by the cmb and targeted by cmb experiments ( 60 - 150 ghz ) .
the basis of this approach is that the dust emission spectrum is reasonably well - understood ( typically represented by a modified blackbody spectrum ) ; and thus the dust polarization pattern measured at 350 ghz can be extrapolated to lower frequencies @xcite .
the polarization direction at 350 ghz is dictated by the magnetic field threading the interstellar clouds where the dust resides .
if for example all emission originates from a single cloud , then the emission will be partially polarized in a direction perpendicular to the plane - of - the - sky ( pos ) projection of the magnetic field .
the dust emission at 150 ghz should then be polarized in the same direction - since the polarization is generated by that same magnetic field .
the validity of such an extrapolation breaks down whenever more than one `` cloud '' ( more than one dusty region with different temperatures and magnetic fields ) is contributing to the integrated signal along the line of sight .
the reason is that the total linearly polarized intensity fraction and polarization direction is obtained by algebraic addition of each of the stokes linear polarization parameters @xmath1 , @xmath2 , and @xmath3 along the line of sight . if the relative contribution from two regions with different polarization directions ( effectively different magnetic field directions ) changes between frequencies , so will the resulting polarization fraction and polarization direction .
indeed , the relative contribution _ will _ change between frequencies if the two regions have different temperatures , due to the temperature - dependent black - body part of the dust emission spectrum .
in this paper we use a simple two - cloud model to demonstrate that this effect could cause decorrelation of the polarization pattern between 350ghz and 150ghz maps .
we calculate conditions under which the effect would be large in configuration ( rather than in fourier ) space .
and we propose ways to correct for this systematic effect .
we consider a simple case in which only two `` clouds '' of temperatures @xmath4 and @xmath5 and dust column densities @xmath6 and @xmath7 contribute to the total dust emission along a line of sight ( see fig . [ cartoon ] ) . two clouds with temperatures @xmath4 and @xmath5 , dust column densities @xmath6 and
@xmath7 , and magnetic fields oriented at different directions contribute to the total dust - emission intensity along a line of sight.,width=226 ] the total intensity emitted by each cloud can be generally well described by a modified black - body spectrum @xcite , @xmath8,\ ] ] where @xmath9 is the blackbody spectral radiance and @xmath10 is the opacity .
if the degree of polarization in a cloud is @xmath11 , then the polarized intensity will be @xmath12 . in this paper
we assume that @xmath11 is independent of frequency , although in the general case this is not so @xcite . in the case of low optical depth @xmath13 ( appropriate for regions of low dust emission , such as the low dust - emission regions typically targeted by cmb experiments ) ,
the ratio of polarized intensities contributed by each cloud can be written as @xmath14 this will be a function of frequency because of the blackbody part of the emission spectrum
. the polarization state of the dust emission from each cloud can be completely described by the stokes parameters @xmath1 , @xmath2 , and @xmath3 ( where @xmath3 is the total intensity ) .
the polarization degree @xmath11 and polarization angle @xmath15 of the emission from a single cloud are related to the stokes parameters through @xmath16 when both clouds contribute to the emission along a given line of sight , the corresponding stokes parameters add : @xmath17 the polarization degree @xmath18 and electric vector position angle @xmath19 of the total emission then are obtained by inserting @xmath20 , @xmath21 , and @xmath22 into eq .
( [ parameters ] ) .
if the pos - projection of the magnetic field in cloud 1 lies in a direction forming an angle @xmath23 with that of cloud 2 , the polarization directions of each cloud s emitted radiation will similarly form an angle @xmath23 . without loss of generality
, we can take the magnetic field of cloud 1 to be at an angle @xmath24 from the reference direction , and that of cloud 2 to be at an angle @xmath25 .
then @xmath26 , @xmath27 , and @xmath28 in both cases ( @xmath29 ) .
we can represent this configuration with stokes parameters @xmath30 and @xmath31 where we have taken @xmath32 , and @xmath33 carry the signs .
the stokes parameters of the total signal will then be @xmath34 the resulting polarization parameters will thus be @xmath35 we next consider the two extreme cases of relative magnetic field directions : parallel and perpendicular .
the simplest and least problematic case is the one where the directions of the pos magnetic field , and thus the polarization directions , are parallel in the two clouds ( i.e. @xmath36 ) . in this case ,
( [ general ] ) reduce to @xmath37 which simplifies to @xmath38 if @xmath39 .
if the polarization properties of both couds are the same , the resulting combined emission will also share these properties .
conversely , the most worrisome case is the one where the pos - projection of the magnetic field in cloud 1 lies in a direction perpendicular to that in cloud 2 . in this case , @xmath40 , so from eqs .
( [ general ] ) we obtain @xmath41 the polarization angle @xmath42 and @xmath43 if cloud 1 dominates in polarized emission ; conversely , @xmath44 and @xmath45 if cloud 2 dominates in polarized emission .
if @xmath46 retains the same sign between two frequencies , the resulting polarization angles will be the same .
if @xmath46 changes sign , the resulting polarization angles _ will abruptly rotate by @xmath47 as one changes from one frequency to another_.
upper panel : @xmath19 as a function of @xmath48 for different values of angle @xmath23 between the pos magnetic fields of the clouds
: black : @xmath49 ; blue : @xmath50 ; red : @xmath51 ; green : 75@xmath52 ; magenta:@xmath53 . lower panel : @xmath54 as a function of @xmath48 .
colours as above .
three different cases are shown , with @xmath55 , @xmath56 , and @xmath57 , distinguished by their different asymptotic behavior at high @xmath48 , when @xmath58 .
, width=264 ] upper panel : ratio of polarized intensities @xmath59 at 350 ghz and 150 ghz as a function of cloud 2 temperature , for three different cloud 1 temperatures .
lower panel : ratio @xmath59 at 150 ghz ( black ) and 350 ghz ( red ) , as a function of cloud 2 temperature .
cloud 1 temperature is fixed at 20 k. line type corresponds to different values of the product @xmath60 : dot - dash : 0.5 ; dash : 0.7 ; dot : 1 ; solid : 1.3 ; thick dot - dash : 1.5 ; thick dashed : 2 ; thick dot : 3 ; thick solid : 4 .
, title="fig:",width=264 ] upper panel : ratio of polarized intensities @xmath59 at 350 ghz and 150 ghz as a function of cloud 2 temperature , for three different cloud 1 temperatures .
lower panel : ratio @xmath59 at 150 ghz ( black ) and 350 ghz ( red ) , as a function of cloud 2 temperature .
cloud 1 temperature is fixed at 20 k. line type corresponds to different values of the product @xmath60 : dot - dash : 0.5 ; dash : 0.7 ; dot : 1 ; solid : 1.3 ; thick dot - dash : 1.5 ; thick dashed : 2 ; thick dot : 3 ; thick solid : 4 .
, title="fig:",width=264 ] the behavior of @xmath19 and @xmath18 as a function of the ratio of polarized intensities , @xmath48 , between the two clouds , is demonstrated in fig .
( [ twoofrho ] ) . when @xmath61 , cloud 2 dominates , and both @xmath19 and @xmath18 tend to the corresponding cloud 2 values ; conversely , when @xmath62 , cloud 1 dominates .
as a result , the maximum change in @xmath19 as @xmath63 changes is in all cases equal to @xmath23 .
this maximu change occurs when we transition from cloud 1 completely dominating the polarized emission to cloud 2 dominating .
the larger the angle @xmath23 , the more abrupt the transition between the two regimes , which occurs at @xmath64 .
the resulting polarization fraction @xmath18 is plotted in units of @xmath65 , and for three values of @xmath66 : @xmath67 , @xmath68 , and @xmath69 , distinguished in the plot by the different asymptotic behavior of the @xmath70 curves as cloud 1 dominates ( for @xmath62 ) .
the possible change in @xmath18 as @xmath48 changes is larger the larger the angle @xmath23 and the larger the difference between @xmath66 and @xmath65 . as discussed in
[ perp ] , for the extreme case @xmath71 , the resulting pos magnetic field has a step - function behavior between @xmath72 for @xmath73 , and @xmath74 for @xmath75 , while the resulting polarization fraction passes through zero at @xmath76 . in conclusion
, the potential change of the observed polarization properties of the combined emission from the two clouds , between 150 and 350 ghz , depends on the change in @xmath48 between the two frequencies . in turn , from eq .
( [ eqr ] ) , @xmath48 depends on the temperature of the two clouds , the ratio of their column densities , and the ratio of their polarization fractions .
note that @xmath48 _ does not _
depend on the ratio of temperatures @xmath77 , but has a more complicated dependence through the exponential part of the planck law .
in fact , when both clouds are hot enough to be in their rayleigh - jeans regime in the frequencies of interest , and the dependence of @xmath63 on the temperature _ is _ dominated by the @xmath77 ratio , the frequency dependence of @xmath48 cancels out , @xmath48 remains unchanged between 150 and 450 ghz , and the polarization properties of the resulting emission remain constant .
however , dust in typical ism temperatures ( @xmath78 k ) is _ not _ in the rayleigh - jeans limit between 150 and 350 ghz .
possible values of @xmath48 for interstellar medium clouds , for different temperatures of the two clouds and different values of the product @xmath60 , as well as likely changes of @xmath48 between @xmath79 and @xmath80 ghz are shown in fig .
[ roverr ] .
the manner in which the polarization properties measured for the combined signal change between 350 ghz and 150 ghz is shown in fig .
different colours correspond to different angles @xmath23 between the pos magnetic fields of the two clouds as in fig .
[ twoofrho ] , and different line types to different values of the product @xmath60 , as in fig .
[ roverr ] .
the upper panel shows the difference of the polarization angle of the combined signal as measured between 150 ghz and 350 ghz , and the lower panel shows the ratio @xmath81 , plotted in both cases against the temperature of cloud 2 .
the temperature of cloud 1 is fixed at the average interstellar dust value of 20 k @xcite . in the case of the degree of polarization ,
we have taken @xmath68 , and the true effect may thus be even greater ( see fig .
[ twoofrho ] , lower panel ) .
upper panel : difference of polarization angles of combined signal as measured at 350 ghz and 150 ghz , as a function of cloud 2 temperature .
lower panel : ratio of polarization fractions of combined signal as measured at 350 ghz and 150ghz as a function of cloud 2 temperature , assuming @xmath68 .
color corresponds to angle @xmath23 between the pos magnetic fields of the clouds : black : @xmath49 ; blue : @xmath50 ; red : @xmath51 ; green : 75@xmath52 ; magenta:@xmath53 .
line types as in fig .
[ roverr ] ( lower).,title="fig:",width=264 ] upper panel : difference of polarization angles of combined signal as measured at 350 ghz and 150 ghz , as a function of cloud 2 temperature .
lower panel : ratio of polarization fractions of combined signal as measured at 350 ghz and 150ghz as a function of cloud 2 temperature , assuming @xmath68 .
color corresponds to angle @xmath23 between the pos magnetic fields of the clouds : black : @xmath49 ; blue : @xmath50 ; red : @xmath51 ; green : 75@xmath52 ; magenta:@xmath53 .
line types as in fig .
[ roverr ] ( lower).,title="fig:",width=264 ] the first point to note is that once we consider the possible polarization pattern decorrelation as a function of all parameters entering the problem , a large range of outcomes is possible .
the physically transparent picture depicted in fig .
( [ twoofrho ] ) becomes much more complicated .
however , there are still generally informatve trends .
when the temperatures of the two clouds are different , severely misaligned cloud magnetic fields will result in significant differences in polarization degree and polarization angle between 150 ghz and 350 ghz for a range of possible parameters of our model .
caution is then warranted when using 350 ghz data to extrapolate dust properties to 150 ghz , as the 150 ghz polarized emission map might look very different from the 350 ghz map .
the polarization degree might differ by large factors ( more than a factor of 2 for the most misaligned magnetic fields , more than 20% if the pos magnetic field of the two clouds forms an angle higher than 75@xmath52 ) .
similarly , the polarization angle can be up to @xmath82 different if the clouds are perfectly misaligned , and certainly higher than 20@xmath52 for clouds with severely but not perfectly misaligned pos magnetic fields ( @xmath83 ) .
on the other hand , there are large parts of the parameter space where the polarization properties of the combined emission would look very similar at the two frequencies : if the pos magnetic fields of the two clouds form an angle of less than 60@xmath52 , then the difference in polarization fraction is less than @xmath84 and the difference in polarization angle less than @xmath85 , independently of the other parameters of our model .
this behavior , as discussed in the next section , offers a straghtforward way to correct for this systematic effect , by masking problematic regions with severe magnetic field misalignments along the line of sight .
the effect we have discussed here is intrinsic to all microwave experiments and is due to information loss due to line - of - sight integration .
future experiments with greater sensitivity / angular resolution will not avoid this systematic uncertainty .
however , if the magnetic fields of the contributing clouds form an angle smaller than @xmath53 , then the resulting change in the polarization properties maps from 350 ghz to 150 ghz is minimal ( at least under the assumptions of the simple model considered here ) . our model can be improved in a variety of ways : e.g. , more than two contributing clouds along the line of sight , a physical model of dust properties and emission , and constraining the observed combined emission spectrum along a line of sight to agree within uncertainties with the observed one .
additionally , in our model we assumed the degree of polarization to be constant across frequencies , and the frequency dependence of the dust opacity to be identical in the two clouds .
any deviation from these assumptions will cause further change between frequencies in the ratio @xmath48 of polarized emission between the two clouds .
further change in @xmath48 will cause further discrepancies between the maps in the two frequencies , up to the maximum difference , which is the difference between a state where cloud 1 completely dominates the polarized emission , and a state where cloud 2 completely dominates .
there are two ways for future experiments to deal with the systematic effect we have identified here .
the first is to calculate , for the specifications for each experiment , the magnitude of the error introduced by this effect in fourier space , and determine whether it is comparable to the level of the primordial signal they might claim .
this would involve simulating the effect using a detailed dust emission model , a simulated 3-dimensional map of high - galactic - latitude interstellar clouds and their magnetic fields , and the specific instrument response functions , mapping frequencies , and analysis techniques .
the second is to correct for the effect .
although a detailed analysis in fourier space is beyond the scope of this letter , some empirical data do exist addresing polarization pattern decorrelation of the dust signal between different frequencies in fourier space .
cross - correlation studies of the polarized microwave sky between frequencies performed by @xcite are limited to mid galactic latitudes ( lower than those targeted by cmb b - mode experiments ) and average on larger ( 10@xmath52 ) scales than those relevant for primordial b - modes .
mid latitudes are llikely to be less decorrelated , because a larger number of clouds is likely to be present along each line of sight , which more frequently than not will average out the strong variations that are possible in the case of only two clouds
. similar studies in @xcite at higher latitudes are limited to a smaller frequency difference ( between 353 and 217 ghz instead of 353 abd 150 ) . for larger frequency differences ,
the decorrelation will be larger , because the difference in polarized emission fraction from each cloud will be larger ( see fig .
( [ twoofrho ] ) ) .
even so , the @xmath86 pattern decorrelation ratio between 353 and 217 ghz is found to be at @xmath87 ( of the dominant dust signal ) ; assuming that it woud similarly be @xmath88 at cmb b - mode mapping frequencies , if the primordial signal is at 30% of the dust , the systematic discussed here would be at 10% of the primordial signal , and if the primordial signal is at 3% of the dust , the systematic would be at the 100% level . depending on the level of primordial signal ,
this effect could be of little concern , or very problematic . in configuration space ,
data are similarly not very constraining . if we were to make the simplistic assumption that each line of sight in a cmb b - mode experiment experiences polarized dust emission from two clouds and the magnetic fields in these cloud are randomly ( uniformly ) distributed , then the probability of a large , problematic ( @xmath89 ) misalignment of magnetic fields is 33% .
the realistic scenario is much more complicated : the probability of severe misalignments depends on the number of clouds within the galactic disk projected to high latitudes , the number of higher - distance high - velocity clouds that may also intervene along specific lines of sight , the size of each of these clouds , the correlation length of the magnetic field within a single cloud , and large - scale correlations of the magnetic field within the galactic disk .
certainly the sky at high latitudes appears `` patchy '' on the plane of the sky @xcite . by copernican arguments ( see also @xcite ) , it will also be patchy along the line of sight , with several distinct clouds , each with its own , potentially different , magnetic field .
additionally , at high latitudes , interstellar clouds are most likely to be located nearby , from a few tens of parsecs to one galaxy scale height ( few hundred parsecs ) , and thus span large angular scales .
( e.g. , the polaris flare cloud , @xcite , over @xmath90 across , galactic latitude @xmath91 up to @xmath92 , and the leo cloud , @xcite , @xmath93 across , @xmath91 up to @xmath94 ) .
halo objects can also emit in microwave frequencies ( e.g. high velocity clouds , such as ivc 135 + 53 , @xcite with @xmath91 up to @xmath53 , @xmath95 across ) . only a few severe misalignments between such large ( in angular scale ) clouds
may end up affecting a large fraction of the cmb sky .
this systematic effect could , however , be _ corrected _ through stellar optopolarimetry .
polarization of starlight induced by dichroic dust absorption traces the direction of the magnetic field in intervening absorbing clouds @xcite . if the optopolarimetric properties of many ( tens ) of stars within a single cmb b - mode beam were measured , and if these stars were located at various , known distances , then the magnetic field direction in the various clouds along each line of sight could be determined .
such _ magnetic tomography _ of the interstellar medium could imeddiatey identify lines of sight with severe misalignments of the magnetic field along the line of sight , and these lines of sight could be simply masked from b - mode analyses , akin to the treatment of point sources . in the era of gaia ( e.g. , @xcite ) which will provide distance measurements to stars down to 20th magnitude , the bottleneck in such an endeavor will be the number of high - accuracy optopolarimetric measurements that can be performed down to very low polarization fractions characteristic of the regions targeted by cmb experiments .
we thank a. readhead , b. hensley , n. kylafis , p. goldsmith , g. panopoulou , a. tritsis , and i. liodakis for useful comments .
we thank an anonymous referee for a constructive review .
we acknowledge support by : the `` robopol '' project , implemented under the `` aristeia '' action of the `` operational programme education and lifelong learning '' , co - funded by the european social fund ( esf ) and greek national resources ; the european commission seventh framework programme ( fp7 ) through grants : pcig10-ga-2011 - 304001 `` jetpop '' ; pcig - ga-2011 - 293531 `` sfonset '' ; pirses - ga-2012 - 31578 `` eurocal '' . | recent _ planck _ results have shown that radiation from the cosmic microwave background passes through foregrounds in which aligned dust grains produce polarized dust emission , even in regions of the sky with the lowest level of dust emission .
one of the most commonly used ways to remove the dust foreground is to extrapolate the polarized dust emission signal from frequencies where it dominates ( e.g. , @xmath0350 ghz ) to frequencies commonly targeted by cosmic microwave background experiments ( e.g. , @xmath0150 ghz ) . in this paper
, we describe an interstellar medium effect that can lead to decorrelation of the dust emission polarization pattern between different frequencies due to multiple contributions along the line of sight . using a simple 2-cloud model
we show that there are two conditions under which this decorrelation can be large : ( a ) the ratio of polarized intensities between the two clouds changes between the two frequencies ; ( b ) the magnetic fields between the two clouds contributing along a line of sight are significantly misaligned . in such cases ,
_ the 350 ghz polarized sky map is not predictive of that at 150 ghz_. we propose a possible correction for this effect , using information from optopolarimetric surveys of dichroicly absorbed starlight .
[ firstpage ] cosmology : inflation polarization ism : magnetic fields ism : dust cosmic background radiation cosmology : observations |
this is a continuation of a series of papers dedicated to the study of the distribution of neighbor denominators of farey fractions whose denominators are in arithmetic progression .
previously , we have treated in @xcite , @xcite the cases of pairs of odd and even denominators , respectively , while here we deal with tuples of consecutive denominators of fractions in @xmath4 , the set of farey fractions with denominators @xmath5 .
( here @xmath11 are integers , with @xmath2 and @xmath3 . )
the motivation for their study comes from their role played in different problems of various complexities , varying from applications in the theory of billiards to questions concerned with the zeros of dirichlet l - functions .
although the present work is mostly self - contained , the reader may refer to the authors @xcite and the references within for a wider introduction of the context and the treatment of some calculations .
two generic neighbor fractions from @xmath0 , the set of farey fractions of order @xmath1 , say @xmath12 and @xmath13 , have two intrinsic properties .
firstly , the sum @xmath14 is always greater than @xmath1 and secondly , @xmath15 .
none of these two properties is generally true for consecutive members of @xmath4 , but we shall see that they may be recovered as initial instances of some more complex connections
. given a positive integer @xmath16 , our main interest lies on the set of tuples of neighbor denominators of fractions in @xmath17 : @xmath18 ( notice that due to some technical constraints , in our notations , the dimension is @xmath19 and not @xmath7 . ) in fact our aim is to show that there is a limiting set @xmath20 of the scaled set of points @xmath21^{s+1}$ ] , as @xmath22 . strictly speaking ,
this is the set of limit points of sequences @xmath23 , where each @xmath24 is picked from @xmath25 .
more in depth information on @xmath20 is revealed if one knows the concentration of points across its expanse .
the answer is given by theorem [ theorem ] below , which shows that there exists a local density function on @xmath20 and gives an explicit expression for it .
next , let us see the formal definition .
let @xmath26 be a generic point in @xmath27^{s+1}$ ] and denote by @xmath28 the function that gives the local density of points @xmath29 in the @xmath19-dimensional unit cube , as @xmath6 , where @xmath9 are consecutive denominators of fractions in @xmath17 .
at any point @xmath30^{s+1}$ ] , we define @xmath31 by @xmath32 where @xmath33 are cubes of edge @xmath34 centered at @xmath35 .
a consequence of the fact that any sequence of consecutive denominators in @xmath36 is uniquely determined by its first two terms is the framework of @xmath20 , built as a union of two dimensional compact surfaces in @xmath37 .
this is the reason for which we have divided in by the area of a square of edge @xmath34 only , and not by @xmath38 .
thus , in reality @xmath39 is a function of two variables . here
@xmath40 will run over a domain that embodies the farey series , the farey triangle with vertices @xmath41 ; @xmath42 ; @xmath43 denoted by @xmath44 .
suppose now that @xmath45 , that is , we are in dimension two .
we conclude this introduction with some remarks on the shape of @xmath46 , for different @xmath47 and @xmath48 that we have tested ( see details , tables and pictures in section [ annex ] ) .
it is likely that our observations extend over all @xmath47 and @xmath2 .
the first thing to be remarked is the fact that for any @xmath2 , @xmath46 is a polygon obtained as a union of some sequences of polygons with constant local density on each of them , one of them always including all the others . but
the most noteworthy property is that each of these constant density polygons has a vertex at @xmath43 and looks like a mosaic composed by polygonal pieces , most of them being quadrangles .
the fact that the mosaics exist is not just an accidental occurrence ; on the contrary , more and more pieces fit into mosaics with a larger and larger number of components as @xmath48 increases .
the mosaics are either symmetric with respect to the first diagonal or they appear in pairs , whose components are symmetric to each other with respect to the first diagonal .
it is not true , as one would guess from tests with many acceptably small @xmath48 s and different @xmath47 s , that the larger mosaic ( most likely equal to @xmath46 ) is always a quadrangle .
the first counter - example is @xmath49 , which is a hexagon ( see fig . [ mosaic_312_11 ] ) .
the shape of the mosaics is more regular when @xmath48 has fewer prime factors . in particular , for each prime modulus @xmath48 , the exterior frame of all the mosaics is the same with that from the case @xmath50 , @xmath51 ( cf .
* fig . 5 , 6 ) ) , which in turn was the same in the case @xmath50 , @xmath52 ( cf .
mosaics having exactly the same form appear , for example , also when @xmath53 , @xmath51 , merely than each of them occurs twice .
as opposed to the prime modulus instance , we have included in section [ annex ] the pictures that appear in the case @xmath54 , @xmath55 .
one may appreciate these mosaics for their unexpected shape and beauty .
suppose the integer @xmath1 is sufficiently large , but fixed .
we also fix @xmath2 , the modulus , @xmath3 , the residue class , and an integer @xmath16 ( @xmath19 is the dimension ) .
then , we define recursively the following objects . for @xmath56 ,
let @xmath57 be given by @xmath58 , @xmath59 and @xmath60 , for @xmath61 , where @xmath62 $ ] .
we say that @xmath63 are _ generators _ of @xmath64 , and of @xmath65 also , or that @xmath64 and @xmath65 are _ generated _ by @xmath63 . in order to get a sequence of consecutive denominators of fractions in @xmath0 it suffices to know only the first two of them .
moreover , any two coprime integers @xmath66 , with @xmath67 , appear exactly once in the sequence of consecutive denominators of fractions in @xmath0 .
then , the subsequent denominators are obtained as follows .
given two neighbor denominators @xmath66 , they are succeeded by @xmath68 , where @xmath69 , for @xmath61 , and @xmath70 $ ] .
we put @xmath71 , @xmath72 .
notice that this values of @xmath73 coincide with those defined above if @xmath74 and @xmath75 .
then , in order to simplify the notation , we write @xmath76 instead of @xmath77 . as before
, we say that @xmath78 are _ generators _ of @xmath79 and of @xmath65 or that @xmath79 and @xmath65 are _ generated _ by @xmath78 .
a good way to look at a tuple @xmath80 is to think that it is associated to the whole @xmath81-tuple @xmath82 of consecutive denominators in @xmath0 .
we remark that the link between @xmath65 and @xmath83 is also made by the relations : @xmath84 , @xmath85 , @xmath86 , @xmath87 , etc .
it is plain that @xmath88 , for @xmath61 .
additionally , we need to extend the definition of @xmath65 to the case @xmath89 . then @xmath65 is empty , that is it has no components , and we say that its order is zero . in general , consecutive fractions in @xmath4
are not necessarily consecutive in @xmath0 , but have intercalated in - between several other fractions from @xmath0 .
we remark also that , in general , consecutive denominators of fractions in @xmath4 are not necessarily coprime .
let @xmath90 be an @xmath7-tuple of positive integers , and denote @xmath91 .
we say that @xmath92 , a tuple of consecutive denominators of fractions in @xmath4 , is _ of type _
@xmath93 if @xmath94 if all the components of @xmath83 are @xmath5 . ] and there exists @xmath95 a pair of consecutive denominators in @xmath0 with @xmath96 , @xmath97 , and @xmath98 , for @xmath99 . in this case we also say that the tuple @xmath100 , with @xmath101 is _ of type _
@xmath102 . to select the components that are @xmath103 , we define the _ choice application _
@xmath104 , with @xmath105 similarly , for @xmath106 $ ] , with @xmath107 , we also put @xmath108 let @xmath109 be the set of all @xmath110 of type @xmath111 , for any @xmath78 .
we remark that the generators of such a @xmath65 are , in general , not unique .
then , for any @xmath112 , we consider the set of residues relatively prime to @xmath113 , given by @xmath114 for example , suppose @xmath115 , @xmath52 , @xmath116 .
one can find in @xmath117 the following series of consecutive fractions : @xmath118 from these only @xmath119 survive , and are consecutive , in @xmath120 .
then , in our terminology , the tuple of denominators @xmath121 is of type @xmath93 , with @xmath122 . in particular , we see that @xmath123 are , respectively , the number of denominators of consecutive fractions in @xmath36 that are @xmath124 intercalated between the fractions with denominators that are @xmath5 . also , @xmath125 has @xmath126 components , and @xmath127
given a set @xmath128 and integers @xmath129 , let @xmath130 be the number of lattice points in @xmath131 with relatively prime coordinates congruent modulo @xmath48 to @xmath132 , respectively , that is , @xmath133 [ lno ] let @xmath134 and let @xmath128 be a convex set of diameter @xmath135 .
let @xmath48 be a positive integer and let @xmath129 , with @xmath136 .
then @xmath137 the proof follows by a standard argument , as in the proof of ( * ? ? ?
* lemma 3.1 ) . as a corollary of lemma [ lno ]
, we obtain an asymptotic formula for the cardinality of @xmath138 . for this
we use and two more facts .
firstly , the area of the farey triangle is @xmath139 and secondly , the number of residue classes @xmath140 that are relatively prime to @xmath47 is @xmath141 .
( here @xmath142 is the euler totient function ) .
then , we have @xmath143
we count separately the contribution to @xmath144 of points of the same type .
thus , we denote by @xmath145 , the local density in the unit cube @xmath27^{s+1}$ ] of the points @xmath29 of type @xmath102 , as @xmath146 . at any point @xmath30^{s+1}$ ] , this local density @xmath147 is defined by @xmath148 where @xmath33 are cubes of edge @xmath34 centered at @xmath35 .
then , we have @xmath149 provided we show that each local density @xmath147 exists , as @xmath146 . in the following we find each @xmath147 .
the variable @xmath150 is for now fixed , but eventually will tend to zero .
when @xmath197 , for each @xmath65 the polygons @xmath162 are smaller and smaller and converge toward a point @xmath198 , which we call the _ core _ of @xmath162 . if @xmath45 , @xmath162 is a parallelogram and the core of @xmath162 coincides with its center given by .
suppose now that @xmath199 is fixed .
the size of @xmath200 depends on the position of the core with respect to @xmath201 .
there are three cases .
if @xmath202 , we denote by @xmath203 , @xmath204 and @xmath205 , the topological interior , the boundary , and the set of vertices of @xmath206 , respectively .
] then , for @xmath193 small enough , @xmath207 , so @xmath208 . then , by , we get @xmath209 suppose now that @xmath210
. then there exists a certain bound @xmath211 such that if @xmath212 , the intersections @xmath213 are polygons similar to each other .
let @xmath214 be the polygon similar to these ones for which the variable @xmath193 equals @xmath215 in all the equations of the boundaries of the strips from , whose intersection is @xmath162 .
so , the size of @xmath214 is independent of @xmath193 .
notice that @xmath216 is generally smaller than @xmath217 , and even smaller than @xmath218 .
then , we have @xmath219 if @xmath220 , the reasoning from the previous case shows that there exists @xmath221 , with the property that for @xmath222 the polygons @xmath223 are similar to each other
. then , we denote by @xmath224 the polygon similar to these ones for which @xmath225 in all the equations of the boundaries of the strips from .
let us observe that the size of @xmath224 is independent of @xmath193 , and although we use the same notation , the polygons @xmath224 are distinct for different vertices of @xmath201 .
these yield @xmath226 inserting the evaluations from , and into , for @xmath227 it yields @xmath228 since the number of tuples @xmath229 of consecutive denominators of fractions in @xmath138 is @xmath230 , making use of and , it follows that @xmath231 by lebesgue differentiation , combining and , we get the following result .
[ theorem1 ] let @xmath2 and @xmath3 be integers . then
, for any @xmath232^{s+1}$ ] , we have : @xmath233 where the sums run over tuples @xmath112 .
we remark that in , the first term is essential , since it gives the local density on @xmath27^{s+1}$ ] , except on a set of area zero .
let @xmath150 be small and let @xmath151^{s+1}$ ] be the point around which we check the density .
we consider the parallelepiped centered at @xmath152 and edge @xmath34 given by @xmath153 .
then , given @xmath90 , we need to estimate the cardinality of @xmath154 this reduces to an area estimate if we put @xmath155 ^ 2\colon\ \begin{array}{l } x+y > q,\ \\
\displaystyle { { \mathbf{k}}}(x , y;|{{\mathbf r}}|-1)\in\a_{{\mathbf r}}(\c,\d),\ f_{{\mathbf r}}(x , y))\in q\cdot \box \end{array } \right\}.\ ] ] then , by lemma [ lno ] , @xmath156 where @xmath157 ^ 2\colon\ \begin{array}{l } x+y>1,\ \\
\displaystyle { { \mathbf{k}}}(x , y;|{{\mathbf r}}|-1)\in\a_r(\c,\d),\ f_{{\mathbf r}}(x , y)\in \box \end{array } \right\}.\ ] ] for any @xmath65 , we denote @xmath158 ^ 2\colon\ x+y,\ { { \mathbf{k}}}(x , y;|{{\mathbf r}}|-1)={{\mathbf{k}}}\big\}\ ] ] and @xmath159 ^ 2\colon\ f_{{\mathbf r}}(x , y)\in\box\big\}\,.\ ] ] also , we put @xmath160 , the farey triangle .
then , we have @xmath161 by a compactness argument it follows that only finitely many terms of the series are non - zero , although @xmath109 may be infinite .
next we need to see the shape of @xmath162 , since we are mainly interested to know @xmath163 .
this is the object of the next section .
the integer values @xmath73 defined in section [ section1 ] satisfy the classical _ mediant _ property of the farey series .
for instance , if @xmath165 are consecutive denominators of three fractions in @xmath0 , then @xmath166 is a positive integer .
hall and shiu @xcite called it the _ index _ of the fractions with denominators @xmath78 , respectively .
more generally , for a series of indices @xmath167 , we consider a sequence of polynomials @xmath168 , defined as follows .
let @xmath169 , @xmath170 , and then , for any @xmath61 , @xmath171 the first polynomials with nonempty argument are : @xmath172 often we write @xmath65 , meaning the sequence of indices starting with @xmath173 , but notice that the polynomial of rank @xmath174 depends only on the first variables @xmath175 .
in particular , one sees that @xmath176 coincides with the index of hall and shiu .
also , we remark the symmetry property : @xmath177 the role played by these polynomials is revealed by the next relation , which shows that for any @xmath178 , @xmath179 is a linear combination of @xmath180 and @xmath181 : @xmath182 turning now to the set @xmath183 defined by , where @xmath184 , by we see that this is the set of points @xmath185 that satisfy simultaneously the conditions : @xmath186 this shows that @xmath162 is the intersection of @xmath19 strips and for @xmath187 the sets @xmath188 and @xmath189 are similar , the ratio of similarity being equal to @xmath190 .
consequently , it follows that @xmath191 and @xmath192 is independent of @xmath193 . in particular , in the case @xmath45 , for @xmath194 ,
the set @xmath162 is a parallelogram of center @xmath195 and area @xmath196
putting together the contribution of points of all types , by and theorem [ theorem1 ] , we get the main result bellow .
[ theorem][theorem2 ] let @xmath2 and @xmath3 be integers .
then , for any @xmath235^{s+1}$ ] , we have @xmath236 as a consequence , we obtain as a natural object the support set .
[ corollary ] there exists a limiting set @xmath237 , as @xmath6 . when @xmath45 , the theorem can be stated more precisely using .
[ corollary1 ] let @xmath2 and @xmath3 be integers .
then , for any @xmath238^{2}$ ] , we have @xmath239 where the sums run over all @xmath240 and @xmath241 .
corollary [ corollary1 ] with @xmath50 retrieves the authors ( * ? ? ?
* theorem 3 ) as a particular instance .
the noteworthy thing hidden in the background of theorem [ theorem2 ] is the geometry of the arrangements of the domains @xmath242 , which we call _ pieces _ or _ tiles_. it is easier to see this in the bidimensional case , @xmath45 , assumed in what follows .
then the tiles are polygons included in @xmath27 ^ 2 $ ] and the choice application becomes @xmath243 for any @xmath80 , we shall call _ kernel _ the integer @xmath244 .
moreover , we say that it is the kernel of the tile @xmath245 .
notice that the inverse of the kernel is the contribution of each @xmath65 to @xmath246 .
the tiles of a given kernel fit into a few larger polygons , which we call _ mosaics_.
their common feature is that always one of their vertices is at @xmath43 .
they are either symmetric with respect to the first diagonal or they appear in pairs , symmetric to each other with respect to the first diagonal .
most of them are quadrangles , but their shape may vary a lot with @xmath48 , @xmath47 and the value of the kernel .
these mosaics behave like successive layers of constant density put over @xmath27 ^ 2 $ ] .
then the local density @xmath246 at a given point @xmath247 ^ 2 $ ] is the sum of the densities on the mosaics ( the inverse of its kernel ) stung by @xmath248 .
the contribution to the sum is halved if @xmath248 touches only an edge of a mosaic , and if the point touches a vertex of a mosaic , it adds to the sum the density reduced proportionally with the size of the angle of the mosaic at that vertex .
the number of mosaics that lay over @xmath249 is finite , and it is endless if @xmath250 . for each given @xmath11 ,
the number of the mosaics is unbounded , but their size has a certain rate of decay as their kernel increases . in the appendix
we have included the larger mosaics in two moduli : @xmath116 and @xmath54 .
it seems that the set @xmath46 is always equal to the first mosaic , which happens to be the largest .
this is known to be true when @xmath48 is small and in the cases @xmath51 and @xmath48 prime @xcite .
many other intriguing questions are raised by these objects . here , we conclude only by pointing out that each of these mosaics has an associated tree . in the nodes
the tree has the tuples @xmath65 that define the tiles and the arcs link nodes whose corresponding tiles are adjacent on the mosaic .
the root node corresponds to the tile with a vertex at @xmath43 . as an example , in figure [ tree ] , it is the tree associated to the mosaic @xmath251 $ ] from figure [ mosaic_25_91 ] .
999 f. p. boca , c. cobeli , a. zaharescu , _ on the distribution of the farey sequence with odd denominators _ , michigan math . j. , * 51 * ( 2003 ) , 557573 . c. cobeli , a. zaharescu , _ the haros - farey sequence at two hundred years _ , acta univ
. apulensis math .
inform . * 5 * ( 2003 ) , 138 . c. cobeli , a. iordache , a. zaharescu , _ the relative size of consecutive odd denominators in farey series _ , integers , * 3 * ( 2003 ) , a7 , 14 pp .
( electronic ) . c. cobeli , a. zaharescu , _ a density theorem on even farey fractions _ , preprint . c. cobeli , a. zaharescu , _ on the small neighbor denominators of the farey series with denominators in arithmetic progression _
, preprint .
r. r. hall and p. shiu , _ the index of a farey sequence _ , michigan math
. j. * 51 * ( 2003 ) , 209223 .
we have assigned names to the mosaics using the following conventions .
the first letter is either @xmath254 or @xmath255 , according to whether the mosaic is or not symmetric with respect to the first diagonal .
since the non - symmetric ones appear in pairs , symmetric to each other with respect to the first diagonal , we have included the picture of only one of them .
the next letter or group of letters indicates the shape of the mosaic .
the possible configuration are : triangle ( t ) , quadrangle ( q ) , pentagon ( p ) , hexagon ( h ) , octagon ( o ) or concave hexagon
v - shape ( v ) .
the argument is the tuple @xmath65 that gives the tile from the north - east corner .
finally , the subscript represents the number of components of @xmath65 . for example , the mosaic @xmath256 $ ] ( see figure [ mosaic_25_73 ] ) is a non - symmetric pentagon , whose tile from the n - e corner is the transformation of @xmath257 through @xmath258 , and @xmath259 $ ] ( figure [ mosaic_312_21 ] ) is a symmetric quadrangle , whose piece from the n - e corner is the image of @xmath260 through @xmath261 . as an exemplification , in figure [ tree ] we have included merely a tree associated to a mosaic
there , nodes are the tuples @xmath65 defining the tiles of @xmath256 $ ] and the arcs connect @xmath65 s that define adjacent tiles of the mosaic from figure [ mosaic_25_73 ] .
more data on the mosaics are entered in tables [ table1 ] and [ table2 ] . on the first column
, one can find the kernel , the number whose inverse gives the local density on the layer given by that mosaic .
the entry on the third column is the number of tiles arranged in the mosaic , while on the forth are the orders the number of components of @xmath65 s ( the smallest and the largest ) that produce the tiles . in the last column
are the coordinates of the vertices of the mosaic . for the modulus @xmath116 , the mosaics are the same in any of the cases @xmath262 or @xmath263 , but they are different when @xmath51 . when @xmath54 , the situation is more complex , mainly due to the larger number of factors of @xmath264 . due to arithmetical constraints , there are no mosaics of kernel @xmath265 when @xmath116 and @xmath262 or @xmath263 .
we remark that @xmath266 $ ] , for @xmath267 or @xmath263 and @xmath268 $ ] . in other words
, this says that the limiting set of pairs of consecutive denominators from @xmath17 equals , as @xmath6 , the largest of the mosaics .
+ & & & & + + & & & & + + 1 & sq_0 [ ] & 21 & 0 - 9 & ( 1,1 ) ; ( 0,1 ) ; ( 1/6,1/6 ) ; ( 1,0 ) + 2 & - & - & - & - + 3 & sq_1[3 ] & 7 & 1 - 5 & ( 1,1 ) ; ( 2/7,1 ) ; ( 3/8,3/8 ) ; ( 1,2/7 ) + 4 & sq_1[4 ] & 27 & 1 - 11 & ( 1,1 ) ; ( 3/13,1 ) ; ( 2/7,2/7 ) ; ( 1,3/13 ) + 5 & shv_4[2,2,2,2 ] & 35 & 4 - 14 & ( 1,1 ) ; ( 1/6,1 ) ; ( 8/43,23/43 ) ; ( 1/2,1/2 ) ; ( 23/43,8/43 ) ; ( 1,1/6 ) + 6 & sh_1[6 ] & 51 & 1 - 11 & ( 1,1 ) ; ( 1/5,1 ) ; ( 4/19,14/19 ) ; ( 3/8,3/8 ) ; ( 14/19,4/19 ) ; ( 1,1/5 ) + 7 & nq_2[2,4 ] & 6 & 2 - 6 & ( 1,1 ) ; ( 6/11,1 ) ; ( 3/5,2/5 ) ; ( 1,3/8 ) + 7 & nq_2[4,2 ] & 6 & 2 - 6 & ( 1,1 ) ; ( 3/8,1 ) ; ( 2/5,3/5 ) ; ( 1,6/11 ) + 7 & np_3[2,2,3 ] & 30 & 3 - 12 & ( 1,1 ) ; ( 3/13,1 ) ; ( 7/17,7/17 ) ; ( 4/5,1/5 ) ; ( 1,6/31 ) + 7 & np_3[3,2,2 ] & 30 & 3 - 12 & ( 1,1 ) ; ( 6/31,1 ) ; ( 1/5,4/5 ) ; ( 7/17,7/17 ) ; ( 1,3/13 ) + 8 & sq_1[8 ] & 21 & 1 - 9 & ( 1,1 ) ; ( 7/17,1 ) ; ( 4/9,4/9 ) ; ( 1,7/17 ) + 8 & sh_3[2,3,2 ] & 36 & 3 - 13 & ( 1,1 ) ; ( 7/37,1 ) ; ( 6/31,26/31 ) ; ( 4/9,4/9 ) ; ( 26/31,6/31 ) ; ( 1,7/37 ) + 9 & sq_1[9 ] & 33 & 1 - 9 & ( 1,1 ) ; ( 2/7,1 ) ; ( 9/19,9/19 ) ; ( 1,2/7 ) + 9 & np_4[2,2,2,3 ] & 7 & 4 - 15 & ( 1,1 ) ; ( 8/43,1 ) ; ( 7/37,32/37 ) ; ( 1/3,2/3 ) ; ( 1,5/7 ) + 9 & np_4[3,2,2,2 ] & 7 & 4 - 15 & ( 1,1 ) ; ( 5/7,1 ) ; ( 2/3,1/3 ) ; ( 32/37,7/37 ) ; ( 1,8/43 ) + & & & & + ( 0,-5)(22,16 ) ( 9,10 ) ( 9,9 ) ( 6,8 ) ( 12,8 ) ( 3,7 ) ( 9,7 ) ( 15,7 ) ( 1,6 ) ( 5,6 ) ( 9,6 ) ( 13,6 ) ( 17,6 ) ( 1,5 ) ( 7,5 ) ( 11,5 ) ( 15,5 ) ( 5,4 ) ( 9,4 ) ( 13,4 ) ( 17,4 ) ( 3,3 ) ( 7,3 ) ( 11,3 ) ( 15,3 ) ( 19,3 ) ( 5,2 ) ( 9,2 ) ( 13,2 ) ( 17,2 ) ( 15,1 ) 3 & sh_1[3 ] & & 1- & ( 1,1 ) ; ( 0,1 ) ; ( 1/13,5/13 ) ; ( 1/5,1/5 ) ; ( 5/13,1/13 ) ; ( 1,0 ) + 3 & sq_2[2,2 ] & & 2- & ( 1,1 ) ; ( 0,1 ) ; ( 1/5,1/5 ) ; ( 1,0 ) + 6 & sq_1[6 ] & 314 & 1 - 41 & ( 1,1 ) ; ( 1/9,1 ) ; ( 1/5,1/5 ) ; ( 1,1/9 ) + 6 & sh_5[2,2,2,2,2 ] & 424 & 5 - 39 & ( 1,1 ) ; ( 1/13,1 ) ; ( 1/9,5/9 ) ; ( 1/5,1/5 ) ; ( 5/9,1/9 ) ; ( 1,1/13 ) + 9 & sq_1[9 ] & 63 & 1 - 13 & ( 1,1 ) ; ( 1/5,1 ) ; ( 3/7,3/7 ) ; ( 1,1/5 ) + 9 & nq_4[2,2,2,3 ] & 56 & 4 - 18 & ( 1,1 ) ; ( 1/5,1 ) ; ( 1/4,1/2 ) ; ( 1,1/5 ) + 9 & nq_4[3,2,2,2 ] & 56 & 4 - 18 & ( 1,1 ) ; ( 1/5,1 ) ; ( 1/2,1/4 ) ; ( 1,1/5 ) + 12 & so_5[3,1,6,1,3 ] & 142 & 5 - 29 & ( 1,1 ) ; ( 1/5,1 ) ; ( 3/11,7/11 ) ; ( 4/13 , 8/13 ) ; ( 1/3,1/3 ) ; ( 8/13,4/13 ) ; ( 7/11,3/11 ) ; ( 1,1/5 ) + 12 & sq_11[2, ,2 ] & 21 & 11 - 20 & ( 1,1 ) ; ( 1/5,1 ) ; ( 1/2,1/2 ) ; ( 1,1/5 ) + 15 & sq_1[15 ] & 38 & 1 - 11 & ( 1,1 ) ; ( 3/7,1 ) ; ( 5/9,5/9 ) ; ( 1,3/7 ) + 15 & sh_7[2,3,1,5,1,3,2 ] & 173 & 7 - 29 & ( 1,1 ) ; ( 1/4,1 ) ; ( 3/11,7/11 ) ; ( 5/13,5/13 ) ; ( 7/11,3/11 ) ; ( 1,1/4 ) + 15 & nq_4[3,2,1,10 ] & 117 & 4 - 27 & ( 1,1 ) ; ( 3/11,1 ) ; ( 5/13,5/13 ) ; ( 1,1/3 ) + 15 & nq_4[10,1,2,3 ] & 117 & 4 - 27 & ( 1,1 ) ; ( 1/3,1 ) ; ( 5/13,5/13 ) ; ( 1,3/11 ) + 15 & nq_7[3,2,2,2,2,2,2 ] & 35 & 7 - 20 & ( 1,1 ) ; ( 1/3,1 ) ; ( 7/19,11/19 ) ; ( 1,1/2 ) + 15 & nq_7[2,2,2,2,2,2,3 ] & 35 & 7 - 20 & ( 1,1 ) ; ( 1/2,1 ) ; ( 11/19,7/19 ) ; ( 1,1/3 ) + 18 & sq_1[18 ] & 246 & 1 - 26 & ( 1,1 ) ; ( 1/5,1 ) ; ( 3/7,3/7 ) ; ( 1,1/5 ) + 18 & nq_6[2,3,1,5,1,4 ] & 128 & 6 - 28 & ( 1,1 ) ; ( 1/5,1 ) ; ( 3/7,3/7 ) ; ( 1,5/13 ) + 18 & nq_6[4,1,5,1,3,2 ] & 128 & 6 - 28 & ( 1,1 ) ; ( 5/13,1 ) ; ( 3/7,3/7 ) ; ( 1,1/5 ) + 21 & sq_1[21 ] & 36 & 1 - 11 & ( 1,1 ) ; ( 5/9,1 ) ; ( 7/11,7/11 ) ; ( 1,5/9 ) + 21 & nt_5[3,2,2,1,14 ] & 64 & 5 - 28 & ( 1,1 ) ; ( 1/3,1 ) ; ( 1,1/5 ) + 21 & nt_5[14,1,2,2,3 ] & 64 & 5 - 28 & ( 1,1 ) ; ( 1/5,1 ) ; ( 1,1/3 ) + 21 & sq_7[4,2,1,7,1,2,4 ] & 16 & 7 - 13 & ( 1,1 ) ; ( 5/9,1 ) ; ( 7/11,7/11 ) ; ( 1,5/9 ) + 21 & nq_8[2,2,3,1,4,2,1,7 ] & 12 & 8 - 14 & ( 1,1 ) ; ( 3/5,1 ) ; ( 7/11,7/11 ) ; ( 1,1/2 ) + 21 & nq_8[7,1,2,4,1,3,2,2 ] & 12 & 8 - 14 & ( 1,1 ) ; ( 1/2,1 ) ; ( 7/11,7/11 ) ; ( 1,3/5 ) + 21 & nq_10[2, ,2,3 ] & 22 & 10 - 21 & ( 1,1 ) ; ( 1/2,1 ) ; ( 5/7,3/7 ) ; ( 1,5/13 ) + 21 & nq_10[3,2, ,2 ] & 22 & 10 - 21 & ( 1,1 ) ; ( 5/13,1 ) ; ( 3/7,5/7 ) ; ( 1,1/2 ) + 24 & sq_9[6,1,3,1,6,1,3,1,6 ] & 79 & 9 - 26 & ( 1,1 ) ; ( 5/13,1 ) ; ( 1/2,1/2 ) ; ( 1,5/13 ) + 24 & shv_9[2,2,3,1,5,1,3,2,2 ] & 161 & 9 - 33 & ( 1,1 ) ; ( 1/5,1 ) ; ( 5/13,9/13 ) ; ( 2/3,2/3 ) ; ( 9/13,5/13 ) ; ( 1,1/5 ) + 24 & sh_11[4,1,4,1,4,1,4,1,4,1,4 ] & 94 & 11 - 27 & ( 1,1 ) ; ( 7/19,1 ) ; ( 17/43,29/43 ) ; ( 1/2,1/2 ) ; ( 29/43,17/43 ) ; ( 1,7/19 ) + 27 & sq_1[27 ] & 36 & 1 - 11 & ( 1,1 ) ; ( 7/11,1 ) ; ( 9/13,9/13 ) ; ( 1,7/11 ) + 27 & nq_6[10,1,2,3,1,6 ] & 124 & 6 - 37 & ( 1,1 ) ; ( 1/5,1 ) ; ( 11/19,10/19 ) ; ( 1,1/2 ) + 27 & nq_6[6,1,3,2,1,10 ] & 124 & 6 - 37 & ( 1,1 ) ; ( 1/2,1 ) ; ( 10/19,11/19 ) ; ( 1,1/5 ) + 27 & nq_8[2,3,2,1,8,1,2,4 ] & 32 & 8 - 23 & ( 1,1 ) ; ( 5/13,1 ) ; ( 7/15,11/15 ) ; ( 1,7/11 ) + 27 & nq_8[4,2,1,8,1,2,3,2 ] & 32 & 8 - 23 & ( 1,1 ) ; ( 7/11,1 ) ; ( 11/15,7/15 ) ; ( 1,5/13 ) + & & & & + | let @xmath0 be the set of farey fractions of order @xmath1 . given the integers @xmath2 and @xmath3 ,
let @xmath4 be the subset of @xmath0 of those fractions whose denominators are @xmath5 , arranged in ascending order .
the problem we address here is to show that as @xmath6 , there exists a limit probability measuring the distribution of @xmath7-tuples of consecutive denominators of fractions in @xmath4 .
this shows that the clusters of points @xmath8^{s+1}$ ] , where @xmath9 are consecutive denominators of members of @xmath0 produce a limit set , denoted by @xmath10 . the shape and the structure of this set are presented in several particular cases |
the double - hybrid ( dh ) approximations @xcite have become ones of the most accurate approximations for electronic - structure calculations within density - functional theory ( dft ) .
they consist in mixing hartree - fock ( hf ) exchange with a semilocal exchange density functional and second - order mller - plesset ( mp2 ) correlation with a semilocal correlation density functional : @xmath0 \nonumber\\ & & + ( 1-a_c ) e_{c}[n ] + a_c e_{c}^{\text{mp2 } } , \label{dh}\end{aligned}\ ] ] where the first three terms are calculated in a self - consistent hybrid kohn - sham ( ks ) calculation , and the last mp2 term is usually evaluated with the previously obtained orbitals and added _ a posteriori _ ( see , however , ref . for a double - hybrid scheme with orbitals optimized in the presence of the mp2 correlation term ) .
the two empirical parameters @xmath1 and @xmath2 are usually determined by fitting to a thermochemistry database .
a variety of such double hybrids have been constructed with different @xmath1 and @xmath2 parameters and various density functionals @xcite .
double - hybrid approximations with more parameters have also been proposed @xcite .
the so - called multicoefficient correlation methods combining hf , dft and mp2 energies can also be considered to be a form of double - hybrid approximation @xcite .
recently , sharkas _ et al . _
@xcite provided a rigorous theoretical justification for double hybrids based on the adiabatic connection formalism and which lead to a density - scaled one - parameter double - hybrid ( ds1dh ) approximation @xmath3 \nonumber\\ & & + e_c[n ] -{\ensuremath{\lambda}}^2 e_c[n_{1/{\ensuremath{\lambda } } } ] + { \ensuremath{\lambda}}^2 e_c^{\text{mp2 } } , \label{excds1dh}\end{aligned}\ ] ] where @xmath4 $ ] is the usual correlation energy functional evaluated at the scaled ( squeezed ) density @xmath5 . in this class of double hybrids only one independent empirical parameter @xmath6
is needed instead of the two parameters @xmath1 and @xmath2 .
the connection with the original double - hybrid approximations can be made by neglecting the density scaling @xmath7 \approx e_c[n ] , \label{ecnoscaled}\end{aligned}\ ] ] which leads to the one - parameter double - hybrid ( 1dh ) approximation @xcite @xmath8 \nonumber\\ & & + ( 1-{\ensuremath{\lambda}}^2 ) e_c[n ] + { \ensuremath{\lambda}}^2 e_c^{\text{mp2}}. \label{exc1dh}\end{aligned}\ ] ] corresponding to the standard double - hybrid approximation of eq .
( [ dh ] ) with parameters @xmath9 and @xmath10 .
ds1dh and 1dh approximations have been constructed using the perdew - burke - ernzerhof ( pbe ) @xcite and the becke - lee - yang - parr ( blyp ) @xcite exchange - correlation density functionals , and it was found that when neglecting density scaling the accuracy on atomization energies largely deteriorates for pbe but in fact improves for blyp ( see , also , ref . for a 1dh approximation based on the modified perdew - wang @xcite exchange functional and the perdew - wang-91 @xcite correlation functional ) .
by appealing to the high - density limit of the correlation functional , toulouse _
et al . _
@xcite argued that a more sensible approximation to the density - scaled correlation functional is @xmath7 \approx ( 1-{\ensuremath{\lambda } } ) e_c^{\text{mp2 } } + { \ensuremath{\lambda}}e_c[n ] , \label{ecscaledlinear}\end{aligned}\ ] ] leading to the linearly scaled one - parameter double - hybrid ( ls1dh ) approximation @xmath11 \nonumber\\ & & + ( 1-{\ensuremath{\lambda}}^3 ) e_c[n ] + { \ensuremath{\lambda}}^3 e_c^{\text{mp2 } } , \label{ls1dh}\end{aligned}\ ] ] which works reasonably well for pbe and corresponds to the form used for the pbe0-dh @xcite and pbe0 - 2 @xcite double hybrids . finally , note that fromager and coworkers @xcite explored the theoretical basis of the two - parameter double - hybrid approximations and proposed new double - hybrid schemes .
the vast majority of all these double hybrids have been applied with generalized - gradient approximation ( gga ) density functionals . only recently , goerigk and grimme @xcite proposed two - parameter double - hybrid schemes based on meta - gga density functionals . in these schemes ,
the opposite - spin mp2 correlation energy is combined with a refitted tao - perdew - staroverov - scuseria ( tpss ) @xcite meta - gga exchange - correlation functional , or with a refitted perdew - wang @xcite gga exchange functional and a refitted becke-95 ( b95 ) @xcite meta - gga correlation functional .
kozuch and martin @xcite also tested numerous spin - component - scaled double hybrids , with four or more optimized parameters , including the b95 meta - gga correlation functional and the tpss , b98 @xcite , bmk @xcite , @xmath12hcth @xcite meta - gga exchange - correlation functionals . in this work
, we reexamine the theoretical basis of double - hybrid approximations using meta - gga functionals , and we construct one - parameter double hybrids using the tpss meta - gga functional . while the 1dh and ls1dh schemes can be readily applied with a meta - gga functional , the ds1dh scheme requires an extension of the density scaling to the non - interacting kinetic energy density @xmath13 that is used in the tpss functional .
we then assess the accuracy of these one - parameter meta - gga double hybrids on test sets of atomization energies and reaction barrier heights .
in addition to the explicit dependence on the density @xmath14 , its gradient @xmath15 and possibly its laplacian @xmath16 , a meta - gga density functional also generally depends on the non - interacting positive kinetic energy density @xmath13 which can be seen as an implicit functional of the density ( for simplicity , we only write the equations for the spin - unpolarized case ; the extension to the general spin - polarized case is straightforward ) : @xmath17({\ensuremath{\mathbf{r } } } ) = { \ensuremath{\langle \phi[n ] \vert } } \hat{\tau}({\ensuremath{\mathbf{r } } } ) { \ensuremath{\vert \phi[n ] \rangle } } = \frac{1}{2 } \sum_i^{\text{occ } } \left| \bm{\nabla}_{{\ensuremath{\mathbf{r } } } } \phi_i[n ] ( { \ensuremath{\mathbf{r}}})\right|^2,\end{aligned}\ ] ] where the positive kinetic energy density operator @xmath18 has been introduced ( with the creation and annihilation field operators @xmath19 and @xmath20 ) , @xmath21 $ ] is the ( ks ) single - determinant wave function minimizing the kinetic energy @xmath22 and giving the density @xmath23 , and @xmath24 ( { \ensuremath{\mathbf{r}}})$ ] are the associated ( ks ) spin - orbitals . defining the scaled non - interacting kinetic energy density @xmath25 as the non - interacting kinetic energy density corresponding the scaled density @xmath26 ( where @xmath27 is an arbitrary positive scaling factor )
, one can show @xmath28({\ensuremath{\mathbf{r } } } ) & \equiv & \tau[n_{{\ensuremath{\gamma}}}]({\ensuremath{\mathbf{r } } } ) = \frac{1}{2 } \sum_i^{\text{occ } } \left| \bm{\nabla}_{{\ensuremath{\mathbf{r } } } } \phi_i[n_{\ensuremath{\gamma } } ] ( { \ensuremath{\mathbf{r}}})\right|^2 \nonumber\\ & = & \frac{{\ensuremath{\gamma}}^3}{2 } \sum_i^{\text{occ } } \left| \bm{\nabla}_{{\ensuremath{\mathbf{r } } } } \phi_i[n ] ( { \ensuremath{\gamma}}{\ensuremath{\mathbf{r}}})\right|^2 \nonumber\\ & = & \frac{{\ensuremath{\gamma}}^5}{2 } \sum_i^{\text{occ } } \left| \bm{\nabla}_{{\ensuremath{\gamma}}{\ensuremath{\mathbf{r } } } } \phi_i[n ] ( { \ensuremath{\gamma}}{\ensuremath{\mathbf{r}}})\right|^2 \nonumber\\ & = & { \ensuremath{\gamma}}^5 \tau[n]({\ensuremath{\gamma}}{\ensuremath{\mathbf{r } } } ) , \label{taug}\end{aligned}\ ] ] where the scaling relation @xmath29 ( { \ensuremath{\mathbf{r } } } ) = { \ensuremath{\gamma}}^{3/2 } \phi_i[n ] ( { \ensuremath{\gamma}}{\ensuremath{\mathbf{r}}})$ ] has been used . the scaling relation on @xmath30({\ensuremath{\mathbf{r}}})$ ] is consistent with the well - known quadratic scaling of the non - interacting kinetic energy @xmath31={\ensuremath{\gamma}}^2 t_s[n]$ ] where @xmath32=\int \tau[n]({\ensuremath{\mathbf{r } } } ) d{\ensuremath{\mathbf{r}}}$ ] @xcite .
furthermore , this scaling relation can easily be verified for one - electron systems with the von weizscker kinetic energy density , @xmath33 , and for the uniform electron gas , @xmath34 .
working with @xmath30({\ensuremath{\mathbf{r}}})$ ] as an implicit functional of the density in a self - consistent ks calculation requires to use the optimized effective potential approach to calculate the functional derivative of the exchange - correlation energy @xmath35 $ ] with respect to the density @xmath23 @xcite , which is computationally impractical .
the standard practice @xcite is to consider the exchange - correlation energy as an explicit functional of both @xmath23 and @xmath12 , @xmath36 $ ] . following this practice
, we define a density - scaled one - parameter hybrid ( ds1h ) approximation with a @xmath12-dependent functional as @xmath37+\bar{e}_{xc}^{{\ensuremath{\lambda}}}[n_{\phi},\tau_{\phi}]\bigl\ } , \label{ds1h}\end{aligned}\ ] ] where @xmath38 is a single - determinant wave function , @xmath39 is the kinetic energy operator , @xmath40 is the external ( e.g. , electron - nucleus ) potential operator , @xmath41 is the coulomb electron - electron interaction operator , @xmath42 $ ] and @xmath43 $ ] are the complement hartree and exchange - correlation functionals evaluated at the density and kinetic energy density of @xmath38 , @xmath44 and @xmath45 .
the complement hartree and exchange functionals are linear with respect to @xmath6 : @xmath46=(1-{\ensuremath{\lambda } } ) e_{{\ensuremath{\text{h}}}}[n],\end{aligned}\ ] ] @xmath47=(1-{\ensuremath{\lambda } } ) e_{x}[n,\tau],\end{aligned}\ ] ] where @xmath48 $ ] and @xmath49 $ ] are the usual ks hartree and exchange functionals .
the complement correlation functional is obtained via the extension of uniform coordinate scaling of the density @xcite to the kinetic energy density @xmath50 & = & e_{c}[n,\tau]-e_{c}^{{\ensuremath{\lambda}}}[n,\tau ] \nonumber\\ & = & e_{c}[n,\tau]-{\ensuremath{\lambda}}^2 e_{c}[n_{1/{\ensuremath{\lambda}}},\tau_{1/{\ensuremath{\lambda } } } ] , \label{}\end{aligned}\ ] ] where @xmath51 $ ] is the usual ks correlation functional , @xmath52 $ ] is the correlation functional corresponding to the interaction @xmath53 , @xmath5 is the scaled density and @xmath54 is the scaled kinetic energy density
. the minimizing single - determinant wave function @xmath55 in eq .
( [ ds1h ] ) is calculated by the self - consistent eigenvalue equation : @xmath56 + \hat{v}_{{\ensuremath{\text{h}}}}^{{\ensuremath{\lambda}}}[n_{\phi^{\ensuremath{\lambda } } } ] + \hat{v}_{xc}^{{\ensuremath{\lambda}}}[n_{\phi^{\ensuremath{\lambda}}},\tau_{\phi^{\ensuremath{\lambda } } } ] \right ) { \ensuremath{\vert \phi^{\ensuremath{\lambda}}\rangle } } \nonumber\\ = { \cal e}_0^{{\ensuremath{\lambda } } } { \ensuremath{\vert \phi^{\ensuremath{\lambda}}\rangle } } , \:\;\;\;\;\;\;\ ; \label{ds1heigenval}\end{aligned}\ ] ] where @xmath57 is the nonlocal hf potential operator , @xmath58 is the complement local hartree potential operator , and @xmath59 is the complement exchange - correlation potential operator @xmath60 = \int \frac{\delta \bar{e}_{xc}^{\ensuremath{\lambda}}[n,\tau]}{\delta n({\ensuremath{\mathbf{r } } } ) } \hat{n}({\ensuremath{\mathbf{r } } } ) d{\ensuremath{\mathbf{r } } } + \int \frac{\delta \bar{e}_{xc}^{\ensuremath{\lambda}}[n,\tau]}{\delta \tau({\ensuremath{\mathbf{r } } } ) } \hat{\tau}({\ensuremath{\mathbf{r } } } ) d{\ensuremath{\mathbf{r } } } , \nonumber\\ \label{vxclnt}\end{aligned}\ ] ] where @xmath61 is the density operator and the second term in eq .
( [ vxclnt ] ) corresponds to a non - multiplicative `` potential '' operator . the nonlinear rayleigh - schrdinger perturbation theory of refs .
can readily be extended to start with the ds1h reference of eq .
( [ ds1h ] ) with a @xmath12-dependent functional ( details are given in the supplementary material @xcite ) .
for the second - order energy correction , due to brillouin s theorem , only double excitations contribute and consequently the nonlinear terms of the perturbation theory vanish since they involve expectation values of the one - electron operators @xmath62 and @xmath63 between determinants differing by two spin orbitals .
the second - order energy correction to be added to the ds1h energy has thus a standard mp2 form @xmath64 where @xmath65 and @xmath66 refer to occupied and virtual ds1h spin - orbitals , respectively , with associated orbital eigenvalues @xmath67 , and @xmath68 are the antisymmetrized two - electron integrals .
the ds1dh exchange - correlation energy for a @xmath12-dependent functional is thus ( dropping from now on the explicit dependence on @xmath69 ) @xmath70 \nonumber\\ & & + e_c[n,\tau ] -{\ensuremath{\lambda}}^2 e_c[n_{1/{\ensuremath{\lambda}}},\tau_{1/\lambda } ] + { \ensuremath{\lambda}}^2 e_c^{\text{mp2}}. \nonumber\\ \label{excds1dhtau}\end{aligned}\ ] ] neglecting the scaling in the correlation functional , @xmath71 \approx e_{c}[n,\tau]$ ] , gives the 1dh exchange - correlation energy @xmath72
\nonumber\\ & & + ( 1-{\ensuremath{\lambda}}^2 ) e_c[n,\tau ] + { \ensuremath{\lambda}}^2 e_c^{\text{mp2 } } , \label{exc1dhtau}\end{aligned}\ ] ] and using the approximate scaling @xmath73 \approx ( 1-{\ensuremath{\lambda } } ) e_c^{\text{mp2 } } + { \ensuremath{\lambda}}e_c[n,\tau]$ ] gives the ls1dh exchange - correlation energy @xmath74
\nonumber\\ & & + ( 1-{\ensuremath{\lambda}}^3 ) e_c[n,\tau ] + { \ensuremath{\lambda}}^3 e_c^{\text{mp2}}. \label{ls1dhtau}\end{aligned}\ ] ] we apply these double - hybrid schemes with the tpss exchange - correlation functional and refer to them as ds1dh - tpss , 1dh - tpss and ls1dh - tpss .
llrrcllrr & & & + method & @xmath6 & mae & me & & @xmath6 & mae & me + blyp@xmath75 & & 6.52 & -1.18 & & & 8.10 & -8.10 + pbe@xmath75 & & 15.5 & 12.4 & & & 9.61 & -9.61 + tpss & & 5.79 & 3.78 & & & 8.41 & -8.41 + mp2@xmath75 & & 6.86 & 4.17 & & & 3.32 & 3.11 + + ds1dh - blyp@xmath75 & @xmath76 & 4.73 & -2.52 & & @xmath77 & 0.60 & 0.24 + 1dh - blyp@xmath75 & @xmath78 & 1.46 & 0.07 & & @xmath79 & 0.80 & -0.18 + b2-plyp@xmath75 & & 1.39 & -1.09 & & & 2.21 & -2.21 + ds1dh - pbe@xmath75 & @xmath77 & 3.78 & 1.30 & & @xmath76 & 1.32 & 0.48 + 1dh - pbe@xmath75 & @xmath80 & 8.64@xmath81 & 7.06@xmath81 & & @xmath76 & 1.42 & 0.12 + ls1dh - pbe@xmath82 & @xmath79 & 3.59 & 0.23 & & @xmath83 & 0.73 & -0.20 + ds1dh - tpss & @xmath83 & 0.91 & 0.07 & & @xmath79 & 0.59 & -0.14 + 1dh - tpss & @xmath84 & 5.74 & 2.37 & & @xmath85 & 0.74 & -0.28 + ls1dh - tpss & @xmath86 & 3.28 & 2.21 & & @xmath85 & 0.96 & 0.13 + + + [ cols="<,^,^,^,^",options="header " , ]
calculations have been performed with a development version of the molpro program @xcite , in which the ds1dh - tpss , 1dh - tpss and ls1dh - tpss approximations have been implemented .
the scaling relations for the scaled exchange - correlation energy and its derivatives for a general meta - gga functional are given in the appendix .
the empirical parameter @xmath6 is optimized on the ae6 and bh6 test sets @xcite .
the ae6 set is a small representative benchmark set of six atomization energies consisting of sih@xmath87 , s@xmath88 , sio , c@xmath89h@xmath87 ( propyne ) , c@xmath88h@xmath88o@xmath88 ( glyoxal ) , and c@xmath87h@xmath90 ( cyclobutane ) .
the bh6 set is a small representative benchmark set of forward and reverse hydrogen barrier heights of three reactions , oh + ch@xmath87 @xmath91 ch@xmath89 + h@xmath88o , h + oh @xmath91 o + h@xmath88 , and h + h@xmath88s @xmath91 hs + h@xmath88 .
all the calculations for the ae6 and bh6 sets were performed at the optimized qcisd / mg3 geometries @xcite using the dunning cc - pvqz basis set @xcite .
the performance of the best double hybrid is then checked on the larger benchmark set of 49 atomization energies of ref .
( g2 - 1 test set @xcite except for the six molecules containing li , be , and na ) at mp2(full)/6 - 31 g * geometries using the dunning cc - pvqz basis set .
core electrons are kept frozen in all our mp2 calculations .
spin - restricted calculations are performed for all the closed - shell systems , and spin - unrestricted calculations for all the open - shell systems .
in figure [ fig : ae6bh6 ] , we plot the mean absolute errors ( maes ) for the ae6 and bh6 test sets as functions of the parameter @xmath6 for the ds1dh - tpss , 1dh - tpss and ls1dh - tpss approximations .
the maes and mean errors ( mes ) of the double hybrids based on the tpss functional at the optimal values of @xmath6 which minimize the maes on the ae6 and bh6 sets are also reported in table [ tab : ae6bh6 ] , and compared to those obtained with standard blyp , pbe , tpss and mp2 , as well as with other double - hybrid approximations based on the blyp and pbe functionals . for @xmath92
, all these double hybrids reduce to a standard ks calculation with the tpss functional , while for @xmath93 they all reduce to a standard mp2 calculation . for the ae6 set ,
ds1dh - tpss gives by far the smallest mae with 0.91 kcal / mol at the optimal value of @xmath83 .
ls1dh - tpss gives a larger mae of 3.28 kcal / mol for an optimal value of @xmath86 , and 1dh - tpss gives a yet larger mae of 5.74 kcal / mol for an optimal value of @xmath84 , providing virtually no improvement over the tpss ks calculation at @xmath92 .
for the bh6 set , the three double - hybrid approximations are very similar for the entire range of @xmath94 , indicating that the scaling in the correlation functional contribution is not as crucial for barrier heights as for atomization energies .
the mae minima are 0.59 kcal / mol at @xmath95 , 0.74 kcal / mol at @xmath96 and 0.96 kcal / mol at @xmath96 for ds1dh - tpss , 1dh - tpss , and ls1dh - tpss , respectively . neglecting the scaling of the density and of the kinetic energy density in the tpss correlation functional , @xmath71 \approx e_{c}[n,\tau]$ ] , i.e. going from ds1dh - tpss to 1dh - tpss , largely deteriorates the accuracy of atomization energies .
a similar deterioration is obtained when neglecting the scaling of the density in the pbe correlation functional , whereas a large improvement of atomization energies is observed when neglecting the scaling of the density in the lyp correlation functional ( see table [ tab : ae6bh6 ] and ref . ) .
these different behaviors between the pbe and tpss correlation functionals on a one hand and the lyp correlation functional of the second hand may be related to the observation on atoms with non - degenerate ks systems that pbe and tpss are more accurate than lyp for the high - density limit ( or weak - interaction limit ) , @xmath97 $ ] @xcite .
contrary to most other double hybrids , the ds1dh - tpss double - hybrid approximation gives very close optimal values of @xmath6 on the ae6 and bh6 sets , i.e. @xmath98 and @xmath95 .
for general applications , we propose to use the value of @xmath99 for this double hybrid when using the cc - pvqz basis set .
finally , in table [ tab : g55 ] , we compare ds1dh - tpss ( with the optimal parameter @xmath100 ) with tpss and the standard double hybrid b2-plyp @xcite on the larger set of 49 atomization energies of ref . .
with maes of 3.9 and 2.3 kcal / mol for tpss and ds1dh - tpss , respectively , it is clear that the improvement in accuracy brought by ds1dh - tpss over tpss observed on the small ae6 set remains ( although smaller ) for this larger set .
ds1dh - tpss is however slightly less accurate on average on this set than b2-plyp ( mae of 1.6 kcal / mol ) .
we have constructed one - parameter double - hybrid approximations using the tpss meta - gga exchange - correlation functional and tested them on test sets of atomization energies and reaction barrier heights .
we have shown that neglecting the scaling of the density and of the kinetic energy density in the correlation functional largely deteriorates the accuracy on atomization energies , in contrast to what was previously found for double hybrids based on the blyp functional .
we thus propose the density - scaled double - hybrid ds1dh - tpss approximation with a fraction of hf exchange of @xmath100 as a viable meta - gga double hybrid for thermochemistry calculations , improving over both standard ks tpss and mp2 calculations .
we hope that this work will lead to more investigations of meta - gga double hybrids with minimal empiricism .
possible extensions of this work include introducing meta - gga functionals in multiconfigurational hybrids @xcite or in coulomb - attenuated double hybrids @xcite .
we thank andreas savin ( upmc / cnrs , paris ) for stimulating discussions .
we give the expressions for the scaled meta - gga correlation functional @xmath101={\ensuremath{\lambda}}^2 e_c[n_{1/{\ensuremath{\lambda}}},\tau_{1/{\ensuremath{\lambda}}}]$ ] and its derivatives ( see , also , refs . ) . starting from a standard meta - gga density functional , depending on the density @xmath14 , the square of the density gradient @xmath102 , the laplacian of the density @xmath103 and/or the non - interacting kinetic energy density @xmath13 @xmath104 & = & \int e_{c } \left(n({\ensuremath{\mathbf{r}}}),\left|\bm{\nabla}_{\ensuremath{\mathbf{r } } } n({\ensuremath{\mathbf{r}}})\right|^2 , \bm{\nabla}_{\ensuremath{\mathbf{r}}}^2 n({\ensuremath{\mathbf{r } } } ) , \tau({\ensuremath{\mathbf{r } } } ) \right ) d{\ensuremath{\mathbf{r } } } , \nonumber\\\end{aligned}\ ] ] the corresponding scaled functional is written as @xmath105 & = & \int e_{c}^{\ensuremath{\lambda}}\left(n({\ensuremath{\mathbf{r}}}),\left|\bm{\nabla}_{\ensuremath{\mathbf{r } } } n({\ensuremath{\mathbf{r}}})\right|^2 , \bm{\nabla}_{\ensuremath{\mathbf{r}}}^2 n({\ensuremath{\mathbf{r } } } ) , \tau({\ensuremath{\mathbf{r } } } ) \right ) d{\ensuremath{\mathbf{r } } } , \nonumber\\\end{aligned}\ ] ] where the energy density is obtained by the scaling relation @xmath106 the first - order derivatives of the energy density are @xmath107 and @xmath108 and @xmath109 and @xmath110 the same scaling relations apply for spin - dependent functionals @xmath111 $ ] . | we extend the previously proposed one - parameter double - hybrid density - functional theory [ k. sharkas , j. toulouse , and a. savin , j. chem . phys . * 134 * , 064113 ( 2011 ) ] to meta - generalized - gradient - approximation ( meta - gga ) exchange - correlation density functionals .
we construct several variants of one - parameter double - hybrid approximations using the tao - perdew - staroverov - scuseria ( tpss ) meta - gga functional and test them on test sets of atomization energies and reaction barrier heights .
the most accurate variant uses the uniform coordinate scaling of the density and of the kinetic energy density in the correlation functional , and improves over both standard kohn - sham tpss and second - order mller - plesset calculations . |
we acknowledge helpful discussions with p. simon , s. kettemann , g. zarnd , a. rosch , g. montambaux , c. texier , h. bouchiat and l.p . lvy .
in addition , we would like to thank g. zarnd for providing us with the numerical data of the nrg calculation of @xmath57 .
we are also indebted to r.a .
webb for the gold evaporation of sample au1 .
this work has been supported by the french ministry of science , grants # 02 2 0222 and # nn/02 2 0112 , the european comission fp6 nmp-3 project 505457 - 1 ultra-1d , and by the ipmc grenoble .
c.b . and g.e .
acknowledge financial support from procope .
hewson , _ the kondo problem to heavy fermions _ , cambridge studies in magnetism ( 1997 ) .
j. kondo , prog .
phys . * 32 * 37 ( 1964 ) .
wilson , rev .
phys . * 47 * 773 ( 1975 ) .
p. nozires , _ j. low temp
. phys . _ * 17 * 31 ( 1974 ) .
f. pierre and n.o .
birge , _ phys .
lett . _ * 89 * 206804 ( 2002 ) .
f. pierre et al . _
b _ * 68 * 085413 ( 2003 ) .
p. mohanty et al .
_ * 78 * 3366 ( 1997 ) .
b.l . altshuler et al .
_ j. phys .
c _ * 15 * 7367 ( 1982 ) ; for a review see g. grner and a. zawadowski , rep . prog . phys . * 37 * 1497 ( 1974 ) .
g. zarnd , et al .
* 93 * 107204 ( 2004 ) .
p. mohanty and r.a .
webb , _ phys .
lett . _ * 84 * 4481 ( 2000 ) .
f. schopfer et al .
. lett _ * 90 * 056801 ( 2003 ) .
e. akkermans and g. montambaux , in _ physique msoscopique des lectrons et des photons _ , ed .
edp sciences , ( 2004 ) . c. van haesendonck et al .
lett . _ * 58 * 1968 ( 1987 ) .
the fact that the extracted kondo temperatures are different for the two samples is somewhat surprising . however , these results are consistent with the maxima observed previously in the spin flip rate@xcite . c. rizutto , rep .
. phys . * 37 * 147 - 229 ( 1974 ) .
peters et al . , _ phys .
lett . _ * 58 * 1964 ( 1987 ) .
g. eska et al . , j. low temp .
phys * 28 * 551 ( 1977 ) .
s. hikami et al . , _ prog .
phys _ * 63 * 707 ( 1980 ) .
s. kettemann and e.r .
mucciolo , cond - mat/0509251 .
i.k.yanson et al . , _ phys .
lett . _ * 74 * 302 ( 1995 ) .
p. mohanty and r.a .
webb , _ phys .
lett . _ * 91 * 066604 ( 2003 ) .
d.s . golubev and a.d .
zaikin , phys .
lett . * 81 * , 1074 ( 1998 ) .
t. micklitz et al .
, cond - mat/0509583 . | we present measurements of the phase coherence time in quasi one - dimensional au / fe kondo wires and compare the temperature dependence of with a recent theory of inelastic scattering from magnetic impurities ( phys .
rev .
lett .
93 , 107204 ( 2004 ) ) .
a very good agreement is obtained for temperatures down to 0.2@xmath0 . below the kondo temperature @xmath0 ,
the inverse of the phase coherence time varies linearly with temperature over almost one decade in temperature .
the kondo problem has been fascinating physicist for more than 30 years .
such a large interest is due to the fact that this model , first introduced to describe transport properties of metals containing magnetic impurities , is a generic model for the description of many solid state physics phenomena@xcite . concerning magnetic alloys , the `` historical '' kondo solution first allowed to describe the transport properties at temperatures larger than the characteristic temperature of the model , the kondo temperature @xmath1@xcite . the major breakthrough has been the pioneering work of wilson who was able to calculate the ground state of the kondo problem at all temperatures using his numerical renormalisation group ( nrg)@xcite . another important contribution was due to nozires who showed that the zero temperature limit of the kondo model can be described within landau s fermi liquid theory@xcite which made precise predictions for transport as well as thermodynamic properties .
since then , many experiments have confirmed the relevance of this approach .
recently , it has been suggested that scattering by magnetic impurities@xcite is responsible for the experimentally observed saturation of the phase coherence time of electrons in metals at low temperatures@xcite , renewing the interest in kondo physics . though the altshuler - aronov - khmelnitzky ( aak ) theory describes the temperature dependence of the phase coherence time of electrons in pure metals@xcite , no exact solution is available for the phase coherence time in the presence of kondo impurities .
only a high temperature expansion , the nagaoka - suhl ( ns ) expression , was able to describe the experimental data at temperatures @xmath2@xcite . in the opposite limit ( @xmath3 ) ,
fermi liquid theory predicts a @xmath4 dependence of the inelastic scattering rate@xcite .
this , however , has never been observed experimentally . only very recently ,
zarnd and coworkers have been able to obtain an exact solution for the inelastic scattering time in kondo metals using wilson s nrg@xcite .
this calculation constitutes a major breakthrough for the decoherence problem , as it allows to compare experimental data with theoretical results _ for all temperatures , ranging from well above @xmath1 down to zero temperature_. it is thus of crucial importance to check whether this theory can describe correctly experimental data , as it could give new insights into the problem of the low temperature decoherence in mesoscopic metallic wires . in this letter
, we compare this recent theory with experimental results and show that the nrg calculation describes very well the experimental data at temperatures around and below the kondo temperature . in the first part of this letter we analyze the phase coherence time of quasi one - dimensional gold wires containing magnetic iron impurities of ref.@xcite , denoted as aufe1 and ref.@xcite , denoted as aufe2 , whereas in the second part we use the nrg calculation to interpret the phase coherence time as a function of temperature of an extremely pure gold sample ( au1 ) . in table
[ table ] , we list the geometrical and electrical parameters of these samples .
c c c c c c c c c sample&@xmath5&@xmath6&@xmath7&@xmath8&@xmath9&@xmath10 + & ( nm)&(nm)&(@xmath11m)&(@xmath12 ) & ( @xmath13/s)&@xmath14 + + aufe1&@xmath15&@xmath16&@xmath17&@xmath18&@xmath19&@xmath20 + aufe2&@xmath21&@xmath22&@xmath23&@xmath24&@xmath25&@xmath22 + au1&@xmath26&@xmath27&@xmath23&@xmath28&@xmath29&@xmath30 the phase coherence time of sample aufe1 and aufe2 , as extracted from standard weak localization measurements is shown in figure [ nrg ] .
both samples display a distinct plateau at a temperature above @xmath31 .
this plateau is caused by kondo spin flip scattering due to the presence of iron impurities which leads to dephasing .
decreasing the temperature , the magnetic impurity spin is screened and as a result , the phase coherence time increases again .
the temperature dependence of the measured phase coherence time in the presence of magnetic impurities can be described in the following way @xmath32 where @xmath33^{2/3 }
\ , t^{2/3}$ ] corresponds to the electron - electron interaction term@xcite , @xmath34 to the electron - phonon interaction , while @xmath35 correspond to the contribution due to magnetic impurities . to account for the scattering off magnetic impurities , we use on one hand the common nagaoka - suhl expression@xcite @xmath36}\quad \quad \pi^2 s(s+1)}{0.6\,{[ns]}\,\,\,(\pi^2
s(s+1 ) + ln^2(t / t_k))}\ ] ] where @xmath37 is the impurity spin , @xmath10 the impurity concentration expressed in parts - per - million(ppm ) and the prefactor of 0.6[ns ] has been calculated taking the electron density of gold@xcite . on the other hand we use the inelastic scattering rate calculated by nrg@xcite @xmath38 where @xmath39 is the inelastic scattering cross section at finite energy @xmath5 , @xmath40 the elastic scattering cross section at zero temperature and @xmath41 a numerical constant in units of [ s@xmath42 to express the impurity concentration @xmath10 in ppm .
sample aufe1 of ref.@xcite , @xmath43 sample aufe2 ( sample b ) of ref.@xcite .
the green dotted line ( denoted aak ) corresponds to the assumption that only electron - electron and electron - phonon interaction contribute to dephasing .
the black dashed - dotted lines ( ns1 and ns2 ) take in addition account for the magnetic scattering using the ns expression , whereas for the red ( nrg1 ) and blue ( nrg2 ) solid lines , the nrg calculation has been employed for the contribution of the magnetic impurities .
the inset shows 1/versus t on a linear scale for sample aufe2 in order to emphasize the _ linear _ regime.,width=257 ] the green dotted line in figure[nrg ] , denoted as aak , corresponds to the assumption that only electron - electron and electron - phonon interaction contribute to the electron dephasing and that there is no other mechanism for decoherence at low temperatures . for the simulation we have used @xmath44 and @xmath45 .
keeping these values fixed , and adding the contribution due to magnetic impurities using the ns expression for s=1/2 , we obtain the black dashed - dotted lines for sample aufe1 and aufe2 , denoted as ns1 and ns2 .
the ns expression describes relatively well the observed temperature dependence of at temperatures above @xmath0 but fails to describe the desaturation of due to the screening of the magnetic impurities at temperatures @xmath46 .
this is not at all surprising since the ns expression results from a perturbative expansion in @xmath47 and breaks down for @xmath48 . instead ,
taking the inelastic scattering rate obtained by nrg we obtain a very good agreement with the experimental data in basically the entire temperature range as shown by the red ( nrg1 ) and blue ( nrg2 ) solid lines for sample aufe1 and aufe2 , respectively . for the fitting procedure
, we have adjusted the magnetic impurity concentration such that the ns expression and the nrg calculation coincide at high temperatures ( @xmath49 ) .
one clearly sees that the ns expression deviates already from the nrg data at relatively high temperatures ( @xmath50 ) as already pointed out in ref.@xcite and fitting of experimental data with this expression at temperatures @xmath51 should therefore be avoided . from the nrg fitting procedure
we obtain an impurity concentration of approximately 3.3ppm and @xmath52 for sample aufe1 and 45ppm and @xmath53 for sample aufe2@xcite , in good agreement with @xmath0 values for au / fe found in the literature@xcite .
it is worth mentioning that the measurement of the phase coherence time is a very precise method to determine the kondo temperature as well as the magnetic impurity concentration . in resistance measurements ,
the determination of the kondo temperature is not straightforward since the entire temperature range down to the unitary limit ( @xmath54 is necessary to extract the kondo temperature with a satisfactory precision .
the nrg fitting procedure of on the contrary allows to extract the kondo temperature with high precision if temperatures only slightly lower than @xmath0 are attained . at the lowest temperatures we observe deviations from the nrg theory . a probable explanation for the deviation is fact that the spin of fe in au is not exactly 1/2 and the impurity spin might not be completely screened .
this eventually leads to interactions of magnetic impurities and a saturation of @xcite .
another possibility is the presence of another magnetic impurity with a much lower kondo temperature .
this latter issue will be discussed in the last part of this article .
the temperature evolution of below @xmath0 deserves several comments : after a slow increase of the temperature dependence is almost _ linear _ in temperature over almost one decade in temperature , as emphasized in the inset of figure [ nrg ] .
this relatively weak temperature dependence explains the fact why the pioneering experiments@xcite have not succeeded to observe the fermi liquid regime . comparing the experimental results with the nrg calculation
, we clearly see that the fermi liquid regime can only be reached for temperatures typically below @xmath55
. moreover , the aak behaviour is only recovered at extremely low temperatures ( @xmath56 ) .
it is noteworthy , that the calculated quantity by nrg is @xmath57 and not the phase coherence time measured in a weak localization experiment .
in fact , @xmath57 has been calculated in the limit of zero temperature and finite energy @xmath58 , whereas in a transport experiment one measures @xmath59 .
the fact that the numerical results describe this well the experimental data let us conclude that these two quantities are not very different , at least for @xmath60 , @xmath61 being the fermi energy . for different bias currents .
the dotted line corresponds to the theoretical expectation for the resistance correction .
the inset shows a sem photograph of the gold wire.,width=264 ] having now a theory which satisfactorily describes the temperature dependence of in the presence of magnetic impurities , let us reexamine the temperature dependence of in extremely pure gold wires . in a recent article@xcite
, the deviation of from the aak prediction at very low temperatures has been assigned to the presence of an extremely small amount of magnetic impurities ( typically on the order of 0.01ppm ) .
for this purpose we have fabricated an extremely pure gold wire ( au1 ) as shown in the inset of figure [ r_t ] .
the fabrication procedure is essentially the same as for the au / fe wires with the only difference that the wire has been evaporated in an evaporator which is exclusively used for the evaporation of extremely pure gold .
the gold of purity 5n5 has been evaporated directly on a silicon wafer @xmath62 a sticking layer .
in addition , special care has been taken for the sample design such that there is no influence on the phase coherence due to the two dimensional contact pads ( see inset of figure 2 ) .
for this wire the phase coherence length at the lowest temperatures is more than @xmath63 . to our knowledge , this is the largest coherence length ever obtained in a metallic wire and confirms the high purity of the sample . to determine the effective electron temperature of this sample
, we have measured the altshuler - aronov correction to the resistivity at very low temperatures .
a magnetic field of 40mt has been applied in order to suppress weak localization correction to the resistivity . in figure [ r_t ]
we plot the resistance correction as a function of @xmath64 . for measuring currents below 0.7na ,
the sample is in thermal equilibrium ( @xmath65 ) in the entire temperature range and the resistance correction follows the expected @xmath64 temperature dependence down to 10mk .
this clearly shows that the electrons of a mesoscopic sample can be cooled to such low temperatures .
fitting the temperature dependence of the resistance correction to @xmath66 ( dotted line in figure[r_t ] ) , we determine @xmath67 and compare it to the predicted value @xcite of @xmath68 , where @xmath69 is the thermal length and @xmath70 .
we obtain a value @xmath71 which is in very good agreement with the theoretical value of @xmath72 . ) .
the solid green line corresponds to the aak prediction , the black ( a ) , red ( b ) and blue ( c ) solid lines correspond to the nrg calculation assuming @xmath73mk , @xmath74mk , and @xmath75mk , respectively .
the inset shows typical magneto - resistance curves at different temperatures.,width=264 ] the phase coherence time is then measured via standard weak localization ( see inset of figure [ nrg_webb ] ) and the phase coherence length is extracted via the hikami - larkin - nagaoka formula@xcite . from the relation @xmath76
we then calculate the phase coherence time as displayed in figure [ nrg_webb ] .
we fit the experimental data with the aak expression such that an almost perfect agreement is obtained at high temperatures @xmath77 , as shown by the green solid line .
the prefactor we extract from this fit @xmath78 is in very good agreement with the theoretical prediction of @xmath44 . at temperatures below 100mk
our data deviate substantially from the aak prediction . to see whether these deviations can be explained by the presence of a very small amount of magnetic impurities , we simulate the temperature dependence of for the presence of a small amount of magnetic impurities using the nrg calculations .
the black ( a ) , red ( b ) and blue ( c ) solid lines correspond to a simulation assuming @xmath0=40mk , @xmath0=10mk , and @xmath0=5mk , with an impurity concentration of @xmath10=0.008ppm , @xmath10=0.013ppm , and @xmath10=0.015ppm , respectively .
it is clear from our simulations that only magnetic impurities with a kondo temperature @xmath7910mk and with a concentration smaller than 0.015ppm describe satisfactorily the experimental data .
a possible magnetic impurity with a kondo temperature in this temperature range is mn ( @xmath803mk)@xcite . for the sake of objectiveness
let us point out , however , the following : assuming magnetic impurities with a kondo temperature below the measuring temperature leads to an almost temperature independent scattering rate for @xmath81 .
any experimentally observed saturation of can therefore always be assigned to magnetic impurities with an unmeasurably low kondo temperature .
one could also argue , that it is curious that for the case of gold wires , the observed temperature dependence of can only be described satisfactorily by assuming the presence of one specific magnetic impurity with a kondo temperature below the measuring temperature @xmath8210mk , whereas it is known that the dominant magnetic impurity in gold is iron .
if we assume an additional iron concentration ( 0.015ppm ; @xmath0=500mk ) of the same order as for instance mn ( 0.015ppm ; @xmath0=3mk ) , the temperature dependence of does not satisfactorily describe the experimental data as displayed by the dotted black line ( d ) .
a possible explanation of these facts , might be the presence of a distribution of kondo temperatures , which could be significant for very diluted kondo impurities @xcite .
such a distribution of kondo temperatures has already be seen in point contact experiments@xcite .
this , however , can only be verified with phase coherent measurements at high magnetic fields@xcite .
we also note that our present results do not allow to rule out the predictions of ref .
@xcite .
in conclusion we have shown that the nrg calculation of the inelastic scattering rate describe very well the experimentally observed temperature dependence of caused by the presence of magnetic impurities .
below @xmath0 the inverse of the phase coherence time varies basically linearly with temperature over almost one decade in temperature .
the @xmath4 temperature dependence predicted by the fermi liquid theory , on the other hand , can only be reached for temperatures smaller than @xmath83 and remains an experimental challenge .
+ note added in proof : an exact calculation of in presence of disorder supports the above findings @xcite . |
among the most studied properties of the nuclear fission of actinides are the distributions of mass , charge and kinetic energy associated to complementary fragments .
however , r. brissot et al .
@xcite observed that the neutron emission from the fragments , occurred before they reach the detector , generates a structure on the final kinetic energy distribution as a function of mass , non - existent on primary fragment distribution .
naturally , neutron emission also erodes odd - even effects on neutron number distribution and consequently on mass number distribution associated to fragments . for these reasons ,
cold fission regions , corresponding to high kinetic energy associated to fragments that do not emit neutrons , are studied . in that sense , in this paper , mass , charge and kinetic energy distribution in the cold regions from thermal neutron induced fission of @xmath0u , @xmath1u and @xmath2pu , namely @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , respectively , is interpreted in terms of nucleon pair breaking and coulomb interaction energy between fragments in scission configuration .
+ + let us suppose that the complementary fragments have kinetic energies @xmath25 and @xmath26 , referred to primary mass numbers @xmath27 and @xmath28 , and proton numbers @xmath29 and @xmath30 , respectively .
after the neutron emission , complementary fragments 1 and 2 end with mass numbers @xmath31 and @xmath32 where @xmath33 and @xmath34 are the numbers of neutrons emitted by the fragments 1 and 2 , respectively .
the corresponding values of the final kinetic energy associated to those fragments are @xmath35 and @xmath36 in the case that only one of the complementary fragments is detected , which is what happens in the lohengrin spectrometer at the high flux reactor ( hfr ) of the laue - langevin institute ( ill ) in grenoble @xcite , subscripts are omitted . for a given primary mass number ( @xmath37 ) ,
the kinetic energy ( @xmath38 ) distribution is characterized by the corresponding average , @xmath39 , and standard deviation , @xmath40 .
+ + using the lohengrin spectrometer in the hfr of ill in grenoble , r. brissot _ et .
measured @xmath41 y @xmath42 values in @xmath1u(@xmath43,f ) .
their experimental curve @xmath42 showed a peak around @xmath44 , as it had been predicted by a monte carlo simulation performed before that the experiment was carried out . in this simulation
it was demonstrated that that broadening @xmath45 distribution is due to the neutron emission and the sharp fall on @xmath39 in the mass region around @xmath44 .
for each mass number @xmath46,the kinetic energy distribution is the result of superposition of distributions associated to fragments having @xmath47 @xcite .
+ + in the final mass region @xmath48 , @xmath49 falls sharply , about 4 mev per mass number .
let us suppose that the @xmath50 behaviour is similar to that of @xmath49 , and @xmath40 is flat ( equal to 5 mev , for example ) .
therefore , the average kinetic energies associated to fragments with @xmath51 is 8 mev lower than that of fragments associated to @xmath52 .
fragments with @xmath52 , that do not emit neutrons , have the high values of their distribution of primary energies @xmath53 ; and fragments associated to @xmath54 , that do emit 2 neutrons , have the low @xmath38 values of their corresponding primary distribution , whose average is already 8 mev lower than that associated to @xmath52 . both fragments , one with primary mass numbers 110 y and the other with 112 , respectively , having @xmath38 values with average differences of around 8 mev , end with @xmath44 .
the superposition of these dispersed values contributes with the widening of @xmath55 .
+ + d. belhafaf _ et .
al . _ , repeating the experiment on @xmath1u(n@xmath24,f ) made by r. brissot _ et .
@xcite , observed a broadening of the distribution of @xmath56 around @xmath57 @xcite .
moreover , for @xmath0u(n@xmath24,f ) , those authors observe a broadening of the distribution of @xmath56 around @xmath58 .
their simulation did not reproduce these experimental results @xcite .
+ + for @xmath0u(n@xmath24,f ) h. faust _ et .
@xcite , using theoretical models on fission dynamics , calculated the distribution of @xmath38 as a function of @xmath37 , but not around @xmath59 .
however the d. belhafaf _ et .
experimental results for both , @xmath0u(n@xmath24,f ) and @xmath1u(n@xmath24,f ) , were reproduced by a monte carlo simulation performed by de m. montoya _ et .
these authors shows that the broadening on the distribution of @xmath56 is generated by the superposition of the distributions of @xmath38 corresponding to @xmath60 . for the case of @xmath1u(n@xmath24,f ) , @xmath49 falls from 88 mev referred to @xmath61 , to 84 mev referred to @xmath62 , while in this region the tendency of @xmath49 is to increase with @xmath46 @xcite .
it is reasonable to assume that for @xmath37 near but higher than 126 , @xmath50 has a similar behavior .
thus , the superposition @xmath38 values , corresponding to primary mass numbers @xmath63 , having dispersive behavior , contribute to the broadening of @xmath45 distribution around @xmath57 .
the odd - even effect on the fragment mass number distribution ( @xmath64 ) is defined by the relation : @xmath65 where @xmath66 and @xmath67 are yields associated to fragments having even and odd mass numbers , respectively .
+ + similarly , we define the odd - even effects @xmath68 and @xmath69 on the distribution of neutron number ( @xmath70 ) and proton number ( @xmath71 ) , respectively . + + for low energy fission of actinides , the odd - even effect on the proton number distribution of the fragments , previously found to a. l. wahl _ et .
_ @xcite and g. mariolopoulos _ et .
@xcite , suggested that must exist high odd - even effects on the mass number distribution , but they are eroded by neutron emission .
+ + to study the mass number distributions undisturbed by the emission of neutrons , c. signarbieux _ et .
al . _ chosen high kinetic energy windows associated to fragments from @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , respectively .
the number of events analyzed were 1.5@xmath7210@xmath73 , 3@xmath7210@xmath73 and 3.2@xmath7210@xmath73 , respectively .
these authors used the difference of time of flight technique to separate fragment mass numbers , with solid detectors to measure the fragment kinetic energy .
the experiment was conducted in the hfr of ill in grenoble @xcite .
+ + although @xmath64 is not zero , it is not as high as was expected .
however , one may show that this small @xmath64 is compatible with nucleon pair breaking and higher odd - even effects in the distribution of proton or neutron numbers . + +
@xcite , for @xmath74 114.1 mev , obtained the mass number distribution presented in fig .
[ fig2 ] , from which one calculates @xmath75 30.14 % . the mass number distribution in the window of light fragment kinetic energy 113.5 @xmath76 114.5 mev , in @xmath0u(n@xmath24,f ) , is presented in fig .
[ fig3 ] @xcite . from this distribution one
gets @xmath75 32.18 % .
+ + in order to interpret values of @xmath64 , @xmath69 and @xmath68 , let s first define , for each @xmath37 , @xmath77 as the light fragment proton number that corresponds to maximal available energy of the reaction ( @xmath20 ) suppose that nucleon pair - breaking occurs , but fragment couple reaching maximum total kinetic energy still corresponding to @xmath78 ; therefore , for @xmath0u(n@xmath24,f ) , @xmath79 66.6 % and @xmath80 33.3 % .
this result is coherent with the stereographic projection of the values obtained by u. quade _ et .
@xcite for lower energy windows , as it is presented in figs .
[ fig4 ] and [ fig5 ] .
+ + for total kinetic energy @xmath81 204 mev , only @xmath82mo ( @xmath83 , @xmath84 ) survives @xcite ; therefore @xmath85 .
+ + the mass number distribution for the kinetic energy window 113.5 @xmath76 114.5 mev , for light fragments from the reaction @xmath1u(n@xmath24,f ) , is presented in fig .
[ fig6 ] , from which one calculates @xmath75 15.3 % . for @xmath74 113.7
mev , w. lang _ et .
_ obtained the distribution presented in fig .
[ fig7 ] , from which one calculates @xmath86 23.82 % . for @xmath74 108
mev , w. lang _ et .
_ obtain @xmath86 0 , @xmath87 35 % and @xmath88 8 % @xcite .
suppose that proton number correspond to @xmath78 , then one calculates @xmath79 66.6 % and @xmath80 33.3 % .
this result is coherent with the stereographic projection of the values obtained by w. lang _ et .
@xcite for windows of lower energies , as it is shown in figs .
[ fig8 ] and [ fig9 ] .
+ + for total kinetic energy @xmath89 203 mev , the surviving mass numbers are @xmath90 104 - 106 , among which there are 16 even and 8 odd mass numbers ; therefore @xmath75 30 % . for @xmath89 200 mev j. trochon _ et .
@xcite measured mass number and proton number distributions . from their results
one obtains @xmath86 20 % ; @xmath87 60 % , and @xmath88 60 % . + + the mass number distribution in the window of light fragment kinetic energy @xmath74 112 mev , in @xmath2pu(n@xmath24,f ) , obtained by c. schmitt _ et .
_ , is presented in fig .
[ fig10 ] @xcite . from this data one calculates @xmath86 3.7 % . for this energy
they obtain @xmath87 15 % , and @xmath88 10 % @xcite . for @xmath91 119 mev ,
one obtains the distribution presented in fig .
[ fig11 ] , from which one calculates @xmath75 7.6 % .
suppose that for @xmath92 mev , the proton numbers correspond to @xmath78 , therefore @xmath79 66.6 % and @xmath80 33.3 % , which are coherent with the stereographic projection of the values obtained by c. schmitt _ et .
for lower energy windows .
[ fig12 ] and [ fig13 ] .
+ + for @xmath93 mev , one obtains masses @xmath94 , with the higher yield corresponding to @xmath95 .
+ + for lower kinetic energy windows , one obtains @xmath96 .
this result does not imply that @xmath69 y @xmath68 values are zero . m. montoya _ et .
_ using a combinatorial analysis demonstrated @xcite that if there is one and only one nucleon ( proton o neutron ) pair breaking , then @xmath97 h. nifenecker _ et .
@xcite developed other combinatorial analysis that produces the same relation .
in cold fission from @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , shell and coulomb effects on mass number and kinetic distribution were observed @xcite .
+ + to describe `` coulomb effect '' let s first define the following curves depending of @xmath37 : i ) @xmath98 , the smoothed curve referred to maximal value of available energy of the reaction ( @xmath20 ) , as a linear curve in the mass region from the lightest @xmath37 to that at which ends the linear trend and , for higher mass numbers , a curve following the approximately horizontal trend of @xmath20 ; ii ) @xmath23 , the electrostatic interaction energy between complementary spherical fragments corresponding to @xmath78,with surfaces separated by 2 fm ( the result is a step function of mass number ) ; and iii ) @xmath99 , the maximal value of total kinetic energy , as the lowest of the ten highest values of total kinetic energy , defined as presented by m. montoya for @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , respectively @xcite .
+ + now let s define the smooth excess of electrostatic energy , as a function of @xmath37 , by the relation @xmath100 and the smoothed values of minimal excitation energy , as a function of @xmath37 , by the relation @xmath101 the curves of @xmath23 , @xmath99 , and @xmath98 values for @xmath0u(n@xmath24,f ) are presented in fig .
[ fig14 ] . + + the fluctuations of @xmath99 in @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f )
are interpreted as effects of the step variation of the electrostatic energy as a function of @xmath78 @xcite . + +
it is found that the smoothed values @xmath22 and @xmath21 are correlated .
this result suggests that , at scission , a higher excess of electrostatic interaction energy produces a higher deformation of fragments , therefore a higher @xmath21 and @xmath37 then lower @xmath99-values .
+ + for the mass regions @xmath102 in @xmath0u(n@xmath24,f ) ; @xmath103 in @xmath1u(n@xmath24,f ) ; and @xmath104 in @xmath2pu(n@xmath24,f ) : @xmath22 and @xmath21 are decreasing functions of @xmath37 .
[ fig15 ] , [ fig16 ] and [ fig17 ] .
+ + in @xmath0u(n@xmath24,f ) , the higher q - values , are around 204 mev , which occurs in the mass region @xmath105 , which corresponds to the lowest values of @xmath22 . see fig .
[ fig15 ] . for complementary fragments ( @xmath106 , @xmath107 ) and ( @xmath108 , @xmath109 )
, the @xmath99 value reach the respective @xmath20 value . + + as it is presented in fig .
[ fig16 ] , in the mass region @xmath105 , in @xmath1u(n@xmath24,f ) , @xmath22 well as _
x _ are very approximately zero .
+ + in@xmath2u(n@xmath24,f ) , the lowest _ x _ value occurs in the mass region @xmath110 .
[ fig17 ] .
+ + for @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , the value of @xmath22 is not zero .
at least one of the fragments must be deformed and spends energy for that .
however , molybdenum ( @xmath83 ) with @xmath111 , @xmath112 y @xmath113 ; and zirconium ( @xmath114 ) with @xmath115 , @xmath112 y @xmath113 ; have prolate geometries in their ground states , in addition they are transitional and soft @xcite . at scission , these fragments are deformed without excitation energy , and the electrostatic energy is equal to the @xmath20-value .
+ + for @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) , the fluctuations of mass number yield are anti - correlated to @xmath22 @xcite .
+ + for @xmath0u(n@xmath24,f ) , in the light fragment kinetic energy window 113.5 @xmath116 114.5 mev , the mass yield peaks toward a maximum around @xmath117 90 and 100 , respectively , which correspond to minimal values of @xmath22 @xcite . for @xmath1u(n@xmath24,f ) , in the light fragment kinetic energy window [email protected] mev , the mass yield peaks toward a maximum around @xmath90 90 y 102 , respectively , which correspond to minimal values of @xmath22 . for @xmath2pu(n@xmath24,f ) , in the light fragment kinetic energy region @xmath118 mev , the mass yield peaks toward a maximum around @xmath90 84 , 89 , 96 , 100 and 106 , respectively , which correspond to minimal values of @xmath22 .
+ + for @xmath0u(n@xmath24,f ) as well as for @xmath1u(n@xmath24,f ) , for@xmath119 , @xmath22 increases sharply with a and , therefore , @xmath99 decreases sharply too .
for the reaction @xmath1u(n@xmath24,f ) , @xmath22 decrease for @xmath120 .
+ + the fluctuations of @xmath21 and @xmath22 are correlated and they increase with asymmetry of fragmentations .
this is because the step - variation of the electrostatic interaction energy between fragments produced , by changing a unit of proton number , is more pronounced in asymmetric region , as one may in fig .
[ fig17 ] , corresponding to @xmath0u(n@xmath24,f ) .
+ + the electrostatic interaction energy between complementary fragments constitutes a barrier to fission in the channel of the corresponding charge and mass .
the probability to produce a couple of fragments will be higher for lower difference between this barrier and the available energy .
thus , for a same @xmath20-value , between neighbor mass numbers , the fragments with higher asymmetry of charge will reach higher values of kinetic energy .
this is equivalent to say for @xmath19 close to the maximal value the prevalent fragmentations will be those with higher asymmetry .
the coulomb effects in cold fission will be present also in proton number yields corresponding to odd proton number .
the higher yield corresponds to higher asymmetry , as it was shown by w. schwab _ et .
_ in the case of @xmath0u(n@xmath24,f ) @xcite .
neutron emission has erosive consequences on fragment mass and kinetic energy distribution in @xmath0u(n@xmath24,f ) , @xmath1u(n@xmath24,f ) and @xmath2pu(n@xmath24,f ) . to avoid those disturbances , the cold fission , i.e. very low excitation region ,
was studied .
contrary to expected , in cold fission , nucleon pair breaking still exists . nevertheless at the maximal total kinetic energy only even - even fragments survive . for lower energies , zero odd
- even effect on mass distribution does not exclude non zero odd - even effects on proton or neutron distributions .
+ + in addition , in cold fission , coulomb effects on charge , mass and kinetic energy distribution are present : for two fragmentations with equal @xmath20-value , the higher yield corresponds to the higher charge asymmetry .
moreover , coulomb effects are presented as fluctuations of the maximal kinetic energy value as a function of mass .
this effect is due to the fact that , each five units of mass number , there is a change of the proton number that maximizes the @xmath20-value , which produces a step variation of electrostatic interaction energy between complementary fragments with the same shapes . a similar effect on mass number yield curve is observed . due to the characteristics of electrostatic interaction ,
those effects are more notorious in higher asymmetry region .
finally , results from several experiments in cold fission suggest that scission explore all configurations permitted by the @xmath20-value .
r. brissot , j.p .
bocquet , c. ristori , j. cranon , c.r .
guet , h.a .
nifenecker , and m. montoya , _ proc . of a symp . on phys . and
chem . of fission _ , iaea .
vienna , ( 1979 ) .
d. belhafaf , j.p .
bocquet , r. brissot , ch .
ristori , j. cranon , h. nifenecker , j. mougey , and v.s .
ramamurthy , z. physik a : atoms and nuclei * 309 * , 253 , ( 1983 ) .
faust and z. bao , nucl .
a * 736 * , 55 , ( 2004 ) .
m. montoya , e. saettone y j. rojas , rev .
. fs . * 53 * , 366 - 370 , ( 2007 ) .
m. montoya , j. rojas , and i. lobato , rev .
. fs . * 54 * , 440 - 445 , ( 2008 ) .
wahl , a.e .
norris y r.a .
williams , _ proc . of a symp . on the phys . and chem .
of fission _ , iaea .
vienna , p. 813 , ( 1969 ) .
g. mariolopoulos , ch .
hamelin , j.p .
blachot , j.p .
bocquet , r. brissot , j. cranon , , h. nifenecker , ch .
ristori , nucl .
a * 361 * , 213 , ( 1981 ) . c. signarbieux , m. montoya , m. ribrag , c. mazur , c. guet , p. perrin , and m. maurel , j. physique lettres * 42 * , ( 1981 ) .
m. montoya , phd .
thesis , university of paris xi , orsay , ( 1981 ) .
clerc , w. lang , m. mutterer , c. schmitt , j.p .
theobald , u. quade , k. rudolph , p. armbruster , f. gnnenwein , h. schrader , and d. engelhardt , nucl .
a * 452 * , 277 - 295 , ( 1986 ) .
u. quade , k. rudolph , s. skorka , p. armbruster , h .-
clerc , w. lang , m. mutterer , c. schmitt , j.p .
theobald , f. gnnenwein , j. pannicke , h. schrader , g. siegert , and d. engelhardt , nucl .
phys . a * 487 * , 1 - 36 , ( 1988 ) .
w. lang , h .-
g clerc , h. wohlfarth , h. schrader , and k .- h schmidt , nucl .
a * 345 * , 34 , ( 1980 ) .
w. mollenkopf , j. kaufmann , f. gnnenwein , p. geltenbort , and a. oed , j. phys .
g * 18 * , l203-l210 , ( 1992 ) .
j. trochon , g. simon , j.w .
behrens , f. brissard , and c. signarbieux , radi .
effe . * 92 * , 327 - 332 , ( 1986 ) . c. schmitt , a. guessous , j.p .
bocquet , h .-
clerc , r. brissot , d. engelhardt , h.r .
faust , f. gnnenwein , m. mutterer , h. nifenecker , j. pannicke , ch .
ristori , and j.p .
theobald , nucl .
a * 430 * , 21 - 60 , ( 1984 ) .
h. nifenecker , g. mariolopoulos , j.p .
bocquet , r. brissot , ch .
hamelin , j. cranon , and ch .
ristori , z. phys .
a : atoms and nuclei * 308 * , 39 - 49 , ( 1982 ) . m. montoya , z. phys .
a : atoms and nuclei * 319 * , 219 - 225 , ( 1984 ) . m. montoya , r.w .
hasse , and p. koczon , z. phys .
a : atoms and nuclei * 325 * , 357 - 362 , ( 1986 ) . w. schwab , h .-
clerc , m. mutterer , j.p .
theobald , and h. faust , nucl .
a * 577 * , 674 - 690 , ( 1994 ) . | in this paper , the distribution of mass and kinetic energy in the cold region of the thermal neutron induced fission of @xmath0u , @xmath1u and @xmath2pu , respectively , is interpreted in terms of nucleon pair - breaking and the coulomb interaction energy between complementary fragments ( `` coulomb effect '' ) .
the fission process ends in the scission configuration , in which complementary fragments 1 and 2 , having proton numbers @xmath3 and @xmath4 , and mass number @xmath5 and @xmath6 , only interact each other with electrostatic repulsion , by what each one acquires kinetic energy @xmath7 and @xmath8 , respectively .
however , before reach detectors , they emit @xmath9 and @xmath10 neutrons .
thus , the final detected fragments have mass numbers @xmath11 ( @xmath12 ) and @xmath13 ( @xmath14 ) , and kinetic energies @xmath15 ( @xmath16 $ ] ) and @xmath17 ( @xmath18 $ ] ) , respectively . in order to avoid the erosive consequences of neutron emission ,
one studies the cold fission regions , corresponding to total kinetic energy ( @xmath19 ) close to the maximum available energy of the reaction ( @xmath20 ) .
contrary to expected , in cold fission is not observed high odd - even effect in mass number distribution .
nevertheless , the measured values are compatible with higher odd - even effects on proton or neutron number distribution , respectively .
in addition , in cold fission , the minimal total excitation energy ( @xmath21 ) is correlated with the `` coulomb energy excess '' ( @xmath22 ) , which is defined as the difference between @xmath23 ( the electrostatic interaction energy between complementary fragments taken as spherical in scission configuration ) and @xmath20 .
these `` coulomb effects '' increase with the asymmetry of the charge fragmentations . in sum , the experimental data on cold fission suggest that scission configurations explore all the possibilities permitted by the available energy for fission . |
understanding the chemical make - up of stars like the sun is fundamental to our understanding of star formation and stellar evolution .
@xcite undertook a classical study of the chemical evolution of the galactic disc , deriving abundances for a sample of 189 nearby field f and g dwarfs .
they found a number of interesting relationships in the data , for example , that the strongest age dependent abundance comes from ba , which they attribute to the efficient s - element synthesis in low - mass agb stars that enrich the interstellar medium long after star formation .
they confirmed that metal - poor stars ( [ fe / h]@xmath3 - 0.4dex ) are relatively overabundant in @xmath4-elements .
since the [ @xmath4/fe ] for these stars shows a gradient that decreases with increasing galactocentric distance of the orbits , they show that star formation was probably more vigorous and started first in the inner parts of the galactic disc .
since this work , various other samples have been studied , a number of which were performed by the exoplanet community ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these works have concentrated on studying the abundance distributions of exoplanet host stars compared with non - exoplanet hosts and a number of interesting trends have been found . along with the well established dependence of giant planet detection probability on host star metallicity @xcite
these abundance analyses indicate that various other atomic abundances are likely enhanced in exoplanet hosts compared to non - exoplanet hosts ( e.g. si & ni ) , at least for the gas giant population .
these over - abundances are explained in the framework of the core accretion scenario of planet formation @xcite where the more disc material present , the higher the probability of planet formation . in general
, the relative yields of different elements in the atmospheres of stars change with time , due to the processes of nucleosynthesis in the galaxy .
these processes are still poorly known , yet performing high quality analyses of homogeneous stellar samples can shed light on the fine underlying processes that are occurring , following the formation of planetary systems .
in recent years , series of large - scale abundance analyses have been published @xcite , mainly driven by the availability of high resolution spectral data from observational campaigns that are dedicated to searching for planets and the proliferation of software that allows us to automatically process the spectra of hundreds of stars in an automatic , or semi - automatic , manner .
the analysis we provide here is used to determine the chemical abundances in the atmospheres of metal rich stars , along with their relative behaviour versus iron .
it is worth noting that our sample is one of the generally homogeneous samples ( e.g. * ? ? ?
* ) in comparison to similar studies , which cover up to a thousand stars , but obtained using different instruments with different resolutions and pipeline reduction techniques ( e.g. * ? ? ?
on the other hand , some homogeneous studies cover rather smaller number of stars ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or with lower resolution ( e.g. * ? ? ?
in this work we combined both large enough sample and high resolution to provide us with a good statistical sample to investigate .
the layout of the manuscript is as follows , in section [ _ observations ] we discuss the observational data , in section [ _ abundance_method ] we describe the method used to derive the abundances , and , our selection criteria of the examined lines , the influence of the microturbulence and rotation .
section [ _ results ] contains the description of our results , and in section [ _ comparison ] we compare these results to the previously published works .
finally , in section [ _ discussion ] we discuss the implication of our results in the field , and summarise our findings in section [ _ summary ] .
the calan - hertfordshire extrasolar planet search ( cheps ) program @xcite is monitoring a sample of the metal rich stars in the southern hemisphere , to improve the statistics for planets orbiting such stars , along with searching for short period planets that have a high probability to transit their host star .
all stars in the sample were initially selected from the _ hipparcos _ catalogue to have @xmath5-band magnitudes in the range 7.5 to 9.5 .
this range ensures that the sample is not overlapping with existing planet search programs , since most solar - like stars with magnitudes below 7.5 are already being examined as part of other planet search programs , and the upper limit is set to ensure that the stars are bright enough to allow the best follow - up to search for secondary eclipses and transmission spectroscopy , essentially bridging the gap between the long - term precision radial velocity programs , and the fainter samples that photometric transit surveys are generally biased towards .
more specifically , the cheps sample primarily contains objects that were drawn from a larger southern sample observed using the eso - feros spectrograph ( see * ? ? ?
* ; * ? ? ?
the secondary selection of these targets was focused on the inactive ( log@xmath6 - 4.5dex ) and metal rich ( [ fe / h]@xmath00.1dex ) subset of this sample @xcite to ensure the most radial velocity stable targets , and to make use of the known increase in the fraction of planet - host stars with increasing metallicity , mentioned above .
we note that , recent work has shown that the fraction of metal rich stars hosting low - mass planets may not follow the metallicity trend observed in the gas giant population @xcite .
our subset of targets have been followed - up using the harps at la silla in chile . in our analysis , we use high s / n ( @xmath7100 ) and high - resolution spectra ( r@xmath8120,000 ) of predominantly single stars from the cheps sample that have been well characterised photometrically based on _ hipparcos _ data @xcite . the general properties of the sample are shown in fig .
[ _ figure_general_properties ] .
@xcite carried out the photometric determinations of the surface gravity parameter for our stars ( shown with open circles ) , and we compare these values to our spectroscopically measured ( filled circles ) .
we work within a narrow range between 5200 to 6200k , with the majority of our stars being metal rich and on the main sequence ( @xmath74.1 ) , although a few are slightly metal deficient ( [ fe / h]@xmath7 - 0.2 ) .
most of the stars in our sample belong to the thin disc galactic population ; see the toomre diagram at the bottom right of fig .
[ _ figure_general_properties ] .
we used a modified numerical scheme developed by @xcite that allows to determine the atomic abundances , rotational velocities , microturbulences and surface gravities in the stellar atmospheres from high resolution and high s / n spectra , all in the framework of iterative approach , i.e. at each new step , the model atmospheres and synthetic spectra were recomputed for the metallicities and gravities derived before in our procedure , and for the fixed effective temperatures that were set photometrically .
the method is based on the minimisation of differences in the profiles of computed and observed lines .
we used the two independent procedures for the final model parameters .
firstly , we required there to be no dependence of the fe abundance on the line strengths to obtain .
secondly , we required the agreement between the abundances of fe and fe obtained for the previously found to determine . for our starting point we used the photometrically determined metallicities and
( see * ? ? ?
* ) , then we ran a few iterations to determine [ fe / h ] , , [ fe / h ] and .
the final step was the determination of abundances and for the model atmosphere parameters found in the procedure before .
all synthetic spectra were computed by the wita6 program @xcite using 1d local thermodynamic equilibrium ( lte ) model atmospheres computed with sam12 code @xcite combined with the minimization routine abel8 @xcite .
the lines to be fitted in the stellar spectra were compiled using the solar spectrum by @xcite and atomic line data from vald-2 and vald-3 @xcite .
in our analysis we made a few basic assumptions : the level of activity is low in these atmospheres , in other words , the continuum in the optical range forms in the photosphere of the star .
both the micro- and macroturbulent velocity fields are similar to the solar case . for simplicity
we adopt that @xmath9 and do not change with the depth in these atmospheres .
the convection zone is similar to the solar convection zone , even in the case of super metal rich stars .
the minimum of the atmospheric temperature is located above the line formation region .
spots on the surface of these stars are not sufficiently numerous to contribute to the formation of absorption lines or continuum , in terms of the emitted fluxes .
naturally , if any of these assumptions is not valid for the physical state of the studied stars , we will obtain a spread of abundances and other determined parameters , even in the case of ideally determined and .
the effective temperatures were estimated using the photometric methods ( see * ? ? ?
* ) , exploiting the existing large photometric databases , along with the latest relationships between stellar broad - band colours and their photospheric effective temperatures .
namely , we used the johnson @xmath5-band photometry that was taken from the _ hipparcos _ database @xcite as our optical anchor point , and combined this with near - infrared photometry from the 2mass database @xcite , in particular the @xmath10-band magnitude that gives a sufficiently large wavelength baseline to sample the shape of the spectral energy distribution .
we then used the relationships provided in @xcite to calculate accurate effective temperatures and place the stars on an hr - diagram to calculate the photometric values using the y2 evolutionary models @xcite .
we choose to use these photometric for the computations , rather than update them spectroscopically .
we believe this constrains uncertainties which may come from the quality of fits to the spectra and atomic line data , and these may also affect the results of our determination of abundances , microturbulent velocities , and surface gravities .
we provide our analysis for the pre - selected list of spectral lines extracted from vald-2 and vald-3 @xcite . to create the list of reliable lines for each element , we computed the solar synthetic spectra and convolved them to get the effective resolving power of r@xmath870,000 , which is the effective resolution due to macroturbulent velocity field in the solar atmosphere .
the theoretical spectra were fitted to the selected observed lines using the minimization routine abel8 . at this stage , we excluded severely blended lines , too weak lines with residual fluxes lower than 20 per cent and strong lines with @xmath110.7 to minimize the distortion of the results due to noise and possible nlte effects in the cores of the strong lines @xcite .
we computed our synthetic spectra using the damping constants provided by vald @xcite . for the absorption lines without damping constants we computed them using the unsold formulas @xcite .
the fitting part of each line profile was adopted manually to reduce the uncertainties introduced by wing blending . in some cases
, very close blends of the lines of the same element were fitted together , resulting in a single spectral range to fit multiple lines . the final list of fitting ranges for the neutral and ionized atoms available on cds .
the microturbulent velocity is an important parameter in 1d line - profile fitting techniques as there is an evident trend with abundances that causes uncertainties in their measurements .
the dependence of fe and fe abundances on microturbulent velocity for hd 102196 is shown in fig .
[ _ figure_method_microturbulence ] . here ,
fe and fe abundances differ by up to @xmath10.1dex at @xmath121.0but the same at 1.4 .
we carried out a set of synthetic spectra fits to fe and fe lines independently , using the grid of adopted microturbulent velocities in range from 0.0 to 2.6 , with a step of 0.2 .
in fact , we followed the procedure of @xcite , but here we investigated the dependence of @xmath13 , where @xmath14 and @xmath15 are the abundance and the central intensity of the corresponding iron absorption line computed for the grid of , the @xmath16 is known to show the dependence on .
therefore , we determine our at @xmath16=0 . our procedure is based on fitting to the line profiles , excluding the shallow parts of the wings , which could be more affected with blending by the weak lines of other elements .
furthermore , the wings of lines are more affected by the pressure broadening than the microturbulent velocities .
accounting both these factors should improve the determination of ( see fig .
[ _ figure_microturbulence_comparison ] ) .
the profiles of absorption lines in the stellar spectra affected by the rotation of a star , depending on the magnitude of rotation .
the rotational profile , in our case , determined by the formula of @xcite , is of a different shape compared to the instrumental broadening and macroturbulence . in our analysis ,
each line of the synthetic spectra was convolved with a profile of a different , and was fitted to the observed spectra until the best result was found .
despite we used only fe lines for the determination of , the procedure was also applied for every other line of all our elements to get a better fit to the observed line profiles .
we found that the majority of our stars are slow rotators , with @xmath34(fig . [ _ figure_rotation ] ) .
likely , a few stars with larger rotational velocity present the cases where sin@xmath171 , rather than large @xmath18 itself , since these stars are of lower activities , with only a few fast rotating active stars left over . on the other hand ,
correlates with , older stars rotate slower , in agreement with the theory , and vice versa , which is in line with their ages given by the preliminary cheps selection .
we can see an uprising trend of for the stars of lower , which represent a younger population .
for the elements other than fe , was used rather as the adjusting parameter . nevertheless , it agreed for the properly fitted lines of different elements within an uncertainty of less than 1 .
we can claim a clear measurement for the values of @xmath72 .
uniform results for the iron lines give credibility to our results .
we carried out a differential analysis with respect to the sun as a star , since our line lists were selected based on a comparison of the synthetic spectra to the observed solar spectrum by @xcite .
this allowed us to minimize possible blending effects , at least in the solar case .
we performed a quantitative analysis of the solar abundances to verify our procedure .
computations were done for the initial model atmosphere with //[fe
/ h]=5777/4.44/-4.40 , computed using sam12 @xcite . for the iron ions , we obtained n(fe)[email protected] and n(fe)[email protected] , which again highlights the known problem of the measured difference between fe and fe abundances obtained in the framework of the classical approach ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the abundances for the other elements ( see the first column of table [ _ table_solar ] ) are in agreement with those in the literature ( e.g. * ? ? ?
* ; * ? ? ?
we also determined the microturbulence and projected rotational velocity for the sun as a star , measuring = [email protected] = [email protected] , which also agree with known results .
the solar abundances determined for the different atomic line data and line lists are discussed in section [ _ comparison ] .
the complete table of abundances available on cds . in our analysis
, we used the parameter @xmath19 to describe the average slope of the distribution of abundances for a given element relative to iron .
we computed @xmath19 using a standard least squares approach to approximate the dependence of [ x / fe ] versus [ fe / h ] by a linear function . by definition , @xmath19 characterises , to first - order , the relative changes of the yield of elements with respect to the iron . in order to make a useful comparison with other works we needed to translate onto a common scale
we found the easiest way to achieve this was to adopt the scale of @xcite rather than in terms of the solar abundances determined by our procedure . in some papers ,
abundances are provided in relation to derived solar scales in order to compensate for the differences in procedures ( e.g. * ? ? ?
* ; * ? ? ?
other authors adopt @xcite or its later derivatives ( e.g. * ? ? ?
* ; * ? ? ?
* ) , though they did not consider their derived solar abundances .
in order to compare the different samples studied by different authors we converted abundances for the various comparison samples to the scale of @xcite ( fig .
[ _ figure_samples ] ) .
this also allowed us to assess the general accuracy level of stellar abundance determinations from the combined spread of abundances . in fig .
[ _ figure_overiron ] we show the dependence of abundances relative to iron with their corresponding error bars .
these uncertainties depend on the number of lines for each element , quality of fit , the scatter of abundances determined from the fits to different lines , the local continuum level , and the atomic line data taken from vald .
the uncertainties for [ fe / h ] are not plotted to make the plots easier to read .
the average [ fe / h ] uncertainty is @xmath10.02dex . below we discuss the specific results for the selected elements : _ fe , z = 26_. differences up to 0.1dex between fe and fe abundances are due to the iterative nature of our computations .
previously determined abundances can change after adoption of the refined model atmosphere .
in other words , it reflects the limit of accuracy implied by the model atmosphere , atomic line data , and line list .
_ na , z = 11_. the distribution of the na abundance versus iron is shown in fig .
[ _ figure_samples ] .
the computed @xmath19 for the metal rich stars in our sample and those of @xcite and @xcite show a well defined positive slope ( table [ _ table_abundance_slopes ] ) .
the abundance distribution of na is shifted toward larger abundances by 0.10.2dex compared to fe .
the general trend is similar to those obtained in all comparison works with higher abundances toward higher metallicities ( fig .
[ _ figure_samples ] ) .
_ mg , z = 12_. only up to 7 lines of mg i were used in the analysis .
we found similar result to that of sodium , with a notable over - abundance for our sample , and a mean abundance of [ mg / fe][email protected] . the shift is larger than the formal accuracy of our abundance determination procedure , but
may depend on the adopted continuum level and line list .
we note that @xmath20 for the stars in our sample .
however , it can be seen in fig .
[ _ figure_samples ] that this could be due to border effects , as this trend is rather marginal .
@xcite show the same order of over - abundance , but also a certain downward trend with metallicity , and for the sample of @xcite we see that their mg distribution is in agreement with [ fe / h ] . at the same time , @xcite show the results similar to our .
al , z = 13_. in our work aluminium show a definite positive slope . to some degree our findings similar to those of @xcite and @xcite . but
different to @xcite and @xcite who show no definite dependence on metallicity , and higher mean [ al / fe ] for their samples . _
si , z = 14_. the results for silicon agree well with most of the studies in comparison , showing the average [ si / fe ] over - abundance to be on the order of 0.1dex and no noticeable trend with metallicity . on the other hand ,
@xcite show no excess of silicon .
it is interesting to note that for the same lines in the spectra , our results for vald-2 show 0.1dex lower abundances than for vald-3 , highlighting the importance of reliable atomic line data .
_ ca , z = 20_. calcium is another element for which we see a high level of agreement between the different authors @xcite . in all these works
there is a clear negative trend with metallicity ( table [ _ table_abundance_slopes ] ) . on the whole , [ ca / fe]@xmath70 , but in the super metal rich domain , we note calcium deficiency .
however , various mean abundances in different works ( see fig .
[ _ figure_samples ] ) reflect that this deficiency is within the accuracy limit of the modern methods of abundance determination .
_ ti , z = 22_. in our sample ti follows the iron abundance and shows no dependence with metallicity . on the other hand ,
all comparison works show a larger spread of stars toward higher [ ti / fe ] abundances and a weak negative trend . at the same time , @xcite show no metallicity dependence for ti . for our stars ,
the ti lines show a large scatter , which makes it difficult to come to any firm conclusions in regards to the metallicity trends .
similar to si and fe , the lines of ionized titanium give rise to a 0.1dex higher abundance when the line parameters are taken from vald-3 .
_ cr , z = 24_. chromium is one of the few elements with reliable results that show no significant scatter between the different lines and stars in our sample . we can observe a very weak negative trend with iron .
the same results are shown by @xcite .
@xcite and @xcite show higher abundances for their stars and a weak positive trend , which could be an indication of the sample bias effect .
once again , the same chromium lines give different abundances using the different versions of vald , but unlike si , ti , and fe , we see a 0.05dex lower mean abundance for the vald-3 data .
_ mn , z = 25_. manganese abundances show a clear positive gradient with metallicity , much like zn , though with a larger scatter despite the relatively large number of lines used in the analysis .
manganese exhibits a higher abundance than iron , on average , by 0.3dex .
a positive trend was also found in all comparison works . as for most of the other elements , the abundance measured by @xcite is slightly higher than that of @xcite . and
@xcite show the results similar to our ( fig .
[ _ figure_samples ] ) . _
ni , z = 28_. [ ni / fe ] also shows an upward trend with iron of the same order in all the samples in comparison .
all the authors show different average over - abundances for ni : up to 0.05dex for @xcite and @xcite , 0.1dex for @xcite and our work , and 0.15dex for @xcite and @xcite .
_ cu , z = 29_. the distribution of cu with [ fe / h ] exhibits a large scatter ( fig .
[ _ figure_overiron ] ) , mainly affected by the quality of the line list and number of lines . formally speaking , within the range of metallicities [ fe / h]=-0.1+0.2 , there are two groups of stars , with mean [ cu / fe ] abundances of @xmath8 + 0.1 and -0.05dex , respectively , however with uncertainties at the level of @xmath10.4dex . the lower average abundances and defined positive trend were found by @xcite . the same trend could be observed for the @xmath80.0dex group of stars in our sample .
_ zn , z = 30_. for zn we found the same significant differences in the abundance distribution with iron that were found for manganese .
unfortunately , we used only one line of zn in our analysis , so the visible slope of [ zn / fe ] versus [ fe / h ] should be considered as a preliminary result at this point .
more work should be done to confirm this result using a wider set of zn atomic lines . a large scatter and similar abundance
are also shown by @xcite and @xcite .
our analysis shows that the different atomic line data can have a strong impact on the average abundances measured for similar samples of stars ( e.g. see silicon in this work ) . on the other hand ,
observed trends change rather marginally .
comparisons to our work and those in the literature highlights the similar tendencies for the na , si , ca , cr , mn , and ni ; for the elements with a small spread like si , ca , cr , ni we see various average abundances between the different samples , and similar trends with iron in all comparison works . and for the mg , al , and ti we see weak trends with iron in all comparison works due to large spread of abundances . in the framework of our work , we investigated the dependence of our abundance results on the effective temperature and surface gravity .
the results of these tests are shown in figs .
[ _ figure_temperature_dependence ] and [ _ figure_gravity_dependence ] in appendix .
we see the evidence for a weak trend of abundances versus in fig .
[ _ figure_temperature_dependence ] .
these rather indistinct trends may be explained by uncertainties in the choice of the photometric of the metal rich stars of earlier spectral classes , as well as limitations of the current model atmospheres and atomic line data . in each case
we obtained the trends of abundances versus for the elements with rather low ionization potentials , i.e. mg , al and si , with the pearson product - moment correlation @xmath21()@xmath80.40.5 ( table [ _ table_abundance_slopes ] ) .
the same we also observe for mn and zn .
however , because our data does nt show a sufficient level of homoscedasticity , it is hard to make definite conclusions in these regards .
we note the non - linear behaviour of the [ x / h ] versus dependence for practically all our elements in the stars with close to 6000k .
this apparent phantom gap is poorly constrained due to low number of stars in this temperature range .
what is more important , we do not see any evident trends of the abundances versus the adopted surface gravities ( table [ _ table_abundance_slopes ] ) . in some sense
, this provides us with the evidence that our adopted procedure performs in the way we had previously envisaged it would .
however , sensitivity of the spectroscopic and to the quality of fits and line data ( these are discussed in section [ _ comparison ] ) emphasizes the importance of at least one independent variable .
there have been a number of extensive works to determine abundances of main sequence dwarfs .
they follow similar approaches to the analysis of the observed high - resolution stellar spectra : lte , 1d model atmospheres .
we tabulate these in table [ _ table_abundance_slopes_procedures ] .
to compare our results to other authors we note that our sample primarily consists of metal rich stars ; the samples of some other authors consist of a larger total number of targets , and they maintain a broader metallicity range .
most of our objects are single stars , where our spectroscopic data have been observed solely using harps , using a photometric pre - selection , and thus our sample is homogeneous . the most of comparison works were done using the equivalent width analysis , we used the synthetic profile fitting . in comparison to @xcite and @xcite , who utilized synthetic spectra fitting for the broad spectral ranges and fixed microturbulent velocity , we adopted line - by - line fitting and microturbulence as a free parameter in addition to the other differences .
as noted in subsection [ _ abundances ] , some authors used their own solar abundances for the [ x / fe ] distribution analysis , whereas we use the @xcite abundances in our study . in our work
, we separate the relative and absolute solar scales .
the first case is defined by the reference solar abundances to which the absolute values are translated . and
the second is dependent on the adopted line lists and gf - values .
the latter we identify as the differences in the atomic line data or line lists , but not in the adopted solar scales . | we report results from the high resolution spectral analysis of the 107 metal rich ( mostly [ fe / h]@xmath07.67dex ) target stars from the calan - hertfordshire extrasolar planet search program observed with harps . using our procedure of finding the best fit to the absorption line profiles in the observed spectra
, we measure the abundances of na , mg , al , si , ca , ti , cr , mn , fe , ni , cu , and zn , and we then compare them with known results from different authors .
most of our abundances agree with these works at the level of @xmath10.05dex or better for the stars we have in common .
however , we do find systematic differences that make direct inferences difficult .
our analysis suggests that the selection of line lists and atomic line data along with the adopted continuum level influence these differences the most . at the same time
, we confirm the positive trends of abundances versus metallicity for na , mn , ni , and to a lesser degree , al .
a slight negative trend is observed for ca , whereas si and cr tend to follow iron .
our analysis allows us to determine the positively skewed normal distribution of projected rotational velocities with a maximum peaking at 3 .
finally , we obtained a gaussian distribution of microturbulent velocities that has a maximum at 1.2and a full width at half maximum @xmath2=0.35 , indicating that metal rich dwarfs and subgiants in our sample have a very restricted range in microturbulent velocity .
[ firstpage ] stars : abundances fundamental parameters
late - type solar type |
in the context of hybrid medical imaging methods , a physical coupling between a high - contrast modality ( e.g. electrical impedance tomography , optical tomography ) and a high - resolution modality ( e.g. acoustic waves , magnetic resonance imaging ) is used in order to benefit from the advantages of both . without this coupling ,
the high - contrast modality , usually modeled by an inverse problem involving the reconstruction of the constitutive parameter of an elliptic pde from knowledge of boundary functionals , results in a mathematically severely ill - posed problem and suffers from poor resolution .
the analysis of this coupling usually involves a two - step inversion procedure where the high - resolution modality provides internal functionals , from which we reconstruct the parameters of the elliptic equation , thus leading to improved resolution @xcite .
a problem that has received a lot of attention recently concerns the reconstruction of the conductivity tensor @xmath0 in the elliptic equation @xmath1 from knowledge of internal power density measurements of the form @xmath2 , where @xmath3 and @xmath4 both solve with possibly different boundary conditions .
this problem is motivated by a coupling between electrical impedance imaging and ultrasound imaging and also finds applications in thermo - acoustic imaging .
explicit reconstruction procedures for the above non - linear problem have been established in @xcite , successively in the 2d , 3d , and @xmath5d isotropic case , and then in the 2d and @xmath5d anisotropic case . in these articles , the number of functionals may be quite large .
the analyses in @xcite were recently summarized and pushed further in @xcite . if one decomposes @xmath0 into the product of a scalar function @xmath6 and a scaled anisotropic structure @xmath7 such that @xmath8 , the latter reference establishes explicit reconstruction formulas for both quantities with lipschitz stability for @xmath9 in @xmath10 , and involving the loss of one derivative for @xmath7 . in the isotropic case ,
several works study the above problem in the presence of a lesser number of functionals .
the case of one functional is addressed in @xcite , whereas numerical simulations show good results with two functionals in dimension @xmath11 @xcite .
theoretical and numerical analyses of the linearized inverse problem are considered in @xcite .
the stabilizing nature of a class of internal functionals containing the power densities is demonstrated in @xcite using a micro - local analysis of the linearized inverse problem .
the above inverse problem is recast as a system of nonlinear partial differential equations in @xcite and its linearization analyzed by means of theories of elliptic systems of equations .
it is shown in the latter reference that @xmath12 functionals , where @xmath5 is spatial dimension , is sufficient to reconstruct a scalar coefficient @xmath0 with elliptic regularity , i.e. , with no loss of derivatives , from power density measurements .
this was confirmed by two - dimensional simulations in @xcite .
all known explicit reconstruction procedures require knowledge of a larger number of internal functionals . in the present work , we study the linearized version of this inverse problem in the anisotropic case , i.e. we write an expansion of the form @xmath13 with @xmath14 known and @xmath15 , and study the reconstructibility of @xmath0 from linearized power densities ( lpd ) .
we first proceed by supporting the perturbation @xmath0 away from the boundary @xmath16 and analyze microlocally the symbol of the linearized functionals , and show that , as in @xcite , a large enough number of functionals allows us to construct a left - parametrix and set up a fredholm inversion .
the main difference between the isotropic and anisotropic settings is that the anisotropic part of the conductivity is reconstructed with a loss of one derivative .
such a loss of a derivative is optimal since our estimates are elliptic in nature .
it is reminiscent of results obtained for a similar problem in @xcite .
secondly , we show how the explicit inversion approach presented in @xcite carries through linearization , thus allowing for reconstruction of fully anisotropic tensors supported up to the boundary of @xmath17 . in this case
, we derive reconstruction formulas that require a smaller number of power densities than in the non - linear case , giving possible room for improvement in the non - linear inversion algorithms . for additional information on hybrid inverse problems in other areas of ( mostly medical ) imaging
, we refer the reader to , e.g. , @xcite .
consider the conductivity equation , where @xmath18 is open , bounded and connected with @xmath19 , and where @xmath20 is a uniformly elliptic conductivity tensor over @xmath17 .
we set boundary conditions @xmath21 and call @xmath22 the unique solution to with @xmath23 , @xmath24 and conductivity @xmath20 .
we consider the measurement functionals @xmath25 considering an expansion of the form @xmath26 , where the background conductivity @xmath14 is known , uniformly elliptic and @xmath27 so small that the total @xmath20 remains uniformly elliptic , we first look for the frchet derivative of with respect to @xmath0 at @xmath14 . expanding the solutions @xmath22 accordingly as @xmath28 the pde at orders @xmath29 and @xmath30 gives rise to two relations @xmath31
the measurements then look like @xmath32 therefore , the component @xmath33 of the frchet derivative of @xmath34 at @xmath14 is @xmath35 where the @xmath36 s are linear functions in @xmath0 according to . in both subsequent approaches ,
reconstruction formulas are established under the following two assumptions about the behavior of solutions related to the conductivity of reference @xmath14 .
the first hypothesis deals with having a basis of gradients of solutions of over a certain subset @xmath37 .
[ hyp : det ] for an open set @xmath37 , there exist @xmath38 such that the corresponding solutions @xmath39 of with boundary condition @xmath40 ( @xmath41 ) satisfy @xmath42 once hypothesis [ hyp : det ] is satisfied , any additional solution @xmath43 of gives rise to a @xmath44 matrix @xmath45 , { \quad\text { where } } \quad z_i:= \nabla \frac{\det ( \nabla u_1,\dots,\overbrace{\nabla u_{n+1}}^i , \dots , \nabla u_n)}{\det ( \nabla u_1,\dots,\nabla u_n)}. \label{eq : zmat}\end{aligned}\ ] ] as seen in @xcite , such matrices can be computed from the power densities @xmath46 and help impose orthogonality conditions on the anisotropic part of @xmath14 .
once enough such conditions are obtained by considering enough additional solutions , then the anisotropy is reconstructed explicitly via a generalization of the usual cross - product defined in three dimensions . in the linearized
setting , we find that _
one _ additional solution such that @xmath47 has full rank is enough to reconstruct the linear perturbation @xmath0 .
we thus formulate our second crucial assumption here : [ hyp : z ] assume that hypothesis [ hyp : det ] holds over some fixed @xmath37 .
there exists @xmath48 such that the solution @xmath43 of with boundary condition @xmath49 has a full - rank matrix @xmath47 ( as defined in ) over @xmath50 .
[ rem : const ] in the case where @xmath14 is constant , then it is straightforward to see that @xmath51 ( @xmath41 ) fulfill hypothesis [ hyp : det ] over @xmath17 .
moreover , if @xmath52 denotes an invertible constant matrix such that @xmath53 , then the boundary condition @xmath54 fulfills hypothesis [ hyp : z ] , since we have @xmath55 . throughout the paper , we use for ( real - valued ) square matrices @xmath56 and @xmath57 the contraction notation @xmath58 , with @xmath59 the transpose matrix of @xmath56 . in the treatment of the non - linear case @xcite ,
it has been pointed out that hypothesis [ hyp : det ] may not be systematically satisfied globally in dimension @xmath60 . a more general hypothesis to consider
would come from picking a larger family ( of cardinality @xmath61 ) of solutions whose gradients have maximal rank throughout @xmath17 . while this additional technical point would not alter qualitatively the present reconstruction algorithms
, it would add complexity in notation which the authors decided to avoid ; see also @xcite . in the reconstruction approach developped in @xcite for the non - linear problem
, it was shown that not every part of the conductivity was reconstructed with the same stability .
namely , consider the decomposition of the tensor @xmath62 into the product of a scalar function @xmath63 and a scaled anisotropic structure @xmath64 with @xmath65 .
the following results were then established .
starting from @xmath5 solutions whose gradients form a basis of @xmath66 over a subset @xmath67 , it was shown that under knowledge of a @xmath68 anisotropic structure @xmath64 , the scalar function @xmath69 was uniquely and lipschitz - stably reconstructible in @xmath70 from @xmath10 power densities .
additionally , if one added a finite number of solutions @xmath71 such that the family of matrices @xmath72 defined as in imposed enough orthogonality constraints on @xmath64 , then the latter was explicitely reconstructible over @xmath50 from the mutual power densities of @xmath73 .
the latter reconstruction was stable in @xmath74 for power densities in @xmath10 norm , thus it involved the loss of one derivative . passing to the linearized setting now ( recall @xmath13 ) , and anticipating
that one scalar quantity may be more stably reconstructible than the others , this quantity should be the linearized version of @xmath75 .
standard calculations yield @xmath76 and thus the quantity that should be stably reconstructible is @xmath77 .
the linearization of the product decomposition @xmath78 above is now a spherical - deviatoric one of the form @xmath79 where @xmath80 is the linear projection onto the hyperplane of traceless matrices @xmath81 .
the above inverse problem in - may be seen as a system of partial differential equations for @xmath82 .
this is the point of view considered in @xcite .
however , @xmath83 may be calculated from and the expression plugged back into .
this allows us to recast @xmath84 as a linear operator for @xmath0 , which is smaller than the original linear system for @xmath82 , but which is no longer differential and rather pseudo - differential .
the objective in this section is to show , following earlier work in the isotropic case in @xcite , that such an operator is elliptic under appropriate conditions .
we first fix @xmath85 and assume that @xmath86 , so that integrals of the form @xmath87 are well - defined , with @xmath88 a matrix - valued symbol whose entries are polynomials in @xmath89 ( see @xcite ) and where the hat denotes the fourier transform @xmath90 .
we also assume that @xmath91 and can be extended smoothly by @xmath92 outside @xmath93 . as pointed out in @xcite , in order to treat this problem microlocally
, one must introduce cutoff versions of the @xmath33 operators , which in turn extend to pseudo - differential operators ( @xmath94do ) on @xmath66 .
namely , if @xmath95 is a domain satisfying @xmath96 and @xmath97 is a smooth function supported in @xmath17 which is identically equal to @xmath98 on a neighborhood of @xmath99 , the operator @xmath100 can be made a @xmath94do upon considering @xmath101 as a second - order operator on @xmath66 and using standard pseudo - differential parametrices to invert it @xcite .
we will therefore not distinguish the operators @xmath33 from their pseudo - differential counterparts .
the task of this section is then to determine conditions under which a given collection of such functions becomes an elliptic operator of @xmath0 over @xmath93 .
using relations and , we aim at writing the operator @xmath102 in the following form @xmath103 with symbol @xmath104 ( pseudo - differential terminology is recalled in sec .
[ ssec : prelim ] ) .
we first compute the main terms in the symbol expansion of @xmath33 ( call this expansion @xmath105 with @xmath106 homogeneous of degree @xmath107 in @xmath89 ) . from these expressions
, we then directly deduce microlocal properties on the corresponding operators .
the first lemma shows that the principal symbols @xmath108 can never fully invert for @xmath0 , no matter how many solutions @xmath109 we pick .
when hypothesis [ hyp : det ] is satisfied , then the characteristic directions of the principal symbols @xmath110 reduce to a @xmath111-dimensional subspace of @xmath112 . here and below
, we recall that the colon `` @xmath113 '' denotes the inner product @xmath114 for @xmath115 and @xmath116 denotes the symmetric outer product @xmath117 for @xmath118 .
[ lem : xieta ] * for any @xmath119 and @xmath120 , the symbol @xmath108 satisfies @xmath121 * suppose that hypothesis [ hyp : det ] holds over some @xmath37 .
then for any @xmath122 , if @xmath123 is such that @xmath124 then @xmath125 is of the form @xmath126 for some vector @xmath127 satisfying @xmath128 .
since an arbitrary number of zero - th order symbols can never be elliptic with respect to @xmath0 , we then consider the next term in the symbol expansion of @xmath33 .
we must also add one solution @xmath43 to the initial collection , exhibiting appropriate behavior , i.e. satisfying hypothesis [ hyp : z ] .
the collection of functionals we consider below is thus of the form @xmath129 and emanates from @xmath12 solutions @xmath130 of satisfying hypotheses [ hyp : det ] and [ hyp : z ] . in order to formulate the result
, we assume to construct a family of unit vector fields @xmath131 homogeneous of degree zero in @xmath89 , smooth in @xmath132 and everywhere orthonormal .
we then define the family of scalar elliptic zeroth - order @xmath94do @xmath133 which can be thought of as a microlocal change of basis after which the operator @xmath134 becomes both diagonal and elliptic . indeed
, we verify ( see section [ ssec : proofs ] ) that for any @xmath135 and @xmath0 sufficiently regular , we have @xmath136 the above estimates come from standard result on pseudo - differential operators @xcite .
the presence of the constant @xmath137 indicates that @xmath138 can be inverted microlocally , but may not injective .
composing the measurements @xmath33 with appropriate scalar @xmath94do of order 0 and 1 , we are then able to recover each component of the operator .
the well - chosen `` parametrices '' are made possible by the fact that the collection of symbols @xmath139 becomes elliptic over @xmath93 when hypotheses [ hyp : det ] and [ hyp : z ] are satisfied .
rather than using the full collection of measurements @xmath140 , we will consider the smaller collection @xmath141 augmented with the @xmath5 measurement operators @xmath142 where @xmath143 , known from the measurements @xmath144 , are the coefficients in the relation of linear dependence @xmath145 we also define the operator @xmath146 with principal symbol @xmath147 .
our conclusions may be formulated as follows : [ prop : microloc ] let the measurements @xmath140 defined in satisfy hypotheses [ hyp : det ] and [ hyp : z ] . * for @xmath148 and @xmath149 ,
there exist @xmath150 such that @xmath151 * for any @xmath152 , there exist @xmath153 such that the following relation holds @xmath154 where the remainder @xmath155 can be expressed as a zeroth - order linear combination of the components @xmath156 and @xmath157 reconstructed in ( i ) . the presence of the @xmath158 term in part _
( ii ) _ of prop . [ prop : microloc ] accounts for the loss of one derivative in the inversion process . from prop .
[ prop : microloc ] , we can then obtain stability estimates of the form @xmath159 the above stability estimate holds for @xmath160 using the results of proposition [ prop : microloc ] and in fact for any @xmath161 updating by standard methods ( not detailed here @xcite ) the parametrices in and to inversions modulo operators in @xmath162 ( i.e. , classical @xmath94do of order @xmath163 @xcite ) provided that the coefficients @xmath164 are sufficiently smooth .
the presence of the constant @xmath137 indicates that the reconstruction of @xmath0 may be performed up to the existence of a finite dimensional kernel as an application of the fredholm theory as in @xcite .
equation means that some components of @xmath0 are reconstructed with a loss of one derivative while other components are reconstructed with no loss .
the latter components are those that can be spanned by the components @xmath165 and @xmath166 .
some algebra shows that the only such linear combination is @xmath167 , which , using the fact that @xmath168 , can be computed as @xmath169 confirming the heuristics of sec .
[ sec : heur ] .
it can be shown that all other components of @xmath0 ( i.e. any part of @xmath170 in ) are , to some extent , spanned by the components @xmath171 , and as such can not be reconstructed with better stability than the loss of one derivative in light of . combining the above results with
, we arrive at the main stability result of the paper : @xmath172 such an estimate holds for any @xmath135 .
the above estimate holds with @xmath173 when @xmath174 is an injective ( linear ) operator .
injectivity can not be verified by microlocal arguments since all inversions are performed up to smoothing operators ; see @xcite in the isotropic setting . in the next section , we obtain an injectivity result , which allows us to set @xmath173 in the above expression . however , the above stability estimate is essentially optimal .
an optimal estimate , which follows from the above and the equations for @xmath82 is the following : @xmath175 the left - hand - side inequality is a direct consequence of and the expression of @xmath84 .
the right - hand side is a direct consequence of the expression of @xmath84 .
the above estimate is clearly optimal .
the operator @xmath176 is of order @xmath177 .
if it were elliptic , then @xmath0 would be reconstructed with no loss of derivative .
however , @xmath176 is not elliptic and the loss of ellipticity is precisely accounted for by the results in lemma [ lem : xieta ] . as we discussed above , it turns out that the only spatial coefficient controlled by @xmath178 is @xmath179 , and hence .
now , allowing @xmath0 to be supported up to the boundary , we present a variation of the non - linear resolution technique used in @xcite . first considering @xmath5 solutions generated by boundary conditions fulfilling hypothesis [ hyp : det ]
, we establish an expression for @xmath0 in terms of the remaining unknowns @xmath180 : @xmath181 h^{-1 } { dh } h^{-1 } [ \nabla u]^t - [ \nabla v ] h^{-1 } [ \nabla u]^t - [ \nabla u ] h^{-1 } [ \nabla v]^t ) \gamma_0 , \label{eq : gammaelim}\end{aligned}\ ] ] where @xmath182 $ ] and @xmath183 $ ] denote @xmath44 matrices whose @xmath184-th columns are @xmath185 and @xmath186 , respectively , and where @xmath187 and @xmath188 . in particular we find from the relation @xmath189[\nabla u]^{-1})^t .
\label{eq : trgamma}\end{aligned}\ ] ] plugging back into the second equation in for @xmath41 , one can deduce a gradient equation for the quantity @xmath179 which in turn allows to reconstruct @xmath77 in a lipschitz - stable manner with respect to the lpd @xmath190 ( i.e. without loss of derivative ) . now turning to the full reconstruction of @xmath0
, we consider an additional solution @xmath43 generated by a boundary condition fulfilling hyp .
[ hyp : z ] .
the following proposition then establishes how to reconstruct @xmath180 from @xmath140 : [ prop : scesv ] assume that @xmath191 fulfill hypotheses [ hyp : det ] and [ hyp : z ] over @xmath17 and consider the linearized power densities @xmath192 .
then the solutions @xmath180 satisfy a strongly coupled elliptic system of the form @xmath193 where the vector fields @xmath194 are known and only depend on the behavior of @xmath14 , @xmath47 and @xmath195 , and where the functionals @xmath196 are linear in the data @xmath33 .
when the vector fields @xmath194 are bounded , system satisfies a fredholm alternative from which we deduce that if with a trivial right - hand side admits no non - trivial solution , then @xmath180 is uniquely reconstructed from .
we can then reconstruct @xmath0 from . in the case where @xmath14 is constant
, choosing solutions as in remark [ rem : const ] , one arrives at a system of the form where @xmath197 if @xmath198 , so that the system is decoupled and clearly injective . the conclusive theorem for the explicit inversion is thus given by [ thm : explicit ] assume that @xmath191 fulfill hypotheses [ hyp : det ] and [ hyp : z ] over @xmath17 and consider the linearized power densities @xmath192 .
assume further that the system with trivial right - hand sides has no non - trivial solution .
then @xmath0 is uniquely determined by @xmath140 and we have the following stability estimate @xmath199 we cover the microlocal inversion in sec . [
sec : microloc ] .
linear algebraic and pseudo - differential preliminaries are given in sec .
[ ssec : prelim ] .
the leading - order symbols of order @xmath177 of the lpd functionals are computed in sec .
[ ssec : symbol0 ] and a proof of lemma [ lem : xieta ] is given .
the symbols of order @xmath200 are then computed in [ ssec : symbolm1 ] and the proof proposition [ prop : microloc ] is given in sec .
[ ssec : proofs ] .
we then treat the explicit inversion in sec .
[ sec : explicit ] .
starting with some preliminaries in sec .
[ ssec : prelim2 ] , we derive some crucial relations in sections [ ssec : der1 ] and [ ssec : der2 ] , before proving proposition [ prop : scesv ] and theorem [ thm : explicit ] in sec .
[ ssec : proofs2 ] .
[ [ linear - algebra . ] ] linear algebra .
+ + + + + + + + + + + + + + + in the following , we consider the @xmath44 matrices @xmath201 with the inner product structure @xmath202 for which @xmath201 admits the orthogonal decomposition @xmath203 . for two vectors @xmath204 and @xmath205 in @xmath66
we denote by @xmath206 the matrix with entries @xmath207 , and we also define the symmetrized outer product @xmath208 with @xmath209 denoting the standard dotproduct on @xmath66 , we have the following identities @xmath210 [ [ pseudo - differential - calculus . ] ] pseudo - differential calculus .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + recall that we denote the set of _ symbols _ of order @xmath211 on @xmath17 by @xmath212 , which is the space of functions @xmath213 such that for all multi - indices @xmath214 and @xmath215 and every compact set @xmath216 there is a constant @xmath217 such that @xmath218 we denote the operator @xmath219 as @xmath220 and the set of _ pseudo - differential operators _ ( @xmath94do ) of order @xmath211 on @xmath17 by @xmath221 , where @xmath222 suppose @xmath223 is strictly decreasing and @xmath224 , and suppose @xmath225 for each @xmath184 .
we denote an _ asymptotic expansion _ of the symbol @xmath226 as @xmath227 if @xmath228 given two @xmath94do @xmath125 and @xmath229 with respective symbols @xmath230 and @xmath231 and orders @xmath232 and @xmath233 , we will make repetitive use of the symbol expansion of the product operator @xmath234 ( see ( * ? ? ?
* theorem ( 8.37 ) ) for instance ) @xmath235 where @xmath236 denotes a symbol of order at most @xmath214 . as we will need to compute products of three @xmath94do @xmath237 , @xmath125 and @xmath229 , we write the following formula for later use , obtained by iteration of @xmath238 in the next derivations , some operators have matrix - valued principal symbols
. however we will only compose them with operators with scalar symbols , so that the above calculus remains valid .
writing @xmath239 and @xmath240 ( understood in the componentwise sense ) , we have @xmath241 thus equation reads @xmath242 , where the operators @xmath243 and @xmath244 have respective symbols @xmath245 for @xmath246 a smooth vector field , we will also need in the sequel to express the operator @xmath247 as @xmath94do , the symbol of which is denoted @xmath248 .
we now write @xmath33 as a @xmath94do of @xmath0 with symbol @xmath249 as in .
@xmath33 belongs to @xmath250 and we will compute in this paper the first two terms in the expansion of @xmath249 ( call them @xmath108 and @xmath251 ) , which in turn relies on constructing parametrices of @xmath252 of increasing order and doing some computations on symbols of products of @xmath94do based on formula . if @xmath229 is a parametrix of @xmath252 modulo @xmath253 , i.e. @xmath254 , then straightforward computations based on the relation @xmath242 yield the following relation @xmath255 for any @xmath256 , @xmath257 denotes the operator @xmath258 where @xmath36 solves , and standard elliptic theory allows to claim that @xmath257 smoothes by one derivative so that the error operator @xmath259 defined in smoothes by @xmath211 derivatives .
in particular , upon computing a parametrix @xmath229 of @xmath252 modulo @xmath253 , the first three terms in are enough to construct the principal part of the symbol @xmath249 modulo @xmath253 .
[ [ computation - of - m_ij_0 . ] ] computation of @xmath108 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in light of the last remark , we first compute a parametrix @xmath229 of @xmath252 modulo @xmath260 , that is , since @xmath261 , we look for a principal symbol of the form @xmath262 .
clearly , we easily obtain @xmath263 . in this case , the principal symbol of @xmath33 at order zero is given by , according to and , @xmath264 @xmath108 admits a somewhat more symmetric expression if pre- and post - multiplied by @xmath265 , the unique positive squareroot of @xmath14 , so that we may write , @xmath266 where we have defined @xmath267 and @xmath268 for any @xmath269 as well as @xmath270 .
this last expression motivates the proof of lemma [ lem : xieta ] .
* proof of _
( i ) _ : * let @xmath127 such that @xmath271 , and denote @xmath272 so that @xmath273
. then using identity and , we get @xmath274:\hxi_0\odot\eta ' \\ & = ( v_i\cdot \hxi_0 ) ( v_j\cdot \eta ' ) + ( v_i\cdot\eta')(v_j\cdot \hxi_0 ) \\ & \quad - ( \hxi_0\cdot v_i ) ( \hxi_0\cdot\hxi_0 ) ( v_j\cdot \eta ' ) - ( \hxi_0\cdot v_i)(\hxi_0\cdot\eta')(v_j\cdot\hxi_0 ) \\ & \quad - ( \hxi_0\cdot v_j ) ( \hxi_0\cdot \hxi_0 ) ( v_i\cdot\eta ' ) - ( \hxi_0\cdot v_j ) ( \hxi_0\cdot\eta ' ) ( \hxi_0\cdot v_i)\\ & = 0 , \end{aligned}\ ] ] where we have used @xmath275 and @xmath276 , thus _ ( i ) _ holds .
* proof of _ ( ii ) _ : * recall that @xmath277 : a_0^{-1 } p a_0^{-1}. \end{aligned}\ ] ] we write @xmath112 as the direct orthogonal sum of three spaces : @xmath278 with respective dimensions @xmath98 , @xmath279 and @xmath111 . decomposing @xmath280 uniquely into this sum , we write @xmath281 .
direct calculations then show that @xmath282 since @xmath283 is a basis of @xmath66 , @xmath284 is a basis of @xmath112 and thus implies that @xmath285 therefore @xmath286 with @xmath287 for some @xmath288 , so @xmath289 with @xmath127 proportional to @xmath290 , i.e. such that @xmath128 , thus the proof is complete . in other words , *
all * symbols of order zero @xmath291 are orthogonal to the @xmath292-dependent @xmath111-dimensional subspace of symmetric matrices @xmath293 .
one must thus compute the next term in the symbol exampansion of the operators @xmath33 , i.e. @xmath251 .
we will then show that enough symbols of the form @xmath139 will suffice to span the entire space @xmath112 for every @xmath294 and @xmath295 , so that the corresponding family of operators is elliptic as a function of @xmath0 .
as the previous section explained , the principal symbols @xmath108 can never span @xmath112 .
therefore , we compute the next term @xmath251 in their symbol expansion . we must first construct a parametrix @xmath229 of @xmath252 modulo @xmath296 , i.e. of the form @xmath297 [ lem : qm2m3 ] the symbols @xmath298 and @xmath299 defined in have respective expressions @xmath300_{pq } \partial_{x_i}[\gamma_0]_{ij } - 2 [ \gamma_{0}]_{ij } \partial_{x_i } [ \gamma_0]_{pq}\right ) .
\label{eq : qm3 } \end{aligned}\ ] ] using formula with @xmath301 , and using the expansions of @xmath231 and @xmath302 , we get @xmath303 in order to match the expansion @xmath304 , the expansion above must satisfy , for large @xmath89 , @xmath305 that is , @xmath263 and @xmath306 now , we easily have @xmath307 and @xmath308_{pq } \xi_p \xi_q { { \bf e}}_i$ ] , where @xmath309 is the natural basis of @xmath66 .
we thus deduce the expression of @xmath310 @xmath311_{pq}\xi_p\xi_q \partial_{x_i } [ \gamma_0]_{ij}\xi_j - 2[\gamma_0]_{ij}\xi_j \partial_{x_i } [ \gamma_0]_{pq}\xi_p\xi_q \right),\end{aligned}\ ] ] from which holds .
@xmath299 is clearly in @xmath312 from this expression , since @xmath313 is of order @xmath314 .
the proof is complete we now give the expression of @xmath251 ( or rather , that of @xmath315 ) .
[ prop : mijm1 ] the symbol @xmath315 admits the following expression for any @xmath316 @xmath317 where we have defined @xmath318 , @xmath319 , as well as the vector field @xmath320 assume @xmath229 is a parametrix of @xmath252 modulo @xmath296 and consider formula .
since the term @xmath321 is of order zero , the computation of @xmath251 consists in computing the second term in the symbol expansion of @xmath322 , and the same term with @xmath119 permuted , where we denote @xmath323 with symbol @xmath324 .
plugging @xmath325 , @xmath326 and @xmath327 into and keeping only the terms that are homogeneous of degree @xmath200 in @xmath89 , we arrive at the expression @xmath328 note that the multiplications commute because the symbols of @xmath229 and @xmath329 are scalar , while that of @xmath330 is matrix - valued . since @xmath331 , equation will be proved when we show that @xmath332 * proof of . * starting from , plugging the expression @xmath333 , using the identity @xmath334 and pre- and post - multiplying by @xmath265 yields the relation @xmath335 gathering the first and third terms recombines into @xmath336 ( the last term of ) . on to the second and fourth terms , we first compute @xmath337 using this calculation , the second and fourth terms in recombine into @xmath338 thus the argument is complete .
* preliminaries : * by virtue of hypothesis [ hyp : det ] , @xmath339 may be decomposed into the basis @xmath340 by means of scalars @xmath341 such that @xmath342 as seen in @xcite , the coefficients @xmath343 are directly computible from the power densities @xmath344 and on the other hand , we have the relation @xmath345 thus @xmath346 as defined in for @xmath41 . in the next proofs
, we will use the following [ lem : mmat ] under hypotheses [ hyp : det ] and [ hyp : z ] , the following matrix - valued function @xmath347 is symmetric and uniformly invertible .
symmetry of @xmath348 is obvious by definition .
taking gradient of , we arrive at @xmath349 pre- and post - multiplying by @xmath265 , we deduce that @xmath350 where @xmath351 $ ] .
the proof is complete since hyp .
[ hyp : det ] ensures that @xmath352 never vanishes and @xmath353 is uniformly invertible , and hyp .
[ hyp : z ] ensures that @xmath47 is uniformly invertible . *
the @xmath354 operators .
proof of prop .
[ prop : microloc ] : * as advertised in sec .
[ ssec : statmicroloc ] , because of the algebraic form of the symbols of the linearized power density operators , it is convenient for inversion purposes to define the microlocal change of basis @xmath355 as in , i.e. @xmath356 to convince ourselves that this collection forms a microlocally invertible operator of @xmath0 , let us introduce the zero - th order @xmath94dos @xmath357 with scalar principal symbol @xmath358 for @xmath359 .
then for any @xmath360 , the composition of operators @xmath361 has principal symbol ( repeated indices are summed over ) @xmath362 where we have used the following property , true for any smooth vector field @xmath363 : @xmath364 thus for any @xmath360 , the composition @xmath361 recovers @xmath365 up to a regularization term .
this in particular justifies the estimates and the subsequent inversion procedure .
we are now ready to prove proposition [ prop : microloc ] .
from the fact that @xmath366 is a basis at every point and given their dotproducts @xmath367 , we have the following formula , true for every vector field @xmath368 : @xmath369 * proof of _ ( i ) _ : reconstruction of the components @xmath165 and @xmath370 .
* we work with @xmath371 . using with @xmath372 ,
straightforward computations yield @xmath373 which means that upon defining @xmath374 with scalar principal symbols @xmath375 relation is satisfied in the sense of operators since the previous calculation amounts to computing the principal symbol of the composition of operators in .
+ it remains to construct appropriate operators that will map @xmath376 to the components @xmath171 for @xmath152 , which is where the additional measurements @xmath377 come into play .
let @xmath378 as in and construct the @xmath94do @xmath379 as in .
it is easy to see that , since the @xmath380 are only functions of @xmath132 , the terms of fixed homogeneity in the symbol expansion of @xmath381 satisfy @xmath382 then from equation and relation , we deduce that @xmath383 , so that @xmath384 .
moreover , using equation together with relation , we deduce that @xmath385 is now the principal symbol of @xmath381 . using relation with @xmath386 , the symmetry of @xmath348 and multiplying by @xmath348
, we have the relation @xmath387 using this relation , we deduce the following calculation , for @xmath152 @xmath388 while the second term gives us the missing components @xmath171 , we claim that the first one is spanned by @xmath389 and @xmath390 .
indeed we have @xmath391 so we deduce that @xmath392 in light of these algebraic calculations , we now build the parametrices .
let @xmath393 , @xmath394 , @xmath395 , @xmath396 and @xmath397 the @xmath94dos with respective principal symbols @xmath398 then the relation implies at the principal symbol level .
the operator @xmath237 can indeed be expressed as the following zero - th order linear combination of the components @xmath156 and @xmath399 : @xmath400 so that the left - hand side of is expressed as a post - processing of measurement operators @xmath33 only .
the proof is complete .
for a matrix @xmath56 with columns @xmath401 and @xmath402 the canonical basis , one has the following representation @xmath403 more generally , for two matrices @xmath404 $ ] and @xmath405 $ ] , we have the relation @xmath406 finally , for @xmath56 a matrix and @xmath407 $ ] , the sum @xmath408 is nothing but the @xmath256-th column of the matrix @xmath409 .
let us start from @xmath5 solutions @xmath39 fulfilling hypothesis [ hyp : det ] , and let @xmath180 the corresponding solutions of .
we also denote @xmath182 : = [ \nabla u_1|\dots|\nabla u_n]$ ] and @xmath183 $ ] similarly .
we first mention that for any vector field @xmath363 , we have the following formulas @xmath410 which also amounts to the following matrix relations @xmath411 from the relation @xmath412 we deduce , using , @xmath413 the previous equation allows us express @xmath0 in terms of the remaining unknowns @xmath180 . indeed , taking the tensor product of with @xmath414 and summing over @xmath256 yields @xmath415 where we have used the identity in the last right - hand side .
we may rewrite this as @xmath416 one may notice that the above expression is indeed a symmetric matrix .
in matrix notation , using the preliminaries , we arrive at the expression .
let us now add another solution @xmath43 with corresponding solution @xmath417 at order @xmath418 . by virtue of hypothesis [ hyp : det ] , as in section [ ssec : proofs ] , @xmath339 may be expressed in the basis @xmath419 as @xmath420 where the coefficients @xmath380 can be expressed as ratios of determinants , or equivalently , computable from the power densities at order @xmath421 , see ( * ? ? ?
* appendix a.3 ) . for @xmath41 , we define @xmath422 , and notice that we have the following two algebraic relations @xmath423 the first one is obtained obtained after applying the operator @xmath424 to and the second one is obtained after applying an exterior derivative to . moving on to the study of the corresponding @xmath417 solution
, we write @xmath425 where we have cancelled sums of the form . using the identity
, we deduce that @xmath426 taking exterior derivative of the previous relation yields @xmath427 we now apply @xmath424 to , the left - hand side becomes @xmath428 thus we arrive at the equation @xmath429 where the vector fields @xmath430 are known from the data @xmath140 . combining the latter equation with
, we obtain @xmath431 which we recast as @xmath432 the left - hand side can be considerably simplified by noticing that the second equation of implies @xmath182z^t = z[\nabla u]^t$ ] . with this fact in mind , the left - hand side looks like @xmath433 , where we compute @xmath434z^t \gamma_0 [ \nabla u ] h^{-1 } { { \bf e}}_p \\ & = \gamma_0 z [ \nabla u]^t [ \nabla u]^{-t } { { \bf e}}_p = \gamma_0 z_p.\end{aligned}\ ] ] in equations and , the only unknown is the matrix @xmath183:= [ \nabla v_1,\dots,\nabla v_n]$ ] .
equations and give us the projection of that matrix onto the space @xmath437 and onto the line @xmath438 respectively .
as in the non - linear case @xcite , we expect that a rich enough set of such equations provided by a certain number of additional solutions @xmath439 leads to a pointwise , algebraic reconstruction of @xmath183 $ ] , however we do not follow that route here .
we now show that provided that we use _ one _ additional solution @xmath43 ( on top of the basis @xmath39 ) such that the matrix @xmath47 is of full rank , then we can reconstruct @xmath180 via a strongly coupled elliptic system of the form , after which we can reconstruct @xmath0 from @xmath440 by formula .
we now show how to derive this elliptic system . according to hypothesis [ hyp : z ] , the matrix @xmath441 $ ] has full rank and we recall the important equations @xmath442 and where @xmath443 is given in . assuming that @xmath47 has full rank , the family @xmath444 is a frame with dotproducts defined as @xmath445 , and in this case we define its dual frame @xmath446 for @xmath41 , such that @xmath447 , i.e. with @xmath448 the matrix with columns @xmath449 , we have the relation @xmath450 .
the second equation of may be rewritten as @xmath451 applying the differential operator @xmath452 to the first equation of , we obtain @xmath453 using , we may rewrite the left - hand side of as @xmath454\cdot\nabla v_p + ( \gamma_0 z_p\cdot\nabla ) ( z_i^\star\cdot\nabla ) v_p \\ & = [ z_i^\star , \gamma_0 z_p]\cdot\nabla v_p + ( \gamma_0 z_p\cdot\nabla ) ( z_p^\star\cdot\nabla ) v_i \dots\\ & \qquad+ ( \gamma_0 z_p\cdot\nabla ) ( \omega ( z_p^\star , z_i^\star ) ) , \end{aligned}\ ] ] where we have introduced the lie bracket of two vector fields , which may be written in the euclidean connection @xmath455 : = ( x\cdot\nabla ) y - ( y\cdot\nabla ) x. \label{eq : lie } \end{aligned}\ ] ] plugging the last calculation into ( repeated indices are summed over ) @xmath456\cdot\nabla
v_p = ( z_i^\star\cdot\nabla ) f - ( \gamma_0 z_p\cdot\nabla ) ( \omega ( z_p^\star , z_i^\star ) ) .
\label{eq : tmpsces2 } \end{aligned}\ ] ] we now look more closely at the principal part of this equation .
the first term may be written as @xmath457 where we have used that @xmath458 .
we thus obtain a strongly coupled elliptic system of the form , where @xmath459 , \quad 1\le i , j\le n , \label{eq : wij } \\ f_i & : = - z_i^\star\cdot\nabla f + ( \gamma_0 z_p\cdot\nabla ) ( \omega ( z_p^\star , z_i^\star ) ) , \quad 1\le i\le n. \label{eq : fi } \end{aligned}\ ] ] this concludes the proof . in order to assess the properties of system
, we recast it as an integral equation as follows : let us call @xmath460 , and define @xmath461 , where @xmath3 is the unique solution to the equation @xmath462 by the lax - milgram theorem ( see e.g. @xcite ) , one can establish that such solutions satisfy an estimate of the form @xmath463 , where @xmath464 only depends on @xmath17 and the constant of ellipticity of @xmath14 , thus @xmath465 is continuous , and by rellich imbedding ( i.e. the fact that the injection @xmath466 is compact ) , @xmath467 is compact .
applying the operator @xmath468 to , we arrive at the integral system @xmath469 where it is easy to establish that for @xmath470 , the operator @xmath471 is compact whenever the vector fields @xmath194 are bounded . in vector notation , if we define the vector space @xmath472 , @xmath473 , @xmath474 and for @xmath475 , @xmath476 we have that @xmath477 is a compact linear operator , and the system is reduced to the following fredholm ( integral ) equation @xmath478 note here that the operator @xmath479 defined in depends only on @xmath14 and the solutions @xmath109 , so that the injectivity properties depend on the @xmath14 around which we pose the problem , in particular , whether one can fulfill hypotheses [ hyp : det ] and [ hyp : z ] . equation satisfies a fredholm alternative . in particular ,
if @xmath200 is not an eigenvalue of @xmath479 , admits a unique solution @xmath475 ( injectivity ) , @xmath480 is well - defined and continuous and @xmath481 satisfies the estimate @xmath482 from which we deduce stability below . in the statement of theorem [ thm :
explicit ] , the fact that `` system with trivial right - hand sides admits no non - trivial solution '' precisely means that @xmath200 is not an eigenvalue of the operator @xmath479 .
when @xmath14 is constant , constructing @xmath130 as in remark [ rem : const ] yields @xmath483 a constant matrix . in particular , the commutators @xmath484 $ ] vanish in the expression of @xmath194 .
thus system is decoupled and clearly injective . by continuity
, we also obtain that is injective for @xmath14 ( not necessarily scalar ) sufficiently close to a constant . starting from the integral version of the elliptic system in the case where @xmath485 , then the fredholm alternative implies . in order to translate inequality into a stability statement
, we must bound @xmath486 in terms of the measurements @xmath487 .
we have for @xmath41 , @xmath488 and since @xmath196 , expressed in involves the @xmath33 and their derivatives up to second order , if we assume all other multiplicative coefficients to be uniformly bounded , we obtain an estimate of the form @xmath489 thus we obtain in the end , an estimate of the form @xmath490 once @xmath481 is reconstructed , we can reconstruct @xmath0 uniquely from @xmath140 and @xmath183 $ ] using formula , with the stability estimate @xmath491 in order to see that @xmath179 satisfies a gradient equation that improves the stability of its reconstruction , the quickest way is to linearize ( * ? ? ?
* equation ( 7 ) ) derived in the non - linear case , which reads as follows : @xmath492 where @xmath493 is the @xmath44 matrix of power densities @xmath494 and @xmath495 is the @xmath496-th entry of @xmath497 . plugging the expansions @xmath26 , @xmath498 , @xmath499 , and using the fact that @xmath500 the linearized equation at @xmath418 reads @xmath501 from this equation , and using the stability estimates and , it is straighforward to establish the estimate @xmath502 and thus the proof is complete . | this paper concerns the reconstruction of an anisotropic conductivity tensor in an elliptic second - order equation from knowledge of the so - called power density functionals .
this problem finds applications in several coupled - physics medical imaging modalities such as ultrasound modulated electrical impedance tomography and impedance - acoustic tomography .
we consider the linearization of the nonlinear hybrid inverse problem .
we find sufficient conditions for the linearized problem , a system of partial differential equations , to be elliptic and for the system to be injective .
such conditions are found to hold for a lesser number of measurements than those required in recently established explicit reconstruction procedures for the nonlinear problem .
guillaume bal and chenxi guo franois monard ( communicated by the associate editor name ) |
topological insulator ( ti ) is a new class of material which is an insulator in its bulk , while having time reversal symmetry protected conducting states on the edge or surface .
@xcite a large number of realistic materials have been theoretically proposed and experimentally confirmed , such as bi@xmath1se@xmath2 and bi@xmath1te@xmath2 .
@xcite however , the layered transition - metal pentatelluride zrte@xmath0 ( and hfte@xmath0 ) is a particular example . zrte@xmath0 and hfte@xmath0
have been studied more than 30 years ago due to the large thermoelectric power @xcite and mysterious resistivity anomaly @xcite .
recently , weng _ et al .
_ predicted that mono - layer zrte@xmath0 and hfte@xmath0 are good quantum spin hall insulators with relatively large bulk band gap ( about 0.1 ev ) by first principles calculations .
@xcite and the three - dimensional bulk phase of zrte@xmath0 and hfte@xmath0 are also predicted to be tis , which are located at the vicinity of a transition between strong and weak ti , but without detailed description .
@xcite nevertheless , subsequent experiments show many contradictions about the topological nature of zrte@xmath0 or hfte@xmath0 .
several experimental works suggested that zrte@xmath0 is a dirac semi - metal without a finite band gap by different characterization methods , such as shubnikov - de haas oscillations , angle - resolved photoemission spectroscopy ( arpes ) , and infrared reflectivity measurements .
@xcite of course there are also other experimental works holding opposite point of view .
for example , in two recent scanning tunneling microscopy ( stm ) experiments , they unambiguously observed a large bulk band gap about 80 or 100 mev in zrte@xmath0 , @xcite implying that there is no surface state on the top surface and therefore it should be a weak ti .
another apres work also favored a weak ti for zrte@xmath0 .
@xcite however , there are two other arpes works which believed that zrte@xmath0 is a strong ti .
@xcite for instance , by using the comprehensive arpes , stm , and first principles calculations , manzoni _ et al .
_ found a metallic dos at fermi energy , which arises from the two - dimensional surface state and thus indicates zrte@xmath0 is a strong ti .
@xcite the divergence of these experiments make zrte@xmath0 ( hfte@xmath0 ) being a very puzzling but interesting material , which needs more further experimental and theoretical studies .
therefore , in order to figure out the physical mechanism behind those contradictory experimental results , we revisited the band structures of zrte@xmath0 and hfte@xmath0 , and carefully studied their relationship with the volume expansion .
we find a clear topological transition between a strong and weak ti in zrte@xmath0 and hfte@xmath0 , accompanied by a dirac semi - metal phase between them .
this work could shed more light on a unified explain about the different experimental results , and propose the crucial clue to further experiments to elucidate the topological nature of zrte@xmath0 and hfte@xmath0 .
the geometric and electronic properties of zrte@xmath0 and hfte@xmath0 are calculated by the density functional theory in the generalized gradient approximation implemented in the vienna ab - initio simulation package ( vasp ) code .
@xcite the projected augmented wave method @xcite and the van der waals ( vdw ) corrected optb86b - vdw functional @xcite are used .
the plane - wave cutoff energy is 300 ev and the k - point mesh is @xmath3 in the calculations . and a denser k - point mesh is used in the density of state ( dos ) calculation . spin - orbit coupling ( soc ) is included in the calculation except in the structural optimization .
the theoretical ground states of zrte@xmath0 and hfte@xmath0 are obtained by fully optimization of the atom positions and lattice constants , until the maximal residual force is less than 0.01 ev /
. then we vary and fix the volumes of unit cell and still optimize the atom positions and lattice constants to study the possible topological transition in zrte@xmath0 and hfte@xmath0 .
as shown in fig .
1(a ) , zrte@xmath0 and hfte@xmath0 share the same base - centered orthorhombic crystal structure , with cmcm ( no .
63 ) space group symmetry .
trigonal prismatic zrte@xmath2 chains oriented along the @xmath4 axis make up the zrte@xmath0 natural cleavage plane .
each chain consists of one zr atom and two different kinds of te atoms .
zrte@xmath2 chains are connected by zigzag te chains along the @xmath5 axis , building a two dimensional structure of zrte@xmath0 in the @xmath4-@xmath5 plane .
one crystal unit cell contains two zrte@xmath0 planes piled along the @xmath6 axis , the stack orientation of zrte@xmath0 .
the brillouin zone and its high symmetry k - points of zrte@xmath0 ( hfte@xmath0 ) are shown in fig .
1(b ) , in which @xmath7 , @xmath8 , and @xmath9 are the reciprocal lattice vectors .
( color online ) ( * a * ) layered crystal stucture of zrte@xmath0 ( hfte@xmath0 ) in the orthorhombic conventional unit cell .
big green and small brown balls represent zr ( hf ) and te atoms respectively . the layers stack along the @xmath6 direction .
( * b * ) brillouin zone and the high symmetry points of the primitive unit cell of zrte@xmath0 ( hfte@xmath0 ) .
, scaledwidth=45.0% ] [ cols="^,^,^,^,^,^",options="header " , ] due to the weak vdw interaction in the layer zrte@xmath0 ( hfte@xmath0 ) @xcite , the vdw corrected correlation functional is necessary in order to obtain the good theoretical lattice constants . in table
i , we present the optimized lattice constants of zrte@xmath0 and hfte@xmath0 based on the optb86b - vdw functional , as well as the experimental ones from reference @xcite .
we find that the theoretical and experimental lattice constants are well consistent with each other and the maximum difference between them is less than 1% . for comparison
, we also optimized the structures of zrte@xmath0 and hfte@xmath0 by using the standard perdew - burke - ernzerhof ( pbe ) exchange correlation .
@xcite it is obvious that there is a large error ( about 9 % ) in the lattice @xmath6 ( in the stacking direction ) , which indicates that standard pbe failed to describe the structures of zrte@xmath0 andhfte@xmath0 , and the vdw correction is necessary .
( color online ) ( * a * ) lattice constants @xmath4 , @xmath6 , and @xmath5 under different unit cell volumes .
( * b * ) relative total energy of a primitive unit cell of zrte@xmath0 under different volumes .
the blue vertical dotted lines represent the experimental volume at 10 k. @xcite , scaledwidth=45.0% ] in order to explore the possible topological phase transition in zrte@xmath0 ( hfte@xmath0 ) , we then study their electronic properties under different volumes . based on the above optimized structure ,
we change volume of the unit cell by hand and then optimize the atom positions and lattice constants under each variant volumes .
this process can simulate the hydrostatic pressure experiments or the thermal expansion effect due to a finite temperatures .
it is noted that we did not change the volume drastically , and the system is far away from the region of superconductivity phase under high pressure found in zrte@xmath0 and hfte@xmath0 .
@xcite in fig .
2(a ) , we present the change of lattice constants @xmath4 , @xmath6 , and @xmath5 under different volume expansion ratio , defined as @xmath10 , where @xmath11 is the unit cell volume at theoretical ground state listed in the table i. it is found that all the lattice constants have similar linear dependence on the volume of the unit cell . but the in - plane lattice constants @xmath4 and @xmath5 changes much slower with the volume than that of the lattice constant @xmath6 , which indicates the weak inter - layer binding energy along the @xmath6 direction in zrte@xmath0 . the parabolic - like relationship between total energy of unit cell and volume
is expected and given in fig . 2 ( b ) .
the blue vertical dotted line represent the experimental volume at low temperature ( 10 k ) , which is nearly 1.9% smaller than our calculated value .
then we have calculated the band structures and the density of states ( doss ) with soc under variant volumes , and three of them are shown in fig .
3 . the calculated band structure ( fig .
3 ( a ) ) at the ground state volume ( @xmath12 ) is similar as previous theoretical computation @xcite , although we use a different high symmetry k - path .
a clear direct band gap about 94.6 mev at @xmath13 point is found in the band structure .
of course it is also found that the valence band maxima is between the @xmath13 and y point and conduction band minima is between the a@xmath14 and t point .
therefore the indirect band gap is much smaller than the direct one at @xmath13 point , which is about 41.7 mev in fig .
the present of a clear band gap is confirmed in its corresponding dos ( fig .
3 ( d ) ) .
our calculated band gap is comparable with the values observed in the previous experiments , @xcite which is 80 or 100 mev .
furthermore , it is found that the slop of dos in the conduction band is much larger than that in the valence band , possibly because the conduction band has more valleys than the valence band .
this characteristic is also found in the experiments .
@xcite when zrte@xmath0 expands from its ground state , the band gap decreases gradually until the valence and conduction bands touch each other at a critical volume expansion ratio about 2.72% .
then , a dirac point is formed at @xmath13 point , which can be clearly seen in fig .
this behavior is also confirmed in its corresponding v - shaped dos near the fermi energy , as shown in fig .
3(e ) , which is the feature of dirac point in band structure .
as the crystal continues to expand , the band gap of zrte@xmath0 opens again , and ultimately reaches a value of about 102.6 ( direct ) or 27.7 mev ( indirect ) under a volume expansion 6.12% .
( see figure 3 ( c ) and ( f ) ) . therefore from fig .
3 , we can clearly see a transition from a semiconductor to a semi - metal and then to a semiconductor again in zrte@xmath0 when it expands . in order to check whether such a transition is topological or not ,
we have calculated the @xmath15 indices under each volume . @xcite it is found that the @xmath15 indices are all ( 1;110 ) when the volume expansion is less than 2.72% , while it is ( 0;110 ) when the volumes expansion is larger than 2.72% .
this definitively confirms that zrte@xmath0 undergoes a topological phase transition from a strong ti , to a dirac semi - metal , and finally turns to a weak ti when its unit cell expands from 0 to 6.12% in our calculation .
( color online ) ( * a * ) calculated absolute value of direct band gap at @xmath13 point of zrte@xmath0 under different volumes .
the light green and blue region represent the phases of strong and weak tis respectively .
the boundary between the strong and weak ti is the semi - metal state .
the red and blue dotted vertical lines represent the experimental volume at 10 k and calculated one at ground state respectively .
( * b * ) same as ( * a * ) but for hfte@xmath0 .
, scaledwidth=45.0% ] the detailed phase diagrams of such a topological phase transition of zrte@xmath0 and hfte@xmath0 are given in fig .
4 , in which all the calculated absolute value of direct band gaps at @xmath13 point under different volumes are plotted . in fig .
4(a ) we can find that the band gaps of zrte@xmath0 decreases linearly as the volume increases from a negative volume expansion ratio about -6% , with a rate around -33 mev per 1% change of volume , where the negative value means a decrease of the band gap when the crystal expands .
the band gap disappears at @xmath16 .
then it raises linearly with volume in a similar rate of 28 mev per 1% change of volume .
therefore zrte@xmath0 undergoes a strong ti to weak ti topological phase transition due to volume expansion .
such a transition must need a zero - gap intermediate state , which is the semi - metal state found at about @xmath16 in our calculation .
similar phase diagram is also proposed by a recent work , @xcite in which they present the band gap at @xmath13 point as a function of the inter - layer distance , but not the volume of the unit cell . in fig .
4(b ) , we also show that hfte@xmath0 has the very similar topological phase transition , with almost the same transition critical volume expansion ratio at about 2.72% .
the band gap of hfte@xmath0 also changes linearly as the volume increases with a rate about -31 and 26 mev per 1% change of volume in the strong and weak ti region respectively .
such a band gap changing rate is quite significant especially in a small band gap semiconductor material .
therefore , we can conclude that the electronic properties of zrte@xmath0 ( hfte@xmath0 ) are indeed very sensitive to the change of the volume and they are indeed located very close to the boundary between the strong and weak ti .
although our optimized and the experimentally measured volume of zrte@xmath0 ( hfte@xmath0 ) both indicate that they should be within the strong ti region , we think it still has the possibility that in experiment zrte@xmath0 ( hfte@xmath0 ) can locate in a weak ti region due to different growth methods and characterization techniques in experiments . according to fig .
4 , it is even possible that zrte@xmath0 ( hfte@xmath0 ) can be in the dirac semi - metal phase if it happens that its unit cell has a proper volume expansion ratio , which possibly been observed in many experiments .
@xcite another more possible reason which can explain the semi - metal behavior found in these experiments is due to the defect and doping , which make the zrte@xmath0 ( hfte@xmath0 ) being a degenerate semiconductor . in a degenerate semiconductor ,
the fermi energy is located within the conduction or valence band due to the doping effect , and the crystal will behave like a metal .
but in this case , the energy gap still exist just below or above the fermi energy and the dirac point is not needed in the energy gap . in order to avoid the possible artificial effect induced by the cleavage in both stm and arpes experiments ,
we suggest that nondestructive optical measurements for the existence of a direct band gap at @xmath13 point , and its change under different temperatures , in the high quality and stoichiometric single crystals are probably useful to elucidate the topological nature in zrte@xmath0 and hfte@xmath0 .
finally we show the importance of our calculated change rate of band gap .
first we can roughly estimate the bulk thermal expansion coefficient from experimental lattice constants of zrte@xmath0 ( table i ) to be about @xmath17 k@xmath18 , which means that the volume will change about 1% when the temperature changes from 0 to 300 k , equivalently the band gap of zrte@xmath0 at @xmath13 point will change about -33 mev for strong ti phase or 28 mev for weak ti phase , according to our calculation . in a recent high - resolution arpes work , @xcite zhang _ et al .
_ found a clear and dramatic temperature dependent band gap in zrte@xmath0 , from which we then can estimate that change rate of observed band gap is about 26 or 37 mev from 0 to 300 k depending on the methods used in their experiment .
these two values are both well consistent with our calcualted result .
moreover , the positive change rate found in the experiment @xcite implies that the zrte@xmath0 crystal used in their experiment is probably a weak ti according to our calculated phase diagram in fig .
in conclusion , we have studied the band structures of zrte@xmath0 and hfte@xmath0 at variant volume by first principles calculations . a clearly volume dependent strong and weak topological phase transition
is found , accompanied by a zero - gap semi - metal phase at the boundary between the transition .
the direct band gap of zrte@xmath0 at @xmath13 point changes linearly with the volume , which is -33 mev and 28 mev in a strong and weak ti phase respectively , if the volume of zrte@xmath0 increases 1% , or equivalently if the temperature increases from 0 to 300 k. the results for hfte@xmath0 is very similar to thoese of zrte@xmath0 .
our calculated results indicate that the electronic properties and topological nature of zrte@xmath0 and hfte@xmath0 are indeed very sensitive to the lattice constants of crystals , which is probably the reason for the divergent experimental results at present .
we suggest that high quality and stoichiometric single crystal with accurate structure refinement at different temperatures would be helpful to resolve the divergent experimental results in zrte@xmath0 and hfte@xmath0 .
this work is supported by the state key program for basic research ( no .
2015cb659400 ) , the national key technologies r&d program of china ( 2016yfa0201104 ) , and the national science foundation of china ( no . 11474150 ) . the use of the computational resources in the high performance computing center of nanjing university for this work is also acknowledged .
m. z. hasan and c. l. kane , rev .
phys . * 82 * , 3045 ( 2010 ) .
qi and s. -c .
zhang , rev .
* 83 * , 1057 ( 2011 ) .
y. ando , j. phys .
jap . * 82 * , 102001 ( 2013 ) . h.
j. zhang , c. x. liu , x. l. qi , x. dai , z. fang , and s. c. zhang , nature physics * 5 * , 438 ( 2009 ) . y. l. chen , j. g. analytis , j. h. chu , z. k. liu , s .- k .
mo , x. l. qi , h. j. zhang , d. h. lu , x. dai , z. fang , s. c. zhang , i. r. fisher , z. hussain , and z .- x .
shen , science , * 325 * , 178 ( 2009 ) .
t. e. jones , w. w. fuller , t. j. wieting , and f. levy , solid state commun . *
42 * , 793 ( 1982 ) .
s. okada , t. sambongi , and m. ido , j. phys .
soc . jap . *
49 * , 839 ( 1980 ) .
f. j. disalvo , r. m. fleming , and j. v. waszczak , phys .
b * 24 * , 2935 ( 1981 )
. r. y. chen , z. g. chen , x .- y .
song , j. a. schneeloch , g. d. gu , f. wang , and n. l. wang , phys .
. lett . * 115 * , 176404 ( 2015 ) .
r. y. chen , s. j. zhang , j. a. schneeloch , c. zhang , q. li , g. d. gu , and n. l. wang , phys .
b * 92 * , 075107 ( 2015 ) .
q. li , d. e. kharzeev , c. zhang , y. huang , i. pletikosi , a. v. fedorov , r. d. zhong , j. a. schneeloch , g. d. gu , and t. valla , nature phys .
* 12 * , 550 ( 2016 ) .
y. w. liu , x. yuan , c. zhang , z. jin , a. narayan , c. luo , z. chen , l. yang , j. zou , x. wu , s. sanvito , z. c. xia , l. li , z. wang , and f. x. xiu , nat .
* 7 * , 12516 ( 2016 ) .
x. yuan , c. zhang , y. w. liu , c. y. song , s. d. shen , x. sui , h. yu , z. h. an , j. zhao , h. g. yan , and f. x. xiu , arxiv:1510.00907 ( 2016 ) .
g. l. zheng , j. w. lu , x. d. zhu , w. ning , y. y. han , h. w. zhang , j. l. zhang , c. y. xi , j. y. yang , h. f. du , k. yang , y. h. zhang , and m. l. tian , phys .
b * 93 * , 115414 ( 2016 ) .
li , w .- k .
huang , y .- y .
lv , k .- w .
zhang , c .-
yang , b .- b .
zhang , y. b. chen , s .- h .
yao , j. zhou , m .- h .
lu , l. sheng , s .- c .
li , j .- f .
jia , q .- k .
xue , y .- f .
chen , and d .- y .
xing , phys .
lett . * 116 * , 176803 ( 2016 ) .
r. wu , j .- z .
nie , l .- x .
zhao , x. huang , j .- x .
yin , b .- b .
fu , p. richard , g .- f .
chen , z. fang , x. dai , h .- m .
weng , t. qian , h. ding , and s. h. pan , physical review x * 6 * , 021017 ( 2016 ) .
l. moreschini , j. c. johannsen , h. berger , j. denlinger , c. jozwiak , e. rotenberg , k. s. kim , a. bostwick , and m. grioni , phys .
b * 94 * , 081101 ( 2016 ) .
g. manzoni , l. gragnaniello , g. auts , t. kuhn , a. sterzi , f. cilento , m. zacchigna , v. enenkel , i. vobornik , l. barba , f. bisti , ph .
bugnon , a. magrez , v. n. strocov , h. berger , o. v. yazyev , m. fonin , f. parmigiani , and a. crepaldi , arxiv:1608.03433 ( 2016 ) .
g. manzoni , a. crepaldi , g. auts , a. sterzi , f. cilento , a. akrap , i. vobornik , l. gragnaniello , ph .
bugnon , m. fonin , h. berger , m. zacchigna , and o.v .
yazyev , f. parmigiani , j. electron spectrosc .
, in press , ( 2016 ) .
y. h. zhou , j. f. wu , w. ning , n. n. li , y. p. du , x. l. chen , r. r. zhang , z. h. chi , x. f. wang , x. d. zhu , p. c. lu , c. ji , x. g. wan , z. r. yang , j. sun , w. g. yang , m. l. tian , y. h. zhang , and h. k. mao , proc .
* 113 * , 2904 ( 2016 ) .
qi , w. j. shi , p. g. naumov , n. kumar , w. schnelle , o. barkalov , c. shekhbar , h. borrmann , c. felser , b. h. yan , s. a. medvedev , phys .
b * 94 * , 054517 ( 2016 ) .
y. liu , y. j. long , l. x. zhao , s. m. nie , s. j. zhang , y. x. weng , m. l. jin , w. m. li , q. q. liu , y. w. long , r. c. yu , x. l. fen , q. li , h. m. weng , x. dai , z. fang , g. f. chen , and c. q. jin , arxiv:1603.00514 ( 2016 ) .
y. zhang , c. l. wang , l. yu , g. d. liu , a. j. liang , j. w. huang , s. m. nie , y. x. zhang , b. shen , j. liu , h. m. weng , l. x. zhao , g. f. chen , x. w. jia , c. hu , y. ding , s. l. he , l. zhao , f. f. zhang , s. j. zhang , f. yang , z. m. wang , q. j. peng , x. dai , z. fang , z. y. xu , c. t. chen , and x. j. zhou , arxiv:1602.03576 ( 2016 ) . | zrte@xmath0 and hfte@xmath0 have attracted increasingly attention recently since the theoretical prediction of being topological insulators ( tis ) .
however , subsequent works show many contradictions about their topological nature .
three possible phases , i.e. strong ti , weak ti , and dirac semi - metal , have been observed in different experiments until now . essentially whether zrte@xmath0 or hfte@xmath0 has a band gap or not is still a question .
here , we present detailed first - principles calculations on the electronic and topological properties of zrte@xmath0 and hfte@xmath0 on variant volumes and clearly demonstrate the topological phase transition from a strong ti , going through an intermediate dirac semi - metal phase , then to a weak ti when the crystal expands . our work might give a unified explain about the divergent experimental results and propose the crucial clue to further experiments to elucidate the topological nature of these materials . |
mixing in the @xmath0-@xmath1 system has been searched for for more than two decades without success until this past year .
both `` @xmath4-factory '' experiments , belle and , have recently published evidence for this phenomenon @xcite . here
we review these measurements and discuss their implications . in total four measurements
are presented , involving the following decay modes @xcite : , , , and . mixing in heavy flavor systems such as that of @xmath5 and @xmath6 is governed by the short - distance box diagram . however , in the @xmath0 system this diagram is both doubly - cabibbo - suppressed and gim - suppressed relative to the amplitude dominating the decay width , and thus the short - distance rate is very small .
consequently , @xmath0-@xmath1 mixing is expected to be dominated by long - distance processes that are difficult to calculate ; theoretical estimates for the mixing parameters @xmath7 and @xmath8 range over two - three orders of magnitude @xcite . here ,
@xmath9 and @xmath10 are the differences in the masses and decay widths , respectively , of the two @xmath0-@xmath1 mass eigenstates , and @xmath11 is the mean decay width .
all methods discussed here identify the flavor of the @xmath12 when produced by reconstructing the decay @xmath13 or @xmath14 ; the charge of the accompanying pion identifies the @xmath15 flavor . because @xmath16 , which is relatively small , the pion has very low momentum and thus is denoted @xmath17 ( `` @xmath18 slow '' ) .
the @xmath0 decay time ( @xmath19 ) is calculated via @xmath20 , where @xmath21 is the distance between the @xmath22 and @xmath0 decay vertices and @xmath23 is the @xmath0 momentum .
the @xmath22 vertex position is taken to be the intersection of the @xmath17 momentum with the beamspot profile .
most of the precision on @xmath19 is due to the vertical ( @xmath24 ) component of @xmath21 , as the spread of the beamspot is only a few microns in this dimension . to reject @xmath25
decays originating from @xmath4 decays , one requires @xmath26 , which is the kinematic endpoint .
the decay is in principle an ideal signature for mixing , as this `` wrong - sign '' ( ws ) final state can be reached only via a @xmath27 transition ( in contrast to the `` right - sign '' ( rs ) final state @xmath28 ) .
however , the neutrino in the final state precludes the decay from being fully - reconstructed , and consequently the decay time measurement is smeared .
the decay time dependence is given by @xmath29 $ ] , and both belle and measure the coefficient @xmath30 by fitting the decay time distribution .
to reduce backgrounds , only the electron channel ( @xmath31 ) has been used .
belle uses 253 fb@xmath32 of data @xcite , making a relatively loose event selection and fitting the resulting @xmath33 and @xmath34 distributions to determine the yield of ws events . to improve the resolution on both @xmath9 and the decay time
, two corrections are made to better determine @xmath35 , where @xmath36 is the total four - momentum of all tracks and photons in the event besides the @xmath17 , @xmath37 , and @xmath38 candidates , and @xmath39 is the four - momentum of the center - of - mass ( cms ) system .
first , @xmath36 is rescaled by a factor @xmath40 such that @xmath41 ; second , @xmath42 is rotated in the plane of @xmath42 and @xmath43 such that @xmath44 .
the ratio of ws to rs events are measured for six bins of decay time ranging from one to ten @xmath0 lifetimes .
summing over all bins , @xmath45 rs and @xmath46 ws events are found .
for each bin the ratio depends on @xmath47 , and the six resulting values of @xmath47 are fit to a constant .
the result is @xmath48 where the first error listed is statistical and the second error is systematic .
the latter is dominated by uncertainty in the background @xmath9 pdf .
as much of the @xmath47 likelihood is in the unphysical ( negative ) region , belle uses a feldman - cousins @xcite approach to calculate an upper limit ; the result is @xmath49 at 90% c.l .
this limit implies both @xmath50 and @xmath51 are @xmath52% .
the experiment uses 344 fb@xmath32 of data @xcite and , in contrast to belle , imposes tight selection criteria to reduce background as much as possible ( with a corresponding loss in efficiency ) .
the most restrictive criterion is that events must have a @xmath25 decay fully reconstructed in the hemisphere opposite that of the semileptonic decay , where the modes @xmath13 , @xmath53 , @xmath54 , @xmath55 , and @xmath56 are used .
this `` double - tagging '' eliminates copious background from real @xmath0 s combining with random @xmath57 tracks to make false @xmath22 candidates , but it reduces the signal efficiency by an order of magnitude . the neutrino momentum is determined via a neural network algorithm , and events are required to have decay times in the range 600 - 3900 fs ( corresponding to 1.5 - 9.5 @xmath0 lifetimes ) .
a separate neural network is used to select signal events , and a final set of kinematic selection criteria are applied to the signal side .
these criteria include cuts on the @xmath17 and @xmath38 momenta , and on the @xmath58 of the electron track in the silicon vertex tracker .
the final candidate samples for ws data and monte carlo ( mc ) simulation are shown in figs .
[ fig : semi_babar]a and [ fig : semi_babar]b , respectively . in the data ,
three events are observed in the signal region @xmath59 , whereas the mc predicts ( for the luminosity of the data ) @xmath60 background events .
together these values give @xmath61% , where the error corresponds to where the log - likelihood function for @xmath62 ( the true number of signal events ) rises by 0.50 units with respect to the minimum value .
the points where the log - likelihood function rises by 1.35 units give a 90% c.l .
constraint @xmath63 .
this upper bound is similar to the upper limit obtained by belle .
distribution for s candidate sample @xcite , in data ( left ) and mc ( right ) .
the mc sample is 1.75 times the size of the data sample .
the dark ( light ) histogram shows events after ( before ) the final kinematic selection . in the left plot
, region `` 1 '' denotes the signal region used to determine @xmath47.,title="fig:",width=151 ] distribution for s candidate sample @xcite , in data ( left ) and mc ( right ) .
the mc sample is 1.75 times the size of the data sample .
the dark ( light ) histogram shows events after ( before ) the final kinematic selection . in the left plot ,
region `` 1 '' denotes the signal region used to determine @xmath47.,title="fig:",width=151 ]
two - body hadronic final states @xmath65 , @xmath66 , and @xmath67 can be reached from either @xmath0 or @xmath1 ; thus @xmath68 and @xmath69 amplitudes both contribute to the decay rate , and detecting the effect of the latter provides evidence for mixing .
the time dependence of the decay rate @xmath70 is given by @xmath71\ , , \label{eqn : master}\end{aligned}\ ] ] where @xmath72 and @xmath73 are complex coefficients relating mass eigenstates to flavor eigenstates : @xmath74 .
the parameter @xmath75 can be written @xmath76 , where @xmath77 is the strong phase difference between amplitudes @xmath78 and @xmath79 , and @xmath80 is a possible weak phase . in the absence of violation ( ) ,
@xmath81 and @xmath82 . for @xmath83
, @xmath78 is doubly - cabibbo - suppressed , @xmath79 is cabibbo - favored , and thus @xmath84 and @xmath85 .
for decays to self - conjugate states @xmath65 and
@xmath66 , @xmath86 and the third term in eq .
( [ eqn : master ] ) can be neglected since @xmath50 and @xmath51 are very small . as @xmath87 ,
if there is no in mixing ( @xmath88 ) then @xmath89 and @xmath90 \approx e^{-\gamma(1+y\cos\phi)\,t}$ ] .
thus @xmath91 .
the observable @xmath92 is denoted @xmath93 and , for no in mixing , equals @xmath94 ; if is conserved , @xmath95 . allowing for arbitrary , one obtains @xcite @xmath96 the ( normalized ) difference in lifetimes @xmath97 is equal to the related expression @xcite @xmath98 this method has been used by numerous experiments to constrain @xcite .
belle s new measurement @xcite uses 540 fb@xmath32 of data and both @xmath65 and @xmath66 final states .
one advance of this analysis is the resolution function , which is constructed as a sum over many gaussian functions @xmath99 : @xmath100 with standard deviations @xmath101 . in this expression
, @xmath102 is the weight of the value @xmath103 taken from the normalized , binned , @xmath53 distribution of @xmath104 , the event - by - event uncertainty in the decay time ( i.e. , @xmath105 ) .
parameter @xmath106 is the weight of value @xmath107 obtained by fitting the mc pull distribution to a sum of three gaussians with widths @xmath108 .
the @xmath109 are scale factors to account for differences between mc and data , and @xmath110 is a common offset .
the parameters @xmath109 and @xmath110 are left free when fitting for .
this resolution function , and a slight variation with an additional offset parameter , yields accurate values of the @xmath53 lifetime over all running periods .
the mean value is @xmath111 fs , which is consistent with the pdg value @xcite ( and actually has greater statistical precision ) .
fitting the @xmath112 , @xmath65 , and @xmath66 decay time distributions ( figs .
[ fig : ycp_belle]a - c ) shows a statistically significant difference between the @xmath112 and @xmath113 lifetimes .
the effect is visible in fig .
[ fig : ycp_belle]d , which plots the ratio of event yields @xmath114 as a function of decay time . performing a simultaneous maximum likelihood ( ml )
fit to all three @xmath115 samples gives @xmath116 which deviates from zero by @xmath117 .
the systematic error is dominated by uncertainty in the background decay time distribution , variation of selection criteria , and the assumption that @xmath110 is equal for all three final states .
the analysis also measures @xmath118 which is consistent with zero ( no ) .
the sources of systematic error for @xmath119 are similar to those for . , and decays , from belle @xcite .
the hatched area represents the background contribution .
plot _ ( d ) _ shows the ratio of decay - time distributions for @xmath120 and @xmath53 ; the solid line is a fit to the points.,width=302 ] for @xmath121 and no , eq .
( [ eqn : master ] ) simplifies to @xmath122 $ ] , where @xmath123 and @xmath124 , @xmath125 are `` rotated '' mixing parameters .
both @xcite and belle @xcite do unbinned ml fits to the decay - time distribution of ws @xmath126 decays to determine @xmath127 , and @xmath128 . because @xmath129 is doubly - cabibbo - suppressed , @xmath128 is small and the mixing terms in the above expression play a larger role ; however , there is substantial background , @xmath13048% .
the largest background component consists of real @xmath53 decays combining with random @xmath131 tracks ; fortunately , the decay time distribution for this background is simple , the same as that for rs @xmath53 decays .
the results of the and belle fits are listed in table [ tab : kpi_results ] .
for , the value for @xmath132 is negative , i.e. , outside the physical region , nominally due to statistical fluctuation .
the likelihood contours are shown in fig .
[ fig : kpi_babar ] .
the no - mixing point @xmath133 has @xmath134 units above the minimum value , corresponding to a cl of only 0.01% ( @xmath135 ) including systematic uncertainty .
this constitutes evidence for mixing .
the largest systematic error is from uncertainty in modeling the tail of the background decay time distribution .
the mixing is visible in fig .
[ fig : kpi_babar_backsubtracted ] , which plots the ratio of the background - subtracted yields of ws to rs decays in bins of decay time . for each bin ,
the yields are determined from two - dimensional fits to variables @xmath136 and @xmath137 .
the plot shows the ratio increasing with decay time , consistent with eq .
( [ eqn : master ] ) but inconsistent with the no - mixing or flat hypothesis . fitting to eq .
( [ eqn : master ] ) gives @xmath138 , whereas fitting to a flat distribution gives @xmath139 .
to allow for , fits the @xmath0 and @xmath1 samples separately ; the results are consistent with each other , showing no evidence of ( see table [ tab : kpi_results ] ) . .
@xcite and belle @xcite results from fitting the decay time distribution of @xmath126 decays .
the errors listed are statistical plus systematic , except for those from s @xmath0 and @xmath1 subsamples , which are statistical only . [
cols="<,^,^,^",options="header " , ] and @xmath140 regions , where sensitivity to @xmath141 is highest.,title="fig:",width=147 ] 0.05 in and @xmath140 regions , where sensitivity to @xmath141 is highest.,title="fig:",width=147 ] 0.10 in and @xmath140 regions , where sensitivity to @xmath141 is highest.,title="fig:",width=147 ] 0.05 in and @xmath140 regions , where sensitivity to @xmath141 is highest.,title="fig:",width=147 ] from belle @xcite : dotted ( solid ) is statistical ( statistical plus systematic ) contour for no ; dashed - dotted ( dashed ) is statistical ( statistical plus systematic ) contour allowing for .
the point is the best - fit value for no .,width=302 ]
all mixing measurements can be combined to obtain world average ( wa ) values for @xmath40 and @xmath24 .
the heavy flavor averaging group ( hfag ) has done such a combination by adding together log - likelihood functions obtained from analyses of @xmath142 , @xmath143 , @xmath126 , @xmath144 , @xmath145 , and decays , as well as cleoc results for double - tagged branching fractions measured at the @xmath146 resonance @xcite .
the combination of likelihood functions preserves correlations among parameters and also accounts for non - gaussian errors .
when using this method , hfag assumes negligible . as a first step , wa values for and @xmath47 are calculated by taking weighted averages of independent experimental measurements see figs . [
fig : hfag_ycp ] and [ fig : hfag_rm].these results are then converted to three - dimensional likelihood functions for @xmath147 .
for example , the measurement of @xmath93 gives a parabolic log - likelihood function in @xmath24 and flat distributions in @xmath40 and @xmath77 .
the @xmath47 likelihood function is an annulus in the @xmath40-@xmath24 plane and a flat distribution in @xmath77 .
the logarithm of the likelihood functions are added , and the result is added to the @xmath147 log - likelihood function obtained from @xmath126 decays .
the latter is determined as follows .
the experiments directly measure a likelihood function @xmath148 ; thus one first projects out @xmath128 by allowing it to take , for any @xmath149 point , its preferred value .
the resulting likelihood for @xmath149 is converted to @xmath150 by scanning values of @xmath147 , calculating the corresponding values of @xmath149 , and assigning the likelihood for that @xmath149 bin .
this method ignores unphysical ( negative ) values of @xmath132 .
the resulting function @xmath150 is added to those obtained from , @xmath47 , and other measurements .
the final likelihood function is projected onto the @xmath141 plane by letting @xmath77 take , for any @xmath141 point , its preferred value .
this projection is shown in fig .
[ fig : hfag_contours ] .
the unusual shape around @xmath151 is mainly due to @xmath126 decays , which disfavor the no - mixing point . at @xmath151 ,
@xmath152 rises by 37 units above the minimum value ; this difference implies that the no - mixing point is excluded at the level of @xmath153 .
0.30 in -@xmath154 contours for @xmath141 , obtained by adding log - likelihoods from measurements of @xmath142 , @xmath143 , @xmath126 , @xmath144 , @xmath145 , and decays , and double - tagged branching fractions measured at the @xmath146 resonance ( from hfag @xcite).,title="fig:",width=302 ] the likelihood function is condensed to one dimension by letting , for any value of @xmath40 @xmath155 , the parameter @xmath24 @xmath156 take its preferred value .
the resulting likelihoods for @xmath141 give central values and 68.3% c.l .
intervals @xmath157 the former is @xmath158 from zero , and the latter is @xmath159 from zero . *
the experimental data consistently indicates that @xmath0 s undergo mixing .
the effect is presumably dominated by long - distance processes , and unless @xmath160 , it may be difficult to identify new physics from mixing alone . *
since is positive , the -even state is shorter - lived , as in the @xmath161-@xmath162 system .
however , since @xmath40 appears to be positive , the -even state is heavier , unlike in the @xmath161-@xmath162 system .
* there is no evidence yet for in the @xmath0-@xmath1 system .
e. m. aitala _ et al . _
( e791 ) , phys .
rev . lett . * 83 * , 32 ( 1999 ) ; j. m. link _ et al .
_ ( focus ) , phys . lett .
b * 485 * , 62 ( 2000 ) ; s. e. csorna _ et al .
_ ( cleo ) , phys . rev
. d * 65 * , 092001 ( 2002 ) ; b. aubert _ et al . _
( ) , phys . rev
. lett .
* 91 * , 121801 ( 2003 ) . | we review searches for mixing and violation in the @xmath0-@xmath1 system and discuss the first evidence for mixing recently obtained by the belle and collaborations .
we also present world average values for the mixing parameters @xmath2 and @xmath3 as calculated by the heavy flavors averaging group . |
these proceedings summarize measurements of @xmath2 decays to two kaons at the @xmath2 factories @xcite @xcite . observations of the @xmath5 and @xmath6 decays have been reported by the experiment @xcite at slac and the belle experiment @xcite at kek .
the collaboration has also performed the first time - dependent @xmath3-violation measurement in @xmath4 . the @xmath7 decay has not been observed by either experiment and the upper limits on its branching fraction were improved .
the @xmath8 penguin transition ( left ) and the annihilation amplitude in @xmath6 ( right).,title="fig:",scaledwidth=27.0% ] the @xmath8 penguin transition ( left ) and the annihilation amplitude in @xmath6 ( right).,title="fig:",scaledwidth=20.0% ] @xmath1 decays are expected to be dominated by the @xmath8 amplitude involving a virtual loop with the emission of a gluon ( fig .
[ fig : btodpenguin ] ) .
these decays are suppressed in the standard model ( sm ) , with branching fractions at the @xmath9 level .
assuming top - quark dominance in the virtual loop @xcite , the direct and mixing - induced @xmath3 asymmetries are expected to vanish in these modes .
however , contributions from up and charm quarks as well as from non - sm particles could induce sizeable cp asymmetries @xcite .
these modes are thus analogous to the @xmath10 decays in the measurement of virtual effects of contributions from non - sm phenomena @xcite .
additional amplitudes could affect @xmath11 penguin decays differently from @xmath12 penguin decays .
in addition , various relations between branching fractions and cp asymmetries in these modes provide a test of sm predictions @xcite .
finally , the @xmath6 decay has an annihilation contribution in the sm ( fig .
[ fig : btodpenguin ] ) .
thus , the difference in decay rate between the charged mode and the neutral mode can be used to constrain the effect of this amplitude in @xmath2 decays .
signal decays are separated from background decays using unbinned extended maximum likelihood fitting to distributions of kinematic and event - shape variables . the primary kinematic variables used to identify a reconstructed signal @xmath2 candidate are the difference between its reconstructed energy in the center - of - mass ( cm ) frame and the beam energy ; and a beam - energy substituted mass , @xmath13 , where the @xmath2-candidate momentum @xmath14 and the four - momentum of the initial @xmath15 state @xmath16 are calculated in the laboratory frame .
event - shape variables are used to suppress the dominant `` continuum '' @xmath17 @xmath18 background further , exploiting angular differences between the jet - like topology of continuum decays and the isotropically distributed decays of @xmath19 events .
particle - identification information is used to separate charged pion from charged kaon candidates in the @xmath6 and @xmath7 decays .
@xmath3 asymmetries in @xmath5 are determined from the difference in the time - dependent decay rates @xmath20 and @xmath21 for @xmath22 and @xmath23 signal decays , respectively , where @xmath24 .
\label{eq : eq1}\ ] ] + here @xmath25 is the time difference between the decays of the signal @xmath2 and the other @xmath2 in the event , @xmath26 is the average @xmath23 lifetime , and @xmath27 is the @xmath28 mixing frequency .
the amplitude @xmath29 describes @xmath3 violation in the interference between mixed and unmixed decays into the same final state , while @xmath30 describes direct @xmath3 violation in decay .
the time - dependent distribution is corrected for detector resolution effects and for mistag rates in the flavor identification of the other @xmath2 in the event .
@xmath5 decays are reconstructed in the @xmath31 and @xmath32 sub - decay modes . as the @xmath33 meson has a relatively long lifetime of @xmath34 ps
, its decay vertex is separated from the @xmath2 decay vertex by a few centimeters , which complicates the measurement of the @xmath35 position of the signal @xmath2 in the absence of prompt charged tracks originating from its decay vertex . to perform the measurement , used a method previously employed in other @xmath33 analyses without prompt charged tracks , wherein the @xmath2 meson is constrained in the vertex fit to decay within the beamspot in the plane transverse to the beam direction @xcite .
this method exploits the threshold nature of the @xmath36 decay , where the @xmath2 mesons are almost at rest in the cm frame ( @xmath37 mev / c ) and therefore have negligible displacement in the lab frame in the non - boosted transverse direction .
the resulting resolution of the @xmath35 position of the signal @xmath2 vertex is still better than the resolution on the vertex position of the other @xmath2 in the event , yielding a @xmath25 precision ( approximately @xmath38 ps ) comparable to that in modes where the signal @xmath2 has prompt charged tracks in the final state .
only @xmath33 mesons that decay within the silicon vertex tracker can be used with this technique . however , as there are two @xmath33 mesons in the final state , while only one is needed to employ this method , the fraction of signal decays suitable for the time - dependent measurement at is approximately @xmath39 in this mode .
if no time - dependent measurement is performed , an integrated flavor or charge asymmetry can be measured : @xmath40 + a non - zero value of this asymmetry signifies the presence of direct @xmath3 violation . in the @xmath41 modes ,
this is the only possible @xmath3 measurement .
both @xmath2 factories observe clear signals in this decay mode , with a branching fraction at the @xmath9 level , in agreement with each other and standard - model expectations .
observes the signal at the @xmath42 level of significance , including systematic uncertainties , while belle s observation is at the @xmath43 level .
the branching fraction results are summarized in table [ tab : ksks ] , while projections of the fitted data in the @xmath44 and @xmath45 variables with the results of the fit overlaid are shown in figs .
[ fig : belleplot ] and [ fig : babarplot ] .
the results of time - dependent @xmath3 measurement are presented in table [ tab : ksks ] and fig [ fig : babarcp ] . while the errors are still large , the measurement already excludes a fraction of the physically allowed region at a confidence level of greater than @xmath46 , as seen in fig .
[ fig : babarcp ] .
in particular , large positive values of @xmath29 are disfavored , although more data is needed to confirm this result .
belle did not perform a time - dependent measurement , but measured the time - integrated cp asymmetry , reporting @xmath47 . .
experimental results on @xmath5 decays , with the number of @xmath48 pairs listed in the last column . [
cols="<,<,^",options="header " , ]
each of the @xmath2 factories is expected to accumulate @xmath49 by the end of operations in @xmath50 . with these datasets ,
the error on the branching fractions of @xmath5 and @xmath51 should decrease to the @xmath52 level for each experiment .
also , each collaboration should achieve an error of @xmath53@xmath54 on @xmath29 and @xmath53@xmath55 on @xmath30 in @xmath5 .
these results will provide meaningful comparisona with theoretical predictions and constrain contributions from beyond - sm physics in @xmath0 fcnc decays .
the effect of the annihilation contribution in @xmath51 will be constrained as well .
the @xmath2 factories have observed the @xmath0 penguin dominated modes @xmath5 and @xmath51 .
the measured branching fractions are consistent with standard model predictions , although experimental errors are still too large for precision comparisons .
has also performed the first time - dependent @xmath3 violation measurement in @xmath5 , excluding a part of the physically allowed region at greater than @xmath46 significance .
this is the first step in the investigation of @xmath3 violation in @xmath0 flavor - changing neutral - current decays . | the @xmath0 penguin - dominated modes @xmath1 have been observed at the @xmath2 factories . in addition , the collaboration has reported the first time - dependent @xmath3-violation measurement in @xmath4 . |
deep colour - magnitude diagrams ( cmds ) from images taken with the advanced camera for survey ( acs ) and the wide field camera 3 ( wfc3 ) onboard the _ hubble space telescope _
( _ hst _ ) revealed that several intermediate - age ( @xmath0 12 gyr ) star clusters in the magellanic clouds host extended main - sequence turn - off ( msto ) regions @xcite , in some cases accompanied by composite red clumps @xcite .
the presence of these extended mstos ( hereafter emstos ) has been interpreted as due to stars that formed at different times within the parent cluster , with age spreads of 150500 myr @xcite .
if so , these objects might be the younger counterparts of the old galactic globular clusters with multiple stellar populations ( e.g. , * ? ? ?
an alternative scenario to explain the emsto phenomenon was originally proposed by @xcite , who suggested that this is due to the presence of a spread in rotation velocity among turn - off stars ( for a detailed discussion , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
finally , @xcite suggested that the emstos and the dual clumps can be photometric features produced by merged binary systems and interacting binaries with mass transfer .
recently , emstos have been detected also in young ( @xmath0 100300 myr ) massive star clusters in the large magellanic cloud ( lmc ) . in particular , @xcite and (
* hereafter c15 ) discovered the presence of an emsto in the star cluster ngc1856 , consistent with the presence of an age spread of @xmath0 80 myr .
interestingly , the cmd of ngc1856 also shows a split in the ms @xcite , which has been interpreted as an indication that the cluster might host two coeval stellar populations with different rotational velocities @xcite .
@xcite found that the cmd of the young ( @xmath0 100 myr ) star cluster ngc1755 shows a similar morphology , including an emsto and a ms split .
they demonstrated that the bimodal ms is well reproduced by a single stellar population with different rotation velocities , whereas the fit between the observed emsto and models with different rotation is not fully satisfactory . on the other hand
, @xcite showed that the @xmath0 300 myr cluster ngc1844 presents only a split in the ms and not an emsto .
it is evident that these new discoveries in young clusters give us a new opportunity to study the nature of the emsto phenomenon and to provide additional constraints to discriminate among the different scenarios . in this context , another interesting object is the young ( @xmath0 100 myr ) star cluster ngc1850 in the lmc .
@xcite used _ hst _
wfc3 observations with the _ f336w _ and _ f438w _ filters to show that the cluster hosts an emsto which can be reproduced by an age spread of @xmath0 40 myr . however , they suggest that this is the age range expected when the effect of rotation in msto stars is interpreted in terms of an age spread , and thus concluded that no age spread is present in the cluster .
they also suggest that the presence of an age spread should cause a `` bimodal '' distribution in the msto region , whereas the observed distribution can be approximated by a single ( possibly skewed ) gaussian distribution . here , we present an analysis of new _ hst _
wfc3 photometry of ngc1850 , exploiting a different set of filters with respect to that used by @xcite . our data , thanks to a significantly larger colour baseline , allow us to study thoroughly the cmd morphology and to perform a detailed analysis of the msto region .
we compare the cluster cmd with theoretical models , using isochrones with different ages and different rotation rates and with monte carlo simulations in order to quantify to what extent the cmd can be reproduced by a simple stellar population ( ssp ) with a distribution of rotation rates .
we derive the cluster mass and escape velocity to test whether the cluster has ( or has had ) the right properties to retain mass - loss material that can be ( or could have been ) used to form a second generation of stars .
this study allows us to have a clearer picture of the cluster s star formation history and in particular on the nature of the observed emsto region .
the remainder of the paper is organized as follows : observations and data reduction are presented in section2 . in section3 , we discuss the observed cmd , whereas in section4 we compare the observed cmd with theoretical models , using isochrones with different ages and different rotation rates . in section5
, we compare the observed cmd with monte carlo simulations , both using one ssp with different rotation rates and a combination of four ssps with different ages and different rotation rates ; we obtain the pseudo - colour distributions for the observed and simulated cmds and compare them . in section6 , we present the analysis of narrow - band h@xmath1 imaging . in section7 , we investigate how the cmd morphology changes as a function of the distance from the cluster center , while the physical and dynamical properties of the cluster are presented in section8 .
finally , in section9 , we discuss our results and their implications on the interpretation of the emsto phenomenon .
ngc1850 was observed with _ hst _ on 2015 october 22 using the uvis channel of the wfc3 as part of the _ hst _ program 14174 ( pi : p. goudfrooij ) .
the cluster was centered on one of the two ccd chips of the wfc3/uvis camera , to avoid the loss of its central region due to the ccd chip gap and to minimize the impact of different photometric zeropoints for the two chips .
the cluster was observed using three filters , namely _
f275w _ , _ f656n _ , and _ f814w_. two long exposures were taken in the _ f275w _ and _ f814w _ filters , and three in the _ f656n
the total exposure times for each filter are 1570 s ( _ f275w _ ) , 2875 s ( _ f656n _ ) , and 790 s ( _ f814w _ ) . in addition , we took two short exposures in the _ f275w _ and _ f814w _ filters ( 150 s and 10 s , respectively ) , to avoid saturation of the brightest stars .
the two or three long exposures were spatially offset from each other by 018 and -2401 in a direction + 8576 with respect to the positive x- and y - axis of the detector .
this was done to move across the gap between the two wfc3/uvis ccd chips , as well to simplify the identification and removal of hot pixels . to investigate the h@xmath1-emitting stars in the cluster
, we complemented our dataset with archival observations with the filter _
( go-11925 , pi : s. e. deustua ) . the main purpose of this filter in this context is to define the continuum level at the wavelength of h@xmath1 along with the _ f814w _ filter .
program go-11925 is a calibration program to test the linearity of wfc3/uvis , hence images in this filter were taken with various exposure times ; for our purpose , only the two long exposures were useful ( 300 s and 500 s , respectively ) .
in addition to the wfc3/uvis observations , we used the wide field camera ( wfc ) of acs in parallel to obtain images @xmath0 6from the cluster center using the _ f555w _ , _ f658n _ , and _
f814w _ filters .
these observations provide a clean picture of the stellar content , star formation history , and h@xmath1-emitting stars in the underlying lmc field . to reduce the images , we use the following method . briefly , we start from the _ flt _ files provided by the _ hst _ pipeline , which constitute the bias - corrected , dark - subtracted , and flat - fielded images .
the _ flt _
files are first corrected for charge transfer inefficiency effects using the dedicated cte correction software .
star positions and fluxes are measured on the cte - corrected _ flc _ files using the publicly - available fortran program _
img2xym_wfc3uv _ with the available library of spatially variable point - spread functions ( psfs , in an array of @xmath2 psfs ) for each filter .
we derive an additional @xmath3 array of perturbation psfs for each exposure , to account for telescope breathing effects @xcite , and we combine it with the library psfs to fit stellar profiles .
stellar positions are corrected for geometric distortion using the solution provided by @xcite .
then , we create a master - frame list using one of the _ f814w _ exposures and we transform and average the exposures for the other filters into the reference system defined by the master frame by means of a six - parameter linear transformation . to obtain the final catalog we use the fortran software package ks2 ( j. anderson , in preparation ) .
this program allows us to find and measure stars in all the individual exposures simultaneously and in particular to measure faint objects that can not be detected in individual exposures .
photometry of saturated stars is measured following the method developed by @xcite for the acs camera and subsequently demonstrated to be valid also for the wfc3/uvis detector @xcite , which allows one to recover the electrons that have bled into neighbouring pixels . to obtain the photometric calibration
, we compare psf - based instrumental magnitudes measured on the _ flc _ exposures with aperture - photometry magnitudes measured on the _ drc _ exposures ( i.e. , the distortion - corrected and resampled _ flc _ files ) using a fixed aperture of 10 pixels .
we then transform the magnitudes into the vegamag system by adopting the relevant synthetic zero - points . finally ,
for our analysis we select only those sources that are recovered in both exposures for the filters _
f275w _ , _ f467 m _ , and _ f814w _ , and in two out of three exposures for the filter _
f656n _ , and we clean the obtained catalog using the quality - of - fit ( qfit ) parameter , provided by ks2 , which identifies how well a source has been fit with the psf model @xcite .
completeness as well as the photometric - error distribution of the final photometry are characterized by performing artificial star tests .
we use the standard technique of adding artificial stars to the images and running them through the photometric routines applied in the reduction process . the method adopted to perform the artificial star tests
is described in @xcite .
briefly , we generate a list of @xmath4 artificial stars and we distribute them in magnitude in order to reproduce a luminosity function similar to the observed one and with a colour distribution that span the full colour ranges found in the cmds .
the overall distribution of the inserted artificial stars follows that of the stars in the image .
each star in the list is inserted one at a time in the images with the appropriate flux and position and it is measured using the same procedure and psf adopted for real stars .
we consider the inserted stars recovered if the input and output magnitudes agree to within 0.75 mag in every filter and if the input and output position agree to within 0.5 pixel .
finally , a completeness fraction is assigned to each individual star in a given cmd as a function of its magnitude and distance from the cluster center .
the left panel of fig.[f : fov ] shows a trichromatic stacked image of the wfc3/uvis field of view ( red channel : _
f814w _ filter ; green channel : _
f656n _ filter ; blue channel : _
f275w _ filter ) . the white circle has a radius of @xmath5 20 which corresponds to the effective radius @xmath6 of the cluster , as derived in sect.[s : king ] . on the left side of ngc1850 ,
the much younger cluster ngc1850b is clearly visible .
the stacked image nicely outlines the gas filaments which cover a large portion of the uvis field of view .
however , as derived in sect.[s : reddening ] , these filaments seem to lie behind the cluster and therefore do not affect our analysis . in the following , we define the region enclosed within a radius of @xmath7 as `` the cluster field '' . in order to minimize the contamination from the stars belonging to ngc1850b , we excluded from the final `` ngc1850 field '' catalog objects in the annular segment in the direction of ngc1850b ( i.e. , the stars that have a position angle between 138 and 217 in the detector frame and for which the distance from the cluster center is larger than @xmath6 ) . in the right panel of fig.[f
: fov ] , we show the spatial positions of the selected stars in the field of view . the objects that satisfy the aforementioned criteria and represent the ngc1850 cluster field are reported as red dots .
fig.[f : cmdobs ] shows the _ f814w _ vs. _ f275w f814w _ cmd of the ngc1850 field , derived as described in sect.[s : data ] and reported in the right panel of fig.[f : fov ] .
the observed cmd presents two interesting features : ( 1 ) a msto region that is wider than what is expected from a single stellar population and ( 2 ) the presence of two mses : a blue and poorly - populated ms , and a red ms , which contains the majority of the stars . the two sequences run parallel from _ f814w
_ @xmath0 18.5 mag to _ f814w _ @xmath0 21.25 mag , where they seem to merge together .
the presence of a small colour difference between the two sequences at fainter luminosity , not detectable due to photometric errors , can however not be excluded .
we calculate the ratio of stars in the blue and the red mses by carefully selecting stars in the two mses in a fixed magnitude interval ( 19.5 mag @xmath8 _ f814w _ @xmath8 21.0 ) .
we derive that the less - populated blue ms contains around 25% of stars , with the remaining 75% belonging to the red ms ( see sect.[s : radial ] for a detailed discussion ) .
this ratio is very similar to the one found by @xcite for ngc1755 . to verify that the observed broadening of the msto and the double ms are `` real '' features in the cluster cmd , we checked for the following potential causes : contamination by field stars , photometric errors , and differential reddening .
we assess the level of contamination by the field population by selecting a region near the top - right corner of the image which has the same surface area as the cluster field .
the stars in this region are shown in fig.[f : cmdobs ] as red open squares .
note that these stars mainly contaminate the lower ms , at fainter magnitudes than the observed splitting , and they do not affect the msto at all .
we therefore conclude that field stars are not the cause of the observed split in the ms and the broadening in the msto region . and shown in the right panel of fig.[f : fov ] .
we determine the contamination from the underlying lmc field population from a region , with the same surface area adopted for the cluster star , near the top - right corner of the image .
stars located in this region are superposed on the cluster cmd ( red squares ) .
magnitude and colour errors are shown in the left side of the cmd .
the reddening vector is also shown for @xmath9 . ]
photometric uncertainties are derived during the reduction process .
magnitude and colour errors are shown in the left panel of fig.[f : cmdobs ] ; photometric errors at the msto level are @xmath10 0.005 mag , whereas they are of the order of 0.015 mag at a magnitude _ f814w _ @xmath0 21.5 mag , where the mses are clearly separated .
these errors are far too small to account for the observed broadening of the msto .
to check whether differential reddening is present in our wfc3/uvis field of view and , in particular , in the cluster region , we adopt the same approach as in @xcite and described in detail in @xcite .
briefly , we first rotate the cmd to a photometric reference frame in which the x - axis is parallel to the reddening line .
this reference system makes it easier to estimate reddening differences than in the original colour - magnitude plane , where the reddening vector is inclined .
then , we generate a fiducial line along the most - populated red ms and , among the stars in the red ms , we select a sample of reference stars with high - precision photometry . for each star in the field of view
, we select a sample of 50 neighbour reference stars and we calculate the residual colours with respect to the fiducial line .
the median of these residuals , calculated along the reddening direction , is assumed as the differential reddening corresponding to each star .
each reference star is excluded in the computation of its own differential reddening .
the results of the differential reddening correction are shown in fig.[f : redmap ] , where we compare a zoom - in portion of the ms for the original cmd ( left panel ) of ngc1850 and the cmd corrected for differential reddening ( middle panel ) .
the two cmds are almost identical , suggesting that the impact of differential reddening is negligible and that the ms split is real .
as a further proof , we show in the right panel of fig.[f : redmap ] the spatial distribution of the differential reddening in our field of view .
each star is reported with a different colour , depending on the final @xmath11 that we applied . in order to better visualize how the differential reddening impacts the area of the cluster field
, we report in fig.[f : redmap ] a circular region , with a radius @xmath12 .
fig.[f : redmap ] clearly shows that the reddening variations in the ngc1850 field are very small , of the order of @xmath13 mag in @xmath14 .
the technique that we applied is able to minimize the amount of differential reddening in the foreground of the cluster .
we find a very small reddening variation across the field of view , and with a very specific spatial pattern that is completely uncorrelated to the h@xmath1 filaments we see in fig.[f : fov ] .
this implies that these filaments are located behind the cluster .
finally , since the reddening variation is negligible , we use the original photometry in the analysis for simplicity .
to understand the nature of the double ms and extended msto in ngc1850 , we compare the cmd with model isochrones .
we investigate the possibility that these features are caused by the two main scenarios presented in the literature : the presence of an age spread in the cluster or of a range of rotation velocities among the cluster stars .
isochrone fitting is performed following the method described in detail in @xcite , and adopted also in c15 , for the case of star clusters that do not yet have a developed rgb . in the following sections , we show the results of the isochrone fitting for the two aforementioned scenarios . to test whether an age spread can reproduce the observed cmd morphology , we superpose two isochrones from parsec @xcite of different age ( @xmath0 90 myr and @xmath0 60 myr ) on the _ f814w _ vs. _ f275w
f814w _ cmd shown in the left panel of fig.[f : cmdage1 ] ( red and blue lines , respectively ) .
the age of the isochrones are chosen to match the maximum and minimum age that can be accounted for by the position and width of the msto .
note ( a ) that the morphology of the msto is not very well reproduced by a ssp and that an age spread of the order of 30 myr may constitute a better fit to the data in this region of the cmd and ( b ) that the observed double ms can not be reproduced at all using a combination of different ages .
interestingly , we note that if we adopt the same parameters in terms of age , metallicity , distance , and reddening and we superpose the isochrones in the _
f467 m _ vs. _ f467m f814w _ cmd ( shown in the right panel of fig.[f : cmdage1 ] ) we see that the selected isochrones do not fit the cmd ms , the isochrones being too red with respect to the observed ms ( note that due to the shorter colour baseline , the split is not observable in this cmd ) . to verify that this discrepancy is not due to systematic uncertainties in the isochrone colour transformations , we also perform the inverse approach .
we use the _
f467 m _ vs. _ f467m f814w _ cmd ( left panel of fig.[f : cmdage2 ] ) to derive the isochrone parameters and then we overplot them onto the _ f814w _ vs. _ f275w f814w _ cmd ( right panel of fig.[f : cmdage2 ] ) .
the latter cmd shows that also in this case a combination of isochrones with different ages provides a satisfactory fit of the msto . and , importantly , _
both isochrones match the less - populated blue ms of the cluster very well_. this is a clear indication that the red ms , containing the majority of stars , can not be reproduced using non - rotating isochrones with different ages and thus _ must be constituted by rotating stars_. 60 and @xmath0 90 myr , respectively ) that can be accounted for by the data are superposed on the cluster cmd , along with the derived metallicity , distance modulus @xmath15 and reddening @xmath14 .
right panel : _ f467 m _ vs. _ f467m f814w _ cmd .
isochrones with the same parameters as in the left panel are overplotted on the cluster cmd . ] as mentioned above , the alternative scenario proposed to explain the emsto phenomenon suggests that it is caused by a spread in rotation velocity among to stars . to test this scenario , we use the isochrones from the geneva syclist database @xcite for which different rotation rates are available .
the left panel of fig.[f : cmdrot ] shows the _ f814w _ vs. cmd with superimposed the isochrones for different rotation rates . in detail
, the red isochrone has no rotation , i.e. @xmath16 = 0.0 , where @xmath17 represents the critical ( break - up ) rotation rate , whereas isochrones with @xmath16 = 0.30 , 0.50 , 0.80 , 0.90 , and 0.95 are reported in magenta , yellow , green , cyan , and blue , respectively .
the non - rotating isochrone has been extended to masses m @xmath8 1.7 m@xmath18 by using the models of ( * ? ? ?
* reported in the cmd as a red dashed line ) . due to slightly different input physics in the two models ,
a small color mismatch is present between the two non - rotating isochrones . to account for it , we shifted the @xcite non - rotating isochrone toward the blue by 0.06 mag . rotating isochrones with m @xmath8 1.7 m@xmath18 are not ( yet ) available . the adopted age , distance modulus , and reddening are also reported in the left panel of fig.[f : cmdrot ] . for what concerns metallicity ,
only three options are provided by the geneva database for the rotating isochrones and we choose the value most appropriate for our case ( i.e. , @xmath19 ) . . in this case , best - fitting isochrones are derived from the _
f467 m _ vs. _ f467m f814w _ cmd ( left panel ) . the adopted age , metallicity , distance , and reddening are also reported .
isochrones with these values are superimposed on the _ f814w _ vs. _ f275w f814w _ cmd ( right panel ) . ]
a combination of different rotation rates seems to reproduce the ms split and the distribution of stars in the blue loop quite well .
in particular , fast - rotating stars ( @xmath16@xmath20 0.80 ) constitute the more populated red ms , whereas non - rotating / slow rotating stars ( @xmath16@xmath8 0.50 ) constitute the blue ms .
the presence of a clear gap between the two mses and thus the absence of intermediate - velocity rotators ( 0.50 @xmath8@xmath16@xmath8 0.80 ) indicates that the distribution of rotation velocities is somewhat bimodal , rather than continuous .
fig.[f : cmdrot ] also shows that the fit in the msto is not satisfactory .
this can be better appreciated in the right panels of fig.[f : cmdrot ] , in which we show zoom - in views of the msto ( top panel ) and the ms in the surroundings of the split ( bottom panel ) . in particular , in the msto region ,
while the `` spread '' in _ f814w _ magnitude is reproduced by a combination of isochrones of different rotation rates , the spread in _ colour _ is larger in the observed cmd with respect to the one produced by the isochrones .
this is further addressed below .
to further determine to what extent a range in rotation velocities can reproduce the observed cmd , and in particular the observed msto morphology , we conduct monte carlo simulations of synthetic clusters with a single age and different stellar rotation velocities ( the method adopted to produce these simulations is described in detail in * ? ? ?
* ; * ? ? ?
briefly , we simulate a ssp with a given age and chemical composition populating the non - rotating and rotating isochrones .
stars are randomly drawn using a salpeter mass function and normalized to the observed ( completeness - corrected ) number of stars . as a first guess
, we arbitrarily populated the different isochrones , in terms of @xmath16 , such that the relative numbers of stars reflect the distribution of @xmath16 for a - type stars in the solar neighbourhood from @xcite .
after inspection of the results , we modified the relative numbers of stars so as to better reproduce the observed cmd morphology , that is to match the observed ratio between the blue and the red ms ( i.e. , we increased the relative numbers of stars for the fast - rotating populations with @xmath16 @xmath21 0.80 ) .
the adopted final values are the following : @xmath10 10% for @xmath16 = 0.00 , @xmath10 5% for @xmath16 = 0.30 , @xmath10 10% for @xmath16 = 0.50 , @xmath10 35% for @xmath16 = 0.80 , @xmath10 30% for @xmath16 = 0.90 , and @xmath10 10% for @xmath16 = 0.95 .
a component of unresolved binary stars , derived from the same mass function , is added to a fraction of the sample stars .
we use a flat distribution of primary - to - secondary mass ratios and we adopt a binary fraction of @xmath10 40% . for the purposes of this work , the results do not change significantly within @xmath22 20% of the binary fraction . finally , we add photometric errors with a distribution derived during the photometric reduction process .
we also obtained syclist cluster simulations from their website for comparison purposes .
simulations include gravity and limb darkening effects @xcite as well as a random distribution of inclination angles ( rotation velocity distribution from * ? ? ?
we found that the msto morphologies of the two sets of simulations are fully consistent with one another ( see sect.[s : pseudo ] below and right panel of fig.[f : cmdsim1 ] ) .
we derive four different sets of synthetic cmds , with the only difference being the adopted age .
in particular , we simulate four ssps with log(age ) values of 7.95 , 8.00 , 8.05 , and 8.10 . to compare in detail the observed msto region with the simulated ones , we create pseudo - colour distributions .
briefly , we construct a parallelogram across the msto ( shown in the left panel of fig.[f : cmdsim1 ] ) and we select the stars within it . we define the distributions with the term `` pseudo - colour '' because it reflects the star distribution along the major axis of the parallelogram , rather than the _ f275w
f814w _ colour .
then , we calculate the pseudo - colour distributions using the non - parametric epanechnikov - kernel density function @xcite .
this is done to avoid possible biases that can arise if fixed bin widths are used .
we apply the procedure to both the observed and the simulated cmd .
the results obtained from the comparison of the observed and simulated pseudo - colour distributions are described in the following sections .
the left panel of fig.[f : cmdsim1 ] compares the observed and simulated cmd , the latter obtained as described above , adopting a single ssp with an age of @xmath0 90 myr ( log(age ) = 7.95 ) and a distribution of rotation rates that fits the morphology of the ms split .
as expected from the isochrone comparison described in the previous section , this simulation reproduces the ms split quite well .
conversely , the msto region of the simulation does not reproduce the observed msto region well , the latter being wider than the former .
this is highlighted in the right panel of fig.[f : cmdsim1 ] , in which we compare the observed and simulated pseudo - colour distributions . indeed , the observed distribution ( shown as a black line ) is wider than the simulated one ( red line ) ; in particular , while the distributions are similar in the left half ( i.e. , blue side ) of their respective profiles , the observed one extends towards a significantly redder colour than the simulated ones .
the same results are obtained comparing the observed distribution with that obtained from the syclist simulation ( reported as a blue line in the right panel of fig.[f : cmdsim1 ] ) .
hence , fig.[f : cmdsim1 ] suggests that the inclusion of gravity and limb darkening effects , as well as the inclusion of a random viewing angle , can not account for the observed msto morphology , since the width and shape of the two simulated pseudo - colour distributions are almost identical . to verify that the obtained results do not depend on the adopted age , in the top panels of fig.[f : cmdsim2 ] , we compare the observed cmd with the other three ssp simulations which have older ages .
the comparison between the observed and simulated pseudo - colour distributions for these ssps is shown in the bottom panels .
this comparison clearly shows that , as expected for older ages , the peak of the simulated distribution moves towards the right ( i.e. , towards redder colours ) , but , independently of the adopted age , the observed distribution is always wider than the simulated ones .
thus , it seems that a range of rotation velocities , which can reproduce the ms split , can not explain the observed msto morphology . in this context , we test the possibility that the emsto can be caused by a combination of the two effects , that is the presence of an age spread _ and _ a range of stellar rotation velocities for the cluster stars . to test this hypothesis
, we derive the combined pseudo - colour distribution of the 4 ssps . at this stage
, we are only interested in verifying whether the combination of a range of ssp ages provide a pseudo - colour distribution width that can match the observed one , and not in a detailed comparison of their global profiles .
hence , we assume that the four ssps have the same number of stars in the cmd , without tailoring the relative weights based on the observed profile .
the left panel of fig.[f : cmdsim3 ] shows the comparison between the observed and simulated cmd , whereas in the right panel we plot the corresponding pseudo - colour distributions .
the right panel clearly shows that the two distributions are similar and that their widths are comparable ( see sect.[s : halpha ] below for a possible explanation of the discrepancy between the two distributions at the red end ) .
it is also worth to note that the obtained distribution is not obviously bimodal and is well approximated by a single gaussian .
this indicates that the profile of the distribution itself can not be used to exclude the presence of an age spread , in contradiction with the claims of @xcite .
hence , the results presented in these sections show that a single ssp with different rotation rates fails in reproducing the observed pseudo - colour distribution of the msto whereas a combination of four ssps with different ages and different rotation rates can provide a better fit .
this suggests that both effects are necessary to reproduce the observed msto region morphology .
one of the original goals of this work was to determine whether ngc1850 hosts _ ongoing _ star formation .
although the results from the analysis of the msto region presented in sect.[s : iso ] and [ s : mc_sim ] seem to exclude such presence , a definitive answer can be provided by performing a search for pre - main sequence stars exploiting narrow - band imaging of the h@xmath1 emission line . to do this
, we use the method adopted in c15 and described in detail in @xcite .
the top panels of fig.[f : excess ] show the _
f467m f656n _ vs. colour - colour diagram for the stars in the cluster field ( top - left panel ) and for the stars in the `` control field '' ( top - right panel ) , the same field we used to derive the background contamination in fig.[f : cmdobs ] . using the median _
f656n _ colour of stars with small photometric uncertainties in each of the three bands as a function of
_ f467m f814w _ , we define the reference sequence with respect to which h@xmath1 emission excess is identified ( shown as the dashed line in fig.[f : excess ] ) .
to select a first candidate sample of stars with h@xmath1 excess emission , we consider all those objects with a _ f467m
f656n _ colour at least 5@xmath23 above the reference sequence , where @xmath23 is the uncertainty of the _ f467m f656n _
colour for each star .
then , we calculate the equivalent width of the h@xmath1 emission line ( @xmath24 ) from the measured colour excess , using the following equation from @xcite : @xmath25 \label{eq : ha_ex}\ ] ] where rw is the rectangular width of the filter , similar in definition to the equivalent width of the line ( for the adopted _
filter , @xmath26 = 27.61 ; see table4 in * ? ? ?
* ) and h@xmath1 - @xmath27 is the difference between the measured h@xmath1 magnitude and the h@xmath1 continuum , represented in this case by the difference between the star _
f656n _ colour and the reference sequence _
f467f656n _ colour , at the corresponding _
f814w_. finally , we keep only the objects for which @xmath28 @xcite .
the stars that satisfy these criteria are reported as red or blue open circles in fig.[f : excess ] , depending on their _
f814w _ colour . excess .
h@xmath29-emitting stars that satisfy the criteria explained in sect.[s : halpha ] are shown as open red or blue circles , depending on their _
f814w _ colour .
bottom panels : cmds of the cluster ( left panel ) and of the control field ( right panel ) . stars with h@xmath1 excess are shown as red or blue open circles as explained above . ] the top panels of fig.[f : excess ] show that for _ f467m
f814w _ @xmath30 0.7 mag , the number of stars with h@xmath1 excess is similar in both fields ( blue open circles ) .
however , there is a large number of objects in the ngc1850 field with _
f467m f814w _
@xmath10 0.00.2 mag ( see red open circles ) , for which there is almost no counterpart in the control field ( only four objects ) . in the lower panels of fig.[f : excess ] , we superimpose these objects on the _ f814w _ vs. _ f275w f814w _ cmds .
note that they are located in the msto region , covering its whole extension .
in particular , we derive that in the parallelogram box used to select msto stars in the previous sections , they represent almost the 40% of the total number of stars .
the presence of h@xmath1 excess in stars with absolute magnitude @xmath31 suggests that these objects are be stars .
the emission lines from these stars arise due to ionized circumstellar material ejected along the poles of rapidly rotating b stars at @xmath32 ( e.g. , * ? ? ?
pending spectroscopic confirmation of these objects as be stars , these results constitute direct evidence that ngc1850 hosts a significant number of fast - rotating stars .
in particular , we note that around @xmath33 _ f275w f814w _ @xmath34 and _ f814w _ @xmath0 18 mag , there is a group of stars with h@xmath1 excess , which are clearly redder than the majority of stars populating the msto region .
we suggest that this may be due to the presence of a dusty circumstellar disk that is oriented in a close to edge - on configuration , thus providing a `` local '' source of reddening that can not be corrected using the differential dereddening technique adopted in this work . to test whether these objects may be responsible for the discrepancy between the observed pseudo - colour distribution and the simulated one ( cf .
sect.[s : pseudo2 ] ) , we adopt the following empirical approach .
we remove all stars that show an h@xmath1 excess ( independently of the derived equivalent widths ) from our ngc1850 catalog . in fig.[f : cmdha ] we show the comparison between the observed cmd ( left panel ) and the one obtained after the stars with h@xmath1 excess are removed from our sample ( right panel ) . for comparison purposes ,
we also superimpose onto both cmds the same simulated single - age cmd as in fig.[f : cmdsim1 ] ( shown as red dots ) . as fig.[f : cmdha ] clearly shows , the removal of h@xmath29-emitting stars results in the removal of almost all stars that were significantly redder than , and detached from , the msto .
then , we derive the pseudo - colour distribution for this subset of stars and compare it with the original one .
this comparison is shown in fig.[f : psha ] .
note that the red tail of the original distribution with pseudo - colour @xmath35 mag is absent in the new one , which is also somewhat narrower than the original one .
however , the comparison with the simulated single - age pseudo - colour distribution ( red line in fig.[f : psha ] ) shows that the latter is still significantly narrower than the observed distribution .
we also show in fig.[f : psha ] the pseudo - colour distribution obtained from the combination of four simulated ssps ( represented by the red dashed line ) . in this case
, the simulated distribution provides a good match to the observed one , better in fact than the original observed one that included the be stars ( cf .
sect.[s : pseudo2 ] ) . however , we note that a small discrepancy still persists at the red end of the distribution
. this seems to indicate that an even older population can be present in the cluster and that the derived age spread of @xmath0 35 myr can be considered as a lower limit .
these results strongly suggest that the population of h@xmath1-emitting stars in the msto region can not account alone for the difference between the observed pseudo - colour distributions and those of the single - age simulations , confirming that rotation alone can not produce the observed morphology of the emsto .
excess ( right panel ) . in both panels
we overplot ( red dots ) the cmd obtained from the monte carlo simulations of a ssp ( the same as in fig.[f : cmdsim1 ] ) .
the parallelogram used to obtain the pseudo - colour distribution is also shown . ]
to investigate whether the population ratio between the two mses ( i.e. , between non - rotating or slow - rotating and fast - rotating stars ) changes as a function of cluster - centric radius we use the following method .
first , we compute a fiducial line for the blue and red ms and we derive the verticalized cmd in such a way that the blue and the red fiducial translate into vertical line with abscissa 0 and 1 , respectively @xcite . to do this
we define for each star @xmath36 $ ] , where _ col _ is _ f275w
f814w_. this particular cmd allows to separate the two mses from each other and from the binary sequence more clearly .
we focus our analysis in the magnitude interval 19.5 @xmath8 _ f814w _ @xmath8 21.0 mag .
in particular , the lower magnitude limit has been chosen to exclude the faint part of the two mses , where they seem to merge together and thus the separation is not clear .
finally , we derive the distribution of stars perpendicular to the ms in five cluster - centric annuli , chosen in such a way that they contain the same number of stars . in the inset of fig.[f : rad ] we show an example of the observed histogram distribution for the third radial interval .
we applied a smoothed nave estimator @xcite to be insensitive to a particular binning starting point . to determine the fraction of blue ms stars with respect to the total number of stars in each region , we adopt a @xmath37 cut in the observed histogram distribution : stars with @xmath38 0.35 mag are associated to the blue ms , whereas stars with 0.35 @xmath39 1.5 mag are associated to the red ms (
the two limits are reported in fig.[f : rad ] as dashed lines ) .
the latter @xmath40 limit has been chosen to avoid the contamination from binary stars . in fig.[f :
rad ] , we show the fraction of blue ms stars with respect to the total number of ms stars as a function of the radial distance from the cluster center .
fig.[f : rad ] shows that this ratio is constant with radius to within the poisson uncertainties . to test the robustness of this result
, we also adopt a different method to derive the fraction of blue and red ms stars in the different cluster - centric annuli .
we use a bi - gaussian function to fit the observed histogram distribution by means of least squares and we infer the fraction of stars in each ms from the areas under the gaussians . the results obtained from the two different methods are consistent within the errors .
excess ( dashed black line ) , the one derived from a ssp ( red line ) , and the one derived from the combination of four ssps ( dashed red line ) .
] taken these results at face value , it seems that the non - rotating / slow - rotating and fast - rotating stars share the same radial distribution .
recently , similar studies have been carried out for other two young star clusters : @xcite studied ngc1856 and found similar results to ours for ngc1850 , that is the two populations share the same radial distribution , whereas in the cluster ngc1866 , @xcite did find a difference in the radial distribution , with the fraction of non - rotating / slow - rotating stars ( blue ms ) increasing slowly outward beyond its effective radius .
on the other hand , @xcite determined the radial distribution of emsto stars in several intermediate - age ( 12 gyr ) massive star clusters .
they find that , for the clusters with the highest escape velocities and longest relaxation times , the stars in the brightest half of the msto region are significantly more centrally concentrated than the stars in the faintest half .
this has been interpreted in terms of an age spread ; the indication that the inner region of the cluster are preferentially populated by second - generation stars is in agreement with the prediction of the _ in situ _
scenario @xcite .
hence , if we assume that the results obtained here are valid also for intermediate - age star clusters ( i.e. , non - rotating and fast - rotating stars have the same radial distributions ) , it seems highly unlikely that the observed different central concentration can be explained in terms of a range of rotation velocities in the massive intermediate - age clusters .
this suggests that the emsto phenomenon in intermediate - age star clusters may be due , at least in part , by a spread in age .
unfortunately , a direct comparison with the radial distribution of msto stars in ngc1850 is hampered by the low number of objects in the cluster msto region .
limits adopted to associate stars to the red and blue ms are reported as dashed lines . ]
one of the important features of the emsto phenomenon in intermediate - age star clusters in the magellanic clouds is that they do not occur in _ all _ such clusters .
a possible explanation of this in the context of multiple populations was provided by @xcite , who suggested that emsto are only hosted by clusters whose escape velocities were higher than the wind velocities of first - generation `` polluters '' stars that are thought to provide the seed material for secondary star formation , during the era when the cluster contained such stars ( this is referred to as the `` early escape velocity threshold '' scenario ) .
the most popular types of such polluter stars are _
( i ) _ agb stars with @xmath41 @xcite , _ ( ii ) _ massive rotating stars ( frms , * ? ? ? * ) . and _ ( iii ) _ massive binary stars @xcite .
subsequently , the studies of @xcite and @xcite suggested that the critical escape velocity for a star cluster to be able to retain the material ejected by first - generation polluter stars seem to be in the range of 1215 kms@xmath42 .
furthermore , the calculations of @xcite show that massive clusters should be able to sweep up significant fractions of their mass from the ambient interstellar medium ( ism ) at young ages if their environment is sufficiently conducive . under the assumptions
that ( 1 ) the line - of - sight velocity dispersion of the lmc ( @xmath43 kms@xmath42 , @xcite ) is a typical relative velocity between a star cluster and the ambient ism in the lmc , and ( 2 ) a typical young cluster in the lmc has a half - mass radius of @xmath44 pc ( @xcite ; c15 ) , equations 58 of @xcite suggest that this can occur in the present - day lmc if the cluster is massive enough ( @xmath45 m@xmath18 ) and the local ism is dense enough but not so dense that it would strip any gas accreted by the cluster by ram pressure in its orbit ( 1 @xmath46 ( n/@xmath47 ) @xmath46 100 for a @xmath48 m@xmath18 cluster ) . with this in mind ,
we determine the structural and dynamical parameters of the cluster as described below .
to determine the radial surface number density distribution of stars , we follow the procedure described by @xcite and adopted in c15 .
we fit the radial surface number density profile using a @xcite model combined with a constant background level , described by the following equation : @xmath49 where @xmath50 is the central surface number density , @xmath51 is the core radius , @xmath52 is the king concentration index ( @xmath53 being the tidal radius ) , and @xmath54 is the equivalent radius of the ellipse ( @xmath55 , where @xmath56 is the semi - major axis of the ellipse and @xmath57 is its ellipticity ) .
fig.[f : king ] shows the best - fit king model , obtained using a @xmath58 minimization routine . the derived core radius @xmath51 = 124 @xmath22 05 , which corresponds to 3.01 @xmath22 0.12 pc , whereas the effective radius @xmath6 = 205 @xmath22 14 ( 4.97 @xmath22 0.35 pc ) .
our estimate for @xmath51 is in reasonable agreement with the 2.7 pc derived by @xcite . [
cols="^,^,^,^,^,^,^,^,^,^ " , ] we estimate the cluster mass and escape velocity as a function of time going back to an age of 10 myr , after the cluster has survived the era of violent relaxation and when the most massive stars of the first generation that are proposed to be candidate polluters in literature ( i.e , frms and massive binary stars ) , are expected to start losing significant amounts of mass through slow winds .
the current mass of ngc1850 is determined from its integrated - light _
i_-band magnitude listed in table[t : param ] and derived as described in c15 . in this particular case , in order to minimize the contamination by ngc1850b , we derive the aperture - photometry magnitudes of stars inside a radius of 20@xmath59 , centered at the coordinates found in the analysis of sect.[s : king ] , measured on the _
f814w drc _ image ( obtained combining long and short exposures ) . the background level and its uncertainty
is derived from five different regions of the image with distances larger than 100@xmath59 from the center of ngc1850 , avoiding the major h@xmath1 emission filaments ( see fig.[f : fov ] ) .
we calculate the escape velocity of ngc1850 and its evolution with time using the prescriptions of @xcite .
briefly , we evaluate the evolution of cluster mass and radius with and without mass segregation .
this property plays a fundamental role in terms of the early evolution of the cluster s expansion and mass - loss rate ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . for the case of a model cluster with initial mass segregation
, we adopt the results of the simulation called sg - r1 in @xcite , which involves a model cluster that features a level of initial mass segregation of @xmath60 = 1.5 , where @xmath61 is the effective radius of the cluster for stars with @xmath62 1 m@xmath18 ( see * ? ? ?
* for a detailed description of the reasons for this choice ) .
escape velocities are calculated from the reduced gravitational potential @xmath63 , at the core radius and at the effective radius .
we choose to calculate the escape velocity at the cluster core radius in accordance with the prediction of the _ in situ _
scenario @xcite , where the second generation stars are formed in the innermost region of the cluster . assuming a @xcite initial mass function , the current cluster escape velocity is 10.5 @xmath22 0.2 kms@xmath42 at the effective radius and 13.8 @xmath22 0.2 kms@xmath42 at the core radius .
for a @xcite initial mass function , we obtain 8.0 @xmath22 0.2 kms@xmath42 and 10.6 @xmath22 0.2 kms@xmath42 , respectively .
fig.[f : escvel ] shows the `` plausible '' escape velocity of ngc1850 as a function of time ( black line ) , derived using the same approach as in c15 and described in detail in @xcite , for a @xcite initial mass function .
briefly , the `` plausible '' escape velocity is calculated adopting a procedure that takes into account the various results from the compilation of magellanic cloud star cluster properties and n - body simulations by @xcite . for comparison fig.[f
: escvel ] also shows the escape velocity as a function of time for the limiting cases in which ( 1 ) the cluster does not have any primordial mass segregation ( blue line ) and ( 2 ) the cluster has the same level of primordial mass segregation as the model sg - r1 by ( * ? ? ?
* red line ) .
the critical escape velocity range of 1215 kms@xmath42 is depicted as the light grey region in fig.[f : escvel ] , whereas the region below 12 kms@xmath42 , representing the velocity range in which emstos are not observed in lmc star clusters , is shown in dark grey . )
whose parameters are shown in the legend .
ellipticity and effective radius @xmath6 are also shown in the legend . ]
fig.[f : escvel ] shows that , independent of the assumed level of primordial mass segregation in the cluster , the initial escape velocity was likely high enough to retain the material ejected through winds by the first - generation polluters
. moreover , it seems to decrease below the critical value of @xmath0 15 kms@xmath42 at an age similar to what we derive as the age spread ( i.e. , 3040 myr ) . in the context of the _ in situ _
scenario , considering the time when first - generation polluters are predicted to be present in the cluster ( i.e. , at ages of @xmath0 530 myr for massive stars and @xmath0 50200 myr for im - agb stars ) and the age ranges derived from our analysis ( i.e. , 3040 myr ) , it seems that massive stars would be the most likely source of `` polluted '' material in ngc1850 .
future investigations of the range of light - element abundances ( e.g. , na or n ) found in ngc1850 should be able to shed more light on the extent to which the cluster has actually retained such polluted material .
furthermore , the mass of ngc1850 is high enough for the cluster to have been able to accrete a significant amount of `` pristine '' gas from its surroundings in the past , which could have constituted an additional source of gas to form stars after the initial burst that formed ngc1850 ( cf.above ; @xcite ; @xcite ; but see also @xcite ) . .
formal @xmath22 1 @xmath23 errors of @xmath64 are shown by the dashed line .
the blue line represents the escape velocity in case ngc1850 had no primordial mass segregation , while the red line does so in case it had the same level of primordial mass segregation as the model sg - r1 in @xcite .
the light grey region represents the critical range of @xmath64 mentioned in sect.[s : dyn ] , i.e. 1215 kms@xmath42 .
the region below 12 kms@xmath42 in which @xmath64 is thought to be too low to permit retention of material shed by the first stellar generation , is shaded in dark grey . ]
we present the results of a study of new , deep _ hst _ wfc3 images of the young ( @xmath0 100 myr ) star cluster ngc1850 in the lmc .
the analysis of its cmd reveal the presence of an emsto and a double ms .
the blue component of the ms hosts about a quarter of the the total number of ms stars .
we verify that these features can not be explained in terms of photometric errors , contamination from background lmc stars , or differential reddening . from a comparison with theoretical models , we show that the ms split can not be reproduced using non - rotating isochrones with different ages .
conversely , it is very well reproduced by a coeval stellar population with a variety of rotation rates .
however , when adopting the latter models , we show that the observed morphology of the msto can not be fit satisfactorily : the pseudo - colour distribution of stars across the msto obtained from the observed cmd is significantly wider than that obtained from monte carlo simulations of the set of single - age isochrones with a range of rotation rates that fit the double ms feature .
a combination of ssps with an age range of @xmath65 myr _ along with _ different rotation rates reproduces the data significantly better . from an analysis of narrow - band h@xmath1 imaging
, we find that the msto region hosts a population of be stars , providing direct evidence that rapidly rotating stars are present in the cluster .
we note that a fraction of these stars are redder than the majority of to stars .
we interpret this in terms of an ( close to ) edge - on configuration of dusty circumstellar disks of be stars , thus providing a source of local reddening that can not be accounted for in our differential reddening correction .
we quantify how this effect impacts the observed pseudo - colour distribution by removing all the stars with an h@xmath1excess from the original catalog and deriving a new pseudo - colour distribution .
we demonstrate that this newly derived pseudo - colour distribution is still significantly wider than the simulated one . as such
, the emsto morphology still can not be explained by an ssp with a variety of stellar rotation rates ; a spread of ages still seems to be required . finally , the dynamical properties of ngc1850 derived from our data show that the cluster had an escape velocity within its core radius of @xmath0 1317 kms@xmath42 , at an age of @xmath0 10 myr , high enough in principle to retain material shed by the slow winds of `` polluter stars '' .
furthermore its mass was high enough to accrete `` pristine '' gas that may have been present in the environment of the cluster shortly after it was first formed .
the important role that stellar rotation can play in the interpretation of the emsto phenomenon in young and intermediate - age star clusters has been acknowledged in the recent literature , including in works that suggested that an age spread is responsible for emstos ( see for example c15 and fig.7 in * ? ? ? * and discussion therein ) .
unfortunately , the current lack of rotating isochrone models for masses m @xmath8 1.7 m@xmath18 prevents us from testing in detail its impact in intermediate - age star clusters . at the same time , the new findings for the young clusters do not yet provide a significant clarification of the picture in terms of the role of stellar rotation in the emsto phenomenon . for example , @xcite concluded that the emsto of intermediate - age star clusters can in principle be explained in terms of stellar rotation , as long as the initial rotation distribution covers the interval from @xmath16 @xmath0 0.0 to @xmath10 0.5 .
conversely , the detailed studies of the split ms of ngc1850 and other young clusters show that they can be reproduced only when assuming that a large number of stars are _ fast _ rotators ( i.e. , @xmath16 @xmath30 0.80 ) .
another open issue is the origin of the non - rotating or slow - rotating stellar populations ( see * ? ? ?
* for a detailed discussion ) .
@xcite suggested that the observed slow - rotating and non - rotating stars could have lost their angular momentum during the pre - ms phase do to magnetic breaking .
an alternative hypothesis is that tidal interaction in binary systems could be responsible for slowing down a fraction of stars . in this context
, @xcite suggested that binary synchronization could be responsible for the slow - rotating population .
spectroscopic studies of the distribution of stellar rotation rates in young and intermediate - age star clusters will be of fundamental importance to test the aforementioned scenarios and address the role of the stellar rotation in the emsto phenomenon .
support for this project was provided by nasa through grant hst - go-14174 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas526555 .
we made significant use of the sao / nasa astrophysics data system during this project .
thp acknowledges support by the fondecyt regular project grant ( n. 1161817 ) and the basal center for astrophysics and associated technologies ( pfb-06 ) . | we use the wide field camera 3 onboard the _ hubble space telescope _ to obtain deep , high - resolution photometry of the young ( @xmath0 100 myr ) star cluster ngc1850 in the large magellanic cloud .
we analyze the cluster colour - magnitude diagram ( cmd ) and find that it hosts an extended main - sequence turn - off ( msto ) and a double ms .
we demonstrate that these features can not be due to photometric errors , field star contamination , or differential reddening . from a comparison with theoretical models and monte carlo simulations ,
we show that a coeval stellar population featuring a distribution of stellar rotation rates can reproduce the ms split quite well . however , it _ can not _ reproduce the observed msto region , which is significantly wider than the simulated ones . exploiting narrow - band h@xmath1 imaging , we find that the msto hosts a population of h@xmath1-emitting stars which are interpreted as rapidly rotating be - type stars .
we explore the possibility that the discrepancy between the observed msto morphology and that of the simulated simple stellar population ( ssp ) is caused by the fraction of these objects that are highly reddened , but we rule out this hypothesis .
we demonstrate that the global cmd morphology is well - reproduced by a combination of ssps that cover an age range of @xmath0 35 myr _ as well as _ a wide variety of rotation rates .
we derive the cluster mass and escape velocity and use dynamical evolution models to predict their evolution starting at an age of 10 myr .
we discuss these results and their implications in the context of the extended msto phenomenon .
[ firstpage ] galaxies : star clusters globular clusters : general magellanic clouds |
we acknowledge quantel laser for the gift of an eylsa laser for our main cooling and trapping light , georgios siviloglou and yuri lensky for experimental contributions , and erich mueller , sayan choudhury , alan jamison , and misha lukin , sankar das sarma , sid parameswaran , ehud altman , and eugene demler for stimulating discussions .
w.c.c . is grateful for the samsung scholarship .
this work was supported by the nsf through grant phy-0969731 , through the center for ultracold atoms , afosr muri grant fa9550 - 14 - 1 - 0035 and aro muri grant no .
w911nf-14 - 1 - 0003 .
the experiment begins with a nearly pure bec confined in a crossed dipole trap in the @xmath25 hyperfine state with @xmath20@xmath26 @xmath27rb atoms .
we turn on a magnetic field gradient to levitate the atoms against gravity and simultaneously weaken the dipole traps by lowering the power to their final values . from here , the non - adiabatic and adiabatic sequences differ . in the nonadiabatic sequence ,
the condensate is adiabatically loaded into a two - dimensional optical lattice composed of one vertical beam and one horizontal beam with lattice constants @xmath28 nm and depth of 11 recoil energies ( @xmath29 ) in both the x- and y - directions , in the presence of a very weak raman lattice ( @xmath30 @xmath29 ) with relative frequency detunings equal to the bloch oscillation frequency @xmath31 hz .
the two beams are beat together on a beam cube located as close to the atoms as our optical setup allows , and the signal is used to stabilize the relative phase of the raman beams .
once a phase lock is achieved , we turn on a large tilt in the regime @xmath32 e@xmath33 by sweeping the frequency of an rf field in 0.29 ms to transfer all the atoms to the @xmath34 hyperfine state . here
@xmath35 is the bare tunneling energy in the lowest band , @xmath36 hz is the energy offset between adjacent sites , and e@xmath33 is the energy gap between the lowest band and the first excited band . upon completion of the rf sweep
, the initial system represents an uncoupled stack of 2d bec s . for topologically trivial superfluids
, resonant tunneling is re - established with amplitude modulation of the vertical ( x - axis ) lattice with modulation frequency @xmath17 .
for the 1/2 flux superfluid resonant tunneling is instead re - established by linearly ramping up both raman beams in @xmath37 ms to @xmath38 , where @xmath39 is the two - photon rabi frequency . after a variable hold time
, all laser beams are switched off to allow 20 ms time - of - flight , followed by absorption imaging . in the adiabatic sequence ,
the condensate enters the mott insulating phase after lattices in all three directions are adiabatically ramped up to 20 @xmath23 .
one of the raman beams is then ramped to its final value in 15 ms , after which the hyperfine state is flipped with a 0.29 ms rf sweep as explained previously .
the lattices are then ramped down to their final values while the second raman beam is ramped up in 35.09 ms , thereby adiabatically connecting the mott insulator to the 1/2 flux superfluid . again
, time - of - flight absorption imaging is performed after a variable hold time .
all authors contributed to experimental work , data analysis , and manuscript preparation .
the authors declare no competing financial interests .
40ifxundefined [ 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 10.1126/science.1227224 [ * * , ( ) ] link:\doibase 10.1126/science.1243326 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.59.1776 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.096407 [ * * , ( ) ] link:\doibase 10.1021/nl303758w [ * * , ( ) ] link:\doibase 10.1126/science.1259327 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.84.806 [ * * , ( ) ] @noop * * , ( ) @noop ( ) http://stacks.iop.org/1367-2630/5/i=1/a=356 [ * * , ( ) ] http://stacks.iop.org/0295-5075/93/i=2/a=20003 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) \doibase doi:10.1038/nphys2750 [ * * , ( ) ] link:\doibase 10.1038/nature13915 [ * * , ( ) ] http://stacks.iop.org/0370-1298/68/i=10/a=304 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.14.2239 [ * * , ( ) ] link:\doibase 10.1038/nphys3171 [ * * , ( ) ] @noop ( ) , link:\doibase 10.1103/physreva.90.043613 [ * * , ( ) ] link:\doibase 10.1103/physreva.91.033601 [ * * ( ) , 10.1103/physreva.91.033601 ] http://arxiv.org/abs/1410.4576 [ ( ) ] , @noop * * , ( ) @noop * * , ( ) @noop ( ) , link:\doibase 10.1103/physreva.83.013612 [ * * ( ) , 10.1103/physreva.83.013612 ] @noop * * , ( ) link:\doibase 10.1103/physreva.87.033614 [ * * ( ) , 10.1103/physreva.87.033614 ] , link:\doibase 10.1103/physrevlett.104.200403 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1126/science.1227831 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.105701 [ * * , ( ) ] link:\doibase 10.1103/physreva.76.055601 [ * * , ( ) ] http://arxiv.org/abs/1412.7615 [ ( ) ] , link:\doibase 10.1103/physrevlett.111.225301 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1088/0034 - 4885/77/12/126401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.185301 [ * * , ( ) ]
we quantify the visibility of the 1/2 flux superfluid diffraction pattern above the broad , thermal background by fitting one - dimensional line cuts of the column density in both the x- and y - lattice directions . to design a peak visibility fitter for the hh superfluid
, one needs to account for multiple degrees of freedom associated with the new peaks that emerge , in addition to the trivial superfluid peaks .
in addition , the uncontrolled phase of the micromotion leads to changes in the diffraction pattern , as seen in the paper . this is because the relative optical path length between the cubic lattice and the running wave raman lattice at the position of the atoms is unstabilized .
when designing a fitter , we must build in the flexibility in handling shot - to - shot changes in the peak locations on the ccd camera . the fitter works by finding the center of mass of the image and taking the average of several rows of pixels in each direction around this center value .
the total visibility is determined by least - squares fitting both the x and y line cuts to the sum of six gaussians one broad thermal background and five sharp diffraction peaks .
the visibility of a given peak is defined as the amplitude of the coherent peak divided by the coherent amplitude plus the twice thermal background at the peak center .
this is in analogy with michelson s definition of optical fringe contrast .
we define the final visibility as the average of visibility of the peaks in both the x- and y - directions .
reliable fitting in such a high dimensional parameter space is robust when the fitting routine starts with accurate initial guesses . to generate accurate initial conditions , the image is first low pass filtered to remove spatial frequencies corresponding to the diffraction peaks and higher , and then fitted to a single gaussian function .
the coefficients resulting from this fit become the initial conditions of the thermal fraction in the final fit .
next , the guess for the thermal component is subtracted from the raw lineout and all remaining local maxima are identified with a peak finding algorithm .
the peaks are sorted by amplitude , and the highest five locations are chosen to represent the initial position and amplitude in the initial conditions for the superfluid diffraction peaks . with these input conditions ,
nonlinear least - squares fitting of the six gaussian peaks gave reproducible fits with small residuals .
an example of a typical fit is shown in figure 1 . to quantify the population imbalance in the two minima of the hh spectrum
, we developed a band mapping technique , in which laser beams are not switched off instantly but instead ramped down to zero smoothly before time - of - flight absorption imaging .
the requirements for band mapping in the @xmath40 hh lattice are different from the standard bandmapping technique in two ways .
first , the two hofstadter sub - bands are connected at dirac points and the total band gap is therefore zero , forbidding a globally adiabatic band mapping procedure . since the atoms are condensed , they are sufficiently localized to quasimomentum states that are far away from the dirac points and are locally gapped .
second , there is a strong potential gradient present , leading to bloch oscillations in the lab frame . for low lattice depths , the bloch oscillations would cause landau - zener transitions at the zone edge .
this effect is minimized by ramping down the lattice relatively quickly , in about one bloch oscillation time . in the case of topologically trivial superfluids ,
all the lattices are linearly ramped down in @xmath41 ms . in the case of hh superfluids ,
raman beams are linearly ramped down in @xmath42 ms first to avoid excitations associated with the raman drive , and then the lattices are linearly ramped down in @xmath41 ms .
[ [ superfluid - diffraction - close - to - the - suerpfluid - to - mott - insulator - transition ] ] superfluid diffraction close to the suerpfluid to mott insulator transition ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ as mentioned in the main text , we observe weak diffraction peaks in three - dimensional lattices for z - lattice depths up to 20 @xmath23 with x and y lattices held fixed at 11 @xmath23 , demonstrating bose - einstein condensation in the hh lattice with strong interactions .
the images in figure 2 represent a selection of shots immediately after the @xmath2035 ms ramp out of the mott insulator state into the 1/2 flux superfluid .
the doubling structure is clearly visible at lattice depths close to the mott insulating transition for filling factors @xmath43 .
[ [ collisional - instability - of - the - ground - state - in - tilted - and - driven - lattices ] ] collisional instability of the ground state in tilted and driven lattices ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the observed decay of a bose - einstein condensate in tilted and driven lattices is much faster than the decay in standard lattices .
although our lifetime seems to be limited largely by technical noise , we wish to discuss the limit given by the time dependent nature of the single - particle wave function representing the ground state of the floquet hamiltonian @xcite .
_ study the stability of a condensate in a tilted one - dimensional lattice with resonant amplitude modulation using a high frequency perturbation analysis .
the ground state of bosons in the corresponding floquet hamiltonian is a superfluid in a geometry with a lattice in one direction and free particle motion in the other two .
the description of this system is simple yet nontrivial due to the metastability of the wannier - stark ladder and the time - modulated many - particle floquet hamiltonian . as pointed out in @xcite , one consequence of these two facts
is that elastic collisions between atoms will deplete the condensate by using one or two quanta of the micromotion ( equal to the offset energy @xmath17 between neighboring sites ) to create excitations .
this will cause decoherence and relaxation to higher temperatures .
there are two main processes contributing to this instability of the condensate .
wannier - stark states which are mostly localized in a given well @xmath44 have small components in lower wells @xmath45 , with amplitude @xmath46 .
this small component , which has energy @xmath17 higher than the wannier - stark state localized in the well @xmath45 , can collide with atoms in the lower lattice site , thus creating two excitations with energy @xmath47 in the transverse direction ( the creation of two exitations is necessary for conservation of transverse momentum ) .
the second mechanism involves the counter - rotating term of the phonon or photon assisted tunneling process .
the co - rotating term enables resonant tunneling to a lower lattice site by stimulated emission of a phonon . in the counter - rotating process , the atoms tunnels to a lower lattice site , but absorbs a lattice phonon .
therefore , it has now an excess energy of @xmath48 which can create two transverse excitations where each of them has energy @xmath17 .
these two - body elastic collisions convert modulation or tilt energy into motion along the transverse directions .
the characteristic relaxation times of the two processes are given by @xmath49 and @xmath50 : @xmath51 @xmath52 we can associate @xmath49 with a process that occurs even in the absence of modulation the rate is proportional to the tunnel coupling squared between neighboring sites , whereas the rate of the @xmath50 process is proportional to the square of the phonon induced tunneling rate .
both decay processes are rather general and should apply to both tilted lattices and lattices modulated by superlattices . as a test of these predictions , we implemented this hamiltonian experimentally using the same experimental procedure described in the main text for resonant amplitude modulation .
we measured the coherence and lower band population lifetimes as shown in figure 3 .
approximate values in our experiment were @xmath53 , with @xmath54 as the chemical potential , @xmath55 the s - wave scattering length , @xmath56 laser - assisted tunnelling rate , @xmath35 the bare tunnelling rate , and @xmath17 the energy tilt per lattice site .
@xmath57 is the size of the wannier statein the x - direction @xmath58 .
these give us the values of @xmath59 and @xmath60 , which are longer than our observed lifetime.uncertainty is driven by uncertainty in our density measurements due to redistribution during lattice ramp - up .
however , during the long coherence lifetime , significant population loss was observed ( due to inelatic two- or three - body collisions , or transfer to higher bands ) , and detailed further study is needed to identify the contributions of different processes to the decay rate . the mechanism proposed by choudhury _
_ appears to give a reasonable upper limit for collisional decay in 1d lattices .
we expect the same mechanism to explain at least part of the decoherence observed for both amplitude modulation in a two - dimensional lattice and the hh hamiltonian . for a quantitative comparison , we modify the calculation in @xcite to reflect that a 2d lattice has only one orthogonal direction of free motion . for resonant modulation , in the floquet basis , the hamiltonian is @xmath61 with @xmath62 , v as the system volume , and @xmath63 and @xmath64 as the annihiliation and creation operators for states with quasimomentum @xmath65 .
the interaction parameter is given by @xmath66 where @xmath58 and @xmath67 are the wannier functions in the tilted ( x ) and untilted ( y ) lattice directions , @xmath68 is the lattice spacing , and @xmath69 is the size of the wannier state . the instantaneous single particle dispersion @xmath70 is given by @xmath71 where @xmath72 and @xmath73 are the bare tunnelling rates in the x and y directions , and @xmath56 is the effective tunnelling in the tilt direction . following a similar argument , we find that the scattering rate is given by @xmath74 @xmath75 @xmath76 with the effective dispersion : @xmath77 + 2j_y[1-\cos(k_y)]+\sqrt{\frac{gn}{m } ( \hbar k_z)^2+\left(\frac{(\hbar k_z)^2}{2 m}\right)^2}.\ ] ] note that ref .
@xcite uses a hartree - fock excitation energy with free particle dispersion .
the high mean field energy in the one dimensional tubes created by a two dimensional lattice , @xmath54 , is similar in energy scale to the tilt frequency , and we thus use the full bogoliubov spectrum for the energy of quasi - particles . for our system ,
representative numbers are @xmath84 , and @xmath85 , @xmath86 , @xmath87 , @xmath88 , @xmath89 , and @xmath90 , resulting in @xmath91 and @xmath92 .
this is much smaller than the decay rates measured in the experiment .
by monitoring the loss of atoms we can show that certain processes are negligible for the loss rate of the superfluid state .
figure 4 shows the lifetime of the population in the lowest band of our lattice or equivalently the total atom number lifetime , since the strong tilt will quickly remove population promoted to higher bands due to landau - zener tunneling to the continuum .
the lifetime for atom loss from the lower band is about an order - of - magnitude longer than the coherence lifetime and shows a slight decrease with increasing lattice depth . for comparison , a bec in an untilted lattice exhibits coherence times exceeding one second in deep lattices . we conclude that the population loss seen in figure 4 is dominated by some process beyond heating in standard lattices ( spinflip collisions , magnetic field noise , lattice intensity noise , and beam pointing noise ) .
however , the dominant decay process for the bec in the hh lattice are excitations within the lowest band
. a 3d lattice should be helpful for reducing collisional heating .
qualitatively , in a fermi s golden rule picture of heating , the decay rate depends on the dispersion relation of the transverse modes @xcite .
thus , if a 3d lattice has a bandgap at the energies @xmath17 and @xmath47 , at which quasiparticles can be created , we would expect to see a `` gapping out '' of collisional heating and a corresponding increase in the lifetime upper limit . since we do not see a qualitative change in lifetime when changing the bandgap
, we conclude that , similarly to a 2d lattice , technical fluctuations are the likely limit of the lifetime . 2ifxundefined [ 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 http://arxiv.org/abs/1410.4576 [ ( ) ] , link:\doibase 10.1103/physreva.91.033601 [ * * ( ) , 10.1103/physreva.91.033601 ] | extensions of berry s phase and the quantum hall effect have led to the discovery of new states of matter with topological properties . traditionally , this has been achieved using gauge fields created by magnetic fields or spin orbit interactions which couple only to charged particles . for neutral ultracold atoms , synthetic magnetic fields
have been created which are strong enough to realize the harper - hofstadter model . despite many proposals and major experimental efforts , so far it has not been possible to prepare the ground state of this system . here
we report the observation of bose - einstein condensation for the harper - hofstadter hamiltonian with one - half flux quantum per lattice unit cell .
the diffraction pattern of the superfluid state directly shows the momentum distribution on the wavefuction , which is gauge - dependent .
it reveals both the reduced symmetry of the vector potential and the twofold degeneracy of the ground state .
we explore an adiabatic many - body state preparation protocol via the mott insulating phase and observe the superfluid ground state in a three - dimensional lattice with strong interactions .
topological states of matter are an active new frontier in physics .
topological properties at the single particle level are well understood ; however , there are many open questions when strong interactions and correlations are introduced @xcite as in the @xmath0 state of the fractional quantum hall effect @xcite and in majorana fermions @xcite . for neutral ultracold atoms
, new methods have been developed to create synthetic gauge fields .
forces analogous to the lorentz force on electons are engineered either through the coriolis force in rotating systems @xcite , by phase imprinting via photon recoil @xcite , or lattice shaking @xcite .
much of the research with ultracold atoms has focused on the paradigmatic harper - hofstadter ( hh ) hamiltonian , which describes particles in a crystal lattice subject to a strong homogeneous magnetic field @xcite . for magnetic fluxes on the order of one flux quantum per lattice unit cell , the radius of the smallest possible cyclotron orbit and the lattice constant are comparable , and their competition gives rise to the celebrated fractal spectrum of hofstadfer s butterfly whose sub - bands have non - zero chern numbers @xcite and dirac points .
so far , it has not been possible to observe the ground state of the hh hamiltonian , which for bosonic atoms is a superfluid bose - einstein condensate .
it is unknown whether this is due to heating associated with technical noise , non - adiabatic state preparation , or inelastic collisions .
these issues are complicated , since all schemes for realizing the hh hamiltonian use some form of temporal lattice modulation and therefore are described by a time - dependent floquet formalism .
the hh model arises after time averaging the floquet hamiltonian , but it is an open question to what extent finite interactions and micromotion lead to transitions between floquet modes and therefore heating @xcite .
bose - einstein condensation has been achieved in staggered flux configurations @xcite and in small ladder systems @xcite further highlighting its noted absence in the uniform field configuration . in this article ,
we report the first observation of bose - einstein condensation in hofstadter s butterfly .
the presence of a superfluid state in the hh lattice allows us to analyze the symmetry of the periodic wavefunction by self - diffraction of coherent matter waves during ballistic expansion
analogous to a von laue x - ray diffraction pattern , which reveals the symmetry of a lattice . using this method
, we show that the unit cell of the magnetic lattice is larger than that of the underlying cubic lattice .
this reflects that the vector potential necessarily has a lower symmetry than a uniform magnetic field , since a transitionally invariant field can only be realized by a vector potential that breaks this translational invariance . in our experiment
, we analyze the wavefunction directly and observe both its non - gauge invariance @xcite and ground state degeneracy .
finally , we explore a many - body adiabatic state preparation protocol where the many - body gap of the mott insulator is used to switch superfluid order parameters without creating excess entropy .
we realize the hh hamiltonian superfluid ground state in a three - dimensional lattice with strong interactions . the hh model with large synthetic magnetic fields
is realized as an effective hamiltonian engineered by laser - assisted tunneling processes in a tilted lattice potential . as in refs .
@xcite , tunneling in the x - direction of a two - dimensional optical lattice is suppressed by an energy offset and then subsequently restored with a resonant raman process with a momentum transfer , @xmath1 .
the z - direction consists of loosely confined tubes of condensate , so interactions are weak . in the time - averaged picture , this gives rise to the hamiltonian : @xmath2 where @xmath3 reflects the specific gauge implemented ( fig .
1a ) .
an atom that travels around a single lattice unit cell picks up a phase of @xmath4 , with @xmath5 being the lattice spacing , so the hamiltonian is equivalent to that of a charged particle in a magnetic field with one half of a magnetic flux quantum per unit cell . after preparing the atoms in the hh lattice ( see methods ) ,
we observe the momentum distribution of the wavefunction by suddenly turning off all laser beams to allow 20 ms time - of - flight , and measure the resulting density distribution with absorption imaging ( fig .
1h - j ) .
we first note that the images show sharp peaks , which are the hallmark of a superfluid bose - einstein condensate in a periodic potential .
this demonstrates for the first time that we have successfully prepared a low entropy state in the bulk hh hamiltonian .
in addition , the diffraction images directly show a reduced symmetry of the superfluid , despite the translational symmetry of both the lattice and the homogeneous synthetic magnetic field .
fundamentally , the vector potential is responsible for the broken translation symmetry of the lattice , and the time - of - flight patterns can be understood by examining these symmetries . in real space , there are two relevant unit cells : one is the unit cell of the cubic lattice , and the other is the unit cell of the hamiltonian determined by the experimental gauge , which we call the magnetic unit cell . in highly symmetric gauges such as our experimental gauge and the landau gauge
a magnetic flux of @xmath6 has a magnetic unit cell that is @xmath7 times larger than the original unit cell and contains @xmath7 indistinguishable sites @xcite . in momentum space ,
these unit cells correspond to the brillouin zone of the underlying lattice and to the magnetic brillouin zone , respectively . due to the indistinguishability of the sites in the magnetic unit cell , every quasimomentum state within the magnetic brillouin zone
is @xmath7-fold degenerate with other quasimomentum states connected by the translation symmetry operators of the original lattice modified by a phase @xmath8 .
this modified translation symmetry connects all degenerate ground states in the magnetic brillouin zone , and corresponds to a third relevant scale @xmath9 useful for understanding the diffraction pattern of a superfluid in the hh hamiltonian .
this length scale defines the doubly reduced brillouin zone , which is the primitive cell of the band structure @xcite .
experimentally , we observe the gauge - dependent diffraction pattern directly when the condensate occupies only one minimum of the band structure ( fig .
1h ) .
the pattern s symmetry reflects that of the gauge potential ( fig .
1d ) , directly demonstrating the symmetry of the reciprocal lattice vectors @xmath10 of the magnetic brillouin zone .
a common belief is that all observables are gauge - independent .
however , gauge - dependent observations can be made in time - of - flight images of ultracold atoms when the momentum distribution of the wavefunction is observed .
the sudden switch - off of all laser beams transforms canonical momentum , which is gauge dependent , into mechanical momentum , which is readily observed @xcite . the diffraction pattern shown in fig .
1i reveals the symmetry of the band structure ( fig .
1b ) .
in the experimental image , both degenerate states are equally populated and the peaks can be regarded as the sum of the diffraction pattern in fig . 1h with an additional copy separated by the gauge - invariant scale @xmath11 in either @xmath12 or @xmath13 directions .
the diffraction pattern reflects this two - fold multiplicity of momentum peaks in both the x and y directions and features a four - fold increase of visible peaks @xcite . to the best of our knowledge ,
this work is the first direct observation of the different symmetries characterizing a simple lattice with a homogeneous magnetic field : figs .
1g - i show the symmetry of the lattice , the gauge potential , and the band structure .
realization of a bose condensed ground state allows direct observation of both gauge - dependent and gauge - independent symmetry breaking .
a final feature of the observed diffraction pattern is the periodic oscillation of the peaks along the x - axis .
this is the result of micromotion coming from acceleration due to the tilt and raman drive . in the limit of a low drive amplitude , this micromotion is a bloch oscillation in the tilted lattice .
more generally , it is the effect of a time - dependent transformation between the rotating frame in which the hh hamiltonian is realized and the lab frame in which it is observed @xcite . figs .
1e , i and f , j show diffraction patterns for the micromotion at zero phase and at @xmath14 phase equivalent to zero and @xmath15 phases in a bloch oscillation due to the doubling of the unit cell . in the experiment ,
the absolute phase of the micromotion is a random variable due to wavelength scale drifts of the relative positions of the stationary and raman lattices .
qualitatively , almost all images show roughly equal population of both degenerate ground states .
occupation of only one state ( as in fig .
1h ) is observed in less than 1% of the shots . for a quantitative determination of the populations
, we developed a band mapping technique that adiabatically connects the hh ground states to momentum states ( see supplement ) .
band mapping reduces the complicated diffraction patterns of fig . 1 to two peaks , one for each degenerate ground state .
fig .
2b shows that the minima are most frequently populated with equal proportion , demonstrating the degeneracy of the minima and the robustness of the loading procedure to technical fluctuations .
however , mean - field calculations predict that a superposition state , which would create spatial density modulation , is energetically unfavorable due to repulsive interactions @xcite . therefore , the simultaneous occupation of both minima most likely indicates the presence of domains , composed of atoms in only one of the dispersion minima . to rule out non - adiabatic transitions to higher bands as an alternative explanation of the two peaks in band mapping
, we tested the technique on a topologically trivial floquet superfluid ( see methods ) .
this system consists of a cubic lattice with a tilted potential where tunneling in the tilted axis is restored by amplitude modulation ( am ) @xcite .
its micromotion , which is similar to that of the hh system , and its trivial band structure allows identification of non - adiabatic excitations attributed to the micromotion .
the resulting momentum distributions for both the am and the hh systems are shown in figure 2b .
non - adiabatic band excitations would appear as additional peaks in the tilted direction , which are not observed .
therefore , we interpret the band mapping procedure in the hh system as an adiabatic process . 15 ms and 71 @xmath16 8 ms , respectively .
the data were taken after a 10 ms hold time after switching on the final hamiltonian using the non - adiabatic procedure ( see methods ) .
uncertainty is given by the statistical error of the mean of five repetitions of the experiment , added in quadrature to uncertainty in the peak visibility fitting . ]
the above results confirm many aspects of the weakly interacting superfluid ground state of the hh hamiltonian . however , there are many open questions about topological physics in the presence of strong interactions , often involving small energy differences . to study this regime , long coherence lifetimes are necessary . in order to evaluate our coherence lifetime and disentangle different sources of decoherence and heating
, we extensively used the am superfluid as a benchmark .
initially , the hh superfluid had a vastly shorter lifetime than the am superfluid .
subsequent improvements in the stability of magnetic fields , beam pointing , raman phase , and gradient alignment to the lattice direction improved the hh lifetime until both systems had a comparable lifetime in the ground state , given in fig .
3 . the lifetimes for the number of trapped atoms
is much longer than the coherence lifetime in either case , with no discernible loss of atom number up to 500 ms .
this leads to the conclusion that two - body dipolar interactions and three - body recombination , which would lead to particle loss from the trap , are not limiting the lifetime of the ground state .
the same applies to excitations to higher lying bands which , due to the tilt , are strongly coupled to the continuum via landau - zener tunneling .
therefore the decay in fig .
3 is dominated by transitions within the lowest band , which can be caused by technical noise or by elastic collisions that transfer energy from the micromotion into heat .
while our heating seems to be largely technical , it is useful to briefly discuss these collisions as an ultimate limit on the lifetime .
for the tilted lattice , one decay path is via overlap of neighboring wannier - stark states with offset energy @xmath17 , which can be transferred to excitations of the lowest bands , or to the free particle motion along the tubes orthogonal to the 2d lattice .
such processes are described by fermi s golden rule for transitions between floquet states @xcite . using this framework
, we derive the scattering rate for the am superfluid to be between @xmath18 , where the uncertainty comes from the uncertainty in our density measurement due to redistribution during lattice ramp - up .
this estimate is smaller than the observed decay rate by a factor of @xmath19 .
see supplement for a derivation and experimental parameters .
in principle , an adiabatic procedure to prepare the ground state of the hh hamiltonian should enable better control and higher fidelity in the final state .
however , we empirically found that a sudden turn - on of the tilt and the raman beams gave the most consistent high contrast images , and was therefore useful in evaluating technical improvements . a slower turn - on could introduce more technical heating , and the entropy created in the sudden turn - on
could be absorbed by other degrees of freedom .
such an entropy reservoir is provided in the 2d lattice schemes by the third dimension in our case tubes of condensates with @xmath20500 atoms . for extensions to 3d lattices ,
we implement an adiabatic state preparation procedure . to enter the mott insulating phase , after which the hyperfine state is flipped with a 0.29 ms rf sweep as done in the non - adiabatic scheme .
the lattices are then ramped down to their final values while the raman lattice is ramped up in 35 ms .
lifetimes of both methods are given in the top corner of each figure ; the lifetime of the adiabatic approach is measured from the end of the 35 ms ramp .
errors are statistical and given by the exponential fit of the peak visibility . ]
adiabatic methods usually require a pathway where the ground state is protected by a gap . for single particle states , this may involve matching the size of the unit cell between a topologically nontrivial lattice and a trivial lattice @xcite .
instead , we explore the use of many - body gap of the bosonic mott insulator state . since the bose - hubbard model and the hh model with interactions have the same mott insulating ground state for large u ( assuming the high frequency limit @xmath21 ) , one can connect a trivial superfluid in the cubic lattice to the ground state of the hh model via two quantum phase transitions .
the first transition `` freezes '' out the original superfluid by entering the mott insulator where the phases of the tunneling matrix elements can be changed to a non - zero flux configuration without adding entropy @xcite .
the second transition `` unfreezes '' the mott state into the hh superfluid . technically , this approach can not be fully adiabatic due to the exact degeneracy of the ground states in the hh spectrum ; a kibble - zurek model implies that this will lead to the spontaneous formation of domains @xcite .
however , this will not create entropy beyond the randomization of domains .
we implement this scheme to load the ground state of the hh hamiltonian .
so far , we have found the adiabatic turn - on to be less robust against technical noise , leading to a higher shot - to - shot variability ; however , the best images have identical visibility and a marginally better lifetime than that of the non - adiabatic state preparation method ( fig .
4 ) . we can now use the lattice in the third dimension , required for the adiabatic loading protocol , to add stronger interactions to the physics of gauge fields . to do this
, we keep the lattice depths and tunneling strengths along the x- and y - directions constant , but vary the depth of the z - lattice , thereby adjusting the hubbard interaction parameter u from nearly zero to @xmath20450 hz . with an estimated filling factor of @xmath205 , we can access interaction energies in excess of @xmath22 , which is close to the superfluid - to - mott insulator transition in high magnetic fields @xcite .
we realize hh superfluids with lattice depths from 0 to 20 @xmath23 and observe at least weak superfluid diffraction peaks ( see supplement ) . coherence lifetimes for z - lattice depths up to 11 @xmath23 are presented in fig .
5 .
the average visibility of the diffraction pattern is reduced at high z - lattices , but the lifetimes are reduced only by a factor of @xmath24 from the lifetime of the weakly interacting superfluid at 0 @xmath23 . however , we observe much larger shot - to - shot variation : some shots will have clearly visible diffraction peaks , while other attempts show no coherence at all .
we attribute the variation to technical fluctuations , which seem to have an enhanced effect at higher 3d lattices . in conclusion ,
we have observed bose - einstein condensation in a strong synthetic magnetic field .
we have directly observed how the gauge potential breaks the symmetry of the original lattice . to reach the regime of strong interactions ,
we have demonstrated that the hh model can be implemented with reasonably long coherence times in the presence of both strong gauge fields and strong interactions .
therefore , this work lays the groundwork for further investigations of topologically nontrivial and/or highly frustrated states .
straightforward modifications of our setup such as the introduction of complex tunneling in the z - direction or pseudospins realize weyl points @xcite , the quantum spin hall hamiltonian , and ultracold atom topological insulators @xcite . recent technological advances in controlling and detecting ultracold atoms combined with this work offer a promising path toward realization of bosonic laughlin states in optical lattices @xcite .
the ability to realize strong gauge fields with strong interactions highlights the capability of ultracold atom systems to realize exotic states of matter with no known analogues in nature . |
our general aim is to build up relativity theories as theories in the sense of mathematical logic .
so we axiomatize relativity theories within pure first - order logic ( fol ) using simple , comprehensible and transparent basic assumptions ( axioms ) .
we strive to prove all the surprising predictions of relativity from a minimal number of convincing axioms .
we eliminate tacit assumptions from relativity by replacing them with explicit axioms ( in the spirit of the foundation of mathematics and tarski s axiomatization of geometry ) .
we also elaborate logical and conceptual analysis of our theories .
logical axiomatization of physics , especially that of relativity theory , is not a new idea , among others , it goes back to such leading scientists as hilbert , reichenbach , carnap , gdel , and tarski .
relativity theory was intimately connected to logic from the beginning , it was one of the central subjects of logical positivism . for a short survey on the broader literature ,
see , e.g. , @xcite .
our aims go beyond these approaches in that along with axiomatizing relativity theories we also analyze in detail their logical and conceptual structure and , in general , investigate them in various ways ( using our logical framework as a starting point ) .
a novelty in our approach is that we try to keep the transition from special relativity to general relativity logically transparent and illuminating .
we `` derive '' the axioms of general relativity from those of special relativity in two natural steps .
first we extend our axiom system for special relativity with accelerated observers ( sec.[acc - sec ] ) .
then we eliminate the distinguished status of inertial observers at the level of axioms ( sec.[gen - sec ] ) .
some of the questions we study to clarify the logical structure of relativity theories are : * what is believed and why ?
* which axioms are responsible for certain predictions ? * what happens if we discard some axioms ?
* can we change the axioms and at what price ?
our aims stated in the first paragraph reflect , partly , the fact that we axiomatize a physical theory .
namely , in physics the role of axioms ( the role of statements that we assume without proofs ) is more fundamental than in mathematics . among others ,
this is why we aim to formulate simple , logically transparent and intuitively convincing axioms.our goal is that on our approach , surprising or unusual predictions be theorems and not assumed as axioms .
for example , the prediction `` no faster than light motion ... '' is a theorem on our approach and not an axiom , see thm.[thm - noftl ] .
getting rid of unnecessary axioms is especially important in a physical theory .
when we check the applicability of a physical theory in a situation , we have to check whether the axioms of the theory hold or not .
for this we often use empirical facts ( outcomes of concrete experiments ) .
however , these correspond to existentially quantified theorems rather than to universally quantified statementswhich the axioms usually are .
thus while we can easily disprove the axioms by referring to empirical facts , we can verify these axioms only to a certain degree .
some of the literature uses the term empirical fact for universal generalization of an empirical fact elevated to the level of axioms , see , e.g. , @xcite , @xcite .
we simply call these generalizations ( empirical ) axioms .
for one thing , einstein s theory of relativity not just had but still has a great impact on many areas of science .
it has also greatly affected several areas in the philosophy of science .
relativity theory has an impact even on our every day life , e.g. , via gps technology ( which can not work without relativity theory ) .
any theory with such an impact is also interesting from the point of view of axiomatic foundations and logical analysis . since spacetime is a similar geometrical object as space , axiomatization of relativity theories ( or spacetime theories in general )
is a natural continuation of the works of euclid , hilbert , tarski and many others axiomatizing the geometry of space .
there are many examples showing the benefits of using axiomatic method . for example , if we decompose relativity theories into little parts ( axioms ) , we can check what happens to our theory if we drop , weaken or replace an axiom or we can take any prediction , such as the twin paradox , and check which axiom is and which is not needed to derive it . this kind of reverse thinking helps to answer the why - type questions . for details on answering why - type questions by the methodology of the present work , see @xcite , @xcite .
the success story of axiomatic method in the foundations of mathematics also suggests that it is worth applying this method in the foundations of spacetime theories @xcite , @xcite .
let us note here that euclid s axiomatic - deductive approach to geometry also made a great impression on the young einstein , see @xcite . among others
, logical analysis makes relativity theory modular : we can change some axioms , and our logical machinery ensures that we can continue working in the modified theory .
this modularity might come handy , e.g. , when we want to unify general relativity and quantum theory to a theory of quantum gravity .
for further reasons why to apply the axiomatic method to spacetime theories , see , e.g. , @xcite , @xcite , @xcite , @xcite , @xcite .
we aim to provide a logical foundation for spacetime theories similar to the rather successful foundations of mathematics , which , for good reasons , was performed strictly within fol .
one of these reasons is that fol helps to avoid tacit assumptions .
another is that fol has a complete inference system while second - order logic ( or higher - order logic ) can not have one .
still another reason for choosing fol is that it can be viewed as a fragment of natural language with unambiguous syntax and semantics .
being a _ fragment of natural language _ is useful in our project because one of our aims is to make relativity theory accessible to a broad audience . _ unambiguous syntax and semantics _
are important , because they make it possible for the reader to always know what is stated and what is not stated by the axioms .
therefore they can use the axioms without being familiar with all the tacit assumptions and rules of thumb of physics ( which one usually learns via many , many years of practice ) . for further reasons why to stay within fol
when dealing with axiomatic foundations , see , e.g. , ( * ? ? ?
* appendix : why fol ? ) , @xcite , @xcite , @xcite , @xcite .
before we present our axiom system let us go back to einstein s original ( logically non - formalized ) postulates .
einstein based his special theory of relativity on two postulates , the principle of relativity and the light principle : `` the laws by which the states of physical systems undergo change are not affected , whether these changes of state be referred to the one or the other of two systems of coordinates in uniform translatory motion . '' and `` any ray of light moves in the ` stationary ' system of co - ordinates with the determined velocity @xmath0 , whether the ray be emitted by a stationary or by a moving body . '' , see @xcite .
the logical formulation of einstein s principle of relativity is not an easy task since it is difficult to capture axiomatically what `` the laws of nature '' are in general .
nevertheless , the principle of relativity can be captured by our fol approach , see @xcite , @xcite . instead of formulating the two original principles , we formulate the following consequence of theirs : `` the speed of light signals is the same in every direction everywhere according to every inertial observer '' ( and not just according to the ` stationary ' observer ) . here
we will base our axiomatization on this consequence and call it light axiom .
we will soon see that the light axiom can be regarded as the key assumption of special relativity .
since we want to axiomatize special relativity , we have to fix some formal language in which we will write up our axioms .
let us see the basic concepts ( the vocabulary " of the fol language ) we will use .
we would like to speak about motion .
so we need a basic concept of things that can move . we will call these object _
bodies_. the light axiom requires a distinguished type of bodies called _ photons _ or _
light signals_. we will represent motion as the changing of spatial location in time . thus we will use reference frames for coordinatizing events ( meetings of bodies ) .
time and space will be marked by _
the structure of quantities will be an _ ordered field _ in place of the field of real numbers . for simplicity , we will associate special bodies to reference frames .
these special bodies will be called `` observers .
'' observations will be formalized / represented by means of the _ worldview relation_. to formalize the ideas above , let us fix a natural number @xmath1 for the dimension of spacetime . to axiomatize theories of the @xmath2-dimensional spacetime , we will use the following two - sorted fol language : @xmath3 where @xmath4 ( bodies ) and @xmath5 ( quantities ) are the two sorts , @xmath6 ( inertial observers ) and @xmath7 ( light signals or photons ) are one - place relation symbols of sort @xmath4 , @xmath8 and @xmath9 are two - place function symbols of sort @xmath5 , and @xmath10 ( the worldview relation ) is a @xmath11-place relation symbol the first two arguments of which are of sort @xmath4 and the rest are of sort @xmath5 .
atomic formulas @xmath12 and @xmath13 are translated as `` _ _
@xmath14 is an inertial observer _ _ , '' and `` _
_ @xmath15 is a photon _ _ , '' respectively . to speak about coordinatization
, we translate @xmath16 as `` _ _ body @xmath14 coordinatizes body @xmath17 at space - time location @xmath18 _ _ , '' ( i.e. , at space location @xmath19 and at instant @xmath20 ) .
sometimes we use the more picturesque expressions _ sees _ or _ observes _ for _ coordinatizes_. however , these cases of `` seeing '' and `` observing '' have nothing to do with visual seeing or observing ; they only mean associating coordinate points to bodies .
the above , together with statements of the form @xmath21 are the so - called _ atomic formulas _ of our fol language , where @xmath22 and @xmath23 can be arbitrary variables of the same sort , or terms built up from variables of sort @xmath5 by using the two - place operations @xmath9 and @xmath8 .
the _ formulas _ are built up from these atomic formulas by using the logical connectives _ not _
( @xmath24 ) , _ and _ ( @xmath25 ) , _ or _ ( @xmath26 ) , _ implies _ ( @xmath27 ) , _ if - and - only - if _ ( @xmath28 ) and the quantifiers _ exists _ ( @xmath29 ) and _ for all _ ( @xmath30 ) .
for the precise definition of the syntax and semantics of fol , see , e.g. , @xcite . to meaningfully formulate the light axiom
, we have to provide some algebraic structure for the quantities .
therefore , in our first axiom , we state some usual properties of addition @xmath8 and multiplication @xmath9 true for real numbers . : : the quantity part @xmath31 is a euclidean field , i.e. , + @xmath32 @xmath33 is a field in the sense of abstract algebra , + @xmath32 the relation @xmath34 defined by @xmath35 is a linear ordering on @xmath5 , and + @xmath32 positive elements have square roots : @xmath36 .
the field - axioms ( see , e.g. , @xcite ) say that @xmath8 , @xmath9 are associative and commutative , they have neutral elements @xmath37 , @xmath38 and inverses @xmath39 , @xmath40 respectively , with the exception that @xmath37 does not have an inverse with respect to @xmath41 , as well as @xmath9 is additive with respect to @xmath8 .
we will use @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath42 as derived ( i.e. , defined ) operation symbols .
@xmath43 is a mathematical " axiom in spirit .
however , it has physical ( even empirical ) relevance .
its physical relevance is that we can add and multiply the outcomes of our measurements and some basic rules apply to these operations .
physicists usually use all properties of the real numbers tacitly , without stating explicitly which property is assumed and why .
the two properties of real numbers which are the most difficult to defend from an empirical point of view are the archimedean property , see @xcite , @xcite , and the supremum property , see the remark after the introduction of axiom @xmath44 on p .. euclidean fields got their name after their role in tarski s fol axiomatization of euclidean geometry @xcite .
by @xmath43 we can reason about the euclidean structure of a coordinate system the usual way , we can introduce euclidean distance , speak about straight lines , etc .
in particular , we will use the following notation for @xmath45 ( i.e. , @xmath46 and @xmath47 are @xmath48-tuples over @xmath5 ) if @xmath49 : @xmath50 we will also use the following two notations : @xmath51 for the _ space component _ and the _ time component _ of @xmath52 , respectively . now let us see how the light axiom can be formalized in our fol language . : : for any inertial observer , the speed of light is the same in every direction everywhere , and it is finite .
furthermore , it is possible to send out a light signal in any direction .
formally : @xmath53 axiom @xmath54 has an immediate physical meaning .
this axiom is not only implied by the two original principles of relativity , but it is well supported by experiments , such as the michelson - morley experiment .
moreover , it has been continuously tested ever since then .
nowadays it is tested by gps technology .
axiom @xmath54 says that `` it is _ possible _ for a photon to move from @xmath46 to @xmath47 iff ... '' .
so , a notion of possibility plays a role here . in the present paper we work in an extensional framework , as is customary in geometry and in spacetime theory .
however , it would be more natural to treat this possibility phenomenon " in a modal logic framework , and this is more emphatically so for relativistic dynamics @xcite . it would be interesting to explore the use of modal logic in our logical analysis of relativity theory
. this investigation would be a nice unification of the works of imre ruzsa s school on modal logic and the works of our tarskian spirited school on axiomatic foundations of relativity theory .
robin hirsch s work can be considered as a first step along this road @xcite .
let us note that @xmath54 does not require that the speed of light be the same for every inertial observer or that it be nonzero .
it requires only that the speed of light according to a fixed inertial observer be a quantity which does not depend on the direction or the location .
why do we not require that the speed of light is nonzero ?
the main reason is that we are building our logical foundation of spacetime theories examining thoroughly each part of each axiom to see where and why we should assume them .
another ( more technical ) reason is that it will be more natural to include this assumption ( @xmath55 ) in our auxiliary axiom @xmath56 on page . our next axiom connects the worldviews of different inertial observers by saying that all observers observe the same external " reality ( the same set of events ) .
intuitively , by the event occurring for @xmath57 at @xmath46 , we mean the set of bodies @xmath57 observes at @xmath46 .
formally : @xmath58 : : all inertial observers coordinatize the same set of events : @xmath59 this axiom is very natural and tacitly assumed in the non - axiomatic approaches to special relativity , too
. basically we are done .
we have formalized the light axiom @xmath54 .
we have introduced two supporting axioms ( @xmath43 and @xmath60 ) for the light axiom which are simple and natural ; however , we can not simply omit them without loosing some of the meaning of @xmath54 .
the field axiom enables us to speak about distances , time differences , speeds , etc .
the event axiom ensures that different inertial observers see the same events . in principle
, we do not need more axioms for analyzing / axiomatizing special relativity , but let us introduce two more simplifying ones .
we could leave them out without loosing the essence of our theory , it is just that the formalizations of the theorems would become more complicated .
: : any inertial observer sees himself on the time axis : @xmath61 the role of @xmath62 is nothing more than making it easier to speak about the motion of reference frames via the motion of their time axes .
identifying the motion of reference frames with the motion of their time axes is a standard simplification in the literature .
@xmath62 is a way to formally capture this simplifying identification .
our last axiom is a symmetry axiom saying that all inertial observers use the same units of measurements .
: : [ axsymd ] any two inertial observers agree about the spatial distance between two events if these two events are simultaneous for both of them ; furthermore , the speed of light is 1 : @xmath63 @xmath64 let us see how @xmath56 states that `` all inertial observers use the same units of measurements . ''
that `` the speed of light is 1 '' ( besides that the speed of light is nonzero ) means only that observers are using units measuring time distances compatible with the units measuring spatial distances , such as light years or light seconds .
the first part of @xmath56 means that different observers use the same unit measuring spatial distances .
this is so because if two events are simultaneous for both observers , they can measure their spatial distance and the outcome of their measurements are the same iff the two observers are using the same units to measure spatial distances .
our axiom system for special relativity contains these 5 axioms only : @xmath65 in an axiom system , the axioms are the price " we pay , and the theorems are the goods " we get for them .
therefore , we strive for putting only simple , transparent , easy - to - believe statements in our axiom systems .
we want to get all the hard - to - believe predictions as theorems .
for example , we prove from @xmath66 that it is impossible for inertial observers to move faster than light relative to each other ( no ftl travel " for science fiction fans ) . in the following
, @xmath67 means logical derivability .
( no faster than light inertial observers ) [ thm - noftl ] @xmath68 for a geometrical proof of thm.[thm - noftl ] , see @xcite .
in relativity theory we are often interested in comparing the worldviews of different observers .
so we introduce the worldview transformation between observers @xmath57 and @xmath14 as the following binary relation : @xmath69 by thm.[thm - poi ] , the worldview transformations between inertial observers in the models of @xmath66 are poincar transformations , i.e. , transformations which preserve the so - called minkowski - distance @xmath70 of @xmath2-tuples @xmath71 . for the definition ,
we refer to @xcite or @xcite .
[ thm - poi ] @xmath72 for the proof of thm.[thm - poi ] , see ( * ? ? ?
* thm.11.10 , 640 . ) or ( * ? ? ?
* thm.3.2.2 , 22 . ) . by thm.[thm - poi ] , all predictions of special relativity , such as `` moving clocks slow down , '' are provable from @xmath66 . for details ,
see , e.g. , @xcite , @xcite , @xcite .
let us illustrate here by a simple example what we mean by logical analysis of a theory .
in @xmath60 we have assumed that all observers see the same ( possibly infinite ) meetings of bodies .
let us try to weaken @xmath60 to an axiom assuming something similar but only for finite meetings of bodies .
a natural candidate is one of the following finite approximations of @xmath60 : : : all inertial observers see the same @xmath48-meetings of bodies : @xmath73 for example , @xmath74 means only that inertial observers see the same bodies .
let us also introduce axiom scheme @xmath75 as the collection of all the axioms @xmath76 . by prop.[prop - meet ]
, @xmath76 is strictly weaker assumption than @xmath77 and @xmath60 is strictly stronger than all the axioms of @xmath75 together .
[ prop - meet ] @xmath78 item follows easily by the formulations of the axioms . to prove item , we are going to construct a model of @xmath76 in which @xmath77 is not valid .
let @xmath79.let all the bodies be inertial observers .
let @xmath80 see all the bodies in @xmath81 and none of them in any other coordinate points , i.e. , let @xmath82 hold iff @xmath83 ; and for all @xmath84 let @xmath85 see all the bodies but @xmath86 at coordinate points @xmath87 for all @xmath88 , i.e. , let @xmath89 hold iff @xmath90 and @xmath91 .
in this model , all inertial observers see all the possible @xmath48-meetings .
so @xmath76 is valid in this model .
however , the only inertial observer who sees the @xmath92-meeting @xmath93 is @xmath80 .
so @xmath77 is not valid in this model .
we are going to prove item by a similar model construction .
the only difference is that now @xmath5 will be infinite . for simplicity ,
let @xmath5 be the set of natural numbers .
let all the other parts of the model be defined in the same way .
now all the inertial observers see all the possible @xmath48-meetings of the bodies for all natural numbers @xmath48 .
so @xmath76 is valid in this model for all natural number @xmath48 .
hence @xmath75 is valid in this model .
however , only @xmath80 sees the event @xmath94 .
so @xmath60 is not valid in this model .
now we will use that there are no stationary ( i.e. , motionless ) light signals .
so let us formalize this statement .
: : inertial observers do not see stationary light signals .
@xmath95 [ prop - m3e ] @xmath96
first let us make some general observations .
by @xmath43 , there is no nondegenerate triangle in @xmath97 whose sides are of slope @xmath0 .
this is clear if @xmath98 ; and in the case @xmath99 , this can be shown by contradiction using the fact that the vertical projection of a triangle of this kind is a triangle whose one side is the sum of the other two sides .
therefore , @xmath43 and @xmath54 together imply that any inertial observer @xmath57 sees the events in which a particular photon participates on a line of slope @xmath100 .
by @xmath43 , @xmath54 and @xmath101 , every inertial observer @xmath57 sees different meetings of photons at different coordinate points .
this is so since ( by @xmath43 ) for every pair of points there is a line of slope @xmath102 containing only one of the points .
hence , by @xmath54 , there is a photon seen by @xmath57 only at one of the two coordinate points .
let us now prove item .
let @xmath57 and @xmath14 be inertial observers and let @xmath46 be a coordinate point . to prove @xmath60 , we have to find a coordinate point @xmath103 such that @xmath104 . to find this @xmath103 ,
let @xmath105 , @xmath106 and @xmath107 , see fig.[fig - meet ] . by @xmath54
, there are photons @xmath108 , @xmath109 and @xmath110 such that @xmath111 , @xmath112 , @xmath113 and @xmath114 .
since @xmath57 sees every photon on a line of slope @xmath100 , he sees the meeting of @xmath108 and @xmath109 only at @xmath46 and does not see the meeting of @xmath108 and @xmath110 .
since @xmath115 implies @xmath116 , @xmath14 sees the same meetings of pairs of photons .
so there is a @xmath103 where @xmath14 sees @xmath108 and @xmath109 meet .
@xmath103 is the only point where @xmath14 sees both @xmath108 and @xmath109 .
this is so because @xmath14 sees different meetings of photons at different points but sees the same @xmath117-meetings as @xmath57 .
so if there were another point , say @xmath118 , where @xmath14 sees @xmath108 and @xmath109 , there were photons @xmath119 and @xmath120 such that @xmath121 , @xmath122 and @xmath14 does not see the meeting of @xmath123 and @xmath124 . by axiom
@xmath115 @xmath57 has to see the meetings @xmath125 and @xmath126 .
the only point where @xmath57 can see these meetings is @xmath46 since @xmath46 the only point where @xmath57 sees @xmath108 and @xmath109 meet .
therefore @xmath57 sees the meeting of @xmath123 and @xmath124 at @xmath46 .
thus , by @xmath115 , @xmath14 also has to see the meeting of @xmath123 and @xmath124 , but @xmath14 does not see it . hence @xmath103 is the only point where @xmath14 sees both @xmath108 and @xmath109 .
let @xmath17 be a body such that @xmath127 .
by @xmath115 , @xmath14 has to see the meeting of @xmath108 , @xmath109 and @xmath17 .
this point has to be @xmath103 since the only point where @xmath108 and @xmath109 meet is @xmath103 .
since @xmath17 was an arbitrary body , we have @xmath128 .
the same argument shows that @xmath129 .
so @xmath104 as desired .
[ l][l]@xmath57 [ l][l]@xmath14 [ br][br]@xmath15 [ tl][tl]@xmath108 [ l][l]@xmath109 [ l][l]@xmath110 [ tl][tl]@xmath46 [ br][br]@xmath47 [ b][b]@xmath130 [ bl][bl]@xmath131 [ tl][tl]@xmath103 [ tl][tl]@xmath17 [ tl][tl]@xmath38 [ tl][tl]@xmath100 ] we are going to prove item , by constructing a model .
let @xmath132 be the field of real numbers .
let us denote the set of natural numbers by @xmath133 .
let @xmath134 .
let @xmath57 and @xmath14 be all the inertial observers and let the lines of slope @xmath38 be all the photons .
let @xmath57 and @xmath14 see the photon @xmath15 at coordinate point @xmath46 iff @xmath135
. let @xmath57 see all the bodies @xmath86 at @xmath46 iff @xmath136 .
let @xmath14 see all the bodies @xmath137 but @xmath86 at @xmath46 iff @xmath138 ( i.e. , iff @xmath46 is in the horizontal hyperplane @xmath139 ) . , vertical lines can be used instead of horizontal hyperplanes , which gives a counterexample with bodies having more natural properties . ]
it is straightforward from this construction that axioms @xmath43 , @xmath54 and @xmath140 are valid in this model .
since every line of slope 1 intersects every horizontal hyperplane , @xmath57 and @xmath14 see the same @xmath141-meetings of bodies .
hence @xmath116 is also valid in this model .
however , the only inertial observer who sees the meeting @xmath142 is @xmath57 .
so @xmath60 is not valid in this model .
we prove item by a similar construction .
the only difference is that now the set of bodies is @xmath143 ; and the photons are the vertical lines .
it is straightforward from the construction that axioms @xmath43 , @xmath54 are valid in this model ( @xmath98 ) .
since every vertical line intersects every horizontal hyperplane , @xmath57 and @xmath14 see the same @xmath48-meetings of bodies .
hence @xmath75 is also valid in this model .
however , only @xmath57 sees the meeting @xmath142 .
so @xmath60 is not valid in this model .
prop.[prop - m3e ] shows that a price to weaken axiom @xmath60 to @xmath115 is to assume that there are no stationary light signals .
since @xmath56 contains this assumption , we can simply replace @xmath60 with @xmath115 in @xmath66 .
a natural continuation of this investigation can be a search for assumptions that allow us to weaken @xmath115 to @xmath116 .
a possible candidate is that bodies move along straight lines and the dimension @xmath2 is at least @xmath117 .
the proof of item shows that assuming only that bodies move along straight lines is not enough , if @xmath144 .
several similar investigations on the logical connections of axioms and predictions , see , e.g. , @xcite , @xcite on dynamics , @xcite , ( * ? ? ?
* , 7 ) , @xcite on twin paradox , @xcite on kinematics , time - dilation and length - contraction , twin paradox , etc .
in @xmath145 we restricted our attention to inertial observers .
it is a natural idea to generalize the theory by including accelerated observers as well .
it is explained in the classic textbook @xcite that the study of accelerated observers is a natural first step ( from special relativity ) towards general relativity .
we have not introduced the concept of observers as a basic one because it can be defined as follows : an _ observer _ is nothing other than a body who observes " ( coordinatizes ) some other bodies somewhere , this property can be captured by the following formula of our language : @xmath146 our key axiom about accelerated observers is the following : @xmath147 : : at each moment of his life , every accelerated observer sees ( coordinatizes ) the nearby world for a short while in the same way as an inertial observer does . for formulation of @xmath147 in our fol language , see @xcite , @xcite or @xcite . axiom @xmath147 ties the behavior of accelerated observers to those of inertial ones .
justification of this axiom is given by experiments .
we call two observers _ co - moving _ at an event if they `` see the nearby world for a short while in the same way '' at the event . by this notion
@xmath147 says that at each event of an observer s life , he has a co - moving inertial observer .
we can think of a dropped spacepod as a co - moving inertial observer of an accelerated spaceship ( at the event of dropping ) . or , if a spaceship switches off its engines , it will move on as a co - moving inertial spaceship would . our next two axioms
ensure that the worldviews of accelerated observers are big enough .
they are generalized versions of the corresponding axioms for inertial observers , but now postulated for all observers . : : if @xmath57 sees @xmath14 in an event , then @xmath14 can not deny it : @xmath148 : : any observer sees himself in an interval of the time axis : @xmath149 our last two axioms will ensure that the worldlines of accelerated observers are tame " enough , e.g. , they have velocities at each moment . in @xmath66 ,
the worldview transformations between inertial observers are affine maps , the next axiom will state that the worldview transformations between accelerated observers are approximately affine , wherever they are defined .
: : the worldview transformations have linear approximations at each point of their domain ( i.e. , they are differentiable ) . for a precise formalization of @xmath150 ,
see , e.g. , @xcite .
we note that @xmath150 implies that the worldview transformations are functions with open domains .
however , if the numberline has gaps , still there can be crazy motions .
our last assumption is an axiom scheme supplementing @xmath150 by excluding these gaps .
: : every definable , bounded and nonempty subset of @xmath5 has a supremum ( i.e. , least upper bound ) . in @xmath44
`` definable '' means `` definable in the language of @xmath151 , parametrically . '' for a precise formulation of @xmath44 , see @xcite or @xcite .
@xmath44 is a
mathematical axiom " in spirit .
it is tarski s fol version of hilbert s continuity axiom in his axiomatization of geometry , see @xcite , fitted to the language of @xmath151 .
when @xmath5 is the field of real numbers , @xmath44 is automatically true . that @xmath44 requires the existence of supremum only for sets definable in the language of @xmath151 instead of every set , is important not only because by this trick we can keep our theory within fol ( which is crucial in a foundational work ) , but also because it makes this postulate closer to the the physical / empirical level .
the latter is true because @xmath44 does not speak about `` any fancy subset '' of the quantities , just those `` physically meaningful '' sets which can be defined in the language of our ( physical ) theory . adding this 5 axioms to @xmath66 , we get an axiom system for accelerated observers : @xmath152 as an example we show that the so - called _ twin paradox _ can be naturally formulated and analyzed logically in @xmath151 .
our axiomatic approach also makes it possible to analyze the details of the twin paradox ( e.g. , who sees what , when ) with the clarity of logic , see @xcite for part of such an analysis . according to the twin paradox ,
if a twin makes a journey into space ( accelerates ) , he will return to find that he has aged less than his twin brother who stayed at home ( did not accelerate ) .
we formulate the twin paradox in our fol language as follows .
: : every inertial observer @xmath57 measures at least as much time as any other observer @xmath14 between any two events @xmath153 and @xmath154 in which they meet ; and they measure the same time iff they have encountered the very same events between @xmath153 and @xmath154 : @xmath155 where @xmath156 .
[ thm - twp ] @xmath157 for the proof of thm.[thm - twp ] , see @xcite or @xcite .
item of thm.[thm - twp ] states that @xmath44 can not be replaced with the whole fol theory of real numbers in @xmath151 if we do not want to loose @xmath158 from its consequences .
our theory @xmath159 is also strong enough to predict the gravitational time - dilation effect of general relativity via einstein s equivalence principle , see @xcite , @xcite .
our theory of accelerated observers @xmath151 speaks about two kinds of observers , inertial and accelerated ones .
some axioms are postulated for inertial observers only , some apply to all observers .
we get an axiom system @xmath160 for general relativity by stating the axioms of @xmath151 in a generalized form in which they are postulated for all observers , inertial and accelerated ones equally . in other words
, we will change all axioms of @xmath151 in the same spirit as @xmath161 and @xmath162 were obtained from @xmath62 and @xmath60 , respectively .
this kind of change @xmath151 @xmath163 @xmath160 can be regarded as a `` democratic revolution '' with the slogan `` all observers should be equivalent , the same laws should apply to all of them . '' here `` law '' translates as `` axiom . ''
this idea originates with einstein ( see his book ( * ? ? ?
* part ii , ch.18 ) ) . for simplicity
, we will use an equivalent version of the symmetry axiom @xmath56 ( see ( * ? ? ?
* thm.2.8.17(ii ) , 138 . ) or ( * ? ? ?
* thm.3.1.4 , 21 . ) ) , and we will require the speed of photons to be 1 in @xmath164 ( as opposed to requiring it in @xmath165 ) .
: : the velocity of photons an observer meets " is 1 when they meet , and it is possible to send out a photon in each direction where the observer stands . :
: meeting observers see each other s clocks slow down with the same rate . for a precise formulation of these axioms ,
see @xcite , @xcite .
we introduce an axiom system for general relativity as the collection of the following axioms : @xmath166 axiom system @xmath160 contains basically the same axioms as @xmath66 , the difference is that they are assumed only locally but for all the observers .
thm.[grcomp - thm ] below states that the models of @xmath160 are exactly the spacetimes of usual general relativity . for the notion of a lorentzian manifold
we refer to @xcite , @xcite and @xcite .
[ grcomp - thm ] @xmath167 is complete with respect to its standard models , i.e. , with respect to lorentzian manifolds over real closed fields . this theorem can be regarded as a completeness theorem in the following sense .
let us consider lorentzian manifolds as intended models of @xmath160 .
how can we do that ? we give a method for constructing a model of @xmath160 from each lorentzian manifold ; and conversely
, we show that each model of @xmath160 is obtained this way from a lorentzian manifold .
after this is elaborated , we have defined what we mean by a formula @xmath168 in the language of @xmath160 being valid in a lorentzian manifold .
then completeness means that for any formula @xmath168 in the language of @xmath160 , we have @xmath169 iff @xmath168 is valid in all lorentzian manifolds over real closed fields .
this is completely analogous to the way in which minkowskian spacetimes were regarded as intended models of @xmath66 in the completeness theorem of @xmath66 , see ( * ? ? ?
* thm.11.28 , 681 . ) and @xcite .
we call the worldline of an observer _ timelike geodesic _ , if each of its points has a neighborhood within which this observer maximizes measured time ( wrist - watch time ) " between any two encountered events . for formalization of this concept in our fol language , see , e.g. , @xcite .
according to the definition above , if there are only a few observers , then it is not a big deal that a worldline is a time - like geodesic ( it is easy to be maximal if there are only a few to be compared to ) . to generate a real competition for the rank of having a timelike geodesic worldline
, we postulate the existence of many observers by the following axiom scheme of comprehension .
: : for any parametrically definable timelike curve in any observers worldview , there is another observer whose worldline is the range of this curve .
a precise formulation of @xmath170 can be obtained from that of its variant in @xcite . an axiom schema @xmath170 guarantees that our definition of a geodesic coincides with that in the literature on lorentzian manifolds .
therefore we also introduce the following theory : @xmath171 so in our theory @xmath172 , our concept of timelike geodesic coincides with the standard concept in the literature on general relativity .
all the other key concepts of general relativity , such as curvature or riemannian tensor field , are definable from timelike geodesics .
therefore we can treat all these concepts ( including the concept of metric tensor field ) in our theory @xmath172 in a natural way . in general relativity , einstein s field equations ( efe )
provide the connection between the geometry of spacetime and the energy - matter distribution ( given by the energy - momentum tensor field ) .
since in @xmath172 all the geometric concepts of spacetime are definable , we can use einstein s equation as a definition of the energy - momentum tensor , see , e.g. , @xcite or @xcite , or we can extend the language of @xmath172 with the concept of energy - momentum tensor and assume einstein s equations as axioms .
as long as we do not assume anything more of the energy - momentum tensor than its connection to the geometry described by einstein s equations , there is no real difference in these two approaches . in both approaches ,
we can add extra conditions about the energy - momentum tensor to our theory , e.g. , the dominant energy condition or , e.g. , that the spacetimes are vacuum solutions .
there is observational evidence suggesting that in our physical universe there exist regions supporting potential non - turing computations .
namely , it is possible to design a physical device in relativistic spacetime which can compute a non - turing computable task , e.g. , which can decide whether zf set theory is consistent .
this empirical evidence is making the theory of hypercomputation more interesting and gives new challenges to the physical church thesis , see , e.g. , @xcite .
these new challenges do more than simply providing a further connection between logic and spacetime theories ; they also motivate the need for logical understanding of spacetime theories .
we have axiomatized both special and general relativity in fol . moreover , via our theory @xmath151 , we have axiomatized general relativity so that each of its axioms can be traced back to its roots in the axioms of special relativity .
axiomatization is not our final goal .
it is merely an important first step toward logical and conceptual analysis .
we are only at the beginning of our ambitious project .
andrka , h. , j. x. madarsz , and i. nmeti , with contributions from a. andai , g. sgi , i. sain and cs .
tke , 2002 , _ on the logical structure of relativity theories_. research report .
budapest , alfrd rnyi institute of mathematics .
http://www.renyi.hu/pub/algebraic-logic/contents.html .
andrka , h. , j. x. madarsz , and i. nmeti , 2006 , logical axiomatizations of space - time .
samples from the literature . in a.
prkopa , et al .
_ non - euclidean geometries_. berlin , springer .
155185 .
madarsz , j. x. , i. nmeti , and g. szkely , 2007 , first - order logic foundation of relativity theories . in d.
gabbay , et al .
( eds . ) , _ mathematical problems from applied logic ii_. berlin , springer . 217252 .
h. andrka , j. x. madarsz , + i. nmeti , g. szkely + alfrd rnyi institute of mathematics + of the hungarian academy of sciences + budapest p.o .
box 127 , h-1364 hungary + [email protected] , [email protected] + [email protected] , [email protected] | the aim of this paper is to give an introduction to our axiomatic logical analysis of relativity theories . |
the hubble deep field ( williams 1996 ) affords us an unprecedented view of the optical sky at small angular scales and faint flux levels .
it thus allows us to study faint ( and presumably high redshift ) galaxies without complicating field crowding effects caused by comparatively poor seeing in ground - based faint galaxy studies ( tyson 1995 ) .
preliminary results ( giavalisco 1996 ) show that source counts in the hdf continue to rise as a power law to the completeness limit of the data .
such an effect may be due to ever larger numbers of galaxies at fainter flux levels
. however , it may also be due to the increasingly clumpy appearance of galaxies at high redshift , which can confuse source detection algorithms into counting parts of each physically distinct galaxy as several faint sources . to test this possibility ,
we consider how redshift effects can lead to over - counting of whole sources in a deep field like the hdf .
@xmath7-correction and surface - brightness dimming tend to enhance the relative prominence uv bright and compact objects , such as active star - forming regions ( oconnell & marcum 1996 ) .
if the enhancement is sufficient , several star - forming regions occurring in a single galaxy will produce the appearance of several small sources separated by an angular scale comparable to the size of a normal galaxy . in the later sections of the paper , we use two different statistics to test the extent to which hdf source counts reflect the subdivision of galaxies .
these tests exploit the weak dependence of angular size on redshift @xmath8 at @xmath3 .
since galaxies of present day size ( @xmath9 ) remain resolved at all redshifts in the hdf ( peebles 1993 ) , we can compare the physical separations and sizes of objects in the hdf to those of galaxies in the low - redshift universe . in section 4 , we discuss the two - point angular correlation function @xmath10 of hdf sources .
considerable correlations may be expected for physical scales @xmath11 if many galaxies in the field break up into multiple giant hregions in the source catalogs . in section 5 ,
we present number - magnitude relations derived from our source catalogs , which show a smooth increase to the completeness limit , with a flatter faint - end slope than in deep ground - based images , and a rough relation @xmath12 , consistent with the hypothesis that many of the faintest images are parts of larger objects .
two redshift effects play an important role in the appearance of very deep fields .
first , the redshift moves the ultraviolet rest - frame light into the optical , so that rest - frame uv - bright objects will be selected over optically bright objects .
schade ( 1995 ) have found that in up to one - third galaxies with @xmath13 , compact blue components dominate the blue light . in such galaxies
, active star - formation is occurring .
abraham ( 1996 ) and clements and couch ( 1996 ) have found the trend toward increasingly blue asymmetric objects to be even more pronounced in the hdf , where yet higher redshifts ( @xmath14 ) bring ever harder rest - frame uv emission into optical bands . moreover , many spectroscopically confirmed high redshift objects in the hdf display noticeable asymmetry and multiple structure .
a nice collection of such objects can be found in figure 2 of steidel ( 1996 ) .
second , compact high redshift objects can appear more prominently than diffuse objects if their angular size is smaller than the point - spread function .
the @xmath15 bolometric surface brightness dimming of resolved sources is less significant for such compact sources : the spreading of light rays is inconsequential if they never spread into more than one psf .
this could lead to a maximum @xmath16 relative enhancement of compact sources over resolved , diffuse sources . also , the actual physical distance changes little beyond redshift @xmath17 , so that a true point source would suffer little @xmath18 dimming .
giant hregions , which have sizes of 0.1 to @xmath2 , ( hodge 1993 ) , remain marginally unresolved by hst , and are thus sufficiently compact for surface brightness dimming to be diminished .
steidel ( 1996 ) , and cowie ( 1996 ) have shown that many sources in the hdf have redshifts @xmath19 ; these could produce a 2 magnitude relative enhancement of compact sources over diffuse sources . both of these factors work to enhance the prominence of compact star - forming regions in a very deep field such as the hubble deep field .
we retrieved the hdf version 2 images and version 1 object catalogs from the space telescope science institute ( williams 1996 ) .
we found that the catalog contained spurious faint sources near the margins of the large , bright ( nearby ) sources .
this created an overabundance of very close pairs , which artificially increases the angular correlation function at small scales .
we therefore created a new catalog , using the _ daofind _ algorithm to identify objects in deep images formed by averaging the @xmath20 and @xmath21 frames .
this algorithm looks for peaks in the image after bandpass spatial filtering ; the filter is a truncated lowered gaussian with breadth comparable to the point - spread function ( psf ) . _
daofind _ detected a total of 2817 objects in the central @xmath22 regions of three wide - field chips .
the bandpass filtering helps control both random noise and spurious detections in the wings of very bright sources .
however , it may also wash out object pairs with separations below about @xmath0 , and appears to detect subgalaxian features in some foreground galaxies . to lessen this effect
, we masked out sources brighter than magnitude 23.5 before running our detection algorithms .
one may compute the two - point angular correlation function by comparing the number of data pairs at given angular separation to the number of simulated random ( window ) pairs at the same separation .
we thus populated the survey area with @xmath23 random points and computed the distribution of pairs of data objects with data objects @xmath24 , data objects with window objects @xmath25 , and window objects with window objects @xmath26 .
this allowed us to compute ( as hamilton 1993 ) the angular correlation for each of the three hdf fields ( wfc chips ) as @xmath27 because the comoving volume of the hdf sample is dominated by @xmath3 , most objects in the catalog are at high redshift . for reference ,
we have used two methods to select higher redshift objects in the catalog .
the first , from steidel ( 1996 ) , measures spectral curvature between filters to find `` uv dropout '' objects , in which the lyman-@xmath28 break has entered the bluest filter . for the hdf filter set , this happens at redshift @xmath29 .
steidel et al ( 1996 ) find that a useful cut is @xmath30 $ ] .
we developed a second , similar cut based changing spectral slope , using the filter selection information provided by stsci ( 1995 ) , their table 3 ; this cut prefers objects with redshift @xmath31 .
any color - based selection criterion ought to place physically associated objects into the same bin , since such objects presumably have similar evolutionary histories and hence colors .
we have plotted in figure 1 the angular correlations yielded by the objects meeting various selection criteria . in each panel
, the lighter curves represent the correlations from each of the three wfc chips , while the heavier curve is the mean of those three .
the one - sigma error bars are derived from comparison of the three values for the individual chips .
the first two rows gives the correlations for the above - mentioned cuts as applied to our catalog .
the third row , for comparison , contains a catalog from couch ( 1996 ) with the steidel ( 1996 ) cut applied .
the first ( leftmost ) column of figure 1 shows the correlations for the higher color - redshift objects ; the second column shows the correlations for the lower redshift objects ; the third column shows the cross - correlations between the high and low color - redshift objects .
first , we examine the cross - correlation .
if the color cuts effectively sort physically associated objects at different redshift , there should be little or no correlation between the high and low cuts .
indeed , the cross - correlation signal is very low , and consistent with zero in the first two rows ( our catalog ) .
we thus surmise that these cuts , applied to our catalog , rarely admit physically associated objects into different bins . the high correlations and low cross - correlations in figure 1
tell us that both moderate and high color - redshift species must exhibit strong correlation on scales of less than one arcsecond .
this is to be expected at low to moderate redshift if the cataloging scheme records subgalaxian structure , such as hregions and bulges .
when one overplots our catalog on the images themselves , one sees that such structure is detected within galaxies as individual objects .
some of these galaxies have many detections within them .
these objects , visibly within single galaxies , likely dominate the correlation signal at low redshift .
however , at higher redshift , the underlying galaxy may vanish due to @xmath7-correction and surface brightness dimming , so that only the compact and uv - bright star - forming regions are left .
many object pairs and groups are visible in the images . we infer that these pairs and groups are in fact objects within the same physically associated galaxy , just as the hregions at low redshift are in the same galaxy . for this to be so
, the correlation scales must correspond to physical scales which allow detected objects to fit within a single galaxy .
peebles ( 1993 ) demonstrates that for most cosmologies the angular size of @xmath9 galaxies at @xmath32 is @xmath33-@xmath34 arcseconds ( @xmath35 , any @xmath36 ) , so that correlations below this scale indicate counting of subgalaxian objects . efstathiou ( 1991 ) has shown that projection effects are unlikely to contaminate correlations , so that small - scale correlation is likely due to physical association of objects .
the increasing correlation down to @xmath0 implies that many detected objects have sizes @xmath37 , which is comparable to the expected size ( @xmath38 ) of a @xmath39 hregion at @xmath3 .
we checked this by computing petrosian radii ( with @xmath40 , following the kron [ 1995 ] notation ) for the catalog objects , which showed a broad peak at about @xmath41 .
if we now compare the high redshift correlations to low redshift correlations in the first two rows of figure 1 , we see that both show substantial correlation strength on sub - arcsecond scales .
in addition , the high redshift bin shows a substantially stronger correlation around @xmath0 .
the statistical significance of this increase ( about one sigma ) is debatable , though a clear excess is visible in all three chips . in all three chips , the correlation appears to turn over at half an arcsecond in the second column , but not in the first .
this excess correlation on very small scales may be due to the increasing prominence of very young , compact , and hot star - forming regions at higher redshifts , where the wien cut - off begins hiding a type and cooler stars ( fitzpatrick , 1996 ) .
we have confirmed that the correlation persists among a pure high redshift sample by measuring the correlation functions of both the high-@xmath8 candidates and the spectroscopically confirmed high-@xmath8 sources of steidel et al ( 1996 ) .
the peak correlation strengths were @xmath42 and @xmath43 respectively .
all of these arguments suggest numerous subgalaxian hregions in our catalog , both at high @xmath44 and lower redshifts . at low redshifts , the parent galaxies are visible in the images , but at high redshift , the parent galaxy is most often invisible , which might lead one to think navely that the objects are physically separate , and overcount `` galaxies . '' finally ,
we compare correlations derived from our catalog to those derived from the couch ( 1996 ) catalog , which he created using the _ sextractor _ ( bertin 1994 ) package .
we have found that his catalog does not often overcount foreground substructure , as does ours .
correlations in the two catalogs agree at high redshift , but disagree significantly in the lower redshift sample .
the smaller correlation at low redshift and small angular scales in his catalog evinces the difference between our multiple counting of foreground subgalaxian structure and his proper counting of one count per one foreground galaxy .
however , the agreement on the excess sub - arcsecond correlation at high redshift suggests that both schemes overcount substructure at high redshift .
couch s sextractor catalog also shows nonzero correlation at large angular separations ( @xmath45@xmath46 ) , absent in our catalog .
this is probably due to the increased sensitivity of sextractor s thresholding to faint objects in the wings of bright objects .
_ daofind s _ high - pass filtering is less sensitive to such objects . as a final test , we made percentile cuts according to angular size ( intensity weighted second moments , @xmath47 and @xmath48 from the couch catalog ) .
we found dramatically enhanced sub - arcsecond correlation in small objects .
for the smallest quartile ( @xmath49 ) , we found correlation of @xmath50 at @xmath0 ; for the smallest 10% of objects ( @xmath51 ) , we found correlation of @xmath52 at @xmath0 as visible in the fourth row of figure 1 .
this sharp , if uncertain , enhancement demonstrates that the bulk of the correlation is , in fact , coming from the smallest objects , in agreement with the hypothesis that the correlation derives from small , subgalaxian objects .
we present number - magnitude relations for both unsmoothed and smoothed @xmath21 images in figure 2(a ) .
the solid histograms represent no smoothing , the dashed histogram @xmath53 smoothing , and the dotted histogram @xmath54 smoothing .
no effort has been made to correct the observed counts for incompleteness or crowding effects , and the decrease in the counts beyond @xmath55 30th magnitude is almost certainly due to incompleteness .
object fluxes in @xmath56 , @xmath57 , and @xmath58 apertures were used to generate the magnitudes at the three smoothing scales .
we chose aperture photometry for its robustness in cases where multiple sources overlap on the sky .
the resulting number - magnitude relation rises monotonically to a completeness limit around ab magnitude @xmath59 in the @xmath21 filter . at the faint end
, it has a slope somewhat shallower than that seen in ground - based r band counts , and substantially shallower than in ground - based u counts ( tyson 1995 ) .
the number - magnitude slope for the unsmoothed image catalogs steepens at the bright end , which can be understood as a side effect of using small aperture photometry .
the smoothed catalogs , on the other hand , show a continued power law behavior to the brightest objects in the hdf ( @xmath60 ) .
the slope and normalization of the counts are consistent with ground - based counts for the overlap region ( @xmath61 ) ( tyson 1995 ) .
we have also plotted the @xmath62 line in figure 3 .
this line separates slopes with convergent and divergent total flux at faint magnitudes , and is a good match to ground - based r band data for @xmath63 .
it is also the number - magnitude slope expected if standard objects are broken into fragments and then counted as sources ( with equal probability for any number of fragments , up to some large maximum ) .
counts from the unsmoothed data match this slope for @xmath64 .
figure 2(b ) shows the number - magnitude relation for only the high - color redshift ( using the steidel cut ) .
the results here are not very different , other than zero - point , from the relation for all sources , which again suggests that we are seeing a similar population at high and moderate redshift .
comparing the results for @xmath53 and @xmath54 smoothing to those for no smoothing , we detect two and four times fewer objects , and the completeness limit worsens by about two and four magnitudes . this shows the deleterious effects of field - crowding in low - resolution fields , where galaxian images begin to overlap in the poor seeing .
the counts in the smoothed fields are less likely to overcount substructure , but more likely to miss faint objects within the seeing discs of bright objects .
therefore , care must be taken when comparing ground - based deep counts with counts in the hubble deep field .
we have cataloged objects in the hubble deep field in a way that is less prone to spurious detections than were previous efforts . from the catalog ,
we have drawn the angular correlation for high and low color - redshift subsets using two different cuts . we have found similar correlations down to @xmath53 for both subsets .
since the signal is ostensibly dominated by hregions in the lower redshift subset , we surmise that that signal is also dominated by subgalaxian structure in the higher redshift subset .
we have compared our results to the correlation derived from an independent catalog ( couch 1996 ) which appears not to over - count nearby objects . as expected
, there is less correlation in the lower redshift subset of this catalog than in ours .
however , at higher redshift , the correlations of the catalogs agree rather well , so that both of our catalogs include as distinct objects what are likely subgalaxian structure at high redshift .
a dramatic increase in sub - arcsecond correlation occurs in the subset of objects with the smallest angular sizes , in agreement that the correlated objects are small , subgalaxian objects .
the qualitative difference between deep space- and ground - based optical data is due to a conspiracy of scales . because the characteristic angular sizes of galaxies at redshifts @xmath65 correspond to the @xmath66 angular resolution of ground - based data , deep optical counts from the ground will see galaxy - sized objects as single peaks . at the higher resolution available from space ,
substructure becomes detectable in galaxies at any redshift , and overcounting becomes a possibility .
we have also computed the magnitude - radius relation , which shows that a large fraction of the objects have characteristic sizes around @xmath41 , corresponding to scale lengths @xmath67 , typical of both high redshift galaxian scale - lengths and diameters of giant star - forming regions .
the peak at @xmath41 allows several objects to fit into a single galaxy , as one requires for the subgalaxian structure scenario .
the number - magnitude relations for our catalogs show convergent flux in all bands with @xmath12 as expected for images broken into fragments .
this physically reassuring result differs from the nave extrapolation of ground - based number - magnitude relations for u , b , and possibly r bands .
we find that after we smoothed the data , the counts drop dramatically at the faint end .
this illustrates how seeing reduces the faint counts in ground - based work , diluting isolated faint objects below detection thresholds while blurring substructures in brighter galaxies together to form single peaks .
the statistical tests presented herein suggest that the most distant objects in the hdf must be some combination of galaxies and star - forming fragments , a distinction increasingly hard to draw in deep fields .
this supports our hypothesis that ultraviolet bright and compact star - forming regions contribute substantially to the flux , and increasingly to the number counts , we receive from high redshift samples .
wnc is most grateful for the continued support of the fannie and john hertz foundation , and partial support from nsf grant ast-9529120 .
jer and dns s work has been supported by nsf grants ast 91 - 17388 , nasa adp grant 5 - 2567 , and jer s nsf traineeship dge-9354937 .
jer also thanks ipac for its hospitality .
jpo s work has been partially supported by nsf grant ast-9424416 .
we would also like to thank robert h. lupton , j. richard gott , iii , warrick couch , sangeeta malhotra , and anonymous referees for their very useful communications .
we thank jill knapp for her kind support and encouragement .
finally , we thank the hdf team for their hard work and generosity in preparing the data for public release .
hodge , p. , 1993
, in massive stars : their lives in the interstellar medium , asp conference series , vol .
35 , joseph p. cassinelli and edward b. churchwell , eds .
( san francisco : astronomical society of the pacific ) oconnell , r. w. & marcum , p. , 1996
, `` the ultraviolet morphology of galaxies '' in _ hst and the high redshift universe ( 37th herstmonceux conference ) _ , eds .
tanvir , a. aragon - salamanca , j.v . wall , 1996 . | the hubble deep field ( hdf ) offers the best view to date of the optical sky at faint magnitudes and small angular scales .
early reports suggested that faint source counts continue to rise to the completeness limit of the data , implying a very large number of galaxies . in this _
letter , _ we use the two - point angular correlation function and number - magnitude relation of sources within the hdf in order to assess their nature .
we find that the correlation peaks between @xmath0 and @xmath1 with amplitude of 2 or greater , and much more for the smallest objects .
this angular scale corresponds to physical scales of order @xmath2 for redshifts @xmath3 .
the correlation must therefore derive from objects with subgalaxian separations . at faint magnitudes ,
the counts satisfy the relation @xmath4 , expected for images which are subdivisions of larger ones .
several explanations for these observed correlations are possible , but a conservative explanation can suffice to produce our results .
since high redshift space ( @xmath5 ) dominates the volume of the sample , observational redshift effects are important .
rest - frame ultraviolet radiation appears in the hdf s visible and near - uv bands , and surface brightness dimming enhances the relative brightness of unresolved objects versus resolved objects .
both work to increase the prominence of compact star - forming regions over diffuse stellar populations .
thus , a `` normal '' gas - rich galaxy at high redshift can appear clumpy and asymmetric in the visible bands .
for sufficiently faint and distant objects , the compact star - forming regions in such galaxies peak above undetectable diffuse stellar backgrounds .
our results do not exclude asymmetric formation or fragmentation scenarios .
@xmath6 email : wes,rhoads,[email protected] ; [email protected] |
the interest in logarithmic conformal field theories ( cfts ) ( see @xcite for an introduction ) has grown over the last ten years or so as their potential role in condensed matter as well as string theory applications has become more evident .
condensed matter applications include critical geometrical models in two dimensions such as percolation and polymers ( see @xcite for a recent contribution and review ) , and critical points in non - interacting disordered fermion models in @xmath10 dimensions ( see @xcite for recent contributions ) , such as the transition between plateaux in the integer quantum hall and spin quantum hall effect .
string theory applications include the study of sigma models on non - compact ( super- ) target spaces such as psu@xmath11 @xcite .
logarithmic cfts are also fascinating mathematical objects in their own right , bound to attract the attention of representation theory experts .
logarithmic cfts were encountered as early as 1987 by knizhnik @xcite . in 1992 ,
rozansky and saleur tackled in some details one of the simplest such theory , the wzw model on the u(@xmath12 ) supergroup .
they found that the non - semisimplicity of u(@xmath12 ) carried over to the cft , with the possibility of a non - diagonal virasoro generator ( here , proportional to the superalgebra casimir ) @xmath13 , indecomposable operator product expansions ( opes ) , and logarithms in some of the correlators @xcite .
the potential generality of these features was realized almost simultaneously by gurarie in 1993 @xcite , who coined the name logarithmic cft .
certainly the theory of symplectic fermions is the best understood logarithmic cft @xcite .
[ a close second is the wzw model on u(@xmath12 ) where powerful results were obtained recently using the minisuperspace approach @xcite .
it is not clear to what extent the current algebra structure makes this example qualitatively different from the logarithmic cfts of more physical interest . ]
a natural extension of the symplectic fermions has given rise to the so - called `` augmented @xmath14 models '' , which exhibit an enlarged triplet @xmath15 algebra @xcite .
an essential feature of these theories is that the identity has a logarithmic partner , and thus that there are at least two fields of vanishing dimension .
there have been some attempts to generalize the structure of the augmented @xmath14 models to other series , in particular the @xmath16 case @xcite .
one of the most interesting cases , with numerous condensed matter applications to disordered systems , corresponds to vanishing central charge .
consideration of singularities in opes as @xmath17 led gurarie to the conjecture that in these theories the stress energy tensor @xmath18 had a logarithmic partner @xmath19 such that @xmath13 had a jordan cell of dimension 2 at the level @xmath20 @xcite . in this and in subsequent papers gurarie and
ludwig @xcite proposed to use the modes of @xmath19 to define an extension of the virasoro algebra characterized by a parameter @xmath21 . [
although appealing intuitively and successful in some cases , the idea has not borne much fruit , in particular because of considerable difficulties in putting together chiral and non chiral sectors . ]
the problem has also been discussed from a slightly different point of view of the replica limit by cardy @xcite .
most of the approaches mentioned so far start from the consideration of abstract conformal field theories
. such generality may not be the easiest way to proceed . in 2001 , we defined families of supersymmetric lattice models , whose continuum limits we identified with the ir limits of conformal supersymmetric non - linear sigma models , or in some cases with landau - ginzburg theories @xcite .
[ for instance , one of these lattice models was defined using alternating fundamental and conjugate representations of sl(@xmath2 ) and argued , in the continuum limit , to be described by the ir limit of the @xmath22 model at @xmath23 .
this model is closely related to the properties of hulls of percolation @xcite .
a similar model was defined for dilute and dense polymers . ] in all these cases , we obtained the full spectrum of the associated logarithmic conformal field theory , and found considerable richness , including an infinite set of virasoro primary fields which all have rational conformal weights , and indications of a large non chiral symmetry algebra commuting with virasoro .
we note that in all our examples with @xmath4 , there was a single field with vanishing dimension .
we expect in general that the logarithmic theories we uncovered do not coincide with any of the extended @xmath14 or @xmath16 models mentioned before ( except for the symplectic fermions ) .
it is worthwhile to consider a broad view of the strategies pursued in these various references .
first , we point out that the traditional idea of a _ rational _ cft contains two essential ingredients @xcite : the local fields ( or corresponding states ) fall into a _ finite _ number of representations of the chiral algebra ( which in the strictest view of rational cft is generated by a finite number of bosonic local fields of integer conformal weights ) , and these representations are semisimple , that is fully decomposable as direct sums of irreducible representations of the chiral algebra .
hence also the operator products of local fields decompose fully into sums of these local fields .
next , we note that the occurrence of logarithms in correlators is due to @xmath24 not being diagonalizable , and this results from the appearance of fields transforming in non - semisimple representations of the virasoro algebra .
one is then led to consider indecomposable representations that are not necessarily irreducible .
the next problem is how to control these theories that do not satisfy this requirement of traditional rational cfts .
many of the papers cited above try to do so by considering extended chiral algebras @xcite , such that the number of indecomposable representations is finite , so that in this sense one may speak of `` rational logarithmic cfts '' .
but many applications , such as to disordered systems , and all the theories we consider in this paper , appear to possess only the virasoro algebra as the chiral algebra , and infinitely many ( non - isomorphic ) indecomposable representations appear as local fields .
in addition , these theories may even fail to be quasirational , that is infinite sums of local fields may be produced in ope s . in these irrational cfts , it is not clear how the analysis is to be organized .
we propose that the use of symmetry algebras that commute with the _ full _ chiral algebra @xcite can be a powerful organizing principle , even when semisimplicity of the algebraic structures is lost .
these symmetries and structures can be usefully studied by starting with lattice models .
the idea of gaining understanding of logarithmic cft ( lcft ) by studying lattice realizations in details has been around for a while , and put forward most recently ( albeit without the algebraic connotations we consider crucial ) in @xcite .
it does open a fruitful new route to progress .
indeed , the lattice equivalent of the reducible but indecomposable virasoro representations in lcft is the non semi - simplicity of the various algebras underlying the lattice models .
[ examples include the temperley - lieb ( tl ) algebra or @xmath25sl@xmath26 , neither of which is semi - simple when @xmath27 is a root of unity ; such values are usually those of greatest interest .
] a great deal of progress has been made in the mathematics literature in this area over the last fifteen years or so .
our strategy will be to exploit this progress and infer from it results about the lcfts .
more precisely , the lattice models we deal with all have a similar structure ( we discuss here only the boundary case , even though many results generalize to the case of the bulk ) . the local hamiltonian densities generate an associative algebra ( in our examples , the tl algebra ) whose commutant , in the lattice hilbert space of interest , is @xmath28(sl@xmath29 ) in the example of the spin-1/2 chain , or in the supersymmetric models in refs.@xcite a remarkably large algebra whose representations are in `` morita equivalence '' with those of @xmath28(sl@xmath29 ) . using results from the mathematics literature @xcite and others we derive @xcite
, we can analyze in full detail the structure of these two commuting algebras ( which are in so - called ringel duality ) and their representations in the lattice model .
it is then a natural step to conjecture the corresponding structure in the continuum limit : the algebra of hamiltonian densities goes over to the virasoro algebra @xcite , while the commutant goes over to an algebra of non - local charges .
the general structure of the theory is one in which the virasoro algebra commutes with a certain symmetry algebra , but the latter is _ not _ any kind of current algebra , and nor is it a yangian .
the structure of indecomposable representations in the lattice model goes over to the continuum limit , and we thus obtain information about the representations of the virasoro algebra in our logarithmic theories .
these representations are typically reducible but indecomposable with a diamond ( or quartet ) shape and made out of two `` standard modules '' connected by some `` glue '' .
they bear a lot of resemblance to the so - called staggered modules introduced abstractly in @xcite .
the full structure of the hilbert space involves big indecomposables representations of virasoro and of its commutant , and can be exhibited in the form of what we will call a `` staircase diagram '' .
although our approach does not so far yield much information on the detailed structure of the virasoro indecomposable representations , it is enough to give access to the fusion rules for our logarithmic theories .
we work these out in detail in some cases .
the remainder of this paper is organized as follows .
section 2 describes the general framework of the lattice models , algebraic structures , and the continuum limit , for @xmath27 not a root of unity , and also the general features of cases in which @xmath27 is a root of unity .
the following three sections go into details about particular examples of special interest .
each one begins with a summary of the physical systems considered , viewed as loop models , or as field theories of certain nonlinear sigma models or landau - ginzburg theories , then describes the algebraic structures leading to the continuum limit , and ending with the fusion rules .
section 3 is the case @xmath30 ( or central charge @xmath3 in the continuum limit ) , which corresponds to symplectic fermions or dense polymers .
section 4 is the case @xmath31 ( @xmath4 in the continuum limit ) , which corresponds to percolation , the spin quantum hall transition , or a certain nonlinear sigma model at @xmath23 .
section 5 is the case of dilute polymers ( @xmath4 ) , or of the critical point in a certain nonlinear sigma model or landau - ginzburg theory .
for this theory we discuss the continuum directly , and not the lattice models .
finally , section 6 is a more physical discussion of the structures for the cases with @xmath4 , using operator product expansions , and making contact with the work of gurarie and ludwig @xcite . to make reading easier
, we provide a list of notations for some of the algebraic objects we introduce , all of which implicitly depend on @xmath27 or the central charge @xmath32 , and also on length @xmath33 for the tl algebra : @xmath34 : tilting modules over @xmath28(sl@xmath29 ) @xmath35 : projective modules over tl@xmath36 @xmath37 : standard modules over the virasoro algebra @xmath38 : simple ( or irreducible ) modules over the virasoro algebra @xmath39 : projective modules over the virasoro algebra
what we will do here is ( i ) take the continuum limit of the lattice model on a strip to obtain the representation content of the states , and ( ii ) also use the lattice to infer fusion rules for the boundary cft . in the companion paper @xcite
, we showed that the commutant algebras of the temperley - lieb algebra in some supersymmetric lattice models are morita equivalent @xcite to @xmath28(sl@xmath29 ) .
accordingly , here we can use the @xmath28(sl@xmath29)-invariant spin-1/2 chain , secure in the knowledge that the symmetry structure is the same in the supersymmetric constructions that apply when @xmath40 is an integer .
we often refer to representations as modules , as these are essentially the same things for our purposes ( however , in the more mathematical discussions , saying `` representation '' in place of `` module '' would be quite misleading in some places , especially when preceded by `` projective '' ) .
then an irreducible representation is the same thing as a simple module , and it is also common to speak of modules being `` over '' the algebra that they represent .
we assume familiarity with basic algebraic concepts such as sub- and quotient- objects , here usually for modules . for a general reference on the algebraic background
, we suggest ref .
@xcite .
the hamiltonian of the @xmath28(sl@xmath29)-invariant spin-1/2 chain is essentially the xxz chain , but with some boundary terms ( here the @xmath41 operators are pauli matrices):@xmath42 the hamiltonian @xmath43 is built from generators @xmath44 , @xmath45 of the tl algebra tl@xmath46 , which is defined as the algebra generated by a set of elements @xmath44 , @xmath47 , @xmath48 , ,
@xmath49 , subject to the relations @xmath50and no other relations algebraically independent of these .
these are satisfied by the above expressions in the spin-1/2 chain .
then @xmath51 ( up to an irrelevant additional constant ) .
all future references to the spin-1/2 or xxz chain are to this model .
first we briefly summarize the case of @xmath27 generic ( i.e. , @xmath27 not a root of unity ) and its continuum limit . for @xmath27
not a root of unity , the tl algebra is semisimple and has irreducible representations ( all such will be referred to as simple modules hereafter ) labeled by @xmath52 , @xmath0 , @xmath48 , , @xmath53 but restricted to values such that @xmath54 is an integer .
the dimensions of these simple modules are @xmath55,@xmath56 these modules are referred to as standard ( or specht ) modules .
the commutant of the tl algebra in the spin-1/2 chain is a finite - dimensional homomorphic image of @xmath28(sl@xmath29 ) , that we will denote @xmath25sl@xmath57 ( it is also called the @xmath27-schur algebra ) . for @xmath27 generic ,
its simple modules are again labeled @xmath52 , @xmath0 , @xmath48 , , @xmath58 , where @xmath54 is an integer .
throughout the paper , we will use the familiar term `` spin '' to refer to @xmath59 , unless otherwise noted .
these modules have dimensions @xmath60 , independent of @xmath61 except that @xmath62 .
they are called weyl modules [ or standard modules of @xmath28(sl@xmath29 ) ] ( we can drop the superscript @xmath61 as they are independent of it , and the limit @xmath63 can be taken for these ) .
they are the deformations to @xmath64 of the usual irreducibles of su(2 ) .
if two chains of lengths @xmath65 , @xmath66 are joined end to end ( with @xmath65 on the left ) , one obtains fusion rules for both algebras .
the states of a single chain of length @xmath65 fall into multiplets ( simple modules ) of the tensor product of commuting algebras , tl@xmath67sl@xmath68 which are labeled by @xmath59 ; these modules are the tensor product of the corresponding simple modules of the two respective algebras .
when the two chains are joined , the resulting chain of length @xmath69 can be analyzed in the same way . if the standard modules of @xmath28(sl@xmath29 ) are denoted @xmath70 , then tensor products of such modules ( where the tensor product is defined as in the usual way for vector spaces ) decompose fully into a direct sum,@xmath71exactly as for su(2 ) .
the standard modules of tl must behave in a similar way , but now the product operation is defined using induction ( which may be familiar from induced representations in group representation theory ) from the product space as a module over tl@xmath72tl@xmath73 to a module over tl@xmath74 , which contains the former as a subalgebra . the general theory described in ref .
@xcite implies that indeed the `` fusion rule '' for this definition of a product of tl modules must take the same form as that above for @xmath28(sl@xmath29 ) ( note this product is a functor , and not just a decomposition of the tensor product of vectors ; in particular the dimensions of the tensor product of the modules on the left and the sum of those on the right are not equal ) .
now we pass to the continuum limit .
from this point on , we consider only @xmath75 even for the time being .
( this does no harm as the modules with @xmath59 integer close on themselves under products , as we see in the fusion rules above , and will see for non - generic @xmath27 below .
this restriction corresponds to the `` oriented loops '' models in ref .
@xcite . ) in the continuum limit we consider only the fourier modes of the @xmath44s of wavevectors @xmath76 with @xmath77 fixed , and also consider only states spanned by eigenstates with energies @xmath78 ( measured from the ground state energy ) with @xmath79 fixed .
then the fourier modes of the @xmath44s become the virasoro generators @xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 ,
the standard modules of the tl algebra , labeled by @xmath59 , become certain modules of the virasoro algebra , which we will also call standard , and which are still labeled by @xmath59 .
these modules are highest weight modules , that is they are generated by the action of the @xmath80s on the state of lowest energy or of lowest @xmath24 eigenvalue in the module .
they are not , however , verma modules , which are constructed from a highest weight state by lowering by @xmath80s with @xmath86 , and using only the commutation relations of the virasoro algebra .
the conformal weight @xmath87 of the module , which is the @xmath24 eigenvalue of the highest weight state , takes one of the values @xmath88 at which a null vector ( the highest weight in a submodule ) appears in the verma module , which must be set to zero in order to obtain a simple module . for generic values of the central charge , there is only a single null vector , and only when the conformal weight takes one of the values in the kac table @xcite@xmath89^{2}-1}{4x(x+1)},\label{kac}\]]in which @xmath90 is determined by the central charge @xmath91 the generic values of @xmath32 are those where @xmath90 is irrational .
when we take the limit of spin-1/2 chain , the resulting central charge follows from @xmath92 . the conformal weights that occur are @xmath93 in the kac table . in the corresponding verma module ,
the null vector is at conformal weight @xmath94 , hence the virasoro character of the simple modules that appear as direct summands in the @xmath95 limit of the chain are ( the trace is over states in the simple module only)@xmath96 where @xmath97 , @xmath98 is a parameter , @xmath99 , where @xmath100 is the inverse temperature ( the length in the imaginary time direction ) , and @xmath101 is the inverse of the euler partition function , and is related to the dedekind @xmath102 function , @xmath103 thus the partition function of the continuum limit of the xxz chain in the generic case is e^-(h - e_0(l))=tr ^ l_0= _ j=0^(2j+1).here the continuum limit is taken with @xmath97 fixed , after subtracting the ground state energy @xmath104 of the finite-@xmath33 chain from the hamiltonian , and the trace is over all the states of the spin-1/2 chain .
[ we emphasize that this is the ordinary trace on the vector space , not a @xmath27-trace ; it corresponds to what were called the modified partition functions in ref.@xcite . ]
if instead the subtraction were the length times the ground state energy density @xmath105 of the chain as @xmath95 , this expression would be multiplied by @xmath106 .
this is the way we will define the partition function here and in the following : z= e^-(h - e_0(l)l)=tr ^ l_0-c/24= _ j=0^(2j+1).let us note that this partition function can also be obtained by field - theoretic ( rather than algebraic ) means , by bosonizing the spin-1/2 chain or the 6-vertex model , that is , by representing it by a scalar field with a background charge , and taking the continuum limit @xcite .
it is known that @xcite , as a consequence of the null vectors that were set to zero , the operator products of the fields @xmath107 that correspond to the states in the modules decompose according to the fusion rules @xmath108 these clearly correspond to those we inferred from the induction product in the chains , or from the @xmath28(sl@xmath29 ) symmetry of the whole construction . in the following we use similar reasoning to infer fusion rules also in the non - generic cases in which representations are not fully decomposable . in this procedure , we have constructed the fusion product algebraically by joining two chains end to end and taking the continuum limit .
one may wonder whether this really yields the operator product of corresponding local boundary fields in the cft , which is usually pictured as bring points on the boundary together .
we will not discuss this fully , but provide some heuristic arguments .
if we view joining the chains as an event in time , then it corresponds to a picture like that shown in fig.[slitstrip ] , in which the continuum limit has been assumed .
( the chains are shown as equal in length , though more generally the ratio of their lengths can be arbitrary , but is held fixed when the continuum limit is taken . ) if we wish to consider the fusion of states that are @xmath24 eigenstates , for example , then the two initial states may be prepared at early imaginary time @xmath109 , and the matrix element with a desired final state can be taken , by preparing that state at @xmath110 . as conformal transformations move points within the interior of the domain ( and even deform the domain ) , it is natural that the tensor product of two copies of virasoro merges into a single copy of virasoro , as a proper subalgebra .
this occurs because the continuum analog of the extra generator @xmath111 must be included along with the others
. then the corresponding actions of these algebras on the states are related by the induction functor .
thus the use of the induction functor for virasoro is natural in the slit - strip picture ( i.e. , in the operator formalism ) .
the conventional picture of the incoming and outgoing states as corresponding to local fields at a point on the boundary , which are brought together in an ope , is recovered by using a conformal mapping from the slit - strip to the unit disk , or the upper half plane , with the two incoming and one outgoing state(s ) at time @xmath112 represented by points on the boundary .
this conformal mapping is described for example in ref .
incidentally , it has often been remarked in the cft literature that the fusion product is not simply a tensor product of representations of virasoro ( or more generally , the chiral algebra ) .
the fusion product defined as the induction functor for joining strips end to end should agree with the `` deformed tensor product '' defined in some of these references @xcite , at least for the present case of boundary cft .
the states or fields we find in this construction based on the tl algebra at generic @xmath27 correspond to the first column of the kac table .
the first row of the kac table has identical fusion rules , and so one would expect a lattice construction to lead to this theory also . such a model is the o(@xmath113 ) loop model @xcite and its supersymmetric versions @xcite in which the critical theory ( which exists for @xmath114 ) is referred to as the dilute theory , and where another algebra ( essentially a two colour version of the tl algebra ) replaces the tl algebra .
we emphasize that it has the same symmetry as the tl ( dense loops ) model , and that it too comes in two versions , oriented loops ( only integer @xmath115 ) and unoriented loops ( including all integer values of @xmath116 ) .
when @xmath27 is a root of unity ( but @xmath117 ) , we will define @xmath118 to be the smallest integer such that @xmath119 .
the structure of both tl and @xmath28(sl@xmath29 ) changes , and this can be traced to the vanishing of the @xmath27-integers @xmath120_q=(q^n - q^{-n})/(q - q^{-1})$ ] for some @xmath121 for such @xmath27 values : @xmath120_q=0 $ ] for @xmath122 ( mod @xmath123 ) @xcite . for @xmath27 a root of unity , standard modules of both tl and 2 can be defined , and have the same dimensions @xmath55 , @xmath124 , respectively , as before . however , they are not all irreducible .
it turns out that for @xmath59 such that @xmath125_q=0 $ ] , they remain irreducible , while for other @xmath59 , they contain a proper submodule but are indecomposable . for both algebras ,
the reducible standard modules have the same structure : the largest proper submodule ( or radical ) of the module is irreducible , and the quotient by this submodule is also
. then we will represent the structure of such a module by a diagram like ccc & & + & -.2 cm & + & & -.3cm[stanmod ] . here
the open circle represents the states in the simple quotient module or `` top '' ( or `` head '' ) , and the closed circle the simple submodule or `` foot '' .
the arrow represents the action of the algebra ; there is some element of the algebra that maps the top to the foot , but not vice versa , as well as elements that map the top into itself and the foot into itself .
the positions of the circles on the page is not particularly important , as the key information is the arrows and ( for these standard modules ) whether a circle is open or closed .
below , we will also represent the structure of other modules by similar , though more complex , diagrams . in these ,
each circle is a nonzero simple subquotient module , and arrows show the action of the algebra other than within the simple subquotients , with the convention that composites of arrows should also be understood as present implicitly . in our examples , a head and a foot can usually be defined uniquely , but there will be other subquotients in between .
later , the positions of the circles will be replaced by additional information such as the spin or dimension of the simple module in question .
the general rule for the spin values can be given here : for either 2 or tl , we label standard modules by @xmath52 , @xmath48 , , @xmath33 .
we choose the labels for the simple modules to be @xmath52 , @xmath48 ,
, @xmath33 also ( or in some cases , a subset of these ) . for reducible standard modules of @xmath25sl@xmath26 , the simple module at
the _ top _ has the same value of @xmath59 ; the foot is a simple module isomorphic to a simple module with _ smaller _ @xmath59 . for the tl algebra ,
it is the same , except that the foot has a larger @xmath59 than the head .
when the xxz chain is analyzed for @xmath27 a root of unity , it decomposes under the tl algebra into a direct sum of tl modules .
when analyzed under @xmath25sl@xmath26 , it likewise is a direct sum .
however , in neither case are the direct summands the standard modules just described , except for those standard modules that remain simple at this @xmath27 , some copies of which appear as in the generic cases . for the reducible standard modules , each algebra action mixes them , mapping states from one standard to another @xcite .
the direct summand modules will be referred to as `` tilting '' modules , in the case of either algebra .
( we should point out that non - semisimple algebras typically have infinitely many distinct , i.e. mutually non - isomorphic , indecomposable modules , even when the algebra is finite dimensional .
it becomes important to explain _ which _ particular types of indecomposable modules are of interest . )
the tilting modules turn out to have a fairly simple structure also , consisting of two standard modules with arrows connecting them , typically in the form : ccccc & & & & + & -.2cm&&-.2 cm & + -.3cm&&&&-.3 cm + & -.2cm&&-.2 cm & + & & & & .[tiltmod ] thus , the module is seen to contain a submodule that is a standard module , consisting of an open and a closed circle together with an explicit arrow that connects them .
the quotient by this submodule is also standard .
the detailed structure of all these modules will be exhibited for several important values of @xmath27 ( or @xmath113 ) in the following sections , as will the inter - relation of the commuting tl and 2 actions .
in addition to the tilting modules , other important modules are the projective modules .
projective modules may be defined as occurring as a direct summand in some free module , or in several other equivalent ways .
we will be especially interested in the indecomposable projective modules of the tl algebra . for finite dimensional algebras ,
there is one indecomposable projective module ( up to isomorphism ) corresponding to each distinct simple module ; the latter appears as the top ( the unique simple quotient module ) of this projective module .
this fact makes it fairly straightforward to read off the indecomposable projective modules from the structure of the xxz chain as a direct sum , since the algebra acts faithfully . in many cases ,
the tilting modules are also indecomposable projective modules . joining chains of lengths @xmath126 , @xmath127 end to end leads to natural operations making a tensor product of modules over 2 into another such module .
these agree with the product operations usually defined for 2 as a quantum group .
this product decomposes as a direct sum of indecomposable modules , and as tilting modules are defined as direct summands in the xxz chain , the tensor product of tilting modules is again tilting .
the fusion rules for the tilting modules are then those of most basic interest ( and are determined by the decomposition of a product of indecomposable tilting modules into indecomposables ) . also , as this product is just the ordinary tensor product operation of vector spaces , the sum of the dimensions of the indecomposables produced is the product of the dimensions of the modules we began with . for @xmath27 a root of unity ,
the fusion rules are however modified from those in the generic cases , and will be discussed in the examples below . in our previous paper @xcite , we also explained how a product is defined on the tl modules when joined end to end , based on induction as mentioned earlier . in this case , the dimensions of modules are not conserved , but there are again fusion rules .
crucially , the module induced from a tensor product of projective modules of the two short chains is a projective module of the chain of length @xmath128 , and can be decomposed as a direct sum of indecomposable projective modules .
hence we can define fusion rules for products of projective modules , which are again determined by the decomposition of a product of indecomposable projective modules into indecomposable projective modules .
most importantly , the fusion rules for indecomposable projective modules of tl are the _ same _ as those for indecomposable tilting modules of @xmath25sl@xmath26 @xcite .
we also discussed previously @xcite how the limit @xmath95 can be taken , both at a purely algebraic level , and focussing on the low - energy states as a continuum limit .
the first can be well - defined mathematically , while the second is presently less rigorous and more heuristic .
for the symmetry structure , given by 2 or a morita equivalent of it , the purely algebraic construction is sufficient , as it yields a limit algebra , which is 2 for the xxz case .
the dimensions of modules are stable as @xmath33 increases , but additional modules appear .
the fusion rules are also given by the finite - dimensional analysis and remain the same as @xmath33 increases [ and in discussing 2 above , it was natural to ignore the distinction between it and its finite - dimensional quotients ] . on the other hand , for the tl algebra , the dimensions of corresponding modules increase with @xmath33 , and there seems to be no purely algebraic way to define the limiting algebra that is suitable for our purposes @xcite .
nonetheless , the fusion rules behave stably ( as they are controlled by the symmetry ) . in the purely algebraic limit ,
the limit of a projective module is still a projective object in the limiting category of modules . in the continuum limit
, all evidence suggests that the modules of tl become virasoro modules .
the limits of projective modules will still be referred to as projective , and continue to play an important role . the standard ( resp . ,
tilting ) modules of tl become standard ( resp . , tilting ) modules of virasoro , also .
we note that the virasoro algebra may not act faithfully on the limiting space of states .
this is to be expected as the representations that occur as summands are quite restricted , so that the algebra appears smaller it is a quotient of virasoro .
we can see this already for generic @xmath27 , where the standard modules became the irreducible representations of virasoro that lie in the first column of the kac table .
it is possible that the modules we call projective do obey the general definition of projective modules over this smaller algebra , even if they do not for the virasoro algebra itself .
also , there could in principle be elements of the algebra not in the virasoro algebra ( though this seems unlikely in the spin-1/2 chains and related theories studied here ) .
when @xmath27 is a root of unity , the corresponding central charge is one of the special values of kac and bpz ( in which @xmath90 above is rational ) , at which the standard modules may acquire additional null vectors .
when this occurs , the standard module of virasoro now has the same submodule structure as the standard tl module illustrated above .
the positions in the first column of kac table at which this occurs agree exactly with what we would expect based on the tl algebra .
similarly , we will infer that the projective and tilting virasoro modules relevant to the analysis of the xxz chain ( to operators as well as states , by the state - operator correspondence ) have structures like that above ( when reducible ) .
finally , the fusion rules must be those inferred from finite size systems , and for the projective modules are again the same as for finite size @xmath33 .
these can in turn be obtained from those on the 2 side , which is very convenient as one can use characters there .
this illustrates how symmetry becomes useful in analyzing the structure of a cft .
thus , in the following sections , the ideas of this part will be worked out in several instances of physical or geometric interest .
in this section , we consider the case of the open xxz chain of even length @xmath130 , with @xmath129 , so @xmath30 and @xmath131 .
when the partition function is expanded as a sum over loops , this model assigns a factor @xmath129 to each loop , so describes dense polymers .
this case also corresponds ( by morita equivalence of symmetry algebras ) to the gl(@xmath1 ) supersymmetric chains , and to nonlinear sigma models with target space @xmath132 at a particular value of the coupling constant , as described in ref.@xcite . for @xmath133 , the gl(@xmath12 )
spin chain actually coincides with the 2 or spin-1/2 chain , and both are free fermion systems . in the continuum limit , the latter theory is the free symplectic fermion theory . the decomposition of the space of states into a sum of indecomposable representations ( tilting modules ) under 2 has been well studied .
it turns out that none of the indecomposable summands are simple ; all are reducible . following the notation described in the previous section , each tilting module @xmath134 has the structure corresponding exactly to that in diagram ( [ tiltmod ] ) @xcite : @xmath135 the nodes correspond to simple modules and are labeled by spin values ; the simple module of spin @xmath136 has dimension @xmath137 .
( nodes with negative spin values are viewed as zero dimensional and can be omitted . ) for some purposes , the module @xmath138 for @xmath52 can also be viewed as a tilting module of 2 but does not occur in the chain because its multiplicity is zero ( this is special to the @xmath129 case ) ; in the present section we will view it also as zero .
each pair of nodes that can be connected by a southwest arrow forms a standard module , as in diagram ( [ tiltmod ] ) .
as advertised , the dimension of the standard module of spin @xmath139 is @xmath140 ; conversely , this together with a knowledge of the structure enables one to calculate recursively the dimensions of the simple modules .
the total dimension of the module is thus @xmath141 .
we emphasize that the same structure applies to the corresponding tilting modules for the enlarged symmetry algebra @xmath142 , by morita equivalence , though the dimensions are different .
similar statements will be true for the symmetry algebras throughout , but will not usually be made explicit . for @xmath143 ( i.e. six spins @xmath0 ) for example , the @xmath59 values of the indecomposable summands @xmath134 are @xmath144 , @xmath145 , @xmath146 .
the full decomposition of the chain under 2 ( where @xmath147 denotes multiplicities of the indecomposable tilting modules ) is @xmath148 and the check on the dimensions is @xmath149 knowing the decomposition under ( an image of ) @xmath25sl@xmath26 , one can also find directly the structure of the modules of the commutant algebra of 2 in the spin-1/2 chain .
this by definition consists of all endomorphisms ( linear maps of the vector space into itself ) that commute with the 2 action .
when the algebra is semisimple , as for 2 at generic @xmath27 , such endomorphisms can only map a copy of a simple module to another copy of the same one , and so the commutant is also semisimple , and has simple modules of dimensions equal to the multiplicities of the simple modules of the given algebra ( this was used in ref.@xcite ) . for non - semisimple algebras , as here , there are such endomorphisms that map the tilting modules to themselves and which combine to form full matrix algebras .
indeed , the multiplicities of the 2 tilting modules are the dimensions of the simple modules of its commutant .
but there are also endomorphisms that map the indecomposable tilting modules into one another .
this can occur when a tilting module has one or more subquotient modules that also occur in another tilting module .
in fact the `` top '' of the indecomposable tilting module , which is a source but not a sink for arrows , must map onto a non - top part of the other tilting module , and all subquotients that lie below the top must also map such that the arrows are preserved . in most cases ,
the foot of the tilting module , and some subquotients above it , are mapped to zero .
thus the endomorphisms of the space of the spin-1/2 chain can be read off from the structure and multiplicities of the indecomposable tilting modules . for the tl representation in the spin-1/2 chain ,
the commutant algebra of 2 obtained in this way is isomorphic to the tl algebra ; the two algebras form a `` dual pair '' .
( if instead the tl structure were given , one could follow the same procedure to find the structure of the modules of its commutant . ) in the present case , the indecomposable tilting modules @xmath150 have non - trivial homomorphisms into neighboring tilting modules @xmath151 only .
there are also ( for @xmath152 ) non - trivial homomorphisms of @xmath134 into itself , which map the head to the foot , which have the same @xmath59 value .
the commuting actions of 2 and its commutant tl@xmath36 can be represented in a `` staircase '' diagram that generalizes those for 2 or tl separately , shown for @xmath143 in fig .
[ overall ] .
this shows the action of both algebras , and thus the tensor product @xmath25sl@xmath153tl@xmath36 , on the states of the chain .
[ the diagram is the same if we consider @xmath154tl@xmath36 for the supersymmetric chains . ]
the circles at cartesian coordinates @xmath155 represent simple subquotient modules over this product algebra .
@xmath25sl@xmath26 acts by the horizontal arrows ( that is , its action preserves @xmath139 ) , while tl@xmath36 acts `` vertically '' ( that is , preserving @xmath59 ) , and these actions commute ( readers are invited to check that this is consistent , because of the arrangement of the arrows as `` commuting squares '' ) . once again , composites of arrows are arrows that are implicitly present .
each circle is a subquotient module with a multiplicity of one when viewed as a @xmath25sl@xmath153tl@xmath36 module ; some values @xmath155 do occur twice , and those circles have been separated slightly for clarity .
the indecomposable tilting modules of 2 can be recovered by ignoring vertical arrows , and stretching out the components of the diagram to recover the diagrams above .
likewise , the indecomposable tilting modules of the tl algebra are obtained by ignoring the horizontal arrows .
( if we had started knowing those , then we could have obtained the tilting modules for the commutant of tl by a similar procedure as we described . )
we now turn to the structure of these in some examples . for this @xmath129 , @xmath33 integer case ,
the general form is always a single indecomposable module under the product algebra , and for @xmath156 simply extends up to larger @xmath59 values . in detail for @xmath143 , the tl indecomposable tilting modules ( direct summands ) have the following structure , in which the nodes are labelled by the dimension of the corresponding simple subquotient module over tl , and arrows now denote action of the tl algebra @xcite : @xmath157 with total dimension @xmath158 notice here the appearance of multiplicities from the 2 analysis as dimensions of simple tl modules , and vice versa . using the labels @xmath59 for the tl simple modules as already described , eq .
( [ tlindec ] ) becomes @xmath159 for general @xmath160 , the pattern is similar , starting with a `` trio '' at @xmath144 , followed by quartets up to @xmath161 , and then a single simple module also at @xmath161 ( here we assign @xmath59 to the tl tilting modules by the @xmath59 of the head ) . the tilting modules for tl can also be compared with the standard modules .
the standard module labeled @xmath59 for tl@xmath162 consists of the simple modules @xmath59 , @xmath163 , together with the tl action that connects them by an arrow , except for @xmath161 when the simple module is standard and one dimensional ( @xmath164 ) , and @xmath52 where the standard module is simple . for six spins
@xmath0 again , one has @xmath165 , @xmath166 , @xmath167 , @xmath168 .
the indecomposable tilting modules consist of pairs of standard modules `` glued '' together by the action of some tl elements , similarly to @xmath25sl@xmath26 .
when the structure of the tilting modules is known as in ( [ tlindeci ] ) , one can recursively calculate the dimensions @xmath169 of the simple modules of tl , starting from @xmath161 . for @xmath143
this gives @xmath170 , @xmath171 , @xmath172 , @xmath173 .
usually there is a simple module of tl for @xmath52 , but here for @xmath129 it vanishes .
( this is related with the fact that the partition function of the chain vanishes exactly ) . for general even @xmath130 ,
one finds the formula for the dimensions of simple tl modules for the @xmath30 case , d_j^0=_j=j , j+1, ,l(-1)^j
- jd_jfor @xmath52 , @xmath48 , , @xmath33 .
we may also consider @xmath25sl@xmath174 and analyze it viewed as a module over itself , acting on the left or on the right ( the left and right regular representations ) , or on both sides ( i.e. as a @xmath25sl@xmath175sl@xmath174 module , as the actions on the two sides commute ) .
from the last of these , we can then make a similar staircase diagram for @xmath25sl@xmath174 as a module over @xmath25sl@xmath175sl@xmath174 , from which the left or right structure can be determined .
the cellular ideals we reviewed in ref .
@xcite can be readily seen in such a picture , as they form sets of circles and arrows mapped into themselves by the left and right action of @xmath25sl@xmath174 .
this is also useful in calculating the total dimension of the algebra , which is equal to @xmath176sl@xmath177 , even though it is not semisimple
. one can do the same for the tl algebra , and then @xmath178 ( @xmath179 for @xmath143 ) . similarly , for the spin-1/2 chain itself , we can point out that the total dimension is @xmath180 , again just as if the algebras were semisimple .
of course all of these expressions for the dimensions follow from the independence of the dimensions of the standard modules from @xmath27 , so that the formulas from the generic case always hold .
this is also true for the other non - generic values of @xmath27 , some of which are considered later . for the models with morita equivalent symmetry algebras ,
@xmath124 should be replaced by the dimensions @xmath181_{q'}$ ] , where @xmath182 @xcite , but again it is true that the formulas can be obtained `` as if '' the standard modules were simple ( similar formulas also hold for the @xmath27-dimensions [ in the 2 case ] and the superdimensions [ in the superalgebra cases @xcite ] ) .
we will give here the expressions for the ( super-)dimensions of the simple modules of our algebras @xmath183 , @xmath133 , @xmath145 , , which are morita equivalent to 2 with @xmath30 @xcite .
first we point out that the dimensions @xmath163 of the @xmath59th simple module of 2 can be deduced from the structure of the standard modules , and the dimension @xmath124 of the latter , much like those for tl .
they are obtained from the relations @xmath184 , with dimension @xmath48 for the simple module with @xmath52 , which follow from the existence of simple sub- and quotient modules of the each standard module .
for the algebras in the supersymmetric models , the standard modules have dimensions @xmath181_{q'}$ ] , where @xmath182 @xcite .
then as the structure of submodules and quotients is the same as for 2 because of the morita equivalence @xcite , we have for the dimensions @xmath185 of the simple modules the corresponding recurrence @xmath186 , with @xmath187 obviously corresponding to a simple module .
this is solved by d_j^0&=&d_j-d_j-1+d_j-2- + & = & q^2j+q^2j-4+ +q^-2j + & = & [ j+1]_q^2 .
similarly , we may look at the quantum dimensions of the modules ; for the algebra @xmath183 the quantum dimensions are just the superdimensions . for the @xmath59th standard module ,
the superdimensions were denoted @xmath188 in ref .
@xcite , given by @xmath189_q$ ] , which is the same as the @xmath27-dimension for 2 because of the morita equivalence with 2 as ribbon hopf algebras ( here for the algebras with integer @xmath59 only ) @xcite . in the present case , @xmath190 .
the same equality applies to dimensions of other modules such as the simple modules , and thus we can compute the quantum dimensions for 2 ( @xmath30 ) simples and @xmath183 simples as @xmath191_{q^2}$ ] similarly . for the tilting modules , the quantum dimensions are @xmath192 ( @xmath144 , @xmath145 , ) , and so @xmath193 for all @xmath194 . for a finite - dimensional algebra ,
the indecomposable summands in the left regular representation are the indecomposable projective ( left ) modules .
thus we can read off the indecomposable projective modules .
for the tl algebra ( for which these are of particular interest ) for @xmath143 , these are the same modules as the three tilting modules other than the singlet
. the singlet can not be an indecomposable projective module as it occurs as a quotient module of an indecomposable tilting module , which is impossible for a projective module ( one of the alternative definitions for a projective module is that whenever it is a quotient of some module , it is isomorphic to a direct summand in that other module ) .
the other three indecomposable tilting modules are acceptable , and are projective as follows automatically if the tl algebra is defined as the commutant of 2 in this chain ( this is an aspect of the key lemma @xcite ; see also ref.@xcite ) .
there are no other indecomposable projective modules in the present case ( for the other cases @xmath195 , we will see that there are also projective modules that are not tilting ) .
so the indecomposable projectives of the tl algebra are @xmath196 this pattern is repeated for general @xmath197 ( integer ) ; the indecomposable projectives @xmath198 of tl for @xmath30 are the indecomposable tilting modules , other than the singlet at @xmath161 : p_j : ccccc & & -.7 cm j & & + & -.2 cm & & + j-1&&&-.3 cm j+1 + & -.2 cm & & + & & -.7 cm j & & , where once again we understand in this case that the simple modules with @xmath199 or @xmath200 are zero and can be omitted .
( for @xmath201 , one does have a two - dimensional indecomposable @xmath202 , because of the implicit arrow from the head @xmath144 to the foot @xmath144 . )
we now pass to the continuum limit
. our strategy will be to deduce as much as possible from the lattice , and compare with known results from the representation theory of the virasoro algebra .
first , we can consider the purely algebraic limit @xmath95 as discussed in ref .
the staircase diagram then becomes infinite ; see fig .
[ contoverall ] , and the limits of the dimensions @xmath169 of the simple tl modules are also infinite . the more physical way of taking @xmath95 , called the continuum limit @xcite , focusses on low - energy states and the long - wavelength fourier components of the @xmath44s .
we expect that the virasoro algebra emerges from the tl algebra in this limit @xcite , and that the staircase diagram still applies to this .
in addition we will use the known information ( as for the case of generic @xmath27 ) of the values of the central charge and a set of scaling dimensions that are associated with the virasoro modules ( or the scaling fields of the theory ) .
this is done for convenience ; we expect that with some additional effort these may themselves be established purely on the grounds of consistency with the algebraic structure encoded in the staircase diagram , since the structure of the 2 modules is known and is unchanged in the @xmath95 limit . on specializing the generic results , eqs .
( [ c ] ) , ( [ kac ] ) for the central charge and the kac table to @xmath30 , so @xmath203 , one finds that @xmath3 , and the kac formula reads @xmath204 where @xmath123 , @xmath205 are positive integers .
the values appearing in the spectrum are @xmath206 where we used symmetries of the kac formula to restrict to the fundamental domain , and again here @xmath207 is integer .
as the limits of the dimensions of the modules over tl are infinite , it is convenient to use instead the graded dimensions or characters defined using a trace over the module of the exponential of the hamiltonian , or @xmath24 virasoro generator . for the standard modules , which we will denote @xmath208 , these were given in sec . 2 , and for the cases of interest here
are @xmath209 we may point out here that the full partition function of the model in the continuum limit is obtained by specializing the result for generic @xmath27 : z = tr ^
l_0-c/24= _ j=0^(2j+1),which as for the total dimension of the finite length chains is still valid at @xmath30 , even though the modules involved are no longer fully decomposable .
this is similar to the result given for the case @xmath4 in the appendix of ref .
this result may be obtained using field - theoretic methods @xcite , and provides a check on our algebraic deductions .
it is perhaps more natural to calculate a partition function as a quantum trace instead of a trace , which corresponds to a boundary condition in the ( periodic , imaginary ) time direction that does not break the 2 or @xmath183 symmetry . in that case
@xmath124 is replaced by @xmath188 , and the result is valid for the continuum limit of the supersymmetric models as well as that of the spin-1/2 chain @xcite . in the present case @xmath30 , these partition functions ( called `` unmodified '' in ref .
@xcite ) vanish identically ( this is also true on the lattice ) .
this follows because we can collect all states into tilting modules of the symmetry algebra , and the quantum dimensions @xmath210 of these tilting modules vanish .
for the modified partition functions of the supersymmetric models , @xmath124 must be replaced by @xmath211 as for the finite-@xmath33 chains . in studying the representation theory for highest - weight modules of the virasoro algebra , it is conventional to use verma modules .
the verma module generated from the highest weight vector with weight @xmath93 has , for generic central charge , a singular vector at level @xmath212 , with value of the conformal weight @xmath94 .
that is , there is a submodule which is also highest weight with that conformal weight , and vectors in the submodule are orthogonal to all those in the verma module . at @xmath3 ,
another singular vector appears at the lower level @xmath213 , while now @xmath214 .
these submodules in turn contain submodules .
this leads to a picture of verma modules embedded into one another in a sequence , as shown in fig.([vermemb ] ) .
the representations obtained by factoring out from the verma module the submodule generated from the `` generic '' singular vector at level @xmath212 are the standard modules @xmath215 . at @xmath3
this representation is reducible ; from fig . [ vermemb ] it contains a submodule that is simple , and its structure is described by a diagram of the form ( [ stanmod ] ) .
it possesses an irreducible quotient @xmath216 , with character @xmath217 , and the character of the submodule in @xmath215 is @xmath218 .
hence we have the equality @xmath219 the calculation of @xmath217 from the embedding diagram fig.[vermemb ] is standard , and gives @xmath220 and note that @xmath221 ( the continuum equivalent of the absence of the @xmath222 node in fig . [ overall ] ) .
this shows that the structure of the standard virasoro modules at @xmath3 parallels that of the finite-@xmath33 tl modules of the same @xmath59 .
we now go a step further and describe the structure of virasoro modules that parallel the indecomposable projective modules of tl ; these will be called indecomposable projective modules of virasoro ( the terminology is subject to caveats already mentioned in the previous section ) .
the modules we want are the continuum limits of the finite-@xmath33 indecomposable projective tl modules . calling these indecomposable projective modules @xmath223
, we expect thus in particular to have the characters @xmath224 the structure of these modules , in the sense of the submodules that appear , is as follows .
all are described by the same diagrams as the corresponding projectives @xmath198 of tl , and all except that for @xmath144 are `` quartets '' since the effects at @xmath161 have disappeared as @xmath95 .
each circle in the diagram is a simple subquotient module , and for the virasoro algebra these are highest weight modules . in more detail , we should first have an irreducible representation @xmath216 generated by a highest weight vector @xmath225 at the `` foot '' ; then a standard submodule @xmath226 generated by a highest weight vector @xmath227 , in which @xmath225 appears as a singular vector .
then a state @xmath228 which forms a two dimensional jordan cell ( of @xmath13 ) with @xmath225 : @xmath229 @xmath228 is not a highest weight vector , and positive modes of virasoro ( i.e. @xmath80 s with @xmath230 ) map it to descendants of @xmath231 .
finally , there should be a @xmath232 which is technically a sub - singular vector , i.e. it is singular in the quotient by the representation generated by @xmath225 ( in other words , positive modes of virasoro send it to descendants of @xmath225 ) .
this is schematically represented in fig.[repstruc ] , where arrows indicate action of the virasoro algebra .
nodes connected by southeast arrows contribute @xmath233 and @xmath234 to the character ; these correspond to a submodule and a quotient module , both of which are isomorphic to standard modules ; the head and foot of each of these are indicated by the open and closed circles . in the cft ,
these vectors that are highest weights modulo submodules will be the closest that we can come to finding primary fields for the virasoro algebra . representations with the quartet structure such as @xmath39 have been abstractly studied before and are called staggered modules in @xcite .
it can be shown that they depend on one characteristic parameter , which can be defined through the equation @xmath235 we do not know the relevant values of @xmath236 in our models ; it would be interesting to obtain them from the spin-1/2 chain , or from the symplectic fermion field theory discussed below .
presumably our modules are special as they are `` projective . ''
now we interpret the continuum limit using the symplectic fermion representation of the theory .
this follows from writing the spin-1/2 chain , or the supersymmetric formulation for @xmath133 , as a free - fermion system .
the continuum limit can then be taken .
as we consider the open chain case , one must treat the boundary conditions on the fermion fields .
the fermion fields can be viewed as a complex scalar fermion , and its conjugate .
these fermions are partners under a supersymmetry ( graded lie algebra ) of free bosons , and the corresponding continuum free boson theory is a complex scalar boson .
the appropriate boundary conditions for the latter were considered in detail in ref .
@xcite , both for the lattice model ( discrete in both space and time directions , and called a network model ) and in the continuum .
hence the boundary condition on the fermions also is a variant of a neumann boundary condition , in which on the boundary , the derivative of the field at some fixed angle ( the `` hall angle '' @xcite ) to the normal to the boundary must vanish ( the hall angle is a free parameter of the theory ) .
this is due to the alternation of the two types of site , in the fundamental and its dual , along the chain when viewed as a representation of gl(@xmath12 ) .
a property of these boundary conditions is that , as for ordinary neumann boundary conditions , when we find the normal modes of the continuum theory on a finite interval , there is a zero mode for the fermion .
indeed , the entire spectrum is independent of the choice of hall angle . for any value of the hall angle
, the theory can be mapped to the hall angle zero case , which is the ordinary neumann boundary condition .
the transformation is a dual rotation , mapping the field to a linear combination of itself and its dual field ; this transformation is nonlocal in terms of the fields , but linear , and commutes with the u(1 ) symmetry of the complex fermion theory . after this transformation ,
the fields of the theory are the fermion fields @xmath237 with two real components @xmath238 , @xmath239 , where @xmath240 , and their derivatives with respect to @xmath241 only ( as they are boundary fields ) , and products of these .
the lagrangian density is @xmath242 , and thanks to the neumann boundary condition , there is a well - known exact symplectic [ sp(2,*r * ) ] symmetry of this theory , that acts locally on the fields .
note however that for the original field variables that were natural for the spin chain , only the u(1 ) subgroup of this symmetry acts locally ( it corresponds to @xmath243 or fermion number symmetry in the chain ) , while the other two generators of the sp@xmath29 lie algebra are nonlocal in those variables and arise as the `` renormalized squares '' of @xmath244 and @xmath245 in the 2 language , or as some of the nonlocal @xmath246 basis elements constructed in ref.@xcite .
now we classify these local operators according to the virasoro [ and @xmath25sl@xmath26 ] module structure we have worked out .
the arrows connecting simple modules in the indecomposables of either algebra are always due to the fermion zero modes @xcite , which correspond to the local fields @xmath248 without derivatives , as in operator products @xmath249 , while @xmath248 without a derivative can not be produced by any operator products of the local fields if it is not already a factor in an operator .
the case @xmath250 is typical .
then @xmath251 , the vector @xmath225 can be identified e.g. as the derivative of either symplectic fermion , say @xmath252 .
the vector @xmath253 has dimension @xmath254 and is a virasoro highest weight : it generates a reducible representation , in which @xmath255 appears as a singular vector .
we still need a field @xmath256 which is a combination involving @xmath257 etc .
it is not primary , nor is @xmath258 . by interchange @xmath259 ,
two such virasoro modules appear , which corresponds to the dimension of the 2 simple module , which is just 2 . for the case
@xmath144 , the cell lacks its left hand side and looks as on figure [ partindec ] .
one has for instance @xmath260 while @xmath261 , etc . for the 2 structure
, we may notice that general local fields that are highest weights for virasoro , modulo submodules , can be written as one of the following form : ( i ) @xmath262 , with @xmath144 , @xmath145 , , where the symmetrizer @xmath263 symmetrizes in the indices @xmath264 , @xmath145 ; ( ii ) the same with @xmath265 inserted inside the symmetrizer @xmath263 ; ( iii ) a similar expressions as in ( ii ) but with the index on @xmath265 contracted with one of the @xmath266 using the symplectic form ; ( iv ) the form ( i ) times @xmath267 . for each @xmath59 , these all have the same scaling dimension ( meaning the eigenvalue of @xmath24 , modulo submodules ) .
those of form ( i ) are the foot of the virasoro module @xmath268 , that is they form a simple module of scaling dimension @xmath269 ; this may be seen using operator products with the stress tensor @xmath270 .
there are @xmath59 distinct such operators , because of the symmetrization , and so they form a simple 2 module , with the 2 spin @xmath271 .
the local fields of form ( ii ) in which @xmath265 are included form a multiplet of dimension @xmath163 , those of type ( iii ) form a multiplet of dimension @xmath271 , and finally those of form ( iv ) in which @xmath267 is included form a multiplet of dimension @xmath59 . for each @xmath59 , these four types of local operators ( or three types if @xmath144 ) make up the tilting module @xmath34 of @xmath25sl@xmath26 , listed here starting with the foot and ending with the head .
the symplectic or sp@xmath29 symmetry of symplectic fermions acts to map the fields of the same type into each other ; if we view the boundary symplectic fermion theory as a chiral theory in the plane , then there are three corresponding noether currents @xmath272 .
the doublet of currents @xmath273 of psl(@xmath12 ) @xcite act ( through operator product ) to remove an undifferentiated @xmath248 from the boundary local fields .
these are the arrows in the 2 quartets .
the 2 algebra can here be viewed as isomorphic to a lie superalgebra generated by the line integrals of these five local currents ( which also correspond to the five generating elements of 2 at a root of unity ) ; this lie superalgebra is not semisimple ( as a lie superalgebra ) , because the psl(@xmath12 ) generators form an ideal . for the virasoro structure
, it suffices to notice that the stress tensor @xmath18 also maps each of the four types into themselves by operator products , and also can reduce the number of undifferentiated fermion fields in a local operator by one or two .
these actions then exhibit the commuting actions of virasoro and @xmath25sl@xmath26 , and show how the staircase diagram structure appears in terms of the local fields .
the partition function can also be interpreted in terms of symplectic fermions . in terms of the characters of simple virasoro modules and dimensions of simple modules of @xmath25sl@xmath26
, it becomes @xmath274 where @xmath275 is the laplacian on the strip with neumann boundary conditions in the space direction and antiperiodic boundary conditions in the time direction . for symplectic fermions
, the vanishing of the `` unmodifed '' ( or quantum trace ) partition function can be viewed as due to the fermion zero modes , as is well known ( in the present language , it is the determinant of the laplacian with periodic boundary conditions in the time direction , and this laplacian has a zero mode ) .
we now discuss fusion of the tilting modules @xmath34 ( of dimension @xmath141 ) labelled by @xmath59 , in other words the clebsch - gordan series for 2 at @xmath30 .
a tensor product of indecomposable tilting modules necessarily decomposes into a direct sum of tilting modules , and given this fact , the decomposition can be easily determined by using characters @xcite .
it takes the form : @xmath276 here @xmath277 means @xmath278 , and for @xmath279 , the term @xmath280 is absent ; for the present we may view this exception as handled by our convention that @xmath281 if @xmath199 .
( a check on the result can be easily obtained by considering the dimensions @xmath282 of the tilting modules . )
it follows from morita equivalence that the same fusion rules apply to the tilting modules of our supersymmetry algebras @xmath183 also @xcite . from the general theory @xcite
, the fusion rules for the product _ functor _ for the indecomposable _ projective _ tl modules have the same form , @xmath283 and so do those for the projective virasoro modules in the continuum limit : @xmath284 ( again , recall that the modules @xmath285 and @xmath286 are zero . )
the summations may be re - organized into the form @xmath287 where the sum is over @xmath288 , @xmath289 , , @xmath290 .
this shows that these results agree with those for @xmath3 in ref .
@xcite . there ,
some indecomposables of virasoro are built somewhat explicitly , and fusion is defined and calculated up to level six through the algorithm in @xcite ; the final results are then conjectured .
the correspondence in the latter reference would be @xmath291 and @xmath292 .
calculations in @xcite are extremely technical , and it is quite remarkable that the results can be derived from fusion of indecomposables of @xmath293sl@xmath294 . we have concentrated here on the fusion rules for tilting modules on the 2 side , and the projective modules on the tl or virasoro side , as these are the most natural starting point , and there is an equivalence between these fusion products .
one can also derive from these the fusion rules for some other types of modules , such as the standard and simple modules .
one must be careful in using the 2 fusion results on the tl / virasoro side as the equivalence may not always hold for the non - tilting modules .
the general definition of fusion for the tl or virasoro modules is always the module induced from the tensor product by the inclusion of the algebras for two chains into that for a single chain , by joining them end to end .
we emphasize that these fusion rules can be viewed as describing operator products of the conformal fields that transform as virasoro modules discussed here , for example for the symplectic fermions , for which the `` pseudoprimary '' fields were written down earlier , each of which is the primary field in a simple subquotient module .
they also apply to the corresponding fields in the supersymmetric models , where the fields have different multiplicities given by the dimensions @xmath185 . in applications the fusion rules given here will have to be combined with those for the symmetry algebra [ 2 or @xmath183 ) ] , as the fields transform under both algebras .
these combined forms can be obtained in the finite-@xmath33 lattice models , where they come from decomposing tensor products simply as vector spaces ( the multiplicities from the symmetry algebra side make the dimensions of spaces conserved ) , and then the continuum limit can be taken . as a final note
, we mention that in the case of unoriented loops , we should consider odd as well as even values of @xmath130 .
then the half integer spins @xmath59 , for which @xmath295 is odd also appear in the spectrum . for @xmath30 ,
these standard modules remain simple , both for 2 and ( hence ) also for tl .
they appear then as extra circles interspersed among the integer @xmath59 values in the staircase diagrams , with no extra explicit arrows , so these modules are both tilting and projective also .
fusion rules for products involving these can be found along the same lines as for @xmath296 , @xmath297 integer ( the fusion rules for a product of @xmath296 with @xmath297 both integers are unchanged , and still close on integer @xmath59 only ) . in terms of dense polymers ,
the unoriented - loops model contains local boundary fields for odd as well as even numbers of legs , the number of legs being ( as always for the loop models ) @xmath295 for fields in a standard module of spin @xmath59 .
in this section we follow a similar procedure as in the previous one , for the xxz chain of even length @xmath130 with @xmath298 , so @xmath31 . in this case
the loop model describes dense , oriented loops which can be identified with the boundaries of clusters in percolation at the critical percolation threshold . in the spin chains with gl(@xmath2 ) supersymmetry @xcite ,
the continuum limit in this case corresponds to the fixed - point theory of the @xmath299 nonlinear sigma model at @xmath23 , with the enlarged symmetry algebra @xmath300 that is morita equivalent to 2 with @xmath31 @xcite .
the @xmath133 case is thus connected with the spin quantum hall transition also @xcite . because the tl algebra for @xmath301 is isomorphic to that for @xmath113 , the first , algebraic part of
the analysis is the same for the @xmath302 finite-@xmath33 spin chain for that case also , up to a change in sign of the hamiltonian . because of the latter sign change
, a distinct continuum limit is obtained for @xmath302 which is not considered in this section .
we note that since the chain has even length , only integer spins @xmath59 occur , and for these the symmetry algebras 2 and @xmath303sl@xmath26 are isomorphic .
we start with the analysis of the spin-1/2 chain under @xmath293sl@xmath294 . for @xmath31 , so @xmath304 , and the representations depend on @xmath124 modulo @xmath146 , instead of modulo @xmath145 as in the previous case . the standard modules for @xmath207 with @xmath305 ( mod 3 ) remain irreducible .
the remaining standard modules for @xmath152 become reducible in the manner described for all roots of unity , and the corresponding tilting modules ( which are the direct summands in the spin-1/2 chain ) contain a submodule and a quotient module that are standard , and which have adjacent @xmath59-values if the values @xmath305 ( mod 3 ) are skipped . similarly , the @xmath59 values assigned to the simple subquotient modules of the standard modules are also adjacent if the values @xmath306 ( mod 3 ) are skipped , and the value at the head is the same as the value for the standard module .
the standard module for @xmath52 is one dimensional , and so is simple , and also appears as a summand in the chain .
thus the tilting modules have the following structure @xcite : t_j : \ { cl ccccc & & -.7cmj-1 & & + & -.2 cm & & + j-3&&&-.3cmj + & -.2 cm & & + & & -.7cmj-1 & & & + & + j & + & + ccccc & & -.7cmj-2 & & + & -.2 cm & & + j-3&&&-.3 cm j + & -.2 cm & & + & & -.7 cm j-2 & & & . for @xmath52 , @xmath48 , , @xmath33 . here again , simple subquotient modules labeled @xmath307 are understood to be zero and can be omitted . in the reducible modules , the pairs connected by southwest arrows
form standard modules . given this structure and the dimensions @xmath124 for the standard modules ,
the dimensions of the simple modules can be found recursively . introducing @xmath308 ( where @xmath309 is the largest integer @xmath310 ) ,
the dimensions of the simple modules are @xmath311 , @xmath312 , @xmath313 for @xmath314 , @xmath48 , @xmath145 , ( mod 3 ) respectively ( @xmath315 ) .
thus in terms of dimensions of the simple subquotients , we have for the two non - trivial classes : @xmath316 the @xmath27-dimensions , and the dimensions and superdimensions of the standard and tilting modules over @xmath300 , can be calculated similarly , as we showed in the case of @xmath129 .
in particular , the @xmath27-dimensions ( superdimensions ) @xmath210 of all the tilting modules over 2 ( resp . , @xmath300 ) are zero except for the @xmath52 tilting module , which has @xmath317 . taking @xmath143 as an example , the decomposition of the hilbert space of the chain under 2 is , in terms of spins @xmath59 , @xmath318 for a total of 64 states indeed .
the staircase diagram is shown in fig .
[ nesti ] . for the tl algebra , the direct summands , with their multiplicities ,
are @xcite @xmath319 in order of increasing @xmath59 .
the dimensions of the simple subquotients are the multiplicities of the 2 tilting modules , and the multiplicities are the dimensions of the 2 simple modules . but notice that dimension @xmath48 appears in connection with two different summands under @xmath25sl@xmath26 , and these are not expected to be isomorphic as tl modules
they have different @xmath59 values .
for this reason , one of them has been distinguished here by writing @xmath320 .
all of these indecomposable modules are projective , except for the tl singlet at the end . however , there must also be an additional projective module , which has the simple module @xmath320 at the top , and must `` cover '' the lower part of the tl tilting module in which it occurs .
this remaining indecomposable projective module thus has the form ccc ( 1) & & + & -.3 cm & + & & -.3cm(4 ) ( and is a standard module ) , for a total of @xmath321 isomorphism classes as expected here , since they must correspond to the four distinct indecomposable tilting modules of @xmath25sl@xmath322 . for general @xmath33 ,
the pattern is similar .
there is always a singlet standard module with spin @xmath161 which is a tl tilting module . for @xmath323 ( mod 3 )
, it is not projective . for @xmath324 ( mod 3 )
, this singlet is projective as well as tilting .
but then there is always a simple summand ( tilting module ) at the largest @xmath325 ( mod 3 ) , which is not a tl projective module .
the remaining indecomposable tilting modules [ including those standard modules with @xmath306 ( mod3 ) which are simple ] are also projective , and there is one additional projective module for @xmath52 , that has a ( different ) singlet at the top .
thus the projective modules for tl are @xcite , in terms of spin labels , p_j : \ { cl ccc 0 & & + & -.2 cm & + & & -.3 cm 2 & + ccccc & & -.7 cm j & & + & -.2 cm & & + j-1&&&-.3 cm j+2 + & -.2 cm & & + & & -.7cmj & & & + & + j & + & + ccccc & & -.7 cm j & & + & -.2 cm & & + j-2&&&-.3 cm j+1 + & -.2 cm & & + & & -.7 cm j & & & . where again simple modules with @xmath326 or @xmath200 are understood to be zero .
the standard sub- and quotient modules are the pairs connected by a southeast arrow .
once this structure is known , the dimensions @xmath169 of the simple modules can be found recursively . for @xmath314 ,
the standard module is simple , and has dimension @xmath327 as before . for the other cases
, one has @xmath328 [ @xmath314 ( mod 3 ) ] and @xmath329 [ @xmath330 ( mod 3 ) ] .
these can be solved starting from @xmath331 .
it is possible to show that @xmath332 for all @xmath33 ( as already stated above ) , and then the recursion can also be solved starting from the @xmath52 end . for @xmath31 , we have not found any very illuminating closed - form expressions for the dimensions @xmath169 for @xmath152 . in the large @xmath33 limit
, the staircase diagram continues to infinity , and resembles that for @xmath30 , except that extra simple modules appear at @xmath305 ( mod 3 ) , and there is now a node at @xmath222 ; see fig .
[ rep8 ] .
let us now consider the continuum limit in more detail .
for @xmath4 , the kac formula reads @xmath333 the conformal weights appearing of the primary fields appearing as subquotients in the lattice partition function are @xmath334 we again consider only the case @xmath59 integer , and often abbreviate @xmath335 . for @xmath336 ( mod 3 ) ,
these weights are integers .
introducing the same notation @xmath337 for the virasoro characters of the standard modules , which we again denote @xmath338 , we have @xmath339 and from the limit of the spin chain we expect to have the relations ( using again @xmath340 ) @xmath341 where @xmath342 are the characters of the associated irreducible virasoro representation with highest weight @xmath93 ( denoted here by @xmath38 ) . as for @xmath129 ,
this expectation is correct as it agrees with the known structure of virasoro representations in the kac table for @xmath4 @xcite ; the standard modules @xmath343 contain a unique simple submodule when @xmath325 ( mod 3 ) .
the irreducible characters are known , and take different forms depending on @xmath59 ( mod 3 ) @xcite . if @xmath344 , we have @xmath345 if @xmath346 , we have @xmath347 } \frac{\widehat{q}^{(12n+12p+7)^{2}/24 } -\widehat{q}^{(12n+12p+11)^{2}/24 } } { p(\widehat{q})},\ ] ] and if @xmath348 @xmath349 } \frac { \widehat{q}^{(12n+12p-1)^{2}/24 } -\widehat{q}^{(12n+12p+7)^{2}/24 } } { p(\widehat{q})}.\ ] ] in particular , the irreducible character of the identity is @xmath350 . for @xmath344 ,
the standard representation @xmath351 is irreducible , and we can view it as our projective module . for @xmath352 ( mod 3 ) we expect that there are indecomposable projective modules @xmath39 that are the limits of the tl modules , and correspond to them in submodule structure , thus r_j : \ { cl ccc r_0 & & + & -.2 cm & + & & -.3 cm r_2 & + ccccc & & -.7 cm r_j & & + & -.2 cm & & + r_j-1&&&-.3 cm r_j+2 + & -.2 cm & & + & & -.7cmr_j & & & + & + r_j & + & + ccccc & & -.7 cm r_j & & + & -.2 cm & & + r_j-2&&&-.3 cm r_j+1 + & -.2 cm & & + & & -.7 cm r_j & & & . for @xmath52 , @xmath48 , . apart from the the @xmath52 case , and the intruding @xmath305 ( mod 3 ) simple modules , these resemble the projective modules of the @xmath3 case .
the characters of the non - trivial ( non - standard ) cases are clearly , for @xmath348 , @xmath353 and for @xmath346 , similarly , @xmath354 the partition function of the continuum limit of the spin-1/2 chain is again given byz=_j=0^(2j+1)k_j= _ j=0^(2j+1).for the @xmath27-trace instead of the trace , one replaces @xmath124 by @xmath125_q$ ] , and the resulting formula also gives the `` unmodified '' partition function of the supersymmetric models with symmetry algebra @xmath300 @xcite .
for the partition function of the latter , one instead replaces @xmath124 by the dimensions @xmath181_{q'}$ ] , where @xmath355 for these @xmath298 models ( see appendix b in ref .
@xcite ) . understanding the representations @xmath39
more explicitly is more difficult than for @xmath129 , since we do not have an explicitly solvable version analogous to the symplectic fermions .
one might think of using the feigin - fuchs free field representation of the theory properly extended by some extra space to allow for the presence of jordan cells , along the lines of @xcite .
we have not done so however , but suspect that the continuum limits of the present models are different anyhow , since they involve a single copy of the @xmath356 field .
this will be discussed in more detail when we turn to the case of the torus in a later paper .
we can now study fusion of the 2 tilting modules @xmath134 .
the decomposition of a tensor product of @xmath357 with @xmath358 has to be considered in six cases , as it depends on @xmath296 and @xmath297 ( mod 3 ) .
we let @xmath359 , @xmath360 . for cases
in which @xmath296 , @xmath361 ( mod @xmath146 ) , we have : @xmath362 where the sum runs over @xmath363 , @xmath364 ,
, @xmath365 .
here and in the following , underlined terms are absent if @xmath366 .
similarly @xmath367 where the sum runs over @xmath363 , @xmath364 , ,
@xmath368 . finally , for the mixed fusion @xmath369 where the sum runs over @xmath370 , @xmath371 , , @xmath365 if @xmath372 and @xmath373 , @xmath374 , , @xmath365 if @xmath375 .
for the cases in which ( without loss of generality ) @xmath376 ( mod 3 ) , we have @xmath377 where the sum runs over @xmath363 , @xmath364 , , @xmath365 .
moreover @xmath378 where the sum runs over @xmath363 , @xmath364 , , @xmath365 , and @xmath379 where the sum runs over @xmath370 , @xmath371 , , @xmath365 if @xmath372 and @xmath373 , @xmath374 , , @xmath365 if @xmath375 .
finally , we note the special case @xmath380 , when t_0_j = t_jfor all @xmath207 .
the fusion rules for the virasoro projective modules follow immediately from the correspondence @xmath381 ( and also those for tl projective modules , from the correspondence @xmath382 ) . as far as we know , these fusion rules for a logarithmic theory at @xmath4 are new .
this is one of the central results of this paper .
we note that the @xmath52 projective module @xmath286 does indeed behave as the identity in the fusion rules , as expected from the relation with @xmath25sl@xmath26 , because it contains the identity and the stress tensor , which map any of the projective modules @xmath383 for @xmath152 into themselves .
none of the fusion rules produce the module @xmath138 ( on the symmetry side ) or @xmath286 ( on the virasoro side ) .
the identity field , and the stress tensor , also occur within the @xmath384 ( resp . , @xmath385 )
modules , so that conformally - invariant correlation functions do exist .
a lattice ( vertex ) model for dilute loops can be formulated on the square lattice with a two - colour tl algebra @xcite , or a spin-1 chain with the states @xmath386 viewed as occupied by parts of loops , and @xmath387 as empty .
it does have 2 symmetry , with the strands transforming as spin-1/2 . from a symmetry point of view , the models are identical to the dense cases . this is also true within the supersymmetric formulation @xcite , in which the obvious supersymmetry algebra is ( the enveloping algebra of ) osp(@xmath388 ) , and acts on the occupied sites as the vector ( defining ) representation , and trivially on the empty sites . as the models forbid the strands to cross , the global symmetry algebra is actually larger than osp(@xmath388 ) @xcite .
we will consider here the unoriented version of dilute loops with @xmath129 , for variety and because this is of natural interest for polymers .
this means the number of strands on any time - slice of the system is unrestricted . because of the existence of empty sites , both parities of @xmath295 occur for each length of system . for @xmath129 ,
this model possesses both a high - temperature phase and a low - temperature phase , separated by a transition .
the continuum limit in the low - temperature phase is in the same universality class as the ( unoriented loops ) @xmath129 dense polymer theory above .
the critical point describes dilute polymers .
it is the same universality class as in an osp(@xmath389)-invariant @xmath390 ( landau - ginzburg ) formulation of dilute polymers , or as in an osp(@xmath389)-invariant nonlinear sigma model with target space a supersphere , @xmath391 @xcite .
we will not define in detail here the lattice model , but we make some comments . for all @xmath27 ,
the algebra generated by the nearest - neighbor interactions in the spin-1 chain that represents dilute loops is clearly somehow similar to the tl algebra , even though the dimensions of the algebras are different .
we suspect that this relation is in fact morita equivalence ( with a partial exception that will be mentioned below ) : the algebra generated by the spin-1 chain is morita equivalent to the tl algebra ( generated by the nearest - neighbor terms in the spin-1/2 chain ) for the same number of sites , and at the same @xmath27 .
this is motivated by the common symmetry structure for all @xmath27 , and by the relation of this with the tl ( or virasoro ) structure .
it remains a conjecture as we have not established that the spin-1 model and 2 form a `` dual pair '' .
the exception occurs when @xmath129 , since as we will see the singlet module at @xmath222 is present in the spin-1 case , but was absent for the spin-1/2 model .
this is connected with a well - known phenomenon in these loop models : the dilute model possesses a low-@xmath18 phase that is in the same universality class as the dense model . yet the dilute model has non - vanishing partition function ( here it is the `` unmodified '' partition function , defined using the @xmath27-trace in the transfer matrix approach , that we mean ) , but that of the dense model vanishes for @xmath129 .
the discrepancy is resolved because the state with no polymer loops exists in the dilute model , and produces the nonzero partition function , but there is no such configuration in the dense model . in the dense phase of the dilute model
, this state occurs at high energy and is dropped in the continuum limit , so the partition function ( in which energies of the spin chain are measured from the ground state energy ) becomes zero .
this configuration with no loops corresponds to the state of the chain at @xmath222 in the staircase diagram for @xmath129 . as we do not consider the lattice algebraic formulation in detail here
, we pass directly to the continuum theory at the critical point for @xmath129 .
the staircase diagram is shown in fig.[polyrep ] .
it has the same form as the dense polymer case , except that the circle at @xmath222 is now included .
the 2 structure is the same as for dense polymers , and we include the @xmath295 odd cases for the unoriented case
. here the singlet tilting module @xmath138 does appear with nonzero multiplicity in the chain .
for the supersymmetric models , the symmetry algebra for the unoriented loops case is called @xmath392 in ref.@xcite . on the virasoro side ,
the conformal weights are taken from the first _ row _ of the @xmath4 kac table , so @xmath393 the characters of the standard modules read @xmath394 using standard character formulas we check the identities @xmath395 similar to the @xmath129 dense case .
the partition function of the continuum limit of the spin one chain is again given by z= _ j=0^(2j+1 ) .
for the @xmath27-trace instead of the trace , one replaces @xmath124 by @xmath189_q$ ] with @xmath30 ( thus @xmath190 for @xmath59 integer , @xmath396 for @xmath59 half - integer ) , which also gives the unmodified partition function of the supersymmetric models with symmetry algebra @xmath392 @xcite .
the latter is simply @xmath397 by the euler identity .
for the `` modified '' partition function of the supersymmetric models , one instead replaces @xmath124 by the dimensions @xmath181_{q'}$ ] , where @xmath182 for these models ( see appendix b in ref.@xcite ) . for @xmath133 ,
the supersymmetry algebra is isomorphic to @xmath25sl@xmath26 , as in the dense case .
projective modules for virasoro are also similar to the dense case in structure , but now there is also @xmath286 for @xmath52 , which is just a standard module , similarly to the @xmath298 ( percolation ) case .
it has the identity operator at its head , and the stress tensor @xmath18 as the highest weight in the foot submodule .
these fields are also part of the @xmath398 module , similarly to percolation ( @xmath298 ) .
fusion rules for @xmath152 are as for dense polymers . in the fusion of tilting modules of 2 ( or projective modules of virasoro ) for @xmath296 with @xmath399 ( and @xmath400 ) , the term @xmath138 ( resp . , @xmath286 )
is again absent , even though the corresponding modules are nonzero ; this applies for @xmath401 odd as well as even .
@xmath138 behaves as the identity for the fusion rules : t_0_j = t_jfor all @xmath207 , and similarly for fusion of @xmath286 with @xmath268 .
we note that the identity field occurs within the @xmath398 module ( unlike the dense case ) , and so conformally - covariant correlation functions can be produced .
in this section , some physical discussion is given , by writing the algebraic statements in terms of operator products .
the discussion is mainly for @xmath4 , and makes contact with that of gurarie and ludwig @xcite .
we also make some remarks on the role of fields of weight 1 in the boundary theories .
the case @xmath4 has probably been the most discussed in the literature on logarithmic cfts .
there , the existence of the `` @xmath32-catastrophe '' ( i.e. , poles in @xmath402 in opes as @xmath403 ) led gurarie @xcite to the introduction of a _ logarithmic partner _
@xmath19 for the stress energy tensor , satisfying the opes ( up to non - divergent terms ) : @xmath404 notice that @xmath18 is primary , but has a vanishing two - point function , while @xmath19 is not primary .
the @xmath405 ope itself is more complicated and starts with a logarithm : @xmath406 it follows from these equations that @xmath407 form a size two jordan cell for @xmath13 : @xmath408 gurarie went on to suggest that @xmath18 , @xmath19 were part of a `` susy '' multiplet containing at least two weight - two primary operators with non - vanishing two - point functions @xcite
. the foregoing algebraic analysis of our @xmath4 lattice models gives rise immediately to the existence of @xmath19 .
indeed , for the continuum limit of the chain , the virasoro tilting module that contains the identity and the stress tensor is @xmath409 each simple subquotient module @xmath410 that appears is a highest weight module , so ( by the correspondence of states and operators ) there is a corresponding pseudoprimary field .
all these states have multiplicity 1 .
thus the pseudoprimary fields consist of the identity with conformal weight @xmath411 , two fields with weight @xmath412 , and one with dimension @xmath413 .
one of those with dimension 2 ( the bottom field , as it is primary ) is naturally identified with the stress energy tensor @xmath18 , and should form a jordan cell with its logarithmic partner - here the top field ( with conformal weight 2 ) which we identify with @xmath19 .
the whole indecomposable representation is generated by acting on @xmath19 with the virasoro algebra .
we now see that there is a fourth member of the multiplet of fields , which we call @xmath414 , and which has conformal weight @xmath413 .
we recall that this module can be analyzed into a submodule that is standard , and which is generated by the descendants of @xmath48 , and a quotient module that is also standard , and the quotient module is generated by @xmath19 .
the arrow from @xmath415 to @xmath416 corresponds to the fact that @xmath19 is not primary , and can be mapped to the identity under the action of virasoro generators - this is directly related with the existence of the number @xmath21 in the second ope ( [ ttope ] ) .
the arrow from @xmath415 to @xmath417 indicates that the second ope in ( [ ttope ] ) must be extended to include the field of dimension @xmath418 : t(z)t(0)~+++ + z c_tt^(0 ) where @xmath419 is some coefficient .
@xmath414 is not primary , but only maps to descendents of @xmath18 under action of virasoro with positive modes : t(z)(0)~+ .modulo the virasoro submodule generated by @xmath18 in this way , the field @xmath414 is simply the null virasoro descendent of @xmath19 that occurs in the virasoro standard module when @xmath4 ; it must not be set to zero .
the possible supersymmetry multiplet of fields of weight 2 @xcite may be compared with the 2 multiplet structure ( horizontal in our diagrams ) . for the weight 2 fields for the @xmath298 theory ,
this is the @xmath384 tilting module ( or its morita equivalent in the supersymmetric theories @xcite ) , which has the trio form ccc 0 & + & -.2 cm + & & -.3cm2 + & -.2 cm + 0 & in terms of the spin labels which are @xmath52 , @xmath145 ( for the @xmath298 theory ) .
this is in agreement with their expectations , since our node @xmath250 is a multiplet , not a single field .
what plays the role of supersymmetry in our case is @xmath293sl@xmath294 or the @xmath420 symmetry algebra @xcite , which is represented acting horizontally on fig .
the dimension of the space of fields of conformal weight two in our supersymmetric models turns out to be : @xmath421 ( for the limit of the spin-1/2 chain , it is @xmath422 . ) the basic doublet @xmath19 , @xmath18 is thus completed by a large number ( the preceding number minus 2 , called @xmath423 ) of other fields with @xmath20 , which themselves lie in the @xmath424 piece of another virasoro indecomposable of structure @xmath425 the whole picture is thus considerably more involved than in ref.@xcite .
a similar discussion applies to the dilute polymer case .
the virasoro tilting module containing the identity and the stress tensor is then @xmath426 and the arrow from @xmath427 to @xmath415 indicates that in this case the ope ( [ ttope ] ) must be extended to include a field of dimension 7 , similar to @xmath414 ; it is a virasoro descendant that is a null vector , but must not be set to zero .
we have for the supersymmetric models @xmath428 for the 2 theory , this number is @xmath429 , so that it indeed turns out to be simply a quartet of fields , and as @xmath30 these can be viewed as a supersymmetry multiplet also , by putting @xmath133 .
it is interesting to observe that reducible but indecomposable modules of virasoro of the forms we find are in general characterized by dimensionless parameters measuring the `` mixing '' of representations , and that the number @xmath21 of gurarie @xcite appears simply as a particular case of what might be an infinite set of parameters .
clearly , a more detailed analysis of indecomposables at @xmath4 along the lines of @xcite would be useful here .
but it is also important to understand how features of the lattice models determine the values of these parameters in our cases .
gurarie and ludwig @xcite also advocated viewing the operator product algebra generated by @xmath19 and @xmath18 as a kind of chiral algebra .
but we now see that we must introduce @xmath414 as a further field , and the fusion rules for this multiplet with itself produce fields in higher-@xmath59 projective modules .
that is , they generate by repeated ope s not only the full set of integer - weight fields , which would then need to be included in the chiral algebra ( and which are not all singlets under the symmetry algebra ) , but also for @xmath298 the non - integer weight fields . in the oriented @xmath129 and @xmath298 theories ,
this space of fields corresponds to the full space of states !
this argument ignored the role of the symmetry algebra in the actual fields , which after all carry representations of the symmetry algebra also .
this might restrict the fields actually produced in iterated ope s .
however , it seems unfair to omit the fields related to @xmath19 and @xmath18 by symmetry from such a chiral algebra .
the fields related by symmetry include some in the @xmath430 virasoro projective module , and fusion of these produces higher @xmath59 s .
thus it seems inescapable that all integer @xmath59 fields , including those with fractional weights , must be in such a chiral algebra . in cases such as the dilute unoriented polymers
, there are also fields with @xmath295 odd , and all of these together should probably be viewed as making up a single representation of this large chiral algebra .
( thus , the structure of the modules over this larger algebra resembles that of the level - one su(2 ) affine lie algebra theory at @xmath431 , of which the 2 theories are a deformation . ) to us , it does not seem likely to be helpful to try to view this as a chiral algebra .
instead , the maximal chiral algebra ( in the strict sense of containing only integer - weight fields ) of these boundary theories seems to be the virasoro algebra . on the other hand
, the symmetry analysis seems powerful .
finally , we turn to another issue : we point out that neither of the boundary cfts with @xmath4 possesses any fields with @xmath432 .
the bulk theories in the supersymmetric formulation do possess such fields of weights @xmath433 and @xmath434 @xcite , and they transform in the adjoint representation of the algebras gl(@xmath435 ) and osp(@xmath389 ) , respectively in the two examples .
the existence of these fields is required by noether s theorem .
( despite the conformal weights , these currents are not generally ( anti-)holomorphic , implying that logarithms will appear in their correlators , and they do not generate affine lie algebras @xcite ) . in the boundary theory , on the other hand , while there are global symmetries , there is in general no such thing as a divergenceless boundary current , because charge can escape into the bulk , and noether s theorem only implies the existence of bulk currents .
thus in general there need not be any corresponding weight-1 fields in the boundary cft .
on the other hand , a more fruitful point of view about such fields is that weight one fields on the boundary correspond to marginal perturbations of the boundary theory , while boundary fields of weight less than one are relevant perturbations on the boundary . in particular ,
exactly marginal perturbations of the boundary theory correspond to the possibility of a continuous change of conformally - invariant boundary conditions . for the @xmath4 theories studied here
, we learn that there are no marginal boundary perturbations , and the only relevant perturbations transform in representations of the symmetry algebra that do not contain any singlets .
thus such a perturbation would necessarily break the global symmetry . on the other hand ,
for the @xmath3 theories , there are a number of weight 1 fields , transforming in a @xmath384 tilting module for @xmath25sl@xmath26 , or a corresponding tilting module for @xmath436 .
for symplectic fermions , there are 8 such fields .
one of the fields in the @xmath250 subquotient is the boundary term that corresponds to the bulk `` topological '' term in the sigma model , and corresponds to changing the hall angle in the boundary condition discussed in sec .
this gives a line of boundary conditions that preserve conformal invariance on the boundary .
however , as we noted above , these theories can be transformed onto one another by a duality rotation , so this perturbation is `` redundant '' .
the behavior under all the other perturbations is currently not known to us .
the analysis we have presented can be extended to the case of other values of @xmath27 a root of unity , in relation with loop models .
it can also be extended to other boundary conditions for open chains , which lead generically to standard modules with @xmath88 , @xmath118 using results from refs .
@xcite and @xcite ; see also ref . @xcite . in all cases ,
the patterns are quite similar , in particular the reducible indecomposable projective virasoro modules usually have the quartet form .
while this paper and its companion were being completed , an interesting preprint appeared on the archive @xcite which has some overlap with our analysis of the dense polymers case . in this reference ,
some of the fusion rules of the conformal field theory are conjectured based on a diagrammatic interpretation of fusion within the tl algebra .
our paper provides the correct algebraic framework to understand the observations of @xcite , together with the tools allowing immediate generalization to all roots of unity cases .
we also would like to mention the recent work @xcite where 2 at roots of unity is used to build indecomposable representations of chiral @xmath437-algebras .
as discussed in @xcite the theories we are considering are not rational ( though most likely quasi - rational ) and we do not think the results of @xcite apply to our case .
m. flohr , int . j. mod .
a * 18 * , 4497 ( 2003 ) ; m. gaberdiel , int . j. mod
a * 18 * , 4593 ( 2003 ) .
m. flohr and a. mueller - lohmann , j. stat . mech . * 0604 * , p002 ( 2006 ) .
gruzberg , a.w.w .
ludwig , and n. read , phys .
lett . * 82 * , 4524 ( 1999 ) .
s. guruswamy , a. leclair and a.w.w .
ludwig , nucl .
b * 583 * , 475 ( 2000 ) .
bhaseen , j .-
s.caux , i.i . kogan and a.m. tsvelik , nucl .
b * 618 * , 465 ( 2001 ) .
g. gotz , t. quella and v. schomerus , `` the wznw model on psu(@xmath438 ) '' , hep - th/0610070 .
knizhnik , comm .
112 * , 567 ( 1987 ) .
l. rozansky and h. saleur , nucl .
b * 376 * , 461 ( 1992 ) . v. gurarie , nucl .
b * 410 * , 535 ( 1993 ) .
gaberdiel and h.g .
kausch , phys .
b * 386 * , 131 ( 1996 ) ; nucl .
b * 538 * , 631 ( 1999 ) ; h.g .
kausch , nucl .
b * 583 * , 513 ( 2000 ) . v. schomerus and h. saleur , nucl .
b * 734 * , 221 ( 2006 ) .
flohr , int .
a * 11 * , 4147 ( 1996 ) ; _ ibid . _ a * 12 * , 1943 ( 1997 ) .
b. l. feigin , a.m. gainutdinov , a.m. semikhatov , and i.yu tipunin , commun .
phys . * 265 * , 47 ( 2006 ) .
j. fjelstad , j. fuchs , s. hwang , a.m. semikhatov and i. yu tipunin , nucl .
b * 633 * , 379 ( 2002 ) .
m. flohr and a. mller - lohmann , hep - th/0510096 .
v. gurarie , nucl .
b * 546 * , 765 ( 1999 ) .
v. gurarie and a. w. w. ludwig , j. phys .
a * 35 * , l377 ( 2002 ) .
v. gurarie and a. w. w. ludwig , hep - th/0409105 . j. cardy , `` the stress tensor in quenched random systems '' , in _ statistical field theories _ , proceedings of the nato advanced research workshop on statistical field theories , june 2001 , a. cappelli and g. mussardo , eds , nato science series , vol .
ii/73 ( kluwer academic , dordrecht , 2002 ) .
n. read and h. saleur , nucl .
b * 613 * , 409 ( 2001 ) .
g. moore and n. seiberg , commun . math
* 123 * , 177 ( 1989 ) .
n. read and h. saleur , `` enlarged symmetry algebras of spin chains , loop models , and @xmath439-matrices '' , cond - mat/0701xxx .
p. a. pearce , j. rasmussen and j.b .
zuber , `` logarithmic minimal models '' , hep - th/0607232 . p.p .
martin , _ potts models and related problems in statistical mechanics _ ( world scientific , singapore , 1991 ) .
p. p. martin , int . j. mod
. phys . a * 7 * , supp .
1b , 645 ( 1992 ) .
martin and d. mcanally , int .
phys . a * 7 * , supp .
1b , 675 ( 1992 ) .
jones , invent . math . * 72 * , 1 ( 1983 ) .
w.m . koo and h. saleur , nucl
b * 426 * , 459 ( 1994 ) .
g. feverati and p. pearce , nucl .
b * 663 * , 409 ( 2003 ) .
f. rohsiepe , `` on reducible but indecomposable representations of the virasoro algebra '' , hep - th/9611160 .
f.w . anderson and k.r .
fuller , _ rings and categories of modules _ , 2nd ed . , graduate texts in mathematics 13 ( springer - verlag , new york , ny , 1992 ) . for a review , see e.g. p. di francesco , p. mathieu , and d. senechal , _ conformal field theory _ ( springer , new york , 1997 ) .
belavin , a.m. polyakov and a.b .
zamolodchikov , nucl .
b * 241 * , 333 ( 1984 ) .
h. saleur and m. bauer , nucl .
b * 320 * , 591 ( 1989 ) .
j. cardy , j. stat .
phys . * 125 * , 1 ( 2006 ) .
green , j.h .
schwarz , and e. witten , _ superstring theory , volume 2 : loop amplitudes , anomalies , and phenomenology _ , ( cambridge university , cambridge , 1987 ) , p. 204
b. nienhuis , phys .
lett . * 49 * , 1062 ( 1982 ) .
p. di francesco , h. saleur , and j.b .
zuber , j. stat . phys .
* 49 * , 57 ( 1987 ) .
b. nienhuis , j. stat . phys . * 34 * , 731 ( 1984 ) .
v. pasquier and h. saleur , nucl .
b * 330 * , 523 ( 1990 ) .
see ref .
@xcite , lemma 29.4 , p. 324 .
s. xiong , n. read , and a.d .
stone , phys .
b * 56 * , 3982 ( 1997 ) .
kausch , `` curiosities at @xmath3 '' , hep - th/9510149 .
m. gaberdiel and h. g. kausch , `` indecomposable fusion products '' , hep - th/9604026 .
gaberdiel , int . j. mod
a * 9 * , 4619 ( 1994 ) .
p. di francesco , h. saleur and j.b .
zuber , nucl .
b * 285 * , 454 ( 1987 ) .
u. grimm and p. pearce , j.phys .
a * 26 * , 7435 ( 1993 ) .
p. martin and h. saleur , lett . in math
. phys . * 30 * , 189 ( 1994 ) .
a. nichols , `` the temperley - lieb algebra and its generalizations in the potts and xxz models '' , hep - th/0509069 .
p. pearce and j. rasmussen , `` solvable critical dense polymers '' , hep - th/0610273 .
b. feigin , a.m. gainutdinov , a.m. semikhatov and i. yu tipunin , `` kazhdan - lusztig dual quantum group for logarithmic extensions of virasoso minimal models '' , hep - th/0606506 . | we set up a strategy for studying large families of logarithmic conformal field theories by using the enlarged symmetries and non semi - simple associative algebras appearing in their lattice regularizations ( as discussed in a companion paper ) .
here we work out in detail two examples of theories derived as the continuum limit of xxz spin-@xmath0 chains , which are related to spin chains with supersymmetry algebras gl(@xmath1 ) and gl(@xmath2 ) , respectively , with open ( or free ) boundary conditions in all cases .
these theories can also be viewed as vertex models , or as loop models .
their continuum limits are boundary conformal field theories ( cfts ) with central charge @xmath3 and @xmath4 respectively , and in the loop interpretation they describe dense polymers and the boundaries of critical percolation clusters , respectively .
we also discuss the case of dilute ( critical ) polymers as another boundary cft with @xmath4 . within the supersymmetric formulations ,
these boundary cfts describe the fixed points of certain nonlinear sigma models that have a supercoset space as the target manifold , and of landau - ginzburg field theories .
the submodule structures of indecomposable representations of the virasoro algebra appearing in the boundary cft , representing local fields , are derived from the lattice .
a central result is the derivation of the fusion rules for these fields . #
1#2#3#4 # 1#2#3#4 spht 06/154 n. read @xmath5 and hubert saleur @xmath6 _ @xmath7 department of physics + yale university , p.o .
box 208120 , new haven , ct 06520 - 8120 usa + _ _ @xmath8 service de physique thorique , cea saclay , + gif sur yvette , 91191 , france + _ _ @xmath9 department of physics and astronomy , university of southern california + los angeles , ca 90089 , usa + _ |
the discovery of iron - based superconductivity @xcite with transition temperatures as high as 55 k @xcite has prompted efforts to understand both the electronic structure and magnetism of these materials , which are interrelated with superconductivity @xcite .
the first class of iron - based superconductors reported , the 1111 family , derive from @xmath4feaso ( @xmath4 = trivalent lanthanide ) .
they adopt the zrcusias structure @xcite , composed of layers of edge - sharing o@xmath5 tetrahedra alternating with layers of edge - sharing feas@xmath6 tetrahedra .
a second class , 122 materials , derive from @xmath7fe@xmath0as@xmath0 ( @xmath7 = ca , ba ) with the thcr@xmath0si@xmath0 structure @xcite , which again contains layers of edge - sharing feas@xmath6 tetrahedra .
the metallic parent phases in both classes undergo structural phase transitions from tetragonal to orthorhombic symmetry just above an antiferromagnetic ( afm ) ordering temperature ( @xmath8 = 137 k for laofeas @xcite and 172 k for cafe@xmath0as@xmath0 @xcite ) with small ordered moments on the fe sites in the @xmath9 plane .
superconductivity has also been observed in the binary iron chalcogenide systems : the properties of fe@xmath10te are very sensitive to the iron content @xcite and superconductivity can be induced by s or se doping @xcite ; @xmath11-fese does not order magnetically and undergoes a transition to a superconducting state at 8 k at ambient pressure @xcite , or 37 k at 7 gpa @xcite .
recently , attention has turned to the potassium - iron - selenide phase diagram , in particular , k@xmath12fe@xmath13se@xmath0 , which adopts a vacancy - ordered thcr@xmath0si@xmath0 structure .
this material is semiconducting @xcite and orders antiferromagnetically below 559 k. interestingly , the ordered fe@xmath14 moments are large ( 3.31 @xmath15 ) and are oriented perpendicular to the layer , in contrast to the 1111 and 122 materials @xcite . the magnetism of the iron sublattice in these materials has been the focus of much study in recent years .
initial studies on the 1111 and 122 materials suggested that the observed stripe magnetic ordering ( ferromagnetic stripes along [ 010 ] in the orthorhombic unit cell ) @xcite arises from the competing nearest - neighbor ( nn ) and next - nearest - neighbor ( nnn ) afm interactions @xcite .
subsequent work has highlighted the roles of other factors which lead to the complexity of the magnetic phase diagram for these materials @xcite . in @xmath4feaso and @xmath4mnaso materials ,
the @xmath16 ions have a significant role not only in tuning the superconducting transition temperature in the doped phases ( e.g. @xmath17 = 26 k for lafeaso@xmath2f@xmath18 @xcite , and 55 k for smfeaso@xmath2f@xmath18 @xcite ) , but also in influencing the magnetism in the undoped parent phases .
for example , the fe@xmath14 moments of cefeaso order antiferromagnetically in the @xmath9 plane at @xmath19 = 140 k @xcite while the ce@xmath20 moments couple strongly with the fe@xmath14 moments at relatively high temperatures @xcite , before developing a long range order below @xmath213.7 k with moments predominantly in the @xmath9 plane @xcite .
recent studies suggested some reorientation of the fe moments within the @xmath9 plane at the onset of the long range order of the ce moments @xcite .
the ce@xmath20 ions influence the iron magnetic sublattice , and can also induce exotic properties such as kondo screening of the local moment in closely - related cefepo @xcite and cerupo @xcite . the synthesis and crystal / magnetic structures of the iron oxyselenide ce@xmath0o@xmath0fese@xmath0 were reported in 2011 @xcite . it adopts a zrcusias - related structurein which the transition metal sites are half occupied by fe@xmath14 cations in a stripe ordered structure ( fig.[structure]@xmath22 ) . the magnetic structure of ce@xmath0o@xmath0fese@xmath0 ( fig.[structure]@xmath23 ) determined from neutron powder diffraction ( npd ) data reported rather surprising observations @xcite .
it undergoes an afm ordering below @xmath8 = 171 k in which the fe@xmath14 spins have a ferromagnetic ( fm ) order within each chain of edge - sharing fese@xmath6 tetrahedra despite the fe - se - fe angle ( 71.94@xmath24 ) deviating strongly from 90@xmath24 ( fig.[structure]@xmath23 ) , so one would have expected an afm ordering according to the goodenough - kanamori rule @xcite . in the present work we re -
examine the magnetic ordering in ce@xmath0o@xmath0fese@xmath0 to confirm these unusual observations on the basis of npd and inelastic neutron scattering ( ins ) experiments as well as density functional theory ( dft ) calculations .
the paper is divided into five sections including this introduction ; experimental and calculation descriptions ; experimental and theoretical results;and finally a discussion and conclusion .
( a ) orthorhombic nuclear unit cell and ( b ) monoclinic magnetic unit cell of ce@xmath0o@xmath0fese@xmath0 ( ce = green , fe = blue , o = red , and se = yellow spheres ) .
( c ) an isolated sheet of edge - sharing chains of fese@xmath6 tetrahedra present in ce@xmath0o@xmath0fese@xmath0 .
( d ) zoomed - in view of the magnetic ordering in the fe and ce sublattices of ce@xmath0o@xmath0fese@xmath0 . for convenience of discussion
, the directions of the orthorhombic unit cell are used to describe the magnetic structure in ( c ) and ( d ) ; the fm chains of edge - sharing fese@xmath6 tetrahedra lie in the @xmath9-plane with the fm chains running along the @xmath22-direction . ]
ce@xmath0o@xmath0fese@xmath0 was prepared as a black , polycrystalline sample ( 2.48 g ) as described previously @xcite .
preliminary characterisation was carried out using a bruker d8 x - ray diffractometer ( reflection mode , cu k@xmath25/k@xmath26 radiation , lynxeye si strip position sensitive detector , step size 0.02@xmath24 with variable slits ) equipped with an oxford cryosystems phenix cryostat .
npd data were collected on the high - flux d20 diffractometer at institut laue langevin ( grenoble , france ) with neutron wavelength 2.41 .
the sample was placed in a 6 mm cylindrical vanadium can ( to a height of @xmath214 cm ) and cooled to 2 k. data were collected over a 2@xmath27 range of 5 - 130@xmath24 at 2 k intervals on warming to 200 k. powder diffraction data were analyzed by the rietveld method @xcite using the topas academic software suite @xcite controlled by local routines .
the diffractometer zero point and neutron wavelength were initially refined using data collected at 12 k with lattice parameters fixed at values determined previously @xcite .
the zero point and wavelength were then fixed in all subsequent refinements .
typically , the background was refined for each data set as well as the unit cell parameters and a caglioti description of the peak shape .
structural characterization using data collected on the hrpd diffractometer at isis revealed no structural changes in this temperature range ( 4 218 k ) @xcite , so the atomic coordinates were fixed and this work focuses on the magnetic ordering .
the web - based isodistort software @xcite was used to obtain a magnetic symmetry mode description of the magnetic structure ; magnetic symmetry mode amplitudes were then refined to determine the magnetic structures .
the same polycrystalline sample was used for ins measurements .
the sample was packed into an al foil envelope and placed in an al can .
two experiments were performed using the mari direct geometry chopper instrument at isis .
the sample was cooled to 5 k in a closed - cycle cryostat .
the energy of the incident beam , @xmath28 , was selected using a gd fermi chopper spinning at 150 hz ( for @xmath28 = 40 mev ) or 400 hz ( for @xmath28 = 150 mev ) .
in addition , a @xmath29 chopper was used to block fast neutrons and a thick disk chopper ( spinning at 50 hz ) was used to improve background from neutrons above the gd absorption edge .
the cold triple - axis spectrometer spins at nist center for neutron research ( gaithersburg , usa ) was used to investigate the temperature dependence of the crystal fields . a pyrolytic graphite ( pg ) monochromator ( 004 reflection )
was used on the incident beam to give good resolution at high energy transfers and a pg(002 ) analyzer ( horizontally focused over 11@xmath24 ) was tuned to select a fixed final energy of e@xmath30 = 5.0 mev .
a be filter was used on the scattered side . in our dft electronic structure calculations for ce@xmath0o@xmath0fese@xmath0 , we employed the projected augmented - wave ( paw ) method encoded in the vienna ab initio simulation package @xcite , and the generalized gradient approximation ( gga ) of perdew , burke and ernzerhof @xcite for the exchange - correlation corrections , the plane wave cutoff energy of 500 ev , and the threshold of self - consistent - field ( scf ) energy convergence of 10@xmath31 ev .
we extract four spin exchange parameters by employing five ordered spin states defined on a ( @xmath22 , 2@xmath23 , @xmath32 ) supercell ( see below ) .
the irreducible brillouin zone was sampled with 4 @xmath33 2 @xmath33 1 @xmath34-points . to describe the electron correlation associated with the 3d states of fe and the 4f states of ce , the dft plus on - site repulsion u ( dft+u ) @xcite calculations were carried out with effective @xmath35 = @xmath36-@xmath37 ( see below ) .
in this section we outline the experimental and computational results of this paper .
we first discuss the neutron diffraction results probing the magnetic structure followed by a section discussing inelastic neutron results from which exchange constants between the fe ions and the ce ions are derived .
finally , these are compared with density functional calculations .
rietveld analysis of npd data collected at 250 k are consistent with the fe - ordered , orthorhombic crystal structure described above .
additional reflections observed below @xmath38 are consistent with the magnetic ordering ( and propagation vector @xmath39 = ( 0 @xmath40 @xmath40 ) ) reported previously @xcite and were indexed using an @xmath41 @xmath33 2@xmath42 @xmath33 2@xmath43 supercell
( where the subscript @xmath44 refers to the nuclear unit cell ) ; @xmath45 indices given subsequently for magnetic reflections refer to this magnetic unit cell .
the intensity of these reflections increases smoothly on cooling to @xmath21100 k. below this temperature , some reflections ( e.g. , ( 0 1 1 ) , ( 0 1 9 ) , ( 0 3 3 ) ) continue to increase in intensity , others ( e.g. , ( 0 1 5 ) , ( 0 1 7 ) , ( 2 1 1 ) ) decrease slightly ( fig.[peaks ] ) , while some additional very weak reflections ( e.g. ( 1 1 1 ) , ( 1 1 3 ) ) are observed below this temperature . temperature - dependence of the magnetic bragg reflections : ( a ) sum of the intensity of the 17 strongest magnetic reflections .
( b - e ) the intensity of the ( 0 1 1 ) , ( 0 1 3 ) , ( 0 1 5 ) and ( 1 1 3 ) reflections ( @xmath45 indices refer to the @xmath41 @xmath33 2@xmath42 @xmath33 2@xmath43 magnetic unit cell ) from the sequential refinements using a pawley phase to fit the magnetic reflections .
( f ) evolution of the fe@xmath14 ( blue , solid ) and ce@xmath20 ( green , open ) magnetic moments on cooling from rietveld refinements .
the solid black line shows fit to the function @xmath46 = @xmath47 for the fe data between 100 - 171 k with @xmath48 = 3.40(4 ) @xmath15 , @xmath38 = 175.5(8 ) k and @xmath49 = 0.28(1 ) . ]
the npd data collected below @xmath21170 k can be fitted by the nuclear structure and a magnetic phase composed of fm chains of edge - sharing fese@xmath6 tetrahedra , with afm coupling between adjacent fm chains ( fig.[structure]@xmath23 ) .
attempts to fit the data with models containing afm chains were not successful .
symmetry adapted ordering mode " approach @xcite was used here to describe the magnetically ordered structure .
mode inclusion analysis ( described elsewhere , @xcite ) was used to confirm that this arrangement of fe moments gives the best fit to the data and does not change below @xmath38 .
other models , including those with afm chains , gave significantly worse fits .
whilst this fm - chain model gives magnetic bragg peaks in the observed positions , the fit to the peak intensities was not perfect .
mode inclusion analyses were carried out at lower temperatures ( 80 k , 4 k ) and confirmed that the arrangement of fe@xmath14 moments does not change on cooling . given the large ordered moment on the fe sites in ce@xmath0o@xmath0fese@xmath0 , we would expect our refinements to be sensitive to slight reorientations of the fe@xmath14 moments , but there is no indication that reorientation of the fe@xmath14 moments occurs .
this is in contrast to the related prfeaso in which the fe moments ca nt slightly along @xmath32 at the onset of pr@xmath20 ordering @xcite . whilst the fm - chain model gives
a better fit than afm - chain models , further analysis indicated that including the ce magnetic ordering modes improves the fit significantly ( @xmath50 decreases from 5.49% to 4.37% at 80 k , and from 7.19% to 4.61% at 4 k for one additional parameter ) .
refinements are very sensitive to the relative signs of the fe@xmath14 and ce@xmath20 magnetic ordering mode amplitudes .
for example , as measured by @xmath50 , a surface plot showing fit for different amplitudes of the ce and fe magnetic ordering modes indicates that the best fit is obtained when both modes have the same sign , corresponding to a fm coupling between nn fe and ce sites ( see supplementary material ) .
refinement profiles and details are shown in fig.[refinement ] .
if canting of the ce moments is included in the model , the ce moments become oriented at @xmath2112@xmath24 to the @xmath9 plane ( i.e. , a @xmath51 component of 0.25(5 ) @xmath15 ) and @xmath50 is reduced by 0.04 % , but this improvement can not be regarded as significant from our data .
analysis using isodistort @xcite suggests that the magnetic structure of ce@xmath0o@xmath0fese@xmath0 can be described by the @xmath52-centered space group @xmath53 [ bns : 15.9 with basis ( 0 , -1 , 1 ) , ( -1 , 0 , 0 ) , ( 0 , 2 , 0 ) and origin at ( 0 , 0 , 0 ) ] shown in figure[structure]@xmath23 and refinement using 4 k data gives moments of 3.14(8 ) @xmath15 and 1.14(4 ) @xmath15 for fe and ce sites , respectively .
the ordered fe@xmath14 moment in ce@xmath0o@xmath0fese@xmath0 is comparable with that reported for the mott insulating oxyselenides ( e.g. , la@xmath0o@xmath0fe@xmath0ose@xmath0 ( 3.50(2 ) @xmath15 ) @xcite and the parent phase to superconducting k@xmath12fe@xmath13se@xmath0 ( 3.31 @xmath15 ) @xcite ) , and is consistent with a high - spin d@xmath54 configuration for fe@xmath14 sites .
it is significantly larger than that observed in @xmath4feaso materials with poor metallic behavior ( e.g. 0.94(3 ) @xmath15 for cefeaso at 1.7 k @xcite ) .
the sequential rietveld refinements using npd data collected on cooling show that the ce moment increases almost linearly at low temperatures .
the fe moment can be fitted well by the critical behavior with @xmath49 = 0.28(1 ) and @xmath8 = 175.8(8 ) k ( fig.[peaks]f ) for 100 k @xmath55 < 171 k. this fe@xmath14 moment ordering is similar to that observed for cefeaso at @xmath8 = 137 k , which can be described by critical behavior with @xmath49 = 0.24(1 ) @xcite .
these values for @xmath49 are larger than those reported for 2d - ising like systems ( including undoped bafe@xmath0as@xmath0 ( @xmath49=0.125 @xcite ) and la@xmath0o@xmath0fe@xmath0ose@xmath0 ( @xmath49=0.122 @xcite ) ) , but smaller than those predicted for three - dimensional critical fluctuations ( @xmath49=0.326 , 0.367 and 0.345 for 3d ising , 3d heisenberg and 3d xy systems , respectively ) @xcite .
the crossover between 2d and 3d universality classes has been suggested to originate from a coupling to an orbital degree of freedom @xcite or the prroximity of a lifshitz point ( see , for example , fe@xmath10te @xcite and bafe@xmath0as@xmath0 @xcite ) . the unusual change in the relative intensities of the different magnetic bragg reflections observed on cooling ce@xmath0o@xmath0fese@xmath0 ( fig.[peaks ] )
can be rationalised in terms of the contribution of the ce and fe moments to peak intensities .
the magnetic modes that describe the ordering of both the fe and ce moments have the same basis vector @xmath39 = ( 0 @xmath40 @xmath40 ) . as a consequence , the ordering on these two sublattices contributes to mostly the same reflections .
based on the magnetic unit cell @xmath41 @xmath33 2@xmath42 @xmath33 2@xmath43 , the @xmath45 reflections with @xmath56 = 2n , @xmath57 @xmath58 2n and @xmath59 = 2n ( i.e. , ( 0 1 1 ) , ( 0 1 3 ) , ( 0 1 5 ) ) have contributions from both ce and fe sublattices , whilst some weaker @xmath45 reflections with @xmath56@xmath582n , @xmath57@xmath582n and @xmath59@xmath582n have contributions only from the ce ordering .
the ordering of the fe and ce sublattices adds constructively for some peak intensities ( e.g. , ( 0 1 1 ) , ( 0 1 9 ) ) and destructively for others ( e.g. , ( 0 1 5 ) , ( 0 1 7 ) , ( 0 1 3 ) ) ( for this latter ( 0 1 3 ) reflection the ce contribution is small and so the intensity is dominated by fe ordering ) . the non - monotonic temperature dependence of the magnetic reflection intensities observed for ce@xmath0o@xmath0fese@xmath0 is similar to those reported for @xmath60cuo@xmath6 @xmath61 = pr , nd , @xmath62 = 250 - 325 k ) @xcite and for cevo@xmath63 ( @xmath8 = 124 - 136 k ) @xcite .
ins was used to obtain experimental estimates for the magnetic exchange interactions .
low - energy fluctuations were studied to probe directly the fe - fe exchange along the chains .
ce crystal electric field ( cef ) excitations were then investigated to determine the fe - ce exchange . before discussing the scattering response from magnetic ions , we first describe how the background was subtracted from the powder averaged data .
the measured neutron scattering intensity @xmath64 is proportional to the structure factor @xmath65 but also includes a temperature - independent background contribution due to instrument effects and sample environment .
using the principle of detailed balance , we employ data collected at different temperatures to account for the temperature - independent background .
this allows us to isolate the inelastic scattering ( appendix a ) , which has both magnetic and lattice ( phonon ) contributions . to extract the purely magnetic scattering
, we assume that the 300 k scattering is dominated by phonons , which is a reasonable approximation . scaling by the bose factor and assuming a harmonic response ( appendix 1 )
, we estimate the phonon cross section at each temperature and subtract it from the background - corrected data . using this method we extract the purely magnetic scattering at a given temperature , as has been used previously to study the hydrogen - containing polymeric magnet cu(quinoxaline)br@xmath0 @xcite and the low - energy magnetic dynamics of fe@xmath2te@xmath66se@xmath67 @xcite .
the purely magnetic contributions to the inelastic scattering are shown in figure [ mari_scans ] for @xmath28 = 40 mev .
a strong , sharp excitation , independent of @xmath68 , is observed at @xmath69 @xmath21 11 mev , which is ascribed to the ce@xmath20 cef excitations . at slightly lower energies ,
a gapped excitation is observed near @xmath68 = 0 .
the gap value is similar to that observed in other parent fe@xmath14 based superconductors such as fe@xmath10te @xcite and la@xmath0o@xmath0fe@xmath0ose@xmath0 @xcite and in 122 systems including bafe@xmath0as@xmath0 @xcite . unlike the crystal field excitation , this scattering is well - defined in momentum and decays quickly with momentum transfer , bearing a strong resemblance to the magnetic excitation observed in powder averaged measurements of la@xmath0o@xmath0fe@xmath0ose@xmath0 @xcite .
the temperature - dependence of this excitation is also different from that of the ce@xmath20 cef excitation : at 4 k , it has a gap of @xmath219 mev , which decreases on warming and softens into the elastic line by 115 k ( fig.[mari_scans]@xmath70 ) .
based on these observations , we conclude that the low energy , low-@xmath68 scattering originates from the fe@xmath14 magnetic sublattice . to separate the ce cef excitations from the fe@xmath14 magnetic excitations , the cef contribution was estimated by taking a cut over the momentum transfer range of @xmath68 = 2.8 - 3.5 @xmath71 , and then scaling by the ce@xmath20 form factor @xcite to estimate the momentum dependence .
this subtraction takes advantage of the fact that the crystal field excitations are dispersionless and flat in momentum transfer , particularly in comparison with the strong momentum dependence of the scattering associated with the fe sites ( as observed for la@xmath0o@xmath0fe@xmath0ose@xmath0 @xcite for example ) .
this analysis leaves only the strongly momentum varying component near @xmath68 = 0 ( fig.[mari_scans]@xmath72 ) , from which the magnetic exchange interactions between fe@xmath14 sites can be estimated .
the single - mode approximation @xcite can be used to compare possible magnetic structures with different signs ( afm @xmath73 , fm @xmath74 ) and magnitudes for the nn interaction @xmath75 ( illustrated in figure [ dft ] ) . using the single mode approximation the structure factor @xmath76
can be written in terms of a momentum - dependent term @xmath77 and a single dirac delta function in energy : @xmath78 . \label{zeroeth_sum}\end{aligned}\ ] ] where @xmath79 is the dispersion .
we approximate @xmath80 as a lorentzian term with full - width equal to the calculated resolution width in energy .
the first moment sum rule @xcite relates @xmath77 to the dispersion : @xmath81 . \label{first_moment_sum}\end{aligned}\ ] ] where @xmath82 is the bond vector connecting nn spins with an exchange interaction @xmath75 . making the assumption that this intrachain interaction dominates , we use the dispersion relation for the one - dimensional ( 1d ) chain system : @xmath83^{2 } ] .
\label{disperion}\end{aligned}\ ] ] where @xmath84 is the gap value determined by anisotropy and @xmath75 is the nn intrachain exchange interaction .
representative calculations using the afm and fm chain models are summarized in figure [ mari_scans ] .
the afm model gives correlations at finite @xmath68 , whereas the fm model gives magnetic scattering only at lowest measurable wave vectors ( near @xmath68 = 0 ) .
from the temperature dependence ( figure [ mari_scans]@xmath85 ) and the subtracted data ( figure [ mari_scans]@xmath72 ) , the strongly temperature dependent magnetic scattering is present near @xmath86 .
this is more consistent with a dominant fm @xmath75 interaction ( simulated in figure [ mari_scans]@xmath87 ) than an afm interaction where the scattering is peaked at finite @xmath68 .
based on this comparison , we conclude that the exchange mechanism is predominately ferromagnetic ( @xmath74 ) , consistent with analysis of npd data described above .
figure [ mari_scans ] @xmath88 shows results of single mode calculations for this fm chain model for various values of @xmath75 .
it is difficult to give an accurate value for this exchange interaction given the scattering is concentrated near @xmath86 , but our calculations indicate that @xmath75 @xmath2110 - 20 mev gives the best qualitative agreement with the observed data .
it should be emphasized that this is an estimate of the coupling and is limited by the kinematics of the scattering geometry described above .
o@xmath0fese@xmath0 showing cef levels for ce@xmath20 site as a function of temperature .
the solid curves are fits to a gaussian function . ]
having discussed the fe - fe exchange , we now discuss the localized ce@xmath20 cef excitations observed in the ins data at @xmath2111 mev and @xmath2137 mev ( fig.[cef ] ) with the goal of extracting the coupling between fe and ce sites .
the magnetic nature of the peak around 11 mev is confirmed by the temperature dependence shown in figure [ cef ] .
the softening of the first crystal field excitation ( fig .
[ cef]d - f ) with increasing temperature could be the result of thermal expansion or of a change in the ground state @xcite .
we note that the softening observed can be reproduced by point charge calculations and is consistent with thermal expansion . to obtain an estimate of fe
ce exchange , it is important to have a heuristic model for the ce crystal fields from which eigenfunctions and transition energies can be derived .
ce@xmath20 ( 4f@xmath89 , @xmath90 ) is a kramers ion ( fig .
[ cef ] ) , and each level remains doubly degenerate for all crystalline electric fields unless a magnetic field is applied .
magnetic ordering on the iron sublattice can give rise to a molecular field at the ce@xmath20 sites if there is coupling between fe@xmath14 and ce@xmath20 ions . in the oxyarsenide cefeaso
, the degeneracy of ce@xmath20 cef states is lifted below @xmath91 suggesting some fe - ce coupling @xcite , which is consistent with muon spin rotation spectroscopy studies @xcite . in the vacancy - ordered ce@xmath0o@xmath0fese@xmath0 structure ( space group @xmath92 ) ,
the ce atoms are on 8@xmath93 sites with local point symmetry @xmath94 .
the resulting crystal field hamiltonian can be expressed in stevens operators formalism which requires five nonzero terms to describe the monoclinic symmetry of the ce@xmath20 site @xcite : @xmath95 ideally , these five terms would be determined by fitting to the experimental data but they can not be uniquely and unambiguously determined with only two cef energies and intensities and the ordered moment from the npd analysis ( which depends on the ce ground state wavefunction determined from the eigenvectors of eqn .
[ full_hamiltonian ] ) .
therefore , a simplified model for the crystal field scheme has been investigated . with no vacancy ordering on the iron sublattice ( fig.[structure ] )
, the symmetry of the ce@xmath20 sites would be tetragonal and the hamiltonan for this scheme contains only three non - zero terms : @xmath96 using the two cef energies and intensities and the ordered moment from npd ( giving five experimental data points " ) , the three coefficients are determined as @xmath97 = 1.5(2 ) mev , @xmath98 = -0.03(1 ) mev @xmath99 = -0.43(7 ) mev .
the ambiguity regarding the sign of these coefficients was resolved with the results from a cluster " point charge calculation integrating over 40 unit cells to ensure convergence of the stevens coefficients ( see appendices for details of this calculation ) .
we now use this heuristic model of the crystal fields to derive an exchange coupling between the fe ce ions based upon the broadening of the crystal field levels in the magnetically ordered low temperature phase . because of kramer s theorem
, the crystal field excitations are doubly degenerate and only split in the presence of a time reversal violating magnetic field .
this splitting can be calculated by adding the following zeeman term to the crystal field hamiltonian above for eigenstates @xmath100 and @xmath93 : @xmath101 where @xmath15 and @xmath102 are the bohr magneton and permeability of free space , respectively , @xmath103 is the effective magnetic field , @xmath104 is an angular momentum operator along @xmath51 , and @xmath105 is the angular momentum matrix element from the ground state to the excited state . to account for the powder averaging ,
all three directions ( @xmath106 , @xmath107 , and @xmath51 ) were averaged .
the molecular field on the ce site is induced by magnetic ordering on the fe sublattice ( fig.[structure]@xmath72 ) . in the absence of a molecular field at the ce sites , any
splitting / broadening of the kramers doublets should arise from the fe ce coupling @xcite and is consistent with muon spin relaxation studies , which indicate a strong non - heisenberg anisotropic fe ce exchange well above @xmath108 in cefeaso @xcite . in the fe - ordered crystal structure of ce@xmath0o@xmath0fese@xmath0 ( fig.[structure]@xmath32 ) , the fm chains of edge - shared fese@xmath6 tetrahedra alternate with vacant stripes along [ 010 ] and each ce site is coupled to two fe sites within a single fm chain ( fig.[structure]@xmath72 ) ) and there are no competing fe ce interactions . the molecular field on the ce site due to the fe magnetic sublattice is equal to 2@xmath109 where @xmath110 is the fe ce exchange coupling .
this provides an opportunity to probe the fe ce coupling by measuring the broadening of the crystal field excitations .
the cef levels observed for ce@xmath0o@xmath0fese@xmath0 are broadened ( fig.[cef ] @xmath111 ) considerably beyond the instrumental resolution ( represented by the horizontal bar in fig.[cef ] @xmath112 ) , but it is difficult to determine the splitting of the kramers doublets ( fig.[cef ] ) in contrast to the case of ceasfeo . to provide an estimate for the fe
ce exchange coupling , we have fitted the low temperature excitation to a single gaussian to obtain a full - width of 2.0(4 ) mev , giving a maximum value for any splitting of @xmath211 mev . using the stevens parameters discussed above for the crystal field analysis
, we obtain an estimate of the fe ce exchange @xmath110 of @xmath210.15 mev .
this estimate is approximately an order of magnitude smaller than the ferromagnetic fe fe exchange @xmath75 interaction discussed above .
( a ) four spin exchange paths of ce@xmath0o@xmath0fese@xmath0 . ( b - f ) five ordered spin arrangements fm and af1 - af4 employed to extract @xmath75
@xmath110 by energy mapping analysis .
the numbers in the parentheses in the first row refer to the relative energies in mev / fu , and the bracketed numbers represent the numbers @xmath113 , @xmath114 , @xmath115 and @xmath116 of eq .
[ energy ] .
af2 is the structure observed experimentally from diffraction with a weak @xmath110 exchange expected from neutron inelastic scattering . ]
summarizing the experimental results above , we observe ferromagnetic fe fe and weaker ferromagnetic fe ce exchange .
this is based on both magnetic neutron diffraction and inelastic scattering results . in this section ,
we provide electronic structure calculations with the goal of understanding these results and comparing them with previous calculations .
figure [ dft]@xmath22 shows the four spin exchanges of ce@xmath0o@xmath0fese@xmath0 we investigate , namely , the intrachain exchange , @xmath75 , and the interchain exchanges , @xmath117 and @xmath118 , between fe@xmath14 ions as well as the exchange @xmath110 between fe@xmath14 and ce@xmath20 ions . to extract the values of @xmath75
@xmath110 by energy - mapping analysis @xcite , we consider five ordered spin states fm and af1
af4 presented in figure [ dft]@xmath119 .
the fm , af1 and af2 states contain fm chains .
the coupling between adjacent fm chains is fm in the fm and af1 states , but afm in the af2 state .
the coupling between the fe@xmath14 and ce@xmath20 ions is fm in the fm and af2 states , but afm in the af1 state .
the af3 and af4 states consist of afm chains so that the net spin exchange between the fe@xmath14 and ce@xmath20 ions vanishes .
the coupling between adjacent afm chains is fm in the af3 state , but afm in the af4 state .
the af2 state is closest to that observed experimentally .
the total spin exchange energies of the fm and af1
af4 states can be expressed in terms of the spin hamiltonian , @xmath120 where @xmath121 = @xmath75 @xmath110 is the spin exchange parameter for the interaction between the spin sites @xmath100 and @xmath93 . by applying
the energy expression obtained for spin dimers with n unpaired spins per spin site ( four for fe@xmath14 , and one for ce@xmath20 ) @xcite , the total spin exchange energies per formula unit ( fu ) of the fm and af1
af4 states can be written as @xmath122 where @xmath123 = 4 and @xmath124 = 1 , and the coefficients @xmath113
@xmath116 for the five spin ordered states are summarized in figure [ dft]@xmath119 .
we examined the relative energies of the fm and af1
af4 states on the basis of dft+u electronic structure calculations with various @xmath1 and @xmath125 values . from their dft+u calculations ,
li et al @xcite found that the experimentally reported magnetic structure ( namely , the af2 state ) is stable for @xmath125 = 12 ev with with @xmath1 = 0 .
our calculations show that structures with fm chains ( af1 and af2 ) are significantly more stable than those with afm chains ( af3 and af4 ) for all combinations of @xmath1 and @xmath125 given here ( table [ relative_energies ] ) ( for @xmath1 @xmath126 2 ev ,
the nn fe se fe spin exchange @xmath75 becomes afm ) .
af1 and af2 spin arrangements differ in the sign of fe ce exchange @xmath110 : in af1 , ce spins are antiparallel to nn fe spins ( i.e. afm @xmath110 exchange ) , whereas ce spins are parallel to nn fe spins ( i.e. fm @xmath110 ) in the af2 arrangement .
we note that the experimentally observed af2 arrangement is the more energetically favourable for all @xmath1 and @xmath125 combinations considered , but that the relative stability of af2 over af1 is much greater with @xmath125 = 12 ev than with @xmath125 = 10 ev .
c c c c c c & & af1 & af2 & af3 & af4 + @xmath1 = 2 , @xmath125 = 12 & & + 1.2 & -33.0 & + 149.5 & + 159.8 + @xmath1 = 2 , @xmath125 = 10 & & -4.9 & -9.7 & + 139.1 & + 178.1 + @xmath1 = 0 , @xmath125 = 12 & & + 0.1 & -35.0 & + 167.5 & + 169.0 + @xmath1 = 0 , @xmath125 = 10 & & 0.0 & -5.9 & + 218.1 & + 222.1 + [ relative_energies ] by mapping the relative energies of the fm and af1 af4 states , determined from the dft+u calculations with @xmath1 = 2 ev and @xmath125 = 12 ev
( see fig .
[ dft]@xmath119 ) , onto the corresponding relative energies determined from eq .
[ energy ] , we obtain @xmath75 = 21.3 mev , @xmath117 = -1.4 mev , @xmath118 = -1.4 mev , and @xmath110 = 0.2 mev ( see table [ table_energy_mapping ] ) .
these spin exchanges are consistent with the observed magnetic structure of ce@xmath0o@xmath0fese@xmath0 .
the calculated value for the fe se fe exchange , @xmath75 @xmath21 21 mev , is comparable in magnitude to the experimental value of about 10 20 mev from ins .
furthermore , the calculated value for the fe ce exchange , @xmath110 = 0.2 mev , is in good agreement with the experimental value of about 0.15 mev .
similar energy - mapping analyses were carried out with other values of @xmath1 and @xmath125 , as shown in table [ table_energy_mapping ] .
these calculations show that @xmath75 is strongly fm and that @xmath110 is weakly fm for the @xmath1 = 2 ev and @xmath125 = 12 ev combination ( all other combinations give @xmath110 weakly afm ) .
c c c c c c & & @xmath75 & @xmath117 & @xmath118 & @xmath110 + @xmath1 = 2 , @xmath125 = 12 & & + 21.3 & -1.4 & -1.4 & + 0.2 + @xmath1 = 2 , @xmath125 = 10 & & + 20.7 & + 1.8 & -1.5 & -0.6 + @xmath1 = 0 , @xmath125 = 12 & & + 23.2 & -2.1 & -1.1 & + 0.0 + @xmath1 = 0 , @xmath125 = 10 & & + 27.9 & -0.1 & -0.3 & 0.0 + [ table_energy_mapping ] figure [ pdos ] shows plots of the projected density of states ( pdos ) obtained for the ce 4f , ce 5d , fe 3d and se 4p states of ce@xmath0o@xmath0fese@xmath0 from the dft+u calculations with @xmath1 = 2 ev and @xmath125 = 6 , 8 , 10 and 12 ev .
it shows a band gap of about 1 ev which is consistent with the semiconducting behavior of ce@xmath0o@xmath0fese@xmath0 observed experimentally ( with band gap of 0.64 ev ) @xcite .
we note from figure [ pdos ] that the fe 3d states overlap with the se 4p states througout the filled energy region , which indicates that the interaction between fe 3d and se 4p orbitals takes place throughout this energy range .
the ce 5d states contribute to the filled region of the fe 3d and se 4p states , and these contributions are not strongly affected by the change in @xmath125 . however , on increasing @xmath125 from 6 ev to 12 ev , the ce 4f states are gradually lowered in energy such that they overlap with the filled fe 3d and se 4p states when @xmath125 @xmath127 12 ev , but do not when @xmath125 @xmath128 12 ev .
likewise , the filled ce 4f states overlap with the ce 5d states when @xmath125 @xmath127 12 ev but do not when @xmath125 @xmath128 12 ev .
the ferromagnetic fe se ce spin exchange @xmath110 ( and increased stability of the experimentally observed af2 spin arrangement ) is found when the ce 5d states do not overlap in energy with the fe 3d and se 4p states in the energy region within 2 ev below the fermi level .
this is understandable because an antiferromagnetic fe - se - ce spin exchange would involve the fe 3d , se 4p and ce 5d orbitals . that the electronic structure of ce@xmath0o@xmath0fese@xmath0 is described by using a small value of @xmath1 ( 2 ev ) suggests a weakly correlated nature of the iron - selenide sheets in ce@xmath0o@xmath0fese@xmath0
this is similar to that found for parent materials to iron - based superconductors ( @xmath1 @xmath21 2 ev for smfeaso and bafe@xmath0as@xmath0 @xcite and @xmath21 4 ev used to describe the electronic properties of k@xmath129fe@xmath130se@xmath0 @xcite .
the need for a large on - site repulsion @xmath125 for ce ( 12 ev ) is comparable to the trend established for cefeaso where @xmath125 = 9 ev was required @xcite .
in the low - temperature magnetic structure of ce@xmath0o@xmath0fese@xmath0 , both ce@xmath20 and fe@xmath14 moments lie within the @xmath9 plane , similar to the structure reported for cefeaso @xcite .
the observation of in - plane ce@xmath20 moments is consistent with the easy - axis along @xmath106 proposed for ce@xmath20 sites in orthorhombic cefeaso @xcite .
the ordered ce@xmath20 moment of ce@xmath0o@xmath0fese@xmath0 at 4 k ( 1.14(4 ) @xmath15 ) is slightly larger than that reported for cefeaso ( 0.83(2 ) @xmath15 at 1.7 k ) @xcite , and is close to that expected for a ce@xmath20 doublet ground state ( 1 @xmath15 ) @xcite .
the high ordering temperature for the ce moments implied by our npd data is surprising ; other systems known to have high ce ordering temperatures include cerh@xmath63b@xmath0 ( 115 k ) in which ce@xmath20 ion is the only magnetic ion @xcite and cevo@xmath63 ( 50 k ) in which ce ordering is thought to arise from fm exchange between ce@xmath20 and v@xmath20 ions @xcite .
the high ce@xmath20 moment ordering temperature in ce@xmath0o@xmath0fese@xmath0 is most probably due to the fm spin exchange between adjacent ce@xmath20 and fe@xmath14 ions , that is , the long range magnetic order of the fe@xmath14 sublattice induces that of the ce@xmath20 ions . in cefeaso , each ce site is coupled to two fm chains of edge - sharing feas@xmath6 tetrahedra with opposite spin orientations @xcite , leading to frustration of any ce fe exchange interactions @xcite .
this is expected to give a negligible field on the ce site ( consistent with the low ce moment ordering temperature ) .
the very small cef splitting ( @xmath211 mev ) observed for ce@xmath0o@xmath0fese@xmath0 is similar to that described for the parasitic ordering of ce@xmath20 moments in cemnaso @xcite .
the nn fe fe magnetic exchange interactions , @xmath75 , determined here experimentally are in good agreement with our dft calculations .
they are similar in magnitude to those reported for cefeaso @xcite but of opposite sign .
they are significantly larger than those reported for la@xmath0o@xmath0fe@xmath0ose@xmath0 , in which the fe@xmath14 cations are coordinated by both oxide and selenide anions , which may give rise to more strongly correlated behavior @xcite .
we note that the nn @xmath75 interactions in ce@xmath0o@xmath0fese@xmath0 are fm , which may reflect some orbital ordering on fe sites , as proposed for the pnictides @xcite .
in conclusion , the fm nature of fe se fe nn interactions @xmath75 has been confirmed by npd and ins measurements .
ins work indicates that this exchange is @xmath2110 20 mev .
this is consistent with dft + u calculations for @xmath1 @xmath131 2 ev and suggest that the fe 3d electrons in the fe@xmath132se sheets in ce@xmath0o@xmath0fese@xmath0 are weakly correlated , similar to the feas sheets in smfeaso and bafe@xmath0as@xmath0 .
weak fm fe se
ce interactions of 0.15 - 0.20 mev ( reproduced by dft + u calculations for @xmath1 = 2 ev ) are not frustrated in this cation - ordered zrcusias - related structure .
therefore `` parasitic '' ordering of ce@xmath133 might be induced by magnetic ordering of the fe sublattice , with ce@xmath20 moments parallel to adjacent fe@xmath14 moments .
see supplemental material at [ url will be inserted by publisher ] for details of analysis of npd data including relative intensity of magnetic reflections on cooling as well as surface plots for magetic refinements .
we acknowledge stfc , epsrc ( ep / j011533/1 ) , royal society of edinbugh , and the nsf ( dmr-0944772 ) for funding .
we thank emma suard ( ill ) , ross stewart ( isis ) for assistance and mark green ( kent ) and efrain rodriguez ( maryland ) for helpful discussions .
the principle of detailed balance can be used to estimate the temperature - independent background contribution to the scattering .
we can approximate that for a fixed wave vector and energy transfer , the neutron energy gain ( negative energy transfer , @xmath134 ) and neutron energy loss ( positive energy transfer , @xmath135 ) are related by the following expression from the detailed balance principle : @xmath136 @xmath137 \label{a2}\end{aligned}\ ] ] where @xmath138 and @xmath139 are temperature - independent background points , @xmath140 is the scattered intensity ( with both magnetic and phonon contributions ) and @xmath141 $ ] is the boltzmann factor .
we assume that the resolution of the inelastic scattering does not change over the energy range investigated . with data collected at two or more temperatures , @xmath138 and @xmath139 can be determined . for @xmath28 = 40 mev ,
data were collected at six temperatures ( 4 k , 75 k , 115 k , 150 k , 200 k and 300 k ) .
these data give us experimental data points in both the energy gain and energy loss spectra ( giving 12 data points in total ) with which the two background points @xmath138 and @xmath139 and the six values for @xmath142 can be determined .
this detailed balance allows us to isolate the inelastic scattering but this has contributions from both magnetic and phonon scattering .
the measured intensity @xmath64 is proportional to the structure factor @xmath65 which is related to the imaginary part of the susceptibility @xmath143 : @xmath144 \chi''(q , e ) \label{a3}\end{aligned}\ ] ] where n(e ) is the bose factor .
the scattering at 300 k is dominated by phonons and so the phonon contribution @xmath145 can be written : @xmath146 } \label{a4}\end{aligned}\ ] ] the phonon contribution at each temperature was then estimated using equation a3 and subtracted to obtain the purely magnetic scattering at each temperature .
to guide the cef analysis , we used a point charge cluster " model ( integrated over 40 unit cells to ensure convergence of the stevens coefficients ) which gave the results shown in table [ table_stephens ] .
10 y. kamihara , t. watanabe , m. hirano and h. hosono , j. am .
soc . * 130 * , 3296 ( 2008 ) .
j. paglione and r. l. greene , nat . phys .
* 6 * , 645 ( 2010 ) .
g. r. stewart , rev .
mod . phys . * 83 * , 1859 ( 2011 ) .
d. c. johnston , adv .
phys . * 59 * , 803 ( 2010 ) .
ren , w. wu , j. yang , w. yi , x .-
shen , z .- c . li , g .- c .
dong , l .- l .
sun , f. zhou , z .- x .
zhao , chin .
lett . * 25 * , 2215 ( 2008 ) .
a. l. wysocki , d. d. belashchenko , v. p. antipov , nat .
* 7 * , 485 ( 2011 ) .
r. j. birgeneau , c. stock , j. m. tranquada , k. yamada , j. phys .
soc . jpn . * 11 * , 111003 ( 2006 ) .
m. a. kastner , r. j. birgeneau , g. shirane , y. endoh , rev .
mod . phys . * 70 * , 897 ( 1998 ) .
v. johnson , w. jeitschko , j. solid state chem . * 11 * , 161 ( 1974 ) .
z. ban , m. sikirica , acta cryst . * 18 * , 594 ( 1965 ) . c. de la cruz , q. huang , j. w. lynn , j. li , w. ratcliff , k. l. zaretsky , h. a. mook , g. f. chen , j. l. luo , n. l. wang , p. dai , nature * 453 * , 899 ( 2008 ) .
a. i. goldman , d. n. argyriou , b. ouladdiaf , t. chatterji , a. kreyssig , s. nandi , n. ni , s. l. budko , p. c. canfield , r. j. mcqueeney , phys .
b * 78 * , 100506 ( 2008 ) .
e. e. rodriguez , p. zavalij , p .- y .
hsieh , m. a. green , j. am .
132 * , 10006 ( 2010 ) .
y. mizuguchi , f. tomioka , s. tsuda , t. yamaguchi , y. takano , appl .
. lett . * 94 * , 012503 ( 2009 ) .
hsu , j .- y .
luo , k .- w .
yeh , t .- k .
cen , t. w. huang , p. m. wu , y .- c .
lee , y .- l .
huang , y. y. chu , d .-
yan , m .- k .
wu , pnas * 105 * , 14262 ( 2008 ) . s. margadonna , y. takabayashi , y. ohishi , y. mizuguchi , y. takano , t. kagayama , t. nakagawa , m. takata , k. prassides , phys .
b * 80 * , 064506 ( 2009 ) .
j. guo , s. jin , g. wang , s. wang , k. zhu , t. zhou , m. he , x. chen , phys .
b * 82 * , 180520(r ) ( 2010 ) .
yan , m. gao , z .- y . lu , t. xiang , phys .
b * 83 * , 233205 ( 2011 ) .
h. lei , m. abeykoon , e. s. bozin , c. petrovic , phys .
b * 83 * , 180503(r ) ( 2011 ) . m. -h .
fang , h .-
wang , c .- h .
dong , z .- j .
li , c .- m .
feng , j. chen , and h. q. yuan , europhys .
* 94 * , 27009 ( 2011 ) .
j. zhao , h. cao , e. bourret - courchesne , d .- h .
lee , r. j. birgeneau , phys .
109 * , 267003 ( 2012 ) .
n. lazarevic , h. lei , c. petrovic , z. v. popovic , phys .
b * 84 * , 214305 ( 2011 ) .
k. wang , h. lei , c. petrovic , phys .
b * 83 * , 174503 ( 2011 ) .
w. bao , q .- z .
huang , g .- f .
chen , m. a. green , d .-
wang , j .- b .
he , y .- m . qiu , chin . phys .
* 28 * , 086104 ( 2011 ) .
j. zhao , q. huang , c. de la cruz , s. li , j. w. lynn , y. chen , m. a. green , g. f. chen , g. li , z. li , j. l. luo , n. l. wang , p. dai , nat . mat . * 7 * , 953 ( 2008 ) . c. fang , h. yao , w .- f .
tsai , j. p hu , s. a. kivelson , phys . rev .
b * 77 * , 224509 ( 2008 ) .
t. yildirim , phys .
* 101 * , 057010 ( 2008 ) .
q. si , e. abrahams , phys .
* 101 * , 076401 ( 2008 ) .
t. j. li , y .- m .
quan , d .- y .
liu , l .- j .
zou , j. magn .
mater . * 324 * , 1046 ( 2012 ) .
h. maeter , h. luetkens , y. g. pashkevich , a. kwadrin , r. khasanov , a. amato , a. a. gusev , k. v. lamonova , d. a. chervinskii , r. klingeler , c. hess , g. behr , b. buchner , h .- h .
klauss , phys .
b * 80 * , 094524 ( 2009 ) . q. zhang , w. tian , h. li , j .- w .
kim , j. yan , r. w. mccallum , t. a. lograsso , j. l. zarestky , s. l. budko , r. j. mcqueeney , d. vaknin , phys .
b * 88 * , 174517 ( 2013 ) .
e. m. bruning , c. krellner , m. baenitz , a. jesche , f. steglich , c. geibel , phys .
* 101 * , 117206 ( 2008 ) . c. krellner
, n. s. kini , e. m. bruning , k. koch , h. rosner , m. nicklas , m. baenitz , c. geibel , phys .
b * 76 * , 104418 ( 2007 ) .
e. e. mccabe , d. g. free , j. s. o. evans , chem .
commun . * 47 * , 1261 ( 2011 ) .
j. b. goodenough , _ magnetism and the chemical bond _ , john wiley and sons , new york - london , ( 1963 ) .
j. b. goodenough , phys . rev .
* 100 * 564 ( 1955 ) .
j. kanamori , j. phys .
solids * 10 * 87 ( 1959 ) .
h. m. rietveld , j. appl . cryst .
* 2 * , 65 ( 1969 ) .
a. a. coelho , j. appl .
cryst . * 36 * , 86 ( 2003 ) .
a. a. coelho , _ topasacademic : general profile and structure analysis software for powder diffraction data _ , bruker axs : karlsruhe , germany ( 2012 ) . b. j. campbell , h. t. stokes , d. e. tanner , d. m. hatch , j. appl . cryst .
* 39 * , 607 ( 2006 ) .
g. kresse , j. hafner , phys .
b * 47 * , 558 ( 1993 ) .
g. kresse , j. furthmller , computational materials science * 6 * 15 ( 1996 ) .
g. kresse , j. furthmller , phys .
b * 54 * , 11169 ( 1996 )
. j. p. perdew , k. burke , m. ernzerhof , phys .
lett . * 77 * , 3865 ( 1996 ) .
s. l. dudarev , g. a. botton , s. y. savrasov , c. j. humphreys , a. p. sutton , phys .
b * 57 * , 1505 ( 1998 ) . a. j. tuxworth , e. e. mccabe , d. g. free , s. j. clark , j. s. o. evans , inorg . chem . *
52 * , 2078 ( 2013 ) .
d. bhoi , p. mandal , p. choudhury , s. pandya , v. ganesan , j. appl . phys .
* 110 * , 113722 ( 2011 ) .
e. e. mccabe , c. stock , e. e. rodriguez , a. s. wills , j. w. taylor , j. s. o. evans , phys .
b * 89 * , 100402(r ) ( 2014 ) .
s. d. wilson , z. yamani , c. r. rotundu , b. freelon , e. bourret - courchesne , r. j. birgeneau , phys .
b * 79 * , 184519 ( 2009 ) .
m. collins , _ magnetic critical scattering ( oxford series on neutron scattering in condensed matter ) _ , oxford university press , new york ( 1989 ) .
chen , b. moritz , j. van den brink , t. p. devereaux , r. r. p. singh , phys .
b * 80 * , 180418(r ) ( 2009 ) .
e. e. rodriguez , d. a. sokolov , c. stock , m. a. green , o. sobolev , j. a. rodriguez - rivera , h. cao , a. daoud - aladine , phys .
b * 88 * , 165110 ( 2013 ) .
d. m. pajerowski , c. r. rotundu , j. w. lynn , r. j. birgeneau , phys .
b * 87 * , 134507 ( 2013 ) .
m. matsuda , k. yamada , k. kakurai , h. kadowaki , t. r. thurston , y. endoh , y. hidaka , r. j. birgeneau , m. a. kastner , p. m. gehring , a. h. moudden , g. shirane , phys .
b * 42 * , 10098 ( 1990 ) .
a. munoz , j. a. alonso , m. t. casais , m. j. martinez - lope , j. l. martinez , m. t. fernandez - diaz , phys .
b * 68 * , 144429 ( 2003 ) .
m. reehuis , c. ulrich , p. pattison , m. miyasaka , y. tokura , b. keimer , eur .
j. b * 64 * , 27 ( 2008 ) . t. hong , m. kenzelmann , m. m. turnbull , c. p.
landee , b. d. lewis , k. p. schmidt , g. s. uhrig , y. qiu , c. broholm , d. reich , phys .
b * 74 * , 094434 ( 2006 ) . c. stock , e. e. rodriguez , m. a. green , phys . rev .
b * 85 * , 094507 ( 2012 ) . c. stock , e. e. rodriguez , m. a. green , p. zavalij , j. a. rodriguez - rivera , phys .
b * 84 * , 045124 ( 2011 ) .
s. d. wilson , z. yamani , c. r. rotundu , b. freelon , p. n. valdivia , e. bourret - courchesne , j. w. lynn , s. chi , t. hong , r. j. birgeneau , phys . rev .
b * 82 * , 144502 ( 2010 ) .
p. j. brown , _ magnetic form factors _ in _ international tables for crystallography _ * c * , edited by a. j. c. wilson , kluwer academic publishers : dordrecht , the netherlands ( 2006 ) .
c. stock , l. c. chapon , o. adamopoulos , a. lappas , m. giot , j. w. taylor , m. a. green , c. m. brown , p. g. radaelli , phys .
lett . * 103 * 077202 ( 2009 ) .
p. c. hohenberg , w. f. brinkman , phys .
b * 10 * , 128 ( 1974 ) .
p. fulde , _ crystal fields _ in _ handbook on the physics and chemistry of rare earths _ , edited by k. a. gschneidner and l. eyring , north - holland publishing company , p.279 ( 1978 ) .
s. chi , d. t. adroja , t. guidi , r. bewley , s. li , j. zhao , j. w. lynn , c. m. brown , y. qiu , g. f. chen , j. l. lou , n. l. wang , p. dai , phys .
* 101 * , 217002 ( 2008 ) .
u. walter , j. phys .
solids * 45 * 401 ( 1984 ) .
whangbo , h .- j .
koo , d. dai , j. solid state chem . * 176 * 417 ( 2003 ) .
d. dai , m .- h .
whangbo , h .- j .
koo , x. rocquefelte , s. jobic , a. villesuzanne , inorg .
* 44 * 2407 ( 2005 ) .
h. j. xiang , c. lee , h .- j .
koo , x. g. gong , m .- h .
whangbo , dalton trans . * 42 * , 823 ( 2013 ) .
d. dai , m .- h .
whangbo , j. chem . phys . * 114 * 2887 ( 2001 ) .
d. dai , m .- h .
whangbo , j. chem . phys . * 118 * 29 ( 2003 ) .
w. li , c. setty , x. h. chen , j. hu , front .
doi 10.1007/s11467 - 014 - 0428-y ( 2014 ) .
w. l. yang , a. p. sorini , c .- c .
chen , b. moritz , w. s. lee , f. vernay , p. olalde - velasco , j. d. denlinger , b. delley , j .- h . chu , j. g. analytis , i. r. fisher , z. a. ren , j. yang , w. lu , z. x. zhao , j. van den brink , z. hussain , z .- x .
shen , t. p. devereaux , phys .
b * 80 * , 014508 ( 2009 ) .
i. a. nekrasov , m. v. sadovskii , j. exp .
. phys . * 99 * , 598 ( 2014 ) .
l. pourovskii , v. vildosola , s. biermann , a. georges , eur .
. lett . * 84 * 37006 ( 2008 ) .
o. v. gornostaeva , k. v. lamonova , s. m. orel , y. g. pashkevich , low temp .
phys . * 39 * 343 ( 2013 ) .
f. givord , j .- x .
boucherie , e. lelievre - berna , p. lejay , j. phys . condens .
matter * 16 * 1211 ( 2004 ) .
li , d .- x .
yao , y .- m .
qiu , h. j. kang , e. w. carlson , j .-
hu , g .- f .
chen , n .- l . wang , p .- c .
dai , front .
china * 5 * 161 ( 2010 ) .
y. tsukomoto , y. okamoto , k. matsuhira , m .- h .
whangbo , z. hiroi , j. phys .
. jpn . * 80 * 094708 ( 2011 ) .
m. j. han , q. yin , w. e. pickett , s. y. savrasov , phys .
lett . * 102 * , 107003 ( 2009 ) . | the discovery of superconductivity in the 122 iron selenide materials above 30 k necessitates an understanding of the underlying magnetic interactions .
we present a combined experimental and theoretical investigation of magnetic and semiconducting ce@xmath0o@xmath0fese@xmath0 composed of chains of edge - linked iron selenide tetrahedra .
the combined neutron diffraction and inelastic scattering study and density functional calculations confirm the ferromagnetic nature of nearest - neighbour fe se fe interactions in the zrcusias - related iron oxyselenide ce@xmath0o@xmath0fese@xmath0 .
inelastic measurements provide an estimate of the strength of nearest - neighbor fe fe and fe ce interactions .
these are consistent with density functional theory calculations , which reveal that correlations in the fe se sheets of ce@xmath0o@xmath0fese@xmath0 are weak .
the fe on - site repulsion @xmath1 is comparable to that reported for oxyarsenides and k@xmath2fe@xmath3se@xmath0 , which are parents to iron - based superconductors .
= 1 |
the lynds 1630 dark cloud ( lynds 1962 ) is located approximately 4@xmath3 north of the optical reflection nebula ngc 2071 in the orion b molecular cloud ( maddalena et al .
l1630 was recognized as a site of recent star formation by strom et al .
( 1975 ) , who identified a young stellar population with an estimated age of a few million years containing early b and a - type stars , optically - revealed h@xmath4 emission line stars ( t tauri stars ) , and several embedded infrared sources
. an overview of star formation in the ngc 2071 region can be found in gibb ( 2008 ) .
the l1630 dark cloud was mapped at 2.2 @xmath5 m by strom , strom , & vrba ( 1976 ) and 70 sources were detected down to a limiting magnitude k @xmath6 10.7 mag .
about one - third of these lacked optical counterparts and they estimated an extinction a@xmath7 @xmath6 20 mag toward the cloud center .
a more sensitive 2.2 @xmath5 m survey of l1630 was undertaken by lada et al .
( 1991 ) who identified 912 sources down to k @xmath6 13.0 mag , about half of which were associated with l1630 itself .
using known mk spectral types and optical photometry for a large sample of non - variable b star members , the distance to this region was estimated to be 390 pc by anthony - twarog ( 1982 ) .
but , the near and far edges of the orion ob1 association span a range of @xmath6320 pc to 500 pc ( brown , de geus , & de zeeuw 1994 ; wilson et al .
2005 ) . using _ spitzer _
irac and mips data along with low - resolution optical spectra , flaherty & muzerolle ( 2008 ) determined a cluster age of 2 @xmath8 1.5 myr and concluded that 79% of the late - type cluster members in their sample have infrared excesses .
they found evidence for active accretion in all of the ir - excess objects .
the extended ngc 2071 cluster identified by lada et al .
( 1991 ) covers an area of @xmath680 arcmin@xmath9 and contains a dense subgroup known as ngc 2071-ir ( figs . 1 and 2 ) which harbors at least nine ir sources in a 40@xmath0 @xmath1 40@xmath0 core region ( walther , geballe , & robson 1991 ; walther et al .
1993 , hereafter w93 ; tamura et al .
2007 , hereafter t07 ) . the position of ngc 2071-ir within ngc 2071 is shown in figure 2 of w93 and a wide - field perspective showing the location of this region within the orion complex can be found in fig .
3 of maddalena et al .
( 1986 ) . within ngc 2071-ir ,
the bright infrared source irs 1 is of particular interest .
the nature of this object is unclear but previous studies have concluded it may be a young embedded b - type star ( w93 ; snell & bally 1986 ) .
the ngc 2071-ir subgroup shows clear signs of active star - formation including powerful bipolar molecular outflows ( bally 1982 ; snell et al .
1984 ; aspin , sandell , & walther 1992 ; eislffel 2000 ) , herbig - haro objects ( zhao et al .
1999 ) , and water masers near irs 1 and irs 3 ( torrelles et al . 1998 ) and irs 2 ( sandell et al .
an oh maser has been detected within @xmath620@xmath0 of irs 1 ( johansson et al . 1974 ; pankonin et al .
1977 ; sandell et al .
1985 ) but the maser position is not well - enough known to determine if it is coincident with irs 1 . _
vla _ maps in nh@xmath10 lines reveal what appears to be a large ring structure or edge - on disk surrounding ngc 2071-ir ( zhou , evans , & mundy 1990 ) .
more than one outflow may be present in ngc 2071-ir . using h@xmath2 1 - 0 s(1 ) images , eislffel ( 2000 ) identified at least three outflows .
these include an east - west outflow that may be driven by irs 1 ( see also aspin , sandell , & walther 1992 ) , and further support for such an outflow comes from high - velocity knots revealed in recent 3.6 cm _ vla _ images ( carrasco - gonzlez et al . 2007 ) .
eislffel ( 2000 ) also concluded that there is an outflow oriented in the northeast - southwest direction that may be driven by irs 3 , and a third outflow close to irs 7 .
additional evidence that irs 3 is driving an outflow comes from extended jet - like structure visible in high - resolution _ vla _ radio continuum images ( torrelles et al .
aspin et al . ( 1992 ) argued that irs 2 may also be driving a compact bipolar outflow .
more recently , stojimirovi et al . ( 2008 ) have reported evidence for at least three outflows in the ngc 2071 region based on moderate resolution ( hpbw @xmath645@xmath0 ) @xmath11co and @xmath12co @xmath13 = 1 @xmath14 0 molecular line maps .
however , the association of individual outflows with specific driving sources remains somewhat uncertain due to the complex morphology of the region .
a previous x - ray observation of the ngc 2071 region with _ xmm - newton _ detected a prominent x - ray source at a position within 1@xmath0 of irs 1 ( skinner et al .
2007 , hereafter s07 ) .
the x - ray source was identified with irs 1 , but was subject to limitations on source identification in the crowded cluster core imposed by _
s @xmath64@xmath0 angular resolution .
we present here the results of a higher angular resolution _
chandra _ x - ray observation of ngc 2071-ir .
our main objectives were to confirm the previous x - ray detection of irs 1 with improved positional accuracy and identify possible mechanisms for its x - ray emission .
although the x - ray emission of optically - revealed ob stars has been extensively studied and is thought arise in shocked winds , much less is known about x - ray production in very young embedded high - to - intermediate mass stars such as irs 1 .
we confirm that irs 1 is indeed an x - ray source and discuss its x - ray properties along with several other x - ray detections in the ngc 2071-ir subgroup ( irs 2 , irs 3 , irs 4 , and irs 6 ) .
_ chandra _ s sharp angular resolution also reveals a new x - ray source associated with the radio continuum source vla 1 lying 2.@xmath02 north of irs 1 .
the vla 1 x - ray source is seen through very high absorption equivalent to a@xmath15 @xmath16 40 mag , suggesting that it is a young heavily - embedded protostar .
we also use _ spitzer _ observations and previous near - ir observations to identify ir counterparts to the x - ray sources and determine their ir spectral energy distributions .
we also use ir color - color diagrams to assign young stellar object ( yso ) classes .
the _ chandra _ observation ( obsid 7417 ) began on 2007 november 6 at 19:40:19 tt and ended on november 7 at 15:18:03 tt . pointing
was centered near ngc 2071 irs 1 at nominal pointing coordinates ( j2000 ) ra @xmath17 05@xmath18 47@xmath19 04.94@xmath20 , dec .
@xmath17 @xmath2100@xmath22 22@xmath3 11.8@xmath0 .
the exposure live time was 67,180 s. exposures were obtained using the acis - i ( advanced ccd imaging spectrometer ) imaging array in faint timed - event mode with 3.2 s frame times .
acis - i has a combined field of view ( fov ) of @xmath616.9@xmath3 x 16.9@xmath3 consisting of four front - illuminated 1024 x 1024 pixel ccds with a pixel size of 0.492@xmath0 .
approximately 90% of the encircled energy at 1.49 kev lies within 2@xmath0 of the center pixel of an on - axis point source .
our discussion here focuses on acis - i , but the s2 and s3 ccds in the acis - s array were also enabled .
more information on _ chandra _ and its instrumentation can be found in the _ chandra _ proposer s observatory guide ( pog ) .
the level 1 events file provided by the _ chandra _ x - ray center ( cxc ) was processed using ciao version 3.4 using standard science threads . the ciao processing applied calibration updates ( caldb vers .
2.26 ) , selected good event patterns , and determined source centroid positions .
additionally , we used the ciao _ wavdetect _ tool to identify x - ray sources on the acis - i array .
wavdetect _ on full - resolution images using events in the 0.3 - 7 kev range to reduce the background .
the _ wavdetect _ threshold was set at @xmath23 = 1.5 @xmath1 10@xmath24 and scale sizes of 1 , 2 , 4 , 8 , and 16 were used .
we identified 207 x - ray sources on acis - i down to a threshold of 5 counts , including the sources of primary interest in this work located in the ngc 2071-ir core region listed in table 1 .
the unabsorbed x - ray luminosity upper limit for non - detections is log l@xmath25 ( 0.3 - 7 kev ) @xmath26 28.7 ergs s@xmath27 .
this upper limit from the portable interactive multi - mission simulator ( pimms ) assumes a 5 count on - axis detection threshold and an absorbed isothermal optically thin plasma spectrum similar to that observed for irs 1 ( n@xmath28 = 2 @xmath1 10@xmath29 @xmath30 , kt = 2 kev ) . at this upper limit ,
the observation was sensitive enough to detect low - mass t tauri stars in orion down to @xmath60.1 - 0.2 m@xmath31 , based on the known correlation between l@xmath25 and stellar mass in t tauri stars ( preibisch et al .
the above upper limit should be considered as a representative value since the detection limit depends on many factors including the spectral properties of the source and extent to which the source is displaced off - axis . at the sensitivity of our observation
, we expect @xmath61 extragalactic background source in a region spanning @xmath61.6 arcmin@xmath9 , or @xmath321 extragalactic source in the @xmath640@xmath0 @xmath1 40@xmath0 ngc 2071-ir region analyzed here .
this estimate is based on hard - band ( 2 - 8 kev ) number counts from the _ chandra _ deep field ( cdf ) and other x - ray surveys ( e.g. brandt et al .
2001 ; moretti et al .
2003 ) and an assumed power - law spectrum with a photon power - law index @xmath33 = 1.4 for extragalactic sources .
the above estimate should be considered as approximate because the accuracy of the log @xmath34 - log @xmath35 distribution for extragalactic sources based on deep - field observations when applied to lower galactic latitude star - froming regions is not well - known .
getman et al .
( 2005 ) concluded that cdf number counts overestimated the number of extragalactic sources in the deep _ chandra _ coup observation of the orion nebula cluster by a factor of @xmath163 - 4 .
ciao _ psextract _ was used to extract source and background spectra for brighter sources , along with source - specific response matrix files ( rmfs ) and auxiliary response files ( arfs ) used in spectral fitting .
light curves of brighter sources were also extracted using ciao tools .
we used the 3@xmath36 source ellipses from _ wavdetect _ to define the extraction regions and background was extracted from adjacent source - free regions .
background is negligible , contributing only 0.21 counts per pixel ( 0.3 - 7 kev ) for the total exposure time , or less than one count in the regions used to extract source spectra and light curves .
spectral fitting and timing analysis were undertaken with the heasoft _ xanadu _ software package including xspec vers .
we also applied the kolmogorov - smirnov ( ks ) test ( press et al .
1992 ) to unbinned photon arrival times to check for x - ray variability . since conventional spectral
fitting is only practical for brighter sources , we used the quantile analysis approach of hong et al .
( 2004 ) to quantify source hardness .
this technique has the advantage that it can be applied to sources that are too faint for spectral fitting .
the quantile method uses sorted photon event energy lists to compute the median energy e@xmath37 and the quartile energies e@xmath38 and e@xmath39 .
the quartile energies are defined such that one - fourth of the counts have energies below e@xmath38 , and three - fourths below e@xmath39 .
the quantile fractions q@xmath40 are computed from quartile energies using q@xmath40 = ( e@xmath41 - e@xmath42)/(e@xmath43 - e@xmath42 ) , where e@xmath42 and e@xmath43 are the lower and upper energy bounds of the extracted events ( 0.3 and 7.0 kev in this work ) .
quantile values ( or functions thereof ) are then overplotted on a grid of neutral hydrogen absorption column density ( n@xmath28 ) and plasma energy ( kt ) values generated from spectral simulations based on a one - temperature ( 1 t ) apec optically thin thermal plasma model .
this provides a graphical representation of the source hardness superimposed on a ( n@xmath28 , kt ) grid that is akin to an x - ray color - color diagram .
we use infrared observations of the ngc 2071-ir region obtained by the _
spitzer space telescope _ to aid in x - ray source identification and to determine yso classes using mid - ir colors .
_ spitzer _ s infrared array camera ( irac ) mapped the ngc 2071-ir region as part of _ spitzer _ orion survey ( gto program i d 043 ; megeath et al .
2005a ; allen et al .
2007 ; flaherty & muzerolle 2008 ; megeath et al .
2009 , in prep . )
further information on irac can be found in fazio et al .
the irac data were obtained in four channels ( 3.6 , 4.5 , 5.8 , and 8.0 @xmath5 m ) using the high - dynamic range ( hdr ) mode , acquiring a short 0.6 s exposure plus a longer 12 s exposure at each pointing position .
four dithers were obtained at each position .
the reduction and analysis of the complete data set will be discussed in detail by megeath et al .
( 2009 , in prep . ) . in brief ,
individual basic calibrated data ( bcd ) frames were mosaicked with custom software to create separate mosaic images for the four irac channels .
point sources were identified using photvis v. 1.10 ( gutermuth et al . 2004 ; gutermuth et al .
2008 ) . using the source positions determined from the mosaics ,
photometry was measured for each point source on the individual bcd images using the idl module @xmath44 ( landsman 1993 ) .
magnitudes were measured in a circular aperture of radius r = 2 pixels centered on the source , using a background annulus of r = 2 - 6 pixels . a gain correction depending on pixel position
was applied to each magnitude .
the final magnitudes are the median magnitude returned from the four dithers .
magnitudes of those sources in the ngc 2071-ir region discussed here are given in table 2 , along with magnitude zero points and aperture correction factors .
_ spitzer _ mips data at longer wavelengths 24 - 160 @xmath5 m are also available ( flaherty & muzerolle 2008 ; megeath et al .
2009 , in prep ) . because the ngc 2071-ir region is confused and heavily saturated , we only provide here the mips 24 @xmath5 m photometry for irs 7 .
mips data were reduced using the procedure given in megeath et al .
we focus here on the heavily - reddened ngc 2071-ir subgroup shown in the 2mass 2.16 @xmath5 m ( k@xmath45 ) image in figure 1 and the _ spitzer _ irac 3.6 @xmath5 m ( ch1 ) image in figure 2 .
this region of high source density includes the infrared sources irs 1 - 8 and irs 8a ( w93 ; t07 ) , as well as the radio source vla 1 located near irs 1 .
we summarize the ir and x - ray properties of each source on an individual basis below .
figure 3 shows near - ir colors of irs 1 - 8 , all of which show some excess reddening . figure 4 is an irac color - color diagram that provides one means of distinguishing between class i ysos ( protostars ) and class ii ysos ( classical t tauri stars , or accreting star@xmath21disk systems ) .
figure 5 plots the infrared spectral energy distributions ( seds ) of those sources in ngc 2071-ir for which reliable near - ir and _ spitzer _ mid - ir data are available ( irs 1 , irs 2 , irs 4 , irs 6 , irs 7 ) .
figure 6 shows the broad - band _ chandra _ acis - i image for the central region of ngc 2071-ir near irs 1 , and figure 7 is a hard - band _ chandra _ image zoomed in on irs 1 and vla 1 .
figure 8 is a quantile diagram allowing the x - ray properties of _ chandra _ detections to be compared .
table 1 summarizes the properties of the x - ray sources detected by _
chandra _ , and
_ spitzer _
irac photometry is given in table 2 .
_ chandra _ detected x - ray sources at or near the positions of irs 1,2,4,6 and vla 1 .
the infrared sources irs 5,7 , 8 , and 8a ( offset @xmath62@xmath0 northeast of irs 8) were undetected by _
chandra_. table 3 summarizes x - ray spectral analysis results for the brighter detections .
irs 1 was identified as a bright 10 @xmath5 m source by persson et al .
it is also clearly visible in the 2mass k@xmath45 image ( fig . 1 ) and in _ spitzer _ irac images ( fig .
2 ) . a very bright source at or near the position of irs 1 saturates the mips 24 @xmath5 m images .
irs 1 is likely the dominant contributor to the mid / far - ir luminosity toward the ngc 2071-ir region , which was determined to be l@xmath46 = 520 l@xmath31 ( d = 390 pc ) by butner et al .
this luminosity suggests that irs 1 may be an embedded early - type star , as discussed further in section 5.1 .
but , its true nature is still somewhat of a mystery .
irs 1 is clearly detected in all four irac channels and its sed rises sharply toward longer wavelengths over the 3.6 @xmath5 m - 8.0 @xmath5 m irac range ( fig .
it is the brightest ir source in the ngc 2071-ir core region at 8.0 @xmath5 m ( table 3 ) . _
spitzer _ s angular resolution of fwhm @xmath6 1.7@xmath0 ( 3.6 @xmath5 m ) is unable to fully resolve vla 1 located 2.2@xmath0 north of irs 1 .
but , the irac images do show what appears to be faint emission extending northward of irs 1 , providing a clue that vla 1 may be weakly contributing to the mid - ir flux measured inside the r = 2.@xmath04 aperture centered on irs 1 .
higher angular resolution mid - ir observations will be needed to confirm that vla 1 is indeed a mid - ir source .
the irac colors of irs 1 are characteristic of a class i protostar ( fig .
4 ) . adopting the yso classification scheme of gutermuth et al . ( 2008 ) , class i sources are defined by [ 4.5 ] @xmath47 [ 5.8 ] @xmath48 1 , where brackets denote magnitudes at the enclosed wavelength ( @xmath5 m ) .
irs 1 has [ 4.5 ] @xmath47 [ 5.8 ] @xmath17 1.67 ( table 3 ) , so the class i criterion is satisfied .
a similar conclusion is reached using the yso classification schemes of allen et al .
( 2004 ) , hartmann et al .
( 2005 ) , and megeath et al .
( 2005b ) .
_ chandra _ detects an x - ray source that is offset by only 0.42@xmath0 from the near - ir position of irs 1 (= 2mass [email protected] ) .
this offset is within _ chandra _ positional uncertainties and we thus associate x - ray source cxo [email protected] with the infrared source irs 1 .
the irac 3.6 @xmath5 m detection of irs 1 has a centroid position [email protected] , which is offset by only 0.18@xmath0 from the _ chandra _ x - ray position .
detection confirms that irs 1 is an x - ray source , as suspected on the basis of the previous lower spatial resolution _
xmm - newton _ observation ( s07 ) .
the x - ray light curve of irs 1 ( fig .
9 ) shows low - amplitude fluctuations about the mean at the @xmath82@xmath36 level but no large flares .
the ks test gives a probability of constant count rate p@xmath49 @xmath17 0.11 .
thus , low - level variability may be present but is not proven at high confidence levels .
irs 1 is clearly detected in the hard - band acis - i image ( fig . 7 ) but its emission is not as hard as that of vla 1 or irs 3 as gauged by median photon energy ( table 1 ) . the x - ray spectrum of irs 1 ( fig .
10 ) is absorbed below 1 kev but the absorption is not as high as that of vla 1 .
the acis - i spectrum of irs 1 shows an emission line from the si xiii he - like triplet complex ( 1.839 - 1.865 kev ) , implying thermal emission .
maximum power for the si xiii triplet is emitted at log t@xmath50 = 7.0 ( k ) so hot plasma is clearly detected .
the spectrum also reveals a faint ( 7 counts ) fluorescent emission line at @xmath66.4 kev from neutral or low ionization stages of fe ( fig .
11 ) , as anticipated from _ xmm - newton _ spectra .
this line forms in `` cold '' material in the vicinity of the star that is irradiated by the hard x - ray source and is discussed further in section 5.3 .
we attempted to fit the acis - i spectrum of irs 1 with various emission models .
all models included a fixed - width gaussian line at 6.4 kev to reproduce the faint fluorescent fe line . a simple one - temperature ( 1 t ) apec solar abundance optically thin plasma model underestimates the flux in the 1.7 - 1.9 kev range and is unacceptable in terms of @xmath51 statistics .
this range includes the si xiii he - like triplet lines at 1.839 , 1.854 , and 1.865 kev , which can not be individually distinguished at the spectral resolution of acis - i .
an acceptable 1 t fit can be obtained by allowing si to be overabundant by a factor of @xmath163 with respect to its solar value .
the apparent si overabundance could either be a real physical effect due to factors such as grain destruction ( e.g via grain - grain collisions in shocks ) or an unphysical consequence of fitting the spectrum with an overly simplistic 1 t model .
an acceptable solar abundance fit can be obtained with a two - temperature ( 2 t ) apec model that includes both a cool and hot plasma component ( table 3 ) .
some further improvement in the 2 t fit can be obtained by allowing si to be overabundant , but this is not required since the solar abundance fit is already statistically acceptable ( @xmath52 = 1.01 ) . because of the heavy absorption below 1 kev , the temperature and emission measure of any cool component are quite uncertain . the 2 t apec model summarized in table 3
uses a fixed temperature kt@xmath53 = 0.55 kev for the cool component .
this temperature results in a minimum value of the c - statistic ( cash 1979 ) when fitting unbinned spectra .
however , there is very little change in the c - statistic for values in the range kt@xmath53 = 0.25 - 0.70 kev .
the 2 t model yields a best - fit temperature for the hot plasma component kt@xmath2 = 3.3 kev , but the true temperature could be higher since the 90% confidence upper limit is not tightly constrained by the data . using the conversion n@xmath28
@xmath17 2.22 @xmath1 10@xmath54 a@xmath55 @xmath30 ( gorenstein 1975 ) , the solar abundance 2 t apec model gives a@xmath55 = 16 [ 12 - 20 ] mags , where square brackets enclose the 90% confidence range .
the slightly different conversion n@xmath28 @xmath17 1.6 @xmath1 10@xmath54 a@xmath55 @xmath30 ( vuong et al . 2003 ) yields a@xmath55 = 22 [ 16 - 28 ] mags .
the standard conversions used above are reliable for low - extinction sources , but some studies have questioned their accuracy for deeply - embedded ysos
( winston et al . 2007 ) .
the above apec model assumes that the x - ray emission arises in an optically thin thermal plasma and is typically used to model coronal emission in magnetically active late - type stars . however , other emission processes might be operating in irs 1 . in particular ,
x - rays could be produced in shocks associated with its powerful outflow .
thus , we tried to fit the irs 1 spectrum with an isothermal plane - parallel shock model ( vpshock ) . the fit statistic for vpshock ( @xmath52 = 1.16 )
is not quite as good as the 2 t apec model and the fit converges to shock temperatures kt @xmath6 2 - 3 kev ( table 3 ) .
such shock temperatures are realistic but do require high - velocity flows , as discussed further in section 5.2.1 . in summary ,
a 1 t apec thermal plasma model is able to reproduce the _ chandra _ spectrum of irs 1 provided that si is allowed to be overabundant .
an acceptable solar - abundance fit can be obtained with a 2 t apec model , but it does not tightly constrain the temperature or emission measure of the heavily - absorbed cool component .
this model requires hot plasma at kt@xmath2 @xmath6 3 kev and an absorption column density equivalent to a@xmath56 @xmath6 16 - 22 mags .
vla 1 is a compact 3.6 cm radio continuum source revealed in high - resolution vla a - configuration images obtained by carrasco - gonzlez et al .
it was also marginally detected in previous 5 ghz vla images shown in snell & bally ( 1986 ) .
the 3.6 cm radio source is located 2.2@xmath0 north of irs 1 , which equates to a projected separation of @xmath6860 au at d = 390 pc .
the _ chandra _ x - ray position is in very good agreement with the radio position of vla 1 ( table 1 ) . due to its close proximity to irs 1 ,
vla 1 was not spatially resolved from irs 1 by _ xmm - newton_. we are not aware of any published high - resolution infrared images which show that vla 1 is an infrared source .
but , _ spitzer _ images ( fig .
2 ) reveal faint northward extension from irs 1 that may be due to vla 1 .
the ks test shows that the x - ray emission of vla 1 is likely variable with p@xmath49 @xmath17 0.02 , though the lightcurve shows no large amplitude flares ( fig .
the acis - i spectrum is heavily absorbed with little or no x - ray emission detected below 2 kev .
the spectrum can be acceptably fitted with a 1 t apec model at kt = 2.2 kev ( table 2 ) .
no significant fe line emission is detected in the 6.4 - 6.7 kev range . the inferred absorption column density log
n@xmath28 = 22.94 gives an equivalent extinction a@xmath55 = 39 [ 22 - 69 ] mag using the gorenstein ( 1975 ) conversion .
based on x - ray spectral fits , vla 1 is more heavily absorbed than irs 1 and figure 8 indicates that it is the most heavily absorbed x - ray source in ngc 2071-ir .
vla 1 could thus be in an earlier evolutionary stage than irs 1 . because of the high absorption the intrinsic ( unabsorbed ) x - ray luminosity of vla 1 is quite uncertain .
but , the value inferred from the 1 t apec model ( table 2 ) is consistent with that of a @xmath16solar - mass pre - main sequence star or protostar , based on the known correlation between l@xmath25 and stellar mass in low - mass ysos ( preibisch et al . 2005
; telleschi et al .
2007a ) .
previous studies have concluded that irs 2 is likely a young stellar object ( w93 ; t07 ) and aspin et al .
( 1992 ) argued that it is driving a bipolar outflow .
irac images show some indication that the source is extended .
irac colors ( fig .
4 ) indicate that irs 2 is either a class i protostar or a heavily - reddened class ii object .
however , some caution is warranted in interpreting irac colors .
high - resolution 3.6 cm radio continuum observations reveal two closely - spaced components a and b separated by @xmath61.@xmath04 with the brighter component a at radio position vla [email protected] ( carrasco - gonzlez et al .
2009 , private comm . ) .
_ spitzer _ can not resolve such a close pair and it is thus possible that the irac colors are the superposition of two sources .
_ chandra _ detects a very faint x - ray source ( 4 net counts ) at an offset of only 0.27@xmath0 from the brighter radio component a ( table 1 ) . because of the good positional coincidence with the radio source , this weak x - ray detection is likely real but is of marginal significance .
there are too few counts to do any substantive x - ray analysis but the high median photon energy e@xmath57 = 5.2 kev suggests that this source is exceptionally hard or very heavily absorbed .
the infrared source irs 3 is about one magnitude fainter at k - band and at 3.6 @xmath5 m and 4.5 @xmath5 m than irs 1
. its k band emission is highly polarized ( w 93 ; t07 ) .
a 1.3 cm radio continuum source lies near irs 3 ( torrelles et al . 1998 ) , but the near - ir peak is offset slightly to the northeast relative to the _ vla _ radio position ( t07 ) . as table 1 shows , the weak x - ray source detected by _
chandra _ is in excellent positional agreement with the 1.3 cm radio source detected by torrelles et al .
the nearest 2mass source lies 2.3@xmath0 from the x - ray peak .
this offset is significant and supports the conclusion of t07 that the near - ir emission is nebulosity .
the object detected with _
chandra _ and the _ vla _ is likely the star that illuminates the near - ir nebula . _
spitzer _ irac images show an infrared source near irs 3 ( fig .
the irac position as measured in the 4.5 @xmath5 m ( ch2 ) image is offset by only 0.42@xmath0 from _
chandra_. thus , it appears that the self - luminous source is also detected by _
spitzer_. we do not have complete irac photometry for irs 3 because it lies in the bright wings of irs 1 .
a ks test shows no significant x - ray variability . the distribution of event photon energies ( fig .
12 ) shows clearly that this is a hard absorbed source , and the energy quantile plot ( fig .
8) substantiates this .
there were too few counts to analyze a binned spectrum but we did attempt to fit the unbinned spectrum with a 1 t apec thermal plasma model using c - statistics ( cash 1979 ) .
the absorption was held fixed at log n@xmath28 = 22.79 @xmath30 , based on the estimated extinction a@xmath55 = 28 mag ( w93 ) .
the fit converges to a high but uncertain temperature kt @xmath58 10 kev , as also indicated by the quartile plot ( fig . 8)
. visual inspection of the unbinned spectrum shows a weak buildup of counts near 6.4 kev , hinting that a weak fluorescent fe line may be present .
the irs 3 event list reveals five events within the 6 - 7 kev range with event energies 6.11 , 6.20 , 6.42 , 6.53 , and 6.69 kev .
the 6.42 and 6.53 kev events could be due to fluorescent fe .
we added a gaussian line component to the 1 t apec model at a line energy of 6.4 kev , fixing the line width at fwhm = 120 ev , corresponding to the acis - i instrumental width at this energy ( _ chandra _ pog , fig .
6.8 ) . with
the gaussian line included , the c - statitistic c = 133.1 is slightly less than without the line ( c = 138.2 ) , and the fit gives log n@xmath28 = 22.79 @xmath30 , kt = 10 kev , and an unabsorbed flux f@xmath59 ( 0.5 - 7.5 kev ) = 1.6 @xmath1 10@xmath60 ergs @xmath30 s@xmath27 .
this equates to log l@xmath25 = 29.5 ergs s@xmath27 ( d = 390 pc ) , which is typical of low - mass ysos in orion ( preibisch et al .
thus , the event list and spectral fits provide some marginal evidence for fluorescent fe , but a deeper exposure would be needed to make a compelling case for a 6.4 kev line detection . the high - resolution _ vla _ 1.3 cm radio continuum images of irs 3 obtained by torrelles et al .
( 1998 ) show elongation in the @xmath6n - s direction , and elongated morphology is also visible in the more recent 3.6 cm _
vla _ images of carrasco - gonzlez et al .
the extension was interpreted as a thermal radio jet , adding support to the belief that irs 3 is driving an outflow ( sec .
1 ) . the orientation of h@xmath2o maser spots approximately perpendicular to the jet axis and maser velocity data led torrelles et al .
( 1998 ) to conclude that the masers trace a compact rotating molecular disk surrounding a low - mass ( @xmath161 m@xmath31 ) yso .
such a disk viewed in a nearly edge - on geometry would act as a strong absorber of soft x - ray photons and would provide a natural explanation for the high n@xmath28 ( fig . 8) and lack of detected low - energy x - ray photons below @xmath162 kev in irs 3 ( fig .
12 ) .
tamura et al . ( 2007 ) concluded that irs 4 is an ir cluster member because it is surrounded by a compact reflection nebula and is associated with faint h@xmath2 emission . the faint x - ray source detected by _
chandra _ is in good positional agreement with 2mass and shows no significant variability .
it has the lowest median and mean photon energy of the six x - ray sources detected in the cluster core .
there are insufficient counts available for spectral analysis .
about half of the detected photons have energies below 2 kev .
quantile analysis shows that this source has lower absorption than the other x - ray detections in ngc 2071-ir ( fig .
near - ir colors ( fig .
3 ) are consistent with a star having intrinsic ctts color but viewed through subsantial reddening ( a@xmath56 @xmath6 13 mag ) .
irac colors ( fig .
4 ) are consistent with a reddened class ii yso but the sed ( fig .
5 ) is nearly flat in the irac bands and is more suggestive of a flat - spectrum source .
the near - ir morphology of this infrared source changes with time and w93 concluded that the near - ir emission could be from a jet , a moving clump of shocked gas , or nebulosity .
irs 5 is located between the two nearby bright mid - ir sources irs 1 and irs 2 .
we were not able to identify a definite point source at the irs 5 position in the irac images , and thus can not provide any reliable irac photometry .
no significant x - ray emission was detected so we find no evidence of a young star and our results are thus consistent with the non - stellar classification proposed by w93 .
near - ir observations have shown that irs 6 is a binary with a 2@xmath0 separation ( t07 ) .
we detect a variable x - ray source that is offset to the northwest of the 2mass position by 1.68@xmath0 at position angle pa = 315@xmath22 ( table 1 ) .
this offset is significant and implies that the 2mass near - ir source is not the same source detected in x - rays . _
spitzer _ images show an irac source at an offset of 1.19@xmath0 from the _ chandra _ position .
irs 6 has similar near - ir and mid - ir colors to irs 4 , and both the near - ir and mid - ir colors of irs 6 are consistent with a reddened class ii object ( ctts ) .
the spectral fit of irs 6 with a 1 t apec thermal model ( table 2 ) suggests a rather hot source with plasma temperature kt = 3.85 kev and moderate absorption n@xmath28 @xmath16 10@xmath29 @xmath30 .
similar values are inferred from the quantile plot ( fig .
8) , but both kt and n@xmath28 have rather large uncertainties due to the low number of x - ray counts .
the absorption from the irs 6 spectral fit ( table 2 ) gives an equivalent extinction a@xmath55 = 5.6 [ 1.7 - 10.4 , 90% conf .
] mag ( gorenstein 1975 conversion ) , which is less than irs 1 and vla 1 . on the basis of its relatively high kt and variability ,
we conclude that the x - ray source lying 1.@xmath068 from the 2mass position is a magnetically - active star .
because this system is a close near - ir pair , higher spatial resolution observations are needed to confirm the ir colors and more accurately determine the relative positions of the two near - ir sources and the x - ray source .
tamura et al .
( 2007 ) argued on the basis of polarization angle that circumstellar material is present around irs 7 .
irac colors support this conclusion , indicating either a class i protostar or a heavily - reddened class ii source ( fig .
the object is visible in mips 24 @xmath5 m images and the sed continues to rise out to 24 @xmath5 m ( fig .
its [ 4.5 ] @xmath47 [ 24 ] color is consistent with a protostar ( megeath et al .
we detect no x - ray emission at or near irs 7 , so it is either intrinsically x - ray faint or its emission is reduced to undetectable levels by high absorption ( e.g. a nearly edge - on disk or dense surrounding envelope ) .
irs 8 and 8a are two faint closely - spaced ir sources separated by 1.6@xmath0 in ra and 0.@xmath07 in dec , with irs 8a lying northeast of irs 8 ( w93 ; t07 ) .
both objects are @xmath481.5 mag fainter at k than the other core sources ( t07 ) .
the closest 2mass positional match to irs 8 has k@xmath45 = 13.54 ( 2mass [email protected] ) and there are no other 2mass sources within 5@xmath0 .
irs 8 is associated with a near - ir nebula and was classified as a self - luminous object , probably an embedded star , by w93 .
irs 8a is thought to be a shocked molecular hydrogen peak ( w93 ) .
the irac images reveal what appears to be extended structure or nebulosity in the ne - sw direction near irs 8 .
a faint emission peak is visible close to the irs 8 position at 3.6 @xmath5 m ( table 2 ) and the peak shifts slightly to the northeast in the 4.5 @xmath5 m image , which is sensitive to shock emission .
the close pair irs 8/8a is not clearly resolved at irac s resolution , but the slight shift of the 4.5 @xmath5 m emission peak toward irs 8a may be an indication that it is a stronger shock - excited source . because of the nebulous morphology in irac images , we were not able to obtain any reliable irac photometry .
there is no x - ray emission at or near the 2mass position of irs 8 or at irs 8a , and consequently no x - ray evidence of an embedded star .
the _ xmm - newton _ epic ccd spectrum of irs 1 analyzed by s07 was extracted using a nominal circular region of radius r@xmath61 = 15@xmath0 ( 68% encircled energy ) centered on irs 1 . as the higher angular resolution _
image in figure 6 shows , four x - ray sources were included in the _ xmm - newton _ spectral extraction region , namely irs 1,2,3 and vla 1 . of these four ,
irs 1 ( 144 counts ) and vla 1 ( 90 counts ) are the brightest ngc 2071-ir core objects in the _ chandra _ image .
irs 2 ( 4 counts ) is very faint in acis - i and it is unlikely that it significantly affected the _ xmm - newton _ spectrum .
irs 3 ( 18 counts ) is also a faint hard _
chandra _ source .
its absorbed flux f@xmath25 ( 0.5 - 7.5 kev ) = 8 @xmath1 10@xmath62 ergs @xmath30 s@xmath27 would have contributed @xmath65% to the observed flux measured in the _ xmm - newton _ pn spectrum .
based on the above , it is clear that the _ xmm - newton _ spectrum was dominated by emission from the close pair irs 1 and vla 1 .
chandra _ data show that vla 1 has much higher x - ray absorption than irs 1 , which explains why fits of the _ xmm - newton _ pn spectrum mysteriously required two spectral components with different n@xmath28 values ( s07 ) .
the fluorescent fe line at 6.4 kev was strongly detected in the _ xmm _ pn spectrum with an epic pn line flux f@xmath25 ( 6.2 - 6.6 kev ) = 3.8 ( @xmath80.2 ) @xmath1 10@xmath60 ergs @xmath30 s@xmath27 and an equivalent width ( ew ) measured relative to the underlying continuum ew = 2.4 kev ( s07 ) . the line flux measured in the _
chandra _ acis - i spectrum ( table 3 ) is a factor of @xmath165 below the _ xmm _ value , but the _ chandra _ ew @xmath63 2.4 kev is consistent with the _ xmm _ value .
the fluorescent fe line is discussed further in section 5.3 .
the heavily - reddened irac colors of irs 1 and its rising sed in the irac bands , along with its bright ( saturated ) detection in mips 24 @xmath5 m images , strongly suggest that it is a class i protostar . based on its association with a radio continuum source and luminosity considerations ,
w93 concluded that irs 1 is most likely a b0-b5 star . under the assumption that the 6 cm radio continuum emission of irs 1 arises in a photoionized compact h ii region at a distance of 500 pc , snell & bally ( 1986 ) found that a single b2 star would be needed to produce sufficient lyman continuum ionizing flux to account for the radio emission .
adjusting their distance downward to 390 pc ( anthony - twarog 1982 ) would require a @xmath16b3 star .
however , we note that the volume emission measure of the radio continuum source computed from their 6 cm radio flux density s@xmath64 = 7.9 mjy is two orders of magnitude below that typically found for optically thin compact hii regions ( eq . [ 3 ] of skinner , brown , & stewart 1993 ) .
thus , the radio continuum emission from irs 1 could very well involve processes other than free - free emission from a compact hii region .
the more recent high - resolution 3.6 cm _
vla _ a - configuration images of irs 1 obtained by carrasco - gonzlez et al .
( 2007 ) reveal a complex source morphology for irs 1 .
a compact core is present along with fainter emission extending in the e - w direction .
more interestingly , compact knots are seen expanding away from irs 1 in an approximate e - w direction with high velocities of up to @xmath16600 km s@xmath27 , as determined from radio proper motions .
these results , along with the positive radio spectral index , suggest that the radio emission includes a contribution from an extended ionized wind or jet .
a possible analog is ngc 7538-irs1 , whose free - free emission is dominated by a highly collimated ionized jet ( sandell et al .
2009 ) . using far - ir data obtained with the _ kuiper airborne observatory _ ( kao ) ,
butner et al . ( 1990 ) determined the total luminosity toward the ngc 2071-ir region to be l@xmath46 = 520 l@xmath31 ( d = 390 pc ) .
this is about 15% greater than the value determined by harvey et al .
( 1979 ) when normalized to a common distance of 390 pc .
because of the rather large @xmath630@xmath0 kao beam size , this total will include any contributions from other sources near irs 1 and should thus be considered an upper limit for irs 1 itself . if due to a single star , a value l@xmath46 = 520 l@xmath31 would correspond to a zams star of mass @xmath165 m@xmath31 ( siess , dufour , & forestini 2000 ) and zams spectral type @xmath16b4 ( thompson 1984 ) .
but , if the l@xmath46 value is interpreted as an upper limit for irs 1 , a lower mass and later spectral type are possible .
the above estimates are based on an assumed distance d = 390 pc .
if irs 1 lies at the far edge of the orion ob1 association then a larger distance d @xmath16 500 pc is possible ( sec .
1 ) . in that case , one infers a slightly higher mass of @xmath166 m@xmath31 and zams spectral type of @xmath16b3 - b4 .
the x - ray luminosity and temperature of irs 1 are within the broad range observed in young b - type stars .
the study of 20 early - type stars in the _ chandra _ orion coup sample by stelzer et al .
( 2005 ) included 12 b stars ranging in spectral type from b0.5 to b9 .
their x - ray luminosities spanned a wide range from log l@xmath25 = 29.4 - 32.4 ergs s@xmath27 , and a hot plasma component at typical temperatures kt@xmath65 @xmath6 1 - 3 kev was present in all b - stars .
different b stars with similar spectral types showed dramatic differences in l@xmath25 and no support was found for earlier claims that l@xmath25 correlates with l@xmath66 in ob stars .
using l@xmath66 = 520 l@xmath31 for irs 1 and the range of l@xmath25 values from the different models in table 3 , we obtain log ( l@xmath25/l@xmath66 ) = @xmath475.7 @xmath80.5 .
this is consistent with values obtained for the coup b - star sample , but some caution is needed in interpreting this ratio for irs 1 since its l@xmath66 is could be dominated by disk emission .
on the basis of the above comparisons , we conclude that a b - star classification for irs 1 is plausible , but a mid - to - late b star is more likely than an early b star based on the luminosity determined by butner et al .
if irs 1 is indeed an embedded intermediate - mass b - star , it is tempting to speculate that it might be a herbig ae / be star progenitor .
ae / be stars are optically - revealed intermediate - mass pre - main sequence stars which in many cases show high - temperature x - ray plasma that is likely of magnetic origin .
but , the question of whether their x - ray emission is intrinsic or instead due to coronal late - type companions remains open . for previous studies of x - ray emission from herbig ae /
be stars , the reader is referred to damiani et al .
1994 ; zinnecker & preibisch 1994 ; preibisch & zinnecker 1996 ; skinner & yamauchi 1996 ; skinner et al . 2004
; hamaguchi , yamauchi , & koyama 2005 ; stelzer et al . 2006 , 2009 ; and telleschi et al .
2007b .
given that the true nature of irs 1 is not well - known and that it may be an embedded b - type star , some further discussion of the origin of its x - ray emission is warranted . very few massive embedded stars have been detected in x - rays and little is known about their x - ray emission during early evolutionary stages prior to becoming optically visible .
we thus consider several possible emission mechanisms below .
the hotter plasma clearly detected at kt@xmath65 @xmath58 2 kev is not consistent with the cool emission ( kt @xmath67 1 kev ) expected from wind shocks in ob stars formed by line - driven flow instabilities ( lucy & white 1980 ; lucy 1982 ; feldmeier et al .
1997 ; owocki et al .
however , if the wind is shocking onto another object or surrounding material , or is magnetically - confined , then high - temperature plasma could be produced in a wind shock .
the predicted shock temperature for a strong adiabatic shock is t@xmath68 = 1.4 @xmath1 10@xmath69 @xmath70v@xmath71 k , or kt@xmath68 = 0.012@xmath70v@xmath71 kev ( lamers & cassinelli 1999 ) . here
, @xmath70v@xmath72 is the shock jump measured relative to the speed of the downstream flow , in units of 100 km s@xmath27 . to achieve an x - ray temperature kt@xmath68 @xmath16 2 kev , a value @xmath70v @xmath16
1300 km s@xmath27 is required .
terminal wind speeds of b5 - b9 stars are not well - determined observationally , but values v@xmath73 @xmath16 1200 - 1400 km s@xmath27 are usually assumed ( cohen et al .
1997 ) . to achieve the required velocity jump
, the wind would have to be virtually stopped by dense surrounding material , e.g. either a disk , thick envelope , or close companion . the wind speed requirement could be reduced if the outflowing wind collides with a dense infalling envelope whose ram pressure exceeds that of the wind . in that case , the infalling envelope would overpower the wind , resulting in a non - stationary shock front ( including both forward and reverse shocks ) moving toward the star .
the above analysis can also be applied to jets and molecular outflows .
high velocity wings extending out to @xmath670 km s@xmath27 are visible in spectra of the 2.12 @xmath5 m h@xmath2 line toward ngc 2071-ir ( persson et al .
if such an outflow shocks onto a stationary target , the maximum expected x - ray temperature is kt@xmath68 = 0.006 kev , much too low to explain the hot x - ray plasma detected in irs 1 . as already noted ( sec .
5.1 ) , high - velocity knots moving outward from irs 1 at speeds up to @xmath16600 km s@xmath27 have been detected with the _ vla _ ( carrasco - gonzlez et al . 2007 ) .
if such a knot shocked onto a stationary object , the maximum shock temperature would be kt@xmath68 @xmath6 0.4 kev , which is again too low to explain the higher temperature plasma in irs 1 .
but , this process could contribute to any cooler heavily - absorbed emission at kt @xmath32 1 kev . since irs 1 is a young class
i object , it is expected to still be accreting .
this is potentially relevant to the x - ray interpretation because x - rays can be produced in an accretion shock formed as infalling material impacts the ( proto)star .
the characteristic x - ray temperature for an accretion shock is kt@xmath74
@xmath6 0.02v@xmath75 kev , where v@xmath76 is the infall speed in units of 100 km s@xmath27 ( ulrich 1976 ) . based on the above
, it is apparent that very high infall speeds v @xmath16 1000 km s@xmath27 would be needed to produce accretion shock temperatures kt@xmath74 @xmath16 2 kev , comparable to the values observed for irs 1 . to achieve such infall speeds under free - fall conditions , a massive central object equivalent to a b0v star ( m@xmath77 @xmath16 18 m@xmath31 , r@xmath77 @xmath16 7 r@xmath31 ; allen 1985 ) would be required .
but , the total luminosity of irs 1 ( sec .
3.1 ) falls well short of the value l@xmath66
@xmath16 10@xmath78 l@xmath31 expected for a b0 zams star . thus , in the absence of any firm observational evidence for very high infall speeds toward irs 1
, it is very unlikely that the high - temperature x - ray emission arises in an accretion shock .
any cooler plasma ( kt @xmath32 1 kev ) that might be present could be accretion - related but , as already noted , the high absorption toward irs 1 masks such cool emission . plasma temperatures
kt @xmath58 2 kev as seen in irs 1 are commonly found in magnetically - active late - type stars , including t tauri stars .
the x - ray emission is accompanied by powerful magnetic reconnection flares , and is thought to arise in structures analogous to the solar corona .
but , magnetic star@xmath21disk coupling undoubtedly leads to more complicated geometries in the case of tts ( reviewed by feigelson & montmerle 1999 ) .
similarly , high plasma temperatures and large flares characteristic of magnetospheric processes have now been observed in low - mass class i protostars such as those in @xmath79 ophiuchus ( imanishi , koyama , & tsuboi 2001 ) .
x - ray luminosities in low - mass ysos generally increase with stellar mass and luminosity , ranging from l@xmath25 @xmath16 10@xmath80 ergs s@xmath27 for low masses m@xmath77 = 0.1 - 0.2 m@xmath31 up to ( and in some cases above ) l@xmath25 @xmath16 10@xmath81 ergs s@xmath27 for more massive tts with m@xmath77 @xmath6 2 m@xmath31 ( preibisch et al . 2005 ; telleschi et al .
2007a ) .
the x - ray temperature and luminosity of irs 1 inferred from the _ chandra _ data are well within the above ranges for low - mass ysos .
it is thus possible that the x - rays observed toward irs 1 arise in a low - mass yso rather than a more massive object such as an embedded b - type star .
the x - ray luminosity of irs 1 could be accounted for by a tts of mass @xmath161 - 2 m@xmath31 .
even if irs 1 is an embedded b - type star , a late - type tts companion could dominate the x - ray emission .
late - type companions have been proposed as a means of explaining the wide range of l@xmath25 values observed for some mid - to - late b stars in orion with weak winds ( stelzer et al .
2005 ) if irs 1 has an ionized wind and a sufficiently strong magnetic field , then the b field can confine the wind into two oppositely directed streams in each hemisphere which collide near the magnetic equator , forming a magnetically - confined wind shock ( mcws )
. such a shock can reach x - ray emitting temperatures .
the mcws mechanism has been used to explain high - temperature x - ray emission in magnetic ap - bp stars ( babel & montmerle 1997a ) and young magnetic o stars such as @xmath82 ori c ( babel & montmerle 1997b ; gagn et al .
2005 ) .
the radio properties of of irs 1 do suggest it has an ionized wind or jet ( carrasco - gonzalez et al .
2007 ) , but there are so far no reports of a magnetic field .
even so , an estimate of the field strength needed to confine the wind can be obtained by making reasonable approximations under the assumption that it is a mid - to - late b star .
the maximum x - ray temperature for a mcws is given by the adiabatic shock equation ( sec .
we thus assume a terminal wind speed v@xmath73 @xmath16 1300 km s@xmath27 for irs 1 to achieve a shock temperature comparable to the observed value ( kt@xmath68 @xmath16 2 kev ) .
the degree to which the wind is confined by the magnetic field is expressed in terms of the confinement parameter @xmath83 = b@xmath84r@xmath85/@xmath86v@xmath73 where b is the equatorial magnetic field strength ( ud - doula & owocki 2002 ) . for values of @xmath83
@xmath6 1 ( critical confinement ) , the b field is able to confine the wind . for a b5v star ( r@xmath77 @xmath6 3.8 r@xmath31 , allen 1985 ) , the above relation with @xmath83 = 1 becomes b@xmath71 @xmath6 47@xmath83@xmath87 , where b@xmath72 is in units of 100 g and @xmath87 is in units of 10@xmath88 m@xmath31 yr@xmath27 .
the mass - loss rates of mid - to - late b stars are not well - determined empirically , but b @xmath16 2 g for @xmath86 = 10@xmath89 m@xmath31 yr@xmath27 ( cohen et al .
1997 ) and b @xmath16 70 g for @xmath86 = 10@xmath90 m@xmath31 yr@xmath27 .
the latter @xmath86 values are typical of intermediate mass herbig ae / be stars ( skinner et al . 1993 ) . in summary ,
the mcws mechanism is a potential means of explaining the high - temperature plasma in irs 1 .
but , without a magnetic field detection and specific estimates of b - field strength and mass - loss parameters , any more definitive comparisons against mcws model predictions can not be made .
based on the above discussion , we conclude that the high - temperature plasma at kt @xmath16 2 kev in irs 1 is most likely due to either a high - speed wind ( v@xmath73 @xmath58 1200 km s@xmath27 ) shocking onto a dense target ( e.g. a disk or dense surrounding envelope ) , or it originates in a low - mass magnetically - active yso companion .
the wind - shock interpretation is plausible if irs 1 is an embedded b - type star whose wind has already turned on , and the extended bright radio continuum emission detected with the _ vla _ does point to an early - type star with a strong wind or jet .
we can not rule out a magnetically - confined wind shock , but there are insufficient observational constraints on magnetic field strength and wind parameters to rigorously test mcws models .
even if irs 1 is an embedded b - type star ( which seems likely ) , a lower mass yso companion at close separation could still be responsible for some ( or even all ) of the observed x - ray emission .
the x - ray luminosity determined from spectral models of irs 1 is at the high end of the range observed for low - mass ysos .
if the x - ray emission is due entirely to a t tauri - like companion , then the known correlation between l@xmath25 and stellar mass suggests a companion mass in excess of 1 m@xmath31 .
a fluorescent fe line at 6.4 kev can be produced by photoionization when `` cold '' material containing neutral or weakly - ionized iron is irradiated by a nearby hard x - ray continuum source .
photons with energies above 7.11 kev are needed to eject a k - shell electron .
the k - shell vacancy is filled by another electron such as an l - shell electron , resulting in fluorescent fe line emission at 6.4 kev .
the line is in fact a doublet consisting of two closely - spaced lines at 6.391 kev and 6.404 kev , but these can not be distinguished at the energy resolution of the acis - i ccds .
the above photoionization process is discussed in more detail by george & fabian ( 1991 ) and kallman et al .
analysis of the fluorescent fe line can potentially provide information on the properties of the absorber and its location relative to the hard continuum source .
the fluorescent fe line detected in irs 1 is faint ( fig .
11 ) , consisting of 7 events distributed in the 6.298 - 6.536 kev energy range , with a median energy 6.395 kev and mean energy 6.419 kev .
three of the seven photons arrived in the first half of the observation .
a ks test applied to the arrival times of the 7 events shows no significant variability and gives a probability of constant count rate p@xmath91 = 0.35 .
the line flux is 7.2 ( @xmath81.2 ) @xmath1 10@xmath62 ergs @xmath30 s@xmath27 .
this is about a factor of five less than measured during the march 2005 _ xmm _ observation ( s07 ) .
the line equivalent width in the acis - i spectrum is uncertain because the continuum flux near 6.4 kev is very faint .
but , our flux estimates give values ew @xmath63 2.4 kev , consistent with the previous _ xmm _ estimate ( s07 ) .
the detection of fluorescent fe line emission in irs 1 is a rare occurrence in ysos , but certainly not unique .
tsujimoto et al .
( 2005 ) examined 1616 x - ray sources detected in the deep _ chandra _ coup observation of the orion nebula cluster and identified seven sources with excess emission at 6.4 kev and line equivalent widths ew @xmath26 0.27 kev .
all seven x - ray sources showed flare - like variability and high line - of - sight absorption n@xmath28 @xmath58 10@xmath29 @xmath30 , as also seen in irs 1 .
the broad fluorescent fe line during a superhot flare in v1486 ori ( coup source 331 ) was analyzed by czesla & schmitt ( 2007 ) .
the 6.4 kev line has also been detected in other ysos including the class i objects ylw 16a ( imanishi et al .
2001 ) and elias 29 ( giardino et al .
2007 ) in @xmath79 oph , and the class i source r cra x@xmath92 (= irs 7b ; hamaguchi et al .
2005 ) .
the fluorescent fe line detection in irs 1 is interesting in three respects : ( i ) it was detected in the apparent absence of any large - amplitude flares ( but low - level variability may be present ) , ( ii ) there is no accompanying line at 6.7 kev from the highly - ionized fe xxv complex , and ( iii ) the line has a very large equivalent width that is difficult to account for by illuminated disk models .
the absence of any detectable large flares in irs 1 is unusual because the fluorescent fe lines in almost all of the ysos mentioned above occurred during flares .
analysis of the v1486 ori data showed that its broad fluorescent fe line appeared during the rise phase of a flare ( czesla & schmitt 2007 ) . an interesting exception is elias 29 , which showed a 6.4 kev line during a large flare .
but the line persisted after the flare had decayed , remaining over a timescale much longer than the radiative lifetimes of fluorescent fe transitions .
giardino et al .
( 2007 ) thus concluded that the x - ray flare was not directly responsible the post - flare 6.4 kev line .
instead , they proposed that the line was produced by collisional ionization from an unseen population of nonthermal electrons .
the 6.4 kev line in irs 1 is similar to elias 29 in the sense that it does not appear to be directly associated with a large flare .
the line was present in both the _ xmm - newton _ spectrum obtained in march 2005 and in the more recent _ chandra _ spectrum , but no large flare was detected in either observation .
if a large flare was involved , then it must have occurred prior to the start of the observations , or it escaped detection ( as could occur if the flare occurred on the back side of irs 1 and was occulted ) . unless large x - ray flares occurred and escaped detection , the existing data indicate that the 6.4 kev line in irs 1 is persistent and does not correlate with large flares . in that case , some mechanism needs to be operating nearly continuously to ionize the absorbing material . at present
, the ionizing source and the ionization mechanism are not known .
the obvious candidate for the ionizing source is irs 1 itself .
but , in terms of hardness , its _ chandra _ spectrum is rather benign , showing no large flares , no fe xxv line and a weak continuum above @xmath167 kev .
however , we can not rule out the possibility of substantial hard emission ( kt @xmath58 20 kev ) above the energy range accessible to _ chandra _ and _ xmm_. nevertheless , the intriguing possibility does remain that the hard ionizing source is not irs 1 and is not directly detected in our x - ray observations .
however , before discounting irs 1 as the ionizing source , two factors should be noted .
first , low - level variations appear to be present in the _ chandra _ light curve of irs 1 ( fig .
these variations may be a signature of persistent low - level flaring .
if that is the case , then quasi - continuous hard energy release by repeated small flares may play a role in photoionizing the absorber .
second , irs 1 is thought to drive a powerful outflow and also shows evidence of a radio jet ( carrasco - gonzlez et al .
if a population of nonthermal electrons is produced in the shocked jet ( which may be magnetically collimated ) they could play a role in collisional ionization of the absorber .
collisional ionization has been invoked as a means of explaining the 6.4 kev line in elias 29 ( giardino et al .
2007 ) and in the active binary system ii peg ( osten et al .
if a population of nonthermal paricles is present , they _ may _ imprint a signature on the radio continuum emission .
thus , further radio observations of irs 1 at lower frequencies where nonthermal emission can dominate over thermal ( free - free ) emission could be informative .
the large equivalent width ew @xmath63 2.4 kev of the irs 1 fluorescent line is a noteworthy feature of its x - ray spectrum .
this value is extreme for a yso , exceeding even the maximum value ew = 1.4 kev observed for v1486 ori during the rise phase of its superhot flare ( czesla & schmitt 2007 ) .
these values conflict with theoretical models of centrally illuminated cold disks , which predict ew @xmath6 0.1 - 0.2 kev ( george & fabian 1991 ) .
the ew of the 6.4 kev line is proportional to the column density n@xmath93 in the fluorescing material ( tsujimoto et al .
if taken literally , the large ew of irs 1 would imply a high absorber column density n@xmath93 @xmath58 10@xmath94 @xmath30 ( s07 ) .
since this value is much larger than the line - of - sight photoelectric absorption inferred from the x - ray spectrum ( table 3 ) , one is led to the conclusion that the fluorescing material lies off the line - of - sight .
but , the above conclusion rests on the assumption that the continuum measured in the x - ray spectrum has undergone negligible absorption .
if the hard irradiating source is not irs 1 itself and is occulted or very heavily absorbed , then some or all of the intrinsic hard continuum will be missing from the observed x - ray spectrum . in that case , the ew measurement would be based on a depressed continuum and thus overestimated , along with n@xmath93 . in summary ,
our main conclusions with respect to the fluorescent fe line in irs 1 are : ( i ) the line is present even in the absence of the fe xxv line ( 6.67 kev ) and any large flare detections , and evidently does not directly depend on large - amplitude x - ray flares for its existence ,
( ii ) the line flux is variable , showing a decrease between the _ xmm _ and _ chandra _ observations taken 2.6 years apart , but no significant variability during the 67 ksec _ chandra _ observation , ( iii ) the line ew is much larger than predicted by centrally - illuminated disk models , suggesting either that the photoionization models do not accurately reflect the line formation process and that other mechanisms ( e.g. collisional ionization ) could play a role , or that the observed continuum is not the same as the intrinsic continuum of the photoionizing source ( e.g. due to the effects of absorption or occultation ) , and ( iv ) the hard ionizing source may have so far escaped x - ray detection .
the main conclusions of this study are the following : 1 .
_ chandra _ s excellent spatial resolution has resolved the ngc 2071-ir core region into six x - ray sources .
x - ray emission was detected at or near the infrared sources irs 1,2,3,4 and 6 and the radio source vla 1 .
x - ray analysis shows that these sources are viewed through moderate - to - high absorption and have hard / high - temperature x - ray emission ( kt @xmath58 2 kev ) , consistent with their classification as embedded young stars . in some cases such as irs 3 ,
the x - ray position is significantly offset from the near - ir position , supporting the conclusion that the near - ir peak traces nebulosity or scattered light and not the star itself .
spitzer _ irac detections of irs 1,2,4,6 , and 7 provide the first reliable mid - ir colors for these objects .
using irac colors along with near - ir colors based on published jhk@xmath45 photometry and the sharply - rising ir sed , we conclude that irs 1 is a class i protostar .
irs 2 and irs 7 are also likely class i protostars , but could be heavily - reddened class ii objects .
irs 4 and irs 6 have colors consistent with reddened class ii objects .
no reliable irac photometry or mid - ir colors were obtained for irs 3 , 5 , 8 , or 8a , but near - ir colors determined from existing jhk@xmath45 photometry show that these objects are heavily - reddened .
chandra _ provides the first x - ray detection of vla 1 , and its x - ray emission is likely variable .
x - ray spectral models require a strong absorption component for vla 1 , equivalent to a@xmath56 = 39 [ 22 - 69 ] mag .
thus , vla 1 appears to be more heavily - embedded than irs 1 and it could thus be an even younger , less - evolved object .
there are as yet no ir data for vla 1 on which to base a yso classification .
the existing data for irs 1 are consistent with a mid - to - late b ( proto)star classification .
the x - ray spectrum of irs 1 is heavily - absorbed below 1 kev and clearly reveals high - temperature plasma at kt @xmath16 2 - 3 kev .
such plasma could originate either in a shocked high - velocity wind / jet or in a magnetically - active low - mass yso companion .
the most important finding of this study is that the fluorescent fe line in irs 1 is persistent and present even in the absence of large flares .
thus , an as yet unidentified quasi - continuous process that operates independently of large flares is needed to explain the k - shell ionization of cold proximate material that leads to formation of fluorescent fe . both the _ chandra _ and _ xmm - newton _ light curves show signs of low - level variability , suggesting that short bursts of hard emission from persistent low - level flaring could play a role in ionizing the absorber .
the large equivalent width ew @xmath63 2.4 kev of the 6.4 kev line is noteworthy because it is well in excess of values predicted by centrally illuminated disks models .
large 6.4 kev line equivalent widths have also been noted in other ysos such as v1486 ori .
if interpreted literally , the large ew values imply either a very high absorption column density in the cold absorber , or perhaps processes other than photoionization ( e.g. collisional ionization in shocks ) .
however , some caution is needed in interpreting the large ew for irs 1 , since the absence of large flares raises the possibility that it is not the hard ionizing source .
if the ionizing source is occulted by irs 1 or buried behind heavy absorption , then the hard continuum measured in the irs 1 spectrum may be less than the intrinsic continuum of the ionizing source ( `` missing continuum '' ) , leading to an overestimate of the equivalent width .
, k.f . , and m.m .
acknowledge _ chandra _ support from sao grant go7 - 8008a .
m.a . acknowledges support from a swiss national science foundation professorship ( pp002110504 ) .
we have utilized data products from the two micron all - sky survey ( 2mass ) , which is a joint project of the university of massachusetts and ipac / caltech .
this work is based in part on data obtained with the _ spitzer _ space telescope , which is operated by the jet propulsion laboratory , caltech , under a contract with nasa .
we would like to thank c. carrasco - gonzlez and g. sandell for radio positions and j. hong for information on quartile analysis computer codes .
allen , c.w .
1985 , astrophysical quantities ( london : athlone ) allen , l.e .
2004 , , 154 , 363 allen , l.e .
2007 , in protostars and planets v , eds .
b. reipurth , d. jewitt , & k. keil ( tucson : u. of arizona press ) , 361 anders , e. & grevesse , n. 1989 , geochim .
acta , 53 , 197 anthony - twarog , b. 1982 , , 87 , 1213 aspin , c. , sandell , g. , & walther , d.m .
1992 , , 258 , 684 babel , j. & montmerle , t. 1997a , , 323 , 121 babel , j. & montmerle , t. 1997b , , 485 , l29 bally , j. 1982 , , 261 , 558 bessell , m.a . &
brett , j.m .
1988 , , 100 , 1134 brandt , w.n .
2001 , , 122 , 2810 brown , a.g.a .
, de geus , e.j . , &
de zeeuw , p.t .
1994 , , 289 , 101 butner , h.m . ,
evans , n.j . ,
harvey , p.m. , mundy , l , g . ,
natta , a. , & randich , m.s .
1990 , , 364 , 164 carpenter , j.m .
2001 , , 121 , 2851 cash , w. 1979 , , 228 , 939 carrasco - gonzlez , c. , anglada , g. , rodrguez , l.f . ,
gmez , j.f .
, torrelles , j.m . ,
& osorio , m. 2007 , in massive star formation : observations confront theory ( heidelberg , germany ) , http://www.mpia.de/msf07/ cohen , d.h . , cassinelli , j.p . , & macfarlane , j.j .
1997 , , 487 , 867 cohen , j.g . , persson , s.e . , elias , j.h . , & frogel , j.a .
1981 , , 249 , 481 czesla , s. & schmitt , j.h.m.m .
2007 , , 470 , l13 damiani , f. , micela , g. , sciortino , s. , & harnden , f.r .
1994 , , 436 , 807 eislffel , j. 2000 , , 354 , 236 fazio , g.g .
2004 , , 154 , 10 feigelson , e.d . &
montmerle , t. 1999 , , 37 , 363 feldmeier , a. , kudritzki .
, palsa , r. pauldrach , a.w.a . , & puls , j. 1997 , , 320 , 899 flaherty , k.m .
, & muzerolle , j. 2008 , , 135 , 966 gagn , m. et al .
2005 , , 628 , 986 george , i.m . &
fabian , a.c .
1991 , , 249 , 352 getman , k.v .
2005 , , 160 , 319 giardino , g. , favata , f. , pillitteri , i. , flaccomio , e. , micela , g. , & sciortino , s. 2007 , , 475 , 891 gibb , a.g .
2008 , in handbook of star forming regions , volume 1 , the northern sky , ed .
b. reipurth ( san francisco : asp ) , 693 gorenstein , p. 1975
, , 198 , 95 gutermuth , r.a .
2004 , , 154 , 374 gutermuth , r.a .
2008 , , 674 , 336 hamaguchi , k. et al .
2005 , , 623 , 291 hamaguchi , k. , yamauchi , s. , & koyama , k. 2005 , , 618 , 360 hartmann , l. et al .
2005 , , 629 , 881 harvey , p.m. , campbell , m.f . ,
hoffmann , w.f . ,
thronson , h.a .
, & gatley , i. 1979 , , 229 , 990 hong , j. , schlegel , e.m . , & grindlay , j.e .
2004 , , 614 , 508 imanishi , k. , koyama , k. , & tsuboi , y. 2001 , , 557 , 747 johansson , l.e.b . , hglund , b , winnberg , a. , rieu , n.q .
, & goss , w.m .
1974 , , 189 , 455 kallman , t.r . ,
palmeri , p. , bautista , m.a . , mendoza , c. , & krolik , j.h .
2004 , , 155 , 675 lada , e.a .
, depoy , d.l . ,
evans , n.j . , & gatley , i. 1991 , , 371 , 171 lamers , h.j.g.l.m . & cassinelli , j.p .
1999 , introduction to stellar winds ( cambridge : cambridge u. press ) , 360 landsman , w.b .
1993 , in astronomical data analysis and software systems ii , a.s.p .
series vol .
hanisch , r.j.v .
brissenden , & j. barnes , 246 lucy , l.b .
1982 , , 255 , 286 lucy , l.b .
& white , r.l .
1980 , , 241 , 300 lynds , b.t .
1962 , , 7 , 1 maddalena , r.j . , morris , m. , moscowitz , j. , & thaddeus , p. 1986 , , 303 , 375 megeath , s.t .
et al . 2005a , in massive star birth : a crossroads of astrophysics , proc .
227 , eds .
r. cesaroni , m. felli , e. churchwell , & m. walmsley ( cambridge : cambridge u. press ) , 383 megeath , s.t . ,
allgaier , e. , young , e. , allen , t. , pipher , j.l . , & wilson , t.l .
2009 , , 137 , 4072 megeath , s.t . ,
hartmann , l. , luhman , k.l .
, & fazio , g.g .
2005b , , 634 , l113 meyer , m.r . , calvet , n. , & hillenbrand , l.a .
1997 , , 114 , 288 moretti , a. , campana , s. , lazzati , d. , & tagliaferri , g. 2003 , , 588 , 696 osten , r.a .
, drake , s. , tueller , j. , cummings , j. , perri , m. , moretti , a. , & covino , s. 2007 , , 654 , 1052 owocki , s.p . , castor , j.i . , & rybicki , g.b .
1988 , , 335 , 914 pankonin , v. , winnberg , a. , & booth , r.s . 1977 , , 58 , l25 persson , s.e . ,
geballe , t.r . , simon , t. , lonsdale , c.j .
, & baas , f. 1981 , , 251 , l85 preibisch , t. et al .
2005 , , 160 , 401 ( p05 ) preibisch , t. & zinnecker , h. 1996 , in _ rntgenstrahlung from the universe _ , ed .
h. zimmerman & j. trmper ( garching : mpe ) , 17 press , w.h . ,
teukolsky , s.a . , vetterling , w.t . , & flannery , b.p .
1992 , numerical recipes in fortran : the art of scientific computing ( 2nd ed . ; new york : cambridge univ .
press ) , 617 sandell , g. , nyman , l.a . , haschick , a. , & winnberg , a. 1985 , in lecture notes in physics , 237 , nearby molecular clouds , eds .
araki et al .
( berlin : springer - verlag , 234 sandell , g.h.l . , wright , m. , goss , w. , & corder , s. 2009 , baas , 41(no .
1 ) , 460 siess , l. , dufour , e. , & forestini , m. 2000 , , 358 , 593 skinner , s.l . , brown , a. , & stewart , r.t .
1993 , , 87 , 217 skinner , s.l . ,
gdel , m. , audard , m. , & smith , k. 2004 , , 614 , 221 skinner , s.l . ,
simmons , a.e . ,
audard , m. , & gdel , m. 2007 , , 658 , 1144 ( s07 ) skinner , s.l . &
yamauchi , s. 1996 , , 471 , 987 snell , r.l . &
bally , j. 1986 , , 303 , 683 snell , r.l . ,
scoville , n.z . , sanders , d.b . , &
erikson , n.r .
1984 , , 284 , 176 stelzer , b. et al .
2005 , , 160 , 557 stelzer , b. , micela , g. , hamaguchi , k. , & schmitt , j.h.m.m .
2006 , , 457 , 223 stelzer , b. , robrade , j. , schmitt , j.h.m.m .
, & bouvier , j. 2009 , , 493 , 1109 stojimirovi , i. , snell , r. l. , & narayanan , g. 2008 , , 679 , 557 strom , k.m . , strom , s.e . , carrasco , l. , & vrba , f.j .
1975 , , 196 strom , k.m . , strom , s.e . , & vrba , f.j .
1976 , , 81 , 308 tamura , m. et al .
, 2007 , , 59 , 467 ( t07 ) telleschi , a. , gdel , m. , briggs , k.r . , audard , m. , & palla , f. 2007a , , 468 , 425 telleschi , a. , gdel , m. , briggs , k.r . , skinner , s.l . , audard , m. , & franciosini , e. 2007b , , 468 , 541 thompson , r.i . 1984 , , 283 , 165 torrelles , j.m . , gmez , j.f . ,
rodrguez , l.f . ,
curiel , s. , anglada , g. , & ho , p.t.p .
1998 , , 505 , 756 tsujimoto , m. et al .
2005 , , 160 , 503 ud - doula , a. & owocki , s.p .
2002 , , 576 , 413 ulrich , r.k .
1976 , , 210 , 377 vuong , m.h . ,
montmerle , t. , grosso , n. , feigelson , e.d . ,
verstraete , l. , & ozawa , h. 2003 , , 408 , 581 walther , d.m . , geballe , t.r . , & robson , e.i .
1991 , , 377 , 246 walther , d.m . , robson , e.i . , aspin , c. , & dent , w.r.f .
1993 , , 418 , 310 ( w93 ) wilson , b.a . , dame , t.m . ,
masheder , m.r.w . , & thaddeus , p. 2005
, , 430 , 523 winston , e. et al .
2007 , , 669 , 493 zhao , b. et al .
1999 , , 118 , 1347 zhou , s. , evans , n.j . , & mundy , l.g .
1990 , , 355 , 159 zinnecker , h. & preibisch , t. , 1994 , , 292 , 152 lllllllll irs 3 & 05 47 04.63 & + 00 21 48.0 & 18@xmath84 & 3.68,4.52,6.06 & 0.76 & 12.18 & vla j054704.62 + 002147.8 ( 0.25 ) + irs 1 & 05 47 04.75 & + 00 21 42.9 & 144@xmath812 & 1.79,2.22,3.27 & 0.11 & 11.21 & 2 m j05470477 + 0021428 ( 0.42 ) + vla 1 & 05 47 04.77 & + 00 21 45.1 & 90@xmath810(v ) & 2.98,3.52,4.48 & 0.02 & ... & vla j054704.758 + 002145.50 ( 0.44 ) + irs 4 & 05 47 05.10 & + 00 22 01.6 & 11@xmath83 & 1.18,1.75,2.53 & 0.14 & 10.86 & 2 m j05470512 + 0022013
( 0.48 ) + irs 2 & 05 47 05.35 & + 00 21 50.4 & 4@xmath81 & ... , 5.22 , ... & ... & 10.41 & 2 m j05470538 + 0021500 ( 0.66 ) + irs 6 & 05 47 05.60 & + 00 22 11.1 & 48@xmath87(v ) & 1.56,2.05,3.21 & 0.01 & 10.10 & 2 m j05470570 + 0022103 ( 1.68 ) + lllll irs 1 & 8.359 @xmath8 0.005 & 6.246 @xmath8 0.005 & 4.576 @xmath8 0.003 & 2.819 @xmath8 0.001 + irs 2 & 7.747 @xmath8 0.046 & 6.382 @xmath8 0.008 & 5.581 @xmath8 0.005 & 4.717 @xmath8 0.005 + irs 3 & ... & 7.250 @xmath8 0.015 & ... & ... + irs 4 & 9.418 @xmath8 0.035 & 8.263 @xmath8 0.007 & 7.776 @xmath8 0.005 & 6.678 @xmath8 0.007 + irs 5 & ... & ... & ... & ... + irs 6 & 8.609 @xmath8 0.023 & 8.052 @xmath8 0.003 & 7.619
@xmath8 0.003 & 6.966 @xmath8 0.006 + irs 7 & 8.546 @xmath8 0.051 & 6.705 @xmath8 0.003 & 5.776 @xmath8 0.002 & 4.654 @xmath8 0.002 + irs 8 & @xmath4811.35 & ... & ... & ... + lllll object & irs 1 & irs 1 & vla 1 & irs 6 model & 2 t apec & vpshock & 1 t apec & 1 t apec n@xmath28 ( 10@xmath29 @xmath30 ) & 3.62 [ 2.75 - 4.52 ] & 3.03 [ 2.20 - 4.52 ] & 8.68 [ 4.88 - 15.3 ] & 1.24 [ 0.37 - 2.31 ] kt@xmath53 ( kev ) & \{0.55 } & 2.31 [ 1.31 - 5.99 ] & 2.22 [ 1.00 - 7.18 ] & 3.85 [ 1.47 - ... ] norm@xmath53 ( 10@xmath95 ) & 2.60 [ 0.38 - 5.03 ] & 0.46 [ 0.23 - 0.82 ] & 1.18 [ 0.95 - 1.39 ] & 0.31 [ 0.17 - 0.65 ] kt@xmath2 ( kev ) & 3.31 [ 1.51 - 21.4 ] & ... & ... & ... norm@xmath2 ( 10@xmath24 ) & 2.82 [ 1.07 - 5.31 ] & ... & ... & ... e@xmath96 ( kev ) & \{6.40 } & \{6.40 } & ... & ...
@xmath97 ( kev ) & \{0.05 } & \{0.05 } & ... & ... norm@xmath96 ( 10@xmath98 ) & 6.89 [ 1.49 - 12.0 ] & 7.10 [ 2.43 - 12.5 ] & ... & ... abundances & solar & solar & solar & solar @xmath51/dof & 8.10/8 & 9.3/8 & 2.13/5 & 1.26/3 @xmath52 & 1.01 & 1.16 & 0.43 & 0.42 f@xmath25 ( 10@xmath60 ergs @xmath30 s@xmath27 ) & 2.99 ( 66.8 ) & 2.84 ( 24.2 ) & 2.54 ( 15.7 ) & 2.65 ( 4.71 ) f@xmath99 ( 10@xmath62 ergs @xmath30 s@xmath27 ) & 7.01 ( 7.01 ) & 8.04 ( 8.04 ) & ... & ... log l@xmath25 ( ergs s@xmath27 ) & 31.08 & 30.64 & 30.45 & 29.93 -band ( 2.16 @xmath5 m ) image of the region near irs 1 .
crosses mark positions of chandra x - ray sources . significant offsets between chandra and 2mass positions exist for irs 3 ( 2.@xmath03 ) and irs 6 ( 1.@xmath07 ) .
see table 1 for positions .
no x - ray emission was detected by chandra at the 2mass positions of irs 5 , irs 7 , and irs 8 .
log intensity scale .
north is up , east is left.,width=566 ] | we present results of a sensitive _ chandra _ x - ray observation and _ spitzer _ mid - ir observations of the infrared cluster lying north of the ngc 2071 reflection nebula in the orion b molecular cloud .
we focus on the dense cluster core known as ngc 2071-ir which contains at least nine ir sources within a 40@xmath0 @xmath1 40@xmath0 region .
this region shows clear signs of active star formation including powerful molecular outflows , herbig - haro objects , and both oh and h@xmath2o masers . we use _ spitzer _ irac images to aid in x - ray source identification and to determine yso classes using mid - ir colors . _
spitzer _ irac colors show that the luminous source irs 1 is a class i protostar .
irs 1 is believed to be driving a powerful bipolar molecular outflow and may be an embedded b - type star or its progenitor .
its x - ray spectrum reveals a fluorescent fe emission line at 6.4 kev , arising in cold material near the protostar .
the line is present even in the absence of large flares , raising questions about the nature of the ionizing mechanism responsible for producing the 6.4 kev fluorescent line .
_ chandra _ also detects x - ray sources at or near the positions of irs 2 , irs 3 , irs 4 , and irs 6 and a variable x - ray source coincident with the radio source vla 1 , located just 2@xmath0 north of irs 1 .
no ir data are yet available to determine a yso classification for vla 1 , but its high x - ray absorption shows that it is even more deeply - embedded than irs 1 , suggesting that it could be an even younger , less - evolved protostar . |
the ejecta from supernova ( sn ) explosions enrich the interstellar medium of the galaxy with heavy elements over thousands of cubic parsecs .
identification of the ejecta products of sne is most easily done via analysis of the x - ray emission from young to middle - aged supernova remnants ( snrs ) , after the reverse shock has propagated back into the ejecta and heated it to x - ray emitting temperatures .
one of the biggest open questions in sn research is what types of stellar progenitor systems lead to what types of sne and snrs . in only a handful of cases
have astronomers directly observed the pre - sn star in archival images , and this situation is unlikely to change in the near future .
the fundamental problem is that sne are plentiful but distant and point - like , while snrs are local and resolvable but limited in number .
snr [email protected] ( also known as msh 10 - 53 ; hereafter g284 ) has been identified through radio @xcite , optical @xcite , and x - ray @xcite observations , and is perhaps @xmath13 years old @xcite with a distance of 5.4@xmath14 kpc @xcite .
it is interacting with a molecular cloud in the north .
an analysis of the extended x - ray emission seen in g284 was first reported by @xcite , who noted that x - ray emission was present in the northern portion of the remnant , just behind a filament identifiable in the radio and optical images . at the center of the remnant
lies the bright x - ray point source 1fgl j1018.6@xmath15856 ( hereafter j1018 ) . identified as a high - mass binary by @xcite , j1018 reveals a 16.6 d periodic modulation in the _ fermi _ lat observations @xcite .
this is one of the few binaries seen in @xmath0-rays @xcite .
@xcite reported results from the first _
chandra _ and _ xmm - newton _ observations of j1018 around binary phases @xmath15 and @xmath16 , respectively ( zero phase corresponds to the maximum of the @xmath17 gev flux , mjd @xmath18 ) , and also found the high - mass companion in 2mass data .
_ suzaku _ @xcite and _ swift _ @xcite observed j1018 covering the entire orbital period , confirming the periodicity seen in @xmath0-rays .
the most recent period measurement , @xmath19 d , was provided by @xcite , who found a large x - ray flux increase at @xmath20 and an additional maximum at @xmath21 .
j1018 exhibits a correlation between x - ray flux and photon index , common to several @xmath0-ray binaries .
j1018 is one of two high - mass @xmath0-ray binaries seen inside of an snr ( along with ss 433 in w50 ) , offering a rare opportunity to study the remains of the explosion that would produce such a system .
the spectral type of the companion star places tighter constraints on the exploded star than would otherwise be possible .
ground - based observations revealed a massive star of spectral type o6v((f ) ) ( @xmath22 ; @xcite ) and mass of @xmath23 @xmath7 , very similar to that of the @xmath0-ray binary ls 5039 .
j1018 was observed with the h.e.s.s .
observatory by @xcite , who report a @xmath24 detection of the source hess j1018@xmath1589a ( spatially coincident with j1018 ) .
the detected @xmath0-ray variability , both in gev and tev , confirms the association between the h.e.s.s .
source and j1018 .
@xcite , based on the shape of the tev phaseogram and source spectrum , suggest that j1018 is a low - inclination , low - eccentricity system . in this _
letter _ , we report our analysis of the x - ray emission from j1018 and g284 .
we use newly - obtained _ chandra _ data , supplemented with archival _ xmm - newton _ data , to examine the extended x - ray emission from the snr and do spectral and spatial analysis of the binary .
we use two sets of _ chandra _ observations ( table 1 ) , totaling 72 ks , taken with the acis - i instrument in timed exposure mode and telemetered in `` very faint '' format .
we processed the data using the _ chandra _ interactive analysis of observations ( ciao ) software ( ver . 4.6 ) and _ chandra _ calibration data base ( caldb ) , ver .
spectra were analyzed using _
xspec _ ( ver .
12.8.2 ) , which contains the latest atomic data from _ atomdb _ 3.0 .
we simulate the _ chandra _ psf using chart ( chandra ray trace ) to study the morphology of the emission in the immediate vicinity of j1018 .
the chart output was supplied to marx v.5.1 software to obtain the final image .
we used a blurring ( aspectblur parameter in marx ) value of @xmath25 , as recommended by the _
help desk .
lccr 14657 & 2013 - 12 - 24 & @xmath26 & 44.40(0.031 ) + 16560 & 2013 - 12 - 29 & @xmath27 & 27.68(0.019 ) + . for our analysis of the g284 emission , discussed below in section [ snr ]
, we supplement our _ chandra _ observations with archival _ xmm - newton _ observations to increase the signal - to - noise ratio .
g284 was observed by _ xmm - newton _ on 2013 jan 9 for a total of 105 ks ( obsid 0694390101 ; pi : de luca ) .
epic - mos observations were taken in full frame mode with a medium filter , while epic - pn data were taken in small window mode with a thin filter .
the pn data captures only emission from the binary ( analyzed by @xcite ) , not the extended emission .
we processed the mos data with xmm science analysis software ( sas ) ver .
13.5 . to filter the data for flares
, we used the _ mos - filter _ routine as part of the esas software , available within sas .
after time filtering , approximately 79 and 91 ks of `` good '' data remained for the mos 1 and 2 detectors , respectively . guided by the _ chandra _
images , we selected two regions of extended emission to analyze , labeled `` north '' and `` west . '' both are shown in figure [ images ] .
these regions have areas of @xmath28 and @xmath29 arcmin@xmath30 , respectively .
background subtraction for these two regions is difficult due to the large size of the remnant compared with the _ chandra _ field of view .
however , based on a comparison of the _ chandra _ and _ xmm _ images , we found off - source regions that appear on both the acis - i and mos arrays and do not appear to have any associated emission from the snr .
we extracted spectra for j1018 from the two _ chandra _ observations .
the spectra suffer from moderate pile - up ( 3,294 and 1,498 counts in the @xmath31 aperture correspond to 0.24 and 0.17 counts per frame in obsid 14657 and 16560 , respectively ) . to correct for pile - up
, we used the davis ( 2001 ) model with the grade migration parameter @xmath32 .
fits with an absorbed power - law model ( with the xspec ` phabs ` model ) give hydrogen column densities @xmath33 and ( @xmath34 @xmath35 ; photon indices @xmath36 and @xmath37 ; absorbed ( unabsorbed ) fluxes @xmath38 ( @xmath39 ) and @xmath40 @xmath41 erg s@xmath10 @xmath35 ; and reduced @xmath42 and 0.75 for obsids 14657 and 16560 , respectively . all uncertainties are @xmath43
. the derived @xmath44 and @xmath45 do not change significantly between the individual fits .
the flux in the second observation is a factor of 1.7 smaller than the first one , in accordance with the orbital phase dependance of the flux . comparing our fit for the first _
observation with the an et al .
( 2015 ) fit for the _ xmm - newton _ observation taken at a close binary phase ( @xmath46 , @xmath47 ) , we find that our @xmath44 is somewhat larger and the slope is somewhat steeper , but the values are consistent at the @xmath5 level . because of the lack of pile - up , the _ xmm - newton _ measurements are probably more accurate .
we also investigate the spatial morphology of j1018 in the acis - i images and compare the surface brightness radial profile with a modeled psf and pile - up parameter @xmath32 . ]
( figure [ surfb ] ) .
overall , the observed profile matches the simulated psf .
there appears to be a mismatch between the simulated profile and the observed profile at @xmath48 , but this is likely due to an inacurate pile - up modeling in marx .
in addition , the psf model is not very reliable at small distances due to the presence of the asymmetric psf anomaly around @xmath49 , which is not simulated by marx . including the pile - up correction in the marx 5.1 simulation does not change the profile substantially beyond @xmath50 , which is expected for moderate pileup .
analyzing the high - resolution images of j1018 , we found an enhancement ( 24 photons in 0.58 kev within the ellipse shown in figure 1 , for the merged image ) located @xmath51 away from the best - fit centroid of the source at a position angle ( pa ) of @xmath52 ( counted east of north ) .
the enhancement , visible in both observations , has @xmath53 significance estimated from comparison with the background ( which includes psf wings ) measured in the @xmath54 annulus for the merged image .
the expected enhancement due to the known mirror artifact has a very similar pa , but is located significantly closer to the source center ( @xmath55 ) and is unlikely the cause of the enhancement . its appearance is similar to the fast - moving extended feature we recently discovered in the vicinity of b125963 @xmath0-ray binary ( pavlov et al .
deeper followup _ chandra _ observations are needed to study this extended emission and its tentative connection to j1018 .
although the optical counterpart of j1018 has been found and the binary nature of this system is firmly established , it remains unknown whether the compact object is a neutron star ( ns ) or a black hole ( bh ) .
we use the known binary properties ( orbital period and mass of the donor star ) to place constraints on the compact object . while there is still some uncertainty about the exact range of masses of progenitors that end their lives as nss and those that become bhs
, most models predict that , for solar or lower metallicities , the transition between nss and bhs occurs for stars with initial masses somewhere in the range of @xmath56 m@xmath57 ( e.g. , @xcite ) . for metallicities much larger than solar , stellar evolution models predict that massive stars develop substantial winds , which cause enough mass loss that the end product is a ns rather than a bh @xcite . for our simulation we assume solar metallicity as the system is in the galactic disk . we first use the single star evolution ( sse ) code @xcite to study the evolution of stars with solar metallicity
we find that stars with initial mass of 20 m@xmath57 and higher would end up as bhs , in agreement with the above .
if j1018 started and evolved as a detached system ( no mass transfer ever happens ) , the progenitor star must have burned its hydrogen faster than the @xmath58 star and hence it should have been even more massive .
therefore , in this scenario the compact object should be a bh .
however , the lack of mass transfer would require a much larger than the currently observed ( @xmath59 d ) initial orbital period , which could only increase ( since the binary is assumed wide and there is no interaction to cause spiraling in ) after the progenitor explodes as supernova .
thus we conclude that the system must have started as a tight binary and undergone mass transfer . however , evolution of massive stars in tight binaries is more complicated because mass transfer can proceed from one star to the other , rendering the sse code inapplicable .
therefore , we used binary stellar evolution ( bse ) models @xcite .
we ran a grid of 80,000 simulations with the following parameters : @xmath60 m@xmath57 ( current donor star ) ,
@xmath61 m@xmath57 ( sn / compact object progenitor with initial mass always greater than m1 ) , binary period @xmath62 d , eccentricity @xmath63 .
we find 115 systems that end up with @xmath64 d , and @xmath65 m@xmath57 right after the sn explosion of star 2 .
a large fraction of these simulated binaries are disrupted immediately after the sn , but we are not interested in such systems .
our findings show that in order to reproduce the observed binary parameters we need the following initial conditions : @xmath66 m@xmath57 , @xmath67 m@xmath57 , @xmath68 d. the simulations resulted in a heavy ( @xmath69 ) ns as the compact object in 95% of the cases , while the o6v donor star gained mass ( from @xmath70 m@xmath57 to 31 m@xmath57 ) via mass transfer .
note that the systems that match the above criteria have high eccentricity ( @xmath71 ) , while @xcite claim ( based on the tev observations ) that the eccentricity of j1018 should be closer to 0 .
the northern region analyzed in @xcite roughly corresponds to our `` north '' region .
we extend their analysis , supplementing it with additional _ chandra _ and deeper _ xmm - newton _ data ( their analysis used a 20 ks _ xmm _ observation ) , while expanding the analysis to our `` west '' region as well .
while we do not show it here , we have also found that the northern filament , identifiable in radio , h@xmath72 , and x - rays , is present in the wise 22 @xmath73 m ir data .
ir emission from the forward shock in snrs is common , even in more evolved snrs @xcite , and generally results from dust heated in the post - shock environment .
we extracted spectra from the north region using only the mos 2 and _ chandra _ observations , since this region falls on one of the dead mos 1 chips .
for the west region , data from mos 1 , mos 2 , and _
chandra _ are used .
after background subtraction , slight residuals remained in the mos 1 & 2 spectra from instrumental lines of al and si k@xmath72 at 1.49 and 1.75 kev , respectively .
we excised small regions around both of these lines ( 1.45 - 1.53 kev and 1.71 - 1.79 kev ) from the mos 1 & 2 spectra , but kept those excised regions in the _ chandra _ spectra .
for each region , we perform joint fits to all available data , with all parameters tied together .
the column density , is taken from the fits to the x - ray spectrum of the point source , and is frozen in the fits to 8@xmath74 @xmath35 , the same value as @xcite used .
we binned the data to a minimum of 50 counts per spectral bin for all snr datasets .
we show the spectra from the north and west regions in figure [ spectra ] .
we fit these spectra with an absorbed plane shock model with variable abundances ( ` phabs*vpshock ` ) , assuming cosmic abundances from @xcite .
the temperature and ionization timescale ( @xmath75 @xmath76 @xmath77 @xmath78 ) were allowed to float freely , as were abundances of ne , mg , and fe .
si was free as well , but was tied together with s due to the poor statistics of the s line .
we kept o frozen at the solar value for two reasons : with the fairly high absorption towards the remnant , o lines are only barely visible , if present at all , and we wanted the abundances of the other elements to be normalized to some fiducial value ( the abundance of o ) .
for the north region , we find normal cosmic abundances for ne and mg , with subsolar abundances for si ( @xmath79s ) and fe .
subsolar abundances are often seen in the x - ray spectra of middle - aged snrs such as the cygnus loop @xcite .
the relatively long ionization timescale and lower temperature leads us to concur with the analysis of @xcite that this emission is dominated by shocked ism .
the ionization timescale is fit as @xmath80 @xmath81 s , but is unbound on the upper end ( lower - bound of 1.9 @xmath82 ) .
an ionization timescale this long makes the spectrum almost indistinguishable from a plasma in collisional ionization equilibrium , again , not unexpected for the shocked ism from an older remnant .
the spectra of the west region show an entirely different emission profile .
they are relatively well - fit ( reduced @xmath83 of 1.4 for 184 d.o.f . ) by the same absorbed plane - shock model with variable abundances with the parameters shown in table [ params ] . immediately obvious
is the strong mg line at 1.35 kev .
ne and fe are both consistent within errors with solar abundances , while si ( and s ) may be slightly overabundant .
however , the overabundance of mg is about a factor of 4.5 higher with respect to o. this overabundance , combined with the higher temperature ( @xmath84 kev ) and much lower ionization timescale ( @xmath85 @xmath81 s ) implies that the emission in this region is dominated by mg - rich ejecta , heated by the reverse shock .
this conclusion is not sensitive to the choice of model ; a regular nei model with variable abundances ( model ` vnei ` in xspec ) returns similar values .
the spectral fits for this region are similar to those reported in @xcite for the lmc remnant n49b .
their analysis is similar to ours , albeit with lmc abundances , different from those in the galaxy .
they find an enrichment of mg and si ( though their si abundance is higher than ours ) , while ne and fe are not enhanced .
they measure a mg / o ratio of 4.2 from a small portion in the interior of n49b , within errors of our measurement of 4.5 in g284 .
as pointed out by @xcite , the high mg / o ratio is surprising , as nucleosynthetic models predict that significant quantities of mg produced in core - collapse sne should be accompanied by significant quantities of o @xcite .
while examples of significant o without mg are plentiful , it is unexpected to find significant mg without o. as @xcite show , deriving a total mass of mg ejected by the sn from x - ray spectral analysis is very difficult , with estimates varying by more than an order of magnitude depending on the assumptions made .
such an estimate is beyond the scope of this work .
nonetheless , the high mg abundances in the ejecta suggest a substantial amount was synthesized in the explosion . in the nucleosynthesis models of @xcite
, significant amounts of mg are only produced in massive progenitors of @xmath12 @xmath7 .
while the northern region of g284 is dominated by shocked ism , we have shown that the ejecta - dominated western region is significantly overabundant in mg . along with n49b in the lmc , this makes only the second known remnant with mg - rich ejecta .
models of nucleosynthesis in ccsn progenitors produce significant amounts of mg relative to other products , like si and fe , only in the explosions of massive stars ( @xmath86 ) .
we find no statistically significant extended emission in the radial profile for @xmath87 from the binary j1018 at the center of g284 , though we do see a slight excess of photons a few arcseconds southeast of the source .
the spectral parameters of j1018 are about the same in the two observations shifted by 0.25 in binary phase , but the flux difference is significant , @xmath88@xmath89 .
our binary evolution calculations support the idea that the progenitor for this sn had an initial mass of @xmath90 25 @xmath7 .
mass transfer prior to the explosion from the progenitor to the current donor star increased that star s mass to @xmath91 .
our simulations imply that the compact object in the binary is a heavy , @xmath92 , ns . with its large size and low surface brightness , g284 would benefit from a deep study with an x - ray telescope with both large effective area and field of view . a long observation with the pn instrument on _ xmm - newton _ in full frame mode or the soft x - ray imager ( sxi ) on the upcoming _ astro - h
_ satellite would be ideal for the purpose of further study on the fainter sections of the ejecta .
we thank the anonymous referee for useful comments .
we acknowledge useful discussions with steve snowden on the _ xmm - newton _ extended source analysis software ( esas ) , a part of the sas software distributed by the _ xmm - newton _ guest observer facility .
we acknowledge _ chandra _ grant g03 - 14047 . north & 0.67@xmath93 & 1.19@xmath94 & 1.06@xmath95 & 0.19@xmath96 & 0.24@xmath97 & 4.6@xmath98 @xmath82 & 9.6@xmath99 @xmath100 & 3.1 & 1.96 ( 112 ) + west & 0.92@xmath101 & 1.30@xmath102 & 4.53@xmath103 & 1.50@xmath104 & 0.97@xmath105 & 1.0@xmath106 @xmath107 & 1.6@xmath108 @xmath100 & 2.5 & 1.40 ( 184 ) + ) .
small black circle shows the best - fit centroid position .
the black sectors show the location of the known _ chandra _ mirror artifact .
a slight excess is seen within the white ellipse , see text for details .
north is up , east is to the left . _
right : _ observed surface brightness radial profile of j1018 ( black ) and modeled psf ( red ) with account for pile - up , see section 3.1.1 for details .
[ surfb],title="fig:",width=264 ] ) .
small black circle shows the best - fit centroid position .
the black sectors show the location of the known _ chandra _ mirror artifact .
a slight excess is seen within the white ellipse , see text for details .
north is up , east is to the left . _
right : _ observed surface brightness radial profile of j1018 ( black ) and modeled psf ( red ) with account for pile - up , see section 3.1.1 for details .
[ surfb],title="fig:",width=359 ] | we present results from two _ chandra _ observations of the 16.6 d x - ray/@xmath0-ray high - mass binary 1fgl j1018.6@xmath15856 located at the center of the supernova remnant [email protected] .
the binary spectra , separated by 0.25 in binary phase , are fit with an absorbed power - law model with @xmath2 for both observations ( the flux during the second observation is a factor of 1.7 smaller ) . in the high - resolution acis - i image we found a hint of extended emission @xmath3@xmath4 southeast of the binary , significant at the @xmath5 level .
binary evolution codes reproduce the system s observed properties with two massive stars with an initial 18-day period , undergoing mass transfer and leaving behind a heavy @xmath6 @xmath7 neutron star . the initial mass of the progenitor star in this scenario is 27 @xmath8 @xmath7 .
_ chandra _ and _ xmm - newton _ images of the remnant show it has a relatively low x - ray surface brightness .
the two brightest regions of extended x - ray emission , with luminosities @xmath9 erg s@xmath10 for @xmath11 kpc , lie in the northern and western portions and show significantly different spectra .
the northern patch is consistent with shocked ism , with a low temperature and long ionization timescale .
however , the western patch is dominated by ejecta , and shows significantly enhanced mg content relative to other ejecta products .
the abundance ratios inferred resemble those from the large magellanic cloud remnant n49b . to our knowledge , this is only the second case of such mg - rich ejecta found in an snr .
nucleosynthesis models for core - collapse sne predict mg - rich ejecta from very massive progenitors of @xmath12 @xmath7 . |
the topic of quantum criticality continuously raises scientific interest because it is believed to be at the heart of the physics of emergent phenomena such as unconventional superconductivity , hidden order and the breakdown of the fermi liquid model.@xcite the phase transition from the disordered paramagnetic to ordered ferromagnetic state ( fm ) in zero magnetic field is a prototypical example of a critical point , i.e. , a second - order phase transition .
the inherently low curie temperatures of metallic fms makes them ideal systems to investigate magnetic quantum criticality , because the curie temperature @xmath5 is often easily suppressed to zero temperature using an external tuning parameter such as pressure @xmath6 or chemical composition @xmath7 , resulting in a fm quantum critical point ( qcp ) .
however , detailed experimental work on archetypal fm metals such as mnsi,@xcite uge@xmath8@xcite , and zrzn@xmath8@xcite , has demonstrated that the situation is more complex than this simple model would suggest .
for these compounds , the fm phase is suppressed to zero temperature at a first - order transition ; i.e. , no quantum critical behavior is observed .
this can be understood theoretically in various ways : e.g. , ( 1 ) additional fermionic modes may couple to the fm critical fluctuations and are expected to generically drive the phase transition to first - order,@xcite or ( 2 ) magnetoelastic coupling may cause a phase transition to become first order .
@xcite on the other hand , each of these systems exhibit additional complicating factors that make it difficult to derive the universal behavior of fm quantum phase transitions ( qpts ) ; mnsi is a long - period helimagnet ( therefore only locally fm ) in which the thermal phase transition is weakly first - order,@xcite uge@xmath8 is a strongly uniaxial fm , and zrzn@xmath8 exhibits a marginal fermi liquid ground state.@xcite these difficulties are further underlined by uru@xmath9re@xmath1si@xmath8 , which is believed to exhibit a fm qcp.@xcite uru@xmath9re@xmath1si@xmath8 has its own set of peculiarities , including unconventional critical scaling of the magnetization , non - fermi - liquid behavior not only in vicinity of the qcp but also within the fm phase , and @xmath10 scaling of the magnetic fluctuations not expected for conventional qcp scenarios in fms.@xcite a further issue may be the disorder associated with tuning via chemical substitution , which , in principle , may smear out the phase transition .
we note that at least some of the unconventional behavior of uru@xmath9re@xmath1si@xmath8 may be understood within a theory that describes a fm kondo lattice.@xcite however , it is still unclear whether quantum phase transitions of fm metals are universally first - order .
moreover , as pointed out by pfleiderer , both first- and second - order qpts may lead to interesting new phenomena in metallic systems.@xcite the interest in the nature of fm qpts has been further intensified by the recent discovery of unconventional superconductivity ( sc ) in uranium - based compounds .
the compounds uge@xmath8 ( under pressure),@xcite urhge,@xcite uir ( under pressure),@xcite and ucoge@xcite exhibit microscopic coexistence of sc and itinerant electron fm that is apparently associated with the uranium ions .
this has led to the suggestion that the scing electrons in the uranium - based fm superconductors may pair in triplet states with parallel alignment of the electron spins , since such a configuration may not be subject to the same magnetic pair - breaking effects for a singlet sc ( although orbital pairbreaking could still be present ) .
furthermore , it has been proposed that the sc in these compounds is mediated by critical fluctuations associated with a fm qcp.@xcite naturally , this raises the question of how the proposed first - order nature of fm phase transitions influences the sc state , since in that case critical fluctuations will be absent .
in addition , sc with triplet pairing is thought to be extremely sensitive to disorder,@xcite demonstrating the need for a model system in which disorder effects on fm sc may be studied systematically . in the present manuscript , we expand on our recent results for the alloy series uco@xmath0fe@xmath1ge which may shed some light on these questions.@xcite we report transport , magnetic , and thermal properties , which reveal an evolution from ferromagnetism in ucoge towards pauli paramagnetism in ufege.@xcite in agreement with a recent report which was restricted to small iron dopant concentrations,@xcite we find that @xmath5 initially increases and the superconducting critical temperature @xmath11 is rapidly suppressed . at higher iron concentrations , @xmath5 ,
the ordered magnetic moment @xmath12 , and the magnetic entropy @xmath13 pass through a maximum and are then suppressed smoothly towards a quantum phase transition near @xmath14 .
non - fermi - liquid ( nfl ) behavior is observed in the electrical resistivity and specific heat near @xmath15 .
we also find that @xmath5 is suppressed with fe substitution in a manner that is consistent with theoretical expectations for a three dimensional itinerant electron ferromagnet.@xcite these findings point to the presence of a fm qcp at @xmath15 where deviations from theoretical predictions in the critical low @xmath16 behaviors are likely due to disorder .
remarkably , we also find that samples with @xmath17 and 0.2 are very similar , suggesting that apart from the absence of sc ( above 50 mk , the lowest temperature to which the samples were measured ) around @xmath15 , the fm dome is symmetric with respect to its magnetic properties .
the substitutional system uco@xmath0fe@xmath1ge thus presents a unique opportunity to investigate how disorder influences fm quantum critical fluctuations and their resulting emergent phenomena , namely superconductivity and/or non - fermi - liquid behavior , by comparing the qcp situated near @xmath18 ( accessible via pressure@xcite ) with the qcp at @xmath15 that is more affected by disorder .
samples of uco@xmath0fe@xmath1ge , with various concentrations spanning the range from @xmath17 to 1 , were prepared using an arc furnace in a zirconium gettered argon atmosphere on a water cooled copper hearth .
the starting materials were cut from solid pieces of pure uranium ( new brunswick lab , 3n ) , cobalt ( alfa aesar , 99.9+@xmath19 ) , iron ( alfa aesar , 99.99+@xmath19 ) and germanium ( alfa aesar puratronic , 99.9999+@xmath19 ) . after melting , each sample was flipped over and re - melted .
this process was repeated five times in order to ensure homogeneous mixing of the starting materials .
each as - grown boule was cut in half using a diamond wheel saw , and half of the sample was sealed under vacuum in a quartz tube and annealed at 900@xmath20c for 10 days .
following our initial studies of polycrystalline material , we produced a single crystal specimen with @xmath7 @xmath21 0.22 using the czochralski technique . in the following figures , we use stars to identify this sample . divided by applied field @xmath22 versus temperature @xmath16 for uco@xmath0fe@xmath1ge with @xmath7 = 0 - 0.25 .
all scans were measured upon field cooling in 10 oe ( except for @xmath7 = 0.25 in 1000 oe ) .
the inset shows an example of the construction used to estimate @xmath5 . ]
powder diffraction patterns were obtained with a bruker d8 diffractometer utilizing cu k@xmath23 radiation and the results analyzed via rietveld refinement using the gsas+expgui@xcite software package .
magnetization data for temperatures 2 - 300 k and fields up to 7 tesla were collected with a quantum design magnetic properties measurement system ( mpms ) squid magnetometer .
resistivity measurements were performed using a standard four wire method with a linear research lr-700 ac resistance bridge operating at @xmath24 hz . for the resistivity measurements , temperatures down to 1 k
were achieved using a home - built @xmath25he bath cryostat while data below 1 k were taken in an oxford kelvinox dilution refrigerator .
several resistivity measurements were repeated with varying currents in order to ensure that spurious heating did not effect the data at the lowest temperatures .
specific heat measurements were obtained from 1.8 - 50 k with a quantum design dynacool physical properties measurement system ( ppms ) utilizing a standard adiabatic heat pulse technique .
specific heat measurements were obtained from 0.4 - 20 k using a quantum design physical property measurement system with a he-3 option .
while ucoge exhibits the orthorhombic tinisi crystal structure , ufege undergoes a monoclinic distortion of the tinisi structure below @xmath26 @xmath20c.@xcite we found that the orthorhombic tinisi structure of ucoge persisted up to @xmath27 for uco@xmath0fe@xmath1ge .
@xcite the @xmath7 = 0.7 , 0.8 , and 0.9 samples showed significant impurity phases and their diffraction patterns could not be fitted well by the calculated patterns for orthorhombic tinisi , monoclinic ufege , or a mixture of these phases .
the as - grown doped samples ( @xmath28 ) showed only small concentrations of an unidentified impurity phase ( the integrated intensity of the largest impurity peak is only @xmath29 of the largest sample peak ) . upon annealing ,
the size of the impurity peaks grows significantly , indicating that annealing of doped uco@xmath0fe@xmath1ge results in significant precipitation of impurity phases . for pure ucoge ,
annealed samples have previously been reported to possess significantly larger residual resistivity ratios than as - grown samples.@xcite for the above reasons , we restrict our analysis to annealed pure ucoge and as - grown doped uco@xmath0fe@xmath1ge samples ( with @xmath28 ) .
versus applied field @xmath22 at 2 k. the dashed line shows an example of the construction used to estimate the ordered magnetic moment @xmath30 . ] in figure [ fig : fig1 ] , we plot the low temperature magnetization @xmath31 divided by magnetic field @xmath22 versus temperature @xmath16 data , from which the evolution of the curie temperature @xmath5 is estimated by extrapolating @xmath31 below the ordering temperature to zero as shown in the inset . as in previously reported measurements,@xcite
the magnetization of pure ucoge exhibits a weak upturn at low temperatures , consistent with fm ordering at @xmath32 k. figure [ fig : fig7](a ) shows @xmath5 versus @xmath7 , demonstrating that @xmath5 initially increases rapidly , passes through a dome with a maximum of @xmath33 k at @xmath34 , and then drops to zero between @xmath35 .
we also find that the value of @xmath36 at @xmath16 @xmath21 2 k passes through a dome as @xmath7 is varied with a maximum near @xmath7 @xmath21 0.1 , suggesting that fe substitution initially drives the magnetism from being itinerant near @xmath7 @xmath21 0 to more localized near @xmath7 @xmath21 0.1 , after which it again becomes itinerant near @xmath7 @xmath21 0.2 .
further insight into the magnetic state is provided by the @xmath31 versus @xmath22 measurements presented in figure [ fig : fig2 ] .
these data were taken upon decreasing the field from 7 tesla , after cooling the samples to 2 k in zero field . for concentrations @xmath37 ,
the data show clear fm characteristics , while for concentrations @xmath38 , the behavior is consistent with paramagnetism .
a rough estimate of the ordered magnetic moment @xmath39 is obtained by linearly extrapolating the high field data to zero field as shown for @xmath40 by the dashed line construction in figure [ fig : fig2 ] .
the resulting values of the ordered moments are plotted versus @xmath7 in figure [ fig : fig7](c ) , demonstrating that @xmath39 passes through a dome , peaked at @xmath41 , and goes smoothly to zero near @xmath42 .
the value of @xmath39 at @xmath41 is more than an order of magnitude larger than in undoped ucoge , indicating a significant strengthening of the magnetic state , while at @xmath43 , the ordered moment has dropped to nearly the same value as in undoped ucoge , suggesting a return to highly itinerant weak ferromagnetism .
divided by temperature @xmath16 versus @xmath16 on a log scale in zero magnetic field @xmath22 for concentrations 0 @xmath44 @xmath7 @xmath44 0.075 .
the arrow indicates the curie temperature estimated for the @xmath45 sample .
the solid line is the estimated lattice contribution to the data ( @xmath46 @xmath21 @xmath47@xmath48 ) for @xmath5 @xmath44 @xmath16 @xmath44 15 k where @xmath47 @xmath49 0.55 mj / mol - k@xmath25 from fits to the data .
( b ) @xmath50 versus @xmath16 on a log scale in zero magnetic field @xmath22 for concentrations 0.1 @xmath44 @xmath7 @xmath44 0.175 .
( c ) @xmath50 versus @xmath16 on a log scale in zero magnetic field @xmath22 for concentrations 0.2 @xmath44 @xmath7 @xmath44 0.5 . ]
figure [ fig : fig3 ] presents the specific heat @xmath51 divided by temperature @xmath16 versus @xmath16 on a log scale .
pure ucoge shows a weak anomaly with an onset near 3 k , and samples with @xmath52 show clear anomalies that are coincident with the onset of ferromagnetic order .
the ordering temperatures estimated from the inflection point at the anomaly are plotted versus @xmath7 in figure [ fig : fig7](a ) and are in agreement with results from magnetization .
the data for @xmath7 = 0.2 , 0.25 , and 0.3 show nearly logarithmic low @xmath16 upturns , but saturate towards constant values at low @xmath16 indicating the presence of a nearby qcp which induces nfl behavior .
the value of @xmath50 at @xmath16 @xmath21 1 k , which offers a measure of the effective charge carrier quasiparticle mass , is plotted versus @xmath7 in figure [ fig : fig7](d ) . here
, we find that @xmath53 reaches a maximum between @xmath7 = 0.2 and 0.25 , consistent with the viewpoint that strong spin fluctuations are present near the region where the ferromagnetism is suppressed towards @xmath16 @xmath21 0 . the magnetic contribution to the entropy @xmath54
is shown in figures [ fig : fig4](a , c ) , where @xmath54 was obtained by subtracting the lattice contribution ( @xmath46 @xmath21 @xmath47@xmath48 for @xmath5 @xmath44 @xmath16 @xmath44 15 k with @xmath47 @xmath49 0.55 mj / mol - k@xmath25 ) from @xmath50 , extrapolating the resulting @xmath55 to zero @xmath16 , and integrating between 0 @xmath44 @xmath16 @xmath44 15 k. results for @xmath55 for selected samples are shown in figures [ fig : fig4](b , d ) .
as summarized in figure [ fig : fig7](d ) , @xmath54(10 k ) increases from 0.07rln2 ( r @xmath21 8.314 j / mol - k is the ideal gas constant ) at @xmath7 @xmath21 0 to 0.15rln2 at @xmath15 @xmath21 0.22 , and finally decreases to 0.1rln2 at @xmath7 @xmath21 0.4 , suggesting a build up of entropy near @xmath15 .
it is noteworthy that peaks in isotherms of entropy versus substituent composition near the qcp were previously observed in the cerh@xmath0co@xmath1in@xmath56 system.@xcite we also find that @xmath13 evolves from 0.03rln2 for @xmath7 @xmath21 0 , through a peak of 0.12rln2 near @xmath7 @xmath21 0.1 , and returns to 0.047rln2 for @xmath7 @xmath21 0.2 , supporting the viewpoint that in the ordered state , fe substitution initially suppresses the conduction electron itineracy near @xmath7 @xmath21 0.1 , after which strong itineracy returns near @xmath15 . versus temperature @xmath16 for concentrations 0 @xmath44 @xmath7 @xmath44 0.15 .
( b ) magnetic contribution to the specific heat divided by temperature @xmath55 versus @xmath16 for selected concentrations 0 @xmath44 @xmath7 @xmath44 0.15 .
( c ) @xmath57 for 0.175 @xmath44 @xmath7 @xmath44 0.4 .
( d ) @xmath58 for 0.175 @xmath44 @xmath7 @xmath44 0.4 . ]
representative normalized electrical resistivity @xmath59 data are presented in figure [ fig : fig5 ] on a log scale .
the residual resistivity @xmath60 changes from 70 @xmath61@xmath62 cm for @xmath7 @xmath21 0 to @xmath49 1 m@xmath62 cm for all other doping concentrations .
all samples exhibit metallic behavior with a broad maximum at high temperatures ( @xmath63 ) that is associated with the onset of the coherent ground state .
@xmath63 gradually increases with increasing @xmath7 , indicating that the hybridization strength between the @xmath64- and conduction electrons is enhanced by fe substitution .
as shown in figure [ fig : fig7](e ) , chemical substitution rapidly introduces disorder , resulting in a rrr @xmath21 @xmath65/@xmath60 @xmath49 2 for the entire doping series . samples with @xmath66 display peaks in @xmath67@xmath68/@xmath67@xmath16 at @xmath5 , as shown for @xmath7 @xmath21 0.075 in figure [ fig : fig5](b ) , which agree with results from @xmath69 and @xmath70 . for @xmath71 @xmath44 @xmath16 @xmath44 @xmath5 and 0 @xmath44 @xmath7 @xmath44 0.2 ,
power law behavior of the form @xmath72 is observed [ representative curves are shown in figure [ fig : fig5](c ) ] . power law behavior is also observed for @xmath7 @xmath21 0.2 and 0.22 for @xmath16 @xmath44 4 k and 2 k , respectively .
the best fits are obtained by plotting @xmath68 versus @xmath73 and adjusting the value of @xmath74 to maximize the range of linear behavior .
the resulting values @xmath74 are plotted in figure [ fig : fig7](e ) , where nfl - like deviations from @xmath48 behavior start near @xmath7 @xmath21 0.15 .
interestingly , it is not possible to extend the power law behavior to the lowest temperature for @xmath7 @xmath75 0.3 .
rather , the resistivity weakly increases with decreasing @xmath16 .
the origin of this behavior is unclear .
superconductivity appears as nearly complete , and partial resistive transitions for the @xmath17 and @xmath76 samples , respectively , while no trace of superconductivity is observed at any of the other iron concentrations .
normalized to @xmath65 versus temperature @xmath16 for selected concentrations @xmath7 on a log scale .
( b ) derivative of electrical resistivity with respect to temperature @xmath67@xmath68/@xmath67@xmath16 versus @xmath16 for @xmath7 @xmath21 0.075 .
( c ) @xmath68 minus residual resistivity @xmath77 versus @xmath73 for concentrations spanning @xmath15 , as described in the text .
the solid lines show the power law fits to the data .
the black arrows denote the fit range . ]
having identified evidence for non - fermi - liquid behavior in the region @xmath7 @xmath21 0.2 - 0.25 , we subsequently synthesized a single crystal specimen with @xmath7 @xmath21 0.22 .
results for @xmath78 and @xmath79 are summarized in figure [ fig : fig6 ] where we find that the single crystal sample follows the trend that we expect from measurements performed on polycrystalline material .
as shown in figure [ fig : fig6](a ) , the electrical resistivity is metallic , has a maximum at @xmath63 , and exhibits power law behavior at low @xmath16 where @xmath74 @xmath21 1.4 .
specific heat measurements reveal a nearly logarithmic divergence at low @xmath16 which is spanned by the low @xmath16 upturns for @xmath7 = 0.2 and 0.25 .
these results strongly suggest the presence of a qcp near @xmath7 @xmath21 0.22 which drives the observed nfl behavior . versus temperature @xmath16 at low @xmath16 for a single crystal specimen of uco@xmath0fe@xmath1ge with @xmath7 @xmath21 0.22 . the solid line is a fit to the data of the form @xmath78
@xmath21 @xmath60 @xmath80 @xmath81 with @xmath74 @xmath21 1.4 , as described in the text .
( b ) @xmath78 for uco@xmath0fe@xmath1ge with @xmath7 @xmath21 0.22 for 0.7 k @xmath44 @xmath16 @xmath44 300 k. ( c ) specific heat @xmath51 divided by temperature @xmath16 versus @xmath16 for several polycrystalline uco@xmath0fe@xmath1ge specimens and the single crystal specimen with @xmath7 @xmath21 0.22 . the black arrow denotes the fit range . ]
dependence of : ( a ) curie temperature @xmath5 as determined from magnetization @xmath31 , specific heat @xmath51 , and electrical resistivity @xmath68 data and onset superconducting critical temperature @xmath11 from electrical resistivity ; ( b ) scaling analysis of the curie temperature @xmath5 , as described in the text .
the grey region represents the error in the critical concentration @xmath15 and the exponent @xmath82 ; ( c ) ordered magnetic moment @xmath39 ; ( d ) @xmath50 at @xmath16 @xmath21 1 k. for data sets where @xmath50 was measured to 1.8 k , the value at 1 k was estimated by linear extrapolation .
; ( e ) the parameter @xmath74 obtained from @xmath72 power law fits to the data and the residual resistivity ratio rrr @xmath21 @xmath65/@xmath60 . ]
our results reveal an interesting phase diagram for uco@xmath0fe@xmath1ge .
initially , ferromagnetism and superconductivity persists over a limited range in @xmath7 .
we note , however , that the resistive superconducting transitions are not complete , indicating that the samples investigated in this study do not exhibit bulk superconductivity . with increasing @xmath7 , the ferromagnetic ordering temperature increases and
the magnetism becomes more localized , as reflected in several quantities . for 0 @xmath44 @xmath7 @xmath44 0.1 ,
electrical resistivity measurements reveal fermi liquid behavior as evidenced by the @xmath78 @xmath83 @xmath48 behavior . near @xmath7 @xmath21 0.075
, @xmath5 goes through a maximum and subsequently decreases with increasing @xmath7 until it disappears between @xmath15 @xmath21 0.2 - 0.25 . in the vicinity of @xmath15 , measurements of @xmath69 , @xmath79 , and @xmath78 provide evidence for a qcp : ( 1 ) @xmath50 obeys a nearly logarithmic divergence in @xmath16 , ( 2 )
a maximum in the electronic coefficient of the specific heat is seen around @xmath15 , ( 3 ) @xmath54 goes through a maximum near @xmath15 , and ( 4 ) power law fits to @xmath78 reveal that @xmath84 evolves from fermi liquid behavior at low @xmath7 through a
v - shaped " region with a minimum between @xmath15 @xmath21 0.2 - 0.25 and recovers towards @xmath74 @xmath21 2 for larger @xmath7 .
we also consider that for a clean itinerant ferromagnetic qcp scenario , it is expected that the ordering temperature should vary as @xmath5 @xmath83 ( @xmath85 @xmath86 @xmath7)@xmath87 where @xmath82 @xmath21 3/4.@xcite as shown in figure [ fig : fig7](b ) , the suppression of @xmath5 is consistent with this prediction , giving @xmath15 @xmath21 0.22 @xmath88 0.01 and @xmath82 @xmath49 0.63 @xmath88 0.07 . altogether , these results agree with the theoretical prediction for a three dimensional itinerant electron ferromagnetic qpt , except that @xmath89 and @xmath82 are slightly reduced from the predicted exponents of 5/3 and 3/4 , which are taken from the moriya scr and hertz - millis theories , respectively.@xcite however , disorder undoubtedly plays a role in determining the ground state behavior in this system and likely perturbs the predicted behaviors , as is also seen for urh@xmath0ru@xmath1ge and cesi@xmath90 .
@xcite we additionally speculate that the slight saturation in @xmath50 for @xmath7 @xmath21 0.22 occurs either due to disorder or being slightly away from the ideal @xmath15 .
since the question of whether ferromagnetic qpts are generally first - order as suggested by theory , or second - order as often indicated by experiment , is still a matter of debate , we discuss this issue in more detail .
notably , it is clear that in chemical substitution studies , in particular , disorder may play an important role .
while the analysis of our data is in agreement with a scenario involving a clean itinerant ferromagnetic qcp , it is possible to obtain similar nfl behavior due to disorder.@xcite from this perspective , the qpt at @xmath15 may be a first - order phase transition that is washed out by disorder .
we note , however , that similar studies of compounds with comparable rrrs ( and thus a comparable amount of disorder ) such as ucoge@xmath0si@xmath91 ( rrr @xmath49 4)@xcite and urh@xmath0ru@xmath91ge ( rrr @xmath49 2)@xcite have usually concluded that the qpt is of second - order .
more detailed studies that go beyond the scope of this work that can access the amount of disorder by determining discrepancies between the local structure and average crystal structure , such as exafs@xcite and pdf@xcite may ultimately allow this issue to be clarified .
a primary question is why samples in the vicinity of @xmath92 do not display sc , in analogy to the undoped parent compound .
one possibility is that if ucoge exhibits spin triplet sc , which only survives in samples with mean - free - paths that are significantly longer than the coherence length , then disorder should rapidly suppress @xmath71.@xcite if disorder is the main difference between these concentrations , then it also follows that the normal state behavior for @xmath7 @xmath21 0.175 - 0.2 should be similar to that of @xmath7 @xmath21 0 .
our results support this viewpoint , particularly for @xmath7 @xmath21 0.175 .
this relationship is more obvious when we compare to previous pressure and doping studies of ucoge . under pressure , @xmath5 decreases and @xmath11 initially increases.@xcite near @xmath93 gpa , @xmath5 is suppressed to zero temperature and @xmath11 passes through a weak maximum and subsequently extends beyond the ferromagnetically ordered state . the magnitude of the feature in the ac magnetic susceptibility at @xmath5 , which may be related to the size of the ordered moment
, is also gradually suppressed with pressure and extrapolates to zero near @xmath94 gpa .
this behavior , together with the possible appearance of nfl behavior in the electrical resistivity near @xmath95,@xcite suggests the presence of a pressure - induced qpt in ucoge that may be analogous to what we observe in uco@xmath0fe@xmath1ge for @xmath7 @xmath21 0.22 .
for ucoge@xmath0si@xmath1 , @xmath11 and @xmath5 decrease until they both vanish simultaneously for @xmath96 at a qpt that is also accompanied by nfl behavior.@xcite from these observations , we propose that ucoge , ucoge@xmath0si@xmath1 , and uco@xmath0fe@xmath1ge exist on a @xmath97 manifold in which the tuning parameters provide access to qpts which are mainly differentiated by the degree of disorder that is present for each case : i.e. , while pressure induces little disorder , si and fe substitution result in structural and structural / electronic disorder , respectively , which act to suppress the superconductivity but do not destroy the quantum fluctuations that are responsible for the nfl behavior .
a related challenge is to understand whether the phase transitions near @xmath7 @xmath21 0 and 0.22 are first or second order .
it has been proposed that clean metallic ferromagnets with low curie temperatures in two or three dimensions generically exhibit first order transitions.@xcite this prediction is significant because it differentiates the types of quantum fluctuations that might be found in the vicinity of a fm qpt from those that are found near a continuous qcp . for pure ucoge
, nmr measurements suggest that the ferromagnetic transition is first order .
@xcite however , bulk measurements such as specific heat and neutron scattering@xcite do not support this viewpoint and , moreover , recent nmr measurements from the same group suggest the existence of critical ferromagnetic fluctuations which drive the superconductivity in ucoge.@xcite therefore , it is not clear whether the ferromagetic phase transition in ucoge is first or second order . from our measurements
, we suggest that the ferromagnetic phase transition near @xmath7 @xmath21 0.22 is second order .
similar behavior is also observed near the critical concentrations in uru@xmath9re@xmath1si@xmath8 and urh@xmath0ru@xmath1ge , where disorder has been suggested as a mechanism for avoiding a first order transition .
@xcite if the nature of the phase transition is different for @xmath7 @xmath21 0 and 0.22 , then this may provide an alternative explanation for their differing low @xmath16 behaviors .
further detailed studies ( _ e.g. _ , inelastic neutron scattering and nmr ) of high quality samples near the @xmath92 critical concentration will be useful to reveal differences between the magnetism in undoped ucoge versus that at @xmath92 , which could help to shed light on the precise conditions which lead to sc in ucoge and nfl behavior in uco@xmath0fe@xmath1ge near @xmath7 @xmath21 0.22 .
we have carried out x - ray diffraction , magnetization , specific heat , and electrical resistivity measurements for the uco@xmath0fe@xmath1ge series for @xmath2 , to investigate the @xmath3 phase diagram .
substitution of fe into ucoge initially increases the curie temperature and a rapidly destroys superconductivity . near @xmath4 ,
the ferromagnetic transition is suppressed to zero temperature at an apparent qcp , where the temperature dependence of the electrical resistivity and specific heat reveal non - fermi liquid behavior .
having established the @xmath3 phase diagram for this doping series , it will be of great interest to pursue further studies ( e.g. , neutron scattering , thermal expansion , photoemission , stm ) to elucidate the nature of the qpt in this system , as well as the role of disorder .
sample synthesis and basic characterization where sponsored by the u.s .
department of energy ( doe ) under research grant de - fg02 - 04er46105 .
low temperature dilution refrigerator measurements were supported by the national science foundation under grant dmr-0802478 .
mj gratefully acknowledges financial support by the alexander von humboldt foundation .
work at los alamos national laboratory was performed under the auspices of the united states department of energy .
s. s. saxena , p. agarwal , k. ahilan , f. m. grosche , r. k. haselwimmer , m. steiner , e. pugh , i. r. walker , s. r. julian , p. monthoux , g. g. lonzarich , a. d. huxley , i. sheikin , d. braithwaite , and j. flouquet , _ nature _ * 406 * , 587 ( 2000 ) . | we have carried out a comprehensive study of the uco@xmath0fe@xmath1ge series across the entire range of compositions @xmath2 , and report the results of x - ray diffraction , magnetization , specific heat , and electrical resistivity measurements to uncover the @xmath3 phase diagram .
substitution of fe into ucoge initially results in an increase in the curie temperature and a rapid destruction of the superconductivity . near @xmath4 ,
the ferromagnetic transition is suppressed to zero temperature at an apparent ferromagnetic itinerant electron quantum critical point , where the temperature dependence of the electrical resistivity and specific heat in this region reveal non - fermi liquid behavior . |
it has long been discussed both observationally and theoretically why some of the galactic gcs show star - to - star inhomogeneity among the light elements of stars and what physical mechanisms are responsible for the inhomogeneity ( e.g. , cottrell & da costa 1981 ; sneden et al .
1992 ; norris & da costa 1995 ; cannon et al . 1998 ; gratton et al . 2004 ; dantona & caloi 2004 ; fenner et al . 2004
; norris 2004 ; lee et al . 2005 ; smith et al .
2005 ; bekki et al . 2007 ; alves - brito et al . 2008 ; calelan 2008 ; kayser et al . 2008 ; piotto 2008 ; da costa et al .
2009 ; marcolini et al . 2009
; yong et al . 2009 ; carretta et al . 2010 ; dercole et al . 2010 ; romano et al . 2010 ; van loon 2010 ) .
although such star - to - star inhomogeneity was discovered in 70 s and 80 s ( e.g. , cohen 1978 ; peterson 1980 ; norris et al . 1981 ; leep et al .
1986 ) , recent statistical studies for a larger number of the galactic gcs have established that the presence of multiple stellar populations is an universal phenomena seen across most of the galactic gcs ( e.g. , carretta et al .
2010 ) . the latest observational results by ferraro et al .
( 2009 ) have revealed that the galactic metal - rich gc tarzan 5 has two different stellar populations with different abundances of heavy elements .
lee et al .
( 2009 ) has also suggested a significant fraction of the galactic gcs have two different populations with different abundances in heavy elements based on color - magnitude diagrams of the gc stars in @xmath8-bands .
now large star - to - star abundance variations have been confirmed in almost unevolved stars in some of the galactic gcs ( e.g. , gratton et al 2001 ; ramirez & cohen 2002 ; bedin et al .
2004 ; gratton 2004 ; carretta et al .
2004 ; piotto et al .
2005 , 2007 ; dorazi et al .
2010a ) , which strongly suggests that after the first generation ( fg ) of stars formed within forming gcs , gas chemically mixed with gaseous ejecta of fg stars was converted into to the second generation ( sg ) of stars ( and even third and fourth generations ) .
observations have now being extensively investigating how chemical properties of sg stars ( e.g. fractions of sg stars ) correlate with their global internal properties ( e.g. , magnitudes and ellipticities ) and with locations and 3d motions with respect to the galactic center to understand the origin of sg stars and physical relationships between formation of fg and sg stars ( e.g. , carretta 2006 , carretta et al .
2010 ) . following these observational developments ,
theoretical studies have considered that sg stars can form from gaseous ejecta either by agb stars ( `` agb scenario '' ; dantona & caloi 2004 ; karakas et al .
2006 ) or by fast rotating massive stars ( `` frms '' scenario ; prantzos & charbonnel 2006 ; decressin et al . 2007 ) and thereby investigated whether the observed physical properties of the galactic gcs can be explained by their models .
although it has not been determined which of the two can explain observations in a self - consistent manner , a number of authors suggested that the agb scenario is a more physically viable because agb ejecta is more likely to be converted into new stars owing to low ejection velocities ( e.g. , renzini 2008 ) .
recent observations have provided the following two key results which can give strong constraints on any theory for gc formation : ( i ) the large fraction ( typically @xmath9 ) of sg stars in each individual galactic gcs and ( ii ) the observed na - o and mg - al anti - correlations between cluster stars ( e.g. , carreta et al .
2010 for a comprehensive study for these two ) .
the first key result suggests that the original stellar systems ( i.e. , fg stars ) need to be much more massive than the present gcs unless we adopt very unusual and unrealistic imfs ( e.g. , smith & norris 1982 ; dantona & caloi 2004 ) : the original systems are either very massive star clusters or dwarf galaxies hosting gcs ( e.g. , bekki & norris 2006 ) .
if the original systems are really massive ones , then the fg stars need to be preferentially lost while sg ones remain the same so that the observed typical fraction ( @xmath10 ) of sg stars can be explained . although a number of theoretical works based on the agb scenario tried to explain the second key observational result in a self - consistent manner ( e.g. , fenner et al .
2004 ; bekki et al .
2007 ; dantona et al . 2005 ; dantona & ventura 2007 ; ventura & dantona 2006 , 2008 ; dercole et al .
2010 ) , their models appear to have not yet explained all of the relevant observations on chemical abundances of the galactic gcs in a fully self - consistent manner .
chemical evolution models based on the frms scenario have not yet been fully explored so that the validity of the frms scenario can not be currently assessed . in these previous models , it is assumed that agb ejecta can be converted into new stars ( i.e. , formation of sg stars ) within already existing clusters .
however it would not be so obvious that such secondary star formation can occur within clusters , given the shallow gravitational potential wells of clusters and the small mass fraction of agb ejecta .
therefore , secondary star formation processes within clusters should be investigated by numerical simulations that can include various physical processes within clusters ( e.g. , retention of agb ejecta ) .
dercole et al .
( 2008 ) first investigated whether sg stars can be formed from the gaseous ejecta of agb stars of fg with @xmath11 .
however , their models are based on one - dimensional hydrodynamical simulations and have limitations in predicting 3d structures and kinematics of final stellar systems .
three - dimensional ( 3d ) stellar and gas dynamical numerical simulations with a plausible model for star formation are ideal to investigate secondary star formation within clusters and can furthermore provide theoretical predictions that can be compared with the observed differences in 3d structures and kinematics between fg and sg stars ( e.g. , norris et al . 1997 ; ferraro et al .
2002 ; sollima et al .
2005 , 2007 ; pancino et al .
2007 bellini et al .
2009 ; anderson & van der marel 2010 ) .
the purpose of this paper is thus to investigate extensively star formation from gaseous ejecta from agb stars within mscs based on self - consistent hydrodynamical simulations with a reasonable model for star formation .
we adopt a scenario in which ( i ) mscs can be formed in their host galactic building blocks embedded in dark matter halos at high redshifts and ( ii ) new stars formed from agb ejecta can finally become the observed sg stars in the present gcs . based on this scenario , we investigate ( i ) how agb ejecta can be retained within mscs , ( ii ) whether and how secondary star formation from agb ejecta proceeds within mscs , and ( iii ) how mscs evolve if they are located in nuclei of their hosts .
we also investigate how mscs lose their stars during tidal interaction with their hosts in order to discuss the observed smaller fractions of fg stars in the galactic gcs .
a number of previous works discussed a scenario in which gcs were originally stellar galactic nuclei or nuclear star clusters in nucleated galaxies and their host galaxies had been already destroyed by strong tidal fields of much larger galaxies to disappear completely ( e.g. , zinnecker et al .
1988 ; freeman 1993 ; bekki & freeman 2003 ; bellazzini et al .
2008 ; bker 2008 ) . following this scenario , bekki et al .
( 2007 ) investigated chemical abundances of fg and sg stars in gcs formed within the central regions of their host galaxies .
the evolution of mscs initially nuclear regions of their hosts in the present study therefore can provide some implications on the validity of the above scenario in explaining observational properties of gcs .
the plan of the paper is as follows : in the next section , we describe the details of the proposed scenario for gc formation . in 3 , we describe the numerical models for evolution of gaseous ejecta of agb stars within mscs . in 4 , we present the numerical results mainly on physical properties of sg stars . in 5 ,
we discuss how long mscs can retain most of their stars when they are influenced by strong tidal fields of their hosts . in
6 , we discuss a number of key issues related to the origin of multiple stellar populations in gcs .
we summarize our conclusions in 7 .
although we describe the scenario in the context of the galaxy formation , the formation processes of gcs within galaxies in general would be similar to those described below .
the scenario is based on the results from our present and previous theoretical studies on gc formation ( e.g. , bekki et al .
2002 ; bekki et al .
2004 ; bekki & chiba 2007 ; bekki 2006 ; bekki et al .
2007 ; hurley & bekki 2008 ; bekki et al .
2008 ) . since the scenario is based partly on results of cosmological simulations ( e.g. , bekki et al .
2008 ) , the scenario has some implications on the observed correlations between abundance properties of gcs and 3d motion and kinematics of gcs ( e.g. , carretta 2006 ; lee et al .
2007 ) . in the present paper , the masses of the present gcs
are represented by @xmath12 so that @xmath0 ( original masses of the gcs ) and @xmath12 can be discriminated with each other .
[ cols="^,^,^,^,^,^,^,^ " , ] fg stars of mscs form from high - density gmcs with masses ( @xmath13 ) as large as or larger than @xmath14 within the galactic building blocks which are later destroyed owing to the strong tidal field of the galaxy when they merge with the galaxy .
most of the hosts form as massive dwarf galaxies with masses ( @xmath15 ) larger than @xmath16 before reionization ( @xmath17 ) , and thus their star formation and chemical enrichment histories can be different from those of the present dwarfs within the galaxy ( bekki et al . 2008 ) .
the formation places of mscs are highly likely to their hosts central regions where mass fractions of gmcs can be much higher ( i.e , not necessarily in the nuclei of the hosts ) .
the mscs formed in the very centers of their hosts start their lives as nuclear mscs so that their evolution can be different significantly from that for mscs formed outside the very centers owing to much deeper gravitational potentials of hosts central regions .
the host gmcs of mscs may well initially have numerous substructures ( smaller gmcs ) owing to their large masses ( e.g. , efremov 1995 ; bonatto & bica 2010 ) so that mscs at their birth can be composed of a number of smaller clusters ( i.e. , not single entities ) .
these smaller sub - units can finally merge with one other to form single massive clusters within the merging timescale depending on @xmath0 and their sizes ( @xmath18 ) .
massive stars and type ii supernova can strongly influence later evolution of molecular gas left behind from the formation of fg stars .
if most of the energy from massive stars and type ii supernova are converted into kinetic energy of gas surrounding the stars , then gaseous ejecta from the stars can not be mixed well with the residual molecular gas ( bekki & chiba 2007 ) .
therefore , sg formation from gaseous ejecta from the above energetic stars and the mass fraction of sg stars formed from the mixed gas is very small ( @xmath19 ) .
however , if massive stars are frmss with very slow ejection _ radial _ velocities ( @xmath20 km s@xmath21 ) , then the gaseous ejecta could be well mixed with the residual gas to form sg stars before supernova explosion expels a significant fraction of gas .
although type ii supernova can heat up and expel a significant fraction of gas from massive gmcs , some fraction of gas can be immune from the effects of supernova owing to the clumpy nature of the massive gmcs .
the smaller residual clouds may well be later accreted onto the forming clusters to dilute agb ejecta .
while supernova explosion is influencing significantly gmcs , star formation rates within the gmcs can drop significantly .
gaseous ejecta from massive agb stars start to be accreted onto the central regions of mscs after all supernova events finish .
however star formation does not resume until an enough amount of gas can be accumulated within the central regions : there should be a threshold gas mass fraction ( @xmath22 ) above which star formation can start within mscs
. therefore there can be significant time - delay ( @xmath23 yr ) between fg and sg formation in this scenario .
sg stars in a msc start to form efficiently in the central region of the msc when the mass fraction of the accumulated agb ejecta to @xmath0 ( @xmath24 ) exceeds @xmath22 .
this efficient secondary star formation _ from agb ejecta _ can occur only if @xmath0 exceeds a threshold cluster mass ( @xmath1 ) , because agb ejecta can not be efficiently retained in mscs with smaller @xmath0 . during this secondary star formation
, the residual molecular gas can be mixed with agb ejecta to form new stars .
owing to the presence of deeper gravitational potential wells of mscs , star formation efficiencies ( @xmath25 ) can be much higher ( @xmath26 ) for this secondary star formation so that very compact clusters can be formed .
mscs thus can initially show `` nested structures '' with more diffuse distributions of fg stars and more compact ones of sg stars .
although mscs with @xmath0 below @xmath1 can hardly form new stars from agb ejecta , such less massive clusters can still capture residual molecular gas and smaller nearby gmcs ( not chemically contaminated by ejecta from fg stars ) .
cold molecular gas with _ initially very high - densities _ obtained by mscs themselves can be converted into new stars within the central regions of mscs ( bekki & mackey 2009 ) .
very low - mass mscs can be destroyed by interaction with gmcs so that secondary star formation can not happen in their central regions .
therefore , there should be a threshold cluster mass below which mscs show differences neither in ages nor in light elements among their stellar populations .
secondary star formation processes within nuclear mscs can be significantly different from those described above owing to much deeper potential wells of their hosts ( as described in the present study ) .
secondary star formation can continue until mscs lose most of their fg stars , though massive stars and type ii supernova of sg stars can also suppress or even truncate further star formation .
strong tidal fields of hosts can efficiently strip stars preferentially from fg stars owing to initially diffuse spatial distributions . this preferential stripping _ by the galaxy _ has been already proposed by dercole et al .
( 2008 ) and discussed in the context of origin of the galactic stellar halo ( vesperini et al .
the present scenario suggests that not the tidal field of the galaxy but those of hosts of mscs are responsible for the tidal stripping of most fg stars in mscs .
the timescale of a msc to lose most of their fg stars due to tidal stripping by its host depends on its position , with respect to its host ( @xmath27 ) , @xmath0 , and @xmath18 . therefore , more massive and denser mscs can continue secondary star formation longer so that agb stars with lower masses can possibly contribute to further star formation .
hosts of mscs are strongly influenced by the tidal field of the forming galaxy during their merging with the galaxy so that they can be completely destroyed by the galaxy .
the stripped stars from hosts , which include fg stars from mscs , can be some parts of the galactic stellar halo .
mscs are stripped during disintegration of their hosts to finally become halo gcs .
strong ram pressure of the galactic halo gas ( e.g. , frank & gisler 1976 ; bekki 2006 ) and no cold gas available in the halo prevent almost completely star formation of the mscs since they become the galactic halo gcs .
thus , multiple stellar populations can be formed within mscs only when they are within their hosts .
gas can be efficiently transferred to the nuclear regions of hosts to be converted into new stars while they are being destroyed by the galaxy ( e.g. , bekki & freeman 2003 ) .
these new stars can have chemical abundances in heavy elements different from those of original nuclear mscs , because they are from the outer regions , where chemical enrichment histories are quite different from those of nuclei .
nuclear mscs can show star - to - star abundance variations not only in light elements but also in heavy ones .
a number of authors have already suggested that if the standard imf is applied for fg stars and if @xmath25 ( star formation efficiency for sg formation ) is 1.0 , then original total masses of fg stars ( @xmath0 ) are required to be at least ten times larger than the present masses of sg stars in the galactic gcs with multiple stellar populations ( e.g. , bekki & norris 2006 ) . given that @xmath25 is not like 100%
as observed in star - forming regions in the galaxy , the required @xmath0 should be even higher than the above .
recent chemical evolution models have shown that the total mass of pristine gas that can mix with agb ejecta to form sg stars needs to be comparable to that of agb ejecta in order to explain the observed levels of star - to - star abundance variations and the o - na anti - correlation in the galactic gcs ( e.g. , dercole et al .
2010 ) : agb ejecta should not be too much diluted by external gas captured / accreted by mscs . thus the required @xmath0
estimated in previous studies can not change significantly even if contribution from external pristine gas to sg formation is considered . in the present scenario , gaseous ejecta from agb stars with masses ranging from @xmath28 to @xmath29
can contribute to the formation of sg stars within a timescale of @xmath30 ( which determines @xmath31 ) .
we here estimate the mass fraction of agb ejecta to the initial mass of a msc ( @xmath24 ) ( i ) for a given @xmath28 and ( ii ) for a given @xmath30 by assuming @xmath32 and power - law initial mass functions with the slopes of @xmath33 .
the adopted imf in number defined as @xmath34 , where @xmath35 is the initial mass of each individual star and the slope @xmath36 corresponds to the salpeter imf .
the normalization factor @xmath37 is a function of @xmath0 , @xmath38 ( lower mass cut - off ) , and @xmath39 ( upper mass cut - off ) : @xmath40 where @xmath38 and @xmath39 are set to be @xmath41 and @xmath42 , respectively .
the total mass of agb ejecta within a msc ( @xmath43 ) is accordingly described as : @xmath44 where @xmath45 describes the total gas mass ejected from an agb star with initial mass @xmath35 and final mass ( @xmath46 ) .
we derive an analytic form of @xmath45 ( @xmath47 ) from the observational data by weidemann ( 2000 ) by using the least - square fitting method , and find : @xmath48 in order to calculate the main - sequence turn - off mass ( @xmath49 ) we use the following formula ( renzini & buzzoni 1986 ) : @xmath50 where @xmath49 is in solar units and time @xmath30 in years .
1 shows that @xmath24 is larger for smaller @xmath31 both for the standard imf ( @xmath51 ) and the top - heavy one ( @xmath52 ) .
although the top - heavy model shows larger @xmath24 than the standard one for @xmath53 , the derived @xmath24 is small : only @xmath54 for @xmath55 and 0.079 @xmath56 .
ventura & dantona ( 2008 ) showed that the observed o - na anti - correlations can be well reproduced only if agb stars with masses larger @xmath57 can contribute to secondary star formation within gcs .
thus , it is highly likely that progenitor clusters for the present gcs . needs to be at least @xmath58 times more massive than the present gcs . as shown in fig . 1
, @xmath24 can be larger than 0.1 for @xmath59 gyr ( or @xmath60 ) for the two imf models .
it should be here stressed that mscs with larger @xmath30 can end up with a large age difference ( @xmath61 gyr ) between fg and sg stars .
original mscs can lose most of their fg stars due to tidal stripping of their hosts within well less than @xmath62 gyr , as shown later in the present study .
therefore gaseous ejecta only from massive agb stars can participate secondary star formation within mscs .
it would be reasonable to consider that @xmath28 is @xmath63 that corresponds to @xmath64 yr .
we investigate star formation of gaseous ejecta from fg agb stars in mscs with @xmath0 ranging from @xmath65 to @xmath66 by using the latest version of grape ( gravity pipe , grape-7 ) which is the special - purpose computer for gravitational dynamics ( sugimoto et al .
we have revised our original grape - sph code ( bekki 2009 ) so that we can investigate star formation processes within the above - mentioned massive stellar system .
mscs with very large ( @xmath67 ) are still referred to as `` clusters '' just for convenience in the present study , though their stellar masses are as massive as those of dwarf galaxies we focus mainly on _ secondary star formation _ from agb ejecta of fg stars within original mscs and describe the results of numerical simulations on the star formation .
we have numerically investigated ram pressure stripping of agb ejecta by ism of their hosts and its strength dependent on @xmath0 and physical parameters of ism .
we however describe the results in our forthcoming papers ( bekki et al .
2010 ) , because ram pressure stripping of agb ejecta of mscs by typical ism of hosts is only effective for low - mass mscs with @xmath68 , ( which can not become ordinary gcs in the present study ) and thus not so important .
we have numerically investigated accretion of warm ism onto mscs by bondi accretion and found that bondi accretion with the accretion rate of @xmath69 yr@xmath21 is possible only if @xmath0 is larger than @xmath70 for typical ism with @xmath71 atom @xmath72 , relative velocity of @xmath73 km s@xmath21 , and gas temperature of 10000 k. although this accretion rate is too small for mscs to obtain fresh gas for the formation of sg stars within @xmath23 yr , this bondi accretion can be an important mechanism for obtaining fresh gas in massive gcs with abundance variations in heavy elements like @xmath74 cen , as briefly discussed later in this paper .
we will describe the details of the numerical results on bondi accretion of ism in our forthcoming papers ( bekki et al .
2010 ) , mainly because this paper becomes too long if the results are included in this paper .
the original mscs are assumed to have a plummer density profile ( e.g. , binney & tremaine 1987 ) with luminosities ( @xmath75 ) and central velocity dispersions ( @xmath76 ) consistent with the relation observed for gcs ( djorgovski et al .
1997 ) : @xmath77 where @xmath78 is a normalization factor for the relation .
the scale length ( @xmath79 ) of a msc is determined by the formula @xmath80 where g is the gravitational constant . since the - mass - to - light - ratio ( @xmath81 )
is assumed to be constant for all scs , @xmath79 and @xmath76 are determined by the equations ( 1 ) and ( 2 ) for a given @xmath82 .
the normalization factor @xmath78 in equation ( 5 ) is determined such that a cluster with @xmath83 @xmath84 can have the size of @xmath85pc ( @xmath86 ) , and the central velocity dispersion of 7 km s@xmath21 .
haegan et al .
( 2005 ) revealed that massive stellar systems with masses larger than @xmath87 appear to have scaling regions different from those of the present gcs .
we accordingly consider that original sizes of mscs with @xmath88 can not be determined by the above equations ( 5 ) and ( 6 ) and therefore investigate models with different @xmath18 for a given @xmath0 for such mscs .
it is also possible that original mscs with @xmath89 have @xmath18 smaller or larger than those described by the above scaling relations because mscs composed only of fg stars have ages of @xmath23 yr in the present study : an order of @xmath90 yr would not be long enough to form gcs on the present scaling - relation owing to dynamical relaxation processes .
we therefore investigate models with @xmath18 smaller and larger than the above equation ( 5 ) and ( 6 ) predict for a given @xmath0 .
a msc is assumed to have a small amount of rotation with the ratio of the initial rotational energy ( @xmath91 ) to the total kinetic one ( @xmath92 ) being a free parameter represented by @xmath93 .
the parameter values of @xmath93 range from 0 ( no rotation ) to 0.3 ( rapid rotation ) .
the initial rotational velocity of a particle at a distance of @xmath94 from the center of msc is @xmath95 , where @xmath74 ( constant angular velocity ) is determined such that @xmath93 can be the adopted value .
therefore , the system has random kinetic energy ( @xmath96 ) of @xmath97 ( due to isotropic velocity dispersion of stars ) .
firstly we estimate @xmath92 for @xmath76 determined by @xmath0 and @xmath18 ( in equations ( 5 ) and ( 6 ) ) and then reduce @xmath76 so that the final system can be in virial equilibrium ( @xmath98 , where @xmath99 is the total potential energy of the system ) .
the fg stars in a msc are represented by equal - mass stellar particles with the particle number of @xmath100 ( @xmath101 ) and some fraction of the particles are assigned as `` agb stars '' with initial masses of @xmath102 and the total number of @xmath103 that can eject sph gas particles with ejection velocities of @xmath104 with respect to the centers of the agb stars .
the mass fraction of agb particles in a msc is a parameter represented by @xmath105 that is determined by the adopted imf .
as outlined in 2 , gaseous ejecta only from massive agb stars with initial masses of @xmath106 need to be converted into new stars ( i.e. , sg stars ) to explain the chemical abundances of sg ( e.g. , ventura & dantona 2008 ) . in the present paper , @xmath104=20 km s@xmath21
is adopted , which is a reasonable choice for massive agb stars ( e.g. , marshall et al .
2004 ) .
the present simulations can not resolve gaseous evolution of each individual agb star owing to the adopted numerical resolution ( @xmath107 pc ) .
we therefore assume that each agb particle initially has an expanding gaseous sphere which is much larger than the agb star itself .
the mass , size , and temperature of the large gaseous sphere represented by sph particles with the particle number of @xmath108 are set to be @xmath109 , @xmath110 and @xmath111 , respectively .
each agb particle accordingly has a gas sphere represented by @xmath108 sph particles with radial velocities ( with respect to the agb particle ) of @xmath104 .
the agb particle therefore has a mass of @xmath112 after gas ejection .
we here assume that each agb particle in a model eject sph particles at @xmath113 when the simulation starts : we do not consider gradual ejection of gas from agb stars at each time step , because an appropriate modeling of such gradual ejection requires a huge number of gas particles and thus is numerically costly ( i.e. , practically not feasible ) .
although we have investigated models with @xmath114k , we mainly describe the results of the models with @xmath115k which corresponding to star - forming warm molecular clouds ( e.g. , wilson et al .
this is because we consider that agb wind can cool down during expansion throughout interstellar space due to radiative cooling ( to finally become molecular gas for further star formation ) so that @xmath111 can becomes much smaller than the original temperature of the wind ( @xmath116 1000k ) .
we adopt an isothermal equation so that agb ejecta can keep initially low temperatures ( @xmath117k ) .
it should be stressed here that if @xmath118k is adopted ( which is not realistic though ) , star formation is possible only for models with @xmath119 . most of agb ejecta can be converted into new stars well before a few myr after commencement of secondary star formation so that feedback effects of massive stars and type ii supernova can not significantly influence gas dynamics during the secondary star formation processes .
we therefore consider that the above isothermal assumption is reasonable . in order to estimate value of @xmath109 corresponding to the total mass ejected from each agb star after the main - sequence turn - off
, we use the formula given in the equation ( 3 ) . about 83% of a agb particle with @xmath120
can be ejected to be used for further star formation in the present model .
if we adopt the salpeter imf , then @xmath105 ( the mass fraction of agb stars with masses larger than @xmath57 ) is @xmath121 in the present study . since initial masses within a msc
are the same between stellar particles , @xmath122 .
although we adopt this @xmath123 for most models , we investigate models with different @xmath105 so that we can find a threshold @xmath105 for a given msc above which secondary star formation can occur in the msc .
it is found that if @xmath105 is larger than 0.008 , secondary star formation is possible in some models , though the star formation efficiency is low .
this means that secondary star formation within a msc can not be possible until a certain amount of gas is accumulated within the msc and thus that there should be a time - delay between the formation of fg stars and that of sg ones .
we investigate whether the gas accumulated in the central regions of mscs can be sufficient to form new stars by adopting a simple prescription for star formation . in the models with `` star formation '' ,
a gas particle is converted into a collision - less new stellar one if the gas particle meets the following three conditions : ( i ) the dynamical time scale of the sph gas particle is shorter than the sound crossing time , ( ii ) the gas is converging ( i.e. , @xmath124 , where @xmath125 is the velocity vector of the gas particle ) , and ( iii ) the local gas density exceeds a threshold gas density ( @xmath126 ) which corresponds to the densities of dense cores of molecular clouds ( @xmath127 atoms @xmath72 ) , where individual star formation is ongoing .
the first two conditions mimic the jeans gravitational instability for gaseous collapse .
the new stellar particles and old ones initially within mscs are referred to as `` new stars '' ( i.e. , sg stars ) and `` old stars '' ( i.e. , fg ones ) , respectively .
also original clusters and new ones formed from agb ejecta are referred to as `` old clusters '' and `` new clusters '' , respectively , just for convenience .
we investigate models with @xmath128 and @xmath129 atoms @xmath72 in order to investigate ( i ) whether the threshold gas density is important for determining star formation histories of sg stars and ( ii ) how the present results , in particular , structural and kinematical properties of final clusters composed of sg stars , depend on @xmath126 : we here consider that @xmath126 has not been observationally well determined yet and therefore could be different for individual star forming clouds .
although these star formation models are less realistic in some points ( not inclusive of magnetic fields and radiative transfer etc ) , we consider that they enable us to grasp essential ingredients of secondary star formation from agb ejecta within original massive clusters .
we mostly investigate `` isolated mscs '' which are located outside nuclei of their host galaxies and therefore short - term ( @xmath130 yr ) evolution of agb eject can not be strongly influenced by the tidal fields of their hosts .
however , dynamical evolution of agb ejecta from nuclear mscs that are initially located in the very centers of their hosts can be significantly influenced by their hosts owing to the much deeper gravitational potential wells of the hosts .
we therefore investigate how hosts influence evolution of agb ejecta from nuclear mscs by assuming that the host dwarfs are dominated by dark matter halos with masses of @xmath15 and thus their gravitational potentials are determined only by the dark halos .
we adopt the density distribution of the nfw halo ( navarro , frenk & white 1996 ) suggested from cdm simulations : @xmath131 where @xmath132 , @xmath133 , and @xmath134 are the spherical radius , the characteristic density of a dark halo , and the scale length of the halo , respectively .
although we investigate models with @xmath135 , we mainly describe the results of the models with @xmath136 , @xmath137 kpc , and the virial radius ( @xmath138 ) of 9.9 kpc ( i.e. , the @xmath139-parameter of 13.9 ) .
these models with @xmath136 are reasonable in the present scenario in which there is a threshold galaxy mass ( @xmath16 ) above which gcs can be formed
. it should be here stressed that low - mass dark matter halos with @xmath140 can influence evolution of agb ejecta in hosts as significantly as those with @xmath136 .
firstly we investigate ( i ) whether agb ejecta can be retained within mscs and ( ii ) whether and how secondary star formation from agb ejecta can proceed for models that are consistent with the scaling relation of gcs shown in the equation ( 2 ) .
secondly we investigate structure and kinematics of the simulated clusters in the models with @xmath11 and @xmath141pc in which the final masses of sg stars can be as large as @xmath142 .
thirdly we investigate whether much deeper gravitational potential wells of host dwarf galaxies can play a role in retaining agb ejecta efficiently in the nuclear mscs .
we have investigated secondary star formation processes for numerous models with different model parameters ( e.g. , @xmath0 , @xmath18 , @xmath126 , and @xmath93 ) ; we mainly describe here the results for the `` standard model '' with @xmath143 , @xmath144pc , @xmath145 , and @xmath146 atoms @xmath72 .
we assume that @xmath147 for most models except those for investigating how agb ejecta can be retained within mscs .
the reason for this adoption is that as long as @xmath148 , the results do not depend on @xmath108 .
since we focus on secondary star formation processes with a timescale of @xmath149 yr , we do not intend to discuss long - term dynamical evolution of clusters due to two - body relaxation .
we accordingly introduce a gravitational softening length ( @xmath150 ) for each simulation and @xmath150 is set to be equal to the mean particle separation at the half - number radius of old stars .
the ranges of model parameters are shown in the table 1 .
we describe in detail only the results of some representative models showing key parameter - dependences in retention processes of agb ejecta and secondary star formation processes within clusters among the investigated models .
secondary star formation processes can continue only for @xmath130 yr so that they can not be strongly influenced by gravitational fields of their hosts unless mscs are located in the very centers of their hosts .
we thus show mainly the results for models without gravitational fields of hosts in order to much more clearly show the importance of the four key parameters @xmath0 , @xmath18 , @xmath93 , and @xmath126 in secondary star formation processes .
as described in 3.2 , we do not consider that different agb stars with different masses eject gas at different @xmath151 owing to different lifetimes of the stars .
therefore all of the gaseous ejecta can be converted into new stars within @xmath152 myr .
this rapid consumption of the agb ejecta is not so realistic , given that there should be at least several tens millions years time delay between the epochs when agb stars with masses of @xmath153 and @xmath57 start to eject their gas .
if we consider the different epochs of gas ejection from agb stars with different masses , then the star formation period would become significantly longer
. however , energy feedback effects of numerous supernovae from the sg stars are likely to truncate star formation in gaseous ejecta from lower - mass agb stars . in our future papers
, we will discuss this point in detail by using a more sophisticated numerical model .
the present model does not include the effect of type ia supernova ( sn ia ) on star formation processes of sg stars .
dercole et al .
( 2008 ) already showed that the cumulative effect of numerous sn ia is important for the star formation processes within massive clusters , because it can drastically alter gas dynamics within the clusters .
therefore the present model that does not include such sn ia effect possibly overestimates the star formation rates in the accumulated agb ejecta in the central regions of mscs .
our future more sophisticated models with sn ia effect will discuss how sn ia can influence gas dynamics and star formation in mscs .
fig . 2 show that time evolution of spatial distributions of agb ejecta is very different between models with @xmath154 and @xmath155 in which model parameters except @xmath0 are exactly the same .
gaseous particles with @xmath156 ejected from an agb star locating in the very center of a msc can expand quickly ( @xmath157myr ) and finally escape from the cluster and never return back in the model with @xmath154 ( @xmath158myr ) .
on the other hand , gaseous ejecta from an agb star in the model with @xmath155 can return back to the initial location within a timescale of a few myr after the ejection owing to the deeper gravitational potential for this more massive cluster .
this clear difference suggests that @xmath0 is an important parameter for determining whether agb ejecta can be retained in mscs during their early formation phases .
3 shows how the mass fractions of agb ejecta retained within mscs ( @xmath159 ) depend on @xmath0 .
3 shows that mscs with @xmath160 can retain a significant fraction of agb ejecta if a scaling - relation similar to the observed one is applied for all mscs .
this result implies that there can be a threshold mass above which agb ejecta can be efficiently retained .
for mscs with @xmath161 , agb ejecta initially in the outer part of the msc can escape from the msc whereas that initially in the inner part can be retained within the msc . as a result of this , a `` core - halo '' structure of agb ejecta can be formed within 13.6 myr . a very compact gas sphere can be formed where star formation would be able to proceed rapidly if star formation is included in the model with @xmath162
. a larger fraction of agb ejecta can be retained in mscs with larger @xmath0 .
gaseous spheres composed of agb ejecta can interact with one another within a msc when numerous stars enter into agb phase almost simultaneously .
the present 3d numerical simulations enable us to investigate whether and how hydrodynamical interaction between the gaseous spheres increase or decrease the total gas mass accumulated within the central region of the msc .
4 shows that if two agb stars with each having @xmath156 are included in the model with @xmath163 ( where no gas accumulation is seen in fig .
2 ) , the gas can lose kinetic energy owing to energy dissipation during hydrodynamical interaction of the gaseous spheres so that a non - negligible amount of gas can be accumulated within the cluster .
about 3.6% of the initial agb ejecta can be accumulated within the central 50pc of the cluster within 13.6 myr .
5 shows how @xmath164 , where @xmath103 are the total numbers of agb stars , depends on @xmath103 : this normalized @xmath165 can measure the efficiency of retaining agb ejecta within mscs .
it is clear from this figure that agb ejecta can be more efficiently retained within clusters for models with larger @xmath103 .
this is mainly because as the larger number of gaseous spheres ( of agb stars ) hydrodynamically interact with one another , the larger amount of the gas can lose their kinetic energy to be accreted onto the central regions of mscs .
the results in figs . 4 and 5 therefore demonstrate that hydrodynamical interaction of agb ejecta can cause efficient gaseous dissipation and thus play a great role in retaining agb ejecta within mscs .
6 shows that ( i ) there is a threshold @xmath0 ( @xmath166 in this parameter study ) above which agb ejecta can be retained effectively and ( ii ) @xmath167 is larger for larger @xmath0 .
the reason of these dependences is that mscs with larger @xmath0 have deeper gravitational potentials so that they can retain agb ejecta more effectively .
differences in @xmath165 between @xmath168pc and @xmath169pc are small for models with @xmath170 , which means that most the accumulated gas can form very compact gaseous regions within the central regions of mscs . thus mscs with larger @xmath0 can obtain a larger amount of agb ejecta for further star formation : @xmath0 is a key for determining whether mscs can have sg stars .
7 shows how gas ejected from agb stars is accumulated within the central region of a msc and consequently converted into new stars ( i.e. , sg ones ) there within 13.6 myr in the standard model with @xmath143 , @xmath144pc , @xmath145 , and @xmath146 atoms @xmath72 .
the gaseous ejecta from agb stars initially located in the central region of the msc can be accumulated there within @xmath62myr so that star formation can start efficiently .
the star formation from agb ejecta can proceed like a `` starburst '' with most gas being consumed rapidly within the first 3 myr evolution of the msc .
the star formation rate ( sfr ) reaches its maximum value ( @xmath171 yr@xmath21 ) at @xmath172 myr and then rapidly declines to be @xmath173 yr@xmath21 ) at @xmath174 myr .
about 86% of the gas can be converted into new stars within 13.6 myr in this model .
this very high star formation efficiency ( @xmath25 ) results from the fact that almost all of the agb ejecta can be accumulated very quickly ( in less than a few @xmath175 yr ) within the central region of the msc owing to the deep gravitational potential and consequently converted into new stars there .
the accumulated gas is strongly self - gravitating ( i.e. , gas mass comparable to stellar mass there ) in the central region of the cluster so that the gas can continue to collapse to form new stars .
this result clearly suggests that _ secondary star formation within mscs _ is responsible for the origin of the observed high stellar densities of the present gcs : a gmc can not be converted into a gc just by one star - formation event .
a new compact star cluster embedded in low - density residual gas can be formed in the central regions of the msc within 13.6 myr in this model .
owing to the high star formation efficiency ( @xmath176 ) , the new cluster is highly likely to survive from gas removal by energetic winds of massive stars and supernova explosion . as shown in fig .
8 , the new cluster is much more compact than the original msc and has a half - mass radius ( @xmath177pc ) significantly smaller than that of the original msc ( @xmath178pc ) . owing to its compactness
, the new cluster is much less susceptible to tidal destruction by its host galaxy .
thus a `` nested stellar system '' composed of a diffuse original cluster ( fg stars ) and a compact new one ( sg ones ) can be formed as a result of secondary star formation from agb ejecta within the msc in this model .
this nested structure is one of common features of the simulated mscs in the present study .
fig.9 shows how the fractions ( @xmath179 ) of the total mass of new stars ( @xmath180 ) to the initial total mass of the msc ( @xmath0 ) depend on @xmath0 for @xmath168pc and @xmath169pc in the models with @xmath146 atoms @xmath72 .
clearly there is a threshold cluster mass ( @xmath1 ) above which secondary star formation is possible : @xmath181 and does not depend on @xmath126 in the present study .
furthermore , @xmath179 is higher in mscs with larger @xmath0 , which implies that @xmath25 is also higher in mscs with larger @xmath0 .
almost all stars can be formed within half - mass radii ( @xmath152pc ) of mscs with @xmath182 so that there can be no differences between @xmath179 for @xmath168pc and @xmath169pc .
although @xmath25 is rather high ( 0.86 ) in the model shown in figs . 7 and 8 , the total mass of new stars can be at most @xmath183 owing to the adopted @xmath123 that is reasonable for the standard imf .
this means that mscs with @xmath184 can become low - mass gcs with multiple stellar populations .
furthermore , @xmath25 is 0.3 for the model with @xmath185 , which suggests that the new cluster can significantly expand after gas expulsion due to energetic winds of massive stars and supernova .
the cluster is likely to become a low - mass , low - density gc with a smaller fraction of sg stars ( like pal 5 ) .
the dependence of @xmath25 on @xmath0 implies that more massive gcs have larger fractions of sg stars .
the details of star formation histories depend weakly on @xmath93 and @xmath126 such that gas can be more rapidly converted into new stars in mscs with smaller @xmath93 and @xmath126 .
10 shows that overall star formation rate is higher in the model with @xmath186 ( @xmath187 ) than in that with @xmath188 ( @xmath189 ) .
this is mainly because initial angular momentum of the msc in the model with @xmath188 can suppress the formation of a very compact gaseous region in the center of the msc .
the secondary peak of the star formation rate in the model with @xmath188 at @xmath190 myr is due to later infall of gas onto the msc s center where a compact gas disk is formed .
the formation of nested structures with very compact new clusters and initially more diffuse old ones is one of essential ingredients of the present simulations .
11 , which shows a typical example of the nested structures , describes how the total stellar mass within @xmath94 depend on @xmath94 for old and new stars in the standard model .
clearly the final cluster is dominated by new stars ( sg ) within @xmath191pc , and the faction ( @xmath192 ) of the total mass of new stars ( @xmath193 ) to the total mass of old stars ( @xmath194 ) and new ones is more than 0.5 for @xmath191pc . fig .
12 shows how @xmath192 depends on @xmath94 for different models with different @xmath126 and @xmath93 .
irrespective of these model parameters , @xmath192 is higher in the inner regions of mscs and lower in the outer ones .
@xmath192 can be higher for higher @xmath126 and smaller @xmath195 . as discussed in previous sections ,
@xmath0 should be at least @xmath196 to explain gcs with @xmath197 ( for @xmath123 ) .
given that @xmath25 is not 1 ( @xmath4 for most models ) and later dynamical evolution can reduce the total masses of mscs , stellar structures and kinematics of the simulated mscs with rather large @xmath0 ( @xmath198 ) can correspond to those of the observed ones of gcs with masses larger than @xmath142 .
13 shows the final spatial distributions of old and new stars in the model with @xmath11 , @xmath141pc , @xmath199 , and @xmath146 atoms @xmath72 .
clearly the compact new cluster in the center of the msc is significantly flattened owing to gaseous dissipation during gas accretion onto the central region .
this significantly flattened shape can be seen in new stars for most models in which @xmath93 is not 0 ( i.e. , even the models with @xmath200 show such flattened shapes ) . the total mass of new stars ( @xmath193 ) is larger than @xmath201 and the effective radius ( @xmath202 ) of the stars is 5.2pc , which is about a factor of 5 smaller than that of the old ones ( @xmath203 ) .
the half - mass ratios ( @xmath204 ) are very similar ( @xmath205 ) in the present models with different model parameters . as shown in fig .
14 , the new stars dominate the inner region ( @xmath206 2pc ) of msc , though @xmath192 there can not become so high as that in the standard model .
it is confirmed that more massive models with @xmath207 with different @xmath93 and @xmath126 also show distinct nested structures .
15 shows clear differences in stellar kinematics between old and new stars ( thus fg and sg ones , respectively ) in the final msc for the model .
new stars clearly have rotational kinematics with the maximum rotational velocity ( @xmath208 ) of 12.0 km s@xmath21 and the smaller central velocity dispersion ( @xmath209 ) of 9.5 km s@xmath21 , which means @xmath210 .
this rotational kinematics can not be seen in old stars , which have @xmath211 in this model .
the models with larger @xmath93 show larger @xmath212 in new stars , which has already been discussed by bekki ( 2010 ) .
these results suggest that the present gcs originate from mscs composed of two stellar populations with significantly different kinematics .
16 shows the time evolution of the projected spatial distribution of agb ejecta for a nuclear msc in the model without star formation yet with the external gravitational field of its host galaxy dominated by dark matter halo ( @xmath136 ) . in this model ,
only one gaseous sphere of a agb star is included and located initially in the center of a nuclear msc with @xmath213 ( @xmath145 ) so that the results can be compared with the model with @xmath154 shown in fig .
2 : it is much better to use these two comparative models with only one gaseous sphere of a agb star , because the effects of msc s host galaxy can be more clearly demonstrated . as shown in fig .
16 , although the gas sphere expands initially to a larger radius ( @xmath214pc ) , the gas can finally return back to the central region of the msc within @xmath152 myr owing to the deep potential well of its host galaxy . fig .
17 shows that @xmath165 in the models with external gravitational fields of hosts do not depend so strongly on @xmath0 as those without the fields do : irrespective of @xmath0 , most of agb ejecta can be retained within the central 50pc of mscs in the models with external gravitational fields of their host owing to their much deeper gravitational potential wells .
however , @xmath165 within @xmath215pc depends more strongly on @xmath0 such that it is larger for larger @xmath0 .
this results indicates that inner gas densities ( @xmath168pc ) can not be so high in nuclear mscs with lower @xmath0 so that star formation can not occur efficiently in the mscs .
it is actually confirmed that star formation can not occur efficiently in the low - mass models with @xmath216 , even if external gravitational fields are included .
thus the original masses of nuclear mscs are still important for secondary star formation processes even if they are located in nuclear regions of their host galaxies
. external gravitational fields can enhance star formation of nuclear mscs , in particular , for those with larger @xmath0 and lower mass - densities .
this enhancement can be most clearly seen in fig .
18 which describes the time evolution of star formation rates for two models with @xmath143 , @xmath217pc , @xmath145 , @xmath128 atoms @xmath72 , and with and without the external gravitational field from a host with @xmath218 .
star formation can start earlier and proceed more efficiently in the model with the external gravitational field ( @xmath219 ) than in the model without ( @xmath220 ) .
these differences between two models with and without external gravitational fields are less remarkable for high - mass and high - density mscs , where star formation can proceed efficiently even in isolation .
in the present scenario , original mscs need to keep substantial masses at least for @xmath23 yr after they form within gmcs so that gaseous ejecta from more massive agb stars can be retained in mscs and finally converted into new stars ( = sg ones ) .
the hosts ( = massive dwarfs ) can strip stars from mscs owing to their tidal fields , even if the fields are not so strong as that of the galaxy .
therefore we here investigate how long mscs can keep their substantial fractions of stars when they are under strong gravitational influences of their hosts .
we consider how a msc dynamically evolves after most of residual gas left behind from the formation of fg stars are expelled from the cluster through supernova events .
accordingly we adopt an assumption that ( i ) the msc is assumed to consists purely of stellar particles and ( ii ) it is not necessarily in virial equilibrium owing to loss of the original mass of the gmc . the cluster with a mass @xmath0 and a size @xmath18 has a plummer density profile with the scale length @xmath79 ( @xmath86 ) .
owing to the above ( ii ) , the central velocity dispersion @xmath76 of the msc can be significantly larger than the one ( @xmath221 ) which the msc can have when it is in virial equilibrium .
therefore we introduce a parameter @xmath222 that is the ratio of @xmath76 to @xmath221 and ranges from 1.0 ( virial equilibrium ) to 1.3 .
thus mscs can expand significantly for models with @xmath223 .
the cluster is influenced by the host with the density distribution of the nfw halo ( i.e. , equation ( 7 ) ) .
the initial location of the msc with respect to the center of the host is set to be ( @xmath224 , @xmath225 , @xmath226 ) .
the initial velocities of the msc in the @xmath227- , @xmath228- , and @xmath229-directions are set to be ( @xmath230 , @xmath231 , @xmath232 ) .
the orbital plane of the msc is coincident with the @xmath227-@xmath228 plane for all models in the present study ( i.e. , @xmath233 and @xmath234 ) .
the initial direction of the velocity vector of the msc is only a parameter for the orbit of the msc owing to the spherically symmetric distributions of the host and the msc .
we therefore assume that the initial direction is parallel to the @xmath228-axis toward the positive @xmath228 ( i.e. , @xmath235 and @xmath236 ) .
we assume @xmath237 , and thus @xmath224 and @xmath231 are free parameters that determine the orbit of the msc .
we consider that the orbit of a msc is not necessarily circular within its host and therefore introduce a free parameter @xmath238 that is the ratio of @xmath230 to the circular velocity @xmath239 at @xmath224 and ranges from 0.5 to 1.0 ( i.e. , circular orbit ) .
we mainly investigate two representative cluster models with @xmath143 and @xmath217pc and with @xmath11 and @xmath141pc for different model parameters @xmath15 , @xmath222 , @xmath224 , and @xmath238 .
although we run numerous models , we mainly describe the results of the models with @xmath218 and @xmath240 kpc ( @xmath241 ) . for each model , we count the total number of stars that are located within 100pc from the center of the msc at each time step in order to investigate the time evolution of the total stellar mass of the msc .
stars that are once stripped and by accident located within the central 100pc without being bound by the msc can be counted as those in the msc .
we consider that since the number fraction of these stars is very small , these stars can only slightly overestimate the total stellar mass of a msc .
the initial stellar mass for a msc is @xmath242 so that we can estimate the time evolution of the _ stellar mass _ for the msc .
fig.19 shows that the host of a msc can gradually remove stars from the outer part of the msc owing to the strong tidal field in the model with @xmath143 , @xmath217pc , @xmath243 , @xmath244 , @xmath245kpc , @xmath246pc , and @xmath247 .
about 60% of original stars can be stripped within @xmath62 gyr and the stripped stars are distributed widely within the original orbital plane and thus regarded as `` field stars '' .
although the mass and the size of the msc can become much smaller than the original ones , the msc is still strongly bound at @xmath248 gyr .
the stripping process is gradual in this model : only 25% of the original mass of the msc can be stripped within @xmath249 gyr .
thus , agb ejecta might well be retained with the msc to finally converted into new stars ( = sg ones ) in this model .
20 shows how the time evolution of total stellar masses within 100pc of mscs depends on @xmath222 and @xmath134 for models with @xmath143 , @xmath217pc , @xmath244 , @xmath246pc , and @xmath247 .
it is clear from this figure that stars can be more rapidly stripped from mscs in the models with larger @xmath222 ( = 1.2 ) and those in more compact hosts ( @xmath250kpc ) .
the msc in the model with @xmath251 and @xmath250kpc can lose @xmath252% of its original mass within @xmath253 gyr owing to the combined effect of the more rapidly expanding stellar system and the stronger tidal field of the host . in this mode
, agb ejecta would not be so efficiently retained and consequently converted into new stars within the msc .
the mass fraction of stripped stars ( @xmath254 ) is larger for models with smaller @xmath134 and larger @xmath222 .
21 shows how the time evolution of total stellar masses within 100pc of mscs depends on @xmath222 and @xmath134 for models with @xmath11 , @xmath141pc , @xmath244 , @xmath246pc , and @xmath247 .
the results shown in figs 20 and 21 indicate that stars can be more slowly and less efficiently removed from msc with larger @xmath0 owing to their stronger gravitational potential wells .
mscs with @xmath11 can not lose more than 50% of their original stellar masses within @xmath62 gyr so that agb ejecta can be retained within the mscs and converted into new stars .
these results imply that original masses of the present gcs with multiple stellar populations should be rather large to keep substantial fractions of their original stellar masses to be immune from tidal destruction by their host within @xmath253 gyr .
initial densities of mscs and @xmath224 also can determine @xmath254 for a given set of parameters @xmath0 , @xmath222 , @xmath15 , @xmath134 , and @xmath238 .
mscs with lower densities and smaller @xmath224 are more susceptible to tidal destruction of their hosts so that they can lose more stars within 1 gyr ( i.e. , larger @xmath254 ) .
thus the time evolution of stellar masses of mscs depends strongly on a number of parameters ( e.g. , @xmath0 and @xmath222 ) so that the time scale within which gaseous ejecta from agb stars can be retained and converted into new stars can also depend on the parameters .
the present parameter study , however , indicates that original mscs are unlikely to be destroyed completely within a timescale of @xmath23 yr for most models : most mscs can slowly ( @xmath61 gyr ) disintegrate owing to the strong tidal fields of their hosts .
we thus suggest that secondary star formation from agb ejecta within mscs is highly likely for most mscs with @xmath255 and can last at least @xmath23 yr .
we have shown that @xmath0 can determine whether agb ejecta from fg stars of mscs can be retained within the mscs and consequently converted into new stars efficiently .
22 briefly summarizes the present key results on how @xmath0 controles @xmath25 for the secondary star formation . mscs with larger @xmath0 can have higher @xmath25 and we have confirmed that his dependence does not depend on @xmath93 and @xmath126 .
mscs with more rotation in fg stars are more likely to show smaller @xmath25 .
secondary star formation is possible if @xmath0 is larger than @xmath256 , though @xmath25 is not so high . as long as @xmath257 , @xmath25 can be larger than 0.5 so that new compact clusters , which would not disintegrate even after gas expulsion , can form within old clusters .
thus our results have shown that there can be a threshold cluster mass ( @xmath258 \times 10 ^ 5 { \rm m}_{\odot}$ ] ) above which secondary star formation with high @xmath25 can occur within original clusters .
this @xmath259 can be as low as @xmath260 , if the original clusters are @xmath152 times more compact than the scaling - relations described by the equations ( 5 ) and ( 6 ) imply .
however @xmath25 is at most 0.1 even for such compact clusters with lower masses . on the other hand ,
even if mscs with @xmath261 are @xmath262 times more diffuse than the scaling - relations described by the equations ( 5 ) and ( 6 ) imply , their @xmath25 can be as large as 0.1 .
therefore , it is reasonable to claim that @xmath1 is @xmath263 .
dercole et al .
( 2008 ) showed that secondary star formation within original clusters is possible for @xmath11 but not for @xmath143 .
this implies that @xmath1 can be somewhere between @xmath264 and @xmath14 and is significantly larger than the derived @xmath1 in the present study .
this is mainly because modeling of star formation and treatment of agb ejecta are different between these two works . both models in dercole et al .
( 2008 ) and the present study have advantages and disadvantages in describing evolution of agb ejecta within clusters .
for example , the present model enables us to investigate self - consistently 3d hydrodynamical interaction of numerous gaseous spheres from agb stars whereas the model by dercole investigated cooling of initially warm agb ejecta self - consistently .
more sophisticated models in our future papers need to investigate @xmath1 in a fully self - consistent manner by including physical effects that are treated by the present model in a simple way .
we here discuss how @xmath0 determines nature of multiple stellar populations in gcs owing to existence of @xmath1 .
we propose that @xmath0 is one of key parameters which can determine differences in ages and various chemical abundances of light and heavy ones between multiple stellar populations .
based on the present numerical results , we divide mscs into the following categories according to their @xmath0 .
it should be stressed here that @xmath0 for a cluster is not the present mass ( @xmath12 ) but its original one .
these low - mass clusters can neither retain agb ejecta nor convert gas into new stars owing to their shallow gravitational potential wells .
therefore these clusters are highly unlikely to show abundance spread even in light elements .
observations showed that the galactic open cluster ( ocs ) show unimodal distributions of cn band strength ( martell & smith 2009 ) and do not have abundance spread in light elements ( e.g. , carretta et al . 2007 ) and o - na anti - correlation ( de silva et al .
these low - mass clusters can be easily destroyed by interaction with gmcs ( e.g. , gieles et al .
2006 ) so that they can not obtain fresh cold gas for further star formation .
the observed low - mass ocs in the galaxy and the magellanic clouds belong to this class .
these clusters can not retain agb ejecta efficiently within their central regions so that secondary star formation can not occur .
however , they can obtain fresh cold gas for further star formation by capture and accretion of cold molecular gas that are either those left behind from previous star formation events or those located close to them ( e.g. , bekki & mackey 2009 ) .
therefore , if secondary star formation occurs within these clusters , they can only show multiple stellar populations with age differences yet no / little abundance spread even in light elements .
the intermediate - age clusters in the lmc and the smc belong to this class . clusters with this mass range can retain most of their agb ejecta within the central regions but can not convert the stars so efficiently ( @xmath268 ) into new stars within a short time scale ( @xmath269 ) yr . although dynamical fate of the accumulated agb ejecta for the cluster remains unclear , `` triggering mechanisms '' such as cluster mergers ( or initially very high - density clusters ) would be necessary to convert the gas so efficiently into new stars .
strong ram pressure stripping of the gas by ism of their host galaxies could remove the gas from the clusters .
the ordinary galactic gcs with abundance spread in light elements can originate from clusters with this mass range , because agb ejecta can be retained in the clusters and converted into new stars very efficiently ( @xmath271 ) .
the mass fraction of sg stars in a present gc depends mainly on @xmath0 and dynamical evolution of its original cluster ( i.e. , destruction by its host and long - term evolution driven by two - body relaxation ) .
if these massive clusters are initially stellar nuclei of massive dwarf galaxies , then agb ejecta from sg stars can be converted into new stars ( i.e. , the formation of third generation of stars ) .
this sequential formation of stars in the nuclear regions of hosts can continue until the hosts are destroyed by much larger galaxies .
very massive clusters with this mass range are progenitors of the present massive gcs with masses larger than @xmath264 .
stellar systems with @xmath273 would be dwarf galaxies themselves rather than clusters , and therefore at least some fraction of massive gcs ( e.g. , @xmath74 cen and g1 ) may well the remnants of nucleated dwarf galaxies .
owing to much deeper gravitational potential wells of nucleated dwarfs , not only agb ejecta but also gas from more energetic massive stars and supernova would be able to be retained and finally converted into new stars .
thus such massive gcs can show abundance spread in both light and heavy elements . using analytical models , pflamm - altenburg & kroupa ( 2009 )
recently have shown that if the mass of a cluster exceeds a threshold mass ( @xmath274 ) , then the cluster can obtain a significant amount of warm ism from outside the cluster by bondi accretion within a reasonable time scale ( i.e. , a few gyr ) .
we have confirmed their results using 3d hydrodynamical simulations , though the threshold mass for gas accretion is larger ( @xmath275 ) in our estimation ( bekki et al .
we suggest that although secondary formation from mixed gas of agb ejecta and the accumulated ism by bondi accretion would be reasonable for massive gcs with possible age differences ( like @xmath74 cen ) , such star formation has the following potentially serious problems in explaining chemical abundances of _ ordinary _ galactic gc .
firstly , if original clusters are within galaxies , then they can obtain gas initially located in different regions with possible different chemical abundances ( owing to radial and azimuthal abundance gradients with the galaxies ) thus are highly likely to have gas and new stars with different [ fe / h ] .
therefore , only a minor fraction of the galactic gcs may well have sg stars formed from mixed gas of agb ejecta and ism obtained by bondi accretion .
secondly , the bondi accretion rate is too small ( @xmath276 yr@xmath21 for @xmath277 ) for typical ism so that an enough amount of sg stars can not form from gas chemically polluted only by massive agb stars ( @xmath278 ) within an order of @xmath90 yr .
it is possible that if densities of warm ism are as large as @xmath279 atoms @xmath72 , then clusters would be able to obtain an enough amount of gas within a timescale of less than @xmath90 yr ( but if this is the case , then such high - density gas may well be cold molecular clouds rather than warm ism ) .
thirdly , even if warm ism with initial temperature of @xmath280k can be accreted by clusters , further efficient radiative cooling ( to 10 - 100k ) within a short timescale ( @xmath269 yr ) is required for further star formation .
our previous and present study have not yet investigated how warm ism and agb ejecta mixed with and consequently convert into cold molecular gas for further star formation using a sophisticated model , and accordingly we can not currently discuss the possibility of secondary star formation from the accumulated warm ism .
our future more sophisticated models will investigate whether these three are really serious problems .
it is possible that the original mscs would have a plenty of cold molecular gas left behind from the formation of fg stars , because the star formation efficiencies are highly likely to be well less than 100% . given that gmcs which are progenitors of fg stars should be very massive ( @xmath281 ) in the present scenario , then the host gmcs would have numerous substructures ( i.e. , smaller molecular clouds ) that remains intact even after the formation of fg stars .
these molecular clouds can be captured by or accreted onto mscs , as a previous simulation demonstrated already ( bekki & mackey 2009 ) . owing to initially low temperatures of the captured cold gas , star formation from gas mixed between agb ejecta and the captured gas
would proceed efficiently .
a potential problem of this process is that it remain unclear whether cold molecular clouds left behind the formation of fg stars can really keep the same abundances as those of fg stars ( i.e. , remain `` pristine '' ) without being chemically polluted by massive stars and supernova of fg stars .
a significant fraction of the residual cold gas would be heated up to become high - temperature and low - density warm and hot gas so that they can not be captured later by mscs owing to their high - temperature and low - density .
owing to clumpy structures of the original massive gmcs , a significant amount of energy from massive stars and supernova can be expelled from low - density inter - substructure regions of the gmcs toward the halos of their hosts .
some smaller cold molecular clouds that can survive from thermal and kinetic feedback effects of energetic stars would be able to be captured later by mscs and consequently mixed with agb ejecta .
the above processes of mixing of cold molecular gas and agb ejecta have not been investigated in detail by any previous 3d hydrodynamical simulations of star - forming gmcs .
bekki & chiba ( 2007 ) investigated ( i ) how massive stars and type ii supernova influence gas dynamics within star - forming gmcs and ( ii ) chemical abundances of gas left behind from star formation .
they however did not include ejection of gas so that they could not investigate how agb ejecta mixed with the residual gas to form new stars .
it is thus doubtlessly worthwhile for our future more sophisticated models to investigate whether and what fraction of residual molecular cloud left behind from the formation of fg stars can be converted into new stars without being chemically polluted by massive fg stars .
a number of recent observational studies have found that a significant fraction of intermediate - age cluster in the lmc and the smc show double main sequence turnoffs ( hereafter referred to as dmstos for convenience ) on their color magnitude diagrams ( mackey & broby nielsen 2007 ; mackey et al .
2008 ; glatt et al .
2008 ; goudfrooij et al .
2009 ; milone et al .
the simplest explanation is that the observed dmstos in each of these clusters represent two distinct stellar populations with differences in age of @xmath282 myr .
recently bastian & de mink ( 2009 ) have proposed that stellar rotation in stars with masses between 1.2 and 1.7@xmath283 can mimic the effects of dmstos without resorting to age differences between stellar populations within clusters of the magellanic clouds .
however , the presence of a dual red clump of giant stars in the color magnitude diagram of ngc 419 in the smc is suggested to be inconsistent with the claim by bastian & de mink ( 2009 ) but be explained naturally by the presence of multiple star - formation episodes ( rubele et al .
2010 ) .
mucciarelli et al . (
2008 ) investigated chemical abundances of light odd - z , @xmath33 , iron - peak , and neutron - capture elements for four intermediate - age clusters in the lmc and found negligible star - to - star scatter for them .
however , mucciarelli et al . ( 2009 ) revealed significant abundance inhomogeneities in [ na / fe ] , [ al / fe ] , [ o / fe ] , and [ mg / fe ] and o - na and mg - al anti - correlations in old gcs ( @xmath284 gyr ) .
these observations imply that intermediate - age clusters have ( at least ) two generations of stars with different ages yet similar chemical abundances whereas old ones are like the galactic gcs with abundance spread in light elements . therefore
, any theory of gc formation need to explain why the lmc appears to have two different types of gcs with multiple stellar populations .
bekki & mackey ( 2009 ) showed that low - mass clusters ( @xmath285 ) can capture cold gmcs and thereby convert the gas into new stars within their central regions in the lmc .
the present study has shown that such low - mass clusters can not efficiently retain agb ejecta and thus are unable to use the ejecta for further star formation .
the results by bekki & mackey ( 2009 ) and the present study therefore combine to suggest that if the observed dmstos in clusters of the lmc are due to age differences of stellar populations , then the two populations are highly unlikely to show abundance spread in light elements owing to their low masses ( @xmath286 ) .
the present study also suggests that the old clusters of the lmc can show abundance spread in light elements , mainly because their original clusters are significantly more massive than the present ones with masses ranging from @xmath201 to @xmath256 . recently
conroy & spergel ( 2010 ) have suggested that clusters with masses larger than a threshold mass of @xmath287 can form sg stars from agb ejecta mixed with ism accreted onto clusters . the observed lack of abundance inhomogeneity in light elements in the intermediate - age clusters of the lmc ( mucciarelli et al .
2008 ) seems to be inconsistent with their model .
the present study suggests that the threshold mass for the formation of sg stars from agb ejecta mixed with pristine gas is much larger than the one suggested by conroy & spergel ( 2010 ) .
old and metal - poor ( [ fe / h]@xmath288 ) clusters in the lmc have masses more than @xmath201 ( mackey & gilmore 2003 ) and would have lost a significant fraction of their masses due to long - term internal evolution effects and external tidal field of the lmc . therefore recent observations on star - to - star abundance variations in light elements in old and intermediate clusters of the lmc by mucciarelli et al .
( 2008 , 2009 ) strongly suggest that the threshold cluster mass ( @xmath1 ) for secondary star formation from agb ejecta mixed with pristine gas would be much more than @xmath201 .
these observations are consistent with the present model with @xmath289 and the one by dercole et al .
( 2008 ) which showed that clusters with @xmath290 can form sg stars from agb ejecta .
dercole et al .
( 2008 ) first claimed that original single stellar systems ( `` clusters '' ) composed purely of fg stars should have masses of @xmath291 to explain the observed masses of massive gcs ( @xmath292 ) .
although our present simulations confirm their claim , it remain unclear how such massive single clusters form from very massive gmcs .
given the possible substructures within host gmcs for fg stars , the original stellar systems can be still clusters of smaller clusters when fg agb stars to eject their gaseous winds
. if this is the case , evolution of agb ejecta in such clusters of clusters would be significantly different from what dercole et al .
( 2008 ) and the present study describe
. it would be also possible that ( i ) original massive stellar systems of gcs are dwarf galaxies themselves and therefore ( ii ) compact stellar systems formed from agb stars of the dwarfs are identical to their stellar galactic nuclei . in this scenario ,
stellar nuclei are dominated by sg stars that can be formed more quickly from agb ejecta of fg ones owing to much deeper gravitational potentials of their host with dark matter halos .
therefore , stellar nuclei of nucleated dwarfs are highly likely to be dominated by he - rich , na - rich , and o - poor stars .
future observations on chemical abundances of stars in nuclei of the sagittarius dwarf and ngc 205 will enable us to discuss whether this scenario is physically viable . the present simulations have shown that secondary star formation from agb ejecta can occur at the central regions of original clusters , where mean stellar number densities exceed @xmath293 stars pc@xmath294 ( for @xmath143 ) .
recently krumholz et al .
( 2009 ) investigated formation of massive stars by gaseous accretion within a gas cloud with a mass of @xmath295 and a size of 0.1 pc and found that two ( binary ) stars with masses of @xmath296 and @xmath297 can be formed within @xmath298 years . therefore the present simulations imply that gas clouds for massive stars ( as those investigated by krumholz et al .
2009 above ) can interact frequently with stars within the central regions of a dense stellar system owing to the small mean separation of stars ( @xmath205 pc ) and the short dynamical time scale ( @xmath299 yr ) .
it would be accordingly possible that massive star formation is severely suppressed owing to this star - cloud interaction which can prevent gas accretion onto clouds forming massive stars : imfs for the formation of sg stars within mscs would be bottom - heavy . if formation of binary star - forming gas clouds , which are progenitors of binary stars , can be also prevented owing to star - cloud interaction , then the observed small binary fractions among sg stars in gcs ( dorazi et al .
2010b ) can be naturally explained .
if only sg stars have a bottom - heavy imf with the power - law slope of @xmath300 with fg ones with @xmath301 , then the number fraction of low - mass stars with masses lower than @xmath302 for sg stars to that for fg ones is 1.7 for @xmath303 .
this number fraction can become as large as 2 by assuming unrealistically low @xmath39 ( @xmath304 ) and steeper @xmath33 ( @xmath305 ) .
therefore , if imfs can be bottom - heavy only for sg stars , then the required original masses of clusters that finally become the present gcs can be up to by a factor of 2 smaller than those derived in the present work with the standard imf both for fg and sg stars .
thus , the possible bottom - heavy imf can not change the present main conclusion that the progenitor clusters of the present gcs are much more massive than the present gcs .
previous works discussed the origin of shapes and rotation ( e.g. , frenk & fall 1982 ; einsel & spurzem 1999 ) , sizes and luminosities ( e.g. , murray 2009 ) , scaling - relations ( e.g. , djorgovski 1993 ; bekki et al .
2004 ; gieles et al . 2010 ) using analytical models and collisionless numerical simulations .
they accordingly did not discuss the origin of gc properties in the context of initially nested structures and ignored the importance of gas dynamics in the formation of gc properties .
if most gcs really originate from nested stellar systems , then conclusions of the above previous works based on models with `` non - nested '' stellar structures must be dramatically modified .
thus it would be reasonable to say that the above conclusions can apply only for low - mass clusters with single stellar population and no nested structures .
bekki ( 2010 ) recently has shown that the origin of the observed rotation in gcs ( e.g. , meylan & mayor 1986 ; meylan & heggie 1997 ; anderson & king 2003 ) can be closely associated with dissipative gas dynamics of agb ejecta from fg stars within forming gcs . given the observed large mass fractions of sg stars in the present gcs ( e.g. , carretta et al .
2010 ) , gas dynamics of agb ejecta within original clusters composed only of fg stars may well be a key determinant for structure and kinematics of the present gcs .
although recent numerical simulations have investigated long - term dynamical evolution of gcs with initially nested stellar structures ( e.g. , decressin et al .
2008 ; dercole et al .
2008 ) , their models do not include initial rotation and flattened shapes of sg stars so that their results can not be used for discussing the origin of rotation and shapes of gcs .
our future long - term ( @xmath284 gyr ) dynamical simulations of gcs based on structures and kinematics of gcs with nested stellar structures , flattened 3d distributions , and rotation simulated in the present study will enable us to discuss structure and kinematics in a consistent manner .
recently vink et al .
( 2009 ) have found two stellar populations with younger and older ones with ages of @xmath306 myr and @xmath307 myr , respectively , in the young , massive stellar cluster sandage-96 in a spiral arm of ngc 2403 .
given the observed possible large age difference ( up to @xmath7 myr ) , this type of young massive clusters can be the candidates that are now forming or have just formed sg stars from agb ejecta from fg ones .
vink et al .
( 2009 ) also have found that the younger population are located closer to the center of the cluster in comparison with the older one .
if the cluster has a mass as large as @xmath264 , the observation would be consistent with the present results on the nested structure of mscs , though quantitative comparison ( i.e. , the mass - ratio and the half - mass - radius - ratio of the two populations ) with the present results is not currently possible .
the present study suggests that if the total mass of the cluster is as large as @xmath264 , the redder colors in the younger population of the cluster can be due to gas and dust that now surrounds the younger population and were previous ejected from the older one .
if future observations find young clusters with large masses ( @xmath308 ) and two populations with the age differences as large as @xmath90 yr in actively star - forming galaxies , they can be forming gcs with nested structures and thus provide strong constraints on the formation models of gcs like the present one .
as suggested above , new compact clusters can be shrouded by gas and dust left behind from the formation of sg stars until supernova events expel the gas and dust : the dust - shrouded new clusters would have very red color for their young ages and possibly they can not be even seen in optical bands if dust extinction is so heavy . if fg stars can not be clearly identified as compact clusters owing to their initially more diffuse distributions , then dust - shrouded new clusters within the fg stars might well be identified as giant hii regions with no optical counterparts .
however it is not clear whether the observed giant hii regions with no near - infrared cluster counterparts in starbursting luminous infrared galaxies with numerous super - star clusters ( e.g. , alonso - herrero et al .
2002 ) can be dust - obscured very young compact clusters ( sg ) embedded in older ones ( fg ) .
massive stellar systems with masses of @xmath291 were discovered in ultra - luminous infrared - galaxies with dust - shrouded strong starbursts ( e.g. , monreal - ibero et al .
numerical simulations showed that such dusty starburst galaxies can evove into `` e+a '' galaxies with poststarburst populations with ages of @xmath309 gyr ( bekki et al .
the present sutdy has shown that mscs with @xmath11 and ages of less than 1 gyr ( before destruction of the old clusters ) can have nested steller structurs .
these results combine to imply that young gcs with nested structures are highly likely to exist in e+a galaxies .
it is thus worthwhile for future observational studies to search for massive young clusters in e+a galaxies ( e.g. , ngc 5102 ) to provide a clue to the origin of multiple stellar populations of gcs .
we have performed 3d numerical simulations of star formation from agb ejecta in mscs formed from gmcs within their host galactic building blocks . we have considered that stars initially in mscs and those newly formed from agb ejecta corresponds to fg and sg stars , respectively , in the present gcs and thus discussed the origin of multiple stellar populations of gcs .
the main results are summarized as follows .
\(1 ) gaseous ejecta from agb stars can be retained within mscs with @xmath0 larger than @xmath310 , if mscs are isolated ( i.e. , not being influenced by ram pressure of ism ) .
the mass fraction of gas retained in a msc to @xmath0 ( @xmath165 ) depends on @xmath0 such that @xmath165 is larger in larger @xmath0 .
the multiple hydrodynamical interaction of gaseous ejecta from numerous agb stars can significantly increase gas mass accumulated within the central regions of mscs , which was first found by the present study .
\(2 ) if @xmath0 of mscs exceed a threshold cluster mass @xmath1 , gaseous ejecta of agb stars can sink into the central regions and then form high - density gaseous regions so that new stars ( i.e. , sg stars ) can be formed there with high star formation efficiencies ( @xmath26 ) .
@xmath1 is demonstrated to be @xmath263 in the present study .
deep gravitational potentials of mscs play a great role both in retaining the accreting agb ejecta onto the central regions and in forming high - density gaseous regions there .
therefore secondary star formation within clusters is inevitable in mscs with @xmath0 larger than @xmath1 .
owing to the existence of @xmath1 , young open clusters and super star clusters with @xmath311 can not evolve into stellar systems with abundance spread in light elements .
\(3 ) owing to the high star formation efficiencies ( @xmath312 ) , very compact stellar systems can be formed from agb ejecta within a time scale of @xmath130 yr ( after massive agb stars start to eject their winds ) .
the half - mass radius of a new compact cluster is by a factor of 5 smaller than that of the old one for most models in the present study .
the nested clusters ( or `` cluster within cluster '' ) with more diffuse old cluster and very compact new one are characteristic for the present numerical models .
\(4 ) at most @xmath6% of all stars can be from new stars formed from agb ejecta for the canonical imfs in the present study .
therefore , in order to explain the present total mass of sg stars with masses of @xmath142 in a gc , @xmath0 is required to be as large as @xmath313 .
the required mass for the progenitor msc for @xmath74 cen can be as large as @xmath66 owing to its larger present mass . given that @xmath314 for secondary star formation , the original masses of progenitors ( @xmath0 ) for the present gcs with masses of @xmath12 need to be larger than @xmath315 .
\(5 ) structural and kinematical properties of new compact clusters formed within original old ones depend on @xmath93 such that new clusters can be more flattened and more strongly supported by rotation in mscs with larger @xmath93 .
the new clusters are much more compact and more strongly supported by rotation than the old ones in all models with initial rotation in old clusters .
these structural and kinematical differences between old and new clusters do not depend on model parameters such as @xmath0 , @xmath18 , and @xmath126 .
given that the present gcs are dominated by sg stars , the result implies that the physical origins of flattened shapes and internal rotation of gcs are closely associated with formation processes of sg stars through gas dynamics within mscs .
\(6 ) evolution of mscs initially in the very centers of their hosts can be significantly different from that of mscs outside the centers , mainly because agb ejecta can be more effectively retained within nuclear mscs owing to much deeper gravitational potentials of their hosts .
irrespective of @xmath0 , most agb ejecta can be retained in these nuclear mscs .
star formation , however , can proceed within nuclear mscs , only if @xmath0 exceeds a threshold value that is almost the same as @xmath1 described above .
strong gravitational fields of hosts can enhance secondary star formation of nuclear mscs .
multiple episodes of star formation ( i.e. , third and fourth generations of stars ) are possible in nuclear mscs with @xmath3 until their hosts are destroyed during accretion of the hosts on larger galaxies .
the origin of multiple generations of stars observed in @xmath74 cen and ngc 2808 can be closely associated with evolution of nuclear mscs .
\(7 ) mscs can lose gradually ( @xmath316 gyr ) significant fractions of their stars during dynamical evolution within their hosts owing to tidal stripping by their hosts .
the stripping processes and the timescale depend strongly on @xmath0 , orbits , and initial locations of mscs within hosts . if stars in a msc enter into their agb phases after the msc has lost most of their stars because of tidal stripping , then the agb ejecta can not be efficiently retained in the msc owing to its much shallower gravitational potential .
therefore gaseous ejecta from agb stars with lower masses are less likely to contribute to secondary star formation owing to the destruction of the msc by its host .
thus the destruction of mscs due to their hosts can provide a mechanism by which star formation can be truncated in mscs .
\(8 ) direct capture and accretion of cold molecular clouds by mscs themselves can be much more efficient ways to dilute the agb ejecta of mscs ( i.e. , mixing of these gaseous components ) in secondary star formation within mscs .
the molecular clouds need to have chemical abundances very similar to those of mscs , and thus they are either ( i ) those that are located very close to mscs or ( ii ) those left behind from the formation of fg stars . the observed hierarchical structures of gmcs are suggested to be responsible for the dilution processes during formation of sg stars .
\(9 ) low - mass mscs with @xmath317 in the lmc can obtain cold molecular gas by capture of the gas but can not retain agb ejecta owing to much shallower gravitational potentials of mscs .
therefore the intermediate - age clusters with dmstos recently discovered in the magellanic clouds can have multiple stellar populations with age differences yet no abundance spread in light elements owing to their low masses .
the initial masses of the old gcs in the lmc were significantly larger than their present masses of @xmath318 so that they could retain agb ejecta to form sg stars with chemical abundances different from those of fg ones .
\(10 ) it is suggested that formation of massive stars can be severely suppressed in very dense central regions of mscs ( i.e. , a bottom - heavy imf ) , because dynamical interaction between stars and forming molecular cores ( where individual stars can form ) can prevent the growth of the cores .
if dense stellar systems are really environments within which only low - mass stars can form ( owing to a bottom - heavy imf ) , then the original masses of the present gcs can be to some extent lower than the proposed one ( @xmath319 ) in the present scenario .
+ thus gcs with the observed high central stellar densities ( @xmath280 stars pc@xmath294 ) can not be formed directly from gmcs with single star formation events within gmcs in a scenario explored in the present study .
the origin of the observed high stellar densities of gcs can be associated with unique star formation processes within central regions of already compact stellar systems ( composed of fg stars ) .
the deep gravitational potential wells of original clusters with much lower stellar densities ( @xmath320 stars pc@xmath294 at half - mass radii ) can play a role in enhancing dramatically star formation efficiencies and consequently forming new compact clusters .
the progenitor stellar systems for the present gcs with multiple stellar populations need to be very massive ( @xmath321 ) to retain effectively agb ejecta from fg stars , dilute the gas with cold molecular gas , form new stars from the gas , and finally have the total stellar masses of sg stars similar to the observed ones in the galactic gcs .
the present scenario explains why open clusters in the galaxy and intermediate - age gcs in the lmc do not show clear abundance spread in light elements .
if gcs really originate from mscs with large @xmath0 , then the host gmcs for such mscs should be even more massive ( @xmath322 ) .
we plan to investigate how very massive gmcs can form and evolve into mscs due to efficient star formation within massive dwarf galaxies at high @xmath229 .
i am grateful to the referee raffaele gratton for valuable comments , which contribute to improve the present paper .
kb acknowledges the financial support of the australian research council throughout the course of this work .
i am grateful to all participants in the last conference `` multiple stellar populations in globular clusters '' held in asiago ( italy ) on 14th @xmath323 16th sep 2010 for their having extensive discussion on origin of globular clusters with me , which helped me better interpret the present numerical results in terms of the corresponding observational ones .
numerical computations reported here were carried out on the grape system at icrar in the university of western australia and that kindly made available by the center for computational astrophysics ( cfca ) of the national astronomical observatory of japan . | we numerically investigate whether and how gaseous ejecta from agb stars can be converted into new stars within originally massive star clusters ( mscs ) in order to understand the origin of multiple stellar populations in globular clusters ( gcs ) .
we adopt a scenario in which ( i ) mscs with masses of @xmath0 can be formed from high - mass , high - density giant molecular clouds ( gmcs ) in their host galactic building blocks embedded in dark matter halos at high redshifts and ( ii ) their evolution therefore can be significantly influenced by @xmath0 , their initial locations , and physical properties of their hosts .
our 3d hydrodynamical simulations show that gaseous ejecta from agb stars can be retained within mscs and consequently converted into new stars very efficiently in the central regions of mscs , only if @xmath0 exceed a threshold mass ( @xmath1 ) of @xmath2 .
the new stars can correspond to the `` second generation ( sg ) '' of stars with higher na and lower o abundances observed in gcs .
star formation efficiencies during the formation of sg stars within mscs with @xmath3 can be rather high ( @xmath4 ) so that very compact new clusters within original mscs can be formed .
@xmath0 should be as large as @xmath5 to explain the observed large fraction of sg stars in the present ordinary galactic gcs , because new stars can consist of only @xmath6% among all stars for the standard imf .
nuclear mscs are found to retain much more effectively the agb ejecta and convert more efficiently the gas into new stars owing to much deeper gravitational potential of their hosts .
capture and accretion of cold molecular gas ( or small gmcs ) by forming mscs themselves can be mechanisms for mixing ( i.e. , dilution ) of agb ejecta with cold pristine gas .
we suggest that both @xmath0 and their locations within their hosts can determine whether abundance spread can be seen only in light elements or even in heavy ones .
we discuss how and in what time scale mscs preferentially lose old stars owing to tidal stripping by their host galactic building blocks .
we also suggest that the origin of the intermediate - age gcs with possible age spread of @xmath7 myr yet apparently no / little abundance spread in light elements in the lmc is closely associated with their incapability to retain the agb ejecta owing to their low masses .
[ firstpage ] galaxies : star clusters globular clusters : general stars : formation |
the aim of this paper is to obtain the equation of state ( eos ) of a plasma in the low coupling regime with a high precision . in this regime standard monte carlo ( mc ) and molecular dynamics simulations techniques
must be handled with care due to huge finite size effects and , in the other hand , the ideal gas approximation or more elaborated analytical expressions commonly used are valid only but asymptotically , for very small values of the coupling parameters .
such thermodynamic conditions are relevant for many astrophysical or laboratory plasmas hydrodynamics applications .
however we shall restrict ourselves to the well known one - component plasma ( ocp ) model , which consists of identical point ions with number density @xmath8 , charges @xmath9 , moving in a neutralizing background , electrons for instance , where @xmath10 , @xmath11 number of particles , @xmath12 volume of the system@xcite . in the very low coupling regime , the virial expansion supplemented by well documented resummation methods , as the well - known debye - hckel ( dh ) theory @xcite and its extensions ( see e.g. cohen @xcite and , more recently
, ortner @xcite expansions for instance ) give reliable results . in the low to intermediate coupling regimes the hypernetted chain ( hnc ) integral equation @xcite
must be solved numerically . finally , in the strong correlation regime
, the ocp has also been extensively studied by monte carlo and molecular dynamics simulations for three decades , see e.g. @xcite and references cited herein . in the more recent of these references one of us
has determined the thermodynamic limit of the excess internal energy per particle @xmath13 of the ocp with a high precision by means of mc simulations in the canonical ensemble within hyperspherical boundary conditions @xcite for @xmath14 .
we recall that in the thermodynamic limit , i.e. for an infinite system of particles , the thermodynamics properties of the model depend solely on the coupling parameter @xmath15 ( @xmath16 , @xmath17 boltzmann constant , @xmath18 temperature , and @xmath19 the ionic radius defined by @xmath20 ) , whereas , for a finite sample , an additional dependance on the number of particles @xmath11 remains . in paper @xcite , henceforth to be referred to as `` i '' , special attention has been brought to describe and take advantage of such finite size effects on the energy @xmath21 to get its thermodynamic limit @xmath13 , using all facilities of work stations available at that time .
recently we have also performed extensive mc simulations of the related yukawa one - component plasma ( yocp ) , i.e. a system made of @xmath11 identical point charges @xmath22 interacting via an effective yukawa pair - potential @xmath23 , where @xmath24 is the so - called screening parameter @xcite , not to be discussed however in this work . for the ocp and the yocp as well , in the low @xmath25 regime , the debye length ( i.e. the correlation length associated with charge fluctuations ) becomes of the order or much larger than the size of the simulation box , yielding huge finite size effects on @xmath21 .
therefore , despite these numerous studies and amount of work it appears that hydrocode applications using the combination of data bases and fits coming from various techniques can be affected by numerical instabilities in the transition regime , around @xmath26 . with nowadays computers
it is now possible to explore this range of small @xmath2 values with the help of performant simulation techniques and to obtain such precise results so that they can be considered as the _ reference _ ones to be used in many applications dealing with degenerate astrophysical or laboratory plasmas .
we also examine carefully in this paper the connection between mc and first principle analytical or hnc results for @xmath27 . we have thus explored and precised the domain of validity of each of these methods .
it turns out to be necessary to combine all of these approaches to obtain a continuous representation of @xmath28 in the range @xmath0 .
finally we extract from these combined approaches the best possible analytical representation for @xmath29 .
our paper is organized as follows .
next section is devoted to a brief presentation of the main features of low @xmath2 expansions ( section [ ortner ] ) , the hnc integral equation ( section [ hnc ] ) and the rather unusual but efficient mc technique used in this paper ( section [ mc ] ) .
note that we have redone , by passing , extremely accurate hnc calculations and obtained new fits of hnc data , presented in section [ hnc ] . in section [ mcdat ] we present and discuss our mc simulations .
fits of the data are described in details and widely illustrated .
finally conclusions are drawn in section [ conclusion ] .
reduced excess energy @xmath30 versus @xmath31 .
diamonds : hnc , black solid line : dh , thick cyan solid line : th1 approximation ( [ a ] ) , thick cyan dashed line : th2 approximation ( [ b ] ) , other curves represent the successive orders of expansion ( [ eqnortner ] ) . ]
the interval @xmath0 covers various correlation regimes from no correlation ( @xmath32 , i.e. the ideal gas ) to an intermediate correlated regime ( @xmath33 , no oscillation or structure in the pair correlation functions ) . in any case
, the long - range nature of the interaction potential between two ionic charges causes mayer graphs to diverge @xcite . a field
theoretical diagrammatic representation of cluster integrals has been proposed recently in @xcite to avoid complicated chain resummations in an attempt to treat the @xmath2 expansion of the classical coulomb system in a more controlled and systematic way . in this interesting paper
the final expansion obtained by the author improves earlier and seminal analytical results of cohen _ et al . _
@xcite obtained by traditional diagrammatic expansions and resummations . from these theoretical analysis it turns out that the physics in this small interval @xmath0 is extremely complicated and exhibits many different correlation regimes , even more than in the widely studied region @xmath3 @xcite .
the low @xmath2 expansions obtained by cohen _
et al . _ and ortner for @xmath29 converge to the hnc results only for @xmath34 as apparent in figure [ fig_u_ortnert ] . for
higher values of @xmath2 these asymptotic expressions do not seem to converge at all and , moreover , the high order terms of the expansions do not improve the results of the lower orders . anticipating the results of sections [ hnc ] and [ mc ] and , as can be observed in figure [ fig_rap ]
, the hnc data deviate from our mc results as soon as @xmath35 .
it results from this sketchy discussion that we must distinguish three different regimes of correlations in the interval @xmath0 , and we confess that this complexity motivated the present study
. ratio of mc excess energies to hnc results versus @xmath2 in the low coupling regime . ] in ref .
@xcite ortner has developed an effective method based on the hubbard - stratonovich ( hs ) transformation and field theoretical approaches to calculate the free energy of classical coulomb systems in the low @xmath2 regime @xcite .
the hs transform was used to obtain the eos of a classical plasma and notably that of the ocp .
the non - trivial part of the helmholtz free - energy density was derived up to order @xmath36 , improving on the previous results of cohen _ et al .
_ at order @xmath37 , obtained by a method of resummation of diverging diagrams .
the author gives an analytical representation of the excess internal energy @xmath38 of the ocp , valid at low @xmath2 , without however any estimation of the error .
it reads as , [ eqnortner ] @xmath39 with the constants , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 and @xmath48 the euler constant . expression [ eqnortner ] ( to be referred to as th2 henceforth ) improves on that given by cohen _ et al._(to be referred to as th1 henceforth ) @xcite , which corresponds to line [ a ] , while the additional terms are those of line [ b ] .
we recognize that the first term ( @xmath49 ) is exactly the well known debye - hckel ( dh ) contribution .
figure [ fig_u_ortnert ] displays the results of the reduced excess energy @xmath30 versus @xmath2 at successive orders in the @xmath2-expansion [ eqnortner ] .
a close examination of the figures reveals that the dh approximation is nearly exact up to @xmath50 , in the sense that higher order contributions do not change the result .
a comparison with hnc results , which are supposed to be nearly exact at least up to @xmath51 ( this point will be fully discussed in next section ) , shows the convergence of the expansions th1 and th2 to hnc at @xmath52 and @xmath53 respectively .
however we do not observe any trend of convergence of these expansions for @xmath54 .
we also notice that the additional terms given by ortner ( cf equation [ b ] ) lead to an oscillatory behavior rather than to an improved convergence radius .
we suspect some misprints in the reported @xmath55 for @xmath56 since the @xmath2 functional @xmath2 dependence of [ eqnortner ] is undoubtedly correct .
reduced excess energy @xmath30 versus @xmath57 .
squares : mc ( the symbols are larger than error bars ) , black line : dh , red line : hnc , blue line : th1 approximation ( [ a ] ) , green line : th2 approximation ( [ b ] ) . ]
comparison between hnc ( red line ) and mc data ( squares , present work and previous results , see ref.@xcite ) for the reduced excess energy @xmath58 versus @xmath59 . ]
we have redone high precision hnc calculations for a hundred of values of @xmath2 in the range @xmath60 ( see figures [ fig_u_ortnert ] and [ fig_comparaison_u ] ) ; additional calculations were also done for some higher values of the coupling parameter , in the range @xmath61 , see figure [ fig_horreur_de_hnc ] .
we used the ng method@xcite with the following control parameters : the pair correlation functions ( direct and non - direct respectively ) @xmath62 and @xmath63 , as well as their fourier transforms @xmath64 and @xmath65 , were tabulated on grids of @xmath66 points with @xmath67 in order to make use of fast fourier transforms with intervals of @xmath68 and @xmath69 in direct and fourier space respectively .
the dimensionless energies were computed according the formulae @xcite @xmath70 where the distances `` @xmath71 '' are measured in the units of the ionic radius @xmath19 and the wave numbers @xmath17 in units of @xmath72 .
the comparison of these two estimations @xmath73 and @xmath74 of the energy , which of course should be equal , give an idea on the relative precision of the numerical resolution of hnc , typically about @xmath75 at @xmath76 and @xmath77 at @xmath78 .
another usefull test is to check the stillinger - lovett ( sl ) sum rules ; recall briefly the two first sl rules ( the third one should not be satisfied by hnc@xcite ) @xmath79 with the control parameters given above the sl rules were satisfied with a relative precision of about @xmath77 .
.[coef4 ] first five cohen - ortner coefficients ( cf eq .
( [ eqnortner ] ) , first line ) compared to the correspondent coefficients of the fits of the energy @xmath80 for hnc and mc data .
second line : hnc , 7 parameters , @xmath81 fixed to its dh value .
third line : hnc , 8 parameters .
last line : mc data in the range @xmath82 , 5 parameters ( @xmath83 ) .
[ cols="<,<,<,<,<,<,<,<,<",options="header " , ] we present and discuss in details the ten values @xmath84 considered in our numerical experiments .
figures [ des_g=01 ] , [ des_g=02 ] , [ des_g=04 ] and [ des_g=07 ] illustrate the cree @xmath85 versus the number of configurations for several caracteristic values of @xmath2 ( @xmath86 , 0.2 , 0.4 , and 0.7 respectively ) typical of the different plasma regimes in the interval @xmath60 .
our previous comments on figures [ des_g=1 ] and [ des_g=2 ] ( for @xmath87 respectively ) are still valid in these cases .
we stress once again the need of large systems together with the need of enough configurations to reach a stable plateau after thermal equilibration .
all generated configurations , @xmath88 , are displayed in figure [ des_g=01 ] ( @xmath86 ) while a zoom of only the first @xmath89 configurations is displayed in figure [ des_g=02 ] ( @xmath90 ) , which exemplifies the plateau reached by the cree for @xmath91 .
on the last two figures [ des_g=04 ] and [ des_g=07 ] , respectively for @xmath92 and @xmath93 and @xmath94 , we see the good convergence with @xmath11 as the interval width between cree values decreases from top to bottom .
by contrast the low @xmath2 runs do not exhibit this regular decrease .
of course beyond visual impressions only the possibility and precision of the fitting process of the mc cree results will give a firm answer on the quality of the tl calculation for each @xmath2 value .
table [ energie ] resumes present work mc calculations of the mc energy @xmath85 of the ocp as a function of @xmath2 for @xmath95 to @xmath91 .
the number in bracket which corresponds to _ one standard deviation @xmath96 _ is the accuracy of the last digits .
results corresponding to @xmath97 , not included in the fits , are not reported .
the thermodynamic limit values of the energy versus @xmath2 are reported in table [ lim ] . the type of extrapolation schemes retained in the fits , i.e. linear or quadratic ( cf equation [ fit_qua ] ) , are specified . in figures
[ fit_g=1 ] , [ fit_g=07 ] , [ fit_g=04 ] and [ fit_g=01 ] we display the quadratic fit of @xmath98 ( solid black line ) and the linear fits for the 3 largest numbers of particles considered , when available ( red dashed line ) for @xmath99 , 0.7 , 0.4 , and @xmath86 respectively .
the error bars on the value of the tl of the energy @xmath100 reported in table [ lim ] are the error bars of the linear ( or quadratic ) regression .
we discuss now the results from high to low @xmath2 s .
figure [ fit_g=1 ] illustrates the high quality of the fits obtained in the case @xmath101 .
indeed the extrapolated tl of @xmath85 coincide for both the linear and the quadratic fits ( the latter involving more states with low number of particles and the former only the 3 largest systems ) with a nice overlap of the error bars .
results are similar down to @xmath93 , as illustrated in figure [ fit_g=07 ] for @xmath93 . in the range @xmath102
the precision of the fits is good but the linear and the quadratic fit extrapolations do not give exactly the same tl values , however the error bars do overlap .
figure [ fit_g=04 ] , corresponding to the case @xmath92 , illustrates the smallest @xmath2 at which a linear fit is possible with the 3 higher values of @xmath11 .
the linear and the quadratic fit extrapolations giving the tl values would coincide within the error bars if the latter were defined to be _ two standard deviations _ rather than only _
one _ according to our choice . for @xmath2
smaller or equal to 0.3 it was impossible to reach an asymptotic form of @xmath85 , linear in the variable @xmath103 , and only a quadratic polynomial fit was possible ( cf table [ lim ] ) .
for that reason it is legitimate to consider the error bars on the extrapolated value @xmath104 as overoptimistic in this range of @xmath2 , see figure [ fit_g=01 ] for an illustration in the case @xmath86 .
simulations involving larger numbers of particles would be necessary but are out of our reach .
for all the states with a @xmath54 the tl limit @xmath29 can thus be obtained with a high precision @xmath105 , after a careful study of finite size effects on the mc energies @xmath21 . for smaller values of @xmath106 , for instance @xmath107 , samples of more than @xmath108 particles should be used to reach the leading order of the asymptotic regime [ scaling ] .
however such an effort would be useless since hnc and cohen approximations are then `` exact '' within the wanted precision on @xmath1 .
the @xmath29 are perfectly well fitted in the range @xmath82 by the cohen s functional form , given by equation [ a ] , involving the five parameters @xmath109 ( @xmath110 ) given in table [ coef4 ] .
in this conclusion we compare at first the cohen - ortner low @xmath2 expansions , hnc and mc data .
figure [ fig_comparaison_u ] shows without ambiguity the good agreement between hnc and mc results in the range @xmath111 and the large departure of both results with analytical expansion ones , dh ( for @xmath112 ) , th1 ( for @xmath113 ) and th2 ( for @xmath114 ) .
note however that the scale of the figure is not large enough to discriminate between hnc and mc results , notably because the errors bars on mc results are smaller than the size of the symbols .
a more enlightening illustration is that of figure [ fig_rap ] which gives the ratio of the mc and hnc energies .
the disagreement for @xmath115 results from a bad evaluation of the tl of @xmath116 due to huge finite size effects spoiling the mc data , while the disagreement for @xmath117 simply reflects the failure of the hnc approximation at high @xmath2 . a nearly perfect agreement between mc and hnc results ,
compatible with _
one standard deviation _ is observed only at @xmath118 ; with _ two standard deviations _
the hnc results are within the error bars of the mc data in the interval @xmath119 . by passing our new hnc calculations for some values of @xmath2 in the range @xmath120
are plotted in figure [ fig_horreur_de_hnc ] were mc data were also included for comparison .
* @xmath122 is the range of validity of debye - hckel theory .
* @xmath123 is the range of validity of cohen low @xmath2 expansion [ eqnortner ] . *
ortner s additional terms do not improve the results .
* @xmath124 is the range of validity of hnc .
the data are perfectly represented by the eight parameters fit of tables [ coef4 ] and [ coef567 ] .
* we were able to extract the thermodynamic limit of the ocp energy from our mc simulations with a precision not smaller than @xmath121 in the range @xmath82 .
our data are well fitted by the five parameters fit of table [ coef4 ] .
99 m. baus and j .- p .
hansen , phys .
* 59 * , 1 ( 1980 ) .
debye , e. hckel physikalische * 24 * , 185 ( 1923 ) .
e. g. d. cohen and t. j. murphy , the phys . of fluids
* 12 * , 1404 ( 1969 ) .
j. ortner , phys .
e * 59 * , 6312 ( 1999 ) .
ng , j. chem . phys . * 61 * , 2680 ( 1974 )
. s. g. brush , h.l .
sahlin , and e. teller , j. chem .
* 45 * , 2102 ( 1966 ) .
e. l. pollock and j. p. hansen , phys .
a * 8 * , 3110 ( 1973 ) . w. l. slattery , g. d. doolen , and h. e. dewitt , phys .
a * 26 * , 2255 ( 1982 ) .
h. e. de witt , proceedings of the international conference on strongly coupled coulomb systems ( boston , 1997 ) .
caillol , j. chem .
phys . * 111 * , 6528 ( 1999 ) .
caillol , j. chem .
phys . * 111 * , 6538 ( 1999 ) .
caillol and d. gilles , j. stat . phys . *
100 * , 933 ( 2000 ) .
stratonovich r l , sov .
. solid state * 2 * , 1824 ( 1958 ) .
hubbard , j 1954 phys .
lett . * 3 * , 77 ( 1954 ) ; hubbard j and shofield p 1972 phys .
a * 40 * , 245(1972 ) .
caillol and j. l. raimbault , j. stat .
phys . * 103 * , 753 ( 2001 ) .
f. h. stillinger and r. lovett , j. chem . phys . *
49 * , 1991 ( 1968 ) .
d. frenkel and b. smit , _ understanding molecular simulation _
( academic , new york , 1996 ) .
solid lines : cumulated reduced excess energy @xmath85 versus @xmath127 for @xmath99 . from left to right @xmath128 .
the error bars correspond to one standard deviation @xmath96 .
solid black line : quadratic polynomial regression of mc data . dashed red line : linear regression of the 3 larger systems mc data . ] solid lines : cumulated reduced excess energy @xmath85 versus @xmath127 for @xmath86 . from left to right @xmath128 .
the error bars correspond to one standard deviation @xmath96 .
solid black line : quadratic polynomial regression of mc data . ] | an accurate equation of state of the one component plasma is obtained in the low coupling regime @xmath0 .
the accuracy results from a smooth combination of the well - known hypernetted chain integral equation , monte carlo simulations and asymptotic analytical expressions of the excess internal energy @xmath1 .
in particular , special attention has been brought to describe and take advantage of finite size effects on monte carlo results to get the thermodynamic limit of @xmath1 .
this combined approach reproduces very accurately the different plasma correlation regimes encountered in this range of values of @xmath2 .
this paper extends to low @xmath2 s an earlier monte carlo simulation study devoted to strongly coupled systems for @xmath3 ( , * 111 * , 6538 ( 1999 ) ) .
analytical fits of @xmath4 in the range @xmath0 are provided with a precision that we claim to be not smaller than @xmath5 .
hnc equation and exact asymptotic expressions are shown to give reliable results for @xmath4 only in narrow @xmath2 intervals , i.e. @xmath6 and @xmath7 respectively . |
the dirac oscillator is a natural model for studying properties of physical systems .
this model is based on the dynamics of a harmonic oscillator for spin-1/2 particles by introducing a nonminimal prescription into free dirac equation @xcite . because it is a exactly solvable model ,
several investigations have been developed in the context of this theoretical framework in the last years .
the interest in this issue appears in different contexts , such as quantum optics @xcite , supersymmetry @xcite , nuclear reactions @xcite , clifford algebra @xcite and noncommutative space @xcite .
recently , the one - dimensional dirac oscillator has been verified experimentally by j. a. franco - villafae et al . ,
based on a tight - binding system @xcite .
a detailed description for the dirac oscillator is given in ref .
@xcite and for other contributions see refs .
@xcite . among the various contexts in which the dirac oscillator can be addressed
, we refer to the cosmic string , a linear defect that change the topology of the the medium when viewed globally .
this framework has inspired a great deal of investigation in recent years .
such works encompass several distinct aspects to investigate the effects produced by topological defects of this nature @xcite . in this work
, we generalize the results in @xcite for a 2d dirac oscillator in the magnetic cosmic string background showing rigorously how the dynamics of this system is affected when the effects of spin are taken into account .
our approach is based on the self - adjoint extension method which is appropriate to address any system endowed with a singular hamiltonian ( due to localized field sources or quantum confinement ) .
we determine the most relevant physical quantities from the present model , such as energy spectrum , wave functions , and self - adjoint extension parameter by applying boundary conditions allowed by the system .
in this section , we study the motion of the particle in the magnetic cosmic string background .
the cosmic string spacetime with an internal magnetic field is an object described by the following line element in cylindrical coordinates ( @xmath1 ) : @xmath2 with @xmath3 , @xmath4 and @xmath5 .
the parameter @xmath6 is related to the linear mass density @xmath7 of the string by @xmath8 runs in the interval @xmath9 $ ] and corresponds to a deficit angle @xmath10 .
geometrically , the metric corresponds to a minkowiski spacetime with a conical singularity @xcite .
we begin with the dirac equation in the curved spacetime ( with @xmath11 ) : @xmath12\psi=0,\ ] ] where @xmath13 is the charge , @xmath14 is mass of the particle , @xmath15 is a four - component spinorial wave function , and @xmath16 is the spinor affine connection , which is given by @xcite @xmath17,\ ] ] where @xmath18 are the standard dirac matrices in minkowski spacetime , and @xmath19 is the spin connection , given by @xmath20 with @xmath21 and @xmath22 ; @xmath23 is the christoffel symbol , @xmath24 is the metric tensor , @xmath25 is the basis tetrad which will be defined below .
the spin connection allows us to construct a local frame through the basis tetrad which gives the spinors in the curved spacetime . also , the @xmath26 matrices are the generalized dirac matrices defining the covariant clifford algebra , @xmath27 and are written in terms of the standard dirac matrices @xmath18 in minkowski spacetime as @xmath28 with @xmath29 , and @xmath30 where @xmath31 are the standard pauli matrices and @xmath32 is the @xmath33 identity matrix . the basis tetrad @xmath34 in eq .
is chosen to be @xcite @xmath35 satisfying the condition @xmath36 the matrices @xmath26 in eq .
are given more explicitly as @xmath37 the starting point for the derivation of eq .
is that the curved - space gamma matrices are covariantly constant @xcite , i.e. , @xmath38 ( see [ sec : appendixa ] ) . for the specific basis tetrad
, the connection is found to be @xmath39 with the non - vanishing element given as @xmath40 details for the calculation of the connection are given in [ sec : appendixb ] . according to
the tetrad postulated @xcite , the matrices @xmath18 could be any set of constant dirac matrices .
thus , we are free to choose a representation for the matrices @xmath18 . making use of the symmetry under z translations of the system , we can reduce the four - component dirac equation to a two two - component spinor equations . to do this , we consider the vector potential @xmath41 as being intrinsically two - dimensional , i.e. , it has only two components and depends on only two spatial coordinates and we take @xmath42 . in this manner , the relevant equation is @xmath43 \psi = e\psi , \label{eq : dirac2}\ ] ] where @xmath44 is a two - component spinor , and @xmath45 is the generalized momentum , @xmath46 is the gradient operator in polar coordinates , and the @xmath47 matrices are given in terms of the pauli matrices as @xmath48 where the parameter @xmath49 , which has a value of twice the spin value , can be introduced to characterizing the two spin states @xcite , with @xmath50 for spin `` up '' and @xmath51 for spin `` down '' @xcite . in the representation , eq .
yields @xmath52 the magnetic vector potential in polar coordinates in the coulomb gauge is chosen to be @xmath53 where @xmath54 is the flux parameter with @xmath55 ( @xmath56 denote the unit vectors in polar coordinates . )
this choice for the vector potential gives a magnetic flux tube , in the background space described by the metric , coinciding with the cosmic string and with the magnetic field strength given by @xmath57 note that , in the limit as @xmath58 , we obtain the magnetic field in euclidean space .
the 2d dirac oscillator is introduced by the non - minimal substitution @xcite , @xmath59 where @xmath60 is the position vector and @xmath61 the frequency of the oscillator ( for a comprehensive discussion of the dirac oscillator see ref .
@xcite ) . in this case , eq . reads
@xmath62 \psi = e\psi .
\label{eq : dirac3}\ ] ] the second order equation implied by eq .
is obtained by applying the matrix operator @xmath63.\ ] ] so , one finds @xmath64 [ \boldsymbol{\alpha}\cdot(\boldsymbol{\pi}-im\omega\beta\mathbf{r } ) ] \psi.\ ] ] inserting eqs . , and the expression for @xmath65 in into eq .
, one obtains @xmath66 where @xmath67^{2}\nonumber \\ & -2m\omega \left[\sigma^{z}+ s\left(\frac{1}{i\alpha}\frac{\partial}{\partial\varphi } + \frac{\phi}{\alpha}+\frac{1-\alpha}{2\alpha}s\sigma^{z } \right ) \right ] \nonumber \\ & + m^{2}\omega^{2}r^{2 } + \frac{\phi s}{\alpha}\frac{\delta(r)}{r}\sigma^{z}.\end{aligned}\ ] ] in eq . , the quantity @xmath68 contributes to the term which depends explicitly on the spin of the particle .
the first term is the contribution due to the magnetic flux while the second is due to the spin connection .
note that , by making @xmath69 ( flat spacetime ) and @xmath70 ( absence of a magnetic field ) in eq . , we obtain , for the planar case , the 2d dirac oscillator as proposed by moshinsky and szczepaniak @xcite and discussed in [ sec : appendixc ] . making use of the underlying rotational symmetry we can express the two - component spinor as @xmath71 with @xmath72 . by replacing eq . into eq .
, we obtain the radial equation for @xmath73 @xmath74 where @xmath75 @xmath76 @xmath77 and @xmath78 the hamiltonian in eq .
governs the dynamics of a dirac oscillator in a magnetic cosmic string background , i.e. , a dirac oscillator problem in the presence of the aharonov - bohm effect in a conical spacetime .
the presence of a two - dimension @xmath0 interaction in the radial hamiltonian @xmath79 , which is singular at the origin , makes the problem more complicated to solve . the most adequate manner to address this kind of point interaction potential
is by making use of the self - adjoint extension approach @xcite .
this is the method adopted in this work and discussed in the next section .
in this section , we review some concepts on the self - adjoint extension approach .
an operator @xmath80 , with domain @xmath81 , is said to be self - adjoint if and only if @xmath82 and @xmath83 , @xmath84 being the adjoint of operator @xmath80 . for smooth functions , @xmath85 with @xmath86 , we should have @xmath87 , and it is possible to interpret the hamiltonian as a self - adjoint extension of @xmath88 @xcite . the self - adojint extension approach consists , essentially , in extending the domain of @xmath89 in order to match @xmath90 . from the theory of symmetric operators , it is a well - known fact that the symmetric radial operator @xmath91 is essentially self - adjoint for @xmath92 , while for @xmath93 it admits an one - parameter family of self - adjoint extensions @xcite , @xmath94 , where @xmath95 is the self - adjoint extension parameter . to characterize this family
, we will use the approach in @xcite , which is based on the boundary conditions at the origin .
all the self - adjoint extensions @xmath96 of @xmath91 are parametrized by the boundary condition at the origin @xmath97 with @xmath98,\end{aligned}\ ] ] where @xmath99 is the self - adjoint extension parameter . for @xmath100
, we have the free hamiltonian ( without the @xmath0 function ) with regular wave functions at the origin , and for @xmath101 the boundary condition in eq .
permit an @xmath102 singularity in the wave functions at the origin .
in this section , we determine the energy spectrum for the dirac oscillator in the cosmic string background by solving eq . .
for @xmath103 , the equation for the component @xmath73 can be transformed by the variable change @xmath104 resulting in @xmath105 with @xmath106 . due to the boundary condition in eq . , we seek for regular and irregular solutions for eq . . studying the asymptotic limits of eq .
leads us to the following regular ( @xmath107 ) ( irregular ( @xmath108 ) ) solution : @xmath109 with this , eq .
is rewritten as @xmath110 equation is of the confluent hypergeometric equation type @xmath111 in this manner , the general solution for eq .
is @xmath112 with @xmath113 in eq .
, @xmath114 is the confluent hypergeometric function of the first kind @xcite and @xmath115 and @xmath116 are , respectively , the coefficients of the regular and irregular solutions . in this point
, we apply the boundary condition in eq . .
doing this , one finds the following relation between the coefficients @xmath115 and @xmath116 @xmath117 .
\label{eq : coef_rel_1}\ ] ] we note that @xmath118 diverges if @xmath92 .
this condition implies that @xmath119 must be zero if @xmath92 and only the regular solution contributes to @xmath73 . for @xmath120 , when the operator @xmath91 is not self - adjoint , there arises a contribution of the irregular solution to @xmath73 @xcite . in this manner ,
the contribution of the irregular solution for the system wave function stems from the fact that the operator @xmath91 is not self - adjoint . for @xmath73 be a bound state wave function
, it must vanish at large values of @xmath121 , i.e. , it must be normalizable .
so , from the asymptotic representation of the confluent hypergeometric function , the normalizability condition is translated in @xmath122 from eq .
, for @xmath120 we have @xmath123 . using this result in eq .
, one finds @xmath124 equation implicitly determines the bound state energy for the dirac oscillator in the cosmic string background for different values of the self - adjoint extension parameter .
two limiting values for the self - adjoint extension parameter deserve some attention . for @xmath100 , when the @xmath0 interaction is absent
, only the regular solution contributes for the bound state wave function . on the other side , for @xmath125 only the irregular solution
contribute for the bound state wave function .
for all other values of the self - adjoint extension parameter , both regular and irregular solutions contributes for the bound state wave function .
the energies for the limiting values are obtained from the poles of the gamma function , namely , @xmath126 with @xmath127 a nonnegative integer , @xmath128 . by manipulation of eq . , we obtain @xmath129\right\}^{1/2}.\end{aligned}\ ] ] in particular
, it should be noted that for the case when @xmath92 or when the @xmath0 interaction is absent , only the regular solution contributes for the bound state wave function ( @xmath130 ) , and the energy is given by eq .
using the plus sign .
note that , for @xmath69 ( flat space ) and @xmath70 ( no magnetic flux ) , eq . coincides with the energy found for the usual 2d dirac oscillator ( cf . eq . in [ sec : appendixc ] ) . without loss of generality ,
let us suppose @xmath131 @xcite . in this interval , and recalling that we are interested in the case where @xmath132 , another interesting feature is present in the energy eigenvalues . for the regular solution , the eigenvalues are independent of @xmath133 , @xmath134 and @xmath135 for @xmath136 .
this situation is shown in fig .
[ fig : fig1](a ) for @xmath137 and @xmath138
. however , this independence is absent for @xmath51 , as shown in fig .
[ fig : fig1](b ) for @xmath137 and @xmath138 . in the other hand , for the irregular solution , the eigenvalues are independent of @xmath133 , @xmath134 and @xmath135 for @xmath51 and dependent for @xmath136 . also , for @xmath51 , decreasing the value of @xmath135 , the energy increase as an effect of the quantum localization .
the unnormalized bound state wave functions for our problem are @xmath139 ( color online ) the energy eigenvalues for the regular solution for a 2d dirac oscillator in the cosmic string background as a function of the declination @xmath135 and the magnetic flux @xmath134 for @xmath137 , @xmath138 and for ( a ) @xmath136 and ( b ) @xmath51 . for convenience ,
we plot @xmath140 instead @xmath141.,title="fig : " ] ( color online ) the energy eigenvalues for the regular solution for a 2d dirac oscillator in the cosmic string background as a function of the declination @xmath135 and the magnetic flux @xmath134 for @xmath137 , @xmath138 and for ( a ) @xmath136 and ( b ) @xmath51 . for convenience ,
we plot @xmath140 instead @xmath141.,title="fig : " ] the self - adjoint extension is related with the presence of the @xmath0 interaction . in this manner
, the self - adjoint extension parameter must be related with the @xmath0 interaction coupling constant @xmath142 .
in fact , as shown in refs .
@xcite ( see also refs .
@xcite ) , from the regularization of the @xmath0 interaction , it is possible to find such a relationship . using the regularization method , one obtains the following equation for the bound state energy : @xmath143 by comparing eqs . and , this relation is found to be @xmath144 where @xmath145 is a very small radius which comes from the @xmath0 regularization @xcite .
we shall now take the nonrelativistic limit of eq . .
using @xmath146 with @xmath147 , we obtain @xmath148 performing the same steps as for the relativistic case , one obtains the shifted energy levels ( cf .
[ sec : appendixc ] ) @xmath149\omega.\end{aligned}\ ] ] in this equation , the @xmath107 ( @xmath108 ) sign is for @xmath100 ( @xmath125 ) when one has regular ( irregular ) solution .
we note that the energy in eq . corresponds to equation ( 54 ) of ref .
@xcite ( cf . also eq . in [ sec : appendixc ] ) with two additional contributions , the spin - orbit coupling and the spin connection .
in this contribution , we have addressed the dirac oscillator interacting with a topological defect and in the presence of the aharonov - bohm potential .
this system has been studied in ref .
however , the authors do not take into account the effects of spin .
in other words , the term proportional to the @xmath0 interaction was discarded , by considering only the regular solution of the problem .
the presence of this term has direct implications in the energy spectrum and wave functions of the oscillator .
the correct approach to this problem must include spin effects , which are explicitly manifested by the spin - orbit coupling term , and , so we have a complete description for the dynamics of the 2d dirac oscillator .
we consider the self - adjoint extension method and show that the spin - orbit coupling term , which results in a @xmath0 interaction , can not be dropped from the hamiltonian .
although being singular at the origin , this term reveals that both regular and irregular solutions contribute for the bound state wave function and , consequently , for the energy spectrum .
expressions for the bound states energy for different values of the self - adjoint extension parameter were obtained . for two specific values for the self - adjoint extension parameter ,
i.e. , @xmath100 and @xmath125 , the bound state energies are given explicitly in eq . .
we also verified that , for the flat space ( @xmath69 ) and no magnetic flux ( @xmath70 ) , the results of the usual 2d dirac oscillator are recovered .
in this appendix , we give the details of the calculation of the covariant derivative of @xmath26 in eq . .
as a consequence of the tetrad postulate @xmath151 together with the condition @xmath152 = \omega_{\mu \bar{b}}^{\bar{a}}\gamma^{\bar{b } } , \label{eq : concv}\ ] ] we have @xmath153 = 0 .
\label{eq : cvtly}\ ] ] to check this , first we need to find the relevant christoffel symbols .
they are found to be @xmath154 moreover , we also make use of eqs . - and eq . and then calculate for each @xmath155 .
then , for @xmath156 , we write7 @xmath157 .\qquad ( \mu = t)\ ] ] since @xmath158 , @xmath159 and @xmath160 , it follows that @xmath161 = 0,\quad ( \lambda = t)\\ \nabla_{t}\gamma^{r } = { } & \partial_{t}\gamma^{r}+\gamma_{t\nu}^{r}\gamma^{\nu } + \left [ \gamma^{r},\gamma_{t}\right ] = 0,\quad ( \lambda = r ) \\
\nabla_{t}\gamma^{\varphi } = { } & \partial_{t}\gamma^{\varphi } + \gamma_{t\nu}^{\varphi}\gamma^{\nu}+ \left [ \gamma^{\varphi},\gamma_{t}\right ] = 0,\quad(\lambda = \varphi)\end{aligned}\ ] ] so that @xmath162 next , we have @xmath163,\quad(\mu = r)\ ] ] and by using @xmath164 ( with @xmath165 ) , @xmath166 , @xmath167 , we get @xmath168 = 0,\quad\left ( \lambda = t\right)\\ \nabla_{r}\gamma^{r } = { } & \partial_{r}\gamma^{r}+\gamma_{r\nu}^{r}\gamma^{\nu } + \left [ \gamma^{r},\gamma_{r}\right ] = 0,\quad\left ( \lambda = r\right)\\ \nabla_{r}\gamma^{\varphi } = { } & \partial_{r}\gamma^{\varphi}+\gamma_{r\nu}^{\varphi}\gamma^{\nu } + \left [ \gamma^{\varphi},\gamma_{r}\right ] , \nonumber \\ = { } & \frac{1}{\alpha r^{2}}\left
( \gamma^{\bar{1}}\sin \varphi -\gamma^{\bar{2}}\cos \varphi \right ) \nonumber \\ & + \frac{1}{\alpha r^{2}}\left
( -\gamma^{\bar{1}}\sin \varphi + \gamma^{\bar{2}}\cos \varphi \right ) , \nonumber \\
= { } & 0,\quad\left ( \lambda = \varphi \right)\end{aligned}\ ] ] and consequently @xmath169 now , for @xmath170 , we write @xmath171 , \quad\left ( \mu = \varphi \right ) \label{eq : lmr}\ ] ] and again , since @xmath172 , and using eqs . and ,
we have @xmath173 , \nonumber \\ = { } & \frac{1}{2}(1-\alpha)\gamma_{\bar{1}}\gamma_{\bar{2}}\gamma^{t } -\frac{1}{2}(1-\alpha)\gamma^{t}\gamma_{\bar{1}}\gamma_{\bar{2 } } , \nonumber \\
= { } & 0 , \quad ( \lambda = t).\end{aligned}\ ] ] @xmath174 , \nonumber \\ = { } & -(1-\alpha ) \gamma^{\bar{2}}\cos
\varphi -(1-\alpha ) \gamma^{\bar{1}}\sin \varphi \nonumber \\ & + \left ( 1-\alpha \right ) \gamma^{\bar{2}}\cos
\varphi -\left ( 1-\alpha \right ) \gamma^{\bar{1}}\sin \varphi , \nonumber
\\ = { } & 0,\quad ( \lambda = r).\end{aligned}\ ] ] @xmath175 , \nonumber \\ = { } & -\frac{1}{\alpha r}(1-\alpha ) \left ( \gamma^{\bar{1}}\cos\varphi + \gamma^{\bar{2}}\sin \varphi \right ) \nonumber \\ & + \frac{1}{\alpha r}(1-\alpha ) \left ( \gamma^{\bar{1}}\cos\varphi + \gamma^{\bar{2}}\sin \varphi \right ) , \nonumber \\
= { } & 0 , \quad ( \lambda=\varphi),\end{aligned}\ ] ] so that @xmath176 therefore , eqs . , and , imply that @xmath177 thus , we have verified that the matrices @xmath26 , in the basis tetrad given in eq .
, are covariantly constant .
the spinor affine connection in eq . can be written more explicitly as @xmath178 + \frac{1}{4}\omega_{t}^{\bar{0}\bar{2 } } \left[\gamma_{\bar{0}},\gamma_{\bar{2}}\right ] + \frac{1}{4}\omega_{t}^{\bar{1}\bar{2 } } \left[\gamma_{\bar{1}},\gamma_{\bar{2}}\right ] , \\
\gamma_{r } = { } & \frac{1}{4}\omega_{r}^{\bar{0}\bar{1 } } \left[\gamma_{\bar{0}},\gamma_{\bar{1}}\right ] + \frac{1}{4}\omega_{r}^{\bar{0}\bar{2 } } \left[\gamma_{\bar{0}},\gamma_{\bar{2}}\right ] + \frac{1}{4}\omega_{r}^{\bar{1}\bar{2 } } \left[\gamma_{\bar{1}},\gamma_{\bar{2}}\right ] , \\
\gamma_{\varphi } = { } & \frac{1}{4}\omega_{\varphi}^{\bar{0}\bar{1 } } \left [ \gamma_{\bar{0}},\gamma_{\bar{1}}\right ] + \frac{1}{4}\omega_{\varphi}^{\bar{0}\bar{2 } } \left [ \gamma_{\bar{0}},\gamma_{\bar{2}}\right ] + \frac{1}{4}\omega_{\varphi}^{\bar{1}\bar{2 } } \left [ \gamma_{\bar{1}},\gamma_{\bar{2}}\right ] , \\
\gamma_{z } = { } & \frac{1}{4}\omega_{z}^{\bar{0}\bar{1 } } \left [ \gamma_{\bar{0}},\gamma_{\bar{1}}\right ] + \frac{1}{4}\omega_{z}^{\bar{0}\bar{2 } } \left[\gamma_{\bar{0}},\gamma_{\bar{2}}\right ] + \frac{1}{4}\omega_{z}^{\bar{1 } \bar{2 } } \left [ \gamma_{\bar{1}},\gamma_{\bar{2}}\right].\end{aligned}\ ] ] in order to calculate the spin connection @xmath19 in eq .
, we use the christoffel symbols given in eqs . and . since @xmath179 , @xmath180 and @xmath181 , we find @xmath182 .
furthermore , we also can verify that @xmath183 . as a result
, we have @xmath184 for @xmath185 , @xmath186 and @xmath187 , @xmath188 .
thus , the only contribution for the spin connection is obtained from @xmath189 which gives @xmath190 thus , for the specific basis tetrad , the connection is found to be @xmath191 with the non - vanishing element given as @xmath192 = -\frac{1}{2}\left ( 1-\alpha \right ) \gamma_{\bar{1}}\gamma_{\bar{2}}.\ ] ]
in this appendix , we briefly discuss the usual 2d dirac oscillator .
we mention that although fully equivalent , the present construction is slightly different from the previous construction one in the literature @xcite .
let us consider eq . with @xmath193 . by using the representation for the @xmath194 matrices in eq .
, we are left with @xmath195\psi=0,\ ] ] with @xmath196 . by squaring eq .
, one obtains @xmath197\psi= ( e^2-m^{2})\psi.\ ] ] equation , restoring the factors @xmath198 and @xmath199 , in terms of components , provides [ eq:2ddiracoscillatorcomp ] @xmath200\psi_1= ( e^2-m^{2}c^{4})\psi_1,\ ] ] @xmath201\psi_2=\\ ( e^2-m^{2}c^{4})\psi_2,\ ] ] where @xmath202 equation for @xmath136 agreed with the expressions found in eq .
( a2 ) of ref .
@xcite and eqs .
( 9 ) and ( 22 ) of ref .
@xcite . using the ansatz in eq .
the energy eigenvalues are determined : @xmath203 showing that the energy eigenvalues are spin dependent .
it should be noted that for @xmath136 ( @xmath51 ) and @xmath204 ( @xmath205 ) the energy eigenvalues are independent of the quantum number @xmath133 . from eq .
, in the nonrelativistic limit @xmath146 with @xmath147 , we have @xmath206\psi= \mathcal{e}\psi.\ ] ] the first term on the left side of eq .
is the hamiltonian of the nonrelativistic circular harmonic oscillator @xcite , explaining why this system is called the dirac oscillator .
the second term is a constant which shifts all energy levels .
the last term is the spin - orbit coupling , which ( restoring the factor @xmath198 ) is of strength @xmath207 .
summarizing , the nonrelativistic limit of the 2d dirac oscillator is the circular harmonic oscillator with a strong spin - orbit coupling term with all levels shifted by the factor @xmath61 .
indeed , the shifted energy levels are @xmath208 as for the relativistic case , for @xmath136 ( @xmath51 ) and @xmath204 ( @xmath205 ) the energy eigenvalues are independent of the quantum number @xmath133 .
we would like to thank r. casana and l. r. b. castro for fruitful discussions .
this work was supported by the fundao araucria ( grant no .
205/2013 ( ppp ) and no .
484/2014 ( pq ) ) , and the conselho nacional de desenvolvimento cientfico e tecnolgico ( grants no .
482015/2013 - 6 ( universal ) , no . 306068/2013 - 3 ( pq ) ) and fapema ( grant no .
00845/13 ) .
finally , we acknowledge some suggestions made by the anonymous referees in order to improve the present work . | in this work the dynamics of a 2d dirac oscillator in the spacetime of a magnetic cosmic string is considered .
it is shown that earlier approaches to this problem have neglected a @xmath0 function contribution to the full hamiltonian , which comes from the zeeman interaction .
the inclusion of spin effects leads to results which confirm a modified dynamics .
based on the self - adjoint extension method , we determined the most relevant physical quantities , such as energy spectrum , wave functions and the self - adjoint extension parameter by applying boundary conditions allowed by the system . |
in physics , stability of a given configuration ( solution of some set of equations ) , is a useful criterium for relevance of that solution .
unstable configurations are likely not to be realizable in practice , and represent an intermediate stage in the evolution of the system . nevertheless , the instability itself is of great interest , since an understanding of the mechanism behind it may help one to better grasp the physics involved . in particular , it is of interest to be able to predict which other systems display similar instabilities , or even have a deeper understanding of the physics behind the instability ( why is the system unstable ?
is there some fundamental principle behind the instability ? ) . in general relativity
, the kerr family exhausts the black hole solutions to the electro - vac einstein equations .
kerr black holes are stable , and can therefore describe astrophysical objects . however , there are many instances of instabilities afflicting objects with an event horizon , such as the gregory - laflamme @xcite , the ultra - spinning @xcite or superradiant instabilities @xcite and other instabilities of higher - dimensional black holes in alternative theories @xcite(for a review see ref .
@xcite ) . konoplya and zhidenko ( hereafter kz ) recently studied small perturbations in the vicinity of a charged black hole in de sitter background , a reissner - nordstrm de sitter black hole ( rnds ) @xcite .
their ( numerical ) results show that when the spacetime dimensionality @xmath0 , the spacetime is unstable , provided the charge is larger than a given threshold , determined by kz for each @xmath1 . because the results are so surprising ( the mechanism behind it is not yet understood ) , we set out to to investigate this instability and hopefully understand it better .
our results can be summarized as follows : ( i ) we can prove analytically the existence of unstable modes for charge @xmath2 higher than a certain threshold .
( ii ) in the near - extremal regime , we are able to find an explicit solution for the unstable modes , determining the instability timescale analytically .
we hope that our incursion in this topic helps to better understand the physics at work .
this work focuses on the higher dimensional rnds geometry , described by the line element ds^2 = -f dt^2 + f^-1 dr^2 + r^2 d_n^2 , [ lineelement ] where @xmath3 is the line element of the @xmath4 sphere and f=1-r^2-+ .
the background electric field is @xmath5 , with @xmath6 the electric charge .
the quantities @xmath7 and @xmath2 are related to the physical mass m and charge @xmath6 of the black hole @xcite , and @xmath8 to the cosmological constant .
the spacetime dimensionality is @xmath9 .
the above geometry possesses three horizons : the black - hole cauchy horizon at @xmath10 , the black hole event horizon is at @xmath11 and the cosmological horizon is at @xmath12 , where @xmath13 , the only real , positive zeroes of @xmath14 . for convenience , we set @xmath15 , i.e. , we measure all quantities in terms of the event horizon @xmath16 .
we thus get 2m=1+q^2- , furthermore , we can also write = . for a fixed @xmath17 and spacetime dimension @xmath1 , the existence of a regular event horizon
imposes that the charge @xmath2 must be smaller than a certain value @xmath18 . with our units
this maximum charge is q_ext^2=.[qext ] gravitational perturbations of this spacetime couple to the electromagnetic field , and were completely characterized by kodama and ishibashi @xcite .
they can be reduced to a set of two second order ordinary differential equations of the form , _ + ( ^2-v_s)_=0,[we ] where the tortoise coordinate @xmath19 and the potentials @xmath20 are defined through r _ * f^-1 dr , v_s = .
[ tortoise ] we have h_+&=&1-x , + h_-&=&m+(1+m)x , and the quantities @xmath21 are given by u_+ & = & y -^3 n^3 ( 3 n-2 ) ( n+1)^4 ( 1+m ) x^4 + & & + 4 ^2 n^2 ( n+1)^2 \{(n+1)(3n-2 ) m + 4 n^2+n-2 } x^3 + & & + 4 ( n+1)\ { ( n-2 ) ( n-4 ) ( n+1 ) ( m+n^2 k ) -7 n^3 + 7 n^2 - 14 n+8 } x^2 + & & + \{16 ( n+1 ) ( -4 m+3 n^2(n-2 ) k ) -16 ( 3 n-2 ) ( n-2 ) } x + & & + 64 m+16 n(n+2 ) k , + u_- & = & y -n^3 ( 3 n-2 ) ( n+1)^4 ( 1+m ) ^3 x^4 + & & -4 n^2 ( n+1)^2 ( 1+m ) ^2 \{(n+1)(3 n-2 ) m -n^2 } x^3 + & & + 4 ( n+1 ) ( 1+m ) \
{ m ( n-2 ) ( n-4 ) ( n+1 ) ( m+n^2 k ) .
+ 4 n ( 2 n^2 - 3 n+4 ) m+n^2 ( n-2 ) ( n-4 ) ( n+1)k } x^2 + & & -16 m \ { ( n+1 ) m ( -4 m+3 n^2(n-2 ) k ) .
+ 3 n ( n-4 ) m+3 n^2 ( n+1 ) ( n-2)k } x + & & + 64 m^3 + 16 n(n+2)m^2 k. the variables @xmath22 and parameters @xmath23 are defined through x & & , yr^2 , + ^2 & & m^2+,mk^2-n k , and the quantity @xmath24 is implicitly given by @xmath25 .
note that the following relations holds @xmath26 .
note also that for the spacetime considered in this paper @xmath27 , which means that the eigenvalues @xmath28 are given by @xmath29 , where @xmath30 is the angular quantum number , that gives the multipolarity of the field .
behavior of @xmath31 for different parameters , for @xmath32 .
here we fix the event horizon at @xmath15 , and the cosmological horizon at @xmath33 .
we consider @xmath34 modes and three different charges , @xmath35.,width=340 ] the behavior of the potentials varies considerably over the range of parameters . in fig .
[ fig : pots ] we show @xmath31 for @xmath32 , @xmath33 , @xmath34 and three different values of the charge , @xmath35 .
a sufficient ( but not necessary ) condition for the existence of an unstable mode has been proven by buell and shadwick @xcite and is the following , _ r_b^r_cdr<0.[vcriterium ] the parametric region of instability in @xmath36 coordinates , according to criterim ( [ vcriterium ] ) , for @xmath34 .
top to bottom , @xmath37.,width=340 ] the instability region is depicted in figure [ fig : threshold ] for several spacetime - dimension @xmath1 , which can be compared with the numerical results by kz , their figure 4 .
it is apparent that condition ( [ vcriterium ] ) very accurately describes the numerical results for @xmath38 , a regime we explore below in section [ sec : ne ] .
as one moves away from extremality criterium ( [ vcriterium ] ) is just too restrictive . an improved analysis and refined criterium
would be necessary to describe the whole range of the numerical results .
nevertheless , figure [ fig : threshold ] is very clear : higher - dimensional ( @xmath0 ) rnds black holes are unstable for a wide range of parameters .
let us now specialize to the near extremal rnds black hole , which we define as the spacetime for which the cosmological horizon @xmath17 is very close ( in the @xmath39 coordinate ) to the black hole horizon @xmath16 , i.e. @xmath40 . the wave equation in this spacetime can be solved exactly , in terms of hypergeometric functions @xcite .
the key point is that the physical region of interest ( where the boundary conditions are imposed ) , lies between @xmath16 and @xmath17 .
thus , f ~2_b , [ approximation2 ] where we have introduced the surface gravity @xmath41 associated with the event horizon at @xmath42 , as defined by the relation @xmath43 . for near - extremal black holes ,
it is approximately _
b ~(1-nq^2 ) . in this limit , one can invert the relation @xmath44 of ( [ tortoise ] ) to get r= .
[ rtortoise ] substituting this on the expression ( [ approximation2 ] ) for @xmath14 we find f = .
[ approximation3 ] as such , and taking into account the functional form of the potentials for wave propagation , we see that for the near extremal rnds black hole the wave equation ( [ we ] ) is of the form + ^2-(,r)=0 , [ waveequation2 ] with v_0= the potential in ( [ waveequation2 ] ) is the well known pshl - teller potential @xcite . the solutions to ( [ waveequation2 ] )
were studied and they are of the hypergeometric type , ( for details see refs .
it should be solved under appropriate boundary conditions : & ~ & e^-ir _ * , r _ * - + & ~ & e^ir _ * , r _ * .
[ behavior1 ] these boundary conditions impose a non - trivial condition on @xmath45 @xcite , and those that satisfy both simultaneously are called quasinormal frequencies . for the pshl - teller potential
one can show @xcite that they are given by = _
b , j=0,1 , .... [ solution ] .[tab : exponent ] the threshold of instability for near - extremal rnds black holes ( i.e. , black holes for which the cosmological and event horizon almost coincide ) for @xmath34 modes .
we show the prediction from the exact , analytic expression obtained in the near extremal limit ( [ analytic ] ) , which we label @xmath46 and the one from criterium ( [ vcriterium ] ) which we label as @xmath47 .
both these results are compared to the numerical results by kz . [ cols="^,^,^,^,^,^,^ " , ] we conclude therefore that an instability is present whenever @xmath48 is negative .
the threshold of stability in the near - extremal regime is therefore given by = 0,[analytic ] the expression for @xmath49 is lengthy , and we wo nt present it here .
the values of the charge @xmath50 that satisfy the condition above are given in table i ( for @xmath34 ) , and compared to the prediction from the analysis in section [ sec : negativev ] , criterium ( [ vcriterium ] ) .
the agreement is excellent .
furthermore , we compare these predictions against the numerical results by kz , extrapolated to the extremal limit ( @xmath51 in kz notation ) .
the agreement is remarkable .
we have shown analytically that charged black holes in de sitter backgrounds are unstable for a wide range of charge and mass of the black hole , confirming previous numerical studies @xcite .
the stability properties of the extremal @xmath52 black hole remain unknown .
our methods and results and inconclusive at this precise point , further dedicated investigations would be necessary .
our analytical result in the near - extremal regime could be used to investigate further the nature of this instability , something we have not attempted to do here .
a possible refinement concerns the large-@xmath1 limit of the instability , where it _ could _
be possible to find an analytical expression throughout all range of parameters .
we have in mind results and techniques similar to those of kol and sorkin @xcite .
it would also be interesting to investigate the stability properties , using this or other techniques , of near - extremal kerr - ds black holes , which have recently been conjectured to have an holographic description @xcite .
we warmly thank roman konoplya and alexander zhidenko for useful correspondence and for sharing their numerical results with us .
this work was partially funded by fundao para a cincia e tecnologia ( fct)-portugal through projects ptdc / fis/64175/2006 , ptdc/ fis/098025/2008 , ptdc / fis/098032/2008 and cern / fp/109290/2009 .
r. emparan and r. c. myers , jhep
* 0309 * , 025 ( 2003 ) ; o. j. c. dias , p. figueras , r. monteiro , j. e. santos and r. emparan , arxiv:0907.2248 [ hep - th ] . v. cardoso , o. j. c. dias , j. p. s. lemos and s. yoshida , phys . rev .
d * 70 * , 044039 ( 2004 ) [ erratum - ibid .
d * 70 * , 049903 ( 2004 ) ] ; v. cardoso and o. j. c. dias , phys .
d * 70 * , 084011 ( 2004 ) ; v. cardoso and s. yoshida , jhep * 0507 * , 009 ( 2005 ) ; v. cardoso , o. j. c. dias and s. yoshida , phys .
d * 74 * , 044008 ( 2006 ) ; v. cardoso and j. p. s. lemos , phys .
b * 621 * , 219 ( 2005 ) ; h. kodama , prog . theor .
* 172 * , 11 ( 2008 ) ; a. n. aliev and o. delice , phys .
d * 79 * , 024013 ( 2009 ) ; h. kodama , r. a. konoplya and a. zhidenko , phys .
d * 79 * , 044003 ( 2009 ) ; k. murata , prog .
phys . * 121 * , 1099 ( 2009 ) ; n. uchikata , s. yoshida and t. futamase , phys .
d * 80 * , 084020 ( 2009 ) .
e. berti , v. cardoso and a. o. starinets , class .
* 26 * , 163001 ( 2009 ) . v. ferrari and b. mashhoon , phys . rev .
d * 30 * , 295 ( 1984 ) .
b. kol and e. sorkin , class .
* 21 * , 4793 ( 2004 ) .
d. anninos and t. hartman , arxiv:0910.4587 [ hep - th ] . | recent numerical investigations have uncovered a surprising result : reissner - nordstrm - de sitter black holes are unstable for spacetime dimensions larger than 6 . here
we prove the existence of such instability analytically , and we compute the timescale in the near - extremal limit .
we find very good agreement with the previous numerical results .
our results may me helpful in shedding some light on the nature of the instability . |
the cluster - update algorithm introduced for simulations of the potts model by swendsen and wang in 1987 @xcite has been a spectacular success , reducing the effect of critical slowing down by many orders of magnitude for the system sizes typically considered in computer simulation studies .
a number of generalizations , including an algorithm for continuous - spin systems and the single - cluster variant @xcite as well as more general frameworks for cluster updates @xcite , have been suggested following the initial work of ref .
the single bond update introduced by sweeny @xcite several years before swendsen s and wang s work is considerably less well known .
this is mostly due to difficulties in its efficient implementation in a computer code , which is significantly more involved than for the swendsen - wang algorithm . in deciding about switching the state of a given bond from inactive to active or _ vice versa _ , one must know the consequences of the move for the connectivity properties of the ensemble of clusters , i.e. , whether two previously disjoint clusters will become connected or an existing cluster is broken up by the move or , instead , the number of clusters will stay unaffected .
if implemented naively , these connectivity queries require a number of steps which is asymptotically close to proportional to the number of spins , such that the resulting _ computational critical slowing down _ outweighs the benefit of the reduced autocorrelation times of the updating scheme .
even though it was recently shown that the decorrelation effect of the single - bond approach is asymptotically _ stronger _ than that of the swendsen - wang approach @xcite , this strength can only be played once the computational critical slowing down is brought under control . here
, we use a poly - logarithmic dynamic connectivity algorithm as recently suggested in the computer science literature @xcite to perform bond insertion and removal operations as well as connectivity checks in run - times only logarithmically growing with system size , thus removing the problem of algorithmic slowing down . as the mechanism as well as
the underlying data structures for these methods are not widely known in the statistical physics community , we here use the opportunity to present a detailed description of the approach . for the convenience of the reader
, we also provide a python class implementing these codes , which can be used for simulations of the random - cluster model or rather easily adapted to different problems where dynamic connectivity information is required .
we consider the random - cluster model ( rcm ) @xcite which is a generalization of the bond percolation problem introducing a correlation between sites and bonds .
it is linked to the @xmath0-state potts model through the fortuin - kasteleyn transformation @xcite , generalizing the potts model to arbitrary real @xmath1 .
special cases include regular , uncorrelated bond percolation ( @xmath2 ) as well as the ising model ( @xmath3 ) . to define the rcm , consider a graph @xmath4 with vertex set @xmath5 , ( @xmath6 ) , and edge set @xmath7 , ( @xmath8 ) .
we associate an occupation variable @xmath9 with every edge @xmath10 .
we say that @xmath11 is _ active _ if @xmath12 and _ inactive _ otherwise .
the state space @xmath13 of the rcm corresponds to the space of all ( spanning is spanning if it contains all vertices of @xmath14 . ] ) sub - graphs , @xmath15 a configuration is thus represented as @xmath16\in\omega$ ] and corresponds uniquely to a sub - graph @xmath17 with @xmath18 the probability associated with a configuration @xmath19 is given by the rcm probability density function ( pdf ) @xmath20q^{k(\vec{\omega } ) } , \label{eq : rcm_pmf}\\ z(p , q ) & \equiv & \sum_{\vec{\omega}\in\omega } \left[\prod_{e\in e}p^{\omega(e)}(1-p)^{1-\omega(e)}\right]q^{k(\vec{\omega } ) } \label{eq : rcm_pmf_2},\end{aligned}\ ] ] where @xmath21 is the number of connected components ( clusters ) and @xmath22 denotes the rcm partition function .
more generally , as a function of the parameters @xmath23 $ ] , the density of active edges , and @xmath24 , the cluster number weight , these expressions define a family of pdfs .
it is worthwhile to mention a number of limiting cases . for @xmath25 eq .
( [ eq : rcm_pmf ] ) factorizes and corresponds to independent bond percolation with @xmath26 . in the limit of @xmath27 with fixed ratio @xmath28
, on the other hand , it corresponds to bond percolation with local probability @xmath29 and the condition of cycle - free graphs .
taking @xmath30 or in the limit of @xmath27 and @xmath31 constant for @xmath32 we obtain the ensemble of uniform spanning trees for connected @xmath14 . naturally , in the latter two limits every edge in a configuration is a bridge .
sweeny s algorithm @xcite is a local bond updating algorithm directly implementing the configurational weight ( [ eq : rcm_pmf ] ) .
we first consider its formulation for the limiting case @xmath33 of independent bond percolation .
for an update move , randomly choose an edge @xmath10 with uniform probability and propose a flip of its state from inactive to active or _
vice versa_. move acceptance can be implemented with any scheme satisfying detailed balance , for instance the metropolis acceptance ratio @xmath34 where @xmath35 and @xmath36 for insertions and deletions of edges , respectively .
this dynamical process is described by the following master equation : @xmath37 where @xmath38 ensures proper normalization of @xmath39 , given the normalization of @xmath40 .
the metropolis transition rates are then given by @xmath41 \label{eq : loc_transr}.\end{aligned}\ ] ] eq .
( [ eq : glob_transr ] ) expresses the uniform random selection of an edge and the corresponding edge dependent transition rate @xmath42 is defined in eq .
( [ eq : loc_transr ] ) .
the product of kronecker deltas ensures the single - bond update mechanism , i.e. , that only one edge per step is changed . from here , generalization to arbitrary @xmath24 is straightforward , leading to a modified transition rate @xmath43 .
\label{eq : transr_rcm}\end{aligned}\ ] ] we note that this metropolis update is more efficient than a heat - bath variant for any value of @xmath0 apart from @xmath2 , where both rules coincide . clearly , for @xmath44
, to compute the acceptance probability @xmath42 of a given trial move one must find @xmath45 , the change in connected components ( clusters ) induced by the move .
this quantity , equivalent to the question of whether the edge @xmath11 is a _ bridge _
, is highly non - local . determining it involves deciding whether there exists at least one alternative path of active edges connecting the incident vertices @xmath46 and @xmath47 that does not cross @xmath48 .
the dynamic connectivity problem is the task of performing efficient connectivity queries to decide whether two vertices @xmath46 and @xmath47 are in the same ( @xmath49 ) or different ( @xmath50 ) connected components for a dynamically evolving graph , i.e. , mixing connectivity queries with edge deletions and insertions . for a static graph
, such information can be acquired in asymptotically constant time after a single decomposition , for instance using the hoshen - kopelman algorithm @xcite . under a sequence of edge insertions ( but no deletions ) , it is still possible to perform all operations , insertions and connectivity queries , in practically constant time using a so - called union - and - find ( uf ) data structure combined with path - compression and tree - balance heuristics @xcite . this fact has been used to implement a very efficient algorithm for the ( uncorrelated ) percolation problem @xcite .
an implementation of sweeny s algorithm , however , requires insertions as well as deletions to ensure balance .
hence , we need to be able to remove edges without the need to rebuild the data structure from scratch .
.asymptotic run - time scaling at criticality of the elementary operations of insertion or deletion of internal or external edges , respectively , using sequential breadth - first search ( sbfs ) , interleaved bfs ( ibfs ) , union - and - find ( uf ) or the fully dynamic connectivity algorithm ( dc ) as a function of the linear system size @xmath51 . [ cols= " < , < , < , < , < " , ] this goal can be reached using a number of different techniques .
building on the favorable behavior of the uf method under edge insertions and connectivity queries , the data structure can be updated under the removal of an external edge ( bridge ) by performing breadth - first searches ( bfss ) through the components connected to the two ends @xmath46 and @xmath47 of the edge @xmath52 .
alternatively , one might try to do without any underlying data structure , answering each connectivity query through a separate graph search in breadth - first manner . in both cases ,
the process can be considerably sped up by replacing the bfss by _ interleaved _ traversals alternating between vertices on the two sides of the initial edge and terminating the process as soon as one of the two searches comes to an end @xcite .
as , at criticality of the model , the sizes of the two cluster fragments in case of a bridge bond turn out to be very uneven on average , this seemingly innocent trick leads to dramatic run - time improvements @xcite .
the asymptotic run - time behavior of insertion and deletion steps for internal and external edges and the algorithms based on bfs or uf data structures is summarized in table [ tab : scaling ] for the case of simulations on the square lattice of edge length @xmath51 .
we expect the same bounds with the corresponding exponents to hold for general critical hypercubic lattices . here
, @xmath53 denotes the finite - size scaling exponent of the susceptibility and @xmath54 is a geometric exponent related to the two - arm crossing behavior of clusters @xcite .
we note that @xmath55 for @xmath56 in two dimensions .
asymptotically , it is the most expensive operation which dominates the run - time of the algorithm and , as a consequence , it turns out that ( for the square lattice ) a simple bfs with interleaving is more efficient than the approach based on union - and - find , cf . table [ tab : scaling ] . in any case , the implementations discussed so far feature a computational effort for a sweep of bond updates that scales faster than linearly with the system size , thus entailing some computational critical slowing down .
it is found in ref .
@xcite that for most choices of @xmath0 , this effect appears to asymptotically destroy any advantage of a faster decorrelation of configurations by the sweeny algorithm as compared to the swendsen - wang method .
an alternative technique based on more complicated data structures allows to perform any mix of edge insertions , deletions and connectivity queries in _ poly - logarithmic _ run - time per operation @xcite .
poly - logarithmic here denotes polynomials of powers of the logarithm of the independent variable , e.g. , system size @xmath51 , of the form @xmath57 here the base of the logarithm is not important and changes only the coefficients . in the following all logarithms are with respect to base @xmath58 .
given the observation of generally faster decorrelation of configurations by the sweeny algorithm in the sense of smaller dynamical critical exponents @xcite , the use of such ( genuinely ) dynamic connectivity algorithms ( dc ) allows for an asymptotically more efficient simulation of the critical random - cluster model @xcite . in the following ,
we discuss the basic ideas and some details of the algorithm employed here .
the dc algorithm is based on the observation that for a given sub - graph @xmath59 it is possible to construct a spanning forest @xmath60 which is defined by the following properties : * @xmath61 in @xmath60 if and only if @xmath61 in @xmath59 * there exists exactly one path for every pair @xmath46 , @xmath47 with @xmath61 in other words a spanning forest of a graph associates a spanning tree to every component , an example is given in fig .
[ spanning_graph_0 ] ( solid lines only ) .
one advantage of @xmath60 is that it has fewer edges than @xmath62 , but represents the same connectivity information .
for the sub - graph @xmath62 , the distinction of _ tree edges _ @xmath63 and _ non - tree edges _
@xmath64 allows for a cheap determination of @xmath45 . for the case of deleting an edge in @xmath65
we know that there is an alternative path connecting the adjacent vertices , namely the path in @xmath60 , so this edge was part of a cycle and we conclude @xmath66 .
if we insert an edge whose adjacent vertices are already connected in @xmath60 then we come to the same conclusion .
if , on the other hand , we want to insert a tree edge , i.e. , an edge with adjacent vertices @xmath46 , @xmath47 not yet connected , we observe that because @xmath46 and @xmath47 belong to separate spanning trees before the insertion of @xmath11 , the new spanning subgraph obtained by linking @xmath46 and @xmath47 via @xmath11 is still a spanning tree .
hence the only modification on the spanning forest for the insertion of a tree edge is the amalgamation of two trees .
this can be done in @xmath67 steps by using the following idea of ref .
@xcite which also supports the deletion of bridges . for a given component @xmath68 in @xmath59
we transform the corresponding tree @xmath69 in @xmath60 into a directed circuit by replacing every edge @xmath70 by two directed edges ( arcs ) @xmath71 $ ] and @xmath72 $ ] and every vertex by a loop @xmath73 $ ] .
figure [ circuit_mapping ] illustrates how to translate edge insertions or deletions into / from the spanning forest to modifications on the directed circuits . deleting an edge from a tree
splits it into two trees .
the directed circuit therefore splits into two circuits corresponding to the two trees . when inserting a tree edge , i.e.
, merging two trees , we join the circuit by the two arcs , corresponding to @xmath11 , at the vertices incident to @xmath11 . by storing
the directed circuits for every component in so called euler tour sequences ( ets ) , @xcite , all necessary manipulations on the directed circuits can be done in @xmath67 operations if we store each ets in a separate balanced search tree such as , for instance , a red - black , avl , or b tree @xcite .
alternatively one can use self - adjusting binary search trees , so called splay trees @xcite . in this case
the bound is amortized , i.e. , averaged over the complete sequence of operations . due to a somewhat simpler implementation and the fact that a monte carlo simulation usually consists of millions of operations in random order naturally leading to amortization , we concentrated on the splay - tree approach .
based on the ets representation we can translate connectivity queries into checking the underlying search tree roots for equivalence .
an in - depth discussion of the exact manipulations on the representing ets is beyond the scope of this article and we refer the interested reader to the literature @xcite .
here we restrict ourselves to considering the deletion of a bridge as an example .
the initial graph is the one illustrated in the left panel of fig .
[ circuit_mapping ] .
one corresponding ets sequence of the linearized directed cycle in the right panel of fig .
[ circuit_mapping ] is the following : @xmath74\rightarrow[1,2]\rightarrow[2,2]\rightarrow[2,4]\rightarrow[4,4]\rightarrow[4,3]\rightarrow[3,3]\rightarrow[3,4]\rightarrow[4,2]\rightarrow[2,1 ] .
\label{eq : sequence}\ ] ] suppose now that we delete edge @xmath75 from the tree ( cf . the dashed line in fig . [ circuit_mapping ] ) .
this will result in a split of the component @xmath76 into two parts @xmath77 and @xmath78 . in this case
the deletion of edge @xmath75 translates into a cut of the original sequence @xmath79 at the two arcs @xmath80 $ ] and @xmath81 $ ] .
the sequence @xmath82 of arcs between these two edges corresponds to @xmath83 and the concatenation of the remaining sequences without the two arcs corresponding to @xmath75 results in @xmath84 representing @xmath85 : @xmath86\rightarrow[1,2]\rightarrow[2,2]\rightarrow[2,1 ] , \label{eq : et_1}\\ \mathcal{e}_{2 } & = & [ 4,4]\rightarrow[4,3]\rightarrow[3,3]\rightarrow[3,4 ] .
\label{eq : et_2}\end{aligned}\ ] ] in summary , we see that by mapping every component in @xmath59 to a tree in @xmath60 and every such tree to a directed circuit which we store in an ets we are able to perform edge insertions / deletions into / from @xmath87 as well as connectivity queries with an amortised @xmath67 computational effort .
the remaining operation not implemented efficiently by the provisions discussed so far is the deletion of edges from @xmath60 which are not bridges , i.e. , for which a replacement edge exists outside of the spanning forest .
the dc algorithm first executes the tree splitting as in the case of a bridge deletion . additionally , however , it checks for a reconnecting edge in the set of non - tree edges .
if such an edge is found , it concludes @xmath66 and merges the two temporary trees as indicated above by using the located non - tree edge , which hence now becomes a tree edge . if , on the other hand , no re - connecting edge is found , no additional work is necessary as the initially considered edge is a bridge . to speed up the search for replacement edges ,
we limit it to the smaller of the two parts @xmath85 and @xmath83 as all potential replacement edges must be incident to both components . to allow for efficient searches for non - tree edges incident to a given component using the ets representation ,
the search - tree data structures are augmented such that the loop arc for every vertex stores an adjacency list of non - tree edges ( vertices ) incident to it .
further , every node in the underlying search tree representing the ets carries a flag indicating if any non - tree edge is available in the sub - tree [ which is a sub - tree of the search tree and not of @xmath60 ] .
this allows for a search of replacement edges using the euler tour and ensures that any non - tree edge can be accessed in @xmath67 time .
it turns out that exploiting this observation is , in general , beneficial but not sufficient to ensure the amortized time complexity bound indicated for the dc algorithm in general .
suppose that the graph consists of a giant homogeneous component with @xmath89 and @xmath90 and the edge deletion results , temporarily setting aside the question of possible replacement edges , in two trees with @xmath91 and @xmath92 incident edges , respectively , where @xmath93 .
then the computational effort caused by scanning all possible non - tree edges is clearly @xmath94 . in amortizing onto the insertions
performed to build up this component , every such non - tree edge carries a weight of @xmath67 . if this case occurs sufficiently frequently , it will be impossible to bound the amortized cost per operation .
this problem is ultimately solved in the dc algorithm by the introduction of an edge hierarchy .
the intuitive idea is to use the expense of a replacement edge search following a deletion to reduce the cost of future operations on this edge .
this is done in such a way as to separate dense from sparse clusters and more central edges from those in the periphery of clusters . by amortizing the cost of non - tree edge scans and level increases over edge insertions it follows that one can reduce the run - time for graph manipulations to an amortized
@xmath95 and @xmath67 for connectivity queries @xcite .
each time an incident non - tree edge is checked and found unsuitable for reconnecting the previously split cluster we promote it to be in a higher level . if we do this many times for a dense component we will be able to find incident non - tree edges very quickly in a higher level .
these ideas are achieved in the dc algorithm by associating a level function to each @xmath10 , @xmath96 based on this level function , one then constructs sub - graphs @xmath97 with the property @xmath98 this induces a hierarchy of sub - graphs : @xmath99 as described above , for every sub - graph we construct a spanning forest @xmath100 . clearly the same hierarchy holds for the family of spanning forests . in other words
the edges in level @xmath101 connect components / trees of level @xmath102 .
if an edge has to be inserted into @xmath59 , then it is associated to a level @xmath103 and hence it is in @xmath104 . to achieve an efficient search for replacement edges ,
the algorithm adapts the level of edges after deletions of tree edges in a way which preserves the following two invariants @xcite : 1 .
the maximal number of vertices in a component in level @xmath102 is @xmath105 .
2 . any possible replacement edge for a previously deleted edge @xmath11 with level @xmath106 has level @xmath107 .
trivially , both invariants are fulfilled when all edges have level 0 .
we now have to specify how exactly the idea of keeping important edges at low levels and unimportant ones at higher levels is implemented . to do this , suppose we deleted an edge from @xmath108 , i.e. , at level @xmath102 , and temporarily have @xmath109 where ( say ) @xmath110 is the smaller of the two , i.e.
, it has less vertices . because of invariant ( i ) it follows that we are allowed to move the tree @xmath110 ( which is now at most half the size of @xmath111 ) to level @xmath112 by increasing the level of all tree edges of @xmath110 by one .
after that we start to search for a replacement edge in the set of non - tree edges stored in the ets of @xmath110 in level @xmath102 where it also remains because of the fact that @xmath113 .
for every scanned non - tree edge we have two options : * it does not reconnect @xmath110 and @xmath114 and has therefore both ends incident to @xmath110 . in this case
, we increase the level of this edge @xmath115 .
this implements the idea of moving unimportant edges in `` dense '' components to higher levels . *
it does reconnect and hence we re - insert it at level @xmath116 .
if we have not found a replacement edge at level @xmath102 we continue at level @xmath101 .
the search terminates after unsuccessfully completing the search at level @xmath116 or when a replacement edge was found . in the first case it follows @xmath117 whereas in the second case
@xmath68 remains unchanged .
implementing this replacement - edge search following any tree - edge deletion introduces an upward flow of edges in the hierarchy of graphs and the level of an edge in the current graph never decreases . focusing on a single edge
, we see that it is sequentially moved into levels of smaller cluster size and hence the cost of future operations on this edge is reduced . taking this into account it follows that the insertion of an edge has a cost of @xmath67 for inserting at level @xmath116 plus it also `` carries '' the cost of all possible @xmath67 level increases with cost of @xmath67 each resulting in @xmath118 amortized per insertion .
deletions on the other hand imply a split of cost @xmath67 in @xmath67 levels . in case of an existing replacement edge another contribution of @xmath67 caused by an insertion at level 0
the contribution of moving tree edges to higher levels and searching for replacement edges ( moving non - tree edges up ) is already paid for by the sequence of previous insertions ( amortization ) .
the only missing contribution is the @xmath67 effort for obtaining the next replacement edge in an ets . in total
, deletions hence have an amortized computational cost of @xmath95 .
we tested the performance of the current dc implementation in the context of sweeny s algorithm in comparison to the simpler approaches based on breadth - first search and union - and - find strategies . while the algorithm discussed here
allows all operations to be performed in poly - logarithmic time , due to the complicated data structures the constants are relatively large .
our results show consistency with the poly - logarithmic run - time bounds derived .
it appears , however , that very large system sizes are required to clearly see the superior asymptotic performance of the dc algorithm as compared to the bfs and uf implementations .
for details see the more elaborate discussion in ref . @xcite . as an example , fig .
[ fig : fig_dc_run_times ] shows the average run - time per edge operation as a function of the system size for three different choices of parameters .
average run - time for sweeny s algorithm for the rcm for different values of @xmath0 at the critical point @xmath119 on the 2d square lattice with periodic boundary conditions . ]
apart from run - time considerations , the implementation has a rather significant space complexity .
since we maintain @xmath67 overlapping forests over the @xmath120 vertices , the space complexity is @xmath121 .
a heuristic suggested in ref .
@xcite to decrease memory consumption is a truncation of higher edge levels as these are , for the inputs or graphs considered in our application , sparsely populated .
we checked the impact on our implementation by comparing run - times and memory consumptions for a truncation @xmath122 .
we did not see any significant change in the run - time . on the other hand we observed a reduction of almost a factor of two in the memory consumption .
this conforms to our observation that during the course of a simulation almost no edges reached levels beyond @xmath123 for system sizes @xmath124 where the actual maximal level according to eq .
( [ eq : levels ] ) is @xmath125 . likewise , a number of further optimizations or heuristics are conceivable to improve the typical run - time behavior .
this includes a sampling of nearby edges when looking for a replacement edge before actually descending into the edge level hierarchy @xcite .
a number of such heuristics and experimental comparisons of fully and partially dynamics connectivity algorithms has been discussed in the recent literature , see refs .
a full exploration of these possibilities towards an optimal implementation of the dc class of algorithms for the purpose of the sweeny update is beyond the scope of the current article and forms a promising direction for future extensions of the present work .
we provide a python class @xcite encompassing four different implementations of sweeny s algorithm based on : * sequential breadth - first searches ( sbfs ) * interleaved breadth - first searches ( ibfs ) * union - and - find with interleaved breadth - first searches ( uf ) *
poly - logarithmic dynamic connectivities as discussed here ( dc ) the package is built on top of a c library and it is therefore possible to use the library in a stand - alone compiled binary . the necessary source code is also provided . for more details see the related project documentation @xcite .
the source code is published under the mit license @xcite . here
we give a basic usage example , which simulates the rcm with @xmath3 ( the ising model ) at @xmath126 , using an equilibration time of @xmath127 sweeps , a simulation length of @xmath128 sweeps , and random number seed @xmath129 using the dc implementation : .... from sweeny import sweeny sy = sweeny(q=2.,l=64,beta = np.log(1 .
+ np.sqrt(2.)),coupl=1 . ,
cutoff=1000,tslength=10000,rngseed=1234567,impl='dc ' ) sy.simulate ( ) .... in order to extract an estimate , say , of the binder cumulant @xmath130 we need to retrieve the time series for @xmath131 and @xmath132 , ....
sec_cs_moment= sy.ts_sec_cs_moment four_cs_moment = sy.ts_four_cs_moment sec_cs_moment * = sec_cs_moment binder_cummulant = four_cs_moment.mean()/sec_cs_moment.mean ( ) .... once an instance of the sweeny class is created , it is easy to switch the algorithm and parameters as follows : .... sy.init_sim(q=1.3,l=64,beta=np.log(1.+np.sqrt(1.3.)),coupl=1 . ,
cutoff=5000,tslength=50000,rngseed=7434,impl='ibfs ' ) ....
we have shown how to implement sweeny s algorithm using a poly - logarithmic dynamic connectivity method and we described the related algorithmic aspects in some detail . we hope that the availability of the source code and detailed explanations help to bridge the gap between the computer science literature on the topic of dynamic connectivity problems and the physics literature related to mc simulations of the rcm , specifically in the regime @xmath133 . the availability of an efficient dynamic connectivity algorithm opens up a number of opportunities for further research .
this includes studies of the tricritical value @xmath134 where the phase transition of the random - cluster model becomes discontinuous for dimensions @xmath135 @xcite as well as the nature of the ferromagnetic - paramagnetic transition for @xmath27 and @xmath136 @xcite .
e.m.e . would like to thank p. mac carron for carefully reading the manuscript .
10 url # 1#1urlprefix[2][]#2 swendsen r h and wang j s 1987 _ phys .
lett . _ * 58 * 8688 | we review sweeny s algorithm for monte carlo simulations of the random cluster model .
straightforward implementations suffer from the problem of computational critical slowing down , where the computational effort per edge operation scales with a power of the system size . by using a tailored dynamic connectivity algorithm we are able to perform all operations with a poly - logarithmic computational effort .
this approach is shown to be efficient in keeping online connectivity information and is of use for a number of applications also beyond cluster - update simulations , for instance in monitoring droplet shape transitions . as the handling of the relevant data structures is non - trivial
, we provide a python module with a full implementation for future reference . |
atmospheric and solar neutrino experiments have demonstrated that neutrinos have mass , implying a lower limit on the total mass of 0.056 ev @xcite .
ongoing and future direct experiments will be sensitive to the absolute neutrino mass scale : for instance , katrin will measure or constrain the electron neutrino mass to the 0.2 ev level .
a thermal neutrino relic component in the universe can impact both the expansion history and growth of structure .
cosmological probes of neutrinos are therefore complementary to direct experiments and provide some of the tightest constraints on both the number of relativistic species present in the early universe and the sum of neutrino masses @xcite . however , up to date , these constraints were dependent on the assumed cosmological model .
for example , bounds on the sum of the neutrino masses could be relaxed by , for example , allowing coupling in the dark sector @xcite or allowing hadronic axions @xcite . in this paper
we summarize how cosmic neutrinos affect cosmic history , examine how recently improved cosmological measurements of the low redshift universe can help constrain these neutrino properties , and scrutinize the impact of our assumptions about the cosmological model . in the standard model for particle physics
there are three massless neutrino species with weak interactions .
neutrinos decouple early in cosmic history , when the temperature was @xmath18 mev , and thereafter contribute to the relativistic energy density with an effective number of species , @xmath19 @xcite .
cosmology is sensitive to the physical energy density in relativistic particles in the early universe @xmath20 which , in the standard model ( for cosmology ) , includes only photons and neutrinos @xmath21 where @xmath22 is the energy density in photons and @xmath23 is the energy density in one active neutrino .
as @xmath22 is extremely well constrained , @xmath20 can be used to constrain neutrino properties ; deviations from @xmath19 would signal non - standard neutrino features or additional relativistic relics .
@xmath24 impacts the big bang nucleosynthesis epoch through its effect on the expansion rate . therefore measurements of the primordial abundance of light elements can constrain @xmath24 @xcite .
these constraints rely on physics at the time of big - bang nucleosynthesis ( @xmath25 mev ) , and in several non - standard models the energy density in relativistic species can change at later time ( e.g. at the last scattering , @xmath26 ev ) .
free - streaming relativistic particles affect the cosmic microwave background ( cmb ) through their relativistic energy density , which alters the epoch of matter - radiation equality , and through their anisotropic stress @xcite .
because the redshift of matter - radiation equality is well constrained by the ratio of the third to first cmb peak height , the first effect defines a degeneracy between @xmath27 and @xmath24 @xcite : @xmath28 because of the second effect , wmap5 data favors @xmath19 over @xmath29 at the 99.5% confidence level @xcite .
neutrinos with mass @xmath30 ev become non - relativistic after the epoch of recombination probed by the cmb , so that allowing massive neutrinos alters matter - radiation equality for fixed @xmath27 .
their radiation - like behavior at early times changes the expansion rate , shifting the peak positions , but this is somewhat degenerate with other cosmological parameters .
therefore , wmap5 alone constrains @xmath31 ev at the 95% confidence in a flat @xmath2cdm universe , and this constraint relaxes to @xmath32 ev for a flat @xmath6cdm universe , assuming @xmath33 @xcite .
figure 17 of @xcite shows that there remain degeneracies between @xmath34 and both @xmath9 and @xmath35 ; in this paper we show that currently available low - redshift measurements can break these degeneracies , yielding improved and robust constraints on @xmath34 . after the neutrinos become non - relativistic , their free - streaming damps power on small scales and therefore modifies the matter power spectrum in the low redshift universe .
the latest large - scale structure constraints from the sloan digital sky survey luminous red galaxy sample in combination with wmap5 yield @xmath36 ev in a flat @xmath2cdm model with @xmath19 @xcite .
in addition , because the measurement of the matter power spectrum provides a separate constraint on the transfer function , this data combination also yields a constraint on @xmath37 in a flat @xmath2cdm cosmology with no massive neutrinos .
we will show that the combination of low redshift data used in this paper provides tighter neutrino constraints in both of these cases . relaxing the assumptions about the cosmic expansion history , e.g. , requiring @xmath38 vs. allowing @xmath6 as
a free parameter @xcite or allowing coupling in the dark sector can relax constraints on neutrino masses @xcite .
as an example we study constraints on neutrino masses in the model of @xcite , allowing @xmath7 , @xmath6 , and coupling strength @xmath8 to vary simultaneously . in section [ whichconstraints ]
we discuss the low - redshift observational constraints we will make use of to improve neutrino constraints , paying careful attention to how the constraints generalize to the models we consider . in sections [ sumneutrinomass ] and [ nrelsec ] we present new constraints on neutrino masses and the effective number of relativistic species at matter - radiation equality .
we also make a detailed comparison with results from different cosmological probes available in the literature . in section [ conc ]
we summarize our findings and argue that they are robust to a wide variety of alterations of the cosmological model .
in subsections [ genh0 ] and [ genmaxbcg ] we discuss the low redshift constraints we make use of in our analysis , paying particular attention to their generalization to models not initially considered in their analyses . in subsection
[ method ] we review our standard bayesian methodology for computing parameter constraints and upper limits , and introduce the profile likelihood as a means to check the dependence on these constraints on our priors . ref .
@xcite presents a redetermination of the hubble constant that makes use of cepheid variables in sn ia host galaxies and in the `` maser galaxy '' ngc 4258 , along with an expanded sn ia sample at @xmath39 .
we reinterpret the reported constraint , @xmath0 km s@xmath40 mpc@xmath40 , as a constraint on the inverse luminosity distance at the effective redshift @xmath41 ( riess , private communication ) .
this correctly accounts for the slight dependence on @xmath6 shown in figure 14 of @xcite , and generalizes the constraint to an arbitrary cosmology . in practice ,
when combining with other data sets , the cosmology dependence of this constraint is quite small . ref .
@xcite derives cosmological constraints from the maxbcg cluster catalog @xcite , which identifies spatial overdensities of bright red galaxies in sdss dr4 + imaging data and results in a nearly volume - limited catalog with photometric redshifts in the range @xmath42 to @xmath43 .
maxbcg clusters are classified by @xmath44 , the number of red - sequence galaxies within a scaled radius such that the average galaxy overdensity interior to that radius is 200 times the mean galaxy density .
the group multiplicity as a function of @xmath44 is one of two primary observable inputs to the maxbcg posterior .
the second is the mean mass measured via gravitational lensing for 5 bins in @xmath44 from @xcite .
the analysis of ref .
@xcite assumes a flat @xmath2cdm cosmology with @xmath33 massless neutrinos , where @xmath35 and @xmath45 are allowed to vary , and @xmath46 , @xmath47 , and @xmath48 are held fixed
. there are four nuisance parameters : two to define the mean power law relationship between @xmath44 and halo mass @xmath49 , one to quantify the size of the log - normal scatter about this relation , and one for the lensing mass bias parameter .
the principal cosmological constraint derived from the maxbcg catalog is @xmath50 in section 5.2.1 , ref .
@xcite addresses the limitations of their assumptions on cosmological parameters .
they find that using a gaussian prior on @xmath51 and @xmath52 does not alter their @xmath53 constraint .
in addition , for a prior with @xmath54 ev , the @xmath53 constraint is not altered .
massive neutrinos suppress the linear power spectrum , but because @xmath35 corresponds to the mass scale to which maxbcg is most sensitive , the impact of the alteration in power spectrum shape is minimal on the final constraints . in principle , massive neutrinos also alter growth in the nonlinear regime . ref .
@xcite shows that power suppression is enhanced in the nonlinear regime for massive neutrinos , implying that the upper bound reported in the present paper is robust .
therefore , since the range of total neutrino mass we explore in this paper is well below @xmath55 ev , we can safely apply the maxbcg constraint . in the case of massive neutrinos
, we also study the maxbcg constraint in combination with wmap5 for a @xmath2cdm model including spectral index running .
when we evaluate the spectral index @xmath56 for all @xmath57/mpc from the wmap5 constraints in this model , the value falls between @xmath58 and @xmath55 for 68% of the time .
this range satisfies the stated maxbcg tolerance to variation in the spectral index , and therefore we may apply the maxbcg constraint unaltered in the models we study including massive neutrinos .
the maxbcg analysis did not directly address models in which @xmath24 is varied .
however , we have verified that within the space of parameters allowed by wmap5 and the @xmath9 constraint of @xcite in a @xmath2cdm+@xmath24 model , the impact of @xmath24 on the shape of the linear power spectrum can be compensated by a change in @xmath59 that is within the maxbcg stated tolerances .
therefore we may apply the constraint on the power spectrum amplitude through @xmath35 in equation [ maxbcgconstraint ] without modification when @xmath24 is varied .
we use the standard markov chain monte carlo ( mcmc ) scheme to probe the posterior distribution for several different cosmological models @xcite . the publicly available wmap5 chains which we resample have uniform priors on the parameters @xmath60 , the baryon density , cold dark matter density , dark energy density , reionization optical depth , scalar spectral index at @xmath61 mpc@xmath40 , amplitude of curvature perturbations at @xmath62 , and sz marginalization factor .
the mcmc chain we produce for the the dark coupling model of @xcite uses the same parameters , but replaces @xmath63 with the parameter @xmath64 , the ratio of the approximate sound horizon to the angular diameter distance , which is used by cosmomc @xcite . for a more `` frequentist '' statistic ,
the profile likelihood can be computed from the mcmc chain @xcite .
since we have 7 or more uninteresting parameters and only one parameter for which we want to report constraints ( say , @xmath65 ) , for each value of @xmath65 we find the maximum likelihood value @xmath66 , regardless of the values assumed by other parameters .
we then consider @xmath67 as a function of the model parameter @xmath65 , or function thereof ; here @xmath68 is the maximum likelihood over the entire chain .
we examine the profile likelihood for both @xmath34 ( section [ profilelike ] ) and @xmath24 ( section [ neffprofilelike ] ) .
as long as the posterior distribution is singly peaked , this profile should be independent of the bayesian priors ; but note that the profile likelihood will become noisier as @xmath69 decreases , since the mcmc method means the chains do not densely sample low @xmath69 regions .
we begin in section [ mnusec ] by studying the @xmath2cdm model and a few one - parameter extensions explored in the wmap5 analysis . in section [ darkcouplingresults ] we study the dark coupling model of @xcite as an example of a model which significantly relaxes the model assumptions about the expansion history by varying three additional parameters .
we consider the profile likelihood to study the dependence of our constraints on our cosmological priors and check the consistency of the data sets we examine .
finally we compare with previous constraints on @xmath34 in the literature using a variety of data sets .
@xmath2cdm wmap5-only constraints with @xmath12 are shown in blue .
the black transparent contours show how the constraints on @xmath70 and @xmath9 degrade when @xmath34 is left as a free parameter .
the dotted lines show contours on which @xmath71 ev for the wmap5-only posterior distribution .
the green shows the constraints when maxbcg and @xmath9 constraints are also included .
the 95% confidence upper limit is reduced from 1.3 ev to 0.4 ev . ] same as figure [ fig : mnulcdm ] but for the @xmath2cdm model with base dataset wmap@xmath72bao@xmath72sn .
the dotted lines show contours on which @xmath73 ev for the wmap@xmath72bao@xmath72sn posterior distribution . ]
llcccc + model & base dataset & & @xmath72maxbcg & @xmath74 & @xmath72maxbcg@xmath74 + @xmath2cdm & wmap5 & 1.3 & 1.1 & 0.59 & 0.40 + @xmath2cdm & wmap5@xmath72bao@xmath72sn & 0.67 & 0.35 & 0.59 & 0.31 + @xmath2cdm @xmath75 & wmap5 & 1.34 & 1.25 & 0.54 & 0.39 + @xmath2cdm @xmath76 & wmap5 & 1.36 & 1.18 & 0.83 & 0.40 + @xmath6cdm & wmap5@xmath72bao@xmath72sn & 0.80 & 0.52 & 0.72 & 0.47 + dark coupling & wmap5@xmath77sn & - & - & 0.51 & - + @xmath2cdm & wmap5 & 1.3 & 0.82 & 0.50 & 0.33 + @xmath2cdm & wmap5@xmath72bao@xmath72sn & 0.66 & 0.32 & 0.56 & 0.30 + @xmath2cdm @xmath75 & wmap5 & 1.28 & 1.15 & 0.63 & 0.43 + @xmath2cdm @xmath76 & wmap5 & 1.23 & 0.86 & 0.72 & 0.30 + @xmath6cdm & wmap5@xmath72bao@xmath72sn & 0.80 & 0.44 & 0.74 & 0.44 + dark coupling & wmap5@xmath77sn & - & - & 0.55 & - + figure [ fig : mnulcdm ] shows the wmap5-only constraints in a @xmath2cdm model when @xmath12 ( blue contours ) .
the solid black contours show that when @xmath34 is allowed to vary , the constraints on @xmath9 and @xmath70 weaken substantially .
while both the maxbcg and @xmath9 constraints are in good agreement with the @xmath2cdm values , they effectively exclude high neutrino masses ( green contours ) . adding both constraints
reduces the 95% confidence limit on @xmath34 from 1.34 to 0.4 in a flat @xmath2cdm cosmology .
figure [ fig : mnulcdmbaosn ] shows the same scenario , but when bao and sn data have also been combined with wmap5 .
since those probes break the degeneracy between @xmath9 and @xmath45 in @xmath2cdm cosmology , @xmath9 is well constrained , almost independently of @xmath34 . in this case
the maxbcg result drives the reduction of the 95% confidence bound from 0.67 ev to 0.31 ev . in table
[ table : mnu ] we present these and several other one dimensional constraints on @xmath34 for the @xmath2cdm model , as well as several other data and model variants . in the upper half of the table
the limits are bayesian , i.e. they represent the @xmath34 value below which 95% of the weighted points in the mcmc chain lay .
the lower half of the table approximates a frequentist interpretation ( a more detailed discussion of this is in sec .
[ profilelike ] ) : we report the largest value of @xmath34 in the chain for which @xmath78 . while this limit can be interpreted in terms of numbers of standard deviations only for gaussian distributions , it is prior - independent .
we have checked that the @xmath34 distribution is well approximated by a gaussian .
the constraint on @xmath34 from wmap5@xmath72maxbcg@xmath74 does not relax when running or tensors are allowed .
the constraint relaxes slightly to @xmath79 ev at the 95% confidence level when @xmath6 is allowed to vary , but bao and sn constraints are also included .
the constraint relaxes because the maxbcg constraint - @xmath34 degeneracy is less steep for this model , since changing @xmath6 also impacts the growth of structure .
these examples indicate that when the expansion history is fixed , @xmath3 ev ( 95% confidence ) is a robust upper limit ; changes in the primordial fluctuations that are consistent with wmap5 do not impact this bound . however , when the expansion history is altered at late times ( @xmath80 but constant ) , this bound relaxes slightly .
as an extreme example of this , we study the dark coupling model of @xcite next . in this subsection
we consider the particular model of coupling in the dark sector proposed by ref .
@xcite . in their model
the densities of dark matter and of dark energy do not evolve independently as the universe expands but are coupled , with an interaction strength parameterized by @xmath8 as follows : @xmath81 where @xmath82 and @xmath83 are the energy momentum tensors for the dark matter and dark energy components , respectively , @xmath84 denotes the dark matter four velocity and @xmath85 the hubble parameter . the coupling strength @xmath8 alters the expansion history through changes in the matter and dark energy density evolution @xcite : @xmath86 where
@xmath87 , the superscript @xmath88 denotes the present - day quantity , and @xmath89 denotes the scale factor .
we adopt this model as a suitable example for which to study whether the constraint on @xmath34 relaxes when several free parameters are allowed to simultaneously vary .
it was shown by ref .
@xcite that allowing for a coupling in the dark sector dramatically relaxes cosmological constraints on neutrino mass . ref .
@xcite presents a slightly different model for dark coupling , which improves on existing models in the literature as it does not suffer from instabilities ; they present cosmological constraints on this dark coupling parameterization in a very general model , allowing the curvature , equation of state of dark energy @xmath90 , total neutrino mass , and the amplitude of dark coupling @xmath91 between dark matter and dark energy to vary along with the standard @xmath2cdm cosmological parameters .
these additional parameters allow the expansion history and growth of structure to deviate appreciably from the @xmath2cdm assumption .
we will explore the cosmological constraints on the coupling strength @xmath8 in a future work alongside this model s non - standard growth of structure ; because of the large deviations from the @xmath2cdm growth of structure , the maxbcg cluster constraint , which is based on the standard spherical collapse - based correspondence between the linear matter power spectrum and the resulting halo mass function in @xmath2cdm cosmologies , can not be applied to the @xcite model . for this model
, we therefore substitute the maxbcg constraints with the sdss dr7 luminous red galaxies @xmath92 constraint of @xcite as a robust low redshift probe of @xmath34 .
we study the dark coupling model using the mcmc method as implemented in the cosmomc package @xcite .
several datasets must be included to avoid long degeneracies in the parameter space .
we include wmap5 data along with several low redshift datasets : the new @xmath9 constraint @xcite , sdss dr7 luminous red galaxies halo power spectrum @xmath92 @xcite , and union sn sample @xcite .
the only important prior we enforce is @xmath93 ; surprisingly , even models consistent with these four cosmological probes can have negligible @xmath94 today since the coupling accelerates the transition to dark energy domination .
note that massive neutrinos suppress @xmath92 in a scale - dependent fashion , thus , by including the @xmath92 measurements , one expects to be able to disentangle the effects of neutrino mass from the effects of alterations of the late - time expansion history .
figure [ fig : mcmc ] shows constraints from these four observational probes ( red contours ) : @xmath34 is not degenerate with dark coupling @xmath8 , and the resulting one dimensional constraint on @xmath34 is 0.51 ev . for completeness , we also show in figure [ fig : mcmc ] how the constraints would tighten _ if _ the maxbcg constraint were applicable ( blue contours ) .
the constraint on the neutrino mass is unaffected .
two - dimensional constraints in @xmath95 in the dark coupling model of ref .
the red contours are the results from the cmb+ @xmath9 @xcite+ @xmath92 @xcite+ union sn @xcite . with this data - set combination
there is no longer a degeneracy between coupling and neutrino mass .
as discussed in section [ darkcouplingresults ] , the strong changes in the growth of structure caused by the dark coupling prevent us from applying the maxbcg constraint to this model .
however , the blue contours , which also include the maxbcg constraint , demonstrate that the constraint on @xmath34 is not tightened even if this additional constraint could be included . ]
we first verify that all three primary datasets are consistent with each other in the @xmath2cdm cosmology by comparing the maximum likelihood value of the wmap5 data alone with the maximum likelihood after adding the additional observational constraints .
we find that @xmath96 decreases by 0.5 with the addition of the two independent gaussian priors from the @xmath9 and maxbcg constraints ; the decrease in @xmath96 is the same when we add massive neutrinos to the model .
therefore , our tight constraints are not arising from tension between the datasets we are using . to compute the profile likelihood as described in section [ method ]
, we find the maximum likelihood value over the mcmc chain in bins of @xmath34 .
figure [ fig : profilelike ] shows that this function is well - fit by a gaussian in each case .
we can therefore tentatively interpret the largest @xmath34 value for which @xmath97 as a prior - independent @xmath98 upper bound on @xmath34 .
we report this value in the bottom half of table [ table : mnu ] ; in general we find excellent agreement with the bayesian results . in table [
table : mnuother ] we compile other @xmath34 constraints available in the literature and compare with the results we have found .
wmap5 provides a robust upper limit in a @xmath2cdm model , @xmath31 ev ; allowing @xmath99 relaxes the constraint only slightly to @xmath32 ev . in a study of wmap3 data , ref .
@xcite find that correcting the likelihood approximation on large angular scales as well as accounting for an over - subtraction of point sources affecting small angular scales , the wmap3-only 95% upper bound improved from 1.9 to 1.57 ev .
@xmath34 remains degenerate with both @xmath9 and @xmath35 within the wmap5 constraint .
using bao and sn reduces the degeneracy with @xmath9 , reducing the limit to 0.67 ev ( @xmath38 ) or 0.8 ev when @xmath6 is also allowed to vary .
using both the @xmath100-dependent suppression of power and bao feature in the matter power spectrum derived from the sdss luminous red galaxy sample , ref .
@xcite finds @xmath36 ev when combined with wmap5 for @xmath38 .
that analysis improves the nonlinear modeling by removing the additional shot noise contribution from galaxies occupying the same dark matter halo and marginalizing over the remaining modeling uncertainties .
@xcite compares the impact of two nonlinear modeling approaches on the cosmological inferences for several extensions of standard @xmath2cdm cosmology , and advocates marginalizing over an additional shot noise term .
additional progress requires probing the amplitude of mass fluctuations .
this has been done with several samples probing the cluster mass function .
@xcite derives a low and high redshift cluster mass function from _ chandra _ observations of clusters in two rosat surveys .
their measurement provides a constraint from the normalization of the local mass function @xmath101 ( statistical and systematic errors , respectively ) ; this is similar to and in agreement with the maxbcg constraint we have used in our analysis .
additional cosmological information is available by comparing the high and low redshift mass functions .
their tighter statistical constraint on the local mass function normalization translates to tighter constraints on @xmath102 ev , but with the caveat that @xmath103 ev of systematic error has not been included .
moreover , ref .
@xcite points out that the analysis of @xcite does not fit for both the cluster scaling relations and cosmological parameters self - consistently
. ref .
@xcite uses the cluster mass function from weak lensing to derive a similar cluster normalization constraint : @xmath104 .
because of both the larger errors , lower central value , and combination with wmap3 rather than wmap5 , they find a relatively weak constraint , @xmath105 ev .
because so far only upper limits can be obtained from cosmology on the neutrino masses , the tightness of the @xmath34 constraints depends drastically on where the lower - redshift @xmath35 determination intersects the cmb @xmath106 degeneracy .
an example of this sensitivity is the constraint obtained including the ly-@xmath107 forest power spectrum ( lya ) measurements @xcite . in that work ,
@xmath98 tension between the power spectrum amplitudes preferred by wmap3 and lya when @xmath12 was reported , with lya preferring a larger amplitude .
this translates into an extremely tight constraint on @xmath108 ev when lya is combined with many other observational probes ( wmap3 , sn , sdss bao , sdss and 2df @xmath109 , and small scale cmb experiments ) . with updated datasets for wmap and sn but including fewer datasets altogether , and in addition allowing for running of the spectral index , we find that lya data still reduces the constraint to @xmath110 ev .
moreover , ref .
@xcite finds possible evidence for an inverted temperature - density relation , which would change the interpretation of the lya power spectrum measurement and significantly weaken the ref .
@xcite bound .
table [ table : mnu ] demonstrated the improvement on @xmath34 when tight hubble constant measurements are included @xcite .
@xcite use the stars in passively - evolving red galaxies as cosmic chronometers to obtain a cosmological - model independent measurement of @xmath111 from @xmath112 to @xmath113 ; adding these additional constraints on @xmath111 also reduces the constraint on @xmath34 to 0.5 ev .
weak lensing provides additional constraints on the amplitude of fluctuations on small scales which can be used to constrain @xmath34 ; both ref .
@xcite and ref .
@xcite find a limit of @xmath114 ev .
finally , there have also been constraints derived from luminosity - dependent biasing @xcite .
the most recent work by @xcite combines wmap5 with the @xcite lrg @xmath109 and luminosity - dependent bias measurements to find @xmath110 ev in @xmath2cdm ; this constraint relaxes to @xmath115 ev in a @xmath6cdm cosmology .
while the modeling procedure can be quite complex , this approach is certainly competitive with other probes . in conclusion
, wmap5 provides a robust 95% confidence upper bound of @xmath116 ev . probes of the redshift - distance or @xmath111 relation break key degeneracies and reduce the constraints to @xmath117 ev .
similarly , adding the measurement of the galaxy power spectrum shape for sdss dr7 @xcite reduces the bound to @xmath118 ev . beyond this , probes of the amplitude of clustering @xmath35 through measurements of the cluster mass function , weak lensing , lya , or luminosity - dependent clustering are necessary .
the most optimistic interpretation of the currently available data reduces the bound to @xmath119 ev .
all of these methods require more complicated modeling and need further quantification of the remaining systematics . from current solar and atmospheric constraints on neutrino mass differences @xmath120 ,
three types of neutrino mass spectra are allowed : the normal hierarchy ( @xmath121 ) , the inverted hierarchy ( @xmath122 ) and the quasi - degenerate hierarchy ( @xmath123 ) .
constraints on the absolute neutrino mass scale would greatly help in distinguishing there three cases as indicated in figure [ fig : hierarchy ] .
direct neutrino mass searches based on tritium beta decay so far have imposed an upper limit of @xmath124 ev at the @xmath125 cl @xcite , but to distinguish e.g. the quasi - degenerate hierarchy from the other cases a mass sensitivity of @xmath126 ev is needed .
the neutrino mass hierarchy is one of the unknowns within the neutrino sector and its extraction would be crucial for determining the neutrino character ( dirac or majorana ) and measuring leptonic cp violation .
a determination of the neutrino mass hierarchy would also rule out some neutrino mass models based on lepton flavor symmetries and/or grand unification schemes @xcite . if future cosmological data finds no evidence for a neutrino mass down to @xmath127 ev , that would imply that the neutrino mass hierarchy is normal , allowing , in principle , a clean measurement of cp violation in the leptonic sector ( provided the mixing angle @xmath128 is not very small ) , see ref . @xcite .
if future cosmological data finds a positive signal above @xmath127 ev it would be impossible to measure unambiguously the hierarchy with cosmological information only .
however , if such a signal is found , direct neutrino mass searches , as tritium beta decay experiments , could infer the mass of the lightest neutrino and provide the key to determine the hierarchy and leptonic cp violation .
moreover , if the neutrino mass hierarchy turns out to be inverted , next generation neutrino - less double beta decay experiments could extract the neutrino character ( for a review of the expected sensitivities , see @xcite ) .
a majorana neutrino character , together with a non - vanishing leptonic cp violation would point to leptogenesis as the responsible mechanism for the matter - antimatter asymmetry of the universe .
lllll + model & base dataset & u.l .
( ev ) & sys .
errors & ref .
+ @xmath2cdm & wmap5 & 1.3 & @xcite & @xcite + @xmath6cdm & wmap5 & 1.5 & @xcite & @xcite + @xmath2cdm & wmap5@xmath72bao@xmath72sn & 0.67 & - & @xcite + @xmath6cdm & wmap5@xmath72bao@xmath72sn & 0.80 & - & @xcite + @xmath2cdm & wmap5+@xmath129 & 0.62 & marginalized & @xcite + @xmath6cdm & wmap5@xmath72bao@xmath72sn+ & & + & _ chandra _ clusters & 0.33 & @xmath130 & @xcite + @xmath2cdm & wmap3+wl mass function & 1.43 & - & @xcite + @xmath2cdm & wmap3+ly-@xmath107+cmb small & & + & + bao+sn+sdss & 2df @xmath109 & 0.17 & igm @xcite & @xcite + @xmath2cdm & wmap5+ly-@xmath107+cmb small & & + + @xmath107 & + @xmath9 2001 @xcite & 0.28 & igm @xcite & - + @xmath2cdm & wmap3+@xmath9+@xmath111 & 0.50 & - & @xcite + @xmath2cdm & wmap5+sn+bao+wl & 0.54 & + 0.04 & @xcite , @xcite + @xmath2cdm & wmap5+@xmath131+@xmath132 2006 & 0.28 & bias modeling & @xcite + @xmath6cdm & wmap5+@xmath131+@xmath132 2006 & 0.59 & bias modeling & @xcite
we now consider constraints on the effective number of neutrino species : in this case we assume that neutrino masses are negligible and in the analysis we keep @xmath34 fixed to @xmath88 . while in principle both @xmath11 and @xmath34 should be varied simultaneously , in practice the two
are no longer degenerate , and so the assumption @xmath12 does not change the constraints much @xcite .
equation [ zequal ] defines the major degeneracy between @xmath11 and @xmath27 in the wmap5 data . with the strict assumptions of a @xmath2cdm cosmology
, the measured angular diameter distance can only be recovered if @xmath9 varies with @xmath11 , while the @xmath45 and @xmath133 constraints do not relax when @xmath11 is allowed to vary .
this explains the strong degeneracy between @xmath11 and @xmath9 in the wmap5 posterior distribution .
there is also a weaker degeneracy between @xmath11 and @xmath59 .
while wmap5 tightly constrains the amplitude of fluctuations on very large scales , @xmath134 mpc@xmath40 , within the wmap5 degeneracy and assuming a power law primordial spectrum , the allowed linear @xmath109 pivots about @xmath62 as @xmath11 varies , so that the amplitude of power near the peak at @xmath135 mpc@xmath40 ( and beyond ) increases with @xmath11 .
this tilt causes the degeneracy between @xmath11 and @xmath35 , and can be constrained both by measuring the galaxy power spectrum shape and constraints on the cluster mass function .
finally , since @xmath45 and @xmath133 are both tightly constrained , the variation of @xmath9 with @xmath11 induces a degeneracy between @xmath11 and @xmath136 .
this could be probed through gas fractions in clusters or eventually through the amplitude of the bao signal in large scale structure data .
these well - defined one - dimensional degeneracies would degrade in a less - restricted model of the expansion history , so that further geometrical constraints would be needed to constrain @xmath11 and other parameters like @xmath7 and @xmath6 simultaneously .
figure [ fig : h0vsneff ] shows the degeneracies in the wmap5+@xmath129 posterior distribution between @xmath11 and both @xmath9 and the maxbcg - constrained parameter @xmath70 ; the results are qualitatively the same when the @xmath129 constraints are excluded .
table [ table : neff2 ] illustrates constraints on @xmath11 in @xmath2cdm for three different base datasets : wmap5 alone , wmap5+bao+sn+@xcite hst prior ( @xmath137 km s@xmath40 mpc@xmath40 ) , and wmap5+@xmath129 from @xcite .
both the maxbcg and updated @xmath9 measurement reduce the error on @xmath11 significantly in all three cases . when both of these constraints are taken together , additional data sets other than wmap5
do not shift the mean value of @xmath11 or reduce its error : @xmath138 . therefore the maxbcg and @xmath9 constraints are sufficient to break the key degeneracies in the wmap5 data in this model . since the geometrical constraints from bao and sn
do not contribute significantly to the @xmath11 constraint , they would probably be useful in constraining more general models ( i.e. , models which vary curvature or the behavior of dark energy ) in the context of freely varying @xmath11 .
lllll + base dataset & - & + maxbcg & @xmath74 & + maxbcg@xmath74 + & & & & + wmap5 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + wmap5+bao & & & & + + sn+hst & @xmath143 & @xmath144 & @xmath145 & @xmath146 + wmap5 & & & & + + @xmath129 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + & & & & + wmap5 & @xmath151 & @xmath152 & @xmath153 & @xmath154 + wmap5+bao & & & & + + sn+hst & @xmath155 & @xmath156 & @xmath157 & @xmath158 + wmap5 & & & & + + @xmath129 & @xmath159 & @xmath160 & @xmath161 & @xmath162 + figure [ fig : h0vsneff ] and table [ table : neff2 ] indicate that the likelihood surface is skewed
contours are more widely separated at large @xmath11 .
this finding is in line with the ref .
@xcite result that the posterior of @xmath11 only approaches a gaussian distribution when several datasets are combined . in figure
[ fig : profilelikenrel ] this skewness is evident in the profile likelihood for @xmath163 .
a gaussian provides a reasonable fit to the profile likelihood only when both @xmath9 and maxbcg constraints are included .
therefore , the `` frequentist '' constraints reported in the lower half of table [ table : neff2 ] are only meaningful for the last column , which includes @xmath9 and maxbcg constraints . in that case
, there is good agreement with the bayesian 95% confidence interval .
table [ table : neffother ] shows constraints on @xmath11 from other data in the literature . while wmap5 provides robust evidence for a lower bound on @xmath11 , by itself it can not provide an upper limit .
the strong correlation between @xmath11 and @xmath9 in the @xmath2cdm model makes direct probes of @xmath9 and @xmath111 the most effective way to constrain @xmath11 with low redshift observations .
the new hst constraint clearly provides a significant improvement over wmap5 in combination with previously available geometrical constraints ( wmap5+bao+sn+hst ) .
the additional constraints on the expansion history from @xcite improve even further , with the wmap5+@xmath9 error on @xmath11 of 1.0 being reduced to 0.5 with the combination wmap5+@xmath9+@xmath111 .
the ly-@xmath107 constraints are also quite promising as a probe of the power spectrum amplitude at large @xmath100 , but should be reevaluated with the new data and in the context of the possible systematics .
comparison of measured isotope abundances with bbn calculations provide still the tightest and completely independent constraints on @xmath11 . [
cols="<,<,<,<",options="header " , ]
in this paper we have demonstrated that probes of the cluster mass function and increased precision on the hubble constant can break key degeneracies with cmb observations and yield excellent constraints on both the number of species and sum of the masses of cosmological neutrinos .
when the expansion history is fixed to @xmath2cdm , current constraints on @xmath9 @xcite and the cluster mass function @xcite constrain @xmath3 ev at 95% confidence . this bound relaxes to 0.5 ev in the two extended models we considered : wcdm , and the dark coupling model of @xcite , which allowed curvature , @xmath6 , and coupling strength @xmath8 to vary . probing the mass function using x - ray clusters , ref .
@xcite combines x - ray cluster data with wmap5+bao+sn and finds @xmath102 ev in a @xmath6cdm cosmology , though the systematic errors still must be clearly quantified .
the optimistic upper bound available from current data , @xmath164 ev , almost excludes the scenario in which the neutrino masses are quasi - degenerate .
the constraint on the number of relativistic species @xmath11 from probes of the cluster mass function has not been widely explored so far .
we find that the combination of wmap5 , @xcite @xmath9 , and the maxbcg cluster mass function constraint provides an excellent constraint : @xmath138 .
however , we point out that this constraint does not improve when bao and sn data are also included ; those data sets have been shown to have excellent constraining power on both @xmath7 and @xmath6 @xcite .
therefore , we may expect that the constraint on @xmath11 will not relax dramatically if we allow for these additional parameters but include all of these datasets , particularly because of the complementarity of the growth of structure and expansion history constraints we have employed .
these constraints are not competitive with the most optimistic errors from bbn @xcite , but provide an important consistency check using probes significantly after the bbn epoch .
b.a.r . acknowledges support from fp7 people-2002irg4 - 4-irg and excellent discussions with steen hannestad and risa wechsler .
lv and rj acknowledge support from micinn grant aya2008 - 03531 .
lv acknowledges support of fp7 people-2002irg4 - 4-irg#202182 .
rj is supported by a fp7-people - irg grant .
om is supported by a _
ramn y cajal _ contract from mec , spain .
10 g. la vacca , s. a. bonometto , and l. p. l. colombo , _ higher neutrino mass allowed if cold dark matter and dark energy are coupled _ , _ new astronomy _ * 14 * ( july , 2009 ) 435442 , [ http://arxiv.org/abs/0810.0127[0810.0127 ] ] .
s. hannestad , a. mirizzi , g. g. raffelt , and y. y. y. wong , _ cosmological constraints on neutrino plus axion hot dark matter _ , _ journal of cosmology and astro - particle physics _ * 8 * ( aug . , 2007 ) 15+ , [ http://arxiv.org/abs/0706.4198[0706.4198 ] ] .
r. h. cyburt , _ primordial nucleosynthesis for the new cosmology : determining uncertainties and examining concordance _ , _ phys .
* 70 * ( july , 2004 ) 023505+ , [ http://arxiv.org/abs/arxiv:astro-ph/0401091 [ arxiv : astro - ph/0401091 ] ] .
r. h. cyburt , b. d. fields , k. a. olive , and e. skillman , _ new bbn limits on physics beyond the standard model from ^4he _ , _ astroparticle physics _
* 23 * ( apr . , 2005 ) 313323 , [ http://arxiv.org/abs/arxiv:astro-ph/0408033 [ arxiv : astro - ph/0408033 ] ] .
f. iocco , g. mangano , g. miele , o. pisanti , and p. d. serpico , _ primordial nucleosynthesis : from precision cosmology to fundamental physics _ , _ physics reports _ * 472 * ( mar . ,
2009 ) 176 , [ http://arxiv.org/abs/0809.0631[0809.0631 ] ] .
r. trotta and a. melchiorri , _
indication for primordial anisotropies in the neutrino background from the wilkinson microwave anisotropy probe and the sloan digital sky survey _ , _ physical review letters _ * 95 * ( june , 2005 ) 011305+ , [ http://arxiv.org/abs/arxiv:astro-ph/0412066 [ arxiv : astro - ph/0412066 ] ] .
e. komatsu , j. dunkley , m. r. nolta , c. l. bennett , b. gold , g. hinshaw , n. jarosik , d. larson , m. limon , l. page , d. n. spergel , m. halpern , r. s. hill , a. kogut , s. s. meyer , g. s. tucker , j. l. weiland , e. wollack , and e. l. wright , _ five - year wilkinson microwave anisotropy probe observations : cosmological interpretation _ , _ apj suppl .
_ * 180 * ( feb . , 2009 ) 330376 , [ http://arxiv.org/abs/0803.0547[0803.0547 ] ] .
r. bowen , s. h. hansen , a. melchiorri , j. silk , and r. trotta , _ the impact of an extra background of relativistic particles on the cosmological parameters derived from the cosmic microwave background _
, _ mnras _ * 334 * ( aug . ,
2002 ) 760768 , [ http://arxiv.org/abs/astro-ph/0110636[arxiv:astro-ph/0110636 ] ] .
j. dunkley , e. komatsu , m. r. nolta , d. n. spergel , d. larson , g. hinshaw , l. page , c. l. bennett , b. gold , n. jarosik , j. l. weiland , m. halpern , r. s. hill , a. kogut , m. limon , s. s. meyer , g. s. tucker , e. wollack , and e. l. wright , _ five - year wilkinson microwave anisotropy probe observations : likelihoods and parameters from the wmap data _ , _ apj suppl . _
* 180 * ( feb . , 2009 ) 306329 , [ http://arxiv.org/abs/0803.0586[0803.0586 ] ] .
b. a. reid , w. j. percival , d. j. eisenstein , l. verde , d. n. spergel , r. a. skibba , n. a. bahcall , t. budavari , m. fukugita , j. r. gott , j. e. gunn , z. ivezic , g. r. knapp , r. g. kron , r. h. lupton , t. a. mckay , a. meiksin , r. c. nichol , a. c. pope , d. j. schlegel , d. p. schneider , m. a. strauss , c. stoughton , a. s. szalay , m. tegmark , d. h. weinberg , d. g. york , and i. zehavi , _ cosmological constraints from the clustering of the sloan digital sky survey dr7 luminous red galaxies _ , _ arxiv e - prints _ ( july , 2009 ) [ http://arxiv.org/abs/0907.1659[0907.1659 ] ] .
l. verde , h. v. peiris , d. n. spergel , m. r. nolta , c. l. bennett , m. halpern , g. hinshaw , n. jarosik , a. kogut , m. limon , s. s. meyer , l. page , g. s. tucker , e. wollack , and e. l. wright , _ first - year wilkinson microwave anisotropy probe ( wmap ) observations : parameter estimation methodology _ , _ apj suppl . _ * 148 * ( 2003 ) 195211 .
a. g. riess , l. macri , s. casertano , m. sosey , h. lampeitl , h. c. ferguson , a. v. filippenko , s. w. jha , w. li , r. chornock , and d. sarkar , _ a redetermination of the hubble constant with the hubble space telescope from a differential distance ladder _
, _ apj _ * 699 * ( july , 2009 ) 539563 , [ http://arxiv.org/abs/0905.0695 [ 0905.0695 ] ] .
e. rozo , r. h. wechsler , e. s. rykoff , j. t. annis , m. r. becker , a. e. evrard , j. a. frieman , s. m. hansen , j. hao , d. e. johnston , b. p. koester , t. a. mckay , e. s. sheldon , and d. h. weinberg , _ cosmological constraints from the sdss maxbcg cluster catalog _ , _ arxiv e - prints _ ( feb . , 2009 )
[ http://arxiv.org/abs/0902.3702 [ 0902.3702 ] ] .
b. p. koester , t. a. mckay , j. annis , r. h. wechsler , a. evrard , l. bleem , m. becker , d. johnston , e. sheldon , r. nichol , c. miller , r. scranton , n. bahcall , j. barentine , h. brewington , j. brinkmann , m. harvanek , s. kleinman , j. krzesinski , d. long , a. nitta , d. p. schneider , s. sneddin , w. voges , and d. york , _ a maxbcg catalog of 13,823 galaxy clusters from the sloan digital sky survey _ , _ apj _ * 660 * ( may , 2007 ) 239255 , [ http://arxiv.org/abs/arxiv:astro-ph/0701265 [ arxiv : astro - ph/0701265 ] ] . d. e. johnston , e. s. sheldon , r. h. wechsler , e. rozo , b. p. koester , j. a. frieman , t. a. mckay , a. e. evrard , m. r. becker , and j. annis , _ cross - correlation weak lensing of sdss galaxy clusters ii : cluster density profiles and the mass richness relation _ ,
_ arxiv e - prints _ ( sept . , 2007 ) [ http://arxiv.org/abs/0709.1159 [ 0709.1159 ] ] .
j. brandbyge , s. hannestad , t. haugblle , and b. thomsen , _ the effect of thermal neutrino motion on the non - linear cosmological matter power spectrum _ , _
journal of cosmology and astro - particle physics _
* 8 * ( aug . , 2008 ) 20+ , [ http://arxiv.org/abs/0802.3700[0802.3700 ] ] .
d. n. spergel , l. verde , h. v. peiris , e. komatsu , m. r. nolta , c. l. bennett , m. halpern , g. hinshaw , n. jarosik , a. kogut , m. limon , s. s. meyer , l. page , g. s. tucker , j. l. weiland , e. wollack , and e. l. wright , _ first - year wilkinson microwave anisotropy probe ( wmap ) observations : determination of cosmological parameters _ , _ apj suppl . _
* 148 * ( sept . , 2003 ) 175194 .
d. n. spergel , r. bean , o. dor , m. r. nolta , c. l. bennett , j. dunkley , g. hinshaw , n. jarosik , e. komatsu , l. page , h. v. peiris , l. verde , m. halpern , r. s. hill , a. kogut , m. limon , s. s. meyer , n. odegard , g. s. tucker , j. l. weiland , e. wollack , and e. l. wright , _ three - year wilkinson microwave anisotropy probe ( wmap ) observations : implications for cosmology _ , _ apj suppl . _
* 170 * ( june , 2007 ) 377408 , [ http://arxiv.org/abs/arxiv:astro-ph/0603449 [ arxiv
: astro - ph/0603449 ] ] .
j. hamann , s. hannestad , g. g. raffelt , and y. y. y. wong , _
observational bounds on the cosmic radiation density _ _ journal of cosmology and astro - particle physics _ *
8 * ( aug . , 2007 ) 21+ , [ http://arxiv.org/abs/0707.0440 [ arxiv : astro - ph/0707.0440 ] ] . w. j. percival , r. c. nichol , d. j. eisenstein , j. a. frieman , m. fukugita , j. loveday , a. c. pope , d. p. schneider , a. s. szalay , m. tegmark , m. s. vogeley , d. h. weinberg , i. zehavi , n. a. bahcall , j. brinkmann , a. j. connolly , and a. meiksin , _ the shape of the sloan digital sky survey data release 5 galaxy power spectrum _ , _ apj _ * 657 * ( mar .
, 2007 ) 645663 , [ http://arxiv.org/abs/arxiv:astro-ph/0608636 [ arxiv : astro - ph/0608636 ] ] .
m. kowalski , d. rubin , g. aldering , r. j. agostinho , a. amadon , r. amanullah , c. balland , k. barbary , g. blanc , p. j. challis , a. conley , n. v. connolly , r. covarrubias , k. s. dawson , s. e. deustua , r. ellis , s. fabbro , v. fadeyev , x. fan , b. farris , g. folatelli , b. l. frye , g. garavini , e. l. gates , l. germany , g. goldhaber , b. goldman , a. goobar , d. e. groom , j. haissinski , d. hardin , i. hook , s. kent , a. g. kim , r. a. knop , c. lidman , e. v. linder , j. mendez , j. meyers , g. j. miller , m. moniez , a. m. mouro , h. newberg , s. nobili , p. e. nugent , r. pain , o. perdereau , s. perlmutter , m. m. phillips , v. prasad , r. quimby , n. regnault , j. rich , e. p. rubenstein , p. ruiz - lapuente , f. d. santos , b. e. schaefer , r. a. schommer , r. c. smith , a. m. soderberg , a. l. spadafora , l .- g .
strolger , m. strovink , n. b. suntzeff , n. suzuki , r. c. thomas , n. a. walton , l. wang , w. m. wood - vasey , and j. l. yun , _ improved cosmological constraints from new , old , and combined supernova data sets _ , _ apj _ * 686 * ( oct . ,
2008 ) 749778 , [ http://arxiv.org/abs/0804.4142[0804.4142 ] ] .
j. r. kristiansen , g. la vacca , l. p. l. colombo , r. mainini , and s. a. bonometto , _ coupling between cold dark matter and dark energy from neutrino mass experiments _ , _ arxiv e - prints _ ( feb . ,
2009 ) [ http://arxiv.org/abs/0902.2737[0902.2737 ] ] .
g. la vacca , j. r. kristiansen , l. p. l. colombo , r. mainini , and s. a. bonometto , _ do wmap data favor neutrino mass and a coupling between cold dark matter and dark energy ? _ , _ journal of cosmology and astro - particle physics _ * 4 * ( apr . , 2009 ) 7+ , [ http://arxiv.org/abs/0902.2711[0902.2711 ] ] .
g. la vacca and j. r. kristiansen , _ dynamical dark energy model parameters with or without massive neutrinos _
, _ journal of cosmology and astro - particle physics _ * 7 * ( july , 2009 ) 36+ , [ http://arxiv.org/abs/0906.4501[0906.4501 ] ] . j. r. kristiansen , h. k. eriksen , and .
elgary , _ revised wmap constraints on neutrino masses and other extensions of the minimal @xmath2cdm model _ , _ phys .
* 74 * ( dec . ,
2006 ) 123005+ , [ http://arxiv.org/abs/arxiv:astro-ph/0608017 [ arxiv : astro - ph/0608017 ] ] .
j. hamann , s. hannestad , a. melchiorri , and y. y. y. wong , _ non - linear corrections to the cosmological matter power spectrum and scale - dependent galaxy bias : implications for parameter estimation _ , _
journal of cosmology and astro - particle physics _
* 7 * ( july , 2008 ) 17+ , [ http://arxiv.org/abs/arxiv:0804.1789[arxiv:0804.1789 ] ] .
a. vikhlinin , a. v. kravtsov , r. a. burenin , h. ebeling , w. r. forman , a. hornstrup , c. jones , s. s. murray , d. nagai , h. quintana , and a. voevodkin , _ chandra cluster cosmology project iii : cosmological parameter constraints _ , _ apj _ * 692 * ( feb . , 2009 ) 10601074 , [ http://arxiv.org/abs/0812.2720[0812.2720 ] ] .
a. mantz , s. w. allen , d. rapetti , and h. ebeling , _ the observed growth of massive galaxy clusters i : statistical methods and cosmological constraints _ , _ arxiv e - prints _ ( sept . , 2009 )
[ http://arxiv.org/abs/0909.3098[0909.3098 ] ] .
j. r. kristiansen , .
elgary , and h. dahle , _ using the cluster mass function from weak lensing to constrain neutrino masses _ , _ phys .
* 75 * ( apr . , 2007 ) 083510+ , [ http://arxiv.org/abs/arxiv:astro-ph/0611761 [ arxiv : astro - ph/0611761 ] ] .
u. seljak , a. slosar , and p. mcdonald , _ cosmological parameters from combining the lyman-@xmath107 forest with cmb , galaxy clustering and sn constraints _ , _ journal of cosmology and astro - particle physics _ * 10 * ( oct . ,
2006 ) 14+ , [ http://arxiv.org/abs/arxiv:astro-ph/0604335 [ arxiv : astro - ph/0604335 ] ] .
j. s. bolton , m. viel , t .- s .
kim , m. g. haehnelt , and r. f. carswell , _
possible evidence for an inverted temperature - density relation in the intergalactic medium from the flux distribution of the ly@xmath107 forest _ , _ mnras _ * 386 * ( may , 2008 ) 11311144 , [ http://arxiv.org/abs/0711.2064[0711.2064 ] ] .
d. stern , r. jimenez , l. verde , m. kamionkowski , and s. a. stanford , _ cosmic chronometers : constraining the equation of state of dark energy .
i : h(z ) measurements _ , _ arxiv e - prints _ ( july , 2009 ) [ http://arxiv.org/abs/0907.3149[0907.3149 ] ] .
j. simon , l. verde , and r. jimenez , _
constraints on the redshift dependence of the dark energy potential _ , _ phys .
* 71 * ( june , 2005 ) 123001+ , [ http://arxiv.org/abs/arxiv:astro-ph/0412269 [ arxiv : astro - ph/0412269 ] ] .
i. tereno , c. schimd , j .-
uzan , m. kilbinger , f. h. vincent , and l. fu , _ cfhtls weak - lensing constraints on the neutrino masses _ , _ arxiv e - prints _ ( oct . , 2008 )
[ http://arxiv.org/abs/0810.0555 [ 0810.0555 ] ] .
u. seljak , a. makarov , r. mandelbaum , c. m. hirata , n. padmanabhan , p. mcdonald , m. r. blanton , m. tegmark , n. a. bahcall , and j. brinkmann , _ sdss galaxy bias from halo mass - bias relation and its cosmological implications _ , _ phys .
* 71 * ( feb . , 2005 ) 043511+ , [ http://arxiv.org/abs/arxiv:astro-ph/0406594 [ arxiv : astro - ph/0406594 ] ] . f. de bernardis , p. serra , a. cooray , and a. melchiorri , _ improved limit on the neutrino mass with cmb and redshift - dependent halo bias - mass relations from sdss , deep2 , and lyman - break galaxies _ , _ phys .
* 78 * ( oct . , 2008 ) 083535+ , [ http://arxiv.org/abs/0809.1095 [ 0809.1095 ] ] .
m. tegmark , d. j. eisenstein , m. a. strauss , d. h. weinberg , m. r. blanton , j. a. frieman , m. fukugita , j. e. gunn , a. j. s. hamilton , g. r. knapp , r. c. nichol , j. p. ostriker , n. padmanabhan , w. j. percival , d. j. schlegel , d. p. schneider , r. scoccimarro , u. seljak , h .- j .
seo , m. swanson , a. s. szalay , m. s. vogeley , j. yoo , i. zehavi , k. abazajian , s. f. anderson , j. annis , n. a. bahcall , b. bassett , a. berlind , j. brinkmann , t. budavari , f. castander , a. connolly , i. csabai , m. doi , d. p. finkbeiner , b. gillespie , k. glazebrook , g. s. hennessy , d. w. hogg , v z. ivezi , b. jain , d. johnston , s. kent , d. q. lamb , b. c. lee , h. lin , j. loveday , r. h. lupton , j. a. munn , k. pan , c. park , j. peoples , j. r. pier , a. pope , m. richmond , c. rockosi , r. scranton , r. k. sheth , a. stebbins , c. stoughton , i. szapudi , d. l. tucker , d. e. v. berk , b. yanny , and d. g. york , _ cosmological constraints from the sdss luminous red galaxies _ , _ phys .
* 74 * ( dec . ,
2006 ) 123507+ , [ http://arxiv.org/abs/arxiv:astro-ph/0608632 [ arxiv : astro - ph/0608632 ] ] . c. e. kraus , _ final results from phase ii of the mainz neutrino mass search in tritium @xmath65 decay _ , _ eur .
phys . j. c _ * 40 * ( 2005 ) 447 , [ http://arxiv.org/abs/arxiv:hep-ex/0412056 [ arxiv : hep - ex/0412056 ] ]
. w. l. freedman , b. f. madore , b. k. gibson , l. ferrarese , d. d. kelson , s. sakai , j. r. mould , r. c. kennicutt , h. c. ford , j. a. graham , j. p. huchra , s. m. g. hughes , g. d. illingworth , l. m. macri , and p. b. stetson , _ final results from the hubble space telescope key project to measure the hubble constant _ , _ apj _ * 553 * ( may , 2001 ) 4772 .
s. hannestad and g. g. raffelt , _ neutrino masses and cosmic radiation density : combined analysis _ , _ journal of cosmology and astro - particle physics _ * 11 * ( nov . , 2006 ) 16+ , [ http://arxiv.org/abs/arxiv:astro-ph/0607101 [ arxiv : astro - ph/0607101 ] ] .
w. j. percival , b. a. reid , d. j. eisenstein , n. a. bahcall , t. budavari , m. fukugita , j. e. gunn , z. ivezic , g. r. knapp , r. g. kron , j. loveday , r. h. lupton , t. a. mckay , a. meiksin , r. c. nichol , a. c. pope , d. j. schlegel , d. p. schneider , d. n. spergel , c. stoughton , m. a. strauss , a. s. szalay , m. tegmark , d. h. weinberg , d. g. york , and i. zehavi , _ baryon acoustic oscillations in the sloan digital sky survey data release 7 galaxy sample _ , _ arxiv e - prints _ ( july , 2009 ) [ http://arxiv.org/abs/0907.1660 [ 0907.1660 ] ] . | we illustrate how recently improved low - redshift cosmological measurements can tighten constraints on neutrino properties .
in particular we examine the impact of the assumed cosmological model on the constraints .
we first consider the new hst @xmath0 measurement by riess et al .
( 2009 ) and the @xmath1 constraint from rozo et al .
( 2009 ) derived from the sdss maxbcg cluster catalog . in a @xmath2cdm model and when combined with wmap5 constraints , these low - redshift measurements constrain @xmath3 ev at the 95% confidence level .
this bound does not relax when allowing for the running of the spectral index or for primordial tensor perturbations . when adding also supernovae and bao constraints , we obtain a 95% upper limit of @xmath4ev .
we test the sensitivity of the neutrino mass constraint to the assumed expansion history by both allowing a dark energy equation of state parameter @xmath5 and by studying a model with coupling between dark energy and dark matter , which allows for variation in @xmath6 , @xmath7 , and dark coupling strength @xmath8 . when combining cmb , @xmath9 and the sdss lrg halo power spectrum from reid et al .
2009 , we find that in this very general model , @xmath10 ev with 95% confidence . if we allow the number of relativistic species @xmath11 to vary in a @xmath2cdm model with @xmath12 , we find @xmath13 for the 68% and 95% confidence intervals
. we also report prior - independent constraints , which are in excellent agreement with the bayesian constraints .
@xmath14 institute for sciences of the cosmos ( icc ) , university of barcelona , barcelona 08028 , spain + @xmath15 icrea ( institucio catalana de recerca i estudis avancats ) + @xmath16theory group , physics department , cern , c-1211 , geneva 23 , switzerland + @xmath17 instituto de fisica corpuscular , ific , csic and universidad de valencia , spain |
it is well known that solution of the electron - electron scattering problem , both classical and quantum , is generally reduced to finding the dependence of differential cross - section on the particles deflection angle . at that , the case of head - on collision
is principally ignored despite the fact that it is in these scatterings quantum properties of electrons reveal themselves maximally .
this is related first of all to their extreme mutual closing in , and therefore the manifestations of anti - symmetry of two - particle wave function become maximal . moreover , the influence of coulomb repulsion energy on two - particle system s behavior must be most significant . in this paper
we have used the coherent states as one - particle states and have examined the head - on collision of coherent electrons in triplet state in center - of - mass system .
let us assume that at an instant @xmath0 the first electron is in coherent state @xmath1 and the second one is in coherent state @xmath2 .
states @xmath3 and @xmath4 are eigenstates for operator @xmath5 with eigenvalues @xmath6 and @xmath7 correspondingly : @xmath8 and may be written as @xcite : @xmath9 where @xmath10 and @xmath11 are mean momentum and mean position of electrons at the instant @xmath0 , @xmath12 is uncertainty of every component of particle s momentum in coherent states @xmath13 : @xmath14 spatial part of two - particle state must be anti - symmetrical for triplet : @xmath15 where @xmath16 is overlapping integral , @xmath17 . at a given instant @xmath18 :
@xmath19 this state satisfies schroedinger equation : @xmath20 average value of kinetic energy in the state @xmath21 comprises three terms @xcite : @xmath22 the first term corresponds to free motion of classical particles , the second ( energy of configuration ) appears to be the contribution of momentum uncertainty ; the third one ( energy of correlation ) appears due to overlapping of two non - orthogonal one - particle states .
average value of coulomb repulsion energy in the triplet state @xmath23 @xmath24 is equal @xcite to : @xmath25 where @xmath26
let us assume , that at the instant @xmath0 the first electron is in coherent state @xmath1 and has mean momentum @xmath27 , and mean position @xmath28 ; the second electron is in coherent state @xmath2 with mean momentum and mean position @xmath29 , @xmath30 , respectively .
as far as we examine the case of collision , the initial momentum is negative for the first electron and positive for the second . as the solution of schroedinger equation in zero - order approximation we have used the state ( [ state ] ) in which @xmath31 and @xmath32 depend upon time according to hamilton equations : @xmath33 where @xmath34 in @xcite it is shown that @xmath35 can be considered as perturbation in wide enough range of @xmath31 and @xmath32 . since analytical solution of the system of equations ( [ h ] ) is not found we use its numerical solution . in fig.[f1 ]
the two typical phase trajectories are shown for head - on collision of two coherent electrons .
the distance between electrons and their momenta are in atomic units , initial distance is 5 a.u .
one can see that according to the ratio of initial relative momentum to momentum uncertainty of each electron , two principally different kinds of behavior of phase trajectories are possible . as long as the initial momentum of electrons is less than some critical momentum @xmath36 the effective deflection occurs .
when initial momentum of electrons exceeds @xmath36 the effective penetration is the case .
let us note that these two cases are experimentally indistinguishable due to identity of electrons . concerning the identity of particles the electron s _ return time _
@xmath18 may be used as an observable feature of collision process .
this term implies time span necessary to revert the distance between electrons to initial value .
for calculation of @xmath18 the plots of dependence of electron s coordinate upon time are used .
the example of such dependence is shown in fig.[f2 ] at the initial distance between electrons 5 a.u .
it should be noticed that a significant delay near the origin is possible both in the case of deflection and penetration .
the dependence of _ return time _
@xmath18 on initial momentum at the initial distance between particles 5 a.u .
is shown in fig.[f3 ] with boxes .
for comparison , analogous dependences for non - interacting particles ( dashed line ) and classical electrons ( solid line ) are given , and they almost coincide for large relative momenta .
fig.[f3 ] shows that dependence of _ return time _
@xmath18 on the initial momentum of electrons @xmath37 coincides with that for classical electrons when @xmath38 only . in the range @xmath39 the coulomb repulsion
may be neglected , though as fig.[f1 ] shows , this range corresponds to effective deflection . in the area @xmath40 a delay occurs . when initial momentum @xmath37 is large , @xmath41 , interaction may be ignored and phase trajectories accord with effective penetration .
the analysis of head - on collision of two coherent electrons has shown that the coulomb repulsion is considerable in the area of small values of relative momenta only . for
almost all values of initial relative momenta the system s behavior practically does not differ from classical .
essentially quantum scattering can be observed only in the area @xmath42 , when the distance between particles at the instant of stop is less than position uncertainty of electrons . | we have examined the head - on collision of two electrons in approximation of coherent states .
we have shown that the character of collision depends mainly on ratio of initial relative electron s momentum to momentum uncertainty of electrons .
when this ratio becomes greater then 1 , the coulomb interaction does not practically influence the scattering . |
recent neutrino experiments confirm that the pontecorvo - maki - nakagawa - sakata ( pmns ) @xcite lepton mixing matrix @xmath9 contains large mixing angles .
for example the atmospheric mixing @xmath10 is compatible with @xmath11 @xcite , and the solar mixing @xmath12 is @xmath13 @xcite . these results should be compared with the third lepton mixing angle @xmath0 which is very small and even compatible with zero @xcite , and with the quark mixing angles in the @xmath14 matrix @xcite .
the disparity that nature indicates between quark and lepton mixing angles has been viewed in terms of a quark - lepton complementarity ( qlc ) @xcite which can be expressed in the relations _ 12^pmns+_12^ckm45^ ; _
23^pmns+_23^ckm45^. possible consequences of qlc have been investigated in the literature @xcite and in particular a simple correspondence between the @xmath9 and @xmath14 matrices has been proposed @xcite and analyzed in terms of a correlation matrix @xcite . the correlation matrix @xmath1 is simply defined as the product of the ckm and pmns matrices , @xmath15 , and efforts have been done to obtain the _ most favorite _ pattern for the matrix @xmath1 @xcite .
unitarity then implies @xmath16 and one may ask where do the large lepton mixings come from ?
is this information implicit in the form of the @xmath1 matrix ?
this question has been widely investigated in the literature , but its answer is still open ( see our section * 2 * ) .
furthermore in some grand unification theories ( guts ) the direct qlc correlation between the @xmath17 and the @xmath18 mixing matrix can be obtained . in this class of models ,
the @xmath1 matrix is determined by the heavy majorana neutrino mass matrix @xcite .
moreover as long as quarks and leptons are inserted in the same representation of the underlying gauge group , we need to include in our definition of @xmath1 arbitrary but non trivial phases between the quark and lepton matrices .
hence we will generalize the relation @xmath15 to @xmath19 where the quantity @xmath20 is a diagonal matrix @xmath21 and the three phases @xmath22 are free parameters ( in the sense that they are not restricted by present experimental evidence ) .
the magnitude disparity between the lepton mixing angle @xmath0 and the other two mixings is a rather striking fact . in this paper
we carry out the investigation of the correlation matrix @xmath1 based on eq .
( [ eq : fund ] ) and prove that it is a zero texture of @xmath1 , namely @xmath23 , that implies a small value for @xmath0 with a sharp prediction @xmath24 we use the wolfenstein parameterization for @xmath14 @xcite in its unitary form @xcite and parameterize @xmath9 with the standard phases and mixing angles . as a zero order approximation
we start inserting by hand the central values of the lepton mixing angles and ckm parameters . owing to the uncertainty in the experimental value for @xmath0 , the possible range for the ( 1,3 ) entry of matrix @xmath1 may or may not include zero
. for example using @xmath25 the ( 1,3 ) entry range does not include zero in accordance with eq .
( 23 ) in ref .
@xcite . for other choices of @xmath0
a vanishing ( 1,3 ) entry is quite possible , as will be seen in section * [ sec : vm]*. it is possible to include bimaximal and tribimaximal forms of the correlation matrix @xmath1 in models with renormalization effects @xcite that are relevant , however , only in particular cases with large @xmath26 @xmath27 and with quasi degenerate neutrino masses @xcite .
the conclusion for matrix @xmath1 is that the possibility of a bimaximal form , or a tribimaximal one is completely open . so in other words , the correlation between the matrices @xmath14 and @xmath9 is rather nontrivial .
the investigation we perform for the @xmath1 matrix starts from the fundamental equation @xmath28 and uses the experimental ranges and constraints on lepton mixing angles .
we resort to a monte carlo simulation with two - sided gaussian distributions around the mean values of the observables .
the input information on @xmath0 is taken from the analysis of ref .
@xcite which uses neutrino data only .
the paper is organized as follows : in section * [ sec : vm ] * we study the numerical ranges of @xmath1 entries with the aid of a monte carlo simulation , emphasizing on specific points of the experimental data .
we will show that the vanishing of the @xmath29 entry is favored by the data analysis . in section * [ sec : pmns ] * we present the matter from a different point of view : we start from a zero @xmath29 @xmath1 entry ( e.g. a bimaximal or tribimaximal matrix ) we derive the consequent prediction for the @xmath9 lepton mixing matrix through u_pmns=(u_ckm)^-1v_m and the corresponding one for @xmath0 in eq . ( [ eq : prediction ] )
. finally we present a summary and our conclusions .
in this section we investigate the order of magnitude of the @xmath1 matrix entries concentrating in particular in the ( 1,3 ) entry and the important mixing angle @xmath30 to which it is directly related .
we then explicitly study the allowed values of the @xmath1 angles , finally concluding that @xmath31 is the value most favored by the data .
we will be using the wolfenstein parameterization @xcite of the @xmath14 matrix in its unitary form @xcite where one has the relation [ ckm ] _
12^ckm=_23^ckm = a^2 _ 13^ckme^-^ckm = a ^3(- ) to all orders in @xmath32 .
the lepton mixing matrix @xmath9 is parameterized in the usual way as u_pmns= u_23u_13^u_12_m . here
@xmath33 and @xmath34 are diagonal matrices containing the dirac and majorana cp violating phases , respectively @xmath35 and @xmath36 , so that [ pmns ] u_pmns= in grand unification models where quarks and leptons belong to the same representation of the gauge group , the quark and lepton fields must acquire different phases once their symmetry is broken . hence one should take into account this phase mismatch at low energy associated with the form of the ckm and pmns matrices ( [ ckm ] , [ pmns ] ) . to this end
we introduced the diagonal matrix @xmath20 [ omega ] = diag(e^_1,e^_2,e^_3 ) in the commonly used relation @xmath37 .
this is therefore generalized to [ eq : thvm ] v_m&=&u_ckmu_pmns .
we use for the observed ckm mixing parameters the values @xmath38 , @xmath39 , @xmath40 , @xmath41 , and for the pmns mixing angles the values @xmath42 , @xmath43 , @xmath25 @xcite . for the @xmath20 phases we resort a monte carlo simulation with flat distributions in the interval @xmath44 $ ] .
we then get the following range of values for the elements of the @xmath1 correlation matrix : [ eq : xing ] v_m= ( ) .
these values are in good agreement with @xcite .
the small differences are due to the fact that we use the full mixing matrix given in eq .
( [ ckm ] ) and not the parameterization given in eq .
( 21 ) of ref .
notice that the @xmath29 entry of the matrix @xmath1 above can not be zero , so @xmath1 can not be bimaximal , i.e. of the form , nor tribimaximal , namely , where only the absolute values have been considered .
the result of eq .
( [ eq : xing ] ) however depends on the assumption about the values used for the mixing angles . for example if we use a different value for @xmath0 , namely @xmath45 ( see ref .
@xcite or our eq .
( [ bestfit2 ] ) for the allowed range of @xmath0 ) , we get v_m= ( ) . for these values
the result is in agreement with the statement that @xmath1 has the @xmath29 entry equal to zero .
it is clear that we need a better investigation of the situation before establishing what are the allowed values of the entries of the correlation matrix @xmath1 that can be deduced from the experimental data .
we next investigate the important entry @xmath29 as it overwhelmingly affects the @xmath0 prediction as will be seen in section * [ sec : pmns]*. at first non trivial orders in @xmath32 we have [ th13vm_a ] ^2_13^v_m&= & |(1- ) e^(_1 -_2- ) _ 13^pmns+_23^pmns _ 13^pmns + o(^3)|^2 it is seen from this expression that the first two terms can cancel each other implying a vanishing @xmath29 entry of the @xmath1 matrix . in order to better investigate this issue we plot in fig.[fig : f1 ] the quantity @xmath48 as a function of @xmath49 .
all other observables are fixed at their best fit points @xcite and we allowed the dirac lepton phase @xmath50 , the majorana ones @xmath51 and @xmath52 , and the unphysical phases of @xmath20 to vary in the interval @xmath53 $ ] with a flat distribution .
as shown in the figure , for the central value of @xmath0 given in @xcite the entry ( 1,3 ) of @xmath1 can not be zero .
however there is a small region @xmath54 for which @xmath55 can be zero .
this fact has the very important consequence of providing a sharp prediction for the unknown mixing angle @xmath0 .
we will investigate this point in detail in the section * [ sec : pmns]*. [ [ sec : thetavm ] ] the allowed values for @xmath56 , @xmath57 , and @xmath48 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ here we further investigate the possibility of @xmath1 to be bimaximal or tribimaximal using the fundamental equation ( [ eq : thvm ] ) .
we start with a monte carlo simulation for the @xmath58 parameters , the @xmath59 mixing angles , the @xmath20 and cp phases . | the complementarity between the quark and lepton mixing matrices is shown to provide a robust prediction for the neutrino mixing angle @xmath0 .
we obtain this prediction by first showing that the matrix @xmath1 , product of the ckm and pmns mixing matrices , may have a zero ( 1,3 ) entry which is favored by experimental data .
hence models with bimaximal or tribimaximal forms of the correlation matrix @xmath1 are quite possible . any theoretical model with a vanishing ( 1,3 ) entry of @xmath1 that is in agreement with quark data , solar , and atmospheric mixing angle leads to @xmath2 .
this value is consistent with the present 90% cl experimental upper limit .
pacs : 14.60.pq , 14.60.lm , 96.40.tv .
bhag c. chauhan@xmath3 , marco picariello@xmath4 , joo pulido@xmath3 , emilio torrente - lujan@xmath5 + _ @xmath6 centro de fsica terica das partculas ( cftp ) departamento de fsica , + instituto superior tcnico av .
rovisco pais , p-1049 - 001 lisboa , portugal + @xmath7 dipartimento di fisica , universit di lecce and infn - lecce , + via arnesano , ex collegio fiorini , i-73100 lecce , italia + @xmath8 dep .
de fisica , grupo de fisica teorica , univ
. de murcia , murcia , spain . _ |
an increasingly relevant aspect of proton - proton collisions at high energies is double parton scattering ( dps ) , where two partons from each proton interact in two separate hard subprocesses .
dps contributes to many final states of interest at the lhc .
they constitute relevant backgrounds to precise higgs boson coupling measurements and searchers for physics beyond the standard model .
our knowledge of dps is fragmentary , and improvements are needed at both conceptual and quantitative level ( see for example @xcite ) .
schematically the dps cross section can be expressed as @xmath0 where @xmath1 represents hard subprocess @xmath2 , @xmath3 is a combinatorial factor equal to two ( one ) if the partonic subprocesses are ( not ) identical and @xmath4 ( @xmath5 ) labels the double parton distribution of the proton with momentum @xmath6 ( @xmath7 ) .
the dpds depend on the longitudinal momentum fractions of the two partons @xmath8 ( @xmath9 ) and the distance between them @xmath10 .
implicit in this expression are the labels for the different flavors , colors , fermion numbers and spins of the four partons .
this quantum - number structure is significantly more complicated in dps compared to the case with only one hard interaction , because of the possibility of interference between the two hard interactions and correlations between the two partons inside each proton .
the correlations can be of kinematical type ( between @xmath8 s and @xmath11 ) , or between the quantum numbers of the two partons . while the kinematical type affects dependence of the dpds on the kinematical variables , the quantum - number correlations lead to a large number of different dpds .
the dpds depend on long distance , non - perturbative physics and can thus not be calculated in perturbative qcd .
including all the correlations and their dpds in phenomenological calculations is cumbersome , and extracting all of them experimentally is unfeasible .
an effective way of reducing the number of dpds of experimental relevance is studying the scale evolution of the dpds and the correlations they describe .
for example , it has been demonstrated for quark and antiquark dpds that color interference terms is suppressed by sudakov factors at large scales @xcite , which when combined with positivity bounds constraining the size of the correlations at low scales @xcite set limits on the scale at which these correlations can be of experimental relevance .
the scale evolution of the dpds are described by generalizations of the usual dglap evolution equations .
two versions of this have been discussed in the literature : a homogeneous equation describing the separate evolution of each of the two partons and an inhomogeneous including also the splitting of one parent parton into the two partons that undergo hard scattering @xcite .
which version is adequate for the description of dps processes remains controversial in the literature @xcite .
these proceedings reviews our study of the effects scale evolution has on dps correlations @xcite . for the numerical results presented in these proceedings we have used the homogeneous evolution equation . to solve the evolution equations numerically , we use a modified version of the code originally described in @xcite .
a number of arguments suggest an interplay between the dependence of dpds on the longitudinal momentum fractions @xmath14 @xmath15 of the partons , as well as between their momentum fractions and their relative transverse distance @xmath10 @xcite . in this section
we study the impact of evolution on the correlations between the momentum fractions and the interparton distance , for this purpose we need a model for the dpds at the starting scale of evolution .
we take a simple ansatz motivated by studies of gpds as explained in @xcite @xmath16\ , , \end{aligned}\ ] ] at the starting scale @xmath17 , with @xmath18 the parameters are set to the values @xmath19 the dpds evolve independently at each value of @xmath13 , but the interplay between @xmath13 and the momentum fractions @xmath20 and @xmath15 in the starting conditions has consequences for the scale evolution at different values of @xmath13 .
@xmath21 dependence of the dpd for two @xmath22 .
the left panel shows the natural logarithm of the dpd and the right panels the corresponding slope in @xmath21 .
longitudinal momentum fractions are fixed at @xmath23.,title="fig:",scaledwidth=46.5% ] @xmath21 dependence of the dpd for two @xmath22 .
the left panel shows the natural logarithm of the dpd and the right panels the corresponding slope in @xmath21 .
longitudinal momentum fractions are fixed at @xmath23.,title="fig:",scaledwidth=51.5% ] the gaussian dependence of our starting condition is approximately preserved by evolution up to large scales , as demonstrated by figure [ fig : ydep ] .
this allows us to take a closer look at the evolution of the width of the @xmath13 dependence .
figure [ fig : fit - result - params](a ) shows the evolution of @xmath24 ( effective gaussian width ) at @xmath25 for @xmath26 , @xmath22 and @xmath27 .
the effective gaussian width decreases under evolution for both @xmath28 and @xmath22 , whereas it changes for the gluon . as the valence combination @xmath28 evolves to higher scales ,
partons move from higher to lower @xmath29 values by radiating gluons . for partons at given @xmath29 and @xmath30
, the width of the @xmath13 distribution is therefore influenced by the smaller values of this width for partons with higher @xmath29 at lower @xmath30 - leading to the decrease of @xmath31 with @xmath30 in figure [ fig : fit - result - params](a ) .
the double @xmath22 distribution mixes with gluons and @xmath32 approaches @xmath33 with increasing scale , although it does so rather slowly .
the difference between the transverse distribution of gluons and quarks , which we have assumed at @xmath34 , remains up to large scales .
( a ) : evolution of the effective gaussian width @xmath24 and evaluated at @xmath25 for @xmath35 , @xmath28 and @xmath27 .
( b ) : evolution of the effective shrinkage parameter @xmath36 obtained by fitting @xmath24 in the range @xmath37 for @xmath38 and @xmath27.,title="fig:",scaledwidth=49.0% ] ( a ) : evolution of the effective gaussian width @xmath24 and evaluated at @xmath25 for @xmath35 , @xmath28 and @xmath27 .
( b ) : evolution of the effective shrinkage parameter @xmath36 obtained by fitting @xmath24 in the range @xmath37 for @xmath38 and @xmath27.,title="fig:",scaledwidth=49.3% ] the dependence of @xmath24 on @xmath29 is shown in figure [ fig : fit - test](a ) , ( b ) and ( c ) for the different parton types .
we see that the evolution is faster at smaller momentum fractions @xmath29 , and at low @xmath29 there is a rapid decrease of @xmath24 with @xmath39 for all parton types . for @xmath22
this results in a region of intermediate @xmath29 where @xmath40 increases with @xmath29 at high @xmath39 . for @xmath28 and @xmath27 the curves for @xmath24 are approximately linear in @xmath41 as long as we stay away from the large-@xmath29 region .
this allows us to extract an effective shrinkage parameter @xmath36 by fitting the effective gaussian width in an appropriate region of @xmath29 .
the scale dependence of @xmath36 is shown in figure [ fig : fit - result - params](b ) .
we find that @xmath36 decreases quite rapidly for @xmath42 and more gently for @xmath26 .
: dependence of @xmath24 on @xmath29 .
the solid sections of the curves in panels ( a ) and ( b ) represent fits in figure [ fig : fit - result - params ] in the range @xmath43 .
, title="fig:",scaledwidth=28.0% ] : dependence of @xmath24 on @xmath29 .
the solid sections of the curves in panels ( a ) and ( b ) represent fits in figure [ fig : fit - result - params ] in the range @xmath43 .
, title="fig:",scaledwidth=28.0% ] : dependence of @xmath24 on @xmath29 .
the solid sections of the curves in panels ( a ) and ( b ) represent fits in figure [ fig : fit - result - params ] in the range @xmath43 .
, title="fig:",scaledwidth=28.0% ] in summary , we find that a gaussian @xmath13 dependence at the initial scale is approximately preserved under evolution , with a noticeable but relatively slow change of the effective gaussian width . despite the mixing between gluons and quarks in the singlet sector ,
the differences between their distributions remains up to high scales .
we next investigate the evolution of spin correlations between two partons inside a proton . for this purpose we assume a multiplicative @xmath13 dependence of the dpds , @xmath44 as our focus is on the degree of parton polarization rather than on the absolute size of the dpds , we set @xmath45 . for the unpolarized dpds
we take a simple factorizing ansatz at the starting scale ( @xmath46 ) , @xmath47 the single parton densities used will be either of the two lo sets mstw 2008 @xcite and gjr 08 @xcite . to model the polarized dpds is more difficult .
there is no reason to believe that a decomposition of a polarized dpd , which describes the spin correlations between two partons , into polarized pdfs , describing the spin correlations between the proton and a parton , should be suitable even as a starting point .
instead in the scenario presented in these proceedings , we make use of the positivity bounds for dpds derived in @xcite . at the starting scale @xmath34 of evolution ,
we maximize each polarized dpd with respect to its unpolarized counterpart .
this gives polarized distributions equal to the unpolarized at the initial scale .
we will show a series of figures with curves for different scales . in each figure
, the upper row shows the polarized dpds and the lower row shows the ratio between polarized and unpolarized dpds .
the ratio indicates how important spin correlations are in the cross sections of dps processes .
we show the polarized distributions as functions of @xmath20 at @xmath48 and as functions of @xmath49 for @xmath50 .
the distribution for longitudinally polarized up quarks and antiquarks is shown in figure [ fig : long - u - max ] .
the polarized distribution @xmath51 evolves very slowly , but the degree of polarization decreases with the evolution scale .
this is due to the increase of the unpolarized dpds .
we find a degree of polarization around 50% at @xmath52 and above 20% at @xmath53 for @xmath54 and a wide range of @xmath49 .
longitudinally polarized up quarks and antiquarks , with initial conditions using the mstw pdfs .
here and in the following figures the upper row shows the polarized dpds and the lower row the ratio between polarized and unpolarized dpds .
, title="fig:",scaledwidth=49.0% ] longitudinally polarized up quarks and antiquarks , with initial conditions using the mstw pdfs .
here and in the following figures the upper row shows the polarized dpds and the lower row the ratio between polarized and unpolarized dpds .
, title="fig:",scaledwidth=49.0% ] transversely polarized up quarks and antiquarks , with initial conditions using the mstw pdfs .
color ( line style ) coding as in figure [ fig : long - u - max].,title="fig:",scaledwidth=49.0% ] transversely polarized up quarks and antiquarks , with initial conditions using the mstw pdfs .
color ( line style ) coding as in figure [ fig : long - u - max].,title="fig:",scaledwidth=49.0% ] transverse quark and antiquark polarization leads to characteristic azimuthal correlations in the final state of dpd processes @xcite .
they do not mix with gluons under evolution , nor with quarks or antiquarks of different flavors .
figure [ fig : trans - u - max ] shows the dpd for transversely polarized up quarks and antiquarks . there is a small decrease of the dpd with @xmath39 over the entire @xmath8 range , but the suppression of the degree of polarization is mainly due to the increase in the unpolarized distributions .
the evolution of the degree of polarization is similar to the case of longitudinal polarization , with a somewhat faster decrease . at intermediate and large @xmath8 values ,
the degree of polarization decreases slowly .
for @xmath50 it amounts to 40% at @xmath52 and to 10% at @xmath53 over a wide rapidity range .
gluons can be polarized longitudinally or linearly .
the unpolarized ( single or double ) gluon density increases rapidly at small momentum fractions due to the @xmath56 behavior of the gluon splitting kernel .
the absence of this low-@xmath29 enhancement in the polarized gluon splitting kernels lead us to expect that the degree of gluon polarization will vanish rapidly in the small @xmath29 region .
as can be seen in figure [ fig : long - ggjr - max ] for longitudinally polarized gluons , this is indeed the case .
the distribution @xmath57 does increase with evolution scale , but at a much lower rate than @xmath58 .
evolution quickly suppresses the degree of longitudinal gluon polarization in the small @xmath8 region .
the degree of polarization at at @xmath52 amounts to a degree of polarization equal to 30% and almost 20% at @xmath53 for @xmath54 , with a very weak dependence on @xmath49 .
our knowledge of the single gluon distribution at the low scale remains poor , and using as an alternative to the gjr distributions the mstw set the degree of polarization is reduced to around half the size or below .
linearly polarized gluons give rise to azimuthal asymmetries in dps cross sections @xcite .
the effect of evolution on the distribution of two linearly polarized gluons is shown in figure [ fig : lin - g - max ] .
we see that even the polarized distribution @xmath59 itself decreases with the scale .
together with the rapid increase of the unpolarized two - gluon dpd this results in a rapid decrease of the degree of linear polarization , especially at small @xmath8 . as in the case of longitudinal gluon polarization , using the mstw distributions at the starting scale results in an even faster suppression . in that case
the degree of polarization is tiny already at @xmath52 .
we conclude that the correlation between two linearly polarized gluons is quickly washed out by evolution and can only be relevant at rather large @xmath8 or rather low scales .
a. v. manohar and w. j. waalewijn , _ a qcd analysis of double parton scattering : color correlations , interference effects and evolution _ , _ phys.rev . _ * d85 * ( 2012 ) 114009 , [ http://xxx.lanl.gov/abs/1202.3794[arxiv:1202.3794 ] ] .
j. r. gaunt and w. j. stirling , _
double parton distributions incorporating perturbative qcd evolution and momentum and quark number sum rules _ , _ jhep _ * 1003 * ( 2010 ) 005 , [ http://xxx.lanl.gov/abs/0910.4347 [ arxiv:0910.4347 ] ] .
j. r. gaunt and w. j. stirling , _
double parton scattering singularity in one - loop integrals _ , jhep * 1106 * ( 2011 ) 048 [ arxiv:1103.1888 ] .
j. r. gaunt , _ single perturbative splitting diagrams in double parton scattering _
, jhep * 1301 * ( 2013 ) 042 [ arxiv:1207.0480 ] .
m. g. ryskin and a. m. snigirev , _ a fresh look at double parton scattering _ , phys .
d * 83 * ( 2011 ) 114047 [ arxiv:1103.3495 ] .
a. v. manohar and w. j. waalewijn , _ what is double parton scattering ? _ , phys .
b * 713 * ( 2012 ) 196 [ arxiv:1202.5034 ] .
b. blok , yu . dokshitzer , l. frankfurt and m. strikman , _ pqcd physics of multiparton interactions _ , eur .
j. c * 72 * ( 2012 ) 1963 [ arxiv:1106.5533 ] .
b. blok , yu . dokshitzer , l. frankfurt and m. strikman , _ perturbative qcd correlations in multi - parton collisions _ ,
m. diehl , t. kasemets , and s. keane , _ correlations in double parton distributions : effects of evolution _ , _ jhep _ * 1405 * ( 2014 ) 118 , [ http://xxx.lanl.gov/abs/1401.1233[arxiv:1401.1233 ] ] .
m. gluck , p. jimenez - delgado , and e. reya , _ dynamical parton distributions of the nucleon and very small - x physics _ , _
eur.phys.j . _ * c53 * ( 2008 ) 355366 , [ http://xxx.lanl.gov/abs/0709.0614 [ arxiv:0709.0614 ] ] . | we review the effect of scale evolution on a number of different correlations in double parton scattering ( dps ) .
the strength of the correlations generally decreases with the scale but at a rate which greatly varies between different types . through studies of the evolution ,
an understanding of which correlations can be of experimental relevance in different processes and kinematical regions is obtained . |
a _ coxeter group _ is a pair @xmath13 consisting of a group @xmath14 and a generating set @xmath15 which is presented using the relations @xmath16 for all @xmath17 and relations of the form @xmath18 for some , but not necessarily all , pairs @xmath19 .
the most common examples of infinite coxeter groups are _ affine weyl groups _ , which are groups generated by the reflections of an affine space across special collections of hyperplanes coming from root systems .
the theory of coxeter groups is both complicated and , especially in the case of affine weyl groups , highly - developed .
affine weyl groups are ubiquitous in the subject of smooth representations of algebraic groups over non - archimedean local fields due to their connection with hecke algebras of reductive groups , and many questions about hecke algebras can be reduced to questions about affine weyl groups .
let @xmath20 be a non - archimedean local field and let @xmath1 be a connected reductive affine algebraic @xmath20-group .
if @xmath21 is a compact - open subgroup and @xmath22 is a smooth complex representation of @xmath23 then the _ hecke algebra _
@xmath24 is the convolution algebra of all compactly - supported functions @xmath25 satisfying @xmath26 for all @xmath27 and @xmath28 .
many of the simple subcategories in the _ bernstein decomposition _ of the category of smooth representations of @xmath29 are equivalent to the category of modules over a hecke algebra of this form .
the _ center _ of such a hecke algebra is important because it consists of the ( functorial ) @xmath29-linear endomorphisms of the representations in the subcategory .
the case that @xmath23 is an _ iwahori subgroup _ and @xmath30 is the trivial @xmath31-dimensional representation yields one particularly important hecke algebra : the _ iwahori - hecke algebra_. in general , an iwahori subgroup is the group of @xmath32-points of a certain connected model @xmath33 of @xmath1 defined in general by bruhat and tits , although in nice cases like @xmath34 there is a much more straightforward description : an iwahori subgroup is the inverse image in @xmath35 of a borel subgroup in @xmath36 under the reduction - mod-@xmath37 map @xmath38 .
it can be shown that any iwahori - hecke algebra @xmath39 has a presentation , called the _ iwahori - matsumoto presentation _ , consisting of a basis of characteristic functions of the double - cosets @xmath40 together with a certain pair of relations which depend on some numerical parameters coming from @xmath1 .
it turns out that a group called the _ iwahori - weyl group _ serves as a system of representatives for these double - cosets .
the iwahori - weyl group is in general merely a semidirect extension of an affine weyl group but its behavior is nonetheless extremely similar to that of a true coxeter group . taken together with the numerical parameters , the group - theoretic structure of this `` quasi - coxeter group '' completely controls the ring - theoretic structure of @xmath39 . in this paper ,
i prove a group - theoretic property of conjugacy classes in iwahori - weyl groups , which is described precisely in the next subsection of this introduction .
the class of all iwahori - weyl groups properly contains the class of all affine weyl groups ( since affine weyl groups arise as the iwahori - weyl groups of semisimple and simply - connected @xmath1 ) and this property is new even in this narrower context .
additionally , there are implications for iwahori - hecke algebras @xmath39 . for example
, this property can be used to show that suitable linearly - independent collections of functions in the center @xmath41 also _ span _ the center .
the prototypical example of such a collection is the _
bernstein basis _ , although the existence and precise definition of such a basis do not appear in the literature for iwahori - hecke algebras of completely general connected reductive groups .
the article @xcite fills this gap , which i explain in more detail next , and also serves as a sample application of the main theorem of this paper . in unpublished work ,
concerning essentially only the case of split @xmath1 , bernstein introduced a particularly important basis for @xmath41 whose elements can be effectively calculated ( by computer , if desired ) and which simultaneously have a straightforward representation - theoretic interpretation : the basis elements are indexed by @xmath42-orbits of cocharacters valued in a certain maximal torus @xmath43 , and if @xmath44 is such an orbit then the corresponding basis element @xmath45 acts on the iwahori - fixed subspace of the ( normalized ) induced representation @xmath46 of an unramified character @xmath47 by the scalar @xmath48 .
this work was extended to the _ affine hecke algebra _ on any reduced root datum with any parameter system by lusztig in @xcite .
an affine hecke algebra on a reduced root datum @xmath49 is constructed by using the _ extended affine weyl group _ @xmath50 as a vector space basis , choosing parameters , and mimicking the iwahori - matsumoto presentation abstractly ( see 3.2 of @xcite and 7.1 of @xcite for details ) .
these abstractly defined affine hecke algebras are useful in the study of reductive groups for the following reason : in some cases , for example if @xmath1 is _ unramified _ , any iwahori - hecke algebra of @xmath1 is naturally isomorphic to an affine hecke algebra for appropriate choice of root datum and parameters .
unfortunately , many iwahori - weyl groups are _ not _ the extended affine weyl group of any root datum .
a specific example of such a @xmath1 is a ramified even - dimensional special orthogonal group ( see 1.16 in @xcite for the definition of such a group ) : such a group has _
torsion _ in the translation subgroup of its iwahori - weyl groups , which simply can not happen in the extended affine weyl group of a root datum because its translation subgroup @xmath51 is free by definition .
consequently , many iwahori - hecke algebras are _ not _ affine hecke algebras and so are not within the scope of the bernstein / lusztig work . in the article @xcite , i extend the bernstein / lusztig results to iwahori - hecke algebras of _ all _ connected reductive @xmath20-groups . after establishing , analogous to @xcite , a bernstein presentation for an iwahori - hecke algebra @xmath39 of a general connected reductive @xmath20-group , i determine , again following @xcite , a bernstein basis for @xmath41 . to prove that these functions indeed span the center ,
i apply the main theorem of this paper as explained in
[ sapplicationtoheckealgebras ] below . _
more precise definitions of everything here are given in [ snotation ] and [ siwahoriweylgroups ] .
_ let @xmath20 be a non - archimedean local field and @xmath1 a connected reductive affine algebraic @xmath20-group .
fix a maximal @xmath20-split torus @xmath52 and let @xmath53 be the corresponding iwahori - weyl group , which acts on the vector space @xmath54 .
it is known that @xmath53 contains as a normal subgroup the affine weyl group @xmath55 of a reduced root system @xmath56 , and that if @xmath57 is the subgroup of elements which act on @xmath58 by translations then @xmath53 splits as @xmath59 ( here @xmath42 denotes the finite weyl group ) .
further , there are sections for @xmath60 , so that @xmath53 also splits as @xmath61 .
if @xmath62 is a coxeter generating set for @xmath55 and @xmath63 is the corresponding length function , then @xmath63 extends to @xmath53 by inflation .
denote by @xmath64 a coxeter generating set for @xmath65 .
the main result of the paper is the following : fix @xmath4 . if @xmath5 then there exists @xmath66 and ( if necessary ) @xmath67 such that , setting @xmath68 , * @xmath69 for all @xmath70 , * both @xmath71 and @xmath72 , and * @xmath10 .
note that the last two properties asserted by the main theorem can be unified into `` @xmath73 '' .
a related result appears in the preprint @xcite : that for any element @xmath6 in the extended affine weyl group of a ( reduced ) root datum there is a sequence of conjugations by simple reflections , each of which preserves or decreases length , resulting in an element @xmath7 which is _ minimal _ length in its conjugacy class .
note however that the @xcite result is also true for _ finite _ weyl groups , originally proven by @xcite , whereas the main theorem here is special to _ infinite _ groups ( length is bounded on a finite group ! ) .
denote by @xmath74 the set of @xmath75-conjugacy classes in @xmath76 and recall that @xmath63 is constant on any @xmath77 .
denote by @xmath78 the @xmath79-coordinate of any @xmath4 .
let @xmath39 be an iwahori - hecke algebra for @xmath1 . by analyzing the equations that define the center @xmath41
, the main theorem can be used to prove dimension bounds for a certain filtration / partition of @xmath41 : if @xmath80 is the @xmath81-subspace of functions supported only on those @xmath6 for which @xmath82 and @xmath83 , and if @xmath84 is the total number of @xmath77 such that @xmath85 and @xmath86 for all @xmath87 then @xmath88 it follows that if @xmath89 is a linearly - independent subset of @xmath41 such that @xmath90 is supported only on those @xmath6 for which @xmath91 and @xmath78 is the same for all @xmath6 supporting @xmath90 then @xmath89 is a _
basis_. in [ snotation ] , i set some notation and define most of the objects that will be used throughout the paper .
i use several non - standard but convenient notations , and almost all of them can be found here .
two exceptions are a `` quasi - coxeter group '' and the hecke algebra on such a group , which are treated in [ siwahoriweylgroups ] and [ ssdefofheckealgebra ] , respectively . in [ siwahoriweylgroups ]
, i recall the notion of _ iwahori - weyl group _ and several of its most important properties , mostly to make explicit the scope of the main theorem .
reductive groups do not appear after this section . in [ smarkedalcoves
] , i define what is a _ marked alcove_. this is just a straightforward generalization of the notion of the _ type _ of a face of an alcove .
this extension is necessary to include iwahori - weyl groups , rather than just affine weyl groups , in the scope of the paper . in [ sdiamonds ]
, i precisely define the _ diamond property _ for an element of a coxeter group . in short , the diamond property is the property asserted by the main theorem .
i then define an equivalent property that refers only to pairs of alcoves and verify the equivalence of the two definitions . in [ slemmas ]
, i prove some simple geometric lemmas about weyl chambers , alcoves , hyperplanes , etc .
that will be used throughout [ sproof ] . in [ sproof
] , i prove the main theorem , divided into three cases . in [ ssdominantcase ] , i prove the `` dominant case '' , proposition [ pdominantnontranslationshavedcp ] , of the main theorem : _ if @xmath5 and @xmath6 sends the base alcove @xmath92 into the dominant chamber @xmath93 then @xmath6 has the diamond property_. it is obvious from the hypothesis that @xmath94 for all @xmath95 , and it is not hard to visualize why there also exists @xmath95 such that @xmath96 . in
[ ssantidominantcase ] , i prove the `` anti - dominant case '' , proposition [ pantidominantnontranslationshavecp ] , of the main theorem : _ if @xmath5 and @xmath97 then @xmath6 has the diamond property_. this is the most difficult case and the general case can be reduced to this one ( the complexity of the anti - dominant case is in some sense maximal while that of the dominant case is minimal this can be quantified somewhat by noting that @xmath98 for all @xmath95 ) .
the basic idea is that , by carefully inspecting the relative position and orientation of the alcoves @xmath92 and @xmath99 , one can perform an infinite sequence of conjugations by @xmath62 which do not _ decrease _ length and which continually move the alcoves in `` different directions '' , guaranteeing an eventual length-_increasing _
conjugation by @xmath62 . in [ ssintermediatecase ] , i finish proving the main theorem , showing that for an _ arbitrary _
@xmath5 one can repeatedly perform conjugations by @xmath62 which do not _ decrease _ length and such that eventually the situation qualifies for the anti - dominant case . in
[ sapplicationtoheckealgebras ] , i use the main theorem to give dimension bounds , proposition [ pdimensionbounds ] , for every subspace in the `` length filtration '' of @xmath41 .
the term `` length filtration '' is a slight abuse , since it is necessary to first filter @xmath41 by the lengths of its supporting elements and then partition each of those subspaces by the @xmath79-components of its supporting elements ( if @xmath79 is infinite then the subspaces in the length filtration are _ infinite _ dimensional , and without refining it further most iwahori - hecke algebras would be outside the scope of the paper ) . in [ sfigures ] , i include several color pictures that illustrate the iterative arguments that occur in the proof of the main theorem .
i use the affine weyl group and apartment of the exceptional type @xmath100 because there are too many coincidences for extremely symmetric types like @xmath101 to correctly explain things , and because i can suppress hyperplane labels for @xmath100 since its labelings are , unlike @xmath101 and @xmath102 , unambiguous ( the alcoves are @xmath103-@xmath104-@xmath105 triangles ) .
i thank thomas haines for carefully reading earlier drafts of this paper and suggesting a very large number of changes to the exposition , and especially for noticing that the original proof occurring in [ ssintermediatecase ] was needlessly labyrinthine .
i also thank my postdoctoral mentor , tonghai yang , for bringing me to the wonderful university of wisconsin at madison .
finally , i thank the referee for suggesting many improvements to the introduction , and also for the care and speed with which the report was composed and returned .
the symbols @xmath106 , @xmath107 , @xmath108 , and @xmath81 refer to the natural numbers ( including @xmath109 ) , the integers , the real numbers , and the complex numbers .
let @xmath110 be a _ reduced _ and _ irreducible _ root system .
let @xmath42 be the finite weyl group of @xmath110 , let @xmath111 be a simple system for @xmath42 .
let @xmath112 be the @xmath108-vector space spanned by the dual root system @xmath113 .
let @xmath114 be the natural pairing @xmath115 and @xmath116 the extension to @xmath117 .
let @xmath118 be the affine root system associated to @xmath110 , i.e. the set @xmath119 of affine functionals on @xmath112 .
the term _ hyperplane _ always means the null - set of an element of @xmath118 , not just an arbitrary codimension-@xmath31 affine subspace .
the term _ root hyperplane _ means the null - set of an element of @xmath110 .
let @xmath120 be the affine weyl group of @xmath118 and @xmath62 the simple system for @xmath120 extended from @xmath111 .
denote by @xmath121 the single element of @xmath122 and by @xmath123 the hyperplane fixed pointwise by @xmath121 .
denote by @xmath124 the usual length function relative to the generating set @xmath62 .
let @xmath125 be the subgroup of translations by elements of @xmath113 , so that @xmath126 .
choose a _ special _ vertex @xmath127 and identify @xmath42 to the finite weyl group at @xmath128 .
let @xmath93 be the weyl chamber at @xmath128 corresponding to @xmath111 and let @xmath129 be the alcove for which @xmath130
. if @xmath131 is a weyl chamber ( at some arbitrary special vertex ) , denote by @xmath132 the opposite chamber .
the term _ face _ always means a codimension-@xmath31 facet of an alcove ( or weyl chamber ) .
the term _ wall _ always means the unique hyperplane containing some face of some alcove ( or weyl chamber ) .
if @xmath133 and @xmath134 are ( distinct ) adjacent alcoves then denote by @xmath135 the wall separating them . the term _ half - space _ refers to one of the connected components of the complement in @xmath112 of a hyperplane ( in particular , half - spaces are open , and therefore also weyl chambers and alcoves ) .
a subset @xmath136 is called _ simplicial _ iff there exists a set @xmath137 of alcoves in @xmath112 such that @xmath138 .
for example , half - spaces and weyl chambers are simplicial .
the _ length _ of a gallery @xmath139 is defined to be @xmath140 ( for consistency with the length function @xmath63 ) .
to say that a hyperplane @xmath141 is an _ intermediate wall _ of a ( non - stuttering ) gallery @xmath139 is the same as to say that there exists an index @xmath142 such that @xmath141 is the unique hyperplane containing the face shared by @xmath143 and @xmath144 .
i sometimes use the fact that if @xmath145 is simplicial and ( topologically ) convex then it is convex in the combinatorial sense , i.e. if @xmath146 are alcoves and @xmath147 is a _ minimal _ gallery from @xmath133 to @xmath148 then @xmath149 for _ all _ @xmath150 ( see theorem 5.11.4 in @xcite ) .
i sometimes refer to this property as _
simplicial convexity_. for example , half - spaces and weyl chambers are simplicially - convex . by _ infinite gallery _
i mean an infinite sequence @xmath151 of alcoves such that for all @xmath152 the alcoves @xmath153 and @xmath154 are adjacent .
an infinite gallery is _ minimal _ iff for _ all _ @xmath155 the finite sub - gallery @xmath156 is minimal in the usual sense .
let @xmath157 be the usual @xmath106-valued metric on the set of all alcoves : @xmath158 is defined to be the minimum length among all galleries from @xmath133 and @xmath134 .
equivalently , @xmath158 is the total number of hyperplanes separating @xmath133 from @xmath134see theorem 5.1.4 in @xcite .
i also use two extensions of this distance function @xmath157 : if @xmath159 is a _ simplicial _ subset and @xmath160 is an alcove then define @xmath161 if @xmath162 is a _ vertex _ , and @xmath145 is as before , then i define @xmath163 in applications , @xmath145 will be either a single half - space or a weyl chamber .
in this section , i define the object which is the main focus of this paper : the _ iwahori - weyl group_. _ in this subsection , i briefly explain what is an iwahori - weyl group and isolate its key properties .
the purpose here is merely to explain the scope of the main theorem , so all proofs are omitted , although i give references whenever possible .
_ let @xmath20 be a non - archimedean local field and let @xmath1 be a connected reductive affine algebraic @xmath20-group .
let @xmath52 be a maximal @xmath20-split torus and set @xmath164 , a minimal @xmath20-levi subgroup .
certain group homomorphisms called _ kottwitz homomorphisms _ are very useful to understand the theory of parahoric subgroups , and in particular iwahori subgroups .
the kottwitz homomorphism of a connected reductive affine algebraic @xmath20-group @xmath141 is a surjective group homomorphism @xmath165 , where @xmath166 is a finitely - generated abelian group whose precise definition is not relevant to this paper see 7 of @xcite for the definitions of @xmath167 and @xmath166 ( the map @xmath167 occurring here is ( 7.7.1 ) in @xcite ) .
the kernel of the kottwitz homomorphism is denoted by @xmath168 .
one may define the _ iwahori - weyl group _ of @xmath169 to be the quotient @xmath170 . note that this is seemingly different from the quotient occurring in remark 9 of the appendix to @xcite , but it can be shown that there is a tautological isomorphism between the two quotients .
proposition 8 combined with remark 9 of the appendix to @xcite shows if @xmath171 is an iwahori subgroup then the double - cosets modulo @xmath172 are naturally represented by @xmath53 .
the iwahori - weyl group @xmath53 acts on the vector space @xmath54 .
there are two extremely important ways to express the iwahori - weyl group @xmath53 as a semidirect product . by the work of bruhat and tits ,
it is known that there exists a reduced root system @xmath56 such that the affine weyl group @xmath55 , in the sense of ch vi 2 no . 1 of @xcite
, is a subgroup of @xmath53 ( this root system is called an _ chelonnage _ in 1.4 of @xcite ; see 4 of @xcite for an extremely nice table listing @xmath56 for every almost - simple group , and much more ) . denoting by @xmath173 the finite weyl group of @xmath56 , it can be shown that @xmath174 and that @xmath175 . further , the subgroup @xmath176 acts on @xmath58 by translations and the subgroup @xmath177 acts on @xmath58 by invertible affine transformations that stabilize any prescribed base alcove in @xmath58 . for more details of all these semidirect products , consult @xcite and @xcite .
a few comments are necessary to emphasize the small but important difference between the notion of an iwahori - weyl group and the possibly more familiar notion of an extended affine weyl group of a ( reduced ) root datum .
first , it is possible that both @xmath176 and @xmath177 have torsion elements ( in fact , it can be shown that the torsion of the former is contained in the torsion of the latter ) .
second , the elements of @xmath176 are not actually translations , but merely _ act _ by translations on @xmath58 .
third , it is possible that some non - identity elements in @xmath176 act by the identity on @xmath58 . using the previous discussion [ ssdefofiwahoriweylgroup ] as a guide , i now isolate the relevant properties of the iwahori - weyl group and present them axiomatically for clarity .
let @xmath178 be the group of invertible affine transformations of @xmath112 which normalize @xmath120 .
fix a finitely - generated abelian group @xmath79 , a group homomorphism @xmath179 , and act by @xmath79 on @xmath112 via this @xmath180 .
the _ quasi - coxeter group _ @xmath53 extended from @xmath120 by @xmath181 is the semidirect product @xmath182 and acts on @xmath112 in the obvious way : @xmath183 for all @xmath184 and @xmath185 .
denote by @xmath78 the projection of @xmath4 into @xmath79 .
note that if @xmath186 are conjugate then @xmath187 , since @xmath79 is abelian .
strictly speaking , the space @xmath112 on which the quasi - coxeter group acts is only a proper subspace of the space @xmath58 on which the iwahori - weyl group acts when @xmath1 is not semisimple .
but due to the way that affine root hyperplanes in @xmath58 are defined , the details of which i omit in this paper , the difference is totally irrelevant from a group - theory perspective .
the setup that i use is essentially the same as that used in ch vi 2 no . 3 of @xcite .
let @xmath57 be the subgroup consisting of all elements that act by translations on @xmath112 , and note that @xmath76 is obviously normalized by @xmath188 .
extend the length function @xmath124 to @xmath53 by inflation along the projection @xmath189 .
i impose the following hypotheses : * assume that @xmath190 for all @xmath191 . *
assume that @xmath76 is a semidirect complement , i.e. that @xmath192 .
* assume that @xmath63 is constant on each @xmath42-conjugacy class in @xmath76 .
* assume that @xmath76 is finitely - generated and abelian .
note that by choice of @xmath178 , the action by @xmath79 on @xmath112 permutes the set of hyperplanes in @xmath112 .
therefore , hypothesis is equivalent to the hypothesis that @xmath193 for all @xmath191 .
the definitions in this section , which are mostly just a variant on the notion of the _ type _ of a face , will be used heavily in [ ssantidominantcase ] and [ ssintermediatecase ] .
a _ labeling _ of an alcove @xmath160 is a bijection from @xmath62 to the set of walls of @xmath133 .
a _ marked alcove _ is a triple @xmath194 such that @xmath133 is an alcove , @xmath195 is a special vertex and @xmath196 is a labeling of @xmath133 . the _ weyl chamber _ of a marked alcove
@xmath194 is the unique weyl chamber at @xmath197 containing @xmath133 . whenever the special vertex @xmath197 and labeling @xmath196 of a marked alcove @xmath194 are understood and there is no danger of confusion , i abuse notation and refer to @xmath133 as the marked alcove .
accordingly , if @xmath133 represents a marked alcove then its special vertex is denoted by @xmath198 , its labeling by @xmath199 , its weyl chamber by @xmath200 , and the hyperplane @xmath201 is called simply `` the wall of @xmath133 labeled by @xmath8 '' .
two marked alcoves @xmath202 and @xmath203 are called _ compatible _ iff there exists @xmath4 such that @xmath204 , @xmath205 and @xmath206 .
the marked alcoves are called _ nt - compatible _ iff @xmath207 .
finally , the base alcove @xmath92 is given the tautological labeling , and all other alcoves inherit ( in general , multiple ) labelings via the action of @xmath53 in the obvious way : the _ base labeling _ is the bijection @xmath208 from @xmath62 to the set of walls of the base alcove @xmath92 defined by assigning to @xmath8 the unique wall of @xmath92 that is fixed pointwise by @xmath8 .
the _ base marking _ is the marked alcove @xmath209 . for each @xmath4 ,
the _ @xmath6-labeling _ is defined to be the bijection @xmath210 from @xmath62 to the set of walls of the alcove @xmath99 .
@xmath6-marked alcove _ is by definition the triple @xmath211 .
as before , i sometimes abuse notation by using @xmath99 to refer to the @xmath6-marked alcove .
note that @xmath99 is compatible with @xmath92 and it is nt - compatible with @xmath92 if and only if @xmath207 . when @xmath212 , alcoves are in bijection with @xmath6-marked alcoves ( due to simple - transitivity of affine weyl groups on alcoves ) and a labeling is essentially just the assignment to every face of every alcove its _ type _ in the usual way .
the following operation will be used frequently in the limiting / inductive arguments of [ ssantidominantcase ] and [ ssintermediatecase ] : for any marked alcove @xmath202 and any @xmath66 , the marked alcove @xmath213 is by definition the triple @xmath214 , where @xmath215 is the wall of @xmath133 labeled by @xmath8 . note that if two marked alcoves @xmath133 and @xmath134 are nt - compatible then @xmath216 and @xmath217 are also nt - compatible for all @xmath66 .
i frequently use the fact that applying a sequence of various @xmath218 operations to a single marked alcove results in a sequence of marked alcoves whose ( un - marked ) alcoves form a _
gallery_. conversely , if @xmath219 is a marked alcove and @xmath220 is a gallery with no repeated alcoves , then each @xmath153 becomes a marked alcove in a unique way , by iteratively applying @xmath218 operations across each intermediate wall of the gallery .
in this situation , the `` label '' of @xmath221 is understood to refer to the element of @xmath62 corresponding to the wall @xmath221 relative to the labeling of @xmath153 ( or @xmath154 ) inherited from @xmath219 .
@xmath186 are _ laterally - conjugate _ iff there exist @xmath67 such that @xmath222 and @xmath69 for all @xmath70 .
any @xmath4 is always considered to be laterally - conjugate to itself . @xmath4 has the _ direct diamond property _
iff there exists @xmath66 such that * @xmath223 , * @xmath94 , and * @xmath96 . by using the well - known lemma [ lhumphreys ] , these three properties _
could _ be replaced by the single property `` @xmath224 '' , but this formulation is not as convenient for me .
@xmath4 has the _ diamond property _ iff it is laterally - conjugate to an element with the direct diamond property .
i frequently use the following geometric characterization of length : [ lgeometriclength ] let @xmath4 and @xmath66 be arbitrary . if @xmath225 and @xmath226 then * @xmath94 if and only if @xmath92 and @xmath99 are on the _ same _ side of @xmath141 , * @xmath96 if and only if @xmath92 and @xmath99 are on the _ same _ side of @xmath227 , and * @xmath228 . when @xmath212 this is all well - known : see proposition ( c ) in 4.4 and theorem ( b ) in 4.5 of @xcite .
the more general statement is immediate by definition of the labeling @xmath229 because @xmath63 factors through @xmath120 and @xmath79 stabilizes @xmath92 .
here are the gallery - theoretic versions of the above 3 definitions : an ordered pair @xmath230 of marked alcoves is _ laterally - conjugate _ to another pair @xmath231 iff there exists a gallery @xmath232 from @xmath133 to @xmath148 and a gallery @xmath139 from @xmath134 to @xmath233 such that * @xmath234 , and * both @xmath235 and @xmath221 have the _ same _ label . for all @xmath70 . note the symmetry in the definition of lateral - conjugacy : @xmath230 is laterally - conjugate to @xmath231 if and only if @xmath236 is laterally - conjugate to @xmath237 . a pair @xmath238 of marked alcoves has the _ direct diamond property
_ iff there exists @xmath66 such that * @xmath239 , * both alcoves are on the same side of @xmath240 , and * both alcoves are also on the same side of @xmath241 . note that , due to the symmetry in the definition , the direct diamond property refers to _ unordered _ pairs of alcoves .
a pair @xmath238 of marked alcoves has the _ diamond property _ iff it is laterally - conjugate to a pair with the direct diamond property .
it is easy to show using the previous lemma that the two notions of diamond property coincide : [ lequivalentdiamonds ] the element @xmath4 has the diamond property if and only if the pair @xmath242 of marked alcoves has the diamond property . by lemma [ lgeometriclength ] , @xmath243 . since @xmath157 is invariant under the diagonal action of @xmath120 , @xmath244 . by definition of the labelings @xmath208 and @xmath229 , @xmath245 .
altogether , @xmath246 .
since the operation @xmath218 always creates galleries , this shows that the two notions of `` lateral - conjugacy '' are equivalent . since @xmath247 if and only if @xmath248 , it follows that the condition @xmath223 is equivalent to the condition @xmath249 .
finally , the fact that the remaining two statements in both direct diamond properties are equivalent follows directly from lemma [ lgeometriclength ] .
although i will have no direct use for this in the remainder of the paper , note that the statement of lemma [ lequivalentdiamonds ] could be made much more specific : if @xmath6 is laterally - conjugate to @xmath7 via the sequence @xmath250 then @xmath251 is laterally - conjugate to @xmath252 via a gallery whose intermediate walls are labeled ( in tandem ) by the same sequence @xmath250 , etc .
[ laffinehyperplanesarenotbipolar ] let @xmath141 be a hyperplane in @xmath112 .
if _ both _ @xmath253 _ and _ @xmath254 then in fact both @xmath255 and @xmath256 .
i usually apply lemma [ laffinehyperplanesarenotbipolar ] in the following way : if @xmath257 then @xmath258 .
suppose @xmath259 and @xmath260 .
since @xmath141 is the null - set of an affine root , there exists @xmath261 such that @xmath262 . on the other hand , @xmath263 and
@xmath264 implies @xmath265 .
it is easy to check from the definitions of @xmath93 and @xmath266 that if @xmath267 or @xmath268 ( or both ) then necessarily @xmath269 . but
@xmath270 since a _ root _
hyperplane can never intersect a weyl chamber at @xmath128 , so this is impossible and therefore both @xmath271 and @xmath272 . [ lnegativechambercontainment ]
let @xmath273 be an alcove and @xmath141 a wall of @xmath133 that is _ not _ a wall of the weyl chamber @xmath93 .
let @xmath134 be any alcove and @xmath274 a vertex . if both @xmath133 and @xmath92 are on the same side of @xmath141 and @xmath275 then @xmath134 is also on the same side of @xmath141 as @xmath133 .
let @xmath276 be the face supported by @xmath141 . if it were true that @xmath277 then necessarily @xmath141 would be a wall of @xmath93 .
this is prohibited by hypothesis on @xmath141 , so @xmath257 .
suppose for contradiction that @xmath141 separated @xmath134 from @xmath133 .
by hypothesis on @xmath133 , @xmath141 must separate @xmath134 from @xmath92 . by hypothesis on @xmath278 , the set
@xmath279 is path - connected and obviously @xmath280 , so necessarily @xmath281 .
but this contradicts lemma [ laffinehyperplanesarenotbipolar ] since @xmath257 is known already .
[ ldoublechamberlemma ] let @xmath131 be a weyl chamber at some ( arbitrary ) special vertex @xmath197 .
if @xmath282 then either @xmath283 or @xmath284 .
suppose for contradiction that both @xmath285 and @xmath286 .
choose an alcove @xmath287 and an alcove @xmath288 .
let @xmath289 be a minimal gallery from @xmath133 to @xmath148 .
by choice of @xmath290 , there exists a wall @xmath141 of @xmath131 separating @xmath133 from @xmath148 .
such an @xmath141 must be an intermediate wall of the gallery ( see lemma 5.1.5 of @xcite ) , say @xmath291 . since the gallery is _ minimal _ and both @xmath292 , simplicial convexity of @xmath93 forces @xmath293 for all @xmath70 .
this means that @xmath294 , ( for example , if @xmath295 is the common face of @xmath296 and @xmath297 then @xmath298 ) .
but by hypothesis @xmath282 and obviously @xmath299 , so @xmath254 also .
this contradicts lemma [ laffinehyperplanesarenotbipolar ] .
[ lgallerytoinfinity ] let @xmath131 be a weyl chamber at some ( arbitrary ) special vertex .
if @xmath300 then there exists a minimal infinite gallery @xmath301 within @xmath93 such that @xmath302 ( see [ ssgalleries ] for the notions of distance @xmath303 and infinite gallery ) it is not always possible to have a sequence of alcoves for which the sequence of distances is _ monotone _ increasing . choose some translation @xmath304 such that @xmath305 ( e.g. translation by @xmath306 where @xmath307 and @xmath308 are the fundamental coweights ) .
i claim that the required gallery can be constructed by iterating @xmath309 .
let @xmath310 ( @xmath311 ) be a reduced expression and let @xmath312 $ ] be the remainder of @xmath70 mod @xmath313 .
give @xmath314 the base marking as usual and define a sequence of marked alcoves inductively by @xmath315 } } $ ] ( so the finite subgallery @xmath316 is just the usual gallery associated to the word @xmath317 ) . by construction , this sequence is an _
infinite gallery _ in @xmath93 . since @xmath318 , a general property of dominant translations in an affine weyl group
, it follows that the infinite gallery @xmath319 is _
i first show that the infinite subsequence @xmath320 , @xmath321 , diverges from @xmath131 , and then i use the triangle - inequality to prove the full limit property . for any alcove @xmath160 and radius @xmath322 ,
let @xmath323 be the set of alcoves @xmath324 such that @xmath325 .
it is clear from the `` cone '' property of weyl chambers and the boundedness of alcoves that for any @xmath322 , there exists @xmath326 such that @xmath327 for all @xmath328 .
it is also clear that @xmath329 for all @xmath328 because otherwise there would be some alcove @xmath330 such that @xmath331 , but this would imply @xmath332 , which contradicts the hypothesis @xmath300 .
this establishes the claim for the subsequence .
let radius @xmath333 be arbitrary .
fix @xmath334 ( recall @xmath335 ) . for any @xmath336 ,
let @xmath337 be the _
@xmath338 such that @xmath339 .
observe that if @xmath340 then @xmath341 .
altogether , if @xmath330 is an _
arbitrary _ alcove and @xmath340 then @xmath342 since @xmath343 by definition of @xmath337 , it follows that @xmath344 .
[ pdominantnontranslationshavedcp ] fix @xmath4 . if @xmath207 and @xmath345 then @xmath6 has the direct diamond property realized by some @xmath95 . let @xmath346 and @xmath347 be such that @xmath348 . by hypothesis , @xmath349 .
let @xmath350 be a wall of @xmath351 which separates @xmath352 from @xmath92 , and note that @xmath353 .
let @xmath141 be the wall of @xmath92 such that @xmath354 and let @xmath95 be the element fixing @xmath141 pointwise ( in other words , @xmath355 ) .
i claim that @xmath8 realizes the direct diamond property for @xmath6 .
i first claim that @xmath356 .
suppose for contradiction that @xmath357 . because @xmath309 is a _ translation _ , @xmath352 and @xmath358 are on the _ same _ side of @xmath357 .
on the other hand , @xmath350 separates @xmath92 from @xmath352 by choice .
together , @xmath350 separates @xmath92 from @xmath359 .
but @xmath345 by hypothesis , so it is impossible for the _ root _ hyperplane @xmath350 to separate @xmath92 from @xmath99 . by hypothesis that @xmath345 , it is automatic that @xmath94 . to show that @xmath96 , it suffices by lemma [ lgeometriclength ] to show that both alcoves @xmath92 and @xmath99 are on the same side of @xmath360 .
let @xmath361 be the _ positive _ root whose null - set is @xmath350 .
by choice of @xmath350 , @xmath362 for all @xmath363 . since @xmath364 and @xmath365 , it must be true that @xmath366 . since @xmath309 is a _ translation _
, this implies that there exists @xmath367 such that @xmath368 is the null - set of @xmath369 and @xmath359 consists of points @xmath370 such that @xmath371 . since @xmath356
, it must be true that @xmath372 .
but @xmath373 for all @xmath374 so @xmath92 and @xmath99 are on the same side of @xmath360 , as desired .
i now show that @xmath223 .
suppose for contradiction that @xmath375 .
then @xmath248 , and since @xmath364 , it follows that @xmath376 . since @xmath309 is a _ translation _ and @xmath377 , @xmath378 .
combining with @xmath248 implies @xmath379 also .
but @xmath380 so this contradicts @xmath356 .
[ rindependenceofchamber ] note that it is not important which of the two alcoves is considered the `` base '' alcove , nor is it important which chamber of the base alcove is considered `` dominant '' . in other words , if @xmath133 and @xmath134 are nt - compatible marked alcoves and @xmath381 then @xmath238 has the direct diamond property .
it is plausible that one might be able to prove the diamond property for general @xmath207 by proving that there always exists a lateral conjugate @xmath7 of @xmath6 such that one of @xmath382 or @xmath92 is contained in some weyl chamber of the other .
this latter statement is _
false_. see figure [ example - g2-annoyingorbit ] for an example in the case of the exceptional affine weyl group @xmath100 .
let @xmath220 be a gallery , @xmath133 an alcove , and @xmath141 a wall of @xmath133 .
the triple @xmath383 is an _
iff 1 .
[ umbsamehalfspace ] _ all _ alcoves @xmath153 are on the same side of @xmath141 as @xmath133 , and 2 .
[ umbextendstominimal ] @xmath220 can be extended to a _
minimal _ gallery from @xmath219 to @xmath133 .
observe that to say @xmath220 can be extended to a minimal gallery from @xmath219 to @xmath133 is the same as to say both that @xmath220 is a minimal gallery itself and that each intermediate wall @xmath221 ( @xmath384 ) separates @xmath153 from @xmath133 ( i use this observation in the proof of induction lemma ) .
let @xmath220 be a gallery and @xmath385 a special vertex .
let @xmath292 be ( distinct ) adjacent alcoves , separated by a wall @xmath141 .
let @xmath350 be a wall of @xmath148 .
assume that 1 .
[ enuminductionhypoth ] @xmath386 is an umbrella , 2 .
[ enumvertexinchamber ] @xmath387 , 3 .
[ enumawayfrombase ] the base alcove @xmath92 is on the _ same _ side of @xmath350 as @xmath148 , 4 .
[ enumnotbasewall ] @xmath350 is _ not _ a wall of the weyl chamber @xmath93 , and 5 .
[ enumcontradictionhypoth ] the wall @xmath388 _ separates _
@xmath389 from @xmath133 .
then @xmath390 is an umbrella . by hypotheses ( [ enumvertexinchamber ] ) , ( [ enumawayfrombase ] ) , and ( [ enumnotbasewall ] ) ,
lemma [ lnegativechambercontainment ] implies that @xmath219 is contained on the same side of @xmath350 as @xmath148 .
since half - spaces are _ simplicially - convex _ , it therefore suffices to show only umbrella property ( [ umbextendstominimal ] ) , i.e. that @xmath220 can be extended to a _
minimal _ gallery connecting @xmath219 to @xmath148 ( because then both endpoints of the gallery , and therefore the whole gallery , must be contained in that half - space ) .
let @xmath391 ( @xmath392 ) be all the intermediate walls of the gallery @xmath220 .
note that @xmath393 separates @xmath153 from @xmath133 for all @xmath394 by hypothesis ( [ enuminductionhypoth ] ) ( more specifically , umbrella property ( [ umbextendstominimal ] ) ) and for @xmath395 by hypothesis ( [ enumcontradictionhypoth ] ) . by the observation preceding this proof
, it therefore suffices to show that the alcoves @xmath133 and @xmath148 are on the _ same _ side of @xmath393 for all @xmath396 .
but this is obviously true : if the claim were _ false _ for @xmath393 , then necessarily @xmath397 , the only hyperplane separating @xmath133 from @xmath148 , which would mean that @xmath141 separated @xmath153 from @xmath133 , a contradiction to hypothesis ( [ enuminductionhypoth ] ) ( more specifically , umbrella property ( [ umbsamehalfspace ] ) ) .
[ pantidominantnontranslationshavecp ] fix @xmath4 .
if @xmath207 and @xmath97 then @xmath6 has the diamond property . view ( sequentially ! ) figures [ example - g2-inductionlemma00 ] to [ example - g2-inductionlemma04 ] for a picture of the use of induction lemma in this proof .
let @xmath134 be the @xmath6-marked alcove @xmath99 .
by lemma [ lequivalentdiamonds ] , it suffices to show that @xmath398 has the diamond property . by lemma [ ldoublechamberlemma ] , either @xmath399 or @xmath400 .
if @xmath400 then the claim follows from proposition [ pdominantnontranslationshavedcp ] .
( using origin @xmath401 and dominant chamber @xmath402 ; see remark [ rindependenceofchamber ] ) .
so , assume that @xmath399 .
applying lemma [ lgallerytoinfinity ] to the chambers @xmath93 and @xmath402 yields a certain infinite minimal gallery @xmath301 within @xmath93 . as usual , give @xmath92 the base marking and let @xmath403 be the infinite sequence in @xmath62 such that @xmath404 , @xmath405 , etc .
let @xmath393 be the wall of the marked alcove @xmath406 labeled by @xmath407 . similarly , use the sequence @xmath403 to define , relative to the prescribed labeling of @xmath134 , a corresponding infinite gallery : @xmath408 as before , each @xmath153 here represents a _ marked _ alcove . note
that by definition of the labeling @xmath409 , the gallery @xmath410 is simply the image under @xmath6 of the gallery @xmath319 . in particular
, @xmath410 is _ minimal _ and @xmath411 for all @xmath70 . because the gallery @xmath319 starts at @xmath92 and is contained completely within @xmath93 , necessarily @xmath412 .
because of this and the hypothesis on @xmath278 , lemma [ lnegativechambercontainment ] says that alcoves @xmath219 and @xmath413 are on the _ same _ side of @xmath414 , i.e. @xmath415 .
let @xmath416 be the wall of @xmath219 labeled by @xmath417 .
if @xmath219 and @xmath418 are on the _ same _ side of @xmath227 then necessarily @xmath419 ( because @xmath414 _ separates _
@xmath219 from @xmath418 ) and both @xmath219 and @xmath413 are on the _ same _ side of @xmath227 ( because @xmath414 is the unique hyperplane separating @xmath413 from @xmath418 and @xmath420 ) .
it is then immediate from the definition that @xmath421 has the direct diamond property ( realized by @xmath417 ) .
otherwise , @xmath227 _ separates _
@xmath219 from @xmath418 and by lemma [ lgeometriclength ] , @xmath422 , i.e. @xmath423 is laterally - conjugate to @xmath424 via @xmath417 . in these circumstances
, induction lemma implies that @xmath425 is an umbrella : * the non - numbered hypotheses of induction lemma are true by choice , * hypothesis ( [ enuminductionhypoth ] ) is true because @xmath426 is trivially an umbrella , * hypothesis ( [ enumvertexinchamber ] ) is true by hypothesis on @xmath6 , * hypothesis ( [ enumawayfrombase ] ) is true by choice of @xmath427 because @xmath319 is _ minimal _ , * hypothesis ( [ enumnotbasewall ] ) is true by choice of @xmath427 because @xmath428 , and * hypothesis ( [ enumcontradictionhypoth ] ) is true by the assumption that @xmath429 did _ not _ have the direct diamond property for @xmath417 ( see previous paragraph : by choice @xmath430 ) .
so , the triple @xmath431 is an umbrella by induction lemma . in particular
, @xmath432 by umbrella property ( [ umbsamehalfspace ] ) .
i now iterate this process .
let @xmath433 be the wall of the marked alcove @xmath434 labeled by @xmath435 .
if @xmath434 and @xmath436 are on the _ same _ side of @xmath227 then @xmath437 ( because @xmath427 _ separates _
@xmath434 from @xmath436 ) and both @xmath434 and @xmath418 are on the _ same _ side of @xmath227 ( because @xmath427 is the unique hyperplane separating @xmath418 from @xmath436 and @xmath438 ) .
it is then immediate from the definition that @xmath439 has the direct diamond property realized by @xmath435 and therefore @xmath398 , being laterally - conjugate to it , has the diamond property .
otherwise , @xmath227 separates @xmath434 from @xmath436 and by lemma [ lgeometriclength ] , @xmath440 , i.e. @xmath441 is laterally - conjugate to @xmath423 via @xmath435 , and therefore also laterally - conjugate to @xmath424 . in these circumstances
, induction lemma implies that @xmath442 is an umbrella : * the non - numbered hypotheses are again true by choice , the status of hypothesis ( [ enumvertexinchamber ] ) has not changed , and hypothesis ( [ enumawayfrombase ] ) is true by choice of @xmath443 for the same reason as before , * hypothesis ( [ enuminductionhypoth ] ) is known by the previous iteration , * hypothesis ( [ enumnotbasewall ] ) is true by choice of @xmath443 because @xmath444 , and * hypothesis ( [ enumcontradictionhypoth ] ) is supplied by the assumption that @xmath439 did _ not _ have the direct diamond property for @xmath435 ( see previous paragraph : by choice @xmath445 ) .
the above induction shows that if @xmath367 and @xmath446 does not have the direct diamond property for all @xmath447 then @xmath448 is laterally - conjugate to @xmath424 for all @xmath447 , and in particular , @xmath449 for all @xmath447 .
but lemma [ lgallerytoinfinity ] says that @xmath450 , so there must exist @xmath451 such that @xmath446 has the direct diamond property .
if @xmath451 is the smallest such index then @xmath424 is laterally - conjugate to a pair with the direct diamond property , as desired .
[ rindependenceofchamber2 ] similar to the `` dominant case '' proposition [ pdominantnontranslationshavedcp ] , the choices of vertex , alcove , and chamber are notationally convenient but otherwise totally unimportant to the conclusion of proposition [ pantidominantnontranslationshavecp ] . in other words , if @xmath133 and @xmath134 are nt - compatible marked alcoves and @xmath452 then @xmath238 has the diamond property . in type
a , there is very simple proof that if @xmath453 then @xmath6 has the diamond property using at most one lateral - conjugation .
because of the extreme symmetry of type a , the inverse image @xmath454 consists of @xmath266 together with _ all _ alcoves @xmath134 such that @xmath455 . because of this ,
if @xmath6 does not already have the direct diamond property , @xmath121 laterally conjugates @xmath6 into the dominant chamber , in which case proposition [ pdominantnontranslationshavedcp ] applies .
of course , this idea fails in ( almost ? ) every other type .
i now prove that the general case can , at worst , be reduced to the anti - dominant case , proposition [ pantidominantnontranslationshavecp ] : suppose @xmath133 and @xmath134 are marked alcoves .
if @xmath133 and @xmath134 are nt - compatible then @xmath238 has the diamond property . in particular , if @xmath4 and @xmath207 then @xmath6 has the diamond property
view ( sequentially ! ) figures [ example - g2 - 2ndinductionlemma-00 ] to [ example - g2 - 2ndinductionlemma-01 ] for a picture of the iteration used in this proof .
let @xmath456 be the set of all @xmath66 such that if @xmath457 then the following three properties are true simultaneously : @xmath458 , both @xmath133 and @xmath134 are on the _ same _ side of @xmath141 , and @xmath459 . if @xmath460 then by definition for every wall @xmath141 of @xmath133 containing @xmath461 either @xmath141 _ separates _
@xmath133 from @xmath134 or @xmath462 . in this case
, @xmath463 and proposition [ pantidominantnontranslationshavecp ] applies .
so assume that @xmath464 and let @xmath465 be arbitrary .
let @xmath466 be a gallery realizing the distance @xmath467 ( see [ ssgalleries ] for this notion of distance ) . by definition
, this means that @xmath468 , the gallery is minimal , and @xmath469 ( note that @xmath470 is possible ) . set @xmath457 . by definition of @xmath456 , @xmath141 separates @xmath134 from every alcove in @xmath471 and @xmath459 so @xmath141 also separates @xmath219 from every alcove in @xmath471 .
therefore , @xmath141 must be an intermediate wall of the gallery @xmath466 ( see lemma 5.1.5 of @xcite ) .
let @xmath472 be the index such that @xmath473 .
then the sequence of alcoves @xmath474 is a _
gallery_. this is obviously a gallery `` from @xmath401 to @xmath475 '' and has fewer than @xmath157 alcoves . by choice of @xmath8 and the compatibility hypothesis , @xmath476 .
altogether , @xmath477 .
on the other hand , by choice of @xmath8 , lemma [ lgeometriclength ] implies that @xmath478 .
since these distances @xmath479 are @xmath106-valued , this means that one may iterate the previous process until a pair of alcoves @xmath231 is constructed which is laterally - conjugate to @xmath230 and such that either @xmath231 has the direct diamond property or @xmath480 , in which case proposition [ pantidominantnontranslationshavecp ] applies . if @xmath481 then @xmath5 if and only if @xmath482 , and it is easy to show that both @xmath483 and @xmath484 directly : combine the exchange property of the coxeter group @xmath485 with the fact that all reduced expressions for a single element must use the _ same _ subset of @xmath62 ( see proposition 7 in ch iv 1 no
. 8 of @xcite ) to conclude that neither length can _ decrease_. it is tempting to think that such @xmath6 always have the direct diamond property realized by @xmath121 but this is not always true : in the affine weyl group @xmath102 ( or @xmath100 ) , there exists @xmath95 such that @xmath486 , so @xmath487 for @xmath488 .
fix a function @xmath489 which is invariant under conjugation by @xmath53 . in the rest of this section [ sapplicationtoheckealgebras ] ,
i assume given a @xmath81-algebra @xmath39 which , as a @xmath81-vector space , has a basis of elements @xmath490 indexed by all @xmath4 .
further , denoting the ring operation by @xmath491 , i assume that the following _ iwahori - matsumoto identities _ are true in @xmath39 : for all @xmath4 and @xmath66 , @xmath492 note that because of the way that the length function @xmath63 was extended to @xmath53 , if @xmath191 and @xmath4 then @xmath493 . if @xmath494 then denote by @xmath495 the coefficient of @xmath490 in the linear combination of @xmath496 with respect to this basis . if @xmath494 and @xmath497 then @xmath6 is said to _ support _
@xmath496 .
it is not difficult to show , and i do so in the article @xcite using ingredients from the appendix to @xcite , that any iwahori - hecke algebra @xmath39 of any connected reductive affine algebraic @xmath20-group is of the form described above .
therefore , the results of this section [ sapplicationtoheckealgebras ] apply to iwahori - hecke algebras .
if greater generality is desired , one can use a pair @xmath498 of parameter systems and a `` generic algebra '' as in 7.1 of @xcite .
i will need the following slight extension of a well - known property of coxeter groups : [ lhumphreys ] fix @xmath4 and @xmath499 . if @xmath500 and @xmath501 then @xmath502 .
when @xmath212 , this is exactly lemma in 7.2 of @xcite .
the general case follows immediately from this since @xmath79 permutes @xmath62 and @xmath63 factors through @xmath120 .
denote by @xmath41 the center of the ring @xmath39 .
fix @xmath494 .
it is clear from the iwahori - matsumoto relations that @xmath503 if and only if @xmath504 and @xmath505 for all @xmath66 and @xmath191 .
fix @xmath66 . for each @xmath506
, one can use the left - handed iwahori - matsumoto relation to compute that the coefficient of @xmath507 in @xmath508 is @xmath509 similarly , one can use the right - handed iwahori - matsumoto relation to compute that the coefficient of @xmath507 in @xmath510 is @xmath511 it is obvious from the iwahori - matsumoto identities that @xmath505 if and only if @xmath512 for all @xmath506 .
it follows that the center @xmath41 is the @xmath81-subspace of vectors @xmath494 whose iwahori - matsumoto coefficients @xmath513 solve the ( infinite ) linear system consisting of the equation @xmath512 for each pair @xmath514 together with the appropriate equation from @xmath515 for each pair @xmath516 .
these equations appeared already in 3 of @xcite for affine hecke algebras on reduced root data .
recall that @xmath78 denotes the projection of @xmath4 into @xmath79 , and that the set of @xmath42-conjugacy classes in @xmath76 is denoted by @xmath74 .
fix @xmath517 and @xmath191 .
define @xmath518 to be the set of all @xmath519 such that @xmath520 if either @xmath521 or @xmath522 .
note that each @xmath518 is a _
finite_-dimensional @xmath81-subspace of @xmath41 and that @xmath41 is the _ union _ of all @xmath518 .
recall that if @xmath77 then @xmath63 is constant on @xmath523 and define @xmath524 to be this constant length .
it follows that any two @xmath525 are _ laterally - conjugate_. fix @xmath517 and @xmath191 .
define @xmath526 to be the total number of conjugacy classes @xmath77 such that @xmath85 and @xmath86 for all @xmath87 .
the following two lemmas show how lateral - conjugacy and the diamond property are related to centers of hecke algebras : [ llateralconjugacyimpliesdependence ] suppose @xmath519 . if @xmath4 is laterally - conjugate to @xmath7 then @xmath527 . if @xmath66 is such that @xmath528 then either @xmath529 or @xmath530 .
choosing @xmath531 in the former case and @xmath532 in the latter case , centrality equation ( [ ecenterequations_bothdecrease ] ) implies that @xmath533 , and the claim follows immediately from this .
[ ldiamondpropertyimpliesdependence ] suppose @xmath519 .
if @xmath4 has the diamond property then there exist @xmath534 satisfying @xmath535 such that @xmath536 is a @xmath81-linear combination of @xmath537 and @xmath538 . by lemma [ llateralconjugacyimpliesdependence ]
, i may assume that @xmath6 has the _ direct _ diamond property .
let @xmath66 be the element realizing the property . by basic coxeter theory , @xmath539 and
it is true that either @xmath540 or @xmath528 . if it were true that @xmath528 then by lemma [ lhumphreys ] it would be true that @xmath375 , but this is explicitly prohibited by the direct diamond property
therefore , @xmath541 . choosing @xmath531 and applying centrality equation ( [ ecenterequations_us_lowest ] ) proves that @xmath536 is a linear combination of @xmath542 and @xmath543 , as desired .
materially , both lemmas [ llateralconjugacyimpliesdependence ] and [ ldiamondpropertyimpliesdependence ] appeared already as lemma 3.1 of @xcite .
[ pdimensionbounds ] fix @xmath517 and @xmath191 .
@xmath88 suppose @xmath4 and @xmath207 .
consider the linear system defining @xmath41 as a @xmath81-subspace of @xmath39 . by applying the main theorem and lemma [ ldiamondpropertyimpliesdependence ] repeatedly
, one can express the variable @xmath495 as a @xmath81-linear combination of variables @xmath513 such that either @xmath544 or @xmath545 . since @xmath518 is the @xmath81-subspace of @xmath41 defined by the additional equations @xmath546 for all @xmath506 such that either @xmath544 or @xmath547 , it follows that @xmath548 is _ at most _ the total number of @xmath347 such that @xmath549 and @xmath86 . on the other hand , if @xmath77 and @xmath525 then @xmath309 is laterally - conjugate to @xmath550 and @xmath551 , so the dimension bound now follows from lemma [ llateralconjugacyimpliesdependence ] .
note one extra detail from the proof : if @xmath552 and both @xmath207 and @xmath553 then @xmath520 .
suppose that for each conjugacy class @xmath77 there is an element @xmath554 such that @xmath91 for all @xmath4 supporting @xmath90 and such that @xmath78 is the same for all @xmath4 supporting @xmath90 . if @xmath89 is linearly - independent then it is a basis for @xmath41 .
fix @xmath517 and @xmath191 .
consider only those @xmath90 for which @xmath85 and for which the uniform @xmath79-component of those @xmath6 supporting @xmath90 is @xmath555 . by hypothesis ,
the set of all such @xmath90 is a linearly - independent subset of @xmath84 vectors in the subspace @xmath518 . by proposition [ pdimensionbounds ] , this set must also _ span _ @xmath518 .
since @xmath41 is the union over all pairs @xmath556 of the subspaces @xmath518 , the claim follows .
( the dominant case ) . the blue alcove in the center is the base alcove @xmath92 and the red alcoves surrounding it constitute a full lateral - conjugacy class .
the light blue / pink cones are the unique dominant weyl chamber containing the various alcoves ( uniqueness is due to the fact that in this @xmath100 example each alcove contains only one special vertex in its closure ) . ] and all possibilities for @xmath350 ( only one in this case ) are also purple .
all other alcoves in the gallery @xmath557 are blue , in particular the alcove @xmath133 adjacent to @xmath148 .
the blue wall is @xmath141 .
the red alcoves constitute the gallery @xmath558 .
observe that the current @xmath141 is always ( one of ) the previous @xmath350 . ] and @xmath559 in the proof of the main theorem ( although in this particular example , the pair of alcoves already has the direct diamond property realized by @xmath121 so no action is necessary ) . the blue alcove is @xmath133 , the red alcove is @xmath134 , and the grey cone is @xmath471 .
the black outline is merely a visual aid . ]
is now @xmath109 and @xmath560 . in this particular example , it is possible to laterally conjugate once more to arrange @xmath561 , but this is not always possible .
the light blue / pink alcove merely represents the previous position of the blue / red alcove . ] | let @xmath0 be an iwahori - weyl group of a connected reductive group @xmath1 over a non - archimedean local field .
the subgroup @xmath2 is a finite weyl group and the subgroup @xmath3 is a finitely - generated abelian group ( possibly containing torsion ) which acts on a certain real affine space by translations .
i prove that if @xmath4 and @xmath5 then one can apply to @xmath6 a sequence of conjugations by simple reflections , each of which is length - preserving , resulting in an element @xmath7 for which there exists a simple reflection @xmath8 such that @xmath9 and @xmath10 . even for affine weyl groups , a special case of iwahori - weyl groups and also an important subclass of coxeter groups ,
this is a new fact about conjugacy classes .
further , there are implications for iwahori - hecke algebras @xmath11 of @xmath1 : one can use this fact to give dimension bounds on the `` length - filtration '' of the center @xmath12 , which can in turn be used to prove that suitable linearly - independent subsets of @xmath12 are a basis . |
consider the linear discrete ill - posed problem @xmath9 where the norm @xmath10 is the 2-norm of a vector or matrix , and @xmath3 is extremely ill conditioned with its singular values decaying to zero without a noticeable gap
. mainly arises from the discretization of the first kind fredholm integral equation @xmath11 where the kernel @xmath12 and @xmath13 are known functions , while @xmath14 is the unknown function to be sought .
if @xmath15 is non - degenerate and @xmath13 satisfies the picard condition , there exists the unique squares integrable solution @xmath14 ; see @xcite . here for brevity we assume that @xmath16 and @xmath17 belong to the same set @xmath18 with @xmath19 .
applications include image deblurring , signal processing , geophysics , computerized tomography , heat propagation , biomedical and optical imaging , groundwater modeling , and many others ; see , e.g. , @xcite .
the theory and numerical treatments of integral equations can be found in @xcite .
the right - hand side @xmath20 is noisy and assumed to be contaminated by a white noise @xmath21 , caused by measurement , modeling or discretization errors , where @xmath22 is noise - free and @xmath23 . because of the presence of noise @xmath21 and the extreme ill - conditioning of @xmath3 , the naive solution @xmath24 of bears no relation to the true solution @xmath25 , where @xmath26 denotes the moore - penrose inverse of a matrix
therefore , one has to use regularization to extract a best possible approximation to @xmath27 .
the most common regularization , in its simplest form , is the direct standard - form tikhonov regularization @xmath28 with @xmath29 the regularization parameter @xcite .
the solutions to and can be fully analyzed by the singular value decomposition ( svd ) of @xmath3 .
let @xmath30 be the svd of @xmath3 , where @xmath31 and @xmath32 are orthogonal , @xmath33 with the singular values @xmath34 assumed to be simple throughout the paper except section [ multiple ] , and the superscript @xmath35 denotes the transpose of a matrix or vector .
then @xmath36 with @xmath37 . throughout the paper
, we always assume that @xmath22 satisfies the discrete picard condition @xmath38 with some constant @xmath39 for @xmath40 arbitrarily large @xcite .
it is an analogue of the picard condition in the finite dimensional case ; see , e.g. , @xcite , @xcite , @xcite and @xcite , where it is known that this condition amounts to assuming that , on average , the fourier coefficients @xmath41 decay faster than @xmath42 and is necessary to enable regularization to compute useful approximations to it . for the ease of analysis
, we assume that the coefficients satisfy a widely used model in the literature , e.g. , ( * ? ? ?
* and 153 ) and @xcite : @xmath43 where @xmath44 is a model parameter that controls the decay of the fourier coefficients . the white noise @xmath21 has a number of attractive properties which play an important role in the regularization analysis : its covariance matrix is @xmath45 , the expected values @xmath46 and @xmath47 , and @xmath48 and @xmath49 ; see , e.g. , @xcite and @xcite .
the noise @xmath21 thus affects @xmath50 _ more or less equally_. with , relation shows that for large singular values @xmath51 is dominant relative to @xmath52 .
once @xmath53 from some @xmath54 onwards , the small singular values magnify @xmath52 , and the noise term dominates @xmath55 and must be suppressed .
the transition point @xmath56 is such that @xmath57 see @xcite and a similar description @xcite .
the @xmath58 are then divided into the @xmath56 large ones and the @xmath59 small ones .
the truncated svd ( tsvd ) method @xcite computes the tsvd regularized solutions @xmath60 it is known from @xcite and @xcite that @xmath61 is the best tsvd regularized solution to and balances the regularization and perturbation errors optimally .
the parameter @xmath5 is a regularization parameter that determines how many large svd components of @xmath3 are used to compute a regularized solution @xmath62 to .
let @xmath63 , @xmath64 and @xmath65 , and define @xmath66 .
then @xmath67 is the best rank @xmath5 approximation to @xmath3 with @xmath68 ( cf .
@xcite ) , and @xmath69 is the minimum - norm least squares solution to @xmath70 that perturbs @xmath3 to @xmath67 in .
this interpretation will be often exploited later .
the solution @xmath71 of the tikhonov regularization has a filtered svd expansion @xmath72 where the @xmath73 are called filters .
the tsvd method is a special parameter filtered method , where , in @xmath62 , we take @xmath74 and @xmath75 . the error @xmath76 can be written as the sum of the regularization and perturbation errors , and an optimal @xmath77 aims to balance these two errors and make the sum of their norms minimized @xcite .
the best possible regularized solution @xmath78 retains the @xmath56 dominant svd components and dampens the other @xmath59 small svd components as much as possible @xcite . apparently , the ability to acquire only the largest svd components of @xmath3 is fundamental in solving .
a number of parameter - choice methods have been developed for finding @xmath77 or @xmath56 , such as the discrepancy principle @xcite , the l - curve criterion , whose use goes back to miller @xcite and lawson and hanson @xcite and is termed much later and studied in detail in @xcite , and the generalized cross validation ( gcv ) @xcite ; see , e.g. , @xcite for numerous comparisons .
all parameter - choice methods aim to make @xmath79 not small for @xmath80 and @xmath81 for @xmath82 .
each of these methods has its own merits and disadvantages and no one is absolutely reliable for all ill - posed problems .
for example , some of the mentioned parameter - choice methods may fail to find accurate approximations to @xmath77 ; see @xcite for an analysis on the l - curve method and @xcite for some other parameter - choice methods .
the tsvd method is important in its own right and plays a central role in analyzing .
it and the standard - form tikhonov regularization produce very similar solutions with essentially the minimum 2-norm error , i.e. , the worst - case error @xcite ; see @xcite , @xcite , @xcite and ( * ? ? ?
* sections 4.2 and 4.4 ) .
indeed , for a linear compact equation @xmath83 including with the noisy @xmath84 and true solution @xmath85 , under the source condition that its solution @xmath86 or @xmath87 , the range of the adjoint @xmath88 of @xmath89 or that of @xmath90 , which amounts to assuming that @xmath85 or its derivative is squares integrable , the best regularized solutions by the tsvd method and the tikhonov regularization not only converge to @xmath85 as the noise tends to zero , but also are _ order optimal , i.e. , the same order as the worst - case error _ ; see , e.g. , @xcite , @xcite and @xcite .
these conclusions carries over to the discrete @xcite .
the tsvd method has been widely used as a general - purpose numerical method for small or medium sized ; see @xcite and the references therein as well as many others .
therefore , we will take @xmath61 as standard reference when assessing the regularizing effects of the iterative solvers under consideration in this paper . for large , the tsvd method and the tikhonov regularization method
are generally too demanding , and only iterative regularization methods are computationally viable .
a major class of methods has been krylov iterative solvers that project onto a sequence of low dimensional krylov subspaces and computes iterates to approximate @xmath27 ; see , e.g. , @xcite .
of krylov iterative solvers , the cgls ( or cgnr ) method , which implicitly applies the conjugate gradient ( cg ) method @xcite to the normal equations @xmath6 of , and its mathematically equivalent lsqr algorithm @xcite have been most commonly used .
the krylov solvers cgme ( or cgne ) @xcite and lsmr @xcite are also choices , which amount to the cg method applied to @xmath7 with @xmath8 and minres @xcite applied to @xmath6 , respectively .
these krylov solvers have been intensively studied and known to have regularizing effects @xcite and exhibit semi - convergence @xcite ; see also @xcite , @xcite , @xcite and @xcite : the iterates converge to @xmath27 and their norms increase steadily , and the residual norms decrease in an initial stage ; then afterwards the noise @xmath21 starts to deteriorate the iterates so that they start to diverge from @xmath27 and instead converge to @xmath55 , while their norms increase considerably and the residual norms stabilize .
if we stop at the right time , then , in principle , we have a regularization method , where the iteration number plays the role of parameter regularization .
semi - convergence is due to the fact that the projected problem starts to inherit the ill - conditioning of from some iteration onwards , and the appearance of a small singular value of the projected problem amplifies the noise considerably .
the regularizing effects of cg type methods were noticed by lanczos @xcite and were rediscovered in @xcite . based on these works and motivated by a heuristic explanation on good numerical results with very few iterations using cgls in @xcite , and realizing that such an excellent performance can only be expected
if convergence to the regular part of the solution , i.e. , @xmath61 , takes place before the effects of ill - posedness show up , on page 13 of @xcite , bjrck and eldn in 1979 foresightedly expressed a fundamental concern on cgls ( and lsqr ) : _ more research is needed to tell for which problems this approach will work , and what stopping criterion to choose .
_ see also @xcite . as remarked by hanke and hansen @xcite ,
the paper @xcite was the only extensive survey on algorithmic details until that time and was for the conference proceedings that unfortunately never appeared , and a strict proof of the regulariziing properties of conjugate gradients is extremely difficult .
an enormous effort has long been made to the study of regularizing effects of lsqr and cgls ( cf .
@xcite and many references therein ) in the hilbert or finite dimensional space setting , but a rigorous regularization theory of lsqr and cgls for is still lacking , and hitherto there has been no definitive answer to the above long - standing fundamental question , and the same is for lsmr and cgme . for @xmath3 symmetric ,
minres and mr - ii applied to directly are alternatives and have been shown to have regularizing effects @xcite , but mr - ii seems preferable since the noisy @xmath2 is excluded in the underlying subspace @xcite . for @xmath3 nonsymmetric or multiplication with @xmath91 difficult to compute , gmres and rrgmres are candidate methods @xcite , and the latter may be better @xcite .
the hybrid approaches based on the arnoldi process have been first proposed in @xcite and studied in @xcite .
gazzola and her coauthors @xcite@xcite have described a general framework of the hybrid methods and presented various krylov - tikhonov methods with different parameter choice strategies .
unfortunately , these methods are not general - purpose regularization methods , and their success is highly problem dependent ; see , e.g. , @xcite and @xcite .
the fundamental cause is that the underlying krylov subspaces generally favor a possibly dominant eigenspace of @xmath3 other than its dominant right singular subspace , which is exactly desired in solving .
the behavior of ill - posed problems critically depends on the decay rate of @xmath92 .
the following characterization of the degree of ill - posedness of was introduced in @xcite and has been widely used @xcite : if @xmath93 , then is mildly or moderately ill - posed for @xmath94 or @xmath95 .
if @xmath96 with @xmath97 , @xmath98 , then is severely ill - posed . here for mildly ill - posed problems we add the requirement @xmath99 , which does not appear in @xcite but must be met for @xmath12 in @xcite . actually , in the one - dimensional @xmath100 case , is severely ill - posed with @xmath15 sufficiently smooth , and it is moderately ill - posed with @xmath101 , where @xmath102 is the highest order of continuous derivatives of @xmath15 ; see , e.g. , @xcite and @xcite .
clearly , the singular values @xmath92 for a severely ill - posed problem decay exponentially at the same rate @xmath103 , while those of a moderately or mildly ill - posed problem decay at the decreasing rate @xmath104 that approaches one with @xmath105 and , for the same @xmath105 , is smaller for the moderately ill - posed problem than for the mildly ill - posed problem . if a regularized solution to is at least as accurate as @xmath61 , then it is called a best possible regularized solution . given , if the semi - convergence of an iterative solver means that a best possible regularized solution has been found , then , by the words of bjrck and eldn , the solver _ works _ for the problem and is claimed to have the _ full _ regularization . otherwise , the solver is said to have the _ partial _ regularization . because of the lack of solid and complete regularization theory , it has been unknown whether or not lsqr , cgls , lsmr and cgme have the full regularization for a given . to this end ,
one commonly combines them with some explicit regularization , so that the resulting hybrid variants ( hopefully ) find best possible regularized solutions @xcite .
a hybrid cgls is to run cgls for several hopefully good regularization parameters @xmath106 and picks up a best regularized solution @xcite .
its disadvantage is that regularized solutions can not be updated with different @xmath106 .
the hybrid lsqr variants have been advocated by bjrck and eldn @xcite and oleary and simmons @xcite , and improved and developed by bjrck @xcite and bjrck , grimme and van dooren @xcite . a hybrid lsqr
first projects onto krylov subspaces and then regularizes the projected problems explicitly .
it aims to remove the effects of small ritz values and expands a krylov subspace until it captures the @xmath56 dominant svd components of @xmath3 @xcite .
the hybrid lsqr and cgme have been intensively studied in , e.g. , @xcite and @xcite .
hybrid lsqr variants are sophisticated or cumbersome , and it is hard to find a near - optimal regularization parameter @xcite .
in contrast , if an iterative solver has the full regularization , it is then straightforward to stop it after semi - convergence , and it is unnecessary to introduce its hybrid variant .
obviously , we can not emphasize too much the importance of completely understanding the regularization of lsqr , cgls , lsmr and cgme . by the definition of the full or partial regularization
, we now modify the concern by bjrck and eldn as : _ do lsqr , cgls , lsmr and cgme have the full or partial regularization for severely , moderately and mildly ill - posed problems ? how to identify their full or partial regularization in practice ?
_ in this paper , assuming exact arithmetic , we first focus on lsqr and make a rigorous and complete analysis on its regularization for severely , moderately and mildly ill - posed problems . due to the mathematical equivalence of cgls and lsqr , the assertions on the full or partial regularization of lsqr
apply to cgls as well .
we then analyze the regularizing effects of lsmr and cgme and draw definitive conclusions .
we prove that lsqr has the full regularization for severely and moderately ill - posed problems with @xmath97 and @xmath95 suitably , and it generally has only the partial regularization for mildly ill - posed problems . in section [ lsqr ] , we describe the lanczos bidiagonalization process and lsqr , and make an introductory analysis . in section [ sine ] ,
we establish accurate @xmath4 theorems for the 2-norm distance between the underlying @xmath5-dimensional krylov subspace and the @xmath5-dimensional dominant right singular subspace of @xmath3 .
we then derive some follow - up results that play a central role in analyzing the regularization of lsqr . in section [ rankapp ]
, we prove that a @xmath5-step lanczos bidiagonalization always generates a near best rank @xmath5 approximation to @xmath3 , and the @xmath5 ritz values always approximate the first @xmath5 large singular values in natural order , and no small ritz value appears before iteration @xmath107 .
this will show that lsqr has the full regularization . in the meantime
, we prove that , for @xmath107 , the @xmath5 ritz values generally do not approximate the first @xmath5 large singular values in natural order and lsqr generally has only the partial regularization for mildly ill - posed problems . in section [ alphabeta ]
, we derive bounds for the entries of bidiagonal matrices generated by lanczos bidiagonalization , which prove how fast they decay and are of practical importance to reliably identify whether or not lsqr has the full regularization without extra cost when the degree of ill - posedness of is unknown in advance . exploiting some of the results on lsqr , we analyze the regularization of lsmr and cgme and prove that lsmr has similar regularizing effects to lsqr for each kind of problem and both of them are superior to cgme . in section
[ compare ] , we present some perturbation results and prove that lsqr resembles the tsvd method for severely and moderately ill - posed problems . in section [ multiple ] , with a number of nontrivial changes and reformulations ,
we extend all the results to the case that @xmath3 has multiple singular values . in section [ numer ] , we report numerical experiments to confirm our theory on lsqr .
finally , we summarize the paper with further remarks in section [ concl ] . throughout the paper , denote by @xmath108 the @xmath5-dimensional krylov subspace generated by the matrix @xmath109 and the vector @xmath110 , and by @xmath111 and the bold letter @xmath112 the identity matrix and the zero matrix with orders clear from the context , respectively . for @xmath113 , we define the nonnegative matrix @xmath114 , and for @xmath115 , @xmath116 means @xmath117 componentwise .
lsqr is based on the lanczos bidiagonalization process , which computes two orthonormal bases @xmath118 and @xmath119 of @xmath120 and @xmath121 for @xmath122 , respectively .
we describe the process as algorithm 1 .
* algorithm 1 : @xmath5-step lanczos bidiagonalization process * take @xmath123 , and define @xmath124 . for @xmath125 @xmath126 @xmath127 @xmath128 @xmath129 algorithm 1 can be written in the matrix form @xmath130 where @xmath131 denotes the @xmath132-th canonical basis vector of @xmath133 , @xmath134 , @xmath135 and @xmath136 it is known from that @xmath137 we remind that the singular values of @xmath138 are all simple , provided that lanczos bidiagonalization does not break down .
this basic fact will often be used later . at iteration @xmath5
, lsqr solves the problem @xmath139 and computes the approximate solution @xmath140 with @xmath141 where @xmath142 is the first canonical basis vector of @xmath133 , and the residual norm @xmath143 decreases monotonically with respect to @xmath5 .
we have @xmath144 and @xmath145 , both of which can be cheaply computed .
note that @xmath146 .
we have @xmath147 i.e. , the iterate @xmath148 by lsqr is the minimum - norm least squares solution to the perturbed problem that replaces @xmath3 in by its rank @xmath5 approximation @xmath149 .
recall that the best rank @xmath5 approximation @xmath67 to @xmath3 satisfies @xmath68 .
we can relate lsqr and the tsvd method from two perspectives .
one of them is to interpret lsqr as solving a nearby problem that perturbs @xmath67 to @xmath149 , provided that @xmath149 is a near best rank @xmath5 approximation to @xmath3 with an approximate accuracy @xmath150 .
the other is to interpret @xmath62 and @xmath148 as the solutions to the two perturbed problems of that replace @xmath3 by the rank @xmath5 approximations with the same quality to @xmath3 , respectively .
both perspectives lead to the consequence : the lsqr iterate @xmath151 is as accurate as @xmath61 and is thus a best possible regularized solution to , provided that @xmath149 is a near best rank @xmath5 approximation to @xmath3 with the approximate accuracy @xmath150 and the @xmath5 singular values of @xmath138 approximate the first @xmath5 large ones of @xmath3 in natural order for @xmath152 .
otherwise , as will be clear later , @xmath151 can not be as accurate as @xmath61 if either @xmath153 is not a near best rank @xmath56 approximation to @xmath3 or @xmath154 has at least one singular value smaller than @xmath155
. we will give a precise definition of a near best rank @xmath5 approximation later .
as stated in the introduction , the semi - convergence of lsqr must occur at some iteration @xmath5 . under the discrete picard condition , if semi - convergence occurs at iteration @xmath56 , we are sure that lsqr has the full regularization because @xmath151 has captured the @xmath56 dominant svd components of @xmath3 and effectively suppressed the other @xmath59 svd components ; if semi - convergence occurs at some iteration @xmath156 , then lsqr has only the partial regularization since it has not yet captured the needed @xmath56 dominant svd components of @xmath3 .
van der sluis and van der vorst @xcite prove the following result , which has been used in hansen @xcite and the references therein to illustrate and analyze the regularizing effects of lsqr and cgls .
we will also exploit it in our paper .
[ help ] for lsqr to solve with the starting vector @xmath159 and cgls to solve the normal equations @xmath6 with the starting vector @xmath160 , the iterate @xmath148 satisfies @xmath161 where @xmath162 and the @xmath163 are the singular values of @xmath138 labeled as @xmath164 .
shows that @xmath148 has a filtered svd expansion of form . if all the @xmath163 , called the ritz values , approximate the first @xmath5 singular values @xmath92 of @xmath3 in natural order , the filters @xmath165 and the other @xmath166 monotonically decay to zero with @xmath167 .
if this is the case until @xmath168 , the @xmath56-step lsqr has the full regularization and computes a best possible regularized solution @xmath151 .
however , if a small ritz value appears before some @xmath107 , i.e. , @xmath169 and @xmath170 with the smallest integer @xmath171 , then @xmath172 tends to zero monotonically with respect to @xmath173 ; on the other hand , we have @xmath174 since the first factor is non - positive and the second factor is positive .
then we get @xmath175 , causing that @xmath148 is deteriorated and lsqr has only the partial regularization .
hansen @xcite summarizes the known results on @xmath166 , where a complicated bound for @xmath176 is given in @xcite but there is no way of estimating the bound itself accurately . as we will see in section [ compare ] , the results to be established in this paper can be used for this purpose , and , more importantly , we will show that the bound there can be sharpened substantially .
the standard @xmath5-step lanczos bidiagonalization method computes the @xmath5 ritz values @xmath163 , which are used to approximate some of the singular values of @xmath3 , and it is mathematically equivalent to the symmetric lanczos method for the eigenvalue problem of @xmath177 starting with @xmath178 ; see @xcite or @xcite for several variations that are based on standard , harmonic , refined projection @xcite or a combination of them .
a general convergence theory of harmonic and refined harmonic projection methods was lacking in the books @xcite and has later been established in @xcite . as is known from @xcite , for a general singular value distribution and a general vector @xmath2 , some of the @xmath5 ritz values become good approximations to the largest and smallest singular values of @xmath3 as @xmath5 increases . if large singular values are well separated but small singular values are clustered , large ritz values converge fast but small ritz values converge very slowly . for
, we see from that @xmath179 contains more information on dominant right singular vectors than on the ones corresponding to small singular values .
therefore , @xmath157 hopefully contains richer information on the first @xmath5 right singular vectors @xmath180 than on the other @xmath181 ones , at least for @xmath5 small .
furthermore , note that @xmath3 has many small singular values clustered at zero . due to these two basic facts ,
all the ritz values are expected to approximate the large singular values of @xmath3 in natural order until some iteration @xmath5 , at which a small ritz value shows up and the regularized solutions then start to be contaminated by the noise @xmath21 dramatically after that iteration .
these qualitative arguments are frequently used to analyze and elaborate the regularizing effects of lsqr and cgls ; see , e.g. , @xcite and the references therein . clearly , these arguments , though helpful to show that lsqr and cgls have regularizing effects , are uncertain , and they are useless to draw any definitive conclusion on the full or partial regularization of lsqr . for a severely ill - posed example from seismic tomography , it is reported in @xcite that the desired convergence of the ritz values actually holds as long as the discrete picard condition is satisfied and there is a good separation among the large singular values of @xmath3 .
unfortunately , there has been no mathematical justification on these observations .
rigorously and quantitatively speaking , a complete understanding of the regularization of lsqr includes accurate solutions of the following basic problems : how well or accurately does @xmath182 approximate or capture the @xmath5-dimensional dominant right singular subspace of @xmath3 ?
how accurate is the rank @xmath5 approximation @xmath183 to @xmath3 ? can it be a near best rank @xmath5 approximation to @xmath3 ? how does the noise level @xmath184 affects the approximation accuracy of @xmath182 and @xmath183 for @xmath107 and @xmath185 , respectively ?
what sufficient conditions on @xmath186 and @xmath187 are needed to guarantee that @xmath183 is a near best rank @xmath5 approximation to @xmath3 ? when do the ritz values @xmath188 approximate @xmath189 in natural order ?
when does at least a small ritz value appear , i.e. , @xmath190 before some @xmath107
? we will make a rigorous and detailed analysis on these problems and some others related closely , present our results , and draw definitive assertions on the regularization of lsqr for three kinds of ill - posed problems . in terms of the canonical angles @xmath191 between two subspaces @xmath192 and @xmath193 of the same dimension @xcite , we first present the following @xmath4 theorem , showing how the @xmath5-dimensional krylov subspace @xmath182 captures or approximates the @xmath5-dimensional dominant right singular subspace of @xmath3 for severely ill - posed problems .
[ thm2 ] let the svd of @xmath3 be as .
assume that is severely ill - posed with @xmath96 and @xmath97 , @xmath98 , and the discrete picard condition is satisfied .
let @xmath194 be the subspace spanned by the columns of @xmath195 and @xmath196 .
then for @xmath197 we have @xmath198 with @xmath199 to be defined by and @xmath200 @xmath201 where @xmath202 whose columns are the first @xmath40 left singular vectors of @xmath3 defined by . then the krylov subspace @xmath203 with @xmath204 partition the diagonal matrix @xmath205 and the matrix @xmath206 as follows : @xmath207 where @xmath208 .
since @xmath209 is a vandermonde matrix with @xmath92 supposed to be distinct for @xmath125 , it is nonsingular .
therefore , from @xmath210 we have @xmath211 where @xmath212 write @xmath213 , and define @xmath214 then @xmath215 , and the columns of @xmath216 form an orthonormal basis of @xmath217 .
so we get an orthogonal direct sum decomposition of @xmath218 : @xmath219 by definition and , for the matrix 2-norm we obtain @xmath220 which is .
next we estimate @xmath221 .
for @xmath222 , it is easily justified that the @xmath105-th column of @xmath223 consists of the coefficients of the @xmath105-th lagrange polynomial @xmath224 that interpolates the elements of the @xmath105-th canonical basis vector @xmath225 at the abscissas @xmath226 .
consequently , the @xmath105-th column of @xmath227 is @xmath228 from which we obtain @xmath229 since @xmath230 is monotonically decreasing for @xmath231 , it is bounded by @xmath232 . with this property and the definition of @xmath233 by
, we get @xmath234 where @xmath235 is a rank one matrix . therefore , by @xmath236 ( cf .
@xcite ) , we get @xmath237 by the discrete picard condition , and the description between them , for the white noise @xmath21 , it is known from @xcite and @xcite that @xmath238 decrease as @xmath105 increases up to @xmath56 and then become stabilized as @xmath238 a constant for @xmath239 . in order to simplify the derivation and
present our results compactly , in terms of these assumptions and properties , in later proofs we will use the precise relationships @xmath240 , for @xmath197 we obtain @xmath241 with @xmath242 replaced by one for @xmath243 . in a similar manner , for @xmath222 ,
from we get @xmath244 from the above and , we finally obtain @xmath245 which proves .
note that the langrange polynomials @xmath246 require @xmath247 .
so , we need to treat the case @xmath248 independently : from and , observe that @xmath249 therefore , we have @xmath250 from which and for @xmath248 it is direct to get . in terms of the discrete picard condition , , and , we have @xmath251 and @xmath252 applying them to and establishes , and , respectively .
we next estimate the factor @xmath253 accurately .
[ estlk ] for the severely ill - posed problem and @xmath222 , we have @xmath254 _ proof_. exploiting the taylor series expansion and @xmath255 for @xmath256 , by definition , for @xmath257 we have @xmath258 by absorbing those higher order terms into the two @xmath259 in the numerator . for @xmath260 , we get @xmath261 which is
. note that for the numerator of we have @xmath262 and @xmath263 whose product for any @xmath5 is @xmath264 on the other hand , note that the denominator of is defined by @xmath265 which , together with the above estimate for the numerator of , proves .
notice that @xmath266 is always _
bigger than one _ for @xmath257 .
therefore , for any @xmath5 , combining and gives .
[ severerem ] and have essentially been shown in @xcite . here
we have given a general and complete proof . from , we get @xmath267 so the results in theorem [ thm2 ] are simplified as @xmath268 and its proof illustrate that @xmath232 exhibits monotonically increasing tendency with respect to @xmath105 . and indicate that @xmath217 captures @xmath269 better for @xmath107 than for @xmath185 .
that is , after iteration @xmath56 , the noise @xmath21 starts to deteriorate @xmath217 and impair its ability to capture @xmath269 . in what follows
we establish accurate estimates for @xmath270 for moderately and mildly ill - posed problems .
[ moderate ] assume that is moderately ill - posed with @xmath271 , where @xmath99 and @xmath272 is some positive constant , and other assumptions and notation the same as in theorem [ thm2 ] . then holds with @xmath273
particularly , we have @xmath274 _ proof_. following the proof of theorem [ thm2 ] , we know that @xmath275 defined by .
so we only need to bound the right - hand side of . for @xmath276 , from we get @xmath277 since the function @xmath278 with any @xmath279 is convex over the interval @xmath280 $ ] , for @xmath281 , from we obtain @xmath282 substituting the above and into establishes , from which and , it follows that and hold . for @xmath248
, we still have , from which and we obtain . from and we get . for a purely technical reason and precise presentation and for the ease of analysis
, we have used the simplified singular value model @xmath283 to replace the general form @xmath93 , where the constant in each @xmath259 is qualitative and implicit .
this model , though simple , reflects the essence of moderately and mildly ill - posed problems and avoids some troublesome derivations and non - transparent formulations . in the spirit of the proof of theorem [ estlk ] , exploiting the first order taylor expansion , we have an estimate @xmath284 where the right - hand side of increases linearly with respect to @xmath5 . [ rem3.5 ] and indicate that @xmath285 is guaranteed for moderately ill - posed problems with @xmath95 .
one may worry that the upper bounds and overestimate @xmath221 and thus @xmath270 because , in the proof , we have bounded the opaque @xmath286 in from above by the compact integral nearest to it , which can be overestimates for @xmath152 .
it is not the case only if @xmath56 is not very small .
in fact , since @xmath99 , we can bound from below by the integral nearest to it : @xmath287 which is near to once @xmath107 is not very small
. the smaller @xmath187 , the smaller the difference between the upper and lower bounds , i.e. , the sharper .
it is easily seen from that @xmath270 increases monotonically with respect to @xmath221 . for @xmath221
reasonably small and @xmath221 large we have @xmath288 respectively . from and , we obtain @xmath289 , where @xmath290 is the gaussian function .
for the white noise @xmath21 , we have @xmath291 . as a result , for moderately ill - posed problems with @xmath95
, @xmath56 is typically small and at most modest for a practical noise @xmath21 , whose relative size @xmath292 typically ranges from @xmath293 to @xmath294 .
this means that for this kind of problem @xmath221 is at most modest and can not be large , so that @xmath295 fairly . for severely ill - posed problems , since all the @xmath296 , a constant , and indicate that @xmath270 is essentially unchanged for @xmath152 and @xmath297 , respectively , that is , @xmath217 captures @xmath269 with almost the same accuracy for @xmath107 and @xmath185 , respectively .
however , the situation is different for moderately ill - posed problems . for them
, @xmath298 increases slowly as @xmath5 increases , and the factor latexmath:[$\sqrt{\frac{k^2}{4\alpha^2 - 1}+\frac{k}{2\alpha-1 } } illustrate that @xmath270 increases slowly with @xmath107 and @xmath300 , respectively .
this means that @xmath217 may not capture @xmath269 so well as it does for severely ill - posed problems as @xmath5 increases . in particular , starting with some @xmath185 , @xmath270 starts to approach one , which indicates that , for @xmath5 big , @xmath217 will contain substantial information on the right singular vectors corresponding to the @xmath181 small singular values of @xmath3 .
[ mildrem ] for mildly ill - posed problems with @xmath301 , there are some distinctive features .
note from and that @xmath56 is now considerably bigger than that for a severely or moderately ill - posed problem with the same noise level @xmath184 and @xmath44 . as a result ,
firstly , for @xmath302 and the same @xmath5 , the factor @xmath303 is bigger than that for the moderately ill - posed problem ; secondly , @xmath304 if @xmath305 and is much bigger than @xmath5 and can be arbitrarily large if @xmath306 ; thirdly , since @xmath307 , for @xmath308 that ensures @xmath309 , we have @xmath310 which also holds for moderately ill - posed problems and is bigger than one considerably for @xmath307 as @xmath5 increases up to @xmath56 .
our accurate bound thus becomes increasingly large as @xmath5 increases up to @xmath56 for mildly ill - posed problems , causing that @xmath221 is large and @xmath311 starting with some @xmath107 .
consequently , @xmath312 can not effectively capture the @xmath56 dominant right singular vectors and contains substantial information on the right singular vectors corresponding to the @xmath59 small singular values . in ( * ?
* thm 2.1 ) , the authors have derived some rough bounds for @xmath221 and thus @xmath270 . there , without realizing the crucial fact that @xmath313 can be effectively bounded by a rank one matrix and the key point that @xmath314 must be treated as a whole other than separately , by the authors made use of @xmath315 and @xmath316 ( cf . ) to obtain bounds for @xmath221 and thus @xmath270 .
these bounds are too pessimistic because of the appearance of the fatal factor @xmath317 , which ranges from @xmath318 to @xmath319 for @xmath222 , too large amplification for @xmath40 large . in contrast , our new estimates , which hold for both @xmath221 and @xmath320 , are much more accurate and @xmath317 has been removed . before proceeding ,
we tentatively investigate how @xmath270 affects the smallest ritz value @xmath321 .
this problem is of central importance for understanding the regularizing effects of lsqr .
we aim to lead the reader to a first manifestation that ( i ) we may have @xmath322 when @xmath295 fairly , that is , no small ritz value may appear only if @xmath217 captures @xmath269 with only _ some _ other than high accuracy , and ( ii ) we must have @xmath323 , that is , @xmath321 can not approximate @xmath58 in natural order , once @xmath270 is sufficiently close to one . [ initial ]
let @xmath324 with @xmath325 , @xmath197 , and let the unit - length @xmath326 be a vector that has the smallest acute angle with @xmath327 , i.e. , the closest to @xmath327 , where @xmath328 is the matrix consisting of the last @xmath181 columns of @xmath329 defined by .
then it holds that @xmath330 if @xmath331 , then @xmath332 if @xmath333 for a given arbitrarily small @xmath334 , then @xmath335 _ proof_. since the columns of @xmath336 generated by lanczos bidiagonalization form an orthonormal basis of @xmath217 , by definition and the assumption on @xmath337 we have @xmath338 with @xmath339 and @xmath340 . since @xmath269 is the orthogonal complement of @xmath327 , by definition we know that @xmath341 has the largest acute angle with @xmath269 , that is , it contains the least information from @xmath269 .
expand @xmath337 as the following orthogonal direct sum decomposition : @xmath342 then from @xmath343 and we obtain @xmath344 from , we next bound the rayleigh quotient of @xmath337 with respect to @xmath177 from below . by the svd of @xmath3 , we partition @xmath345 and @xmath346 where @xmath347 and @xmath348 . making use of @xmath349 and @xmath350 as well as @xmath351
, we obtain @xmath352 observe that it is impossible for @xmath353 and @xmath354 to be the right singular vectors of @xmath355 and @xmath356 associated with the smallest singular values @xmath357 and @xmath358 of @xmath359 and @xmath360 simultaneously , which are the @xmath361-th canonical vector @xmath362 of @xmath363 and the @xmath5-th canonical vector @xmath364 of @xmath365 , respectively ; otherwise , we have @xmath366 and @xmath367 simultaneously , which are impossible as @xmath368 .
therefore , from we obtain the strict inequality @xmath369 from which it follows that the lower bound of holds .
similarly , from we obtain the upper bound of : @xmath370 from the lower bound of , we see that if @xmath371 satisfies @xmath372 , i.e. , @xmath331 , then @xmath373 , i.e. , holds . from and
, we get @xmath374 .
note that @xmath375 is the smallest singular value , i.e. , eigenvalue , of the symmetric positive definite matrix @xmath376 .
therefore , we have @xmath377 where @xmath378 is , in fact , the ritz vector of @xmath177 from @xmath217 corresponding to the smallest ritz value @xmath375 .
therefore , for @xmath337 defined in theorem [ initial ] we have @xmath379 from which it follows from that @xmath380 . as a result , for any @xmath334
, we can choose @xmath381 such that @xmath382 i.e. , holds , solving which for @xmath383 gives @xmath333 .
we analyze @xmath321 when @xmath331 .
a key observation and interpretation is that , in the sense of @xmath384 in , @xmath385 is the optimal vector that extracts the least information from @xmath269 and the richest information from @xmath327 . from theorem [ initial ] , since @xmath269 is the orthogonal complement of @xmath327 , we know that @xmath341 has the largest acute angle with @xmath269 , that is , it contains the least information from @xmath269 and the richest information from @xmath327 .
therefore , @xmath378 and @xmath337 have a similar optimality , so that we have @xmath386 combining this estimate with , we may have @xmath322 when @xmath331 .
we inspect the condition @xmath387 for and get insight into whether or not the true @xmath371 resulting from three kinds of ill - posed problems satisfies it . for severely ill - posed problems ,
this lower bound is basically the constant @xmath103 ; for moderately ill - posed problems with @xmath95 , the bound increases with increasing @xmath107 , and it can not be close to one provided that @xmath95 suitably or @xmath56 not big ; for mildly ill - posed problems with @xmath388 , the bound increases faster than it does for moderately ill - posed problems , and it may well approach one for @xmath107 . therefore , means that @xmath270 is not close to one for severely and moderately ill - posed problems , but it requires that @xmath270 must be close to zero for mildly ill - posed problems . in view of and @xmath324
, we have @xmath389 .
thus , the condition @xmath387 for amounts to requiring that @xmath221 is at most modest and can not be large for severely and moderately ill - posed problems but it must be fairly small for mildly ill - posed problems .
unfortunately , theorems [ thm2][moderate ] and the remarks followed indicate that @xmath221 increases with increasing @xmath5 and is generally large for a mildly ill - posed problem , while it increases slowly with @xmath107 for a moderately ill - posed problem with @xmath95 suitably , and by it is approximately a constant @xmath390 considerably for a severely ill - posed problem with @xmath97 not close to one .
consequently , for mildly ill - posed problems , because the actual @xmath221 can hardly be small and is generally large , the true @xmath371 is small and may well be close to one , so that the condition @xmath391 generally fails to meet as @xmath5 increases , while it is satisfied for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably .
[ appear ] shows that there is at least one ritz value @xmath323 when @xmath270 is sufficiently close to one since we can choose @xmath392 small enough such that @xmath393 is close to @xmath150 arbitrarily .
as we have shown , @xmath270 can not be close to one for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably , but it is generally so for mildly ill - posed problems .
this means that for some @xmath107 it is very possible to have @xmath323 for mildly ill - posed problems .
however , we must be aware that our above analysis on @xmath322 is not rigorous because we can not quantify _ how small _ @xmath394 is . from @xmath395 , it is apparent that the condition @xmath391 may not be sufficient for @xmath322 though it is so for @xmath373 .
we delay our detailed and rigorous analysis to section [ rankapp ] , where we present a number of deep - going and accurate results on the key problems stated in the last second paragraph before theorem [ thm2 ] , including the precise behavior of @xmath321 .
one of the results will be on the sufficient conditions for @xmath322 , which are satisfied when certain deterministic and mild restrictions on @xmath186 or @xmath187 are imposed for severely or moderately ill - posed problems .
however , we will see that @xmath388 for mildly ill - posed problems never meets the sufficient conditions to be presented there .
theorems [ thm2][moderate ] establish necessary background for answering the fundamental concern by bjrck and eldn , and their proof approaches also provide key ingredients for some of the later results .
we next present the following results , which will play a central role in our later analysis .
[ thm3 ] assume that the dicrete picard condition is satisfied , let @xmath396 be defined as and @xmath397 be defined as , and write @xmath398 . then for severely ill - posed problems and @xmath197 we have @xmath399 and @xmath400 for moderately or mild ill - posed problems with the singular values @xmath283 and @xmath272 a positive constant we have @xmath401 and @xmath402 _ proof_. from and , for @xmath125 and @xmath403 we have @xmath404 and from , for @xmath248 we have @xmath405 therefore , for severely ill - posed problems , @xmath197 and @xmath125 , from we obtain @xmath406 for moderately or mildly ill - posed problems , @xmath197 and @xmath125 , from we obtain @xmath407 combining the above with , and , we obtain , while follows from the above and directly . for @xmath248 , from and the above we get and , respectively . by , for @xmath403 we have @xmath408 therefore ,
we get @xmath409 by for @xmath248 we have @xmath410 we have derived the bounds and for @xmath411 for severely and moderately or mildly ill - posed problems , respectively , from which we obtain and for @xmath248 . in order to bound @xmath412 for @xmath403
, we also need to estimate @xmath413 .
we next carry out this task for severely and moderately ill - posed problems , respectively , for each kind of which we study @xmath107 and @xmath185 separately .
case of @xmath107 for severely ill - posed problems : from the discrete picard condition and , we obtain @xmath414 case of @xmath300 for severely ill - posed problems : from above and , , we obtain @xmath415 substituting the above two relations for the two cases into and combining them with and , we get .
case of @xmath107 for moderately or mildly ill - posed problems : from we have @xmath416 case of @xmath300 for moderately or mildly ill - posed problems : from and we have @xmath417 substituting the above two bounds for the two cases into and combining them with , we get . and
indicate that @xmath412 decays swiftly as @xmath5 increases .
as has been seen , we must take some cares to accurately bound @xmath412 .
indeed , for @xmath418 , if we had simply bounded it by @xmath419 the factors @xmath150 in and @xmath58 in would have become @xmath420 and @xmath421 , fixed constants , by substituting the estimates and for @xmath221 into the above .
this way will overestimate @xmath412 too much as @xmath5 increases , and the resulting bounds for @xmath412 are useless to precisely analyze the regularization of lsqr , cgme and lsmr for ill - posed problems since they make us impossible to get those predictively accurate results which will be presented in section [ rankapp]section [ compare ] . as a byproduct
, we consider an interesting problem that has its own right , though its solution will not be used in this paper : how close to the krylov subspace @xmath217 is the individual right singular vector @xmath422 for @xmath423 and @xmath197 ?
denote by @xmath424 the distance between @xmath422 and @xmath217 , which is defined as @xmath425 with @xmath426 the orthogonal projector onto @xmath217 .
then we present the following result .
[ indiv ] let @xmath398 be defined by .
then for @xmath197 and @xmath125 we have @xmath427 where @xmath428 denotes the smallest singular value of a matrix . _
proof_. we first prove the upper bound of . since the columns of @xmath429 defined by form a basis of @xmath217 , its @xmath105-th column @xmath430 . as a result
, we get @xmath431 recall from that the columns of @xmath218 form an orthonormal basis of @xmath217 , and suppose that @xmath432 is orthogonal . then the columns of @xmath433 are an orthonormal basis of the orthogonal complement of @xmath217 with respect to @xmath434 .
particularly , @xmath435 meets the requirement . by definition ,
we obtain @xmath436 from which and @xmath437 it follows that @xmath438 by taking @xmath439 .
so the upper bound of holds .
we next derive the lower bound of .
we obtain from above that @xmath440 we remark that the lower bound in is just the sine of the smallest canonical angle of @xmath269 and @xmath217 . since @xmath441 , it is natural that @xmath442 lies between the smallest and largest angles of @xmath269 and @xmath217 , as indicates . combining with and
, we see that the smaller @xmath105 , the closer @xmath422 is to @xmath217 .
meanwhile , for fixed @xmath5 , since , from , the bounds and for @xmath443 tend to increase monotonically with respect to @xmath105 , indicates that the subspace @xmath444 is as close to @xmath217 as the last column @xmath422 of @xmath445 .
making use of theorems [ thm2][thm3 ] , we are able to solve those key problems stated before theorem [ thm2 ] and give definitive answers to the fundamental concern by bjrck and eldn , proving that lsqr has the full regularization for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably and it , in general , has only the partial regularization for mildly ill - posed problems . define @xmath446 which measures the accuracy of the rank @xmath5 approximation @xmath183 generated by lanczos bidiagonalization to @xmath3 .
recall and the comments followed .
it is known that the full or partial regularization of lsqr uniquely depends on whether or not @xmath447 holds , where the precise meaning ` @xmath448 ' will be clear by introducing the definition of near best rank @xmath5 approximation to @xmath3 , and on whether or not the @xmath5 singular values @xmath188 ,
i.e. , ritz values , of @xmath138 , approximate the @xmath5 large singular values @xmath42 of @xmath3 in natural order for @xmath449 .
if both of them hold , lsqr has the full regularization ; if either of them is not satisfied , lsqr has only the partial regularization .
this section consists of three subsections . in section [ rankaccur ] ,
we present accurate estimates for @xmath450 for three kinds of ill - posed problems under consideration .
we prove that , under some reasonable conditions on @xmath186 or @xmath187 , the matrix @xmath183 is a near best rank @xmath5 approximation to @xmath3 . in section [ ritzapprox ] , we deepen the results in section [ rankaccur ] and show how the @xmath5 ritz values @xmath188 behave .
we derive the sufficient conditions on @xmath186 and @xmath187 for which they approximate the first @xmath5 large singular values @xmath42 of @xmath3 in natural order . in section [ morerank ] ,
we consider general best and near best rank approximations to @xmath3 with respect to the 2-norm . for @xmath3 with @xmath451
, we analyze the nonzero singular values of such a rank @xmath5 approximation , and prove that they approximate the first @xmath5 large singular values of @xmath3 for @xmath95 suitably but can fail to do so for @xmath452 .
these results will help understand the regularization of lsqr .
we first present one of the main results in this paper .
[ main1 ] assume that the discrete picard condition is satisfied .
then for @xmath197 we have @xmath453 with @xmath454 for severely ill - posed problems and @xmath455 for moderately or mildly ill - posed problems with @xmath456 , where @xmath457 for @xmath458 and @xmath459 for @xmath460 with @xmath461 defined by . _
proof_. since @xmath67 is the best rank @xmath5 approximation to @xmath3 for the 2-norm and @xmath68 , the lower bound in holds .
next we prove the upper bounds . from
, we obtain @xmath462 from algorithm 1 , , and , we obtain @xmath463 with @xmath336 and @xmath218 being orthonormal , and the orthogonal projector onto @xmath217 is thus @xmath464 keep in mind that @xmath66 .
it is direct to justify that @xmath465 .
therefore , exploiting this and noting that @xmath466 and @xmath351 , we get from , and that @xmath467 where the last inequality follows by using @xmath468 and the definition of the induced matrix 2-norm to amplify the second term in .
we estimate @xmath371 accurately below . to this end
, we need to use two key identities and some results related . by the svd of @xmath461 ,
it is direct to justify that @xmath469 and @xmath470 define the function @xmath471 with @xmath472 . since the derivative @xmath473
, @xmath474 is monotonically increasing for @xmath475 $ ] and decreasing for @xmath476 , and the maximum of @xmath474 over @xmath472 is @xmath477 , which attains at @xmath478 .
based on these properties and exploiting the svd of @xmath461 , for the matrix 2-norm we get @xmath479 for @xmath458 and @xmath480 for @xmath460 ( note : in this case , since @xmath461 may have at least one singular value smaller than one when @xmath460 , we do not have an expression like ) .
it then follows from , , and @xmath481 that @xmath482 for @xmath458 and @xmath483 for @xmath460 .
replace @xmath412 by its bounds and in the above , and insert the resulting bounds for @xmath484 into .
let @xmath485 .
then we obtain with @xmath486 satisfying and for severely and moderately or mildly ill - posed problems , respectively .
note from that @xmath487 therefore , in @xmath486 of , for @xmath107 we have @xmath488 [ decayrate ] for severely ill - posed problems , from , and the definition of @xmath489 we know that @xmath490 for both @xmath107 and @xmath185 .
from and , for @xmath107 we have @xmath491 by ignoring the smaller term @xmath492 , and for @xmath185 we have @xmath493 which increases slowly with @xmath5 . for the moderately or mildly ill - posed problems with @xmath283 , from the derivation on @xmath486 and its estimate , by comparing and with , for @xmath107 we approximately have @xmath494 and for @xmath185 , from and we approximately have @xmath495 which increases faster than the right - hand side of with respect to @xmath5 . [ decayrate2 ] from , and , for severely ill - posed problems we have @xmath496 and
@xmath450 is an accurate approximation to @xmath150 for @xmath107 and marginally less accurate for @xmath185 .
thus , the rank @xmath5 approximation @xmath183 is as accurate as the best rank @xmath5 approximation @xmath67 within the factor @xmath497 for @xmath107 and @xmath97 suitably . for
moderately ill - posed problems , @xmath450 is still an excellent approximation to @xmath150 , and the rank @xmath5 approximation @xmath183 is almost as accurate as the best rank @xmath5 approximation @xmath67 for @xmath107 .
therefore , @xmath183 plays the same role as @xmath67 for these two kinds of ill - posed problems and @xmath107 , it is known from the clarification in section [ lsqr ] that lsqr may have the full regularization .
we will , afterwards , deepen this theorem and derive more results , proving that lsqr must have the full regularization for these two kinds of problems provided that @xmath97 and @xmath95 suitably . for both severely and moderately ill - posed problems , we note that the situation is not so satisfying for increasing @xmath185 .
but at that time , a possibly big @xmath486 does not do harm to our regularization purpose , since lsqr has already found a best possible regularized solution at iteration @xmath56 and we simply stop performing it . [ mildre ] for mildly ill - posed problems , the situation is fundamentally different . as clarified in remark [ mildrem ] , we have @xmath498 and @xmath499 considerably as @xmath5 increases up to @xmath56 because of @xmath307 , leading to @xmath500 substantially .
this means that @xmath501 is substantially bigger than @xmath155 and can well lie between @xmath502 and @xmath503 , so that the rank @xmath56 approximation @xmath504 is much less accurate than the best rank @xmath56 approximation @xmath505 and lsqr has only the partial regularization . for a given ill - posed problem
, the noise level @xmath184 only affects @xmath56 but has no effect on the overall decay rate of @xmath450 .
there are several subtleties in the proof of this theorem , each of which turns out to be absolutely necessary .
ignoring or missing any one of them would be fatal and make us fail to obtain accurate estimates for @xmath484 defined by : the first is the careful treatment of @xmath506 . by the definition of @xmath270 , if we had amplified it by @xmath507 we would have obtained a too large overestimate , which is almost a fixed constant for severely ill - posed problems and @xmath152 and increases with @xmath152 for moderately and mildly ill - posed problems .
such rough estimates are useless to get a meaningful bound for @xmath450 .
the key is to treat @xmath508 as a whole in order to bound it accurately .
the second is use of and .
the third is the extraction of @xmath412 from as a whole , and one must not amplify it to @xmath509 . the fourth is accurate estimates for it ; see and in theorem [ thm3 ] and the comment on its fatal overestimate .
for example , without using and , we would have no way but to obtain @xmath510 from , and the previous estimates for @xmath221 , such bound is too pessimistic and completely useless in our context , and it even does not decrease and could not be small as @xmath5 increases , while our new estimates for @xmath485 in theorem [ main1 ] are much more accurate and decay swiftly as @xmath5 increases , as indicated by and . to make things clearer and
rigorously prove the full or partial regularization of lsqr for , we need to introduce a precise definition of the near best rank @xmath5 approximation @xmath183 to @xmath3 , i.e. , the precise meaning of @xmath447 . by definition ,
we call @xmath183 a near best rank @xmath5 approximation to @xmath3 if it satisfies @xmath511 that is , @xmath450 lies between @xmath58 and @xmath150 and is closer to @xmath150 . for an ill - posed problem ,
since there is no considerable gap of @xmath58 and @xmath150 , the definition means that @xmath450 must approximate @xmath150 more accurately as @xmath5 increases .
we mention in passing that a near best rank @xmath5 approximation to @xmath3 from an ill - posed problem is much more stringent than it is for a matrix from a ( numerically ) rank - deficient problem where the large singular values are very well separated from the small ones and there is a substantial gap between two groups of singular values .
it is naturally deduced that obtaining a near best @xmath5 rank approximation to @xmath3 from the ill - posed problem is much harder than doing so for the ( numerical ) rank deficient matrix . based on theorem
[ main1 ] , for the severely and moderately or mildly ill - posed problems with the singular value models @xmath512 and @xmath513 , we next derive the sufficient conditions on @xmath186 and @xmath187 that guarantee that @xmath183 is a near best rank @xmath5 approximation to @xmath3 for @xmath152 .
we also analyze if and how the sufficient conditions are satisfied for three kinds of ill - posed problems .
[ nearapprox ] for a given , assume that the discrete picard condition is satisfied .
then , in the sense of , @xmath183 is a near best rank @xmath5 approximation to @xmath3 for @xmath152 if @xmath186 or @xmath187 satisfies @xmath514 for the severely ill - posed problems with @xmath512 and the moderately or mildly ill - posed problems with @xmath513 , @xmath183 is a near best rank @xmath5 approximation to @xmath3 for @xmath152 if @xmath515 and @xmath187 satisfies @xmath516 _
proof_. by , we see that @xmath517 .
therefore , @xmath183 is a near best rank @xmath5 approximation to @xmath3 in the sense of provided that @xmath518 and @xmath519 from which follows . from , for the severely ill - posed problems with @xmath512 and @xmath515 we have @xmath520 from which it follows that @xmath521 since @xmath522 , holds provided that @xmath523 i.e. , @xmath524 , solving which for @xmath186 we get @xmath515 . for the moderately or mildly ill - posed problems with @xmath513 , it is direct from to get @xmath525 since @xmath526 decreases monotonically as @xmath5 increases , its minimum over @xmath152 is @xmath527 . therefore , we obtain from the above . given the noise level @xmath184 , the discrete picard condition and , from the bound for @xmath528 , we see that the bigger @xmath95 is , the smaller @xmath56 and @xmath486 are . therefore , there must be @xmath95 such that holds . here
we should remind that it is more suitable to regard the conditions on @xmath186 and @xmath187 as an indication that @xmath186 and @xmath187 must not be close to one other than precise requirements since we have used the bigger and simplified models @xmath529 and @xmath513 .
for the mildly ill - posed problems with @xmath513 , theorem [ moderate ] has shown that @xmath221 is generally not small and can be arbitrarily large for @xmath152 . from , we see that @xmath486 has comparable size to @xmath221 .
note that the right - hand side @xmath530 of for @xmath152 and @xmath307 .
consequently , can not be met generally for mildly ill - posed problems .
the rare possible exceptions are that @xmath56 is only very few and @xmath187 is close to one .
so , @xmath183 is generally not a near best rank @xmath5 approximation to @xmath3 for @xmath449 for this kind of problem . in what follows , starting with theorem [ main1 ]
, we prove that , under certain sufficient conditions on @xmath186 and @xmath187 for the severely and moderately ill - posed problems with the models @xmath531 and @xmath532 , respectively , the @xmath5 ritz values @xmath188 approximate the first @xmath5 large singular values @xmath42 in natural order for @xmath152 and no small ritz value appears . combining this with theorem [ nearapprox ]
, we draw the definitive conclusion that lsqr must have the full regularization for these two kinds of problems provided that @xmath97 and @xmath95 suitably .
[ ritzvalue ] assume that is severely ill - posed with @xmath531 and @xmath97 or moderately ill - posed with @xmath532 and @xmath95 , and the discrete picard condition is satisfied .
let the ritz values @xmath188 be labeled as @xmath533 .
then @xmath534 if @xmath535 or @xmath95 satisfies @xmath536 then the @xmath188 strictly interlace the first large @xmath537 singular values of @xmath3 and approximate the first @xmath5 large ones in natural order for @xmath152 : @xmath538 meaning that no ritz value smaller than @xmath155 appears for @xmath107 .
_ proof_. note that for @xmath152 the @xmath539 are just the nonzero singular values of @xmath183 , whose other @xmath181 singular values are zeros .
we write @xmath540 the sum of @xmath3 and @xmath541 with @xmath542 by definition . then by the mirsky s theorem of singular values ( * ? ? ?
* , thm 4.11 ) , we have @xmath543 since the singular values of @xmath3 are simple and @xmath2 has components in all the left singular vectors @xmath544 of @xmath3 , lanczos bidiagonalization , i.e. , algorithm 1 , can be run to completion , producing @xmath545 and the lower bidiagonal @xmath546 such that @xmath547 with the @xmath548 matrix @xmath549 and @xmath550 matrix @xmath551 orthogonal and all the @xmath552 and @xmath553 , @xmath256 , of @xmath554 being positive .
note that the singular values of @xmath555 are all simple and that @xmath138 consists of the first @xmath5 columns of @xmath554 with the last @xmath181 _ zero _ rows deleted .
applying the cauchy s _ strict _ interlacing theorem ( * ? ? ?
* , corollary 4.4 ) to the singular values of @xmath138 and @xmath554 , we have @xmath556 therefore , becomes @xmath557 which proves .
that is , the @xmath188 approximate @xmath42 from below for @xmath558 with the errors no more than @xmath517 . for @xmath558 ,
notice that @xmath559 .
then for @xmath515 , from , and @xmath531 we obtain @xmath560 provided that @xmath561 , solving which we get @xmath535 .
together with the upper bound of , the above proves . for the moderately ill - posed problems with @xmath562 and @xmath152 , we get @xmath563 i.e.
, holds , provided that @xmath564 and @xmath95 is such that @xmath565 which means that @xmath566 it is easily verified that the above right - hand side monotonically decreases with respect to @xmath558 , whose minimum attains at @xmath567 and equals @xmath568 .
furthermore , since @xmath568 decreases monotonically as @xmath5 increases , its minimum over @xmath152 is @xmath569 , which is just the condition .
similar to , there must be @xmath95 such that holds .
again , we stress that the conditions on @xmath186 and @xmath187 should be regarded as an indicator that @xmath186 and @xmath187 must not be close to one other than precise requirements since we have used the amplified and the simplified models @xmath570 and @xmath532 . comparing theorem [ nearapprox ] with theorem [ ritzvalue ] ,
we find out that , as far as the severely or moderately ill - posed problems is concerned , for @xmath152 the near best rank approximation @xmath183 essentially means that the singular values @xmath188 of @xmath138 approximate the first @xmath5 large singular values @xmath42 of @xmath3 in natural order , provided that @xmath97 or @xmath95 suitably . under the conditions of theorems [ nearapprox][ritzvalue ] ,
let us explore how the results in them depend on @xmath270 . and indicate that , for the severely ill - posed problems with @xmath512 , ignoring higher order small terms , we have @xmath571 and @xmath572 for @xmath107 ; for the moderately ill - posed problems with @xmath513 , indicates that @xmath486 and @xmath221 are comparable in size for @xmath107 , while shows that @xmath221 is at most of modest size for @xmath107 . as a result ,
theorem [ thm2 ] and theorem [ moderate ] demonstrate that @xmath573 and @xmath295 fairly for severely and moderately ill - posed problems , respectively .
in other words , the largest canonical angle between @xmath217 and @xmath269 does not exceed @xmath574 and is considerably smaller than @xmath575 for these two kinds of problems , respectively .
[ extract ] theorems [ main1][ritzvalue ] show that , for @xmath152 , the @xmath5-step lanczos bidiagonalization is guaranteed to extract or acquire the first @xmath5 dominant svd components for the severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably , so that lsqr has the full regularization for these two kinds of ill - posed problems and can obtain best possible regularized solutions @xmath151 at semi - convergence .
let us have a closer look at the regularization of lsqr for mildly ill - posed problems .
we observe that the sufficient condition for is never met for this kind of problem because @xmath576 for any @xmath56 and @xmath577 .
this implies that the @xmath56 ritz values @xmath578 may not approximate the the first @xmath56 large singular values @xmath42 in natural order and there is at least one ritz value @xmath579 , causing that @xmath151 is already deteriorated and can not be as accurate as the best tsvd solution @xmath61 .
we can also make use of theorem [ initial ] to explain the partial regularization of lsqr : theorem [ moderate ] has shown that @xmath221 is generally not small and can become arbitrarily large as @xmath5 increases up to @xmath56 for mildly ill - posed problems , meaning that @xmath311 for @xmath307 and some @xmath107 , as the sharp bound indicates , from which it follows that a small ritz value @xmath579 generally appears .
we investigate the general best or near best rank @xmath5 approximations to @xmath3 with @xmath513 and @xmath99 .
we aim to show that , for each of such rank @xmath5 approximations , its smallest nonzero singular value may be smaller than @xmath150 for @xmath307 , that is , its nonzero singular values may not approximate the @xmath5 large singular values of @xmath3 in natural order , while the smallest nonzero singular value of such a rank @xmath5 approximation is guaranteed to be bigger than @xmath150 if only @xmath95 suitably . as it will turn out ,
this can help us further understand the regularization of lsqr for mildly and moderately ill - posed problems .
finally , we investigate the behavior of the ritz values @xmath539 when @xmath183 is not a near best rank @xmath5 approximation to @xmath3 for mildly ill - posed problems .
first of all , we point out an intrinsic fact that the best rank @xmath5 approximations to @xmath3 with respect to the 2-norm are _ not _ unique , let alone its near best rank @xmath5 approximations .
this subtlety is important for further understanding theorem [ ritzvalue ] .
let @xmath580 be a best or near best rank @xmath5 approximation to @xmath3 with @xmath581 with any @xmath582 satisfying @xmath583 , i.e. , @xmath584 is between @xmath150 and @xmath58 and closer to @xmath150 , by which we get @xmath585 it is remarkable to note that @xmath580 is not unique .
for example , among others , all the @xmath586 with any @xmath587 and @xmath588 is a family of best or near best rank @xmath5 approximations to @xmath3 .
the smallest nonzero singular value of @xmath589 is @xmath590 . since @xmath513 and @xmath591 for any @xmath403 and @xmath307
, we obtain @xmath592 for @xmath593 sufficiently close to one .
this shows that @xmath590 does not lie between @xmath150 and @xmath58 and interlace them for @xmath403 .
furthermore , for a given @xmath594 $ ] , the bigger @xmath5 is , the smaller @xmath595 is , and the further is @xmath590 away from @xmath150 . on the other hand , for @xmath593 sufficiently small we have @xmath596 that is , @xmath590 interlaces @xmath150 and @xmath58 for @xmath593 sufficiently small . for @xmath3 with @xmath513 and @xmath95 ,
the situation is much better since , for any @xmath5 , the requirement is met for _ any _ @xmath587 provided that @xmath95 suitably , leading to @xmath597 , meaning that the smallest singular value @xmath590 of a near best rank approximation @xmath589 interlaces @xmath150 and @xmath58 .
however , we should be aware that the above analysis is made for the _ worst - case _ : for any best or a near best rank @xmath5 approximation @xmath580 to @xmath3 , the minimum of the smallest nonzero singular values of all the @xmath580 is exactly @xmath598 .
we now prove this .
suppose that @xmath599 is the smallest nonzero singular value of a given such @xmath580 .
then from @xmath581 , by the standard perturbation theory we have @xmath600 clearly , the minimum of all the @xmath599 is attained if and only if the above equality holds , which is exactly @xmath598 . on the other side , by construction
, we also see that the smallest singular value @xmath590 of @xmath580 is arbitrarily close to or equal to @xmath58 by taking @xmath593 arbitrarily small or zero , which means that holds . in this case
, we observe from the equality in that @xmath601 and interlaces @xmath150 and @xmath58 . as far as lsqr is concerned ,
notice that the condition for the interlacing property is derived by assuming the worst case that @xmath602 , i.e. , @xmath321 is supposed to be the smallest possible nonzero one among all the @xmath599 , where @xmath580 belongs to the set of near best @xmath5 approximations that satisfy @xmath603 . for mildly ill - posed problems ,
the above arguments indicate that although in the worst case some of the @xmath5 ritz values @xmath188 may not approximate the first @xmath5 large singular values @xmath42 of @xmath3 in natural order , it is possible so in practice in case @xmath183 is _ occasionally _ a near best rank @xmath5 approximation to @xmath3 for some small @xmath107 . unfortunately , as we have shown previously , @xmath183 is rarely a near best rank @xmath5 approximation to @xmath3 for mildly ill - posed problems , i.e. , @xmath604 generally .
recall the second part of theorem [ initial ] and remark [ appear ] , which have shown rigorously that there is at least one ritz value @xmath605 if @xmath371 is sufficiently small there , i.e. , @xmath486 or equivalently @xmath221 is large .
this is exactly the case that @xmath183 is not a near best rank @xmath5 approximation to @xmath3 , causing that lsqr has only the partial regularization .
we can make a further analysis on the behavior of @xmath539 when is mildly ill - posed .
suppose that @xmath606 $ ] for some @xmath423 , which means that @xmath183 is definitely not a near best rank @xmath5 approximation to @xmath3 when @xmath607 .
below we derive the smallest upper bound for @xmath608 and obtain the biggest lower bound for @xmath163 . for @xmath283 with @xmath307
we have @xmath609 in which @xmath610 decreases with increasing @xmath105 for @xmath388 and is one for @xmath611 . therefore ,
the smallest upper bound for @xmath608 is no more than @xmath612 , which is smaller than @xmath150 once @xmath613 . in view of the above and , for
@xmath606 $ ] , since @xmath614 and has the biggest lower bound @xmath612 , we may have @xmath615 provided that @xmath613 . moreover , when @xmath615 , by the labeling rule there are @xmath616 ritz values @xmath617 smaller than @xmath150 . as a result , for @xmath168 , there are @xmath618 ritz values smaller than @xmath155 that deteriorate the lsqr iterate @xmath151 , so that lsqr has only the partial regularization .
in this section , we will present some results on the decay rates of @xmath621 and @xmath450 and on certain other rank @xmath5 approximations to @xmath3 and @xmath177 constructed by lanczos bidiagonalization .
the decay rates of @xmath619 and @xmath620 turn out to be particularly useful for practically detecting the degree of ill - posedness of and identifying the full or partial regularization of lsqr and lsmr .
the results on the new rank @xmath5 approximations will play a central role in analyzing the regularization of the krylov iterative regularization solvers lsmr @xcite and cgme @xcite . in section [ decay ] , we prove how @xmath619 and @xmath620 decay by relating them to @xmath450 and the estimates established for it .
then we show how to exploit the decay rate of @xmath622 to identify the degree of ill - posedness of m and the regularization of lsqr . in section [ lsmr ]
, we prove that the regularization of lsmr resembles lsqr for each of the three kinds of ill - posed problems . in section [ cgme ]
, we prove that the regularizing effects of cgme have intrinsic indeterminacy and are inferior to those of lsqr and lsmr . in section [ rrqr ] ,
we compare lsqr with some standard randomized algorithms @xcite and strong rank - revealing qr , i.e. , rrqr , factorizations @xcite , and show that the former solves ill - posed problems more accurately than the latter two ones at no more cost .
we consider how @xmath619 and @xmath620 decay in certain pronounced manners and show how to use them to identify the full or partial regularization of lsqr in practice .
[ main2 ] with the notation defined previously , the following results hold : @xmath623 _ proof_. from , since @xmath624 and @xmath551 are orthogonal matrices , we have @xmath625 with @xmath626 resulting from deleting the @xmath627 leading principal matrix of @xmath554 and the first @xmath628 zero rows and @xmath5 zero columns of the resulting matrix . from the above , for @xmath197 we have @xmath629 which shows that @xmath630 and @xmath631 since @xmath632 and @xmath633 .
so from , we get and . on the other hand , noting that @xmath634 we get .
note that @xmath635 and @xmath636 .
by @xmath637 and , note that @xmath638 equals the 2-norm of the submatrix deleting the first column of @xmath639 .
applying the cauchy s strict interlacing theorem to the singular values of this submatrix and @xmath639 , we obtain .
for severely and moderately ill - posed problems , based on the results in the last section , and show that @xmath640 and @xmath641 decay as fast as @xmath150 for @xmath107 and they may slow down a little bit for @xmath185 .
for mildly ill - posed problems , since @xmath486 are generally bigger than one considerably for @xmath107 , @xmath640 and @xmath641 can not generally decay as fast as @xmath150 , and their decays become slower for @xmath185 .
gazzola and his coauthors @xcite claim without rigorous proofs that @xmath642 and @xmath643 for severely ill - posed problems with the constants in @xmath259 unknown ( see proposition 4 of @xcite ) , but they do not show how fast each of them decays ; see proposition 6 of @xcite .
in contrast , our , and are rigorous and quantitative for all three kinds of ill - posed problems . in (
* corollary 3.1 ) , the authors have derived the product inequality @xmath644 whether or not this inequality is sharp is unknown , as they point out . by it , they empirically claim that @xmath645 may decay as fast as @xmath358 when the inequality is sharp ; conversely , if it is not sharp , nothing can be said on how fast @xmath645 decays .
we now shed light on and . for a given
, its degree of ill - posedness is either known or unknown .
if it is unknown , is of practical importance and can be exploited to identify whether or not lsqr has the full regularization without extra cost in an automatic and reliable way , so is . from the proofs of and , we find that @xmath640 and @xmath641 are as small as @xmath450 .
since our theory and analysis in section [ rankapp ] have proved that @xmath450 decays as fast as @xmath150 for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably and it decays more slowly than @xmath150 for mildly il - posed problems , the decay rate of @xmath58 can be judged by that of @xmath619 or @xmath620 or better judged by that of @xmath622 reliably , as shown below . given ,
run lsqr until semi - convergence occurs at iteration @xmath646 .
check how @xmath622 decays as @xmath5 increases during the process .
if , on average , it decays in an obviously exponential way , then is a severely ill - posed problem . in this case
, lsqr has the full regularization , and semi - convergence means that we have found a best possible regularized solution .
if , on average , @xmath619 decays as fast as @xmath647 with @xmath95 considerably , then is surely a moderately ill - posed problem , and lsqr also has found a best possible regularized solution to it at semi - convergence . if , on average , it decays at most as fast as or more slowly than @xmath647 with @xmath187 no more than one , is a mildly ill - posed problem .
if a hybrid lsqr is used , regularization applied to projected problems is performed only from iteration @xmath648 onwards until a best possible regularized solution is found , and , for a hybrid lsmr , regularization is applied to projected problems resulting from lsmr in the same way .
based on the previous results , we can rigorously analyze the regularizing effects of of lsmr @xcite and draw definitive conclusions on its regularization for three kinds of ill - posed problems .
lsmr is mathematically equivalent to minres applied to @xmath6 , and its iterate @xmath649 minimizes @xmath650 over @xmath651 , and the residual norm @xmath652 decreases monotonically with respect to @xmath5 . in our notation , noting from algorithm 1 that @xmath653 with rank @xmath5 , it is known from section 2.2 of @xcite that @xmath654 which can be efficiently computed and updated .
so lsmr amounts to solving the modified problem that perturbs the matrix @xmath177 in @xmath6 to its rank @xmath5 approximation @xmath655 , and the iterate @xmath649 is the minimum - norm least squares solution to the modified problem @xmath656 it is direct to verify that the tsvd solution @xmath62 is exactly the minimum - norm least squares solution to the modified problem @xmath657 that replaces @xmath177 by its 2-norm best rank @xmath5 approximation @xmath658 in @xmath6 . as a result ,
the regularization problem for lsmr now becomes that of accurately estimating @xmath659 , investigates how close it is to @xmath660 and analyzes whether or not the singular values of @xmath661 approximate the @xmath5 large singular values @xmath662 of @xmath177 . [ aprod ] for lsmr and @xmath663 , we have @xmath664 with @xmath665 . _
proof_. for the orthogonal matrix @xmath551 generated by algorithm 1 , noticing that @xmath666 , from and we obtain @xmath667 and @xmath668 where @xmath669 is the @xmath670 matrix that results from deleting the @xmath627 leading principal matrix of the symmetric tridiagonal matrix @xmath671 and the first @xmath628 zero rows and @xmath5 zero columns of the resulting matrix . note that @xmath671 has the diagonals @xmath672 , @xmath122 and the super- and sub - diagonals @xmath673 .
we have @xmath674 according to , it is direct to check that the @xmath675 symmetric tridiagonal matrix @xmath676 is a submatrix of @xmath669 that deletes its first row @xmath677 .
therefore , we have @xmath678 with the last equality from , which proves the lower bound of . on the other hand , noting the strict inequalities in and , since @xmath679 from @xcite we obtain @xmath680 with @xmath681 and @xmath682 if @xmath683 , from which the upper bound of follows directly .
recall that lsqr is mathematically equivalent to cgls that implicitly applies the cg method to @xmath6 . by , , and , noting that @xmath684 , we obtain the lsqr iterates @xmath685 which is the minimum - norm least squares solution to the modified problem @xmath686 that replaces @xmath177 by its rank @xmath5 approximation @xmath687 in @xmath6 . as a result , in the sense of solving the @xmath6 , the regularization of lsqr becomes that of estimating @xmath688
. we can establish the following result which relates this quantity to the counterpart of lsmr .
[ lsqrmr ] for the rank @xmath5 approximations in lsmr and lsqr , we have @xmath689 similar to the proof of theorem [ aprod ] , it is direct to verify that @xmath690 where @xmath691 is an @xmath692 matrix whose first column is @xmath693 and last @xmath181 columns are just the matrix @xmath669 defined by
. therefore , we have @xmath694 , which is just .
this theorem indicates that , as far as solving the normal equations @xmath6 is concerned , the rank @xmath5 approximation used by lsmr is at least as accurate as that used by lsqr .
however , regarding lsqr , theorem [ main1 ] is much more attractive since it not only deals with the rank @xmath5 approximation to @xmath3 directly but also estimates the accuracy of the rank @xmath5 approximation in terms of @xmath150 more compactly and informatively . according to the results and analysis in section [ rankapp ] , we have @xmath695 for severely ill - posed problems , and @xmath696 at most for moderately and mildly ill - posed problems . in comparison with theorem [ main1 ] , noting the form of the lower and upper bounds of , we see that @xmath697 as a rank @xmath5 approximation to @xmath177 is basically as accurate as @xmath183 as a rank @xmath5 approximation to @xmath3 . from @xcite ,
the singular values of @xmath698 are correspondingly bigger than those of @xmath376 , i.e. , @xmath699 .
therefore , the smallest singular value of @xmath698 is bigger than or equal to @xmath375 . as a result
, @xmath698 has no singular values smaller than @xmath700 before @xmath107 , provided that @xmath701 for @xmath107 .
this implies that the noise deteriorates the lsmr iterates @xmath649 no sooner than it does for the lsqr iterates @xmath148 .
a combination of theorem [ lsqrmr ] and the above two remarks means that the regularizing effects of lsmr are highly competitive with and not inferior to those of lsqr for each kind of ill - posed problem .
consequently , from the theory of lsqr in section [ rankapp ] , we conclude that lsmr has the full regularization for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably .
however , theorem [ aprod ] indicates that lsmr generally has only the partial regularization for mildly ill - posed problems since @xmath501 is generally bigger than @xmath155 considerably ; see remark [ mildre ] . we can define a near best rank @xmath5 approximation to @xmath177 similar to .
based on , if , in lsmr , we simply take @xmath702 for the ease of presentation , we can establish an analog of theorem [ nearapprox ] for lsmr .
in the meantime , completely parallel to the proof of theorem [ ritzvalue ] , we can also derive an analog of theorem [ ritzvalue ] for lsmr , in which the sufficient conditions on @xmath486 that ensure that the singular values of @xmath698 approximate the first @xmath5 large singular values of @xmath177 in natural order are found to be @xmath703 for @xmath3 with @xmath532 .
since lsmr and lsqr have similar regularizing effects for each kind of ill - posed problem , we can judge the full or partial regularization of lsmr by inspecting the decay rate of @xmath622 with respect to @xmath5 , as has been done for lsqr , in section [ rankapp ] we have interpreted lsqr as solving the modified problem that perturbs @xmath3 in to its rank @xmath5 approximation @xmath183 , and the regularization of lsqr has been related to the accuracy of such rank @xmath5 approximation to @xmath3 and how the @xmath5 large singular values of @xmath3 are approximated by the nonzero singular values of @xmath138 .
we will treat cgme in the same way later
. it might be hopeful to treat lsmr in this preferable and more direct way . from , lsmr is also equivalent to computing the minimum - norm least squares solution to the modified problem @xmath704 which perturbs @xmath3 in to its rank @xmath5 approximation @xmath705 .
however , an analysis of such formulation appears intractable because there is no explicit way to remove two generalized inverses @xmath26 in such rank @xmath5 approximation , which makes it impossible to accurately estimate @xmath706 in terms of @xmath150 . by and , we get @xmath707 where @xmath708 , and @xmath709 is lower bidiagonal with rank @xmath537 .
thus , it follows from that cgme is the cg method applied to @xmath7 and @xmath8 , where the @xmath5-th iterate @xmath710 minimizes the error @xmath711 , i.e. , @xmath712 , over @xmath651 , and the error norm @xmath713 decreases monotonically with respect to @xmath5 . by lanczos bidiagonalization
, it is known from @xcite that @xmath714 with @xmath715 and the residual norm @xmath716 with @xmath364 the @xmath5-th canonical vector of dimension @xmath5 .
noting that @xmath717 , we have @xmath718 therefore , @xmath710 is the minimum - norm least squares solution to the modified problem that replaces @xmath3 in by its rank @xmath5 approximation @xmath719 .
[ cgmeappr ] for the rank @xmath537 approximation @xmath720 and the rank @xmath5 approximation in cgme , we have @xmath721 since @xmath722 , we obtain @xmath723 which , together with , establishes . from and
we obtain @xmath724 the upper bound of is direct and by noting that @xmath725 along the proof path of theorem [ main2 ] , we obtain @xmath726 with @xmath639 defined by . it is straightforward to justify that the singular values of @xmath727 strictly interlace those of @xmath728 by noting that @xmath729 is an _ unreduced _ symmetric tridiagonal matrix ,
from which we get the lower bound of . by the definition of @xmath450
, this theorem says that @xmath730 is a less accurate rank @xmath5 approximation than @xmath183 in lsqr .
moreover , a combination of it and theorem [ main1 ] indicates that @xmath730 may never be a near best rank @xmath5 approximation to @xmath3 even for severely and moderately ill - posed problems because , unlike lsqr , there do not exist sufficient conditions on @xmath97 and @xmath95 to guarantee this . for mildly ill - posed problems ,
cgme generally has only the partial regularization since @xmath450 has been proved to be generally bigger than @xmath150 substantially and is rarely close to @xmath150 .
next we consider the equally important issue : the behavior of the singular values of @xmath731 .
let @xmath732 be its singular values labeled in the decreasing order , and observe that @xmath731 consists of the first @xmath5 rows of @xmath138 .
since @xmath733 is an @xmath734 unreduced symmetric tridiagonal matrix , whose eigenvalues are @xmath735 , and @xmath736 is the @xmath737 leading principal submatrix of @xmath733 , whose eigenvalues are @xmath738 , by the strict interlacing property of eigenvalues , we obtain @xmath739 on the other hand , note that @xmath666 and @xmath740 , i.e. , the singular values of @xmath741 are @xmath742 and zero , which is denoted by @xmath743 . observe that @xmath744 , and the first rows of @xmath741 are @xmath745 , whose singular values are @xmath746 .
therefore , applying the strict interlacing property of singular values to @xmath747 and @xmath741 , we have @xmath748 from which it follows that @xmath749 and indicate that , unlike @xmath321 that lies between @xmath150 and @xmath58 and approximates @xmath58 for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably ( cf . ) , the lower bound for @xmath750 is simply zero , a seemingly trivial one , and there does not exist a better one for it .
this means that @xmath750 may be much smaller than @xmath155 and actually it can be arbitrarily small , independently of the degree @xmath186 or @xmath187 of ill - posedness .
in other words , the size of @xmath186 or @xmath187 does not have any intrinsic effects on the lower bound of @xmath750 , and one can not control @xmath750 from below by choosing @xmath186 or @xmath187 . in the meantime , tells us that @xmath751 .
these facts , together with theorem [ cgmeappr ] , show that the regularization of cgme is inferior to that of lsqr and lsmr for each kind of problem . on the one hand , they mean that cgme has the partial regularization for mildly ill - posed problems ; on the other hand , the regularizing effects of cgme have indeterminacy for severely and moderately ill - posed problems , that is , it may or may not have the full regularization for these two kinds of problems . clearly , cgme has the full regularization only when @xmath730 is as accurate as the rank @xmath5 approximation @xmath183 and @xmath752 for these two kinds of problems with @xmath97 and @xmath95 considerably , but unfortunately there is no guarantee that these requirements are satisfied mathematically .
the above analysis indicates that cgme itself is not reliable and can not be trusted to compute best possible regularized solutions . in principle
, one can detect the full or partial regularization of cgme in the following way : one first exploits the decay rate of @xmath622 to identify the degree of ill - posedness of .
if is detected to be mildly ill - posed , cgme has only the partial regularization .
if is recognized as severely or moderately ill - posed , one then needs to compute the singular values of both @xmath138 and @xmath731 and check if @xmath753 using some reliable criterion .
if yes , cgme has the full regularization , otherwise it has only the partial regularization .
we can informally deduce more features on cgme .
for the lsqr iterate @xmath148 , note that the optimality requirement of cgme means that @xmath754 .
since @xmath755 and @xmath756 with the first terms in the right - hand sides being the same constant , it is normal that @xmath757 until the semi - convergence of cgme . keep in mind that the regularizing effects of cgme are inferior to or are at most as good as lsqr for each kind of ill - posed problem .
both @xmath758 and @xmath759 first decrease until their respective semi - convergence and then become increasingly large as @xmath5 increases . as a result
, we deduce that ( i ) @xmath710 is at least as accurate as @xmath148 until the semi - convergence of cgme and ( ii ) cgme reaches semi - convergence faster and at least no later than lsqr ; otherwise , indicates that the optimal regularized solution by cgme at semi - convergence would be more accurate than that by lsqr at semi - convergence , which contradicts the property that lsqr has better regularization than cgme .
the experiments in @xcite justify this assertions ; see figure 3.1 and figure 5.2 there .
next let us return to and show how to extract a rank @xmath5 approximation to @xmath3 from the rank @xmath537 approximation @xmath760 as best as possible .
[ approx ] let @xmath761 be the best rank @xmath5 approximation to @xmath762 with respect to the 2-norm
. then @xmath763 where @xmath764 is the smallest singular value of @xmath762 . _
proof_. write @xmath765 . then from we obtain @xmath766 by the assumption on @xmath580 and
, @xmath767 is the best rank @xmath5 approximation to @xmath768 .
keep in mind that @xmath67 is the best rank @xmath5 approximation to @xmath3 . since @xmath769 is a rank @xmath5 approximation to @xmath720 ,
we get @xmath770 from which , and it follows that holds . note that @xmath771 therefore , from it follows that holds .
we point out that may be conservative since we have amplified @xmath772 twice and obtained its bound @xmath150 , which can be a considerable overestimate . in comparison with and
, the bound indicates that @xmath767 may not be as accurate as @xmath183 .
illustrates that @xmath767 can be as accurate as @xmath183 because @xmath773 from and .
moreover , as we have explained , @xmath764 can be arbitrarily small , so that @xmath764 is negligible in and @xmath767 is at least as accurate as @xmath183 .
we now present an informal analysis to show that @xmath767 may be at least as accurate as @xmath183 as a rank @xmath5 approximation to @xmath3 .
keep in mind that @xmath774 be the singular values of @xmath762 .
then the singular values of @xmath761 are @xmath775 .
since @xmath632 for @xmath776 , applying the strict interlacing property of singular values to @xmath138 and @xmath762 , we have @xmath777 the above relationships , together with , prove that @xmath778 that is , the @xmath779 are more accurate than @xmath188 as approximations to @xmath189 . by the standard perturbation theory , note from that @xmath780 while the singular value differences between @xmath3 and @xmath767 are @xmath781 and @xmath782 , all of which , from and , are no more than @xmath450 . based on these rigorous facts and the relationship between @xmath580 and @xmath762 , it is well possible that @xmath783 , and if it is so , then by definition @xmath767 is a more accurate rank @xmath5 approximation to @xmath3 than @xmath183 is .
we compare the rank approximations @xmath183 and @xmath767 by lanczos bidiagonalization with those by some standard randomized algorithms and rrqr factorizations , and demonstrate that the former ones are much more accurate than the latter ones for severely and moderately ill - posed problems . note .
compare , and or with the corresponding results ( 1.9 ) , ( 5.6 ) , ( 6.3 ) and theorem 9.3 in @xcite for standard randomized algorithms and those on the strong rrqr factorization @xcite , where the constants in front of @xmath150 are like @xmath784 and @xmath785 , respectively , which are far bigger than one . within the framework of the rrqr factorizations ,
it is known from @xcite that the optimal factor of such kind is @xmath786 but to find corresponding permutations is an np - hard problem , whose cost increases exponentially with @xmath40 ; see also @xcite .
clearly , the strong rrqr factorizations are near - optimal within the framework , and they suit well for finding a hight - quality low rank @xmath5 approximation to a matrix whose @xmath5 large singular values are much bigger than the @xmath181 small ones . unfortunately , the standard randomized algorithms and rrqr factorization do not very nicely fit into solving ill - posed problems . since there are no considerable gaps of singular values , the rrqr factorization techniques can hardly find a near best rank @xmath5 approximation to @xmath3 in the sense of for three kinds of ill - posed problems , which is vital to solve reliably and makes us possible to ultimately find a best possible regularized solution .
in contrast , for the first two kinds of problems , the rank @xmath5 approximation @xmath183 can be near best for @xmath107 and no singular value smaller than @xmath155 appears . besides , it is easy to check that the @xmath5-step lanczos bidiagonalization costs fewer flops than the standard randomized algorithms do for a sparse @xmath3 , and it is more efficient than the strong rrqr factorization for a dense @xmath3 , which includes @xmath787 flops and the overhead cost of searching permutations . for further developments and recent advances on randomized algorithms , we refer to gu s work @xcite , where he has considered randomized algorithms within the subspace iteration framework proposed in @xcite , presented a number of new results and improved the error bounds for the rank @xmath5 approximations that are iteratively extracted .
such approaches may be promising to solve ill - posed problems .
based on proposition [ help ] , exploiting theorem [ main1 ] , theorem [ ritzvalue ] and theorem [ main2 ] , we present the following results , which , from the viewpoint of tikhonov regularization , explain why lsqr has the full regularization for severely and moderately ill - posed problems and it generally has the partial regularization for mildly ill - posed problems .
[ main3 ] for the severely or moderately ill - posed problems with @xmath97 or @xmath95 , under the assumptions of theorem [ ritzvalue ] , let @xmath166 be defined by .
then for @xmath152 we have @xmath788 _
proof_. for @xmath152 , it follows from that @xmath789 to simplify presentations and illuminate the essence , from theorem [ ritzvalue ] we simply replace @xmath790 by one and @xmath188 by @xmath42 in the above . then by we have @xmath791 for @xmath125 but @xmath54 , replace @xmath163 by @xmath92 approximately . then follows . by , since @xmath792 for @xmath793 , the factors @xmath794 and decay to zero with increasing @xmath54 for each fixed @xmath423
therefore , for @xmath793 we get @xmath795 replace the @xmath163 by @xmath796 in the above , and note that the second term is higher order small , compared with the first term
. then follows . for @xmath152 ,
since @xmath797 are dominant singular values , the factors @xmath798 are modest .
consequently , indicates that the @xmath799 with the errors @xmath800 for @xmath558 , while shows that the @xmath166 are at least as small as @xmath801 for @xmath793 and decrease with increasing @xmath54 . for mildly ill - posed problems and @xmath152 ,
as we have shown in remark [ appear ] and section [ rankapp ] , it is generally the case that @xmath802 .
suppose @xmath803 with the smallest integer @xmath171 .
then we have shown in the paragraph after proposition [ help ] that @xmath804 . as a result
, lsqr has only the partial regularization .
recall that @xmath398 , and define @xmath805 . in terms of , hansen ( * ? ? ?
* , theorems 6.4.1 - 2 ) presents the following bounds @xmath806 @xmath807 and @xmath808 where @xmath809 is the infinity norm of a vector .
we now address a few points on the bounds and .
first , there had no estimates for @xmath221 and @xmath810 ; second , what we need is @xmath811 other than @xmath812 , and as is seen from its proof in @xcite , it is relatively easy to obtain the accurate bound for @xmath813 , whereas it is hard to derive an accurate one for @xmath814 .
because of lacking accurate estimates , it is unclear how small or large the bound and are .
moreover , as it will appear soon , the factor @xmath815 in the numerator of may be a too crude overestimate , such that the bound is pessimistic and is useless to estimate @xmath816 and thus @xmath817 .
let us have a closer look at these points
. obviously , exploiting , we can only obtain the bounds @xmath818 from which it follows that @xmath819 as a result , the estimates for both @xmath814 and @xmath221 are too crude for @xmath40 large and @xmath5 small . indeed , as we have seen previously , their accurate estimates are much involved and complicated . in theorem
[ thm3 ] , we have derived accurate estimates for @xmath820 ; see , for severely ill - posed problems and , for moderately and mildly ill - posed problems . theorems [ thm2][moderate ] have given sharp estimates for @xmath221 for three kinds of problems , respectively .
the factor @xmath815 itself , though simple and elegant in form , does not give clear and quantitative information on its size . as a matter of fact , one must analyze its size carefully for the two cases @xmath107 and @xmath185 , respectively , for each kind of ill - posed problem ; see the discrete picard condition and .
for each of these two cases , using our proof approach used for theorems [ thm2 ] , [ moderate ] and [ thm3 ] , we can obtain accurate estimates for @xmath821 for three kinds of ill - posed problems , respectively . however , the point is that the factor @xmath815 results from a substantial amplification in the derivation .
it is seen from the last line of @xcite that this factor results from simply bounding it by @xmath822 where @xmath823 . for our context , this amplification is fatal , and it is subtle to obtain sharp bounds for @xmath824 .
we observe that @xmath825 is nothing but the first square root factor in , for which we have established the accurate estimates and for severely , moderately and mildly ill - posed problems , respectively , which hold for @xmath197 and are _ independent _ of @xmath40 .
it can be checked that these bounds are substantially smaller than @xmath826 . after the above substantial improvements on and
, we can exploit the accurate bounds for @xmath221 and @xmath814 in theorems [ thm2][moderate ] and theorem [ thm3 ] , as well as the remarks on them , to accurately estimate the bounds and . from them
we can draw the full regularization lsqr for severely and moderately ill - posed problems with @xmath97 and @xmath95 suitably and its partial regularization for mildly ill - posed problems .
making use of some standard perturbation results from hansen @xcite , we can quantitatively relate lsqr to the tsvd method and analyze the differences of their corresponding regularized solutions and differences of @xmath827 and @xmath828 predicting the right - hand side @xmath2 for @xmath152 .
[ lsqrtsvd ] for the severely or moderately ill - posed problem , let @xmath67 be the rank @xmath5 best approximation to @xmath3 , and assume that @xmath829 . then for @xmath152 we have @xmath830 @xmath831 where @xmath832 _ proof_. for the problem @xmath833 that replaces @xmath3 by @xmath67 in , we regard the rank @xmath5 matrix @xmath183 as a perturbed @xmath67 with the perturbation matrix @xmath834 . then by the standard perturbation results on the tsvd solutions @xcite ,
we obtain and directly .
write @xmath835 . since the rank @xmath5 matrices @xmath183 and @xmath67 have the @xmath5 nonzero singular values @xmath188 and @xmath42 , @xmath558 , respectively , from mirsky s theorem ( * ? ? ?
* , theorem 4.11 ) we get the bounds @xmath836 where the lower bound in is no more than the one in .
it is then expected that @xmath837 for severely and moderately ill - posed problems .
therefore , we have @xmath838 , and indicates that @xmath839 , is basically no more than @xmath484 , @xmath152 . from ,
since the possibly not small factor @xmath840 enters the bound , two regularized solutions @xmath148 and @xmath62 may differ considerably even though @xmath827 and @xmath828 predict the right - hand side @xmath2 with similar accuracy for @xmath152 .
this is the case for the inconsistent ill - posed problem @xmath0 with @xmath841 , where @xmath842 decreases with respect to @xmath5 until @xmath843 with @xmath844 and @xmath845 the first @xmath40 columns of the @xmath548 left singular vector matrix @xmath846 and @xmath847 the incompatible part of @xmath2 lying outside of the range of @xmath3 ( cf .
@xcite ) , which may be not much smaller than @xmath848 . here
we remark that the term @xmath847 appears in the relation ( 4.17 ) of @xcite but is missing in the above right - hand side @xcite . for the consistent @xmath1 , since @xmath849 ,
the right - hand side of is @xmath850 we see from the above and that two different regularized solutions can be quite different even if their residual norms are of similar very sizes , as addressed by hansen ( * ? ? ?
* , theorem 5.7.1 ) . however , we point out that the accuracy of different regularized solutions as approximations to @xmath27 can be compared .
if the norms of errors of them and @xmath27 have very comparable sizes , they are equally accurate regularized solutions to .
note that the assumption @xmath829 is only for severely and moderately ill - posed problems . as the previous analysis has indicated
, we have @xmath851 .
furthermore , for these two kinds of problems , we have shown @xmath852 .
as a result , it is easily justified that this assumption is valid for these two kinds of problems provided that @xmath97 and @xmath95 suitably .
however , the assumption generally fails to hold for the mildly ill - posed problems with @xmath451 and @xmath307 since , for @xmath403 , we have @xmath853
previously , under the assumption that the singular values of @xmath3 are simple , we have proved the results and made a detailed analysis on them .
recall the basic fact that the singular values @xmath539 of @xmath138 are always simple mathematically , independent of whether the singular values of @xmath3 are simple or multiple .
in other words , the lanczos bidiagonalization process works as if the singular values of @xmath3 are simple , and the ritz values @xmath539 , are the approximations to some of the _ distinct _ singular values of @xmath3 . in this section
, we will show that , by making a number of suitable and nontrivial changes and reformulations , our previous results and analysis can be extended to the case that @xmath3 has multiple singular values .
assume that @xmath3 has @xmath16 distinct singular values @xmath854 with @xmath42 being @xmath855 multiple and @xmath856 . in order to treat this situation
, we need to make a number of preliminary preparations and necessary modifications or reformulations .
below let us show the detail .
first of all , we need to take @xmath2 into consideration and present a new form svd of @xmath3 by selecting a specific set of left and right singular vectors corresponding to a multiple singular value @xmath42 of @xmath3 , so that holds for one particularly chosen left singular vector associated with @xmath42 .
specifically , for the @xmath855 multiple @xmath42 , the orthonormal basis of the corresponding left singular subspace , which is unique , can be chosen so that @xmath2 has a nonzero orthogonal projection on just one unit length left singular vector @xmath857 in the singular subspace and no components in the remaining @xmath858 ones .
precisely , let the columns of @xmath859 form an orthonormal basis of the left singular subspace associated with @xmath42 , each of which satisfies the discrete picard condition .
then we take @xmath860 where @xmath861 is the orthogonal projector onto the left singular subspace with @xmath42 , and define the corresponding unit length right singular vector by @xmath862 .
we select the other @xmath858 orthonormal left singular vectors which are orthogonal to @xmath857 and , together with @xmath857 , form the left singular subspace associated with @xmath42 , and define the corresponding unit length right singular vectors in the same way as @xmath180 , which and @xmath180 form an orthonormal basis of the unique right singular subspace with @xmath42 .
after such treatment , we get the desired svd of @xmath3 .
we stress that @xmath857 defined above is unique since the orthogonal projection of @xmath2 onto the left singular subspace with @xmath42 is unique and equal to @xmath863 for a given orthonormal @xmath859 .
now we need to prove that @xmath857 satisfies the discrete picard condition essentially . to see this , by the cauchy
schwarz inequality , and the assumption that each column of @xmath859 satisfies the discrete picard condition , we get @xmath864 therefore , the fourier coefficients @xmath865 , on average , decay faster than the singular values @xmath866 .
this is exactly what the discrete picard condition means ; see the description before .
recall that is a simplified model of this condition .
based on the estimate , we recover by simply resetting as latexmath:[\[\label{newpicard } with help of the svd of @xmath3 described above , it is crucial to observe that @xmath62 in is now the sum consisting of the first @xmath5 _ distinct _ dominant svd components of @xmath3 .
furthermore , for and such reformulation of , the matrix @xmath3 in them can be equivalently replaced by the new @xmath868 matrix @xmath869 where @xmath870 , @xmath871 and @xmath872 are the first @xmath16 columns of @xmath846 and @xmath329 , respectively , the last @xmath873 columns of @xmath846 are the other columns of the left singular vector matrix of @xmath3 excluding @xmath874 , which are orthogonal to @xmath2 by the construction stated above , and the last @xmath873 columns of @xmath329 are the other corresponding columns of the right singular vector matrix of @xmath3 . obviously , for the new svd of @xmath3 defined above , @xmath875 is of rank @xmath16 with the @xmath16 simple nonzero singular values @xmath876 , its left and right singular vector matrices @xmath846 and @xmath329 are the corresponding ones of @xmath3 with proper column exchanges , respectively .
we have @xmath877 and the tsvd regularized solutions @xmath878 , where @xmath879 is the best rank @xmath5 approximation to @xmath875 with respect to the 2-norm .
in addition , we comment that from the discrete picard condition @xmath880 independently of @xmath40 and @xmath16 , we can obtain directly in the same way as the introduction of for .
another fundamental change is that the @xmath5-dimensional dominant right singular space of @xmath3 now becomes that of @xmath875 , i.e. , @xmath194 with @xmath195 associated with the first @xmath5 large singular values of @xmath875 .
it is the subspace of concern in the case that @xmath3 has multiple singular values .
we will also denote @xmath881 and @xmath882 .
as for krylov subspaces , by the svd of @xmath3 and that of @xmath875 , expanding @xmath2 as @xmath883 , we easily justify @xmath884 and @xmath885 by noting that @xmath886 for any integer @xmath887 and @xmath888 for any integer @xmath889 . thus , for the given @xmath2 , lanczos bidiagonalization works on @xmath3 exactly as if it does on @xmath875 .
that is , generated by algorithm 1 hold when @xmath3 is replaced by @xmath875 , and the @xmath5 ritz values @xmath188 approximate @xmath5 nonzero singular values of @xmath875 . moreover , and indicate @xmath890 as a result , since @xmath891 has nonzero components in all the eigenvectors @xmath892 of @xmath893 associated with its nonzero distinct eigenvalues @xmath894 , lanczos bidiagonalization can not break down until step @xmath895 , and the singular values @xmath896 of @xmath897 are exactly the singular values @xmath876 of @xmath875 . at step @xmath16
, lanczos bidiagonalization on @xmath3 generates the @xmath898 lower bidiagonal matrix @xmath899 and @xmath900 having done the above , what we need is to estimate how @xmath901 approximates or captures the @xmath5-dimensional dominant right subspaces @xmath269 , @xmath902 .
this is a crucial step and the starting point of all the later analysis . in what follows
let us show how to adapt the beginning part of the proof of theorem [ thm2 ] to the case that @xmath3 has multiple singular values .
observe the krylov subspace @xmath903 with @xmath904 and @xmath905 partition the diagonal matrix @xmath205 and the matrix @xmath206 into the forms @xmath906 where @xmath208 and @xmath907 .
since @xmath209 is a vandermonde matrix with @xmath92 distinct for @xmath125 , it is nonsingular .
therefore , from @xmath908 and the structures of @xmath909 and @xmath910 , we obtain @xmath911 with @xmath912 meaning that @xmath913 is orthogonal to the last @xmath873 columns of @xmath329 .
write @xmath914 and define @xmath915 then @xmath215 , the columns of @xmath216 form an orthonormal basis of @xmath913 , and we get the orthogonal direct sum decomposition @xmath916 denote @xmath917 . for @xmath918 , based on the above
, we get by replacing @xmath328 in by @xmath919 defined as and noting that @xmath218 is orthogonal to @xmath920 .
then it is direct to derive the same bounds for @xmath918 as those established previously in completely the same way . as for the extension of theorem [ initial ] , by definition and
, we need to replace @xmath328 in by @xmath919 defined as .
the unit - length @xmath326 is now a vector that has the smallest acute angle with @xmath921 , and we modify as @xmath922 recall that the columns of @xmath920 are the right singular vectors of @xmath875 corresponding to zero singular values .
it disappears when forming the rayleigh quotient of @xmath923 with respect to @xmath337 .
the proof of theorem [ initial ] then carries over to @xmath875 , and the results hold for the case that @xmath3 has multiple singular values .
another fundamental change is that , when speaking of a rank @xmath5 approximation , we now mean that for @xmath875 . note that the best rank @xmath5 approximation @xmath924 to @xmath875 is @xmath925 , @xmath926 , where @xmath927 and @xmath928 are defined as before , and @xmath929 .
the @xmath5-step lanczos bidiagonalzation process on @xmath3 now generates rank @xmath5 approximations @xmath183 in lsqr and @xmath930 in cgme to @xmath875 and the rank @xmath5 approximation @xmath931 in lsmr to @xmath893 , where @xmath932 and @xmath336 are the first @xmath537 and @xmath5 columns of @xmath933 and @xmath934 in .
we then need to estimate the approximation accuracy of these rank @xmath5 approximations and compare them with that of @xmath879 and @xmath893 , respectively .
meanwhile , for each of these three rank @xmath5 approximation matrices , we need to analyze how its @xmath5 nonzero singular values approximate @xmath5 singular values of @xmath875 or @xmath923 . for the rank @xmath5 approximation @xmath183 to @xmath875 in lsqr ,
similar to , we define @xmath935 then , without any changes but the replacement of the index @xmath40 by @xmath16 , all the results in section [ rankapp ] and in theorem [ main2 ] carry over to the multiple singular value case . the final important note is how to extend the results presented section [ decay][cgme ] to the multiple singular value case .
we have to derive the three key relations similar to , and , where the fact that lanczos bidiagonalization can be run to @xmath40 steps without breakdown is exploited . in the case
that @xmath3 has multiple singular values , since lanczos diagonalization on @xmath3 must break down at step @xmath895 , there are no @xmath936 and @xmath551 as in . to this end , from we augment @xmath933 and @xmath934 to the @xmath548 and @xmath550 orthogonal matrices @xmath937 and @xmath938 , respectively , from which and we obtain @xmath939 having this relation , like and , we get @xmath940 where @xmath639 is the right bottom @xmath941 matrix of @xmath897 , similar to @xmath639 in .
then theorem [ main2 ] extends naturally to the multiple singular value case without any change but the replacement of the index @xmath40 by @xmath16 , and all the other results and analysis in section [ decay][cgme ] carry over to this case as well .
the results in section [ compare ] hold without any change whenever @xmath3 , @xmath67 and the index @xmath40 are replaced by @xmath875 , @xmath924 and @xmath16 , respectively . in summary , based on the above reformulations , changes and preliminary work , except section [ rrqr ]
, we have extended all the results and analysis in sections [ lsqr][compare ] to the case that @xmath3 has multiple singular values , just as we have done for the simple singular value case . in the analysis , derivation and results ,
the index @xmath40 is often replaced by @xmath16 whenever needed , and when this is necessary is clear from the context related .
for a number of problems from hansen s regularization toolbox @xcite , huang and jia @xcite have numerically justified the full regularization of lsqr for severely and moderately ill - posed problems and its partial regularization for mildly ill - posed problems , where each @xmath3 is @xmath942 . in this section , we report numerical experiments to confirm our theory and illustrate the full or partial regularization of lsqr in much more detail .
for the first two kinds of problems , we demonstrate that @xmath943 and @xmath641 decay as fast as @xmath150 .
we compare lsqr and the hybrid lsqr with the tsvd method applied to projected problems . in the experiments
, we use the l - curve criterion , the function @xmath944 in @xcite , to determine an actually optimal regularization parameter and depict the l - curve . for each of
severely and moderately ill - posed problems , we show that the regularized solution obtained by lsqr at semi - convergence is at least as accurate as the best tsvd regularized solution , indicating that lsqr has the full regularization . in the meantime , we show that the regularized solution obtained by lsqr at semi - convergence is considerably less accurate than that by the hybrid lsqr for mildly ill - posed problems , demonstrating that lsqr only has the partial regularization . as a byproduct , we compare lsqr with gmres and rrgmres and illustrate the failure of the latter ones for general nonsymmetric ill - posed problems .
we choose several ill - posed problems from hansen s regularization toolbox @xcite , which include the severely ill - posed problems @xmath945 , the moderately ill - posed problems @xmath946 , and the mildly ill - posed problem @xmath947 .
all the codes are from @xcite , and the problems arise from discretizations of .
we remind that , as far as is concerned , our primary goal consists in justifying the regularizing effects of iterative solvers for , which are _ unaffected by the size _ of and only depends on the degree of ill - posedness , the noise level @xmath184 and the actual discrete picard condition , provided that the condition number of , measured by the ratio between the largest and smallest singular values of each @xmath3 , is large enough . therefore , for this purpose , as extensively done in the literature ( see , e.g. , @xcite and the references therein as well as many other papers ) , it is enough to report the results on small and/or moderate discrete ill - posed problems since the condition numbers of these @xmath3 are already huge or large , which , in finite precision arithmetic , are roughly @xmath948 and @xmath949 for severely , moderately and mildly ill - posed problems with @xmath950 , respectively . indeed , for @xmath40 large , say , 10,000 or more , we have observed the same behavior as that for small @xmath40 , e.g. , @xmath950 used in this paper . also , an important reason is that such choice enables us to fully justify the regularization effects of lsqr by comparing it with the tsvd method , which suits only for small and/or medium sized problems because of its computational complexity for @xmath40 large . for each example
, we generate a @xmath951 matrix @xmath3 , the true solution @xmath27 and noise - free right - hand side @xmath22 . in order to simulate the noisy data , we generate white noise vectors @xmath21 such that the relative noise levels @xmath952 , respectively .
we mention that , to illustrate the behavior of the hybrid lsqr more generally , after this section we , in the concluding section , will report some important observations on @xmath953 and @xmath947 of @xmath954 and @xmath955 , whose condition numbers are as large as @xmath956 and @xmath957 for @xmath958 , respectively .
to simulate exact arithmetic , lsqr uses full reorthogonalization in lanczos bidiagonalization .
all the computations are carried out in matlab 7.8 with the machine precision @xmath959 under the miscrosoft windows 7 64-bit system . * example 1*. this problem @xmath960 arises from one - dimensional image restoration and is obtained by discretizing with @xmath961 $ ] as the domains of @xmath16 and @xmath17 , where @xmath962 * example 2*. this problem @xmath963 has a discontinuous solution and is obtained by discretizing with @xmath964 $ ] as the domains of @xmath16 and @xmath17 , where @xmath965 the problems @xmath960 and @xmath963 are severely ill - posed with the singular values @xmath966 for @xmath960 and @xmath967 for @xmath963 , respectively . in figure [ fig1 ] , we display the decay curves of the @xmath450 for @xmath968 , respectively .
we observe that the three curves with different @xmath969 are almost unchanged .
this is in accordance with our remark [ decayrate ] , where it is stated that the decay rate of @xmath450 is little affected by noise levels for severely ill - posed problems , since @xmath450 primarily depends on the decay rate of @xmath150 and different noise levels only affect the value of @xmath56 other than the decay rate of @xmath450 .
in addition , we have observed that @xmath450 and @xmath150 decay until they level off at @xmath970 due to round - off errors in finite precision arithmetic .
most importantly , the results clearly illustrate that @xmath450 decreases as fast as @xmath150 , and we have @xmath852 , whose decay curves are almost indistinguishable .
these results are in accordance with our theory . in figure
[ fig2 ] , we plot the relative errors @xmath971 with different @xmath969 for these two problems . as we see , lsqr exhibits the clear semi - convergence .
moreover , for a smaller @xmath969 , we get a more accurate regularized solution at cost of more iterations , as @xmath56 is bigger from and . and
@xmath150 for the problem @xmath960 with @xmath972 ( left ) and @xmath973 ( right ) ; ( c)-(d ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath963 with @xmath973 ( left ) and @xmath974 ( right).,width=226,height=188 ] ( a ) and @xmath150 for the problem @xmath960 with @xmath972 ( left ) and @xmath973 ( right ) ; ( c)-(d ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath963 with @xmath973 ( left ) and @xmath974 ( right).,width=226,height=188 ] ( b ) and @xmath150 for the problem @xmath960 with @xmath972 ( left ) and @xmath973 ( right ) ; ( c)-(d ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath963 with @xmath973 ( left ) and @xmath974 ( right).,width=226,height=188 ] ( c ) and @xmath150 for the problem @xmath960 with @xmath972 ( left ) and @xmath973 ( right ) ; ( c)-(d ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath963 with @xmath973 ( left ) and @xmath974 ( right).,width=226,height=188 ] ( d ) with @xmath968 for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( a ) with @xmath968 for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( b ) * example 3*. this problem @xmath975 arises from the inverse heat equation and is obtained by discretizing with @xmath964 $ ] as integration interval , where the kernel @xmath976 with @xmath977 this is a moderately ill - posed problem .
* example 4*. this problem is the @xmath953 famous problem , a moderately ill - posed one .
it is obtained by discretizing with @xmath978 $ ] as the domains of @xmath16 and @xmath17 , where @xmath979 and @xmath150 for the problem @xmath975 with ( left ) and ( b ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath953 with @xmath973 ( right).,width=226,height=188 ] ( a ) and @xmath150 for the problem @xmath975 with ( left ) and ( b ) : decay curves of the sequences @xmath450 and @xmath150 for @xmath953 with @xmath973 ( right).,width=226,height=188 ] ( b ) from figure [ fig3 ] , we see that @xmath450 decreases almost as fast as @xmath150 for moderately ill - posed problems . however ,
slightly different from severely ill - posed problems , @xmath450 , though very good approximations to @xmath150 , may not be so accurate .
this is expected , as the constants @xmath486 in are generally bigger than those in for severely ill - posed problems .
also , different from figure [ fig1 ] , we observe from figure [ fig3 ] that @xmath450 deviates more from @xmath150 with increasing @xmath5 , especially for the problem @xmath953 .
this confirms remarks [ decayrate][decayrate2 ] on moderately ill - posed problems . in figure [ fig4 ]
, we depict the relative errors of @xmath148 , and from them we observe analogous phenomena to those for severely ill - posed problems .
the only distinction is that lsqr now needs more iterations , i.e. , a bigger @xmath56 is needed for moderately ill - posed problems with the same @xmath969 , as is seen from and .
with @xmath968 for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( a ) with @xmath968 for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( b ) * example 5*. the mildly ill - posed problem @xmath947 is obtained by discretizing with @xmath964 $ ] as the domains of @xmath16 and @xmath17 , where the kernel @xmath15 is the green s function for the second derivative : @xmath980 and the solution @xmath14 and the right - hand side @xmath13 are given by @xmath981 and @xmath150 for @xmath947 with @xmath973,width=226,height=188 ] ( a ) and @xmath150 for @xmath947 with @xmath973,width=226,height=188 ] ( b ) figure [ figmild ] ( a)-(b ) display the decay curves of the partial and complete sequences @xmath450 and @xmath150 .
we see that , different from severely and moderately ill - posed problems , @xmath450 does not decay so fast as @xmath150 and deviates from @xmath150 significantly .
recall that theorem [ main1 ] holds for mildly ill - posed problems , where @xmath486 defined by is considerably bigger than one .
these observations justify our theory and confirm that the rank @xmath5 approximations to @xmath3 generated by lanczos bidiagonalization are not as accurate as those for severely and moderately problems .
consequently , lsqr has only the partial regularization . for the severely ill - posed @xmath960 , @xmath963 and the moderately ill - posed @xmath975 , @xmath953
, we compare the regularizing effects of lsqr and the hybrid lsqr with the tsvd method applied to the projected problems , and demonstrate that they compute the same best possible regularized solution for each problem and lsqr thus has the full regularization . for the mildly ill - posed problem @xmath947 ,
we show that lsqr has only the partial regularization and a hybrid lsqr can compute a best possible regularized solution . in the sequel , we only report the results for the noise level @xmath973 .
results for the other two @xmath969 are analogous and thus omitted unless stated otherwise .
we first have a close look at the severely and moderately ill - posed problems .
figure [ fig5a ] ( a)-(b ) and figure [ fig6 ] ( a)-(b ) plot the relative errors of regularized solutions obtained by the two methods for @xmath982 and @xmath946 .
clearly , we see that for each problem the relative errors reach the same minimum level .
after semi - convergence of lsqr , the tsvd method applied to projected problems simply stabilizes the regularized solutions with the minimum error and does not improve them .
this means that lsqr has already found best possible regularized solutions at semi - convergence and has the full regularization , and regularization applied to projected problems does not help and is unnecessary at all . in practice
, we simply stop lsqr after its semi - convergence for severely and moderately ill - posed problems .
for these four problems , for test purposes we choose @xmath983 for lsqr .
figure [ fig5a ] ( c)-(d ) and figure [ fig6 ] ( c)-(d ) show that the regularized solutions @xmath984 are generally excellent approximations to the true solutions @xmath27 .
the exception is the problem @xmath963 with a discontinuous solution , as depicted in figure [ fig5a ] ( d ) . for it ,
the regularized solution deviates from @xmath27 considerably and the relative error is not small .
this is because all cg type methods applied to either or its normal equations @xmath6 or @xmath7 with @xmath8 compute smooth regularized solutions .
in some sense , lsqr and cgls are equivalent to implicitly solving the tikhonov regularization problem , and their regularized solutions are of the filtered form .
it is well known that the regularization term @xmath985 in tikhonov regularization does not suit for the discontinuous solution .
the ill - posed problems with discontinuous or non - smooth solutions are from numerous important applications , including linear regression , barcode reading , gravity surveying in geophysics , image restoration and some others @xcite . for them
, a better alternative is use the 1-norm @xmath986 as the regularization term , which leads to the total variation regularization @xcite or errors - in - variables modeling called in @xcite , where @xmath987 is some @xmath988 matrix with no restriction to @xmath102 and is typically taken to be the discrete approximation to the first or second derivative operator ( * ? ? ? * ch.8 ) . by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 , i.e. , @xmath151 , by lsqr for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( a ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 , i.e. , @xmath151 , by lsqr for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( b ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 , i.e. , @xmath151 , by lsqr for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( c ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 , i.e. , @xmath151 , by lsqr for @xmath960 ( left ) and @xmath963 ( right).,width=226,height=188 ] ( d ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 by lsqr for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( a ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 by lsqr for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( b ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 by lsqr for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( c ) by lsqr and the hybrid lsqr for @xmath973 ; ( c)-(d ) : the best possible regularized solutions @xmath984 by lsqr for @xmath975 ( left ) and @xmath953 ( right).,width=226,height=188 ] ( d ) now we investigate the behavior of lsqr and the hybrid lsqr for @xmath947 .
figure [ figmild2 ] ( a ) indicates that the relative errors of @xmath148 by the hybrid lsqr reach a considerably smaller minimum level than those by lsqr , illustrating that lsqr has only the partial regularization . precisely , we find that the semi - convergence of lsqr occurs at iteration @xmath989 , but the regularized solution is not acceptable .
the hybrid lsqr uses a larger six dimensional krylov subspace @xmath990 to construct a more accurate regularized solution .
we also choose @xmath983 for lsqr and the hybrid lsqr , respectively .
figure [ figmild2 ] ( b ) indicates that the best regularized solution by the hybrid lsqr is a considerably better approximation to @xmath27 than that by lsqr , especially in the non - smooth middle part of @xmath27 . and
the regularized solutions @xmath984 by lsqr at semi - convergence and the best possible regularized solution by the hybrid lsqr for deriv2.,width=226,height=188 ] ( a ) and the regularized solutions @xmath984 by lsqr at semi - convergence and the best possible regularized solution by the hybrid lsqr for deriv2.,width=226,height=188 ] ( b ) for the severely ill - posed @xmath991 and the moderately ill - posed @xmath992 , we now illustrate that @xmath619 and @xmath620 decay as fast as the singular values @xmath58 of @xmath3 .
we take the noise level @xmath973 .
the results are similar for @xmath972 and @xmath293 .
figure [ fig7 ] illustrates that @xmath619 and @xmath620 both decay as fast as @xmath58 , and for @xmath960 and @xmath963 all of them decay swiftly and level off at @xmath970 due to round - off errors in finite precision arithmetic .
precisely , they reach the level of @xmath970 at @xmath993 and @xmath994 for @xmath960 and @xmath963 , respectively . such decay behavior has also been observed in @xcite , but no theoretical support was given .
these experiments confirm theorem [ main1 ] and theorem [ main2 ] , which have proved that @xmath450 decreases as fast as @xmath150 and that @xmath619 and @xmath620 decay as fast as @xmath58 .
, @xmath620 and @xmath58 for @xmath995 ( from top left to bottom right).,width=226,height=188 ] ( a ) , @xmath620 and @xmath58 for @xmath995 ( from top left to bottom right).,width=226,height=188 ] ( b ) , @xmath620 and @xmath58 for @xmath995 ( from top left to bottom right).,width=226,height=188 ] ( c ) , @xmath620 and @xmath58 for @xmath995 ( from top left to bottom right).,width=226,height=188 ] ( d ) we compare the performance of lsqr and the tsvd method for the severely ill - posed @xmath982 and moderately ill - posed @xmath946 .
we take @xmath973 . for each problem , we compute the norms of regularized solutions , their relative errors and the residual norms obtained by the two methods .
we plot the l - curves of the residual norms versus those of regularized solutions in the @xmath996-@xmath996 scale .
figures [ lsqrtsvd1][lsqrtsvd2 ] indicate lsqr and the tsvd method behave very similarly for @xmath960 and @xmath963 .
they illustrate that , for @xmath963 , the norms of approximate solutions and the relative errors by the two methods are almost indistinguishable for the same @xmath5 , and , for @xmath960 , the residual norms by lsqr decreases more quickly than the ones by the tsvd method for @xmath997 and then they become almost identical starting from @xmath989 .
the l - curves tell us that the two methods obtain the best regularized solutions when @xmath998 and @xmath999 for @xmath960 and @xmath963 , respectively .
the values of @xmath56 determined by the l - curves are exactly the ones at which semi - convergence occurs , as indicated by ( b ) and ( c ) in figures [ lsqrtsvd1][lsqrtsvd2 ] .
these results demonstrate that lsqr has the full regularization and resembles the tsvd method very much . for each of @xmath975 and @xmath953 , figures [ lsqrtsvd3][lsqrtsvd4 ] demonstrate that the best regularized solution obtained by lsqr is at least as accurate as , in fact , a little bit more accurate than that by the tsvd method , and the corresponding residual norms decreases and drop below at least the same level as those by the tsvd method .
the residual norms by the two methods then stagnate after the best regularized solutions are found .
all these confirm that lsqr has the full regularization .
the fact that the best regularized solutions by lsqr can be more accurate than the best tsvd solutions is not unusual .
we can explain why .
we deduce from figure [ fig6 ] that the true solutions of the problems @xmath975 and @xmath953 are at least first order differentiable . as a matter of fact , in the hilbert space setting , for a linear compact operator equation @xmath83
, the tsvd method and standard - form tiknonov regularization method have been shown to be order optimal only when the true solution is continuous or first order differentiable , and they are not order optimal for stronger smoothness assumptions on the true solutio .
in contrast , cgls is order optimal , and the smallest error of the iterates is of the same order as the worst - case error for the arbitrarily smooth true solution , that is , given the same noise level , the smoother the true solution is , the more accurate the best regularized solution is .
in other words , for the smoother true solution , the best regularized solution by cgls is generally more accurate than the counterpart corresponding to the continuous or first order differentiable true solution ; see , e.g. , @xcite and @xcite .
consequently , in the finite dimensional space case , provided that the mathematically equivalent lsqr has the full regularization , its best regularized solution is at least as accurate as and can be more accurate than the best regularized solution by the tsvd method or the standard - form tiknonov regularization method when the true solution is smoother than only continuous or first order differentiable . from the figures we observe some obvious differences between moderately and severely ill - posed problems . for @xmath975
, it is seen that the relative errors and residual norms converge considerably more quickly for the lsqr solutions than for the tsvd solutions .
figure [ lsqrtsvd3 ] ( b ) tells us that lsqr only uses @xmath1000 iterations to find the best regularized solution and the tsvd method finds the best regularized solution for @xmath1001 , while the l - curve gives @xmath1002 and @xmath1003 , respectively .
similar differences are observed for @xmath953 , where figure [ lsqrtsvd4 ] ( b ) indicates that both lsqr and the tsvd method find the best regularized solutions at @xmath998 , while the l - curve shows that @xmath1004 for lsqr and @xmath1005 for the tsvd method .
therefore , unlike for severely ill - posed problems , the l - curve criterion is not very reliable to determine correct @xmath56 for moderately ill - posed problems .
we can observe more .
figure [ lsqrtsvd3 ] shows that the tsvd solutions improve little and their residual norms decrease very slowly for the indices @xmath1006 .
this implies that the @xmath180 corresponding to these indices @xmath54 make very little contribution to the tsvd solutions .
this is due to the fact that the fourier coefficients @xmath1007 are very small relative to @xmath42 for these indices @xmath54 .
note that @xmath157 adapts itself in an optimal way to the specific right - hand side @xmath2 , while the tsvd method uses all @xmath1008 to construct a regularized solution , independent of @xmath2 .
therefore , @xmath157 picks up only those svd components making major contributions to @xmath27 , such that lsqr uses possibly fewer @xmath5 iterations than @xmath56 needed by the tsvd method to capture those truly needed dominant svd components .
the fact that lsqr ( cgls ) includes fewer svd components than the tsvd solution with almost the same accuracy was first noticed by hanke @xcite . generally , for severely and moderately ill - posed problems , we may deduce that lsqr uses possibly fewer than @xmath56 iterations to compute a best possible regularized solution if , in practice , some of @xmath1009 , @xmath80 are considerably bigger than the corresponding @xmath42 and some of them are reverse . for @xmath953 ,
as noted by hansen @xcite , half of the svd components satisfy @xmath1010 for @xmath54 even , only the odd indexed @xmath1011 make contributions to @xmath27 .
this is why the relative errors and residual norms of tsvd solutions do not decrease at even indices before @xmath61 is found .
.,width=226,height=188 ] ( a ) .,width=226,height=188 ] ( b ) .,width=226,height=188 ] ( c ) .,width=226,height=188 ] ( d ) .,width=226,height=188 ] ( a ) .,width=226,height=188 ] ( b ) .,width=226,height=188 ] ( c ) .,width=226,height=188 ] ( d ) .,width=226,height=188 ] ( a ) .,width=226,height=188 ] ( b ) .,width=226,height=188 ] ( c ) .,width=226,height=188 ] ( d ) .,width=226,height=188 ] ( a ) .,width=226,height=188 ] ( b ) .,width=226,height=188 ] ( c ) .,width=226,height=188 ] ( d ) gmres applied to solving with @xmath3 square computes the iterate @xmath1012 the quantity @xmath1013 measures the accuracy of the rank @xmath5 approximation @xmath1014 to @xmath3 , where the columns of @xmath1015 and @xmath1016 are orthonormal bases of @xmath1017 and @xmath1018 , respectively , generated by the arnoldi process starting with @xmath1019 , and @xmath1020 is the @xmath627 upper hessenberg matrix .
the sizes of @xmath1021 measure the regularizing effects of gmres for solving . we should address that , different from @xmath450 defined by for lsqr , which has been proved to decrease monotonically as @xmath5 increases ( cf . ) , mathematically @xmath1021 has no monotonic property .
similar to the lsqr iterates @xmath148 and @xmath450 , qualitatively speaking , if @xmath1021 decays smoothly in some definitive manner , then , to some extent , gmres has regularizing effects ; if they do not decay at all or behave irregularly , then gmres does not have regularizing effects and fails to work for .
we test gmres on the general nonsymmetric @xmath975 and the following example 6 , and compare it with lsqr .
* example 6*. consider the general nonsymmetric ill - posed problem @xmath1022 , which is severely ill - posed and arises from inverse laplace transformation .
it is obtained by discretizing the first kind fredholm integral equation with @xmath1023 the domains of @xmath16 and @xmath17 .
the kernel @xmath15 , the right - hand side @xmath13 and the solution @xmath14 are given by @xmath1024 we investigate the regularizing effects of gmres with @xmath973 .
let @xmath1025 denote the upper hessenberg matrix obtained by the @xmath5-step arnoldi process .
we notice that the @xmath1026 decay quickly with @xmath5 increasing , generally faster than @xmath58 ; see figure [ fig8 ] ( a)-(b ) .
this phenomenon may lead to a misbelief that gmres has general regularizing effects .
however , it is not the case .
in fact , a small @xmath1026 exactly indicates that all the eigenvalues of @xmath1027 may approximate some @xmath5 eigenvalues of @xmath3 well and the arnoldi method finds an approximate @xmath5-dimensional invariant subspace or eigenspace @xcite .
we also refer to @xcite for a detailed convergence analysis of the arnoldi method .
unfortunately , for a general nonsymmetric matrix @xmath3 , a small @xmath1026 does not mean that the singular values , i.e. , the ritz values , of @xmath1028 are also good approximations to some @xmath5 singular values of @xmath3 . as a matter of fact , as our analysis in section [ rankapp ] and section [ alphabeta ] has indicated , the accuracy of the singular values of a rank @xmath5 approximation matrix , here the singular values of @xmath1028 , as approximations to the @xmath5 large singular values of @xmath3 heavily relies on the size of @xmath1021 defined by other than @xmath1026 .
indeed , as indicated by figure [ fig8 ] ( c)-(d ) , though @xmath1026 is small , some of the singular values of @xmath1028 are very poor approximations to singular values of @xmath3 , and some of those good approximations are much smaller than @xmath150 and approximate the singular values of @xmath3 in disorder rather than in natural order .
it is important to note that , for a general nonsymmetric or , more rigorously , non - normal @xmath3 , the @xmath5-dimensional krylov subspace @xmath1017 that underlies the arnoldi process mixes all the left and right singular vectors of @xmath3 , and the arnoldi process generally fails to extract the dominant svd components and can not generate a high quality rank @xmath5 approximation to @xmath3 , causing that gmres has no good regularizing effects . and
@xmath150 ; ( c)-(d ) : the singular values ( star ) of @xmath1028 and the ones ( solid line ) of @xmath3 for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( a ) and @xmath150 ; ( c)-(d ) : the singular values ( star ) of @xmath1028 and the ones ( solid line ) of @xmath3 for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( b ) and @xmath150 ; ( c)-(d ) : the singular values ( star ) of @xmath1028 and the ones ( solid line ) of @xmath3 for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( c ) and @xmath150 ; ( c)-(d ) : the singular values ( star ) of @xmath1028 and the ones ( solid line ) of @xmath3 for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( d ) figure [ fig5 ] ( a)-(b ) gives more justifications .
we have a few important observations : for the two test problems , the quantities @xmath450 decay as fast as the @xmath150 for lsqr , while the @xmath1021 diverge quickly from the @xmath150 for gmres and do not exhibit any regular decreasing tendency . for @xmath1022 , the @xmath1021 decrease very slowly until @xmath1029 , then basically stabilize for three iterations followed , and finally start to increase from @xmath993 onwards . as for @xmath975 , the @xmath1021 are almost constant from beginning to end .
since all the @xmath1021 are not small , they indicate that the arnoldi process can not generate any reasonable and meaningful rank @xmath5 approximations to @xmath3 for @xmath1030 .
this is especially true for @xmath975 .
consequently , we are sure that gmres fails and does not have regularizing effects for the two test problems
. we plot @xmath1031 by lsqr and @xmath1032 by gmres in figure [ fig5 ] ( c)-(d ) .
obviously , lsqr exhibits semi - convergence , but gmres does not and the relative errors obtained by it even increase from the beginning ; see figure [ fig5 ] ( d ) .
this again demonstrates that gmres can not provide meaningful regularized solutions for these two problems .
let @xmath983 .
figure [ fig5 ] ( e ) and ( f ) show that lsqr obtains excellent regularized solutions , while gmres fails .
it is known that mr - ii @xcite for @xmath3 symmetric and rrgmres @xcite for @xmath3 nonsymmetric work on the subspace @xmath1033 .
they were originally designed to solve singular or inconsistent systems , restricted to a subspace of range of @xmath3 , and compute the minimum - norm least squares solutions when the ranges of @xmath3 and @xmath1034 are identical . however , for the preferred rrgmres @xcite , we have observed phenomena similar to those gmres , illustrating that rrgmres does not have regularizing effects for the test problems . from these typical experiments ,
we conclude that gmres and rrgmres are susceptible to failure for general nonsymmetric ill - posed problem and they are not general - purpose regularization methods .
in fact , as addressed in @xcite and @xcite , gmres and rrgmres may only work well when either the mixing of svd components is weak or the krylov basis vectors are just well suited for the ill - posed problem , as addressed in @xcite .
, @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( a ) , @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( b ) , @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( c ) , @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( d ) , @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( e ) , @xmath1021 , denoted by @xmath450-lsqr and @xmath450-gmres in the figure , and @xmath150 ; ( c)-(d ) : the relative errors @xmath1031 ; ( e)-(f ) : the regularized solutions @xmath984 obtained by lsqr and gmres for @xmath1022 ( left ) and @xmath975 ( right).,width=226,height=188 ] ( f )
for the large - scale ill - posed problem , iterative methods are the only viable approaches . of them , lsqr and cgls are most popularly used for general purposes , and cgme and lsmr are also choices .
they have regularizing effects and exhibit semi - convergence .
however , if a small ritz value appears before they capture all the needed dominant svd components , then semi - convergence is not enough and the methods have only the partial regularization . in this case ,
their hybrid variants have often been used to compute best possible regularized solutions . if semi - convergence means that they have already found best possible regularized solutions , they have the full regularization , and we simply stop them after semi - convergence .
we have considered the fundamental open question in depth : do lsqr , cgls , lsmr and cgme have the full or partial regularization for severely , moderately and mildly ill - posed problems ? we have first considered the case that all the singular values of @xmath3 are simple . as a key and indispensable step ,
we have established accurate bounds for the 2-norm distances between the underlying @xmath5 dimensional krylov subspace and the @xmath5 dimensional dominant right singular subspace for the three kinds of ill - posed problems under consideration .
then we have provided other absolutely necessary background and ingredients . based on them ,
we have proved that , for severely or moderately ill - posed problems with @xmath97 or @xmath95 suitably , lsqr has the full regularization .
precisely , for @xmath107 we have proved that a @xmath5-step lanczos bidiagonalization produces a near best rank @xmath5 approximation of @xmath3 and the @xmath5 ritz values approximate the first @xmath5 large singular values in natural order , and no small ritz value smaller than @xmath155 appears before lsqr captures the @xmath56 needed dominant svd components , so that the noise @xmath21 in @xmath2 can not come into play deteriorating regularized solutions until a best possible regularized solution has been found .
furthermore , we have shown that lsqr resembles the tsvd method .
for mildly ill - posed problems , we have proved that lsqr generally has only the partial regularization since a small ritz value generally appears before all the needed dominant svd components are captured .
since cgls is mathematically equivalent to lsqr , our assertions on the full or partial regularization of lsqr apply to cgls as well .
we have given bounds for the diagonals and subdiagonals of bidiagonal matrices generated by lanczos bidiagonalization for a general ill - posed problem .
particularly , we have proved that they decay as fast as the singular values of @xmath3 for the severely ill - posed problems or moderately ill - posed problems with @xmath97 or @xmath95 suitably and decay more slowly than the singular values of @xmath3 for mildly ill - posed problems .
these bounds are of practical value and can be used to identify the degree of ill - posedness without extra cost and decide the full or partial regularization of lsqr .
based on some of the results established for lsqr , we have derived accurate estimates for the quality of the rank @xmath5 approximations to @xmath3 and @xmath177 that are involved in cgme and lsmr , respectively .
we have analyzed the behavior of the smallest singular values of the projected matrices associated with cgme and lsmr . using these results
, we have shown that lsmr has the full regularization for severely and moderately ill - posed problems with @xmath95 and @xmath95 suitably , and it generally has only the partial regularization for mildly ill - posed probolems . in the meantime , we have shown that the regularization of cgme has indeterminacy and is inferior to lsqr and lsmr .
in addition , our results have indicated that the rank @xmath5 approximations to @xmath3 generated by lanczos bidiagonalization are substantially more accurate than those obtained by standard randomized algorithms @xcite at less cost of flops and the strong rrqr factorizations @xcite at comparable cost of flops .
remarkably , with a number of nontrivial modifications and reformulations , we have shown how to extend all the results obtained for lsqr , cgme and lsmr to the case that @xmath3 has multiple singular values .
we have made illuminating numerical experiments and confirmed our assertions on lsqr .
we have also compared lsqr with gmres and rrgmres , showing that the latter two methods do not have general regularizing effects and fail to deliver regularized solutions for general nonsymmetric ill - posed problems .
rrgmres may work well only for ( nearly ) symmetric or , more generally , ( nearly ) normal ill - posed problems , for which the left and right singular vectors are ( nearly ) identical to with the eigenvectors of @xmath3 .
our analysis approach can be adapted to mr - ii for symmetric ill - posed problems , and similar results and assertions are expected for three kinds of symmetric ill - posed problems . using a similar approach to that in @xcite ,
the authors @xcite has made an initial regularization analysis of mr - ii and derived the corresponding @xmath4 bounds , which are too large overestimates .
our approach are applicable to the preconditioned cgls ( pcgls ) and lsqr ( plsqr ) @xcite by exploiting the transformation technique originally proposed in @xcite and advocated in @xcite or the preconditioned mr - ii @xcite , all of which correspond to a general - form tikhonov regularization of the matrix pair @xmath1035 , in which the regularization term @xmath1036 is replaced by @xmath1037 with some @xmath988 matrix @xmath987 .
it should also be applicable to the mathematically equivalent lsqr variant @xcite that is based on a joint bidiagonalization of the matrix pair @xmath1035 that corresponds to the above general - form tikhonov regularization . in this setting , the generalized svd ( gsvd ) of @xmath1035 or the mathematically equivalent svd of @xmath1038 will replace the svd of @xmath3 to play a central role in analysis , where @xmath1039 is call the _ @xmath3-weighted generalized inverse of @xmath1040 _ and @xmath1041 if @xmath1040 is square and invertible ; see @xcite and @xcite . finally , we highlight on hybrid krylov iterative solvers and make some remarks , which deserve particular and enough attention in our opinion . because of lack of solid and rigorous regularized theory on lsqr , in order to find a best possible regularized solution for a given , one has commonly been using some hybrid lsqr variants without considering the degree of ill - posedness of ; see , e.g. , @xcite and the related papers mentioned in the introduction . the hybrid cgme @xcite and cgls @xcite
have also been proposed and used .
however , bjrck @xcite has addressed that the hybrid lsqr variants are mathematically complicated , and pointed out that it is hard to find reasonable regularization parameters and tell when to stop them reliably . for a hybrid lsqr variant , or more generally , for any hybrid krylov solver that first projects and then regularizes @xcite
, the situation is much more serious than what one might think .
it has long commonly accepted that _
`` first - regularize - then - project '' _ is equivalent to _
`` first - project - then - regularize '' _ and they produce the same solution ; see section 6.4 and figure 6.10 of @xcite .
this equivalence seems natural . as a matter of fact
, it is not the case , and they are generally _ not _ equivalent when solving .
assume that the same regularization parameter @xmath106 in tikhonov regularization is _
given_. both of them solve the the same well - posed problem and compute the same regularized solution . however , the fundamental point is that each of them must determine the _ unknown _ regularization parameter @xmath106 which is not given in advance .
mathematically , the approach of `` first - regularize - then - project '' can determine an optimal @xmath106 since satisfies the picard condition , though it is generally not computationally viable for a large .
on the contrary , for the approach of `` first - project - then - regularize '' , one must determine its optimal @xmath106 for each projected problem , so one will have a sequence of @xmath106 . for the discrete regularization parameter involved in the tsvd method for and projected problems ,
the situation is similar .
unfortunately , this may lead to fatal mathematical difficulties , as we will explain below .
as is well known , the picard condition is an absolutely necessary condition for the existence of the squares integrable solution to a linear compact operator equation ; without it , regularization would be out of the question ; see , e.g. , @xcite .
this is also true for the discrete linear ill - posed problem , where the discrete picard condition means that @xmath1042 uniformly with some ( not large ) constant @xmath39 such that regularization is useful to compute a meaningful approximation to it @xcite .
nevertheless , to the best of our knowledge , the discrete picard conditions for projected problems arising from lsqr or any other krylov iterative solver have been paid little attention until very recently @xcite .
unfortunately , a fatal problem is that _ the discrete picard conditions are not necessarily satisfied for the projected problems_. in @xcite , under the assumption that the right - hand side @xmath2 of is _ noise free _
, i.e. , the noise @xmath1043 and @xmath1044 , the authors have proved that the discrete picard conditions are satisfied or inherited for the projected problems under the _ absolutely necessary assumption _ that the @xmath5 ritz values , i.e. , the singular values of the projected matrix at iteration @xmath5 , approximate the @xmath5 large singular values of @xmath3 in natural order , regularization makes sense and can be used to solve the projected problems . however , as have been stated in @xcite and highlighted in this paper , under such assumption , krylov solvers themselves will find best possible regularized solutions at semi - convergence , and there is no need to continue iterating and regularize the projected problems at all . on the other hand , if the @xmath5 ritz values do not approximate the @xmath5 large singular values of @xmath3 in natural order and at least one ritz value smaller than @xmath155 appears before @xmath107 , the discrete picard conditions are _ essentially not satisfied any longer _ for the projected problems starting from such @xmath5 onwards .
if so , regularization applied to projected problems is mathematically groundless and meaningless , leading to invincible failure .
we take lsqr as an example for a precise statement on the discrete picard conditions for projected problems .
recall that , in the projected problem , the noisy right - hand side is @xmath1045 with @xmath20 and the noise - free right - hand side is @xmath1046 .
then for @xmath197 and for @xmath40 arbitrarily large and @xmath1047 ( cf .
@xcite ) , the discrete picard conditions for the projected problems are @xmath1048 uniformly with some constant @xmath39 . numerically , for a given @xmath40 , once @xmath1049 is large for some @xmath5 , then the discrete picard condition actually fails for the corresponding projected problem , such that regularization applied to the projected problem may work poorly .
indeed , for the moderately ill - posed phillips and mildly ill - posed deriv2 of order @xmath954 and @xmath955 , we have observed that the hybrid lsqr exhibits considerable _
erratic _ other than smooth curves of the errors between the regularized solutions and @xmath27 in the dampening and stabilizing stage , causing that the hybrid lsqr is unreliable to obtain a best regularized solutions to the problems ; see figure [ figerratic ] .
actually , the regularized solutions obtained by the hybrid lsqr after its stabilization are considerably less accurate than those by the pure lsqr itself .
for deriv2 of order @xmath954 , similar phenomena have also been noticed for the hybrid minres and mr - ii @xcite . by lsqr and the hybrid lsqr for phillips of @xmath954 and @xmath955 with @xmath973 and @xmath293 , respectively ; ( c)-(d ) : the relative errors @xmath1031 by lsqr and the hybrid lsqr for deriv2 of @xmath954 and @xmath955 with @xmath973.,width=226,height=188 ] ( a ) by lsqr and the hybrid lsqr for phillips of @xmath954 and @xmath955 with @xmath973 and @xmath293 , respectively ; ( c)-(d ) : the relative errors @xmath1031 by lsqr and the hybrid lsqr for deriv2 of @xmath954 and @xmath955 with @xmath973.,width=226,height=188 ] ( b ) by lsqr and the hybrid lsqr for phillips of @xmath954 and @xmath955 with @xmath973 and @xmath293 , respectively ; ( c)-(d ) : the relative errors @xmath1031 by lsqr and the hybrid lsqr for deriv2 of @xmath954 and @xmath955 with @xmath973.,width=226,height=188 ] ( c ) by lsqr and the hybrid lsqr for phillips of @xmath954 and @xmath955 with @xmath973 and @xmath293 , respectively ; ( c)-(d ) : the relative errors @xmath1031 by lsqr and the hybrid lsqr for deriv2 of @xmath954 and @xmath955 with @xmath973.,width=226,height=188 ] ( d ) this is due to the actual failure of the discrete picard conditions for the projected problems resulting from krylov solvers because each of the projected matrices starts to have at least one singular value _
considerably _ smaller than @xmath155 from some iteration @xmath107 onwards , which and whose corresponding ( left and right ) ritz vectors does not approximate any singular triplet of @xmath3 well .
a consequence of such actual failure is that it is hard to reliably stop the hybrid variants at right iteration in order to find best regularized solutions .
therefore , for mildly or moderately ill - posed problems with @xmath95 not enough , it is appealing to seek other mathematically solid and computationally viable variants of lsqr , lsmr and mr - ii in order to find best possible regularized solutions .
i thank dr . yi huang and mrs .
yanfei yang for running the numerical experiments in this paper . , _ methods in numerical algebra for ill - posed problems _
, report lith - r-33 - 1979 , dept . of mathematics , linkping univeristy , sweden , 1979 .
proceedings of the international symposium on ill - posed problems : theory and practice , university of delaware , newark , delaware , oct . 26 , 1979 . ,
_ a refined harmonic lanczos bidiagonalization method and an implicitly restarted algorithm for computing the smallest singular triplets of large matrices _ , siam j. sci .
comput . , 32 ( 2010 ) , | for the large - scale linear discrete ill - posed problem @xmath0 or @xmath1 with a noisy @xmath2 , lsqr , which is a krylov iterative solver based on lanczos bidiagonalization , and its mathematically equivalent cgls are most commonly used .
they have intrinsic regularizing effects , where the number of iterations plays the role of regularization parameter .
however , there has been no answer to the long - standing fundamental concern : _ for which kinds of problems lsqr and cgls can find best possible regularized solutions _ ?
the concern was actually expressed foresightedly by bjrck and eldn in 1979 . here
a best possible regularized solution means that it is at least as accurate as the best regularized solution obtained by the truncated singular value decomposition ( tsvd ) method , which and the best possible solution of standard - form tikhonov regularization are both of the same order of the worst - case error and can not be improved under the assumption that the solution to an underlying linear compact operator equation is continuous or its derivative squares integrable . in this paper
we make a complete analysis on the regularization of lsqr for severely , moderately and mildly ill - posed problems .
we first consider the case that the singular values of @xmath3 are simple .
we establish accurate @xmath4 theorems for the 2-norm distance between the underlying @xmath5-dimensional krylov subspace and the @xmath5-dimensional dominant right singular subspace of @xmath3 .
based on them and some follow - up results , for the first two kinds of problems , we prove the following results : ( i ) the @xmath5-step lanczos bidiagonalization always generates a near best rank @xmath5 approximation to @xmath3 ; ( ii ) the @xmath5 ritz values always approximate the first @xmath5 large singular values in natural order ; ( iii ) the @xmath5-step lsqr always captures the @xmath5 dominant svd components of @xmath3 , so that lsqr can find a best possible regularized solution ; ( iv ) the diagonals and subdiagonals of the bidiagonal matrices generated by lanczos bidiagonalization decay as fast as the singular values of @xmath3 . however , for the third kind of problem , the above results do not hold generally . the decay rates of diagonals and subdiagonals of the bidiagonal matrices can be used to decide if lsqr can find a best possible regularization solution .
we also analyze the regularization of other two krylov solvers lsmr and cgme that amount to minres and the cg method and minres applied to @xmath6 and @xmath7 with @xmath8 , respectively , proving that lsmr has similar regularizing effects to lsqr for each kind of problem and both are superior to cgme .
we extend all the results to the case that @xmath3 has multiple singular values .
numerical experiments confirm our theory on lsqr . discrete ill - posed , full or partial regularization , best or near best rank @xmath5 approximation , tsvd solution , ritz value , lanczos bidiagonalization , lsqr , cgls , lsmr , cgme 65f22 , 65f10 , 65j20 , 65r30 , 15a18 |
massive stars affect the structure and dynamics of galaxies , the planet formation process , and even the formation of low mass stars via outflows , winds , and supernova explosions that drive mixing and turbulence in the interstellar medium ( e.g. , @xcite , and references therein ) . yet
, our understanding of their formation and evolution remains unclear .
moreover , the importance of magnetic fields in the formation of stars , both low and high mass , has long been acknowledged ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ) .
indeed , ongoing theoretical work continues to probe the effects of magnetic fields on the star formation process ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
however , observational data on magnetic fields is still scarce and theoretical work is hampered by the lack of information on magnetic field strengths in different density regimes . with these considerations in view
, we have observed the high mass star forming region s88b with the karl g. jansky very large array ( vla ) for the zeeman effect in thermal absorption lines of oh at 1665 and 1667 mhz .
observations of the zeeman effect in absorption lines with interferometers like the vla provide an excellent method of @xmath1 the magnetic field in regions that are along the line of sight toward strong background continuum sources ( e.g. , @xcite ; @xcite ) . in the next subsection
, we will present some details of the source s88b . in
[ sodr ] , we present details of the observations and reduction of the data .
the results are presented in
[ sr ] , and discussed in [ sdisc ] . in
[ sconc ] , we state our conclusions .
s88b is a high mass star forming region ( near @xmath2 ) with a bolometric luminosity of @xmath3 ( @xcite ) .
it has been observed at a wide variety of wavelengths .
there is some confusing nomenclature in the literature , and so we begin by discussing the morphology of its surroundings .
s88 is itself a diffuse extended nebula , and is likely a very evolved region ( e.g. , @xcite ) .
about 14 to its southeast lie two bright knots of optical nebulosity : s88a and s88b ( e.g. , see the red palomar sky survey print in fig . 1 of @xcite ) .
no radio continuum emission is observed from the western source , s88a .
strong radio emission is observed from an region centered slightly to the east of the optical nebulosity s88b , and this is the target source for our zeeman observations .
@xcite first proposed that the radio source in s88b was composed of two components , based on their 5 ghz continuum observations .
the more extended component to the west ( e.g. , see
[ ssc ] , and fig . 1 in this paper ) that intersects the optical nebulosity is s88b-1 , whereas the compact component to the east is s88b-2 . in their high resolution 6
cm vla image ( hpbw 0.4 ) , @xcite resolved out many details of these two sources what they call g61.48 + 0.09a is referenced as s88b-2 in this paper ( and most of the literature ) , whereas their source g61.48 + 0.09b is a group of low intensity sources to the southwest of s88b-1 .
following @xcite , we adopt the distance to s88b of 2.0 kpc .
more recently , @xcite have claimed a distance of 4.1 @xmath4 2.4 kpc to s88b on the basis of their h110@xmath5 recombination line and h@xmath6co absorption observations ; we note that our adopted value is within their error limits .
moreover , @xcite have estimated on the basis of foreground star counts that the distance to s88b can not be substantially greater than 2.0 kpc .
the observations were carried out with the very large array ( vla ) of the nrao in 2003 in the b - configuration , and combined with c - configuration data observed in 1997 .
important parameters of the observations are listed in table [ top ] .
both circular polarizations and both main lines ( 1665 and 1667 mhz ) of oh were observed simultaneously in absorption . in order to mitigate instrumental effects ,
a front - end transfer switch was used to alternate the sense of circular polarization passing through each telescope s if system every @xmath710 minutes .
the editing , calibration , fourier transformation , deconvolution , and calculation of optical depths for the oh data were carried out using the astronomical image processing system ( aips ) of the nrao . following standard procedure for zeeman effect observations , bandpass correction was applied only to the stokes @xmath8 data set , since bandpass effects subtract out to first order in the stokes @xmath9 data , and the bandpass correction process adds noise to the @xmath9 profiles .
the b and c configuration data were combined in the uv - plane using the aips task @xmath10 .
the rms in the continuum image of the combined b and c configuration data is 0.8 mjy @xmath11 , whereas the rms in a channel of the line data is 1.7 mjy @xmath11 .
magnetic fields were determined as described in
[ ssb ] by using the multichannel image reconstruction , image analysis and display ( miriad ) software processing package , formerly of the berkeley - illinois - maryland array ( bima ) and now incorporated into the combined array for research in millimeter - wave astronomy ( carma ) telescope .
figure [ fc ] shows our highest resolution 18 cm continuum image of s88b made with uniform weighting ( beam hpbw @xmath12 ) .
the extended western component s88b-1 and the compact component on the eastern side s88b-2 , first identified by @xcite , are clearly visible in this figure .
we determined the integrated flux for s88b-1 to be @xmath13 jy , and the integrated flux for s88b-2 to be @xmath14 jy . in a 5 ghz continuum survey with the nrao 140-ft telescope ( hpbw 6@xmath155 ) ,
@xcite obtained 6.1 jy for the source 61.5 + 0.1 which is a combination of s88b-1 and s88b-2 . in a 12-hr synthesis map with westerbork ( wsrt ) at 5 ghz ( hpbw 6@xmath166 ) ,
@xcite obtained a total flux density of 6.1 @xmath4 0.2 jy for s88b . at 4.9 ghz ,
@xcite measured 4.4 jy toward s88b-1 and 0.8 jy toward s88b-2 in a combined c and d configuration observation with the vla ( hpbw @xmath17 ) .
the integrated flux densities measured with these shorter wavelength data are in good agreement with the 18 cm vla results , suggesting that the emission is predominantly either near the turnover point or optically thin ( @xmath18 , @xmath19 )
. figure [ fcc ] shows the 350 @xmath0 m dust continuum image of @xcite superposed on our 18 cm continuum image , along with other pertinent information these are discussed in [ ssdc ] .
the oh absorption profiles toward s88b are complicated , with several blended components toward a number of positions . in order to characterize the velocity structure , optical depth profiles for
the oh lines were calculated with the aips task @xmath20 using standard procedure ( e.g. , see @xcite ) .
the oh column density was then determined using the relation @xmath21 where @xmath22 is the excitation temperature of oh and the constant @xmath23 and @xmath24 @xmath25 ( km @xmath26)@xmath27 for the oh 1665 and 1667 mhz lines respectively ( @xcite ) .
the optical depth profiles reveal complex kinematics with a number of heavily blended components present toward much of the region . in order to highlight the prominent absorption components , contour plots of the oh optical depth in several velocity channels
are shown in figure [ ftau ] , superposed on a grayscale image of the 18 cm continuum .
every third velocity channel is displayed , and the plots are shown over the velocity range 17.4 km @xmath28 to 24.0 km @xmath28 .
figure [ fnoh ] shows plots of @xmath29(oh)/@xmath22 toward s88b at 1667 mhz . in principle , variations in @xmath29(oh)/@xmath22 may arise due to variations in the oh column density , or due to variations in @xmath22 , or both .
however , in the results and discussion , we quote only variations in @xmath29(oh)/@xmath22 , since the excitation temperature can not be measured based on absorption studies alone . further discussion about possible variations in @xmath22 appears in [ ssdc ] . from figures [ ftau ] and [ fnoh ] , we identified four prominent absorption enhancements , and these are marked in figure [ fnoh ] as oh - a , oh - b , oh - c , and oh - d respectively .
the absorption enhancement oh - a lies to the southwest of the continuum peak of s88b-1 ; it is seen in the 19.1 km @xmath28 panel in figure [ ftau ] , with a peak @xmath30 .
oh - a is also seen in the 19.9 km @xmath28 panel of figures [ ftau ] with a higher peak @xmath31 . in this 19.9 km @xmath28 panel
, we also see the enhancements oh - b and oh - c , the former to the northwest of the continuum peak of s88b-1 , and the latter about midway between s88b-1 and s88b-2 .
finally , oh - d is seen in the 20.7 , 21.5 , and 22.4 km @xmath28 panels of figure [ ftau ] , with high peak @xmath32 in two of these panels ; it lies northward of the s88b-2 continuum peak ; oh - d is also visible at the highest velocities at which oh absorption is observed ( @xmath33 km @xmath28 ) . the high @xmath34 s in oh - d account for the peak in @xmath29(oh)/@xmath22 to the north of the continuum peak of s88b-2 . the magnetic field strengths in s88b were determined using the zeeman effect , by fitting a numerical frequency derivative of the stokes @xmath8 spectrum to the stokes @xmath9 spectrum for each pixel in the absorption line cube .
the stokes parameters have the standard definition of @xmath8 = ( rcp + lcp ) , and @xmath9 = ( rcp @xmath35 lcp ) ; rcp is right- and lcp is left - circular polarization ; rcp has the standard radio definition of clockwise rotation of the electric vector when viewed along the direction of wave propagation . since the observed @xmath9 spectrum may also contain a scaled replica of the @xmath8 spectrum itself , the stokes @xmath9 spectral profile can be described by the equation ( @xcite ; @xcite ) : @xmath36 in this equation , the fit parameter @xmath37 represents the scaled - down replica of the @xmath8 profile in the @xmath9 profile ; @xmath37 is usually the result of small calibration errors in @xmath38 vs. @xmath39 ; for all results reported here , @xmath40 . for thermally excited lines such as the oh 1665 and 1667 mhz lines reported in this paper , @xmath41 , where @xmath42 is the magnetic field , @xmath43 is the angle of the magnetic field to the line of sight , and @xmath44 is the zeeman splitting factor which is equal to @xmath44 = 3.27 hz @xmath0g@xmath27 for the 1665 mhz line of oh , and 1.96 hz @xmath0g@xmath27 for the 1667 mhz line of oh ( @xcite ) .
we determined magnetic field strengths in s88b by using a least- squares method to fit equation ( [ evlco ] ) to the stokes @xmath9 spectra to solve for @xmath37 and @xmath45 .
the results of the fits give the line - of - sight component of the magnetic field , @xmath46 = @xmath47 .
it is usual practice to consider the results to be significant only if the derived value of @xmath46 is greater than the 3@xmath48 level . for zeeman effect observations in oh absorption lines ,
we usually impose a stronger condition we consider the detections to be significant only if both 1665 and 1667 mhz @xmath46 values are greater than the 3@xmath48 level .
in the current observations , however , this stronger criterion for significance would leave out some important information on the field . toward s88b-2 ,
the 1665 mhz profiles show a @xmath7200 @xmath0 g field with about 5@xmath48 level of significance , whereas the 1667 mhz profiles show a field below the 3@xmath48 level .
therefore , for s88b-2 only , we have relaxed the criterion above to consider the measurement to be significant if it displays a 3@xmath48 detection only in the 1665 mhz line .
we believe this is reasonable , since the measured value of @xmath46 toward this source does not appear to be wildly discrepant from the measured value toward s88b-1 .
the non - detection at 1667 mhz toward s88b-2 is not surprising .
the 1667 mhz stokes v signal is weaker than the 1665 mhz stokes v signal because the zeeman splitting factor for oh 1667 mhz lines ( @xmath44 = 1.96 hz @xmath0g@xmath27 ) is less than that for oh 1665 mhz lines ( @xmath44 = 3.27 hz @xmath0g@xmath27 ) .
this inequality in the zeeman factors would not impact the result as much if the oh gas was thermalized and homogenous , since the intensity ratios would then be @xmath49 .
however , as described below ( [ ssdc ] ) , @xmath50 toward s88b , which makes the sensitivity of the 1667 mhz line to the zeeman effect less than that of the 1665 mhz line . the resulting @xmath46 map ( taken from the 1665 mhz data ) is shown in figure [ fblos ] . in the area enclosed by the thick - lined ellipse in this figure ,
the detected field is above the 3@xmath48 level only in the 1665 mhz line , whereas in the rest of the displayed @xmath46 map , the field is above the 3@xmath48 level in both 1665 and 1667 mhz lines , with peak signal - to - noise ratios of @xmath710 .
examples of the stokes @xmath8 and @xmath9 profiles together with the derivative of @xmath8 scaled by the fitted value of @xmath46are shown in figure [ fb1ivd ] and figure [ fb2ivd ] ; the locations toward which these profiles have been plotted are marked in figure [ fblos ] by a `` + '' and a `` @xmath51 '' respectively . from figure [ fb1ivd ] and
figure [ fb2ivd ] , it is evident that the magnetic field is detected in the component with @xmath52 km @xmath28 ; no fields were detected in any of the other velocity components . toward s88b-1 , we see from figure [ fblos ] that @xmath46 increases from about 90 @xmath0 g in the southwest to about 400 @xmath0 g in the northeast ; this is the case in both 1665 and 1667 mhz lines .
meanwhile , toward s88b-2 , the 1665 mhz line reveals values of @xmath46 as high as @xmath7270 @xmath0 g .
early optical , radio , and infrared observations of s88b by @xcite revealed that the peak of the near infrared ( nir ) position is eastward of the optical nebula , whereas the radio continuum peak is even farther to the east than the nir peak .
this eastward progression is consistent with our observation of the location of the absorbing oh gas .
it is clear from figure [ ftau ] that almost all of the oh absorbing gas is located in the eastern half of s88b ( i.e. , all of s88b-2 , and the eastern part of the extended source s88b-1 ) .
we did not detect any oh absorption in the western parts of s88b-1 , where it is coincident with the h@xmath5 nebulosity . the molecular cloud associated with s88b
was mapped by @xcite in several co transitions at low resolution ( hpbw 1@xmath53 - 2@xmath53 ) , and found to extend over 10@xmath53 ( 6.5 pc ) with a total mass of @xmath54 m@xmath55 .
@xcite speculated that the regions s88b-1 and s88b-2 are a blister formation in the molecular cloud which we are observing from the side .
the velocities of the absorbing oh gas match up well with molecular line velocities ; e.g. , the co lines are singly peaked at 21 km @xmath28 ( @xcite ) .
moreover , recombination line observations by @xcite with the vla ( hpbw @xmath7 2@xmath56 ) revealed that the center velocity of the single - peaked h92@xmath5 line emission toward s88b-1 increases from 22 km @xmath28 at the eastern edge of s88b-1 to 32 km @xmath28 at the western edge .
the 22 km @xmath28 velocity in the h92@xmath5 recombination line at the eastern edge of s88b-1 matches well with the oh absorption data e.g. , see the panels between 21.5 km @xmath28 and 22.4 km @xmath28 in figure [ ftau ] . on the other hand , the recombination line velocity at the western edge matches well with the optical h@xmath5 velocities which exceed 28 km @xmath28 ( @xcite ) .
in other words , we are observing an edge - on transition from the ionized gas to the molecular gas in s88b , similar to m17 ( @xcite ) .
while the velocity structure of oh absorption observed by us toward s88b is complicated , the highest values for @xmath57 are to the north of the continuum peak s88b-2 .
while this may , in principle , be due to variations in @xmath58 or @xmath59 or both , other tracers indicate that the column of absorbing gas is indeed greater in the eastern parts of s88b ( i.e. , toward s88b-2 ) . toward the h@xmath5 peak in the west , @xcite found a relatively low @xmath60 = 5 mag . based on their observations of ir fine structure lines ,
@xcite measured @xmath60= 26 @xmath4 9 mag at the position of the 2.2 nir @xmath0 m peak .
this 2.2 @xmath0 m peak was resolved by @xcite into a system of three stars ( 82 , 83 , 84 ) located on the eastern edge of the optical nebulosity their positions are marked in figure [ fcc ] .
in contrast , toward s88b-2 in the east , @xcite found @xmath60@xmath7 70 mag by comparing their observed br@xmath5 flux with 5 ghz radio continuum data from the literature .
s88b-2 also coincides with the 350 @xmath0 m dust continuum peak from @xcite , as shown in figure 2 .
there is also an extension in the 350 @xmath0 m dust continuum image to the northwest , which coincides with the location of s88b-1 ( fig . 2 ) .
in fact , due to the high obscuration toward the eastern parts of s88b , a complete census of the exciting / ionizing stars of s88b-1 and s88b-2 has proved impossible so far .
@xcite suggested that star 82 could be the ionizing / exciting source of s88b-1 , based on its high luminosity and central location ( see fig .
[ fcc ] ) , but admitted the likelihood that other stars also contribute to the excitation / ionization in the region .
other possible candidates for exciting sources are the stars l1 ( for s88b-2 ) and l2 ( for s88b-1 ) , detected by @xcite at a near - infrared wavelength of 3.5 @xmath0 m ( l@xmath61-band ) and marked in figure [ fcc ] ; however , the issue remains open because @xcite could not determine the spectral type of l1 or l2 as they did not detect them in their j , h , or k@xmath61 near - infrared bands .
if the absorbing oh gas is thermalized and homogenous ( i.e. , not clumpy within the observing beam ) , the ratio of the optical depths of the 1665 mhz and 1667 mhz lines should be @xmath62 .
however , the observed profiles toward s88b do not show this ratio . instead , observed ratios @xmath63 are in the range 1.0 - 1.3 over most parts of the source . while it is likely that the excitation temperatures for the two lines at 1665 mhz and 1667 mhz are not the same , it is unlikely that this difference would cancel out the expected ratio of optical depths over the whole source .
more likely , the oh gas is clumped on size scales smaller than the beam .
such clumpiness has been found also in molecular emission lines .
the co j=2@xmath351 and @xmath64co j=2@xmath351 and j=1@xmath350 data ( hpbw 22@xmath56 - 33@xmath56 ) of @xcite reveal a horseshoe - like structure surrounding the optical nebulosity , which they ascribe to material excited along the periphery of an outflow cavity that encloses the optical nebula . indeed , @xcite have reported a low collimation bipolar outflow extending over 3@xmath53 from their co observations ; this co outflow was also observed by @xcite .
both the co and @xmath64co images in @xciteshow highly fragmented structure , in agreement with the clumpy structure suggested by the oh absorption observations .
more evidence of clumping comes from observations of nh@xmath65 lines at high angular resolution ( hpbw 4@xmath56 ) with the vla by @xcite .
they found compact ammonia structures with sizes of 0.2 pc toward both s88b-1 and s88b-2 .
they suggest that these ammonia clumps correspond to compact molecular structures embedded within the larger molecular cloud , and that these clumps have been heated by the radiation from the star(s ) that ionize(s ) the associated regions .
finally , recent c@xmath66o observations by @xcite with a beam hpbw of 15 provide additional evidence of clumpy structure ( also see [ ssmfe ] below ) ; their clump b is marked in figure [ fcc ] .
the oh excitation temperature @xmath22 ( which will be required in the calculations below ) can not be measured based on absorption studies alone , but reasonable limits for it can be set from our observations , and a likely value estimated based on other tracers .
the minimum continuum @xmath67 against which the line is detected is @xmath7 10 mjy @xmath11 we can use this to set an upper limit on @xmath22 . using @xmath68
1 mjy @xmath11@xmath69 17 k , we get an upper limit of 170 k for @xmath22 .
it is also unlikely that @xmath22 is less than the typical dark cloud value of 10 k. estimates of the kinetic temperature from other molecular tracers are also relevant if we assume lte . from observations of the ( 2,2 ) and ( 3,3 ) lines of nh@xmath65
, @xcite found @xmath70 = 80 k toward s88b-1 ( their western nh@xmath65 clump ) and 70 k toward s88b-2 ( their eastern nh@xmath65 clump ) .
based on co observations , @xcite estimated @xmath70 @xmath7 60 k. @xcite found a dust temperature for s88b equal to @xmath71 = 42 k. @xcite fitted a 2-component dust model for s88b , and obtained @xmath71 = 40 k and 116 k. the higher @xmath72 may be due to compact embedded sources or it may reflect an enhanced population of small grains ( @xcite ) .
indeed , @xcite have reported observations of pah emission features toward s88b .
@xcite found t@xmath73 = 75 k from their 350 @xmath0 m dust continuum observations .
we adopt @xmath22 = 60 k for our calculations , but note that @xmath22 likely varies over the source ; still , it must lie between 10 - 170 k. our observations of the zeeman effect in oh absorption lines at 1665 and 1667 mhz have yielded @xmath46 values in the range of 90 to 400 @xmath0 g .
these values are typical of magnetic fields in molecular clouds detected via the zeeman effect in the density regimes accessible via and oh thermal lines , as tabulated in @xcite .
note , however , that our detected fields are higher than the single dish measurements by @xcite , who found @xmath46 = 69 @xmath4 5 @xmath0 g toward s88b with the nancay single dish .
this is expected , since the low angular resolution ( hpbw @xmath74 ) of nancay will cause fields to be averaged over a large area , leading to much lower @xmath46 values .
the magnetic field in a cloud may consist of a static component @xmath75 , and a time - dependent or wave component @xmath76 .
the static component connects the cloud to the external medium and determines the total magnetic flux throughout the cloud , but it can only provide support to the cloud perpendicular to the field lines .
the wave component is associated with mhd waves in the cloud , and it can provide three dimensional support to the cloud . in principle , the observed @xmath46may consist of contributions from each of these two components . in reality , however , the zeeman effect likely traces the strength of the static component of the magnetic field only .
this is because spatial averaging over the beam and along the line of sight tends to cancel the contribution from the wave component ( e.g. , @xcite , and references therein ) . a principal goal of zeeman effect measurements , therefore
, is to estimate the importance of the magnetic field to the dynamics and evolution of star forming regions like s88b . as summarized in @xcite
, such estimates also require other physical parameters such as internal velocity dispersion , hydrogen column density , and radius , to be known .
the relevant physical parameters mentioned above for estimating the importance of the magnetic field are available either from the present observations of s88b or from the literature .
first , we must decide on an appropriate value for the radius .
the highest resolution molecular line emission observations available to date are the c@xmath66o observations of @xcite referred to above ( with beam hpbw 15@xmath56 ) .
clumps b , c , and d in their observations all have radius @xmath7 0.2 pc . therefore
, we adopt @xmath77 pc for the radius in all the following calculations .
next , the hydrogen column density can be estimated from the oh line data , subject to certain assumptions .
we use an average value of @xmath78/@xmath22 = 6.0 @xmath51 10@xmath79 @xmath80 k@xmath27 ( see fig . [ fnoh ] ) . for the excitation temperature ,
we use @xmath22 = 60 k , based on the discussion in [ ssdc ] above .
then , to get @xmath81 , we use the conversion ratio @xmath82 from @xcite for s106 .
this makes the oh abundance a factor of 10 greater than that in dark clouds found by @xcite .
note , however , that if we consider the standard conversion ratio @xmath83 , and use @xmath84 = 70 mag for s88b ( see [ ssdc ] above ) , we obtain @xmath85 , which justifies our use of the @xcite value rather than the dark cloud value ; similar results were found for m17 by @xcite . using this conversion ratio , we get @xmath86 @xmath80 .
this gives a hydrogen density @xmath87 @xmath88 , which compares well with the value of @xmath89 @xmath88 found by @xcite for the clump designated as @xmath42 in their c@xmath66o observations ( see fig . [ fcc ] ) .
it is worth noting here that at lower resolution ( hpbw 30@xmath56 ) , @xcite determined a mean @xmath90 @xmath88 .
together , this is further proof of the clumping in s88b discussed in [ ssdc ] above , since the larger beam filling factor in a higher angular resolution observation of a clumpy region will naturally give a larger value for the density .
note , however , that the density can not be substantially greater than @xmath710@xmath91 @xmath88 , a conclusion reached by @xcite based on the absence of h@xmath6co emission ( @xcite ) toward s88b . other parameters of interest are the velocity dispersion ( which can be obtained from the observed line width ) and the magnetic field ; the adopted values for the average observed line width of oh lines and the average @xmath46 in s88b are listed in table [ tparms ] . for the calculations below , we will follow @xcite , and use total ( static ) magnetic field strength equal to 2 times @xmath46
. we can now proceed to assess the importance of the magnetic field in s88b , using a number of approaches .
first , we consider the parameter @xmath92 , which is the ratio of thermal to magnetic pressures ( @xcite ) ; it is a crucial parameter in any theory or simulation of the structure and evolution of molecular clouds that incorporates magnetic fields . if @xmath93 , then magnetic fields are important ; the lower the value of @xmath92 , the more dominant is the magnetic field .
based on the adopted values in table [ tparms ] , we find @xmath94 in s88b , indicating the importance of the magnetic field in this region .
our value for @xmath92 in s88b is similar to that in high mass star forming regions like dr 21 oh 1 , where @xmath95 , and w3 ( main ) , where @xmath96 ( @xcite , and references therein ) . another way to assess
the importance of the magnetic field in s88 is to use the relation @xmath97 where @xmath98 is the average proton column density in the cloud .
equation [ ebs ] is a result of equating the static magnetic energy of the cloud to its gravitational energy ; @xmath99 is then the average static magnetic field in the cloud that would completely support it against self - gravity . if the actual static magnetic field in the cloud @xmath100 , then the cloud can be completely supported by the magnetic field , and is said to be magnetically subcritical .
further evolution of the cloud perpendicular to the field lines will occur primarily via ambipolar diffusion .
the field can be judged to be dynamically important to the region even if it is less than @xmath99 , but is comparable to it . for @xmath86 @xmath80 from table [ tparms ]
above , we get @xmath101 g . our average total ( static ) magnetic field strength (= 2 times @xmath46 in table [ tparms ] ) is 400 @xmath0 g ; that is , the observed magnetic field strength is within a factor of @xmath72 less than the critical field .
therefore , the magnetic field should be dynamically significant , providing an important source of support against self gravity .
yet another parameter to judge the extent to which a static magnetic field can support a cloud against gravitational collapse is the observed mass to magnetic flux ratio @xmath102}$ ] , defined by @xcite in units of the critical value .
this is equivalent to the consideration of @xmath99 , and may be redundant here , but we state it nevertheless to express our results in this alternative language . if the ratio of observed @xmath102}$ ] to the critical value is greater than 1 , then the region is magnetically supercritical , meaning that the magnetic field can not by itself provide support against gravitational collapse .
for s88b , we find the observed @xmath102}$ ] in units of the critical value to be 1.8 .
this is consistent with observations to date ; @xcite found the observed @xmath102}$ ] in terms of the critical value to have an average value equal to 2 in a summary of available zeeman measurements in molecular clouds .
the calculated value for s88b indicates that whereas the static magnetic field by itself is not enough to support the molecular cloud against gravitational collapse , it is high enough to provide a significant source of support .
perhaps more important , however , in a highly supersonic medium like the s88b molecular cloud , is that for the magnetic field to be important , its energy density must be comparable to , or greater than , the turbulent energy density .
the supersonic nature of the medium is clear from the value of the sonic mach number @xmath103 given in table 3 ; from it , we have that the ratio of the one - dimensional velocity dispersion ( @xmath48 ) to the isothermal sound speed ( @xmath104 ) in s88b is @xmath105 .
this is a common result in molecular clouds ; in his compilation of results for 27 molecular clouds , @xcite found that motions are supersonic by about a factor of 5 .
table 3 also lists the alfvenic mach number ( @xmath106 ) to be equal to 1 .
this implies that the magnetic energy density is comparable to the turbulent energy density , that is , these two energies are in approximate equipartition .
finally , we can use the measured @xmath46 to assess the relative importance of the gravitational , kinetic , and magnetic energy terms @xmath107 , @xmath108 , and @xmath109 , in the virial equation ( @xcite ) . to calculate their values
, we use the expressions for @xmath107 , @xmath108 , and @xmath109 given in @xcite ; the values are given in table [ tve ] .
we find that @xmath110 , again indicating that magnetic fields are a significant , but not the sole , means of support against gravitational collapse .
finally , we find that @xmath111 ; however , the surface pressure term in the virial equation , which is not included here , will act in the same sense as gravity ; therefore , s88b is in approximate virial equilibrium .
we have observed oh thermal lines in absorption at 1665 and 1667 mhz toward the star forming region s88b with the aim of mapping magnetic fields via the zeeman effect .
the oh absorption line profiles toward this source are complicated , and contain several blended components toward a number of positions .
the oh absorbing gas is located in the eastern parts of s88b , toward s88b-2 and the eastern section of s88b-1 .
the ratio of optical depths @xmath50 over most parts of the source , instead of the value of 1.8 for thermalized and homogenous oh gas .
this is most likely due to clumping of oh gas , a conclusion supported by observations of co , @xmath64co , nh@xmath65 and c@xmath66o emission lines from the literature . from the @xmath57 plots
, we identified 4 prominent oh clumps , which we have designated as oh - a , oh - b , oh - c , and oh - d .
the highest values for @xmath57 are to the north of the continuum peak s88b-2 .
although this could , in principle , be due to an enhancement in @xmath58 or @xmath22 ( or both ) , it is likely that north of s88b-2 is truly the location of the largest column of absorbing oh gas , as supported by independently measured extinction values toward s88b from the literature .
based on the observed quantities and adopted values of parameters from the literature , we obtain a gas density @xmath87 @xmath88 , which compares well with the value of @xmath112 @xmath88 found by @xcite for a c@xmath66o clump coincident with the s88b-1 and s88b-2 sources . in brief , s88b appears to have an overall geometry similar to m17 , in that there is an edge - on progression from ionized to molecular gas going eastward .
our magnetic field map appears to mirror this transition , at least over the extended s88b-1 region , where we detected line - of - sight magnetic fields ranging from 90 - 400 @xmath0 g , with the lowest values to the southwest of the s88b-1 continuum peak , and the highest values to its northeast . toward s88b-2 ,
the fields are significant at the 3-@xmath48 level in the 1665 mhz line only , and are detected over a small region but if we had the sensitivity to keep going eastward , the field would likely follow the same increasing trend .
the detected fields allow us to assess the importance of the magnetic field in s88b , using a number of methods .
we find that the ratio of thermal to magnetic pressures , @xmath113 ; a value of @xmath93 means that magnetic fields are important ; the lower the value of @xmath92 , the more dominant the magnetic field . a second method to assess
the importance of the magnetic field is to compare the observed magnetic field with the critical field that would be required to completely support the cloud against its self - gravity .
we find that the observed magnetic field is within a factor of 2 less than the critical field .
a third method to assess the importance of the magnetic field , equivalent to the second , is to find the ratio of observed mass to magnetic flux in terms of the critical value ; for s88b , we find this to be 1.8 .
these considerations lead us to conclude that the magnetic field in s88b is dynamically significant , and provides an important source of support against gravity .
moreover , the magnetic energy density is comparable to the turbulent energy density , implying that these two energies are in approximate equipartition , and pointing again to the importance of the magnetic field in this region .
this work has been partially supported by a cottrell college science award ( ccsa ) grant from research corporation to a.p.s . , and
by start - up funds to a.p.s . at depaul university . t.h.t .
acknowledges nsf grant ast 03 - 07642 .
we thank kaisa young ( _ nee _ mueller ) and h. saito for sending us digital versions of their data for comparison .
we have used extensively the nasa astrophysics data system ( ads ) astronomy abstract service , the astro - ph web server , and the simbad database .
lccccccrrrrr date & 2003 nov 28/29 , dec 1/2 & 1997 aug 15/16 configuration & b & c r.a . of field center ( j2000 ) & @xmath114.7 & @xmath114.7 decl . of field center
( j2000 ) & 25@xmath11512@[email protected] & 25@xmath11513@[email protected] total bandwidth ( mhz ) & 0.20 & 0.20 no . of channels & 128 & 128 channel spacing ( km @xmath26 ) & 0.28 & 0.28 approx .
time on source ( hr ) & 6.7 & 8.9 rest frequency ( mhz ) & 1665.402 & 1665.402 & 1667.359 & 1667.359 lccccrrrrr radius ( @xmath116 ) & & 0.2 pc column density ( @xmath117 ) & & @xmath118 particle density ( @xmath119 ) & & @xmath120 kinetic temperature ( @xmath121 ) & & 60 k adopted average fwhm linewidth ( @xmath122 ) & & 2.5 km s@xmath27 average magnetic field ( @xmath123 ) & & 200 @xmath0 g 1 @xmath51 3@xmath160 ( p.a.=@xmath1271 ) , and the rms in the image is 0.8 mjy @xmath11 .
the contours are at 6 , 12 , 24 , 48 , 96 mjy @xmath11 .
the extended western component s88b-1 , and the compact component to the east s88b-2 , are marked in the figure as b1 and b2 respectively .
[ fc],scaledwidth=90.0% ] m dust emission toward s88b from @xcite superposed on the 18 cm continuum image .
the hpbw of the 350 @xmath0 m data is 14@xmath56 , and contours are at 1.4 , 2 jy beam@xmath27 , and then in increments of 1 jy beam@xmath27 up to 10 jy beam@xmath27 .
our 18 cm continuum image has the same resolution and contours as in figure [ fc ] .
the positions of the nir sources 82 , 83 , and 84 from @xcite , the 3.5 @xmath0 m sources l1 and l2 from @xcite , and clump b from the c@xmath66o observations of @xcite are also marked in the figure .
[ fcc],scaledwidth=90.0% ] /t@xmath128 toward s88b overlaid on the 18 cm continuum .
the continuum image has the same resolution and contours as in fig .
the 1667 oh beam is shown in the lower right ; its parameters are given in fig .
[ fnoh],scaledwidth=90.0% ] toward s88b , taken from the 1665 mhz data .
the contours depict the 18 cm continuum , and are the same as in fig [ fc ] .
the thick - lined ellipse shows the region in which the detected @xmath46 is at the 3-@xmath48 level in 1665 mhz only .
the `` + '' sign indicates the position toward which the profiles are shown in fig .
[ fb1ivd ] below , while the `` @xmath51 '' indicates the position toward which the profiles are shown in fig .
[ fb2ivd ] .
[ fblos],scaledwidth=90.0% ] ( @xmath129 ) and @xmath9 ( @xmath130 ) profiles at 1665 mhz ( left ) and 1667 mhz ( right ) toward the position in s88b-1 marked in fig .
[ fblos ] by a `` + '' ( @xmath131 ) .
the curve superposed on @xmath9 in the lower frame on the left shows the derivative of @xmath8 at 1665 mhz scaled by @xmath46 = @xmath132 g , whereas the curve superposed on @xmath9 in the lower frame on the right shows the derivative of @xmath8 at 1667 mhz scaled by @xmath46 = @xmath133g.[fb1ivd],scaledwidth=90.0% ] ( @xmath129 ) and @xmath9 ( @xmath130 ) profiles at 1665 mhz ( left ) and 1667 mhz ( right ) toward the position in s88b-1 marked in fig .
[ fblos ] by a `` + '' ( @xmath131 ) .
the curve superposed on @xmath9 in the lower frame on the left shows the derivative of @xmath8 at 1665 mhz scaled by @xmath46 = @xmath132 g , whereas the curve superposed on @xmath9 in the lower frame on the right shows the derivative of @xmath8 at 1667 mhz scaled by @xmath46 = @xmath133g.[fb1ivd],scaledwidth=90.0% ] ( @xmath129 ) and @xmath9 ( @xmath130 ) profiles at 1665 mhz toward the position in s88b-2 marked in fig .
[ fblos ] by a `` @xmath51 '' ( @xmath134 ) .
the curve superposed on @xmath9 in the lower frame shows the derivative of @xmath8 at 1665 mhz scaled by @xmath46 = @xmath135 g ; the fit for the field was made over a restricted range of velocity channels , and the channels over which the fit was made are indicated by the solid line .
[ fb2ivd],scaledwidth=50.0% ] | we present observations of the zeeman effect in oh thermal absorption main lines at 1665 and 1667 mhz taken with the karl g. jansky very large array ( vla ) toward the star forming region s88b .
the oh absorption profiles toward this source are complicated , and contain several blended components toward a number of positions .
almost all of the oh absorbing gas is located in the eastern parts of s88b , toward the compact continuum source s88b-2 and the eastern parts of the extended continuum source s88b-1 .
the ratio of 1665/1667 mhz oh line intensities indicates the gas is likely highly clumped , in agreement with other molecular emission line observations in the literature .
s88-b appears to present a similar geometry to the well - known star forming region m17 , in that there is an edge - on eastward progression from ionized to molecular gas .
the detected magnetic fields appear to mirror this eastward transition ; we detected line - of - sight magnetic fields ranging from 90 - 400 @xmath0 g , with the lowest values of the field to the southwest of the s88b-1 continuum peak , and the highest values to its northeast .
we used the detected fields to assess the importance of the magnetic field in s88b by a number of methods ; we calculated the ratio of thermal to magnetic pressures , we calculated the critical field necessary to completely support the cloud against self - gravity and compared it to the observed field , and we calculated the ratio of mass to magnetic flux in terms of the critical value of this parameter .
all these methods indicated that the magnetic field in s88b is dynamically significant , and should provide an important source of support against gravity .
moreover , the magnetic energy density is in approximate equipartition with the turbulent energy density , again pointing to the importance of the magnetic field in this region . |
in a recent paper , levin @xcite discusses the set of results presented over the last decade by various prominent physicists which led to the conclusion that black holes seem to be susceptible to chaos .
levin argues that the most realistic description available of a spinning pair of black holes is chaotic motion , and goes on to complain that in physics and cosmology chaos has not received the attention it deserves in part because the systems studied have been highly idealized . in contrast , in economics we have the interesting fact that even some of the most simple and highly idealized models describing modern economies can easily lead to chaotic dynamics .
in this paper we apply the techniques of symbolic dynamics to the analysis of a labor market which shows in almost developed economies large volatility in employment flows .
the possibility that chaotic dynamics may arise in modern labor markets had been totally strange to economics until recently , at least as far as we are aware of . in an interesting paper , bhattacharya and bunzel @xcite
have found that the discrete time version of the pissarides - mortensen matching model , as formulated in ljungqvist and sargent @xcite , can easily lead to chaotic dynamics under standard sets of parameter values .
however , in order to conclude about the existence of chaotic dynamics in the three numerical examples presented in the paper , the authors apply the li - yorke theorem or the mitra sufficient condition which should not be generally applied to all specific simulations because they may lead to misleading conclusions .
moreover , in a more recent version of the paper , bhattacharya and bunzel @xcite present new results in which chaos is completely removed from the dynamics of the model .
this paper explores the matching model so interestingly developed by bhattacharya and bunzel with the following objectives in mind : ( i ) to show that chaotic dynamics may still be present in the model for standard parameter values , for high values of the measure of labor tightness ( that is the ratio of vacancies to the number of workers looking for jobs ) which can occur in economic booms ; ( ii ) to clarify some open questions raised by the authors in their first paper , by providing a rigorous proof of the existence of chaotic dynamics in the model through the computation of topological entropy in a symbolic dynamics setting .
therefore , if one is studying whether there are chaotic dynamics or not under certain ranges of parameters values , we suggest that a bifurcation diagram , the variation of the lyapunov exponent , the existence of a periodic orbit of period not equal to a power of two and positive topological entropy are some techniques that can give a clear answer to this problem .
why would the labor market in most of the developed economies behave in such a volatile way as the evidence points out ?
one possibility , and usually the most favoured one in the dominant view of economics , is that the economyhas an inherently linear structure and is hit by permanent exogenous shocks . as these shocks are entirely unpredictable , they render the dynamics and the cycles hardly predictable and controllable .
another more recent view , which should also be considered for discussion because it seems consistent and realistic , is based on the possibility that the economy has a structure that is nonlinear and the cycles are an endogenous manifestation of this characteristic , either with or without external shocks added to the structure . in what follows
, a very simple and fully deterministic model will be presented that is capable of generating such type of volatility with standard parameter sets and no noise .
let us assume that in every period of time there are large flows of workers moving into and out of employment : a certain number of job vacancies is posted by firms @xmath0 and there is a total measure of workers looking for jobs @xmath1 .
when a worker and a firm reach an agreement there is a successful match , and the total number of these matches is given by the aggregate matching function @xmath2the intuition behind ( [ eqm(u , v ) ] ) is very simple : the higher is @xmath3 the easier it will be for firms to get a worker with the desired qualifications ; and the higher is the level of vacancies posted by firms @xmath4 , the higher is the probability that a worker will find an appropriate job . for simplicity we will assume @xmath5 as a constant .
however , a more adequate treatment would consist of treating @xmath5 as a variable dependent on the level of public provision of information by public agencies with the objective of increasing the number of successful matches .
the measure of labor tightness is given by the ratio @xmath6 then , the probability of a vacancy being filled at @xmath7 is given by@xmath8 let @xmath9 be the total number of employed workers at the beginning of @xmath10 and let @xmath11 be defined as the probability of a match being dissolved at @xmath7 . therefore we have @xmath12 where @xmath13 notice that @xmath14 gives the number of undissolved matches prevailing at @xmath7 and passed on to @xmath15 , while @xmath16 represents the number of new matches formed at @xmath7 with the available number of unemployed workers and vacancies .
as shown in @xcite , the model can be solved for the decentralized outcome of a nash bargaining game between workers and firms but to keep the model as close as possible to the presentation in @xcite and @xcite we should focus upon the central planner solution to the matching model .
the objective function of the central planner is given by @xmath17where @xmath18 , @xmath19 and @xmath20 are parameters that represent , respectively , the productivity of each worker , the lost value of leisure due to labor effort , and the cost that firms incur per vacancy placed in the market .
therefore , the planner chooses @xmath21 and the next period s employment level , @xmath22 by solving the following dynamic optimization problem@xmath23\]]subject to@xmath24where @xmath25 is the time discount rate and an initial condition @xmath26 is given .
the lagrangian can be written as @xmath27 + \lambda _ { t}\left [ \left ( 1-s\right ) n_{t}+q\left ( \frac{v_{t}}{1-n_{t}}\right ) v - n_{t+1}\right ] \right\ } .\]]the first order conditions ( foc ) , for an interior solution , are given by@xmath28 = 0 \\ \frac{\partial l}{\partial n_{t+1}}=-\lambda _
{ t}+\beta ^{t+1}\left ( \phi -z\right ) + \lambda _ { t+1}\left [ \left ( 1-s\right ) + q^{\prime } \left ( \theta _ { t+1}\right ) \theta _ { t+1}^{2}\right ] = 0.\end{gathered}\ ] ] the very interesting point in @xcite and @xcite was the manipulation of these foc to arrive at a reduced equation that can lead to chaotic dynamics . from the first foc we get @xmath29 and
substituting this and the corresponding expression for @xmath30 into the second foc we obtain @xmath31with the following parameter definitions and restrictions@xmath32 equation ( [ eqpricipal ] ) gives the law of motion for the index of labor market tightness in the economy under the planner s solution .
in other words , given an initial condition @xmath33 equation ( [ eqpricipal ] ) completely characterizes the trajectory of @xmath34 and the whole economy@xmath35
so , the backward dynamics of this model can be characterized by the four - parameter one - dimensional family of maps @xmath36 \rightarrow \left [ 0,g_{\max } \right ] , $ ] where@xmath37with the parameter restrictions in ( [ eqparam ] ) and @xmath38 where @xmath39 is implicitly defined as the lowest positive root of the equation @xmath40 the first derivative of the map @xmath41 can be calculated as @xmath42 , \]]which implies that @xmath41 is unimodal with a unique maximum ( critical point ) at @xmath43 in addition , @xmath41 has a unique fixed point located to the right of @xmath44 if @xmath45 the unique fixed point of @xmath41 is denoted by @xmath46 and is implicitly given by @xmath47 despite the impossibility of the computation of an explicit expression for @xmath48 the unicity of this solution is obvious by considering @xmath49 and @xmath50where @xmath51 is monotonically decreasing for @xmath34 from @xmath44 to @xmath52 and @xmath53 is monotonically increasing for @xmath34 from @xmath54 to @xmath55 therefore @xmath56 has a unique solution for @xmath57 the fixed point is an attractor in the case of backward dynamics if @xmath58 and in forward dynamics if @xmath59 for @xmath60 a period - doubling bifurcation occurs and the fixed point changes stability . since it is not possible to obtain a closed form expression for @xmath48 this condition can not be checked in general but can be checked for each set of parameters separately .
the first main objective is to show that the matching model leads to chaotic motion in the backward dynamics and the second purpose is to provide a very rigorous method to compute topological entropy for uni(multi)-modal maps .
generally , the search for chaos in one - dimensional maps is mainly based on the application of the well - known li - yorke theorem , but when a period three fails to be encountered it is necessary to use some other criteria . in what follows ,
we use the kneading sequence ( the trajectory initiated at the critical point of the unimodal map ) and the associated markov transition matrix in order to compute the topological entropy of the map .
these are tools from symbolic dynamics which provide simple and exact methods to conclude about the existence of chaos in multimodal maps . in @xcite
the authors conclude that the unique fixed point @xmath61 of the map ( [ eqmapa ] ) is always stable in backward dynamics , for almost any choice of the parameters values with respect to conditions ( [ eqparam ] ) .
we obtain a different result : while for most reasonable sets of parameter values the fixed point is stable , however there are also other sets of parameter values that allow for unstable chaotic motion in the model as we show next . for @xmath62 @xmath63 @xmath64 @xmath65 @xmath66
we obtain chaotic motion for the nonlinear map presented in ( eqmapa ) .
moreover , this is true for almost any value of @xmath67 in the interval @xmath68 $ ] .
the first point which we would like to emphasize is that all parameter values satisfy the restrictions presented in equation ( [ eqparam ] ) . that is@xmath69this is illustrated in figure [ fig1 ] , where the unimodal map and the typical chaotic time series associated to the unstable fixed point @xmath70 are presented .
the first derivative of the unimodal map is @xmath71 , which shows that the equilibrium is unstable . by varying the parameter @xmath67 in the interval
@xmath72 $ ] we obtain the classical period - doubling route to chaos which is also illustrated in figure [ fig1 ] .
so , for any value of @xmath67 arising after the @xmath73 bifurcation point , we encounter chaotic dynamics .
we should point out that in order to obtain the bifurcation diagram illustrated in figure [ fig1 ] the parameters @xmath74 and @xmath75 have also to have their values changed , as they are continuously depending on the values of @xmath76 the parameter restrictions are always satisfied , since @xmath77 moreover , in figure [ fig2 ] the bifurcation diagram is illustrated when the parameter @xmath78 is varied between @xmath79 and @xmath80 and taking the same parameter calibration as in the above proposition : @xmath81 and @xmath82 there is also a period - doubling route to chaos , but the map @xmath83 dynamics escapes to infinity before the occurrence of a period three orbit .
this presents one more reason to consider the complete sharkowski relation between the periodic orbits of the map and to apply some other techniques in the search for chaos in this specific model . in the same figure
we also represent the variation of the lyapunov exponent for @xmath78 belonging to the interval @xmath84.$ ] we recall that positive lyapunov exponent is a necessary condition for chaos .
bhattacharya and bunzel @xcite suggest three examples for the study of the dynamics of the map @xmath41 . in the first case a period 3-cycle is found , which implies the existence of chaos in the li - yorke sense if the sharkovsky order is applied . in the second example
, it was argued that a period 3-orbit could not be found for the following parameter values : @xmath85 @xmath86 @xmath87 @xmath88 since the equation @xmath89 has no solution . since the map @xmath41 is unimodal and mitra s sufficient condition for chaos in unimodal maps
is verified , the conclusion was that for this parameter setting chaotic motion in the backward dynamics could also be found . where @xmath90 is a non - negative interval
, @xmath91 is the critical point such that @xmath92 and @xmath93 is the unique fixed point of the map such that @xmath94 mitra states that : if @xmath95 satisfies @xmath96 and @xmath97 , then @xmath98 shows topological chaos . ] finally , in the third example the set of parameters are : @xmath99 @xmath100 @xmath101 @xmath102 for these values no period three orbit is found , neither the sufficient condition of mitra is verified .
therefore , it was argued that for this case the very existence of chaos for the unimodal map is questioned on the grounds of a lack of logical proof of such dynamics . in order to clarify some of the issues raised by @xcite ,
a symbolic dynamics approach is developed for the unimodal map @xmath41 , which allow us to perform the computation of the topological entropy for any choice of parameters , and , of course , permit us to classify the complexity of the map since positive topological entropy implies the existence of chaotic dynamics .
we will concentrate on their third example .
we consider again the unimodal map
@xmath36 \rightarrow % \left [ 0,g_{\max } \right ] .$ ] this kind of map has symbolic dynamics relative to a topological markov partition generated by the orbit of the critical point @xmath43 this is illustrated in figure [ figmarkov ] for the parameter values presented in example 2 .
so , any numerical trajectory @xmath103 for the map @xmath41 corresponds to a symbolic sequence @xmath104 where @xmath105 depending on where the point @xmath106 falls in , i.e.,@xmath107all symbolic sequences made of these letters may be ordered by the natural lexicographical order @xmath108 and defining the fullshift @xmath109 to be the set of all possible infinite symbolic strings of @xmath110 s and @xmath111 s , then any given infinite symbolic sequence is a singleton in the fullshift space .
the bernoulli shift map @xmath112 is defined by @xmath113 . in general ,
not all symbolic sequences correspond to the trajectory of an initial condition @xmath114 restricting the shift map to a subset of @xmath115 consisting of all the itineraries that are realizable yields the subshift @xmath116 we formulate the result in terms of topological markov chains , a special class of subshifts of finite type where the transition in the symbol sequence is specified by a @xmath117 matrix .
any @xmath118 binary matrix @xmath119 generates a special subshift @xmath120which is called the topological markov chain associated with the markov matrix @xmath121 .
we say that @xmath122 if the transition from @xmath123 to @xmath124 is possible .
the matrix @xmath121 gives a complete description of the dynamics of the unimodal map .
the premier numerical invariant of a dynamic system is its topological entropy defined by @xmath125where @xmath126 is the set of words of length @xmath127 occurring in sequences of @xmath128 .
moreover , if the topological markov matrix @xmath121 is given then , the topological entropy is the natural logarithm of the spectral radius of @xmath121 . for the parameter values @xmath85
@xmath86 @xmath129 @xmath88 , we found a period 5 orbit : @xmath130 which is shown in figure [ figmarkov ] with the corresponding 4 interval @xmath131 markov partition .
the critical point assumes the value @xmath132 and generates the symbolic partition for the map @xmath133 the periodic orbit has the following symbolic address : @xmath134 and in consequence we have the following markov matrix:@xmath135 .\]]the maximal eigenvalue of @xmath121 is given by @xmath136 which implies that the topological entropy is positive : @xmath137 and this shows very clearly that we are dealing with chaotic motion in this set of parameter values .
it should be noted that for any other kneading sequence , for a suitable choice of parameters values , we can obtain a markov partition and a markov matrix which totally determine the complexity of the unimodal map .
this is a very simple and rigorous way to estimate the topological entropy of a one - dimensional model and to check for the existence of chaos .
the general use of the li - yorke theorem or the mitra condition may lead to misleading conclusions about the existence of chaotic dynamics , as it was done in @xcite .
therefore , in order to obtain relevant answers to whether there are or not chaotic dynamics under certain ranges of parameters values in a 1-dimensional particular model , we suggest that a bifurcation diagram , the variation of the lyapunov exponent , the existence of a periodic point of period not equal to a power of two , and symbolic dynamics are very powerful techniques for that purpose . moreover , the application of these techniques to the matching labor market model so interestingly developed by bhattacharya and bunzel clearly confirmed that a very simple model of the labor market , with well behaved aggregate functions ( continuous , twice differentiable and linearly homogeneous ) do really produce chaotic behavior for a range of parameter sets which had been questioned in @xcite , for high values of the measure of labor market tightness . financial support from the fundao cincia e tecnologia , lisbon , is grateful acknowledged , under the contract no pocti/ eco /48628/ 2002 , partially funded by the european regional development fund ( erdf ) . | in this paper we apply the techniques of symbolic dynamics to the analysis of a labor market which shows large volatility in employment flows . in a recent paper , bhattacharya and bunzel @xcite
have found that the discrete time version of the pissarides - mortensen matching model can easily lead to chaotic dynamics under standard sets of parameter values . to conclude about the existence of chaotic dynamics in the numerical examples presented in the paper , the li - yorke theorem or the mitra sufficient condition
were applied which seems questionable because they may lead to misleading conclusions . moreover , in a more recent version of the paper , bhattacharya and bunzel @xcite present new results in which chaos is completely removed from the dynamics of the model . our paper explores the matching model so interestingly developed by the authors with the following objectives in mind : ( i ) to show that chaotic dynamics may still be present in the model for standard parameter values ; ( ii ) to clarify some open questions raised by the authors in @xcite , by providing a rigorous proof of the existence of chaotic dynamics in the model through the computation of topological entropy in a symbolic dynamics setting .
symbolic dynamics , periodic orbits , chaos conditions , backward dynamics , matching and unemployment . |
we discuss the first passage time problem in the semi - infinite interval , for homogeneous stochastic markov processes with lvy stable jump length distributions @xmath0 ( @xmath1 ) , namely , lvy flights ( lfs ) .
in particular , we demonstrate that the method of images leads to a result , which violates a theorem due to sparre andersen , according to which an arbitrary continuous and symmetric jump length distribution produces a first passage time density ( fptd ) governed by the universal long - time decay @xmath2 .
conversely , we show that for lfs the direct definition known from gaussian processes in fact defines the probability density of first arrival , which for lfs differs from the fptd .
our findings are corroborated by numerical results .
lvy flights ( lfs ) and lvy walks ( lws ) are the prime example in the investigation of non - standard transport processes whose stationary solution does not converge towards the boltzmann form @xcite .
being subject to the generalised central limit theorem @xcite , lfs correspond to a markov process in which extremely long excursions can occur with appreciable probability , whereas in lws long excursions are penalised through a time cost introduced via a spatiotemporal coupling @xcite .
applications of lfs and lws range from the famed flight of an albatross @xcite , the spreading of spider - monkeys @xcite , or the grazing patterns of bacteria @xcite , over economical data @xcite to molecular collisions @xcite and plasmas @xcite . despite their broad usage , the detailed behaviour of even the simpler , uncoupled lf processes , on which we concentrate in the following , in external potentials and under non - trivial boundary conditions is still not fully explored .
thus , there have recently been discovered bifurcations between multimodal states of the probability density function ( pdf ) of lfs in steeper than harmonic external fields , in whose presence also the variance becomes finite @xcite , and rich band structures have been reported for lfs in periodic potentials @xcite . of particular interest in random processes is the first passage time density ( fptd ) @xcite .
for lfs , the fptd was determined through the method of images on a finite domain in reference @xcite , and with similar methods in reference @xcite .
these methods lead to results for the fptd in the semi - infinite domain , whose long - time behaviour explicitly depends on the lvy index @xmath3 .
in contrast , a theorem due to sparre andersen proves that for any discrete - time random walk process starting at @xmath4 with each step chosen from a continuous , symmetric but otherwise arbitrary distribution , the fptd asymptotically decays as @xmath5 with the number @xmath6 of steps @xcite , being _ fully independent _ of the index of the lf , i.e. , universal . in the case of a markov process
, the continuous time analogue of the sparre andersen result reads @xmath7 the analogous universality was proved by frisch and frisch for the special case in which an absorbing boundary is placed at the location of the starting point of the lf at @xmath8 @xcite , and numerically corroborated by klafter and zumofen @xcite . in the following ,
we demonstrate that the method of images is generally inconsistent with the universality of the fptd , and therefore can not be applied to solve fptd - problems for lfs .
we also show that for lfs the fptd differs from the pdf for first arrival .
let us start by recalling that an unbiased lf can be defined through the space - fractional diffusion equation for the pdf @xmath9 @xcite @xmath10 where we define the fractional derivative @xmath11 by its fourier transform .
( here and in the following , we restrict ourselves to @xmath12 . ) in position space , the fractional derivative is defined in terms of the convolution ( see @xcite for the case @xmath13 ) @xmath14 equivalently , lfs can be described in terms of continuous time random walks with long - tailed jump length distributions @xmath0 @xcite .
the associated pdf @xmath9 for natural boundary conditions ( @xmath15 ) with initial condition @xmath16 is the lvy stable law @xmath17 @xcite . in fourier - laplace space
@xcite , this pdf corresponds to @xmath18 .
a characteristic of lfs is the divergence of the variance of both @xmath9 and @xmath19 . equipping equation ( [ fde ] ) with a @xmath20-sink of strength @xmath21
, we obtain the diffusion - reaction equation for the non - normalised density function @xmath22 , @xmath23 i.e. , @xmath21 is the negative time derivative of the survival probability . by definition of the sink term , @xmath21 is the pdf of _ first arrival _ :
once a random walker arrives at the sink , it is annihilated . by solving equation ( [ heat ] ) through standard methods ( determining the homogeneous and inhomogeneous solutions ) , it is straightforward to calculate the solution @xmath24 in terms of the propagator @xmath25 of equation ( [ fde ] ) with initial condition @xmath26 yielding @xmath27/\left(s+d|k|^ { \alpha}\right)$ ] , from which , in turn , we find that @xmath21 satisfies the chain rule ( @xmath28 implicitly depending on @xmath29 ) @xmath30 which corresponds to the laplace space relation @xmath31 . equation ( [ mfpt ] ) is well - known and for any sufficiently well - behaved continuum diffusion process is commonly employed to define the fptd @xcite . for gaussian processes with propagator @xmath32\right)$ ]
, one obtains by direct integration of the diffusion equation with appropriate boundary condition the fptd @xcite @xmath33 including the asymptotic behaviour @xmath34 for @xmath35 . in this gaussian case , the quantity @xmath21 is equivalent to the fptd . from a random walk perspective
, this is due to the fact that individual steps are of the same increment , and the jump length statistics therefore ensures that the walker can not hop across the sink in a long jump without actually hitting the sink and being absorbed .
this behaviour becomes drastically different for lvy jump length statistics : there , the particle can easily cross the sink in a long jump . thus , before eventually being absorbed , it can pass by the sink location numerous times , and therefore the statistics of the first arrival will be different from the one of the first passage .
in fact , with @xmath36 , we find @xmath37 by use of the de moivre identity @xmath38 . with @xmath39 and @xmath40
we obtain the limiting form @xmath41 where @xmath42 .
we note that the same result is obtained by the exact expressions for @xmath43 and @xmath44 in terms of fox @xmath45-functions and systematic expansion @xcite .
the inverse laplace transform of the small @xmath46-behaviour ( [ limit3 ] ) can be obtained by completing ( [ limit3 ] ) to an exponential , and then computing the laplace inversion by the identification @xmath47 $ ] with the fox @xmath45-function @xcite , for which the exact laplace inversion can be performed @xcite .
finally , a series expansion of this result leads to the long-@xmath48 form @xmath49 with @xmath50/2\right)\sin^2(\pi/\alpha)/(\pi^2(\alpha-1))$ ] .
clearly , in the gaussian limit , the required asymptotic form @xmath51 for the fptd is consistently recovered , whereas in the general case the result ( [ limit4 ] ) is slower than in the universal fptd behaviour in equation ( [ universal ] ) , as it should as the @xmath20-trap used in equation ( [ heat ] ) to define the first arrival for lfs is weaker than the absorbing wall used to properly define the fptd . for lfs
, the pdf for first arrival thus scales like ( [ limit4 ] ) ( i.e. , it explicitly depends on the index @xmath3 of the underlying lvy process ) , and , as shown below , it
_ differs _ from the corresponding fptd .
( 10,7.2 ) ( 1.8,8 ) before calculating this fptd , we first demonstrate the validity of equation ( [ limit4 ] ) by means of a simulation the results of which are shown in figure [ fig2 ] .
random jumps with lf jump length statistics are performed , and a particle is removed when it hits a certain interval of width @xmath52 around the sink ; for our simulations we found an optimum value @xmath53 .
as seen in figure [ fig2 ] ( note that we plot @xmath54 ! ) and for analogous results not shown here , relation ( [ limit4 ] ) is nicely fulfilled for all @xmath12 , whereas for larger @xmath52 , the slope increases .
the proper dynamical formulation of an lf on the semi - infinite interval with an absorbing boundary condition at @xmath55 , and thereby the determination of the fptd , has to make sure that in terms of above random walk picture jumps across the sink are forbidden .
this can be consistently achieved by setting @xmath56 on the left semi - axis , i.e. , actually removing the particle when it crosses the point @xmath55 .
this formally corresponds to the modified dynamical equation @xmath57 in which the fractional integral is truncated to the semi - infinite interval .
after laplace transformation and integrating over @xmath58 twice , one obtains @xmath59 where @xmath60 is the fptd and the kernel @xmath61 .
this equation is formally of the wiener - hopf type of the first kind @xcite .
after some manipulations similar to those applied in reference @xcite , we arrive at the asymptotic expression @xmath62 , where @xmath63 , in accordance with the universal behaviour ( [ universal ] ) and with the findings in reference @xcite .
thus , the dynamic equation ( [ neu ] ) consistently phrases the fptd problem for lfs .
we note that due to the truncation of the fractional integral it was not possible to modify the well - established grnwald - letnikov scheme @xcite to numerically solve equation ( [ neu ] ) with enough computational efficiency to numercially obtain the direct solution for @xmath22 .
however , to corroborate the validity of the sparre anderson - universality , we perform a simulation of an lf in the presence of an absorbing wall , i.e. , random jumps with lf jump length statistics are performed along the right semi - axis , and a particle is removed when it jumps across the origin to the left semi - axis .
results of such a detailed random walk study are displayed in figures [ fig ] and [ fig1 ] .
the expected universal @xmath64 scaling is nicely obtained for various initial conditions and @xmath3 . clearly , the scaling for the first arrival as well as the image method - fptd derived below are significantly different .
( 10,7.2 ) ( 0.94,-1.8 ) ( 10,7.2 ) ( 0.94,-1.8 ) we now demonstrate that the method of images produces a result , which is neither consistent with the universal behaviour of the fptd ( [ universal ] ) nor with the behaviour of the pdf of first arrival ( [ limit4 ] ) , given the initial condition @xmath65 , the solution @xmath66 for the absorbing boundary value problem with the analogous dirichlet condition @xmath67 according to the method of images is given via the difference @xcite @xmath68 in terms of the free propagator @xmath25 , i.e. , a _ negative _ image solution originating at @xmath69 balances the probability flux across the absorbing boundary .
the corresponding pseudo - fptd is then calculated in the same way as in equation ( [ surv ] ) . for the image solution in fourier - laplace space ,
we obtain @xmath70/\left(s+d|k|^{\alpha } \right),\ ] ] for a process which starts at @xmath71 and takes place in the right half space . in laplace space ,
the image method - fptd becomes @xmath72 after some transformations , we arrive at @xmath73.\ ] ] in the limit of small @xmath46 , this expression reduces to @xmath74 , with @xmath75 . following the same procedure
as outlined above , we find the long-@xmath48 form @xmath76 for the image method fptd . in the gaussian limit @xmath77 , expression ( [ limit2 ] )
produces @xmath78 , in accordance to equation ( [ bmfpt ] ) .
conversely , for general @xmath12 , @xmath79 according to equation ( [ limit2 ] ) would decay faster than @xmath80 .
the method of images breaks down for lfs due to their special non - local nature , displayed by the integrals in equations ( [ fde ] ) and ( [ riesz ] ) , and ( [ neu ] ) , namely having a long - tailed jump length distribution .
this leads to leapovers beyond the absorbing boundary .
the method of images is expected to work when the boundary is also a turning poing of the trajectory , as actually happens for nearest neighbours random walks , or the wiener process .
qualitatively , the following argument may be brought forth in favour of the observed universality of the lf - fptd : the long - time decay is expected to be governed by short - distance jump events , corresponding to the central region of very small jump lengths for the lvy stable jump length distribution .
but this region is , apart from a prefactor , indistinguishable from the gaussian distribution , and therefore the long - time behaviour should in fact be the same for any continuous jump length distribution @xmath19 .
in fact , the universal law ( [ universal ] ) can only be modified in the presence of non - markov effects such as broad waiting time processes or spatiotemporally coupled walks @xcite . in terms of the special case covered by the theorem of frisch and frisch @xcite ,
in which the absorbing boundary coincides with the initial condition , we can understand the general situation for finite @xmath71 , as in the long - time limit , the distance @xmath29 becomes negligible in comparison to the diffusion length @xmath81 : therefore the asymptotic behaviour is necessarily governed by the same universality .
concluding , we demonstrated that the method of images , which has been developed a powerful tool in gaussian diffusion also beyond the homogeneous case @xcite and in the presence of long - tailed waiting times @xcite , fails for lf processes , leading to a false result for the fptd .
moreover , we showed that for such broad jump length statistics , the pdf of first arrival at a point differs from the fptd .
we also provided a framework in terms of a truncated fractional diffusion equation to solve the fptd problem for an lf .
this study is expected to significantly contribute to the understanding of the , at instances , non - trivial behaviour of lfs .
we are happy to acknowledge discussions with igor sokolov .
we also acknowledge financial support through the intas project 00 - 0847 from the european commission . |
|
we do not discuss here the nature of the source responsible for the initial energy release leading to the grb .
we assume that a relativistic wind has emerged at radius @xmath1 with a variable distribution of the lorentz factor @xmath2 ( with @xmath3 ) .
the energy and mass injection rates are @xmath4 and @xmath5 .
the wind production lasts for a total duration @xmath6 ( total ejected mass : @xmath7 ) . at @xmath8 ,
the wind is still optically thick , whereas at the radius @xmath9 where the internal shock phase starts , it has to be already transparent , as indicated by the dominant non - thermal nature of the observed grb spectra .
the radius at which each layer of the relativistic wind becomes transparent can be estimated ( see daigne & mochkovitch 2002 ) : photons emitted by a layer at @xmath10 will successively cross all the layers emitted earlier by the source before they can escape from the relativistic wind at radius @xmath11 : as @xmath12 , we usually have @xmath13 .
the optical depth of a layer at radius @xmath10 is approximatively given by @xmath14 , where the opacity @xmath15 is mainly due to the ambient electrons ( thomson opacity ) .
therefore the photospheric radius ( @xmath16 ) of the layer is given by : @xmath17 .
1 shows this radius for an example where @xmath18 , @xmath19 and the lorentz factor is given in the same figure .
the photospheric luminosity and temperature of each layer can be estimated in the context of the fireball model ( see e.g. piran 1999 ) : @xmath20 where the initial fireball temperature is given by @xmath21 ( @xmath22 is the mass of the central black hole in units of @xmath23 ) . the emitted photons will closely follow a planck distribution , so that the phostopheric emission leads to a spectrum and a count rate in an energy band @xmath24 $ ] : @xmath25 where @xmath26 , @xmath27 and @xmath28 .
the internal shock luminosity is assumed to be @xmath29 where @xmath30 is the efficiency for the conversion of kinetic energy into @xmath31-rays by internal shocks .
this leads to a spectrum and a count rate in energy band @xmath24 $ ] given by @xmath32 where @xmath33 is the observed peak energy of the internal shock spectrum ( @xmath34 ) and @xmath35 is the usual band spectral shape ( with @xmath36 ) , characterized by a low and high - energy slope @xmath37 and @xmath38 ( band et al .
2 shows the spectrum and the time profile obtained in the example of fig . 1 including both the internal shock and the photospheric emission ( with @xmath39 , @xmath40 and @xmath41 ) .
the internal shock emission is computed with the model developped by daigne & mochkovitch ( 1998 ) .
the global observed peak energy of the non - thermal photons is 200 kev .
[ fig : resultat1 ]
the ratio of the photospheric over internal shock count rates is given by : @xmath42 this expression indicates that the photospheric thermal emission is usually dominant in the x- and @xmath31-ray bands ( see fig .
2 ) ! this is clearly in disagreement with the observations showing that the grb spectrum is non - thermal in this spectral range . from eq . 4 ,
there are two possible solutions to recover grbs dominated by the non - thermal internal shock component : ( i ) increase of @xmath30 , i.e. more efficient internal shocks .
however , our calculations show that even for @xmath30 close to unity , the thermal component would still be easily detectable ; ( ii ) decrease of @xmath43 and @xmath44 , i.e. a less hot and bright photosphere , compared to the prediction of the standard fireball model . if we define @xmath45 as the fraction of the energy released by the source which is initially under thermal form , the photospheric luminosity and temperature are given by : @xmath46 so that the expression of @xmath47 in eq .
4 has now to be multiplied by a factor @xmath48 . in the standard fireball model
@xmath49 but much smaller values can be expected if a large fraction of the energy realeased by the source is initially under magnetic form .
3 shows exactly the same example as fig . 1 ,
the only difference being that the photospheric emission has been computed with @xmath50 .
a non - thermal burst is recovered in agreement with the observations .
a small thermal precursor is still present in x - rays
. it can be more intense for intermediate values of @xmath45 , so that the photospheric emission predicted in this case could explain the x - ray precursors observed in a few grbs .
we have computed the photospheric thermal emission of a relativistic wind with a highly variable initial distribution of lorentz factor .
we find that : ( 1 ) the photosphere predicted in the standard fireball model for grbs is usually too hot and luminous .
the thermal photospheric emission dominates the non - thermal internal shock emission , even for large internal shock efficiencies .
this is in disagreement with the observations .
( 2 ) a less luminous photosphere is expected if the energy released by the source is initially injected mainly under magnetic rather than thermal form . in this case , we recover grb spectra and temporal profiles dominated by the non - thermal internal shock emission .
( 3 ) we then obtain x - ray precursors which may correspond to the precursor activity observed in a few grbs by ginga and watch / granat . these resultats are presented in more details in daigne & mochkovitch ( 2002 ) . | the prompt emission of gamma - ray bursts ( hereafter grbs ) probably comes from a highly relativistic wind which converts its kinetic energy into radiation via the formation of shocks within the wind itself .
such `` internal shocks '' can occur if the wind is generated with a highly non uniform distribution of the lorentz factor @xmath0 .
taking into account such a variable distribution of @xmath0 , we estimate the expected thermal emission of the relativistic wind when it becomes transparent . we compare this emission ( temporal profile + spectrum ) to the emission produced by the internal shocks . in most cases
we predict a rather bright thermal emission that could easily be detected .
this favors acceleration mechanisms for the wind where the main energy reservoir is under magnetic rather than thermal form .
such scenarios can produce thermal x - ray precursors comparable to those observed by ginga and watch / granat .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the globular cluster m54 ( ngc6715 ) is a special object under several aspects . in fact
, it is quite massive ( @xmath5 , pryor & meylan 1993 ) , it displays a complex horizontal branch ( hb ) morphology ( extending up to the highest temperatures achievable by these stars ; rosenberg et al .
2004 ) and its stars present an intrinsic spread in iron abundances ( carretta et al .
2010 , c10 hereafter ; see bellazzini et al . 2008 , hereafter b08 , for previous studies and references ) .
moreover , it resides at the center of the sagittarius dwarf spheroidal galaxy ( sgr dsph ; ibata , gilmore & irwin 1994 ) , a satellite of the milky way that is currently disrupting under the tidal strain of the galaxy ( see , for example , majewski et al .
several authors suggested the possibility that m54 could be the nucleus of sgr ( bassino & muzzio 1995 ; sarajedini & layden 1995 ) .
later studies ( layden & sarajedini 2000 ; majewski et al . 2003 ; monaco et al .
2005a ) , however , demonstrated that the metal - poor cluster ( [ fe / h]@xmath6 ; brown , wallerstein & gonzalez 1999 , c10 ) co - exists with a stellar nucleus made of metal rich stars of the sgr galaxy ( @xmath7\rangle\simeq -0.45 $ ] ; b08 and references therein ) .
a recent study based on accurate velocity and metallicity data ( b08 ) strongly supports the idea that m54 formed independently , and plunged into the sgr nucleus as a result of significant decay of its original orbit due to dynamical friction ( monaco et al . 2005b ; see also c10 for independent support to this hypothesis ) . in this context , rr lyrae ( rrl ) variables can provide further insight into the nature of this stellar system , in addition to reliable estimates of the distance and foreground extinction .
the first ( photographic ) survey for variable stars in m54 was performed by rosino & nobili ( 1959 ) who identified 82 variables .
their pioneering analysis was followed up and expanded by layden & sarajedini ( 2000 ; ls2000 hereafter ) , who considered the variables as part of their v , i ccd photometric study aimed at obtaining the colour - magnitude diagram ( cmd ) of m54 + sgr field to derive the star formation history of the sgr dsph galaxy . in that work ls2000 found 117 variable stars , 93 of which are candidate rrl stars .
they analysed the best observed subset of rrl ( 67 stars ) to obtain a distance modulus of @xmath8 and e(v - i)=@xmath9 .
on the basis of the pulsational properties of their sample of rrl , these authors classified m54 as an oosterhoff type i cluster ( oosterhoff 1939 ; see catelan 2009 for a recent review ) .
in this paper we present new bvi photometric data and discuss in detail the characteristics of the rrl stars of m54 , that were presented in preliminary form by cacciari , bellazzini & colucci ( 2002 ) . in sect .
2 we describe our data , the observations , the data reduction , the calibration procedure and the period search procedure . in sect .
3 we present and briefly discuss the characteristics of the cmd resulting from our photometric data and present our sample of detected variables . in sect .
4 we analyze the rrl pulsational properties and use them to estimate their membership to the m54 or sgr stellar population .
5 presents the results of the fourier analysis of the rrl light curves and metallicity estimates .
the reddening and distance are derived in sect .
6 , and the properties of the other variables of our sample are briefly presented and discussed in sect .
we summarize and discuss our results in sect . 8 . in appendix
a we comment on individual variables .
the bvi photometric data presented in this paper were obtained on 13 , 14 and 19 july 1999 at the eso - danish 1.54-m telescope in la silla ( chile ) using the 2k@xmath102k ccd of the danish faint object spectrograph and camera ( dfosc ) .
the images cover a @xmath11 field of view roughly centered on m54 , with a scale of 0.39 @xmath12/pixel .
the bvi images were taken sequentially with exposure times of 360 - 900 sec in b , 150 - 600 sec in v , and 150 - 450 sec in i , depending on the atmospheric conditions .
the average seeing during the three nights ranged between @xmath13 and @xmath14 ( fwhm ) . in total , it was possible to obtain 54 b , 57 v and 52 i images .
the frames were overscan corrected , bias subtracted and flat fielded ( using twilight sky flats ) by means of the standard iraf tasks .
the data reduction was performed using the isis package ( alard 2000 ) , which is based on the method of image subtraction .
isis is able to transform a series of images of the same field to ( a ) the same astrometric system , ( b ) the same flux scale , and ( c ) the same point spread function ( psf ) , without flux losses .
once all the above transformations have been performed , the data reduction proceeds as follows : 1 . a high signal - to - noise reference frame ( per filter ) is constructed , by stacking a suitable number of images ; in the present case the reference frames were built by stacking 14 b , 15 v and 14 i frames . 2 .
each individual image is subtracted from the reference frame ; the squares of the images of the residuals are co - added into a single image where there are only the variable sources .
variable stars are identified on the stacked images of the residuals and their positions are recorded .
the residual flux is measured on each single subtracted image at the position of the detected variables , using a psf fitting routine .
a light curve in units of flux difference is obtained for each variable .
photometry of the reference frames is obtained independently ( in our case using dophot ; schechter , mateo & saha 1993 ) to provide the flux zero point to the light curves ( and convert into a scale of magnitudes ) . in our case
we transformed the instrumental magnitudes obtained with dophot into absolute johnson - cousins magnitudes using several hundreds of stars in common with the photometry by ls2000 ( for v and i ) and by rosenberg , recio - blanco & garcia - marin ( 2004 ) for b magnitudes .
the image - subtraction technique has proven very powerful in detecting even low amplitude variables in crowded fields , ensuring a higher accuracy in relative photometry with respect to traditional techniques ( see e.g. olech et al .
1999 ; mochejska et al .
2004 ; corwin et al . 2006 ; greco et al . 2009
; we refer to these papers also for further details on the technique ) . as an example , in fig .
[ f : compar ] we show for comparison the v light curves of 3 rrl variables in m54 that were measured by ls2000 with traditional techniques ( dophot ) and by us with isis : the higher quality of the new light curves is readily evident .
there are two additional features of the specific application of the method performed in the present case that deserve to be described : * the images of a few bright foreground stars were saturated in our frames . in the process of image subtraction
these sources left spurious residuals due to different degree of saturation in the different frames , blooming etc .
this in turn led to the identification of large numbers of spurious variable sources in the surroundings of these saturated stars . to remove contaminants from the list of variables before obtaining their light curves
, we masked the surroundings of heavily saturated stars .
the dimension of the masked areas was decided case by case , by inspecting the stacked image of the residuals .
* dophot automatically masks the regions surrounding ( even moderately ) saturated stars and/or sources for which the psf fit does not reach convergency .
for this reason there are variable stars for which we obtain a good differential flux light curve from isis but are not measured by dophot on the reference image .
hence they lack a zero - point and their light curves can not be transformed into a magnitude scale . for these stars we can not measure an absolute amplitude , but we can obtain reliable periods and classify them on the basis of their periods , the shape of the light curve and the similarity of the differential flux amplitudes with calibrated variables . for the period search , we used the graphical analyzer of time series ( gratis ) program ( see clementini et al .
2000 for details ) , which was run on the isis differential flux data .
this code ( developed by p. montegriffo at the osservatorio astronomico di bologna ) employes two different algorithms : ( 1 ) a lomb periodogram ( lomb 1976 ) and ( 2 ) a best fit of the data with a truncated fourier series ( barning 1963 ) .
the adopted period search procedure performs the lomb analysis on a wide period interval .
the fourier algorithm refines the period definition and find the best - fitting model .
the r.m.s . of the observed points about the best fit model
is taken as the error associated to the mean magnitudes .
this value is reported along with the photometric data for each variable in table 1 .
the x , y coordinates in pixels have been transformed into j2000 equatorial coordinates with a second order polynomial fitted to @xmath15 stars in common with 2mass ( skrutskie et al 2006 ) , that were used as astrometric standards .
the residuals of the adopted astrometric solution were @xmath16 r.m.s . in both ra and dec
in fig . [ cmd ] the v-(b - v ) and v-(v - i ) cmds obtained with dophot from the reference bvi images are shown .
the cmd reaches a limiting magnitude of @xmath17 sampling the population of m54 and sgr down to @xmath18 2 magnitudes below the cluster main sequence turn - off .
the vertical plume of stars at @xmath19 is composed by foreground / background galactic main sequence stars at various distances along the line of sight . the steep and narrow red giant branch ( rgb ) of m54 is clearly visible , bending from @xmath20 at @xmath21 to @xmath22 at @xmath23 .
the wider ( and sparser ) rgb of sgr runs nearly parallel to the red of the m54 one . a well - defined blue hb ( mostly associated to m54 ;
see monaco et al .
2004 ) and a well populated red clump ( associated to sgr ) are also visible at [email protected] . a detailed discussion of the cmd is out of the scope of the present paper ; we address the interested reader to papers more focused on the interpretation of the cmd ( sarajedini & layden 1995 ; monaco et al . , 2003 , 2004 , and references therein )
for a detailed analysis of the stellar content in the innermost region of the cluster , see the exquisite cmd obtained by siegel et al .
( 2007 ) from hubble space telescope data . for the purposes of the present analysis , it is important to recall that within the limiting radius of m54 ( @xmath24 ; see ibata et al .
2009 ) , the surface density of the cluster is nearly everywhere a factor of 7 - 8 larger than the sgr metal - rich nucleus ( b08 ) .
moreover , the population of the sgr nucleus is dominated by intermediate - age stars that are unlikely to evolve into rrl ( siegel et al .
2007 and references therein ) producing a fraction of rrl which should be much lower than in m54 ( monaco et al . 2004 ) .
this is in agreement with cseresnjes ( 2001 ) , who estimated a density of 139 rr0 per square degree in the central region of sgr , leading to about 6.5 rr0 in our 13 arcmin square field of view .
therefore , we conclude that the sample of rr lyrae assembled in the present study must be dominated by members of m54
. our photometric search allowed us to identify 177 variable stars ( 94 new discoveries ) , significantly increasing the number ( 117 ) of variables identified by the previous studies of rosino & nobili ( 1959 ) and ls2000 .
the identification and photometric information are given in table 1 .
in fig . [ cmdvar ] the variables detected in this study are identified in the v-(b - v ) cmd . among the 89 stars studied by ls2000 :
2 ( v23 and v91 ) were not identified in the present study because they lie out of our fov ; 4 ( v25 , v108 , v111 and v112 ) were not identified because they were masked out during the variable detection phase ( see sect .
[ s : dr ] ) ; 4 ( v86 , v89 , v90 and v95 ) were identified as variables but no light curve could be derived and one ( v100 ) was found to be non variable . of the variables reported in the former study of rosino & nobili ( 1959 ) 5 of them ( v21 , v26 , v73 , v79 and v81 ) were not identified , neither by us nor by ls2000 , because no stars were detected at the locations indicated by rosino & nobili ( 1959 ) and 5 ( v20 , v22 , v27 , v53 and v72 ) were found to be non variable ( as also found by ls2000 ) .
these stars are not listed in table 1 .
finally , 18 stars were identified and analysed using only internally calibrated photometry ; they do appear in table 1 but no photometric information is provided .
the agreement between the light curves and the estimated periods between us and ls2000 is very good , in general .
a few discordant cases are discussed in appendix a. for the 61 variable stars of the ls2000 sample for which we have calibrated photometry , we obtained light curves and derived the ephemerides using ls2000 photometric data in combination with ours to obtain a longer time baseline and more accurate results .
for the 94 newly identified variables ( v118-v211 ) we derived the classification and the periods .
we found : 51 rr0 , 29 rr1 , 3 sx phoenicis , 3 w uma ( ew ) , one algol binary ( ea ) , 4 long period ( lp ) candidates ( with no period determination ) and 3 with uncertain classification ( v145 , v147 and v211 ) of which two ( v147 and v211 ) might be eclipsing binaries ( eb ) . from the present calibrated data we find no evidence of double - mode pulsation in any of the studied stars . for 58 of these stars ,
mostly located in the central region of m54 , the photometry could not be externally calibrated for the reasons described in sect .
[ s : dr ] .
the light curves of a sample of newly discovered rrl and other variables are shown in fig.s [ lc_rrlyr ] and [ lc_other ] .
the epoch photometry for all the variables with calibrated photometry is available in its entirety at the cds . in the following
we will focus the analysis only on rr lyrae stars ; variables of other types will be briefly discussed in sect .
[ s : other ] .
[ t : all ] @lccccccccccccccccl@ i d & ty & ra ( j2000 ) & dec ( j2000 ) & period & epoch & @xmath25 & @xmath26 & @xmath27 & rmsv & @xmath28 & @xmath29 & @xmath30 & ( b - v@xmath31 & ( v - i@xmath31 & @xmath32b - v@xmath33 & @xmath32v - i@xmath33 & note + n. & & deg & deg & d & hjd & & & & & & & & & & & & + & & & & & 2451370 + & & & & & & & & & & & & + + 01 & 0 & 283.7906329 & -30.4768056 & 1.3476492 & 2.11223 & 17.73 & 17.24 & 16.60 & 0.01 & 1.26 & 0.94 & 0.56 & 0.675 & 0.790 & 0.532 & 0.662 & + 02 & 0 & 283.7621242 & -30.4543426 & 1.0945410 & 2.65622 & 17.73 & 17.34 & 16.72 & 0.02 & 1.04 & 0.89 & 0.57 & 0.424 & 0.759 & 0.411 & 0.635 & + 03 & 1 & 283.7591376 & -30.4295744 & 0.5726301 & 9.60328 & 18.70 & 18.19 & 17.52 & 0.03 & 1.09 & 0.76 & 0.55 & 0.610 & 0.745 & 0.536 & 0.687 & + 04 & 1 & 283.7454891 & -30.3931179 & 0.4832307 & 9.75278 & 18.57 & 18.32 & 17.64 & 0.03 & 1.41 & 0.77 & 0.60 & 0.590 & 0.717 & 0.338 & 0.693 & bl + 05 & 1 & 283.7218320 & -30.4670001 & 0.5790976 & 9.70703 & 18.66 & 18.11 & 17.47 & 0.02 & 1.23 & 0.95 & 0.66 & 0.583 & 0.754 & 0.562 & 0.664 & + 06 & 1 & 283.8307177 & -30.5287519 & 0.5384199 & 3.51936 & 18.41 & 18.03 & 17.43 & 0.04 & 1.17 & 0.71 & 0.68 & 0.596 & 0.583 & 0.420 & 0.610 & bl + 07 & 1 & 283.7810458 & -30.5232817 & 0.5944093 & 4.57644 & 18.73 & 18.21 & 17.54 & 0.01 & 0.86 & 0.74 & 0.47 & 0.574 & 0.762 & 0.537 & 0.681 & + 12 & 2 & 283.6919599 & -30.5478954 & 0.3226390 & 9.30830 & 17.14 & 16.71 & 16.15 & 0.03 & 0.48 & 0.34 & 0.22 & & & 0.435 & 0.569 & f + 14 & 1 & 283.8410760 & -30.4206737 & 0.4807214 & 3.04538 & & & & - & - & - & - & & & & & + 15 & 1 & 283.8053859 & -30.4962929 & 0.5868540 & 3.54588 & 18.69 & 18.21 & 17.53 & 0.02 & 0.95 & 0.93 & 0.61 & 0.493 & 0.800 & 0.481 & 0.704 & + 28 & 1 & 283.7859926 & -30.4341522 & 0.5088183 & 3.83630 & 18.48 & 18.08 & 17.51 & 0.05 & 1.40 & 1.18 & 0.78 & 0.518 & 0.734 & 0.432 & 0.614 & bl , sgr + 29 & 1 & 283.7226484 & -30.4924474 & 0.5897109 & 2.90001 & 18.77 & 18.25 & 17.52 & 0.02 & 0.92 & 0.72 & 0.41 & 0.575 & 0.801 & 0.532 & 0.742 & + 30 & 1 & 283.7645154 & -30.4570863 & 0.5738690 & 9.65573 & 18.74 & 18.24 & 17.55 & 0.02 & 1.10 & 0.87 & 0.56 & 0.588 & 0.777 & 0.524 & 0.701 & + 31 & 1 & 283.7323449 & -30.4985297 & 0.6466228 & 3.56726 & & 18.18 & 17.43 & 0.02 & - & 0.40 & 0.23 & & 0.802 & & 0.746 & + 32 & 1 & 283.7063948 & -30.4623137 & 0.5191334 & 3.85942 & 18.49 & 18.08 & 17.46 & 0.07 & 1.27 & 0.90 & 0.72 & 0.517 & 0.772 & 0.443 & 0.644 & bl + 33 & 1 & 283.7890960 & -30.5073081 & 0.4930339 & 3.51936 & 18.75 & 18.13 & 17.59 & 0.06 & 1.12 & 0.77 & 0.74 & 0.647 & 0.604 & 0.633 & 0.560 & bl + 34 & 1 & 283.7473555 & -30.5219780 & 0.5034071 & 3.83630 & 18.68 & 18.16 & 17.61 & 0.03 & 1.38 & 1.14 & 0.70 & 0.527 & 0.767 & 0.524 & 0.604 & + 35 & 1 & 283.7377861 & -30.4650676 & 0.5267239 & 9.78589 & 18.66 & 18.20 & 17.59 & 0.03 & 1.41 & 1.09 & 0.74 & 0.587 & 0.734 & 0.500 & 0.639 & + 36 & 1 & 283.8053328 & -30.4637899 & 0.5990345 & 9.66790 & 18.70 & 18.16 & 17.47 & 0.02 & 0.80 & 0.60 & 0.41 & 0.619 & 0.757 & 0.551 & 0.701 & + 37 & 1 & 283.7784287 & -30.4907654 & 0.6274528 & 3.77083 & 18.74 & 18.21 & 17.49 & 0.01 & 0.62 & 0.48 & 0.23 & 0.566 & 0.794 & 0.529 & 0.727 & + 38 & 1 & 283.7424905 & -30.4697269 & 0.6111295 & 9.64313 & 18.58 & 18.04 & 17.30 & 0.02 & 0.60 & 0.43 & 0.28 & 0.565 & 0.788 & 0.542 & 0.743 & + 39 & 1 & 283.7323257 & -30.4975579 & 0.5995711 & 3.77083 & & 18.19 & 17.53 & 0.02 & - & 0.76 & 0.48 & & 0.755 & & 0.680 & + 40 & 1 & 283.7483469 & -30.5102175 & 0.5863672 & 2.66921 & 18.74 & & 17.51 & - & 0.90 & - & 0.50 & & & & & + 41 & 1 & 283.8052158 & -30.4654648 & 0.6176814 & 3.68530 & 18.77 & 18.15 & 17.43 & 0.05 & 0.97 & 1.04 & 0.61 & 0.594 & 0.870 & 0.613 & 0.752 & ev + 42 & 2 & 283.7866761 & -30.4629200 & 0.3264348 & 9.07901 & & 18.19 & 17.65 & 0.02 & - & 0.49 & 0.31 & & & & 0.553 & + 43 & 1 & 283.7150492 & -30.4659484 & 0.5928533 & 4.64542 & 18.73 & 18.20 & 17.46 & 0.02 & 0.87 & 0.80 & 0.57 & 0.587 & 0.830 & 0.542 & 0.757 & bl + 44 & 1 & 283.7688542 & -30.5014846 & 0.6207049 & 9.49051 & 18.77 & 18.25 & 17.52 & 0.02 & 0.61 & 0.39 & 0.27 & 0.590 & 0.770 & 0.538 & 0.732 & + 45 & 1 & 283.8036281 & -30.5080007 & 0.4888223 & 3.47103 & 18.62 & 18.24 & 17.63 & 0.03 & 1.32 & 1.02 & 0.62 & 0.530 & 0.752 & 0.427 & 0.640 & + 46 & 1 & 283.7530186 & -30.4903405 & 0.6047506 & 4.52668 & 18.65 & 18.08 & & 0.02 & 0.64 & 0.56 & - & 0.603 & & 0.581 & & sgr + 47 & 1 & 283.7542761 & -30.4532285 & 0.5069847 & 2.88684 & 18.62 & 18.22 & 17.61 & 0.02 & 0.44 & 0.70 & 0.41 & 0.332 & 0.670 & 0.397 & 0.618 & bl +
the sample of variables detected in our survey is mainly constituted by rrl stars , as expected from an old metal - poor globular cluster ( gc ) . in this section
we discuss the mebership of the various rrl , attempting to define a sample of m54 rrl as clean as possible from interlopers and/or stars with uncertain or possibly problematic data .
then , we will use the cleaned sample to analyze the pulsational properties of the cluster rrl . )
are marked with black crosses .
the two dashed lines indicate the edges of the instability strip ( from bono et al .
1995 ) . the @xmath34 colour and @xmath35 magnitude distributions are shown in the upper - left and bottom - right panels , respectively.,width=328 ] we show in fig . [ cmdrr ] the distribution in magnitude and colour of the rrl stars for which we obtained calibrated photometry .
the magnitude histogram in the lower - right panel shows that the distribution is dominated by a strong and relatively narrow peak around @xmath36 , clearly associated to the cluster population . the bulk of this population is well bracketed between the edges of the instability strip predicted by theoretical models ( bono , caputo & marconi 1995 ) for a 0.65 @xmath37 rrl star with z=0.0004 , overplotted to the cmd of fig .
[ cmdrr ] . here
we assume @xmath38 and @xmath39 , according to the results of sect .
[ s : dist ] , below .
there are several stars that lie outside the m54 peak , both in magnitude and colour .
there are 11 variables ( 9 rr0 and 2 rr1 ) having @xmath40 , clearly brighter than the bulk of the rrl population , i.e. , in order of increasing magnitude , v119 , v12 , v128 , v132 , v93 , v142 , v65 , v121 , v75 , v92 , and v69 . assuming @xmath41 and @xmath39 we can obtain rough estimates of the distances of these stars .
v119 has @xmath42 kpc , thus may be possibly located in the galactic bulge .
the others range between @xmath43 kpc and @xmath44 kpc ; while some of these are likely interlopers from the galactic thick disc and/or halo , the faintest ones are more likely members of m54 ( or sgr dsph ) that appear over - luminous because of blending with other sources and/or some unidentified problem in the photometry ; in fact four of them ( v92 , v128 , v132 and v142 ) lie in the most crowded region of our images , within @xmath45 from the cluster center .
v55 ( the faintest of the outliers , @xmath46 kpc ) is likely a genuine galactic star in the background .
finally there are five stars that have at least one colour clearly outside the rrl window ( v75 , v123 , v124 , v133 , v143 ; only v75 appears anomalous in both colours ) : this is likely due to a problem in the dophot photometry of the reference frame in one of the three bands .
v123 , that appears rather normal from fig .
[ cmdrr ] has a very red v - i . to keep our final sample of m54 rrl as clean and safe as possible , all the stars discussed above have been excluded from the analysis of the pulsational properties , in the following .
the rrl of this clean sample ( 55 rr0 and 9 rr1 ) have @xmath47 mag and @xmath48 . in fig .
[ f : rad ] the cumulative radial distribution of the rrl in the clean sample is compared with those of the blue hb stars ( bhb ; expected to be dominated by m54 population ) and red clump stars . as expected , both the rrl and the bhb stars are significantly more concentrated than the red - clump stars .
this indicates that the contamination by galactic and/or sgr variables should be minimal ( see also monaco et al .
2003 ) . in fig .
[ theo ] the zero age horizontal branch ( zahb ) model ( from cariulo , deglinnocenti & castellani 2004 ) for a population with metallicity @xmath49 and helium abundance @xmath50 is overplotted to the @xmath51 cmd of m54 .
two evolutionary tracks corresponding to mass values of @xmath52 and @xmath53 are overplotted .
absolute magnitudes have been converted to apparent ones adopting the reddening and distance modulus derived in sect .
[ s : redd ] and [ s : dist ] .
the zahb model fits quite well the lower envelope of the rrl distribution .
three stars are about 0.2 mag fainter than the average , namely v4 , v118 and v130 .
the variable v4 is affected by the blazhko effect and we have likely missed the maximum , as suggested by the unusually low v amplitude ( 0.71 mag ) for its period ( 0.48 d ) .
the other two stars have normal well defined light curves , and no metallicity estimates are available to check the origin of their unusual magnitude . and helium abundance @xmath50 , and the evolutionary tracks ( dashed lines ) for the two stellar masses m=0.65 and 0.70 @xmath37 ( from cariulo et al .
2004 ) are overplotted.,width=328 ] if we consider the entire sample of rrl , including also those for which only internal calibrated photometry could be obtained ( but excluding those flagged as non - member or problematic in the previous section ) , then we have 95 rr0 stars with @xmath54 = 0.584d and 33 rr1 stars with @xmath55 = 0.351d . for comparison , the average values for oosterhoff i type clusters are @xmath54 = 0.559d and @xmath55 = 0.326d , while for oosterhoff ii type clusters they are @xmath54 = 0.659d and @xmath55 = 0.368d ( smith 1995 ) .
the fraction of rr1 over the entire sample of rrl is 0.26 , which is slightly higher than the typical fraction detected in oosterhoff i clusters ( 0.22 ) and significantly smaller than the typical value ( 0.48 ) in oosterhoff ii clusters ( clement et al .
this indicates that the entire sample of rrl stars detected in the m54 region has intermediate characteristics between the classical oosterhoff types i and ii , being much closer to oosterhoff i. if we consider only the subset of stars that are most likely m54 members according to luminosity and metallicity criteria , as we discuss in sect .
[ s : met ] , we find @xmath54 = 0.600 d from 49 rr0 stars and @xmath55 = 0.347 d from 7 rr1 stars in the middle panel of fig .
[ pa ] the period - v amplitude diagram for the sample of rrl is shown .
the mean loci of the unevolved rr0 and rr1 variables in the ggcs m3 ( oosterhoff i ) and m92 ( oosterhoff ii ) are overplotted ( from cacciari , corwin & carney 2005 )
. taken at face value , the majority of the rr0 fall around the ridge line of m3 slightly on the side of smaller amplitudes or shorter periods .
some of these stars could well be unrecognised blazhko variables observed during the low amplitude phase of the blazhko modulation . indeed ,
blazhko effect has been clearly detected for 9 rrls of our sample . according to previous studies ( smith 1995 ; corwin & carney 2001 ; cacciari et al .
2005 ) as much as 30% of the total rrl population in several clusters is affected by blazhko modulation .
if the m54 population shares the same behaviour , then at least twice as many blazhko rrl variables , whose anomaly could not be detected because of the short time coverage of our observations , are expected to exist in the studied sample . within these uncertainties ,
the m54 data appear to be in good agreement with the mean relations derived for m3 , albeit with a somewhat larger scatter ( that may be due to the intrinsic metallicity spread of the cluster , see c10 ) . in the bottom panel of fig .
[ pa ] the period distribution of the rrl of m54 is compared with that of m3 .
the apparent contadiction between the location in the p - av diagram ( slightly on the side of shorter periods ) and the ( longer ) mean period of the m54 rrab stars with respect to m3 is an optical effect of the different amplitude distributions : while 44% of m3 rrab have amplitudes @xmath56 , this fraction is only 11% in m54 .
at least part of this difference might be due to the presence of a significant fraction of unrecognised blazhko variables .
as for the period distributions , which are not affected by the blazhko stars , it is evident that the m54 and m3 rrl period distributions are different : the m3 distribution is skewed towards the short period end whereas in m54 it is wider and more symmetrical ( bottom panel of fig.9 ) .
this accounts for a longer mean rrab period in m54 ( @xmath180.60d ) than in m3 ( @xmath180.56d ) .
one can also note a group of 8 stars with longer periods that seem to stand out from the main distribution .
these stars are v41 , v48 , v76 , v83 , v120 , v125 , v137 and v140 .
this location in the period - amplitude diagram can be due to evolution off the zahb or to lower metallicity . in both cases
the star is expected to be somewhat overluminous .
two of them ( v125 and v137 ) are indeed brighter than the average by more than 0.1 mag and might be evolved , for the other 6 stars the luminosity is quite normal .
the metallicity was estimated for 4 of these stars ( v41 , v48 , v76 and v83 ) and falls within about 0.2 dex of the mean value except for v83 that is very metal - poor ( see sect . [
s : met ] ) . in conclusion m54
rrl show oosterhoff i characteristics as far as the population ratio of fundamental / first overtone pulsators is concerned , while display average periods quite intermediate between the two types , mainly due to the presence of a significant number of long period rrl .
distribution , cleaned of field contaminants , possible blends and photometric errors .
fundamental ( rr0 ) and first overtone ( rr1 ) pulsators are marked with filled and open circles , respectively .
blazhko stars are indicated as asterisks .
the limiting magnitude of our selection criterion are marked with dashed lines .
middle panel : period - v amplitude diagram for the rrl of m54 .
the symbols are the same as the upper panel .
the mean loci of the unevolved rrl in m3 ( solid line ) and m92 ( dashed line ) are overplotted ( from cacciari et al . 2005 ) .
lower panel : period distributions for the rrl of m54 ( grey histogram ) and m3 ( open histogram).,width=328 ]
it was shown that appropriate combinations of terms of a fourier representation of rrl light curves , along with the periods , correlate with intrinsic parameters of the stars , such as metallicity , mass , luminosity and colours ( simon 1988 ; clement , jankulak & simon 1992 ; jurcsik & kovacs 1996 ; kovacs & walker 2001 ; morgan , wahl & wieckhorst 2007 ) . sandage ( 2004 ) interpreted the metallicity dependence of the fourier components of rrl light curves as equivalent to the oosterhoff - arp - preston period - metallicity effect .
this can be found in both fundamental and first overtone pulsators , that we consider here separately .
however , in several cases ( jurcsik & kovacs 1996 ; pritzl et al . 2001 ; nemec 2004 ; cacciari et al 2005 ; stetson , catelan & smith 2005 ) the coefficients as well as the range of validity of such relations have been discussed , often leading to different and conflicting results .
therefore , [ fe / h ] estimates derived from the fourier parameter @xmath57 should be used with caution , especially in the high metallicity regime . in the following we use these metallicites mainly to clean the sample of m54 rrl stars from possible members of sgr
, that would lie at the same distance but may stand out as significant outliers in the metallicity distribution . to this purpose
, we have decomposed the v light curves of all rrl variables of our sample in fourier series of cosines .
[ t : ffec ] @lcccl@ i d & period & @xmath57 & @xmath58_{zw}$ ] & note + & ( d ) & & & + + v12 & 0.3226390 & 4.277 & -1.01 & * + v42 & 0.3264349 & 3.140 & -1.73 & + v49 & 0.3312773 & 3.867 & -1.40 & + v56 & 0.3676695 & 3.741 & -1.81 & + v74 & 0.2889211 & 4.267 & -0.49 & * * + v78 & 0.4019678 & 4.473 & -1.72 & + v97 & 0.3367274 & 3.653 & -1.58 & + v126 & 0.3516532 & 3.810 & -1.65 & + v133 & 0.3648148 & 3.782 & -1.77 & + v138 & 0.3116047 & 3.059 & -1.62 & + v141 & 0.3339446 & 4.327 & -1.13 & * * + v142 & 0.2978301 & 2.333 & -1.79 & * + v148 & 0.2670186 & 4.149 & -0.19 & * * + the use of fourier coefficients to characterise rr1 variables has been less extensive than that of the rr0 stars .
the most recent critical analysis of this method has been presented by morgan et al .
we refer to this work for an extended discussion and references . according to these authors the metallicity of the rr1 stars
can be estimated using the following relation : @xmath59_{zw } = 52.466p^2 - 30.075p + 0.131\phi_{31}^2 + 0.982\phi_{31}\ ] ] @xmath60 where [ fe / h]@xmath61 is expressed in the zinn & west ( 1984 ) metallicity scale , and the coefficient @xmath57 comes from fourier light curve decomposition in series of cosines .
the standard deviation of this relation , obtained from 106 stars , is 0.145 dex .
the standard deviation of this relation , obtained from 106 rrc stars in 12 globular clusters , is 0.145 dex .
the [ fe / h ] values of the calibrating clusters are between approximately -1 and -2 , so this relation likely gives poorer results at very high or very low metallicities .
our sample contains 13 rr1 stars for which we have estimated the fourier parameter @xmath62 and hence the metallicity with the above relation .
the results are listed in table 2 .
two stars ( v12 and v142 ) have been assigned to the galactic field population based on their unusually bright v magnitude ( @xmath63 ) .
three more stars ( v74 , v141 and v148 ) have rather high metallicities suggesting that they may belong to the sgr dsph population .
the average metallicity of the 8 stars that are most likely m54 members is [ fe / h]= -1.66 , with a standard deviation of 0.13 dex . and @xmath47 .
the range of the adopted selection for the _ bona fide _ sample of m54 are marked with dashed lines.,width=325 ] from the analysis of v light curves of 272 rr0 stars , jurcsik & kovacs ( 1996 ) and jurcsik ( 1998 ) found a tight correlation ( r.m.s .
error of the fit 0.14 dex ) between the metallicity of the variables and their @xmath57 ( sine series ) fourier coefficents : @xmath58=-5.038 - 5.394~p + 1.345~\phi_{31}$ ] . based on the layden ( 1994 ) [ fe / h ] values of a large number of field rrab stars , sandage ( 2004 )
recalibrated the above relation and obtained a lower zero - point by about -0.25 dex that is consistent with the layden metallicity scale .
this relation , that we use in the present analysis , is : @xmath59_l = -5.492 - 5.666~p + 1.413~\phi_{31}\ ] ] where the r.m.s .
error of the fit remains @xmath18 0.14 dex .
this new formulation gives values of metallicity that are in good agreement with those obtained from spectroscopy or other methods , as we comment later .
since we are also interested in differential estimates , to separate the m54 from the ( milky way and sgr ) field populations , this method appears sufficiently adequate to identify at least the most obvious outliers .
we corrected the values of our cosine series @xmath57 by + @xmath64 to report them to sine series decomposition , and applied the above relation to calculate the metallicity .
the @xmath65 parameter ( defined by jurcsik & kovacs 1996 ) represents a quality test on the regularity of the shape of the light curves , and hence on the reliability of the derived physical parameters .
we considered as sufficiently reliable only stars with @xmath66 .
the values of @xmath57 and [ fe / h]@xmath67 for these 38 rr0 stars are listed in table 3 .
blazhko stars should be excluded , unless observed at or near maximum blazhko modulation amplitude with a good degree of confidence .
in fact , at small - amplitude blazhko phase the light curves are more likely distorted ( even if dm @xmath685 ) and tend to overestimate the metallicity , whereas at large - amplitude blazhko phase they are quite similar to regular pulsators ( see cacciari et al . 2005 ) . following these considerations ,
only v33 has been excluded from the sample . on the other hand , photometric blends or stars affected by calibration errors
can still produce good metallicity estimates , as only the zeropoint of the photometric scale is incorrect .
seven variables with a reliable metallicity estimate have been associated to the foreground galactic population because of their bright magnitudes . the distribution of the remaining 30 rr0 variables with reliable metallicity estimates is shown in fig .
[ f : met ] .
as can be noted , the distribution has a prominent peak at @xmath58\sim-1.6 $ ] and the majority of rr0 variables have metallicities that range between @xmath69<-1.2 $ ] .
a group of four variables ( v28 , v46 , v63 and v123 ) form a secondary peak at @xmath58\simeq -1.0 $ ] .
one variable ( v139 ) has a solar - like metallicity . finally , two variables ( v50 and v83 ) present a very low metallicity ( @xmath58<-2.2 $ ] ) .
the membership of these 7 stars with metallicities outside the range @xmath69<-1.2 $ ] is not clear cut .
the extreme metallicity of v139 , v50 and v83 strongly suggests a membership to sgr dsph , whose metallicity distribution tails clearly reach values from @xmath58\sim-2 $ ] up to nearly solar ( b08 ; monaco et al 2005b ; bonifacio et al .
in particular cseresnjes ( 2001 ) found a very wide metallicity distribution also for the rrl of sgr , with a mean of [ fe / h]=-1.6 , a dispersion of @xmath70 dex , and a minor but significant population at @xmath58\la$]-2.0 dex .
the membership of the four stars at @xmath58\simeq -1.0 $ ] is harder to establish , especially considering that m54 itself has been found ( from spectroscopic analyses ) to display a spread of @xmath71 dex ( da costa & armandroff 1995 ; b08 ; c10 ) . to be conservative we decided to exclude also these four stars from our final sample .
the remaining 23 rr0 variables that are likely members of m54 have a mean metallicity [ fe / h]= -1.65 @xmath1 0.18 dex .
the mean metallicity of the m54 rr0 stars is in excellent agreement with the mean value found for the rr1 in sect .
[ s : met_rrc ] .
so we can assume that the rr1 and rr0 metallicity estimates can be dealt with jointly , and the mean value of the combined sample is [ fe / h]= -1.65 , with a standard deviation of 0.16 dex , in good agreement with the spectroscopic estimates by da costa & armandroff ( 1995 ) , brown et al .
( 1999 ) , b08 and c10 . the uncertainties on individual measures , on the reliability of the @xmath72 $ ] relations and the conservative selection described above prevent any conclusion about the intrinsic metallicity spread .
by rejecting those rrl with measured metallicities outside the range @xmath69<-1.2 $ ] , in addition to the luminosity criteria described in sect .
[ s : cmd ] , we end up with a sample of 49 rr0 and 7 rr1 variables , that will be used in the following sections to estimate the distance modulus and the reddening of m54 . in the following
, we will refer to these stars as the _ bona - fide _ m54 sample .
[ t : ffe ] @lcccl@ i d & period & @xmath57 & [ fe / h ] & note + & ( d ) & & l84 & + + v3 & 0.5726301 & 1.880 & -1.64 & + v5 & 0.5790976 & 1.702 & -1.93 & + v7 & 0.5944093 & 1.956 & -1.65 & + v15 & 0.5868540 & 1.906 & -1.69 & + v28 & 0.5088183 & 2.030 & -1.07 & blazhko , * * + v29 & 0.5897109 & 2.115 & -1.41 & + v30 & 0.5738690 & 1.994 & -1.49 & + v32 & 0.5191334 & 1.521 & -1.85 & blazhko + v33 & 0.4930339 & 1.894 & -0.66 : & blazhko , low - ampl .
+ v34 & 0.5034071 & 1.786 & -1.38 & + v35 & 0.5267239 & 1.573 & -1.81 & + v41 & 0.6176814 & 2.202 & -1.44 & + v43 & 0.5928533 & 2.020 & -1.56 & blazhko + v45 & 0.4888223 & 1.404 & -1.84 & + v46 & 0.6047506 & 2.629 & -0.76 & * * + v48 & 0.6870139 & 2.262 & -1.75 & + v50 & 0.5636133 & 1.258 & -2.47 & * * + v55 & 0.4260544 & 2.154 & -0.42 & * + v58 & 0.5486673 & 1.797 & -1.62 & + v61 & 0.6039009 & 2.165 & -1.42 & + v63 & 0.5901396 & 2.361 & -1.06 & * * + v65 & 0.5748202 & 2.045 & -1.42 & * + v67 & 0.5900106 & 1.896 & -1.72 & + v69 & 0.6797878 & 2.006 & -2.07 & * + v75 & 0.5846341 & 1.907 & -1.67 & * + v76 & 0.7064914 & 2.378 & -1.70 & + v77 & 0.5769039 & 1.867 & -1.68 & + v82 & 0.5878700 & 2.020 & -1.53 & + v83 & 0.5783485 & 1.398 & -2.35 & * * + v85 & 0.5213111 & 1.381 & -2.06 & + v92 & 0.4840892 & 1.557 & -1.60 & * + v93 & 0.5585378 & 1.780 & -1.70 & * + v96 & 0.5585253 & 1.883 & -1.56 & + v119 & 0.6366165 & 2.356 & -1.33 & * + v123 & 0.5141767 & 2.112 & -0.98 & * * + v124 & 0.5822650 & 1.812 & -1.79 & + v129 & 0.6027318 & 2.060 & -1.56 & + v139 & 0.4039767 & 2.337 & -0.04 & * * +
our large sample of rrl allows us to determine the reddening to m54 with good accuracy .
we consider two approaches to estimate the reddening , using the colours of the rr0 variables at minimum light ( i.e. averaged over the phase interval 0.5 - 0.8 ) : * based on a recalibration of the original relation by sturch ( 1966 ) , walker ( 1998 ) proposed a relation linking @xmath73 and reddening @xmath74 : @xmath75 - 0.356\ ] ] applying this relation to the sample of _ bona - fide _ rr0 variables of m54 we find @xmath76 if we adopt an average metallicity @xmath58=-1.65 $ ] ( see sect .
[ s : met ] ) and @xmath77 if we adopt for each star the metallicity estimated through the @xmath72 $ ] relation defined in sect .
[ s : met_rrab ] * from the analysis of a sample of field rr0 stars mateo et al .
( 1995 ) proposed a constant value of @xmath78 for rr0 stars , independent on metallicity and period . by using our sample of _ bona - fide _
rr0 stars we find @xmath79 which translates into @xmath80 , and hence @xmath81 using the relation e(b - v)=0.8 e(v - i)(dean , warren & cousin 1978 ) . summarizing the above results
, we estimate an average value of @xmath82 .
this estimate is in very good agreement with that provided by the extinction map by schlegel , finkbeiner & davis ( 1998 , @xmath83 ) , and with the results by ls2000 from a similar analysis of rrl stars ( @xmath84 ) .
the distance to m54 has been estimated from the mean characteristics of the _ bona fide _ sample using the absolute visual magnitude m@xmath85 that can be derived from the luminosity - metallicity relation @xmath86 + 0.885\ ] ] ( clementini et al .
2003 ) where the adopted slope 0.214 appears to be supported by the most accurate studies of field rrl stars in the milky way ( fernley et al .
1998 ; chaboyer 1999 ) , in the large magellanic cloud ( lmc ; gratton et al . 2004 ) and gcs in m31 ( rich et al . 2005 ) .
the zero - point of such a relation has been set to be consistent with the mean dereddened v magnitude of the rr0 variables observed in the lmc by clementini et al .
( 2003 ) ( @xmath87 ) assuming a distance modulus of @xmath88 ( freedman et al .
2001 ) .
we use the mean metallicity @xmath58= -1.65 \pm 0.16 $ ] dex ( derived from the subset described in sect . [
s : met ] ) to estimate @xmath89 mag from the above equation , and the mean v magnitude @xmath90 mag ( derived from the stars that have been included in the _ bona fide _ sample ) .
therefore , assuming @xmath91 ( savage & mathis 1979 ) and e(b - v)=0.16 @xmath1 0.02 ( see sect .
[ s : redd ] ) we estimate a distance modulus to m54 @xmath92 , which translates into a distance to m54 of @xmath93 kpc . as a sanity check
, we estimate the distance modulus also adopting the individual fourier metallicities listed in tables 2 and 3 for the smaller sample of 30 + 7 rr0+rr1 variables , and we obtain @xmath94 .
the above estimates are in excellent agreement with those derived by monaco et al .
( 2004 ; @xmath95 ) , and by ls2000 , once their estimate is corrected using the most recent metallicity estimates ( @xmath96 ; see monaco et al .
a significantly smaller distance ( @xmath97 ) has been derived by kunder & chaboyer ( 2009 ) on the basis of the analysis of differential distance modulus of a sample of sgr rrl with respect to a sample of rrl belonging to the galactic bulge .
these authors adopt a different luminosity - metallicity relation ( from bono , caputo & di criscienzo 2007 ) , and their sample is affected by larger amount of extinction ( e(b - v)=0.36 ) , lying in fields at lower galactic latitude , from @xmath98 to @xmath99 apart from m54 .
moreover , the derived distance clearly depends on the assumed distance to the galactic bulge .
finally , a lower distance in that wing of the sgr dsph may be a real effect of the three dimesional orientation of the galaxy major axis .
all these factors can contribute to account for the difference between their distance estimate and our results ( as well as those by ls2000 and monaco et al .
2004 ; see the latter paper for a detailed discussion and references on previous distance estimates ) .
apart from the rrl , a number of other variables have been found in the present survey .
variables v1 and v2 are cepheids already discovered by rosino & nobili ( 1959 ) and classified by ls2000 as a population ii and an anomalous cepheid , respectively .
their magnitudes are consistent with those predicted by the period - luminosity relation of nemec , nemec & lutz ( 1994 ) and are located at 3.2 and 3.5 arcmin from the cluster center , respectively . on the basis of these considerations they should be likely members of m54 ( or sgr dsph ) .
among eclipsing binaries , stars v116 and v195 are algol binaries ( ea ) , whereas v147 and v211 are suspected eb types .
five variables ( namely v99 , v117 , v134 , v135 , and v144 ) are probably w uma ( ew ) . by applying the period - colour - v luminosity relations by rucinski ( 2000 ) we estimated the distance moduli of these stars . among these 5 stars , only v134 and v144 have distance moduli @xmath100 which are compatible with that of m54 estimated in sect . [
s : dist ] , being probable cluster members .
we found 3 sx phoenicis variables ( v146 , v149 and v150 ) with periods in the range @xmath101 . adopting the period - luminosity relation by poretti et al .
( 2008 ) we estimated the distances for these stars . v146 and v149 are significantly less distant than m54 ( @xmath102 mag ) and are likely foreground galactic variables .
the distance modulus of v150 is compatible ( within 0.3 mag ) with that of m54 and could be a cluster or sgr member . finally , we identified 18 candidate long - period variables ( lpv ) .
four of them ( v153 , v154 , v155 and v156 ) are new discoveries .
they are located in the cmd close to the tip of the rgb at colours @xmath103 ( see fig .
[ lpv ] ) .
most of them are located far from the rgb ridge line of m54 , on the expected locus of late m giants of sgr , being likely associated to sgr . of course , the short time coverage of our observations does not allow us to establish their pulsation periods and their average magnitudes and colours are uncertain .
we presented a survey of variable stars in a 13 arcmin square field of view centred on the gc m54 using deep multi - epoch observations collected with the eso - danish 1.54 m telescope . from this analysis we detected 177 variables , 94 of them never observed before .
we provide the ephemerides for the entire sample of variables ( except for 18 long - period candidates ) , and the epoch - photometry and light curves for the large subset of variables with calibrated photometry .
the period - amplitude diagram of the rrl of m54 and the relative fraction of rr1 variables indicate that this cluster shares some of the properties of oosterhoff i clusters ( as already found by ls2000 ) , but contains a larger number of long - period variables .
the average period of its rr0 and rr1 stars are indeed slightly larger than typical oosterhoff i cluster values , even if significantly smaller than oosterhoff ii .
accurate estimates of the cluster reddening ( @xmath82 ) and distance modulus ( @xmath92 ) were also derived , in excellent agreement with previous estimates in literature .
the metallicity distribution of the rrl , derived through the fourier parameter @xmath57 , shows a clear peak at the metallicity of m54 ( @xmath58 \sim -1.65 \pm 0.16 $ ] dex ) and a few stars with a significantly different metallicity ( outside the range @xmath69<-1.2 $ ] ) , likely associated with the population of the sgr dsph .
the presence of a extended blue hb ( reaching the instability strip ) associated with the sgr galaxy has been already discovered by monaco et al .
we note that the presence of rrl stars as metal - rich as @xmath58\ga -0.8 $ ] ( see also cseresjnes 2001 ) , if confirmed , would indicate that sgr was able to reach such relatively high metallicity at a very early epoch ( @xmath104 gyr ago ) .
unfortunately , the small number of stars belonging to this group and the large uncertainties in the metallicities do not allow a firm conclusion on this issue until spectroscopic confirmation is obtained .
this work is largely based on the master thesis of by s. colucci ( 2000 ) , at the bologna university .
we warmly thank christine clement , the referee of our paper , for her helpful comments and suggestions that have helped us improve the paper significantly .
we also thank alfred rosenberg for providing his photometric catalog of m54 that was used for the calibration of the b photometry .
this research was partly supported by the prin - inaf 2007 grant cra 1.06.10.04 assigned to the project : `` the local route to galaxy formation '' and by the spanish ministry of science and innovation ( micinn ) under the grant aya2007 - 65090 .
199 alard c. , 2000 , a&as , 144 , 363 barning f. j. m. , 1963 , ban , 17 , 22 bassino l.p .
, muzzio j.c . , 1995 , the observatory , 115 , 256 bellazzini m. , ibata r.a . , chapman s.c . , et al . 2008 ,
aj , 136 , 1147 ( b08 ) bonifacio p. , zaggia s. , sbordone l. , et al .
2006 , in chemical abundances and mixing in stars in the milky way and its satellites , ed .
s. randich & l. pasquini , berlin : springer , p. 232
bono g. , caputo f. , di criscienzo m. , 2007 , a&a , 476 , 779 bono g. , caputo f. , marconi m. , 1995 , aj , 110 , 2365 brown j. a. , wallerstein g. , gonzalez g. , 1999 , aj , 118 , 1245 cacciari c. , bellazzini m. , colucci s. , 2002 , extragalactic star clusters , 207 , 168 cacciari c. , corwin t. m. , carney b. w. , 2005 , aj , 129 , 267 cariulo p. , deglinnocenti s. , castellani v. , 2004 , a&a , 421 , 1121 carretta e. , bragaglia a. , gratton r. , et al . , 2010 ,
apj , in press ( arxiv:1002.1963 ; c10 ) catelan , m. 2009 , ap&ss , 320 , 261 chaboyer b. , 1999 , post - hipparcos cosmic candles , 237 , 111 clement c. m. , jankulak m. , simon n. r. , 1992 , apj , 395 , 192 clement c. m. , muzzin a. , dufton q. , et al . , 2001 , aj , 122 , 2587 clementini g. , di tomaso s. , di fabrizio l. , et al . , 2000 ,
aj , 120 , 2054 clementini g. , gratton r. , bragaglia a. , carretta e. , di fabrizio l. , maio m. , 2003 , aj , 125 , 1309 corwin t. m. , carney b. w. , 1998 , information bulletin on variable stars , 4547 , 1 corwin t. m. , carney b. w. 2001 , aj , 122 , 3183 corwin , t.m . , et al . , 2006 ,
aj , 132 , 1014 cseresnjes p. , 2001
, a&a , 375 , 909 da costa g. s. , armandroff t. e. , 1995 , aj , 109 , 2533 dean j. f. , warren p. r. , cousins a. w. j. , 1978 , mnras , 183 , 569 fernley j. , skillen i. , carney b. w. , cacciari c. , janes k. , 1998 , mnras , 293 , l61 freedman w. l. , madore b. f. , gibson b. k. , et al . , 2001 ,
apj , 553 , 47 gratton r. g. , bragaglia a. , clementini g. , carretta e. , di fabrizio l. , maio m. , taribello e. , 2004 , a&a , 421 , 937 greco c. , clementni g , catelan m. , et al . , 2009 , apj , 701 , 1323 ibata r. a. , gilmore g. , irwin m. j. , 1994 , nature , 370 , 194 ibata r. , bellazzini m. , chapman s. c. , et al . , 2009 , apj , 699 , l169 jurcsik j. , 1998 , a&a , 333 , 571 jurcsik j. , kovacs g. , 1996 , a&a , 312 , 111 kovcs g. , walker a. r. , 2001 , a&a , 374 , 264 kunder a. , chaboyer b. , 2009 , aj , 137 , 4478 layden a. c. , 1994 , aj , 108 , 1016 layden a.c .
, sarajedini a. 2000 , aj , 119 , 1760 ( ls2000 ) lomb n. r. , 1976 , ap&ss , 39 , 447 majewski s.r . ,
skruskie m.f .
, weinberg a.d . , ostheimer j.c . , 2003 , apj , 599 , 1082 mateo m. , kubiak m. , szymanski m. , kaluzny j. , krzeminski w. , udalski a. , 1995 , aj , 110 , 1141 mochejska , b.j . , 2004 ,
aj , 128 , 312 monaco l. , bellazzini m. , ferraro f. r. , pancino e. , 2003 , apj , 597 , l25 monaco l. , bellazzini m. , ferraro f. r. , pancino e. , 2004 , mnras , 353 , 874 monaco l. , bellazzini m. , ferraro f.r , pancino e. , 2005a , mnras , 356 , 1396 monaco , l. , bellazzini m. , bonifacio p. , et al . , 2005b , a&a , 441 , 141 morgan s.m . , wahl j.n . , wieckhorst r.m
, 2007 , mnras , 374 , 1421 nemec j. m. , 2004 , aj , 127 , 2185 nemec j. m. , nemec a. f. l. , lutz t. e. , 1994 , aj , 108 , 222 olech a. , woniak p. r. , alard c. , kaluzny j. , thompson i. b. , 1999 , mnras , 310 , 759 oosterhoff p.th . , 1939 , observatory , 62 ,
104 poretti e. , clementini g. , held e. v. , et al .
, 2008 , apj , 685 , 947 pritzl b. j. , smith h. a. , catelan m. , sweigart a. v. , 2001 , aj , 122 , 2600 pryor , c. & meylan , g. 1993 , in structure and dynamics of globular clusters , asp conf .
s. djorgovsky & g. meylan ( san francisco , ca : asp ) , 357 rich r. m. , corsi c. e. , cacciari c. , federici l. , fusi pecci f. , djorgovski s. g. , freedman w. l. , 2005 , aj , 129 , 2670 rosenberg a. , recio - blanco a. , garca - marn m. , 2004 , apj , 603 , 135 rosino l. , nobili f. , 1959 , asiago contrib .
, 97 , 1 rucinski s. m. , 2000 , aj , 120 , 319 sandage a. , 2004 , aj , 128 , 858 sarajedini a. , layden a.c . , 1995 ,
aj , 109 , 1086 savage b. d. , mathis j. s. , 1979 , ara&a , 17 , 73 schechter p. , mateo m. , saha a. 1993 , pasp , 105 , 1342 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 siegel m.h .
, dotter a. , majewski s.r . ,
et al 2007 , apj , 667 , l57 simon n. r. , 1988 , apj , 328 , 747 skrutskie m.f .
, cutri r. m. , stiening r. , et al . , 2006 ,
aj , 131 , 1163 smith h. a. , 1995 , in `` rr lyrae '' , cambridge astrophysics series , cambridge , new york , cambridge university press stetson p. b. , catelan m. , smith h. a. , 2005 , pasp , 117 , 1325 sturch c. , 1966 , apj , 143 , 774 walker a. r. , 1998 , aj , 116 , 220 zinn r. , west m. j. , 1984 , apjs , 55 , 45
* v4 : v magnitude @xmath18 0.2 mag fainter than the bulk m54 rrl population . affected by the blazhko effect , the maximum has likely been missed , as suggested by the unusually low v amplitude ( 0.71 mag ) for its period ( 0.48 d ) . *
v12 : we have derived a more accurate period ( p=0.322639 ) , which fits both our data and ls2000 data , whereas ls2000 estimate ( p=0.31985 ) produces out - of - phase curves . the star is bright and its fourier metallicity is [ fe / h]=-1.01 , this suggests that it may belong to the galactic field population . *
v14 : classified as rr0 with p=0.70489 by ls2000 , this star shows a good light curve ( internal photometry only ) with p=0.4807214 in clear disagreement also with rosino & nobili ( 1959 ) determination . *
v28 : metal - rich ( [ fe / h]=-1.07 ) .
normal v magnitude and position in the av - logp diagram .
possible sgr member .
* v41 : slightly shifted in the av - logp plane towards long periods .
normal magnitude .
fourier metallicity [ fe / h]= -1.44 .
* v46 : metal - rich ( [ fe / h]=-0.76 ) .
normal v magnitude and position in the av - logp diagram .
possible sgr member .
* v48 : shifted in the av - logp plane towards long periods .
normal magnitude .
fourier metallicity [ fe / h]= -1.75 .
possibly belonging to sgr .
* v50 : metal - poor ( [ fe / h]=-2.47 ) .
normal v magnitude and position in the av - logp diagram .
possible sgr member .
* v55 : faint magnitude , likely background galactic field star .
this is confirmed by its location in the p - av diagram , much to the short - period side of the main distribution for the m54 stars , also supported by the metallicity obtained from the fourier parameters ( [ fe / h]=-0.43 ) .
* v63 : metal - rich ( [ fe / h]=-1.06 ) .
normal v magnitude and position in the av - logp diagram .
possible sgr member .
* v65 : bright .
fourier metallicity [ fe / h]= -1.42 .
normal position in the av - logp diagram , likely field star .
* v69 : bright .
fourier metallicity [ fe / h]= -2.07 . shifted toward long periods in the av - logp diagram , field star or evolved .
* v74 : metal - rich ( [ fe / h]=-0.49 ) .
normal v magnitude and position in the av - logp diagram .
possible evolved or sgr member .
* v75 : unusually red colours ( b - v = 0.78 ; v - i = 0.45 ) . slightly brighter than average .
fourier metallicity [ fe / h]=-1.67 dex .
normal position in the av - logp diagram .
photometric calibration error ( see sect . [
s : cmd ] ) .
* v76 : shifted in the av - logp plane towards long periods .
normal magnitude .
fourier metallicity [ fe / h]= -1.70 . possibly belonging to sgr .
* v83 : metal - poor ( [ fe / h]=-2.35 ) .
normal v magnitude . slightly shifted in the av - logp plane towards long periods .
possible sgr member .
* v92 : slightly brighter than average and redder than expected from its period .
fourier metallicity [ fe / h]=-1.60 dex . located within central 1 arcmin , possibly a blend .
* v93 : bright .
fourier metallicity [ fe / h]=-1.71 dex .
likely field star . *
v118 : v magnitude @xmath18 0.2 mag fainter than the bulk m54 rrl population .
normal location in the av - logp plane .
* v119 : very bright .
fourier metallicity [ fe / h]=-1.33 dex .
foreground star ( galactic population ) .
* v120 : shifted in the av - logp plane towards long periods .
normal magnitude . possibly belonging to sgr .
* v121 : bright . likely field star . *
v123 : unusually red v - i colour ( v - i = 1.00 ) in spite of normal b - v ( b - v=0.47 ) .
fourier metallicity [ fe / h]=-0.98 .
photometric calibration error ( see sect . [
s : cmd ] ) .
* v124 : unusually red b - v colour ( b - v = 0.94 ) in spite of normal v - i ( v - i=0.50 ) .
fourier metallicity [ fe / h]=-1.79 .
photometric calibration error ( see sect . [
s : cmd ] ) .
* v125 : shifted in the av - logp plane towards long periods .
v magnitude @xmath18 0.1 brighter than the bulk m54 rrl population
. probably evolved variable . *
v128 : bright .
normal location in the av - logp plane . located within the central 1 arcmin , possibly a blend with a field `` blue plume '' star .
* v130 : v magnitude @xmath18 0.2 mag fainter than the bulk m54 rrl population .
normal location in the av - logp plane .
* v132 bright , and redder than the assumed red edge of the instability strip .
normal location in the av - logp plane . located within the central 1 arcmin , possibly a blend with a `` red clump '' star .
* v133 : unusually blue b - v colour ( b - v = 0.18 ) .
photometric calibration error ( see sect . [
s : cmd ] ) .
* v137 : shifted in the av - logp plane towards long periods .
v magnitude @xmath18 0.1 brighter than the bulk m54 rrl population
. probably evolved variable .
* v139 : the period - amplitude values locate the star among the rr1 variables , however the light curve shows the typical shape of a fundamental pulsator . the metallicity obtained from the fourier parameters is [ fe / h]=-0.04
. therefore this star likely belongs to the metal - rich ( @xmath18 solar metallicity ) tail of the sgr field stellar population .
* v140 : shifted in the av - logp plane towards long periods .
normal magnitude . possibly belonging to sgr .
* v141 : metal - rich ( [ fe / h]=-1.13 ) .
normal v magnitude and position in the av - logp diagram .
possible sgr member .
* v142 : bright . located within central 1 arcmin , possibly a blend .
* v143 : unusually red colour ( b - v = 0.99 ) .
photometric calibration error ( see sect . [
s : cmd ] ) .
* v145 : the classification of this star is highly uncertain . the data from the first two night would be compatible with a rr1 , those from the third night are not .
the period and the complete light curve may suggest a classification as a rs cvn variable , with a possible alternative period of 1.053375 days .
its position in the cmd ( straight on the cluster hb ) may indicate a similarity with the case of the superposition of two rrl discussed by corwin & carney ( 1998 ) .
* v147 : period and the b light curve are quite typical of rr0 stars , with an amplitude of @xmath105 mag .
the v light curve does not seem compatible with this interpretation , probably due to an error in the photometric zero point ( from dophot photometry of the v reference frame ) .
for this reason we flag this variable as uncertain .
* v211 : uncertain classification . | we present new b , v and i ccd time - series photometry for 177 variable stars in a @xmath0 field centered on the globular cluster m54 ( lying at the center of the sagittarius dwarf spheroidal galaxy ) , 94 of which are newly identified variables .
the total sample is composed of 2 anomalous cepheids , 144 rr lyrae stars ( 108 rr0 and 36 rr1 ) , 3 sx phoenicis , 7 eclipsing binaries ( 5 w uma and 2 algol binaries ) , 3 variables of uncertain classification and 18 long - period variables .
the large majority of the rr lyrae variables likely belong to m54 .
ephemerides are provided for all the observed short - period variables .
the pulsational properties of the m54 rr lyrae variables are close to those of oosterhoff i clusters , but a significant number of long - period ab type rr lyrae are present .
we use the observed properties of the rr lyrae to estimate the reddening and the distance modulus of m54 , e(b - v)=0.16 @xmath1 0.02 and ( m - m)@xmath2=17.13 @xmath1 0.11 , respectively , in excellent agreement with the most recent estimates . the metallicity has been estimated for a subset of 47 rr lyrae stars , with especially good quality light curves , from the fourier parameters of the v light curve . the derived metallicity distribution has a symmetric bell shape , with a mean of @xmath3\rangle= -1.65 $ ] and a standard deviation @xmath4 dex .
seven stars have been identified as likely belonging to the sagittarius galaxy , based on their too high or too low metallicity .
this evidence , if confirmed , might suggest that old stars in this galaxy span a wide range of metallicities .
[ firstpage ] methods : observational techniques : photometric stars : distances stars : variables : rr lyrae
globular clusters : m54 |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.