Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
users personal data , such as a user s historic interaction with the search engine ( e.g. , submitted queries , clicked documents ) , have been shown useful to personalize search results to the users information need @xcite . crucial to effective search personalization is the construction of user profiles to represent individual users interests @xcite . a common approach is to use main topics discussed in the user s clicked documents @xcite , which can be obtained by using a human generated ontology as in @xcite or using an unsupervised topic modeling technique as in @xcite . however , using the user profile to directly personalize a search has been not very successful with a _ minor _ improvement @xcite or even _ deteriorate _ the search performance @xcite . the reason is that each user profile is normally built using only the user s relevant documents ( e.g. , clicked documents ) , ignoring user interest - dependent information related to input queries . alternatively , the user profile is utilized as a feature of a multi - feature learning - to - rank ( l2r ) framework @xcite . in this case , apart from the user profile , dozens of other features has been proposed as the input of an l2r algorithm @xcite . despite being successful in improving search quality , the contribution of the user profile is not very clear . to handle these problems , in this paper , we propose a new _ embedding _ approach to constructing a user profile , using both the user s input queries and relevant documents . we represent each user profile using two projection matrices and a user embedding . the two projection matrices is to identify the user interest - dependent aspects of input queries and relevant documents while the user embedding is to capture the relationship between the queries and documents in this user interest - dependent subspace . we then _ directly _ utilize the user profile to re - rank the search results returned by a commercial search engine . experiments on the query logs of a commercial web search engine demonstrate that modeling user profile with embeddings helps to significantly improve the performance of the search engine and also achieve better results than other comparative baselines @xcite do . we start with our new embedding approach to building user profiles in section [ ssec : profile ] , using pre - learned document embeddings and query embeddings . we then detail the processes of using an unsupervised topic model ( i.e. , latent dirichlet allocation ( lda ) @xcite ) to learn document embeddings and query embeddings in sections [ ssec : topics ] and [ ssec : query ] , respectively . we finally use the user profiles to personalize the search results returned by a commercial search engine in section [ ssec : rank ] . let @xmath0 denote the set of queries , @xmath1 be the set of users , and @xmath2 be the set of documents . let @xmath3 represent a triple @xmath4 . the query @xmath5 , user @xmath6 and document @xmath7 are represented by vector embeddings @xmath8 , @xmath9 and @xmath10 , respectively . our goal is to select a _ score function _ @xmath11 such that the implausibility value @xmath12 of a correct triple @xmath3 ( i.e. @xmath13 is a relevant document of @xmath14 given @xmath15 ) is _ smaller _ than the implausibility value @xmath16 of an incorrect triple @xmath17 ( i.e. @xmath18 is not a relevant document of @xmath19 given @xmath20 ) . inspired by embedding models of entities and relationships in knowledge bases @xcite , the score function @xmath11 is defined as follows : @xmath21 here we represent the profile for the user @xmath14 by two matrices @xmath22 and @xmath23 and a vector embedding @xmath9 , which represents the user s topical interests . specifically , we use the interest - specific matrices @xmath22 and @xmath24 to identify the interest - dependent aspects of both query @xmath15 and document @xmath13 , and use vector @xmath9 to describe the relationship between @xmath15 and @xmath13 in this interest - dependent subspace . in this paper , @xmath25 and @xmath8 are pre - determined by employing the lda topic model @xcite , which are detailed in next sections [ ssec : topics ] and [ ssec : query ] . our model parameters are only the user embeddings @xmath9 and matrices @xmath22 and @xmath24 . to learn these user embeddings and matrices , we minimize the margin - based objective function : @xmath26 where @xmath27 is the margin hyper - parameter , @xmath28 is the training set that contains only correct triples , and @xmath29 is the set of incorrect triples generated by corrupting the correct triple @xmath30 ( i.e. replacing the relevant document / query @xmath31 in @xmath30 by irrelevant documents / queries @xmath32 ) . we use stochastic gradient descent ( sgd ) to minimize @xmath33 , and impose the following constraints during training : @xmath34 , @xmath35 and @xmath36 . first , we initialize user matrices as identity matrices and then fix them to only learn the randomly initialized user embeddings . then in the next step , we fine - tune the user embeddings and user matrices together . in all experiments shown in section [ sec : expsetup ] , we train for 200 epochs during each two optimization step . in this paper , we model document embeddings by using topics extracted from relevant documents . we use lda @xcite to _ automatically _ learn @xmath37 topics from the relevant document collection . after training an lda model to calculate the probability distribution over topics for each document , we use the topic proportion vector of each document as its document embedding . specifically , the @xmath38 element ( @xmath39 ) of the vector embedding for document @xmath13 is : @xmath40 where @xmath41 is the probability of the topic @xmath42 given the document @xmath13 . we also represent each query as a probability distribution @xmath43 over topics , i.e. the @xmath38 element of the vector embedding for query @xmath15 is defined as : @xmath44 where @xmath45 is the probability of the topic @xmath42 given the query @xmath15 . following @xcite , we define @xmath45 as a mixture of lda topic probabilities of @xmath42 given documents related to @xmath15 . let @xmath46 be the set of top @xmath47 ranked documents returned for a query @xmath15 ( in the experiments we select @xmath48 ) . we define @xmath45 as follows : @xmath49 where @xmath50 is the exponential decay function of @xmath51 which is the rank of @xmath52 in @xmath53 . and @xmath54 is the decay hyper - parameter ( @xmath55 ) . the decay function is to specify the fact that a higher ranked document is more relevant to user in term of the lexical matching ( i.e. we set the larger mixture weights to higher ranked documents ) . we utilize the user profiles ( i.e. , the learned user embeddings and matrices ) to re - rank the original list of documents produced by a commercial search engine as follows : ( 1 ) we download the top @xmath47 ranked documents given the input query @xmath15 . we denote a downloaded document as @xmath13 . ( 2 ) for each document @xmath13 we apply the trained lda model to infer the topic distribution @xmath56 . we then model the query @xmath15 as a topic distribution @xmath43 as in section [ ssec : query ] . ( 3 ) for each triple @xmath3 , we calculate the implausibility value @xmath12 as defined in equation [ equa : stranse ] . we then sort the values in the ascending order to achieve a new ranked list . * dataset : * we evaluate our new approach using the search results returned by a commercial search engine . we use a dataset of query logs of of 106 anonymous users in 15 days from 01 july 2012 to 15 july 2012 . a log entity contains a user identifier , a query , top-@xmath57 urls ranked by the search engine , and clicked urls along with the user s dwell time . we also download the content documents of these urls for training lda @xcite to learn document and query embeddings ( sections [ ssec : topics ] and [ ssec : query ] ) . bennett _ et al . _ @xcite indicate that short - term ( i.e. session ) profiles achieved better search performance than the longer - term profiles . short - term profiles are usually constructed using the user s search interactions within a search session and used to personalize the search within the session @xcite . to identify a search session , we use 30 minutes of user inactivity to demarcate the session boundary . in our experiments , we build short - term profiles and utilize the profiles to personalize the returned results . specifically , we uniformly separate the last log entries within search sessions into a _ test set _ and a _ validation set_. the remainder of log entities within search sessions are used for _ training _ ( e.g. to learn user embeddings and matrices in our approach ) . * evaluation methodology : * we use the sat criteria detailed in @xcite to identify whether a clicked url is relevant from the query logs ( i.e. , a sat click ) . that is either a click with a dwell time of at least 30 seconds or the last result click in a search session . we assign a positive ( relevant ) label to a returned url if it is a sat click . the remainder of the top-10 urls is assigned negative ( irrelevant ) labels . we use the rank positions of the positive labeled urls as the ground truth to evaluate the search performance before and after re - ranking . we also apply a simple pre - processing on these datasets as follows . at first , we remove the queries whose positive label set is empty from the dataset . after that , we discard the domain - related queries ( e.g. facebook , youtube ) . to this end , the training set consists of 5,658 correct triples . the test and validation sets contain 1,210 and 1,184 correct triples , respectively . table [ table:1 ] presents the dataset statistics after pre - processing . .basic statistics of the dataset after pre - processing [ cols="^,^,^,^,^,^",options="header " , ] by directly learning user profiles and applying them to re - rank the search results , our embedding approach achieves the highest performance of search personalization . specifically , our mrr score is significantly ( @xmath58 ) higher than that of _ sp _ ( with the relative improvement of 4% over sp ) . likewise , the p@1 score obtained by our approach is significantly higher than that of the baseline _ sp _ ( @xmath59 ) with the relative improvement of 11% . in table [ tb1 ] , we also present the performances of a simplified version of our embedding approach where we fix the user matrices as identity matrices and then only learn the user embeddings . table [ tb1 ] shows that our simplified version achieves second highest scores compared to all others . ) than our simplified version with 4% relative improvement . ] specifically , our simplified version obtains significantly higher p@1 score ( with @xmath58 ) than _ in this paper , we propose a new embedding approach to building user profiles . we model each user profile using a user embedding together with two user matrices . the user embedding and matrices are then learned using lda - based vector embeddings of the user s relevant documents and submitted queries . applying it to web search , we use the profile to re - rank search results returned by a commercial web search engine . our experimental results show that the proposed method can stably and significantly improve the ranking quality . * acknowledgments * : the first two authors contributed equally to this work . dat quoc nguyen is supported by an international postgraduate research scholarship and a nicta nrpa top - up scholarship .	recent research has shown that the performance of search personalization depends on the richness of user profiles which normally represent the user s topical interests . in this paper , we propose a new embedding approach to learning user profiles , where users are embedded on a topical interest space . we then directly utilize the user profiles for search personalization . experiments on query logs from a major commercial web search engine demonstrate that our embedding approach improves the performance of the search engine and also achieves better search performance than other strong baselines .
despite the large progress in @xmath0-ray source localization made by the _ fermi _ large area telescope ( lat ) @xcite coupled with the improvements in our knowledge of the diffuse galactic @xmath0-ray emission ( e.g. , * ? ? ? * ; * ? ? ? * ) , the positional uncertainties of the _ fermi _ sources are still large with respect to typical localization precisions at x - ray and lower energies , making a significant fraction of the sources in the @xmath0-ray sky yet unknown @xcite . about 43% of the _ fermi _ sources detected in the _ fermi_-large area telescope ( lat ) first source catalog ( 1fgl * ? ? ? * ) were listed as ugss while there were @xmath133% @xmath0-ray objects unassociated in the _ fermi_-lat 2-year source catalog ( 2fgl * ? ? ? * ) . to decrease the number of unidentified gamma - ray sources ( ugss ) many methods based on multifrequency approaches or statistical analyses have been recently adopted . radio follow up observations of the _ fermi _ ugss have already been performed ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? ? * ; * ? ? ? * ) and the _ swift _ x - ray survey for all the ugss listed in the _ fermi _ source catalogs is still on going[multiblock footnote omitted ] ( e.g. , * ? ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? additional x - ray observations performed with _ chandra _ and _ suzaku _ improved our knowledge on the ugss ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? statistical studies based on the @xmath0-ray source properties also allowed us to recognize the nature of the potential counterparts for the ugss ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? moreover a tight connection between the infrared ( ir ) sky seen by wide - field infrared survey explorer ( _ wise _ * ? ? ? * ) and the _ fermi _ one has been recently discovered for blazars , the rarest class of active galaxies @xcite . these works greatly decreased the fraction of ugss with no assigned counterpart at low energies @xcite . more recently also low frequency radio observations ( i.e. , below @xmath11 ghz ) revealed a new spectral behavior that allowed us to search for blazar - like sources lying within the positional uncertainty regions of the ugss @xcite . additional studies have been also carried out with near - infrared observations @xcite as well as in the sub millimeter range ( e.g. , * ? ? ? * ; * ? ? ? optical spectroscopic campaigns also have been crucial to disentangle and/or confirm the natures of the low - energy counterparts selected with different methods ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the main aim of the analysis presented here is to confirm the current status of the associations for the 1fgl and the 2fgl @xmath0-ray sources . thus we prepare a single list , created by merging the unique sources in the 1fgl and in the 2fgl catalogs . on the basis of the information reported in the first _ fermi_-lat catalog of sources above 10 gev ( 1fhl * ? ? ? * ) and in both the first and the second catalog of active galactic nuclei detected by _ ( 1lac and 2lac , respectively * ? ? ? * ; * ? ? ? * ) , together with an extensive literature search in multifrequency archives , we verify the current status of the @xmath0-ray associations . in addition , we also present new optical spectroscopic observations of eight @xmath0-ray blazar candidates selected during our multifrequency study . the result is a refined and merged list of all the associations for the 1fgl - plus-2fgl catalog with updated information on the low - energy counterparts including specific multifrequency notes . the paper is organized as follows : in section [ sec : cat ] we discuss the _ fermi _ procedures to assign identifications and associations with potential counterparts at low energies , and the categories of @xmath0-ray source associations . section [ sec : original ] describes the content and the main properties of both the 1fgl and the 2fgl catalogs . in section [ sec : assoc ] we introduce an updated @xmath0-ray source classification for the counterparts of the _ fermi _ detected objects . details on the correlation with multifrequency databases and catalogs are in section [ sec : crossmatches ] . in section [ sec : refined ] we present the refined version of the associations listed in the _ fermi _ catalogs ( hereinafter designated 1fglr and 2fglr , respectively , for brevity ) . the analysis of the infrared colors for the sources associated in the 1fgl is given in section [ sec : kde ] and new optical spectroscopic observations of @xmath0-ray blazar candidates are presented in section [ sec : optical ] . in section [ sec : statistical ] we compare our results with those achieved from statistical analyses performed on the ugss listed in both the 1fgl and the 2fgl . in section [ sec : connections ] we speculate on both the radio-@xmath0-ray and the ir-@xmath0-ray connections for the _ fermi _ blazars to check the consistency of sources classified as blazar candidates with these multifrequency behaviors . the summary and conclusions are given in section [ sec : conclusions ] . we use cgs units unless stated otherwise . spectral indices , @xmath2 , are defined by flux density , s@xmath3 and _ wise _ magnitudes at [ 3.4 ] , [ 4.6 ] , [ 12 ] , [ 22 ] @xmath4 m ( i.e. , the nominal bands ) are in the vega system . for numerical results a flat cosmology was assumed with @xmath5 km s@xmath6 mpc@xmath6 , @xmath7 and @xmath8 @xcite . * the astronomical survey and source class acronyms used in the paper are listed in table [ tab : acronyms]. 1fgl & _ fermi_-large area telescope first source catalog + 1fglr & 1fgl refined association catalog + 1fhl & first _ fermi_-large area telescope catalog of sources above 10 gev + 1lac & first catalog of active galactic nuclei detected by _ fermi _ + 1xsps & deep swift x - ray telescope point source catalog + 2fgl & _ fermi_-large area telescope 2-year source catalog + 2fglr & 2fgl refined association catalog + 2lac & second catalog of active galactic nuclei detected by _ + 2mass & two micron all - sky survey + 3c & third cambridge catalogue of radio sources + 6dfgs & six - degree - field galaxy redshift survey + agn & active galactic nucleus + agu & agn of uncertain type + at20 g & australia telescope 20 ghz survey + atnf & australia telescope national facility + bzb & bl lac object + bzq & flat spectrum radio quasar + crates & combined radio all - sky targeted eight - ghz survey radio catalog + csc & chandra source catalog + ct & classification tree method + first & vla faint images of the radio sky at twenty - centimeters + gb6 & green bank 6-cm radio source catalog + glc & globular cluster + hmb & high mass x - ray binary + kde & kernel density estimation + iraf & image reduction and analysis facility + iras & infrared astronomical satellite + lat & large area telescope + lba & australian long baseline array + lcs1 & lba calibrator survey + lorcat & low - frequency radio catalog of flat - spectrum sources + lr & logistic regression method + mgps & molonglo galactic plane survey + msp & millisecond pulsar + ned & nasa / ipac extragalactic database + nov & nova + nrao & national radio astronomy observatory + nvss & nrao vla sky survey catalog + oagh & observatorio astrofisico guillermo haro + oan & observatorio astronmico nacional + pks & parkes southern radio source catalog + pmn & parkes - mit - nrao surveys + psr & pulsar + pwn & pulsar wind nebula + rbsc & rosat bright source catalog + rfsc & rosat faint source catalog + sdss & sloan digitized sky survey + sed & spectral energy distribution + sfr & star forming - hii regions + snr & supernova remnant + sumss & sydney university molonglo sky survey + texas & texas survey of radio sources at 365 mhz + ugs & unidentified gamma - ray source + vla & very large array + vlss & vla low - frequency sky survey discrete source catalog + usno & united states naval observatory + xmmsl & _ xmm - newton _ slew survey + wenss & westerbork northern sky survey + wise & wide - field infrared survey explorer + wish & westerbork in the southern hemisphere source catalog + wsrt & westerbork synthesis radio telescope + wsrtgp & wsrt galactic plane compact 327-mhz source catalog + [ tab : acronyms ] in all the _ fermi _ catalogs there is an important distinction between _ identification _ of low - energy counterparts for the _ fermi _ sources and _ association_. identification is based on i ) spin or orbital periodicity ( e.g. , pulsars , binary systems ) or on ii ) correlated variability at other wavelengths ( e.g. , blazars , active galaxies ) or on iii ) the consistency between the measured angular sizes in @xmath0-ray and at lower energies ( e.g. , supernova remnants ) . * on the other hand , the _ association _ designation indeed depends on the results of different procedures adopted in both _ fermi _ catalogs these procedures are : 1 . _ the bayesian association method _ : initially applied to associate egret sources with flat - spectrum radio sources ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , this method assesses the probability of association between a @xmath0-ray source and a candidate counterpart taking into account their local densities . local density is estimated simply by counting candidates in a nearby region of the sky . 2 . _ the likelihood ratio method _ : used to search for possible counterparts in uniform surveys in the radio and in x - ray bands , this procedure was originally proposed by richter ( 1975 ) and subsequently applied by and modified by de ruiter , willis & arp ( 1977 ) , prestage & peacock ( 1983 ) , wolstencroft et al . ( 1986 ) and by sutherland & saunders ( 1992 ) . the logn - logs association method _ : this is a modified version of the bayesian method for blazars taking into account their @xmath9-@xmath10 ( see * ? ? ? * ; * ? ? ? * for more details ) . there are several differences between the source associations presented between the 1fgl and the 2fgl catalogs mostly related to the @xmath0-ray analysis and to the improvements achieved in the models of galactic @xmath0-ray diffuse emission as well as in evaluation of the lat response functions . improving the _ fermi _ source localization not only makes easier the search for low - energy counterparts but allows more accurate estimates of the association probability since they depend on the @xmath0-ray positional uncertainty ( see e.g. , * ? ? ? * ; * ? ? ? these differences also relate to updates of comparison surveys and catalogs used to search for the low - energy counterparts and on multifrequency observations performed on ugs samples . the use of catalogs / surveys that were previously not considered for the _ fermi _ catalog preparation , as for example the infrared all - sky survey performed by _ wise _ and the low - frequency radio catalogs , revealing new connections between the low and the high - energy skies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) has also decreased the unknown fraction of the @xmath0-ray sky ( e.g. , * ? ? ? * ; * ? ? ? * ) . this affects the _ fermi _ associations because the methods adopted to assign low - energy counterparts depend on the source densities of the catalogs @xcite . a better estimate of the counterpart density leads to a better estimate of the false positive associations and of the association probability and this implies that a previously unassociated source could have an assigned counterpart in a new release of the _ fermi _ catalog . motivated by the changes occurring between the 1fgl and the 2fgl associations here we propose to introduce , together with identified and associated sources , the category of _ `` candidate associations''_. these are @xmath0-ray sources that have a potential low - energy counterpart in a specific class of well - known @xmath0-ray emitters lying within the _ fermi _ positional uncertainty region at 95% level of confidence and/or having angular separations between the _ fermi _ and the counterpart position smaller than the maximum one for all the associated sources of the same class . candidate associations have the potential to be promoted to associations in the future ( see section [ sec : assoc ] ) . we considered as counterpart classes for the candidate associations only a few kinds of sources : i ) blazar - like sources , as defined in the following , ii ) known pulsars ( psrs ) , iii ) pulsar wind nebulae ( pwne ) , supernova remnants ( snrs ) and star forming and hii regions ( hereinafter simply indicated as sfrs ) . the first two source classes constitute the two largest populations of known @xmath0-ray emitters while the other choices were considered because 15 out of the 44 snrs and 10 out of the 63 psrs and pwne associated in the 1fgl lie within sfrs and a similar situation occurs for the 2fgl . our choice concerning the sfrs , especially for the most massive cases , is indeed motivated by the possibility that they could host type o and/or b stars , potentially associated with open star clusters , and this being relatively likely to have some members that have progressed to become a psr , pwn , and/or snr , all known classes of @xmath0-ray sources ( see also * ? ? ? * ; * ? ? ? in addition , considering sfrs as candidate associations is also motivated by the recent idea , partially supported by the _ fermi _ discovery of @xmath0-ray emission arising from eta carinae @xcite , that shells detected in the radio , infrared , and x - rays could mark termination shocks of stellar winds . these winds interacting with the stars natal molecular clouds can be considered as plausible sites to accelerate particles that emit @xmath0-rays ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? bowshocks are associated with many source classes such as psrs , cataclysmic variables , colliding wind binaries , cometary h ii regions and particles accelerated therein can radiate up to the mev - gev energy range ( see e.g. * ? ? ? . however we remark that evidence of @xmath0-ray emission from colliding wind binaries associated with wolf - rayet stars was not found by _ , for a selected sample of 7 sources @xcite ; although , as highlighted by the same authors , searching for such @xmath0-ray detections is extremely complicated due to the uncertainties of the diffuse emission in the galactic plane . some candidate associations lack multifrequency information as well as correct spectroscopic identifications but * exhibit some characteristic * features , typical of known @xmath0-ray emitters , mainly blazars and psrs . they were not associated in the previous releases of the _ fermi _ catalogs mostly because they were not listed in one of the appropriate comparison catalogs of potential counterparts used to associated _ fermi _ sources . however once follow up observations * were * performed ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * for blazar - like candidates ) , then they could be promoted to associated sources in new releases of the _ fermi _ catalogs . in addition , within this category we also considered _ fermi _ sources having sfrs within their @xmath0-ray positional uncertainty regions for which multifrequency observations could reveal the presence of unknown psrs , pwne and snrs , already counterparts of @xmath15% of the @xmath0-ray sources . we emphasize three advantages obtained from introducing the concept of candidate associations : 1 . designating candidate association category allows us to flag potential counterparts that might be associated or identified in future releases of the _ fermi _ catalogs , as previously occurred to 171 1fgl sources then recognized , classified and associated in the 2fgl . this also facilitates planning follow up multifrequency observations , in particular spectroscopic ones , necessary to determine and/or confirm their nature . the category of candidate associations permits us to establish which is the number of _ fermi _ sources not yet having potential counterparts . the precise knowledge of this number is extremely relevant to set better constraints on dark matter scenarios ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? 3 . for all the methods adopted to associate sources in the _ fermi _ catalogs , the resulting associations depend on the comparison catalog / survey used . the same low - energy counterpart may be associated with a _ fermi _ source using a certain catalog but not with a different survey , even when the positions in the two catalogs are the same . this occurs because the source density of the comparison catalogs , which is used to determine the probability of false positive associations , affects the association probability . thus observing these candidate associations with follow up campaigns to determine their natures will permit us to add blazars and pulsars to the correct comparison catalogs and associate them in future releases of the _ fermi _ surveys . to reiterate the importance of having candidate associations we highlight the recent work of schinzel et al . ( 2014 ) on radio very - long - baseline interferometry ( vlbi ) observations of _ fermi _ unassociated sources in the 2fgl ( see also * ? ? ? * ) . in the final list of high - confidence associations presented in schinzel et al . ( 2014 ) more than 90% of the sources are also detected in the major radio surveys such * as the national radio astronomy observatory ( nrao ) very large array ( vla ) sky survey catalog ( nvss ; * ? ? ? * ) and the sydney university molonglo sky survey ( sumss ; * ? ? ? * ) . * these counterparts were not associated using the methods adopted in _ fermi _ catalogs due to the high density of sources at low flux level in the above surveys . however thanks to the high resolution radio follow up observations , which reveled the presence of radio compact cores , it has been possible to use a different selected sample of potential counterparts to find new associations via the likelihood procedure @xcite . thus it is extremely important to have targets for follow up observations that , once performed , will allow us to insert the potential counterparts in the correct catalogs used to search @xmath0-ray associations . identifications , associations and candidate associations of @xmath0-ray sources are labelled in both the refined version of the _ fermi _ catalogs as the _ category of the @xmath0-ray source association_. these are defined as : _ identified _ ( i ) , _ associated _ ( a ) , _ candidate association _ ( c ) in the tables while sources for which no potential counterpart was found are assigned to the ugs category ( u ) . the 1fgl catalog[multiblock footnote omitted ] contains 1451 sources detected and characterized in the 100 mev to 100 gev energy range . for each _ fermi _ source listed therein , positional uncertainty regions at 68% and 95% level of confidence , results from spectral fits as well as flux measurements in five energy bands are reported @xcite . firm identification or plausible association with a low - energy counterpart was provided for each 1fgl source on the basis of the comparison with other astronomical catalogs and on the basis of the association procedures previously mentioned ( see section [ sec : cat ] and abdo et al . 2010a for details ) . the published 1fgl catalog lists several classes of counterparts : 63 pulsars ( psrs ) , 7 associated plus 56 identified ( 24 of them discovered thanks to the _ fermi_-lat observations ) , 2 pulsars wind nebulae ( pwne ) 44 @xmath0-ray sources corresponding to 38 snrs , 41 associated and 3 identified , 8 globular clusters ( glc ) , 2 high mass x - ray binaries ( hmb ) , all identified , 295 bl lac objects ( bzbs ) 278 flat spectrum radio quasars ( bzqs ) , 28 non blazar - like active galactic nuclei ( agns ) 92 agn of uncertain type ( agus ) 6 normal galaxies and 2 starburst galaxies . the remaining 630 _ fermi _ sources are all ugss . a detailed comparison between the 1fgl and the 2fgl catalog is reported in nolan et al . there are 1099 sources included in both _ catalogs with only 381 1fgl sources being listed in the 1fhl @xcite . it is worth noting that 267 out of 1099 sources in common between 1fgl and 2fgl changed classification from 1fgl to 2fgl and among them 172 , previously unassociated in 1fgl , obtained an assigned low - energy counterpart in 2fgl . this sample of modified associations includes : 16 sources indicated as generic agns that have been classified plus one that became a ugs ; 54 agus that have been recognized as blazars ( either bzbs or bzqs ) and 2 as agns ; then 17 blazars actually changed their classification from 1fgl to 2fgl thanks to new spectroscopic data available and , among them , 6 actually disappeared since they were listed in the 2fgl as unassociated as occurred for other 5 sources of galactic origin ( i.e. , psrs , pwne and snrs ) . on the other hand , there are 172 ugss of the 1fgl that appear to be associated in the 2fgl : 20 associated with psrs , 11 with snrs and 1 with a glc while all the others with various agn classes ( i.e. blazars , radio galaxies , seyfert galaxies etc . ) . the 1fgl catalog includes 216 sources with @xmath0-ray analysis flags and their detection has to be considered carefully since they could have been artifacts of the analysis @xcite . in particular , 103 of these 216 1fgl flagged sources have the `` c '' flag following the 1fgl name ; this indicates that the source is found in a region with bright and/or possibly incorrectly modeled diffuse emission . on the other hand any non - zero entry in the flags column indicates inconsistencies found during the analysis ( see * ? ? ? * for details ) . however we note that 36 out of 103 `` c''-flagged 1fgl sources are also detected in the 2fgl and only 25 of them were still indicated with a `` c '' flag therein . thus we decided to keep all the c - flagged sources in our analysis since they could be useful for reference in future releases of _ fermi _ catalogs . the 2fgl catalog[multiblock footnote omitted ] based on 24 months of _ fermi _ operation lists 1873 @xmath0-ray sources @xcite . it includes : * 108 psrs , 25 associated plus 83 identified ; * 3 pwne , all 3 identified ; * 61 @xmath0-ray sources corresponding to 50 snrs , 55 associated plus 6 identified ; * 11 associated globular clusters ( glc ) , 4 high mass x - ray binaries ( hmb ) , all identified and 1 nova ( nov ) , among the galactic sources , while in the extragalactic sky : * 435 bl lac objects ( bzbs ) , only 7 identified ; * 370 flat spectrum radio quasars ( bzqs ) , 17 identified ; * 11 non blazar - like active galactic nuclei ( agns ) , 1 identified ; * 257 agns of uncertain type ( agus ) ; * 6 normal galaxies and 4 starburst galaxies . the remaining 577 _ fermi _ sources were are all ugss . the 2fgl catalog * lists * 472 sources with @xmath0-ray analysis flags and 155 of them also have the `` c '' flag following the 2fgl name . it is also worth noting that with respect to the 1fgl catalog the fraction of ugss is decreased from 630 out of 1451 sources ( @xmath143% ) to 577 out of 1873 ( @xmath131% ) in the 2fgl . we adopt the following class definitions for galactic counterparts of the _ fermi _ sources , summarized in table [ tab : classes ] : high mass x - ray binary ( hmb ) , globular cluster ( glc ) , nova ( nov ) , millisecond pulsar ( msp ) , pulsar ( psr ) , pulsar wind nebula ( pwn ) , supernova remnant ( snr ) , star forming region or hii region ( sfr ) and binary star ( bin ) . the classes that we assign for extragalactic counterparts are bl lac object ( bzb ) , quasar radio loud of flat spectrum type ( bzq ) , blazar of uncertain type ( bzu ) , radio galaxy ( rdg ) , normal galaxy ( gal ) , starburst galaxy ( sbg ) and seyfert galaxy ( sey ) . * we note that we used the three - letter designators to indicate the class for each counterpart ; these are shown in lower case to differentiate them from the acronyms used throughout the paper ( see table [ tab : acronyms ] ) . * we emphasize that the labels used for the extragalactic blazar subclasses are used only if the counterpart corresponds to a source classified in the roma - bzcat @xcite . * we used version 4.1 of the roma - bzcat in our analysis . * we also emphasize that the proposed classification depends on the availability of the optical spectroscopic information for the low - energy counterparts . we also considered as a counterpart class blazar candidates ( bcn ) to indicate sources having at least one of the following requirements : 1 ) being classified as bl lac candidate in the roma - bzcat ; 2 ) sources that belong to the combined radio all - sky targeted eight - ghz survey radio catalog ( crates ; * ? ? ? * ) with a bl lac - like or a quasar - like * optical * spectrum available in literature ; 3 ) radio sources with a blazar - like optical spectrum available in literature ( i.e. , a published spectrum or a description of the features found in the near infrared optical band ) and with a _ wise _ counterpart . finally , we labelled as unclassified counterparts ( unc ) those sources satisfying at least one of the following criteria : 1 ) _ wise _ infrared colors similar to those of @xmath0-ray blazars selected according to the association methods developed by dabrusco et al . ( 2013 ) and/or with the kernel density estimator analysis ( * ? ? ? * ; * ? ? ? * see also section [ sec : kde ] for a refined analysis ) ; 2 ) a flat radio spectrum at frequencies below @xmath11 ghz ( i.e. , @xmath110.5 ) ; 3 ) a radio source with or without an x - ray counterpart associated with the methods adopted in the 1lac and the 2lac ; 4 ) a radio and/or an x - ray source that lies at angular separation between the _ fermi _ and the counterpart positions smaller than the maximum one for the blazar class ( i.e. , 0.4525 corresponding to 2fgl j1746.0 + 2316 ) . all the remaining sources are still unidentified ( ugs ) . the proposed classification scheme has three main advantages . 1 . to have a homogeneous classification of the blazars listed in our analysis and to avoid misclassifications we adopted the following criterion . we indicate as blazars only sources that were previously recognized and classified in the roma - bzcat . there are indeed several counterparts classified as bzb or bzq listed in the 1lac and in the 2lac catalogs that do not have the same classification according to the roma - bzcat or for which no optical spectra were found in literature to confirm their classification . 2 . we do not classify a source with a radio and/or x - ray counterpart as an active galaxy of uncertain type ( agu ) since these requirements are not strictly sufficient to claim a source as an agn or as blazar - like and the lack of multifrequency information does not allow us to determine precisely its nature . consequently , we introduced the bcn and the unc classes . once optical spectra and multifrequency observations become available a refined classification will be provided . we introduced the sfr class for cases where _ fermi _ position is consistent with a known massive sfr for three reasons . first , to highlight the possibility that an unknown snr , pwn and/or psr is embedded therein , for which multifrequency observations are necessary to confirm their presence . second , because the use of this information could be adopted to refine @xmath0-ray diffuse models in future releases of the _ fermi _ catalogs . in particular , the largest fraction of these sfr potential associations were determined using the recent _ wise _ catalog of galactic hii regions[multiblock footnote omitted ] @xcite , which was not available during the preparation of the currently - published _ fermi _ catalogs . this could also reveal new correspondences between the uncertainties of the diffuse @xmath0-ray background models and regions of interstellar medium in the galactic plane . third , this could be useful for future investigations of the environments of the galactic @xmath0-ray sources . finally , we remark that for the sources indicated as candidate associations and classified as sfrs , snrs , pwns and psrs we investigated the radio and infrared images available in the nvss , the westerbork synthesis radio telescope ( wsrt ) galactic plane compact 327-mhz source catalog ( wsrtgp ; * ? ? ? * ) , the molonglo galactic plane survey[multiblock footnote omitted ] ( mgps ; * ? ? ? * ) catalogs and in the _ wise _ archives searching for signatures of potential @xmath0-ray emitters . in figures [ fig:1fglj0541.1 + 3542c ] and [ fig:1fglj1729.1 - 3641c ] we show two cases of @xmath0-ray sources having extended structures associated with an open star cluster in a sfr found within their _ fermi _ positional uncertainty regions . these are two examples of star clusters where some members could have been progressed becoming a psr , pwn , and/or a snr , and so being a plausible counterpart for the @xmath0-ray emission . then in figures [ fig:1fglj1726.2 - 3521c ] and [ fig:1fglj1925.0 + 1720c ] we report the best two cases of sfrs for which a signature of a shell - like or ring - like extended structures was found within the _ fermi _ positional uncertainty region that could be ascribed as a shock signature potentially responsible for the @xmath0-ray emission . m ( top ) , 4.6@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the red contours on the _ wise _ image at 3.4@xmath4 m indicate the radio brightness at 1.4 ghz from the nvss . the radio - infrared extended structure , marked by the green arrow , suggests that a shock occurred in the interstellar medium that could be responsible for the @xmath0-ray emission . the radio source is also detected in the westerbork northern sky survey ( wenss ; * ? ? ? * ) and it is associated with an open star cluster.,title="fig:",width=317,height=272 ] m ( top ) , 4.6@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the red contours on the _ wise _ image at 3.4@xmath4 m indicate the radio brightness at 1.4 ghz from the nvss . the radio - infrared extended structure , marked by the green arrow , suggests that a shock occurred in the interstellar medium that could be responsible for the @xmath0-ray emission . the radio source is also detected in the westerbork northern sky survey ( wenss ; * ? ? ? * ) and it is associated with an open star cluster.,title="fig:",width=317,height=272 ] 3641c in two different bands at 3.4@xmath4 m ( top ) , 4.6@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white radio contours drawn from the nvss image are overlaid on the _ wise _ image at 3.4@xmath4 m . this is an example of a candidate association with a sfr that includes the open star cluster 165 from the northern and equatorial milky way catalog built with the two micron all - sky survey ( 2mass ) observations @xcite.,title="fig:",width=317,height=272 ] 3641c in two different bands at 3.4@xmath4 m ( top ) , 4.6@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white radio contours drawn from the nvss image are overlaid on the _ wise _ image at 3.4@xmath4 m . this is an example of a candidate association with a sfr that includes the open star cluster 165 from the northern and equatorial milky way catalog built with the two micron all - sky survey ( 2mass ) observations @xcite.,title="fig:",width=317,height=272 ] m ( top ) , 12@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white contours are overlaid on the _ wise _ image at 12@xmath4 m are the 1.4 ghz radio emission from the nvss . the radio - infrared extended structure , marked by the green arrow , clearly indicates that a shock occurred in the interstellar medium within the _ fermi _ positional uncertainty region . this could be responsible for the @xmath0-ray emission.,title="fig:",width=317,height=272 ] m ( top ) , 12@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white contours are overlaid on the _ wise _ image at 12@xmath4 m are the 1.4 ghz radio emission from the nvss . the radio - infrared extended structure , marked by the green arrow , clearly indicates that a shock occurred in the interstellar medium within the _ fermi _ positional uncertainty region . this could be responsible for the @xmath0-ray emission.,title="fig:",width=317,height=272 ] m ( top ) , 12@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white contours are overlaid on the _ wise _ image at 12@xmath4 m are the 1.4 ghz radio emission from the nvss . the ring - shaped structure , indicated by the green arrow , is extended for about 87 and resembles an unknown snr or a pwn . it clearly lies within the _ fermi _ positional uncertainty region and is also emitting in the radio band . this is a clear example of a candidate association with a sfr including a potential snr / pwn for which additional multifrequency observations are necessary to confirm its origin.,title="fig:",width=317,height=272 ] m ( top ) , 12@xmath4 m ( bottom ) . in each image the yellow ellipse corresponds to the _ fermi _ positional uncertainty region at 68% level of confidence . the white contours are overlaid on the _ wise _ image at 12@xmath4 m are the 1.4 ghz radio emission from the nvss . the ring - shaped structure , indicated by the green arrow , is extended for about 87 and resembles an unknown snr or a pwn . it clearly lies within the _ fermi _ positional uncertainty region and is also emitting in the radio band . this is a clear example of a candidate association with a sfr including a potential snr / pwn for which additional multifrequency observations are necessary to confirm its origin.,title="fig:",width=317,height=272 ] galactic sources & + high mass x - ray binary & hmb + globural cluster & glc + nova & nov + millisecond pulsar ( msp ) & msp + pulsar ( psr ) & psr + pulsar wind nebula ( pwn ) & pwn + binary star & bin + star forming - hii region ( sfr ) & sfr + _ ( for potential associations with unknown snrs and/or pwne and/or psrs ) _ & + supernova remnant ( snr ) & snr + extragalactic sources & + bl lac object ( bzb ) & bzb + quasar radio loud with flat radio spectrum ( bzq ) & bzq + blazar of uncertain type ( bzu ) & bzu + normal galaxy & gal + radio galaxy & rdg + starburst galaxy & sbg + seyfert galaxy & sey + uncertain classification & + blazar candidate & bcn + unclassified sources & unc + unidentified gamma - ray source & ugs + [ tab : classes ] we searched the following major radio , infrared , optical and x - ray surveys and both the nasa extragalactic database ( ned)[multiblock footnote omitted ] and the simbad astronomical database[multiblock footnote omitted ] to verify whether multifrequency information can confirm the natures of uncertain counterparts and blazar candidates . * in the following we list all the catalogs and the surveys searched for our investigation ; the abbreviations correspond to the designators used for the multifrequency notes in table [ tab : main ] . * below @xmath11 ghz radio frequency we searched the vla low - frequency sky survey discrete source catalog ( vlss ; * ? ? ? * - v ) and the recent revision vllsr[multiblock footnote omitted ] @xcite , both the westerbork northern sky survey ( wenss ; * ? ? ? * - w ) and the westerbork in the southern hemisphere source catalog ( wish ; * ? ? ? * - w ) , the sydney university molonglo sky survey ( sumss ; * ? ? ? * - s ) , the parkes - mit - nrao surveys ( pmn ; * ? ? ? * - pm ) , the texas survey of radio sources at 365 mhz ( texas ; * ? ? ? * - t ) the parkes southern radio source catalog ( pks ; * ? ? ? * - pk ) , and the low - frequency radio catalog of flat - spectrum sources ( lorcat ; * ? ? ? * - l ) . for the radio counterparts ( i.e. , above @xmath11 ghz ) , we searched the nrao vla sky survey ( nvss ; * ? ? ? * - n ) , the vla faint images of the radio sky at twenty - centimeters ( first ; * ? ? ? * ; * ? ? ? * - f ) , * the 87 green bank catalog of radio sources ( 87 gb ; * ? ? ? * - 87 ) and the green bank 6-cm radio source catalog ( gb6 ; * ? ? ? * - gb ) , * the australia telescope 20 ghz survey ( at20 g ; * ? ? ? * - a ) , the combined radio all - sky targeted eight - ghz survey ( crates ; * ? ? ? * - c ) and the australian long baseline array ( lba ) calibrator survey of southern compact extragalactic radio sources[multiblock footnote omitted ] ( lcs1 * ? ? ? * - lcs ) . for the infrared , we queried the _ wise _ all - sky survey in the allwise source catalog[multiblock footnote omitted ] ( * ? ? ? * - w ) and the two micron all sky survey ( 2mass ; * ? ? ? * - m ) since each _ wise _ source is automatically matched to the closest 2mass potential counterpart ( see * ? ? ? * for details ) . then , we also searched for optical counterparts , with or without spectra available , in the sloan digitized sky survey data release 9 ( sdss dr9 ; e.g. * ? ? ? * - s ) , in the six - degree - field galaxy redshift survey ( 6dfgs ; * ? ? ? * ; * ? ? ? we also queried the united states naval observatory ( usno)-b1 catalog ( * ? ? ? * - u ) for the optical counterparts of the bcn and the unc sources and reported their magnitudes since they could be useful to plan future follow up observations . in x - rays , we searched the rosat all - sky survey in both the rosat bright source catalog ( rbsc ; * ? ? ? * - x ) and the rosat faint source catalog ( rfsc ; * ? ? ? * - x ) . as well as _ xmm - newton _ slew survey ( xmmsl ; * ? ? ? * ; * ? ? ? * - x ) , the deep swift x - ray telescope point source catalog ( 1xsps ; * ? ? ? * - x ) the chandra source catalog ( csc ; * ? ? ? * - x ) and _ swift _ x - ray survey for all the _ fermi _ ugss @xcite . * we used the same designator for the x - ray catalogs of _ xmm - newton _ , _ chandra _ and _ swift_. these x - ray observatories perform only pointed observations thus there is the chance that the pointed observations present in their archives and related to the field of the _ fermi _ source was not requested as follow up of the @xmath0-ray source but for a different reason . consequently , the discovery of the x - ray counterpart for an associated and/or identified source could be serendipitous . * to perform the searches over all the surveys and the catalogs cited above we considered the 1@xmath12 positional uncertainties reported therein , with the only exceptions being the _ wise _ all - sky survey and the sdss dr9 . in these two cases we searched the closest ir and optical counterpart within a maximum angular separation of 3.3 and of 1.8 for the allwise survey and the sdss dr9 , respectively . these values have been derived on the basis of the statistical analysis described in dabrusco et al . ( 2013 ) and massaro et al . ( 2014b ) developed following the approach also presented in maselli et al . ( 2010 ) and stephen et al . ( 2010 ) . to verify the presence of sfrs consistent with the positional uncertainty of the _ fermi _ sources , we searched in the following surveys and catalogs : 1 ) the sharpless catalog of hii regions @xcite , 2 ) the hii regions in the ( infrared astronomical satellite ( iras ) point source catalog @xcite , 3 ) the radio survey of hii regions at 4.85 ghz @xcite , 4 ) the new catalog of hii regions in the milky way @xcite , 5 ) the radio catalog of galactic hii regions from decimeter to millimeter wavelength @xcite 6 ) the scuba imaging survey of ultra compact hii regions @xcite ; 7 ) the recent _ wise _ catalog of galactic hii regions @xcite 8) the quad galactic plane survey @xcite 9 ) the 5 ghz vla survey of the galacitc plane @xcite and 10 ) the wsrtgp @xcite . we remark that for the cross - matching performed between the sfr catalogs and the _ fermi _ sources we took into account the size of each hii region in combination with the _ fermi _ uncertainty position . to check for the presence of psrs we used the australia telescope national facility ( atnf ) pulsar catalog @xcite , the catalogue of galactic supernova remnants @xcite , and the census of high - energy observations of galactic snrs[multiblock footnote omitted ] @xcite . table [ tab : main ] summarizes all the notes derived from the multifrequency analysis . in particular , for the psrs , we report the information shown in the public list of lat - detected gamma - ray pulsars[multiblock footnote omitted ] and by the pulsar search consortium ( psc ) as well as from literature ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? symbols used for their multifrequency notes are labelled as reported in table [ tab : psr ] ( see * ? ? ? * for more details ) . b & psr in a binary system + e & psr detected in @xmath0 rays by the compton gamma - ray observatory + g & psr discovered in lat @xmath0-ray data + m & msp + p & pulsar discovered by the pulsar search consortium ( psc ) + r & psr discovered in the radio + u & psr discovered using a _ fermi_-lat seed position + x & psr discovered in x - rays + [ tab : psr ] the refined associations for the 1fgl and the 2fgl catalog ( hereinafter 1fglr and 2fglr , respectively ) is presented in table [ tab : main ] . for each source we report : the 1fgl name together with the 2fgl and the 1fhl ones as derived from the associations published in those catalogs , the counterpart name , and in 37 cases also an alternative association ; the category of @xmath0-ray source association ; the class of each counterpart as found in the literature ; notes derived from the multifrequency investigation ; we also report whether the counterpart lies inside a sfr , and specifically for the psrs in a pwn or a snr . in addition , in table [ tab : main ] we also report if a pwn or a snr contains a known psr . the coordinates of each counterpart are also included in table [ tab : main ] together with the optical magnitudes in b and r band from the usno - b1 catalog @xcite . within the multifrequency notes in table [ tab : main ] we also indicate if the spectral energy distribution ( sed ) of the source is shown in takeuchi et al . ( 2013 ) , if a redshift estimate is present in literature ( indicating the reference ) and if the radio counterpart has a flat radio spectrum ( i.e. , designated rf in the notes of table [ tab : main ] whenever @xmath110.5 in the radio band ) . in the on - line version of the table we also report the infrared analysis flag @xcite for those sources , classified as blazar - like with a counterpart in the allwise catalog within 3.3 as well as the column with the confirmed redshifts . for the counterpart names we used the following priorities : if the source is a known blazar we adopted the roma - bzcat nomenclature while for the known pulsars that of the atnf pulsar catalog[multiblock footnote omitted ] @xcite . radio galaxies , seyfert galaxies and starburst names were indicated from the third cambridge catalogue of radio sources and revised versions ( 3c and 3cr * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , the new general catalog and the messier catalog or using their radio names reported in one of the major surveys ( see section [ sec : crossmatches ] ) . then galactic sources as snrs and pwne have their names in galactic coordinates as reported in the catalogue of galactic supernova remnants[multiblock footnote omitted ] @xcite , while the most common names were used for globular clusters and binaries . the counterparts that were associated with sfrs were also labelled with their name in galactic coordinates . finally , the remaining uncertain sources were identified by their radio names whenever possible or by their rosat name one . a handful of counterparts were indicated using the nomenclature of optical catalogs . adopting these choices in the names we remark that all the counterparts can be retrieved from the nasa / ipac extragalactic database ( ned ) and simbad archives without using their coordinates . column description . ( 1 ) : 1fgl name ; ( 2 ) : 2fgl name ; ( 3 ) : 1fhl name ; ( 4 ) : @xmath0-ray source category ( i = identified ; a = associated ; c = candidate ; u = unidentified ; see section [ sec : cat ] for details ) ; ( 5 ) : name of the first low - energy counterpart ; ( 6 ) : multifrequency notes on the first low - energy counterpart ; ( 7 ) : redshift estimate if the source is extragalactic ( in the on line version of the catalog there is also a note if the @xmath13 measurement is uncertain ) ; ( 8) : notes if the first low - energy counterpart lies inside a sfr / snr / pwn ; ( 9 ) : notes if the first low - energy counterpart includes a psr / msp ; ( 10 ) : source class for the first low - energy counterpart ( see section [ sec : assoc ] for details ) ; ( 11 ) : r.a . ( j2000 ) for the first low - energy counterpart ; ( 12 ) : dec . ( j2000 ) for the first low - energy counterpart ; ( 13 ) : b magnitude for the first low - energy counterpart taken from the usno - b1 catalog @xcite , only for the bcn and the unc sources ; ( 14 ) : r magnitude for the first low - energy counterpart taken from the usno - b1 catalog @xcite , only for the bcn and the unc sources ; ( 15 ) : name of the second low - energy counterpart ; ( 16 ) : multifrequency notes on the second low - energy counterpart ; ( 17 ) : source class for the second low - energy counterpart ( see section [ sec : assoc ] for details ) ; ( 18 ) : r.a . ( j2000 ) for the second low - energy counterpart ; ( 19 ) : dec . ( j2000 ) for the second low - energy counterpart ; ( 20 ) : b magnitude for the second low - energy counterpart taken from the usno - b1 catalog @xcite , only for the bcn and the unc sources ; ( 21 ) : r magnitude for the second low - energy counterpart taken from the usno - b1 catalog @xcite , only for the bcn and the unc sources ; ( 22 ) : probability that the first low - energy counterpart has infrared colors similar to the known 1fgl bzbs estimated with the kde method ( see section [ sec : kde ] for details ) ; ( 23 ) : probability that the first low - energy counterpart has infrared colors similar to the known 1fgl bzqs estimated with the kde method ( see section [ sec : kde ] for details ) . symbols used for the multifrequency notes are all reported in section [ sec : crossmatches ] together with the references of the catalogs / surveys . references for the optical spectra reported in the table : ackermann et al . ( 2011 ) , ahn et al . ( 2012 ) , baker et al . ( 1995 ) , bauer et al . ( 2000 ) , bade et al . ( 1995 ) , beuermann et al . ( 1999 ) , bikmaev et al . ( 2008 ) , britzen et al . ( 2007 ) , drinkwater et al . ( 1997 ) , healey et al . ( 2008 ) , hewett & wild ( 2010 ) , hewitt & burbidge ( 1989 ) , johnston et al . ( 1995 ) , jones et al . ( 2004 ) , jones et al . ( 2009 ) , landoni et al . ( 2014 ) landt et al . ( 2001 ) , lister et al . ( 2011 ) , mao ( 2011 ) , maza et al . ( 1995 ) , marti et al . ( 2004 ) , masetti et al . ( 2013 ) , massaro et al . ( 2014c ) mitton et al . ( 1977 ) , paggi et al . ( 2013 ) , quintana & ramirez ( 1995 ) , schwope et al . ( 2000 ) , shaw et al . ( 2013a ) , shaw et al . ( 2013b ) , stern & assef ( 2013 ) , stickel & kuehr ( 1996a , b ) titov et al . ( 2011 ) , vandenbroucke et al . ( 2010 ) , vettolani et al . ( 1989 ) , takeuchi et al . ( 2013 ) , thompson et al . ( 1990 ) , tsarevsky ert al . ( 2005 ) , white et al . [ tab : main ] the 1fglr lists 111 identified sources , 880 associated , 306 candidate associations , and 154 ugss . the sky distribution of the ugss is shown in figure [ fig : skydist1 ] in comparison with the 1fgl associations in which there were 66 identified objects , 755 associated , and 630 unidentified . the results of our multifrequency analysis for the 1fglr are summarized in table [ tab : results ] where we list all the sources identified , associated , and candidate associations for each class separately . we remark that the globular cluster gcl 94 ( alias m28 ) associated with 1fgl j1824.5@xmath142449 contains a known msp psr j1824@xmath142452a while 2mass@xmath14gc01 lies in an sfr . among the psrs and the msps are 21 sources that lie within a known pwn and 4 in snrs as reported in the notes of table [ tab : main ] . twenty psrs and msps out of 39 indicated as candidate associations in both _ catalogs have been identified after the release of the 2fgl catalog , as presented in the public list of lat - detected gamma - ray pulsars . this strongly supports our introduction of this new category of @xmath0-ray sources . in five of the 10 associations of pwne a psr also has been found within the nebula and the same situation occurs for 8 snrs . we find that 113 1fgl sources have an sfr within their positional uncertainty regions obtained by combining in quadrature the @xmath0-ray localization at 95% level of confidence with the size of the sfr . three of these sources are the hmbs and one is the globular cluster 2mass@xmath14gc01 . fifty - six of these 113 have been tentatively associated ( i.e. , indicated as candidate associations ) with previously ugss while the remaining _ fermi _ objects all have psrs , pwne and/or a snrs embedded in gas clouds and/or interacting with the interstellar medium . among the 54 candidate associations with sfrs , two contain snrs , one a pwn , and one a known psr . however , the potential associations with sfrs could be used to refine the @xmath0-ray galactic diffuse emission models and the associations in future releases of the _ fermi _ catalogs , as well as lead to the discovery of new pwne and/or unknown snrs . none of the msps has been found lying within a known sfr while this occurs for 21 psrs , 7 pwne , and 18 snrs . it is worth noting that 1fgl j0503.2 + 4526 , associated according to the 2fgl and the 2lac analyses with an unclassified source , has as an alternative potential counterpart a snr that includes a psr . a similar situation occurs for 1fgl j1837.5@xmath140659c and 1fgl j1846.8@xmath140233c whose @xmath0-ray emission could be ascribed to a psr in a pwn . in addition , 1fgl j0622.2 + 3751 has as an alternative association with a psr . within the unclassified candidate associations ( i.e. , unc ) , we noticed that 8 1fglr sources are positionally consistent with sfrs . in addition , we also listed in the 1fglr the tentative association of 1fgl j1653.6@xmath140158 ( also known as 2fgl j1653.6@xmath140159 ) with the binary system that includes psr j1653@xmath140158 @xcite . * finally , we remark that in the extragalactic sky the small magellanic cloud ( alias ngc 292 ) is identified with the _ fermi _ source 1fgl j0101.3@xmath147257 and classified as a normal galaxy while the large magellanic cloud is associated with all the remaining five _ fermi _ sources listed as normal galaxies . * class & i d . & & can . & tot . & i d . & ass . & can . & tot . + galactic & & & & & & & & + bin & 0 & 1 & 0 & 1 & 0 & 1 & 0 & 1 + hmb & 3 & 0 & 1 & 4 & 4 & 0 & 0 & 4 + glc & 0 & 9 & 0 & 9 & 0 & 11 & 0 & 11 + nov & 0 & 0 & 0 & 0 & 1 & 0 & 0 & 1 + msp & 26 & 16 & 8 & 50 & 26 & 17 & 8 & 51 + psr & 53 & 8 & 31 & 92 & 56 & 12 & 30 & 98 + pwn & 0 & 8 & 2 & 10 & 2 & 6 & 1 & 9 + snr & 3 & 38 & 5 & 46 & 6 & 55 & 5 & 66 + sfr & 0 & 0 & 54 & 54 & 0 & 0 & 63 & 63 + extragalactic & & & & & & & & + bzb & 7 & 231 & 1 & 239 & 7 & 270 & 0 & 277 + bzq & 15 & 271 & 1 & 287 & 18 & 320 & 0 & 338 + bzu & 2 & 46 & 0 & 48 & 2 & 52 & 0 & 54 + gal & 1 & 5 & 0 & 6 & 2 & 4 & 0 & 6 + rdg & 0 & 5 & 0 & 5 & 1 & 7 & 0 & 8 + sbg & 0 & 3 & 0 & 3 & 0 & 3 & 0 & 3 + sey & 0 & 2 & 0 & 2 & 0 & 3 & 0 & 3 + uncertain & & & & & & & & + bcn & 0 & 188 & 25 & 213 & 0 & 242 & 29 & 271 + unc & 1 & 49 & 178 & 228 & 1 & 173 & 146 & 320 + ugs & 0 & 0 & 0 & 154 & 0 & 0 & 0 & 289 + col . ( 1 ) source class . ( 2 ) id.= identified . ( 3 ) ass.= associated . ( 4 ) can.= candidate associations . [ tab : results ] in the analysis of the 1fglr we have already considered the associations for 1099 _ fermi _ sources that are in common between the 1fgl and the 2fgl , however , we report the results for the whole refined 2fgl associations ( hereinafter 2fglr ) in table [ tab : results ] . as for the 1fglr , the 2fglr catalog is dominated by blazar - like sources in the extragalactic sky and by pulsars around the galactic plane . the 2fglr lists 126 identified sources as in the original catalog , 1176 associated , 282 candidate associations and 289 ugss . * the 2fglr sources are classified as * : 4 high mass x - ray binaries,11 globular clusters , 51 msps , and 98 psrs , 21 that lie in a pwn and 6 in an snr , and 17 associated with a sfr indicated in our table [ tab : main ] . in the case of 2fglr , among the psr candidate associations that are not in the 1fglr only one source has been recently identified according to the public list of lat - detected gamma - ray pulsars ( see also * ? ? ? * ) . there are also 9 pwne and 66 sources associated with snrs , 11 of them with a known psr lying inside the remnant . then we list 63 _ fermi _ sources as candidate associations with sfrs , one also including a snr . with respect to the 1fgl there is also a new association with a nova ( e.g. , * ? ? ? the extragalactic sky includes 277 bzbs , 338 bzqs , 54 bzus , 3 seyfert galaxies , and 3 starburst galaxies . four _ fermi _ sources are associated with the @xmath0-ray emission arising from the large magellanic cloud , as occurred for the objects classified as galaxy in the 1fglr , and one with m31 ( i.e. , the andromeda galaxy ) @xcite . in addition we find 271 blazar candidates ( bcn ) , 320 _ fermi _ unclassified sources and 289 ugss . in figure [ fig : skydist2 ] we report the comparison between the sky distributions of all the ugss previously listed in the 2fgl and the remaining ones determined in this refined association analysis . as already done for the agus @xcite and the ugss ( e.g. , * ? ? ? * ) listed in the 2fgl we performed a non - parametric analysis of the infrared colors for the _ fermi _ sources classified as bcn and unc in the merged _ fermi _ catalog 1fglr+2fglr . we used kernel density estimation ( kde ; see e.g. , * ? ? ? * ; * ? ? ? * and reference therein ) to verify the consistency of their infrared colors with the so - called _ wise _ gamma - ray strip ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) the kde technique allows us to estimate the probability function of a multivariate distribution and does not require any assumption about the shape of the parent " distributions . consequently , for all the bcn and unc sources with a _ wise _ counterpart we can provide an estimate of the probability @xmath15 that such source is consistent with the _ wise _ gamma - ray strip . in particular we differentiated between @xmath16 and @xmath17 considering the comparison with the infrared colors of the bzb and bzq subclasses , respectively . we considered as training samples to build the _ wise _ gamma - ray strip@xcite and to compare the infrared colors , the roma - bzcat blazars listed in both 1fglr and 2fglr . we remark that _ wise _ counterparts of the bcn and of the unc sources with the infrared analysis flags @xcite have been considered in our analysis and their flags are reported in table [ tab : main ] together with the probabilities derived from the kde analysis . in figure [ fig : kde ] , the isodensity contours drawn from the kde density probabilities are plotted for the _ fermi_-roma - bzcat blazars in the [ 3.4]-[4.6]-[12 ] @xmath4 m color - color diagram . the ir colors of the sources listed in the refined association list of the _ fermi _ catalogs classified as bcn and unc are also shown . values of @xmath16 and @xmath17 greater than 5% indicate that the source has infrared colors consistent with the portion _ wise _ gamma - ray strip constituted by the bzbs and by the bzqs , respectively , at 95% confidence level . -[4.6]-[12 ] @xmath4 m color - color plot . blue contours were computed from the infrared colors of the bzbs while the red ones were derived from the bzqs . the infrared colors of the sources classified as bcn ( top - black circles ) and unc ( bottom - green squares ) are also reported to show their consistency with the _ wise _ gamma - ray strip . the numbers appearing close to each contour correspond to the values of @xmath15 in both panels . the last contour correspond to 95% confidence level ( see section [ sec : kde ] for details ) . a similar analysis for the active galaxies of uncertain type listed in the 2fgl was already performed in massaro et al . ( 2012a ) . here this study has been repeated for both the 1fglr and 2fglr lists . refined association list of the _ fermi _ catalogs . , title="fig:",width=362 ] -[4.6]-[12 ] @xmath4 m color - color plot . blue contours were computed from the infrared colors of the bzbs while the red ones were derived from the bzqs . the infrared colors of the sources classified as bcn ( top - black circles ) and unc ( bottom - green squares ) are also reported to show their consistency with the _ wise _ gamma - ray strip . the numbers appearing close to each contour correspond to the values of @xmath15 in both panels . the last contour correspond to 95% confidence level ( see section [ sec : kde ] for details ) . a similar analysis for the active galaxies of uncertain type listed in the 2fgl was already performed in massaro et al . ( 2012a ) . here this study has been repeated for both the 1fglr and 2fglr lists . refined association list of the _ fermi _ catalogs . , title="fig:",width=362 ] our analysis confirms that 419 out of 499 total bcn ( 262 ) and unc ( 237 ) sources in the refined and merged associations list of the _ fermi _ catalogs are consistent with the _ wise _ gamma - ray strip at 95% confidence level . for the unc sources this situation occurs for 172 sources out of 237 objects while for the bcn in 242 out of 262 cases . we remark that 374 out of the 499 sources analyzed with the kde are associated , 1 is identified , and the remaining 124 are candidate associations . this analysis supports the introduction of the category for the candidate associations , since a large fraction of the bcn and the unc with a _ wise _ counterpart are likely to be blazar - like sources being consistent with the _ wise _ gamma - ray strip . in our analysis , we took into account the correction for galactic extinction for all the _ wise _ magnitudes according to the draine ( 2003 ) relation . as shown in dabrusco et al . ( 2013 ) , this correction affects only marginally the [ 3.4]-[4.6 ] @xmath4 m color , mostly at low galactic latitudes ( i.e. , @xmath1815 ) . the total number of _ fermi _ sources listed uniquely in the 1fglr and in the 2fglr are 2219 . we found spectroscopic information for 177 of them not reported in the previous 1fgl and 2fgl catalogs , including 8 observed during our spectroscopic campaign ( see below ) . we have been able to confirm spectroscopically 117 bl lacs , 21 with firm redshift estimates and 27 quasars ( qsos ) ; an additional eight sources appear to have a bl lac nature but the lack of good optical spectra in the literature or , if a spectrum is available , a low signal to noise ratio did not allow us to verify their natures . thus 24 out of these 177 sources still have an uncertain classification and an uncertain redshift estimate . in particular among them we found 11 sources listed as blazars in ackermann et al . ( 2011 ) plus one listed in healey et al . ( 2007 ) with no optical spectra published or described in literature . these 177 sources are all listed in table [ tab : main ] as bcn with the only exception being one radio source that appears to be a radio galaxy ; in the table notes we report their possible classifications and their redshift estimates with a question mark ( ? ) indicating that further investigation is required . optical spectroscopic observations of four candidate associations and of three associated with unclassified sources ( unc ) were performed with the 2.1-m telescope of the observatorio astronmico nacional ( oan ) in san pedro mrtir ( mxico ) on the nights between 28 june and 2 july 2014 ( ut ) as reported in table [ tab : optical ] . the telescope carries a boller & chivens spectrograph and a 1024@xmath191024 pixel e2v-4240 ccd . the spectrograph was tuned in the 4000@xmath208000 range ( grating 300 l / mm ) , with a resolution of 4.5 per pixel , which corresponds to 8 ( full - width - half - maximum ) , and a 2@xmath215 slit . data were wavelength calibrated using copper - helium - neon - argon lamps , while for flux calibration spectrophotometric standard stars were observed twice during every night of the observing run . in addition we also observed the @xmath0-ray blazar candidate bzb j2340 + 8015 associated with 1fgl j2341.6 + 8015 at the observatorio astrofisico guillermo haro ( oagh ) located at cananea , sonora in mxico . the telescope was equipped with a spectrograph boller & chivens with a ccd site 1k@xmath191k , tuned in the 4000@xmath207000 range ( grating 150 l / mm ) . we used a slit width of 2@xmath215 with a resolution @xmath115 ( fwhm ) . the data reduction for both telescopes and instruments was carried out using the image reduction and analysis facility ( iraf ) package developed by the national optical astronomy observatory , including bias subtraction , spectroscopic flat fielding , optimal extraction of the spectra and interpolation of the wavelength solution . all spectra were reduced and calibrated employing standard techniques in iraf and our own idl routines ( see also * ? ? ? * for additional details ) . these eight sources for which new spectroscopic data are provided in our analysis are listed in table [ tab : main ] as @xmath0-ray blazar candidates ( i.e. , bcn ) . five of them belong to the 1fglr and are all classified as bl lac objects according to our analysis ; for three of them the presence of several spectral features allowed us to determine their redshifts . the log of the spectroscopic observations and the results of our analysis are reported in table [ tab : optical ] , while in figure [ fig : j1548 ] , figure [ fig : j1844 ] , figure [ fig : j2014 ] and figure [ fig : j2133 ] we show their optical spectra together with their finding charts . the remaining three sources are two quasars corresponding to 2fgl j1848.6 + 3241 and 2fgl j2021.5 + 0632 as shown in figures [ fig : j1848 ] and [ fig : j2021 ] at redshifts 0.981 and 0.217 , respectively , and 2fgl j2031.0 + 1938 , a bl lac object with an uncertain redshift since there is a unique emission line visible , potentially identified as mgii ( see figure [ fig : j2031 ] ) . during our observing nights we also obtained the spectra for two 2fgl sources , 2fgl j1719.3 + 1744 and 2fgl j1801.7 + 4405 , each with uncertain redshift as reported in the roma - bzcat ( see the figures in the appendix ) . we were able to confirm the redshift for 2fgl j1801.7 + 4405 while 2fgl j1719.3 + 1744 ( alias bzb j1719 + 1745 ) is completely featureless . results for these two additional spectra are also reported in table [ tab : optical ] . we also observed 1fgl j1942.7 + 1033 , which is a bl lac object already classified in literature ( * ? ? ? * ; * ? ? ? * and appendix for the figure ) , and 1fgl j2300.4 + 3138 for which a series of unidentified absorption features are clearly visible in its optical spectrum . assuming that these unidentified absorption features are due to mgii intervening systems along the line of sight , the source should lie at redshift @xmath220.96 . this source was in the sample observed by shaw et al . ( 2013a ) but we can not confirm their results since we did not find any spectral feature identifiable as civ ( see figure [ fig : j2300 ] ) . finally we also report the spectrum of bzb j2340 + 8015 associated with 1fgl j2341.6 + 8015 observed at oagh that was originally classified as a bl lac candidate in the roma - bzcat at redshift 0.274 . we have been able to confirm the bl lac nature of this source but not its redshift due to its featureless spectrum ( see figure [ fig : j2341 ] ) . all our finding charts are from the digitized sky survey[multiblock footnote omitted ] . 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j1548.7 + 6311 , the target of our observation , is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j1548.7 + 6311 , the target of our observation , is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j1844.1 + 1547 , the target of our observation , is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j1844.1 + 1547 , the target of our observation , is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2014.4 + 0647 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2014.4 + 0647 is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2133.4 + 2532 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2133.4 + 2532 is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j1848.6 + 3241 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j1848.6 + 3241 is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j2021.5 + 0632 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j2021.5 + 0632 is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j2031.0 + 1938 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 2fgl j2031.0 + 1938 is indicated by the red circle.,title="fig:",width=249,height=173 ] spectral + name & yyyy - mm - dd & ( sec ) & & & features + 1fgl j1548.7 + 6311 & 2014 - 06 - 28 & 2x1800 & bl lac & 0.269 & caii h&k , g , h@xmath23 , mgi + 1fgl j1844.1 + 1547 & 2014 - 06 - 29 & 2x1800 & bl lac & ? & none + 1fgl j2014.4 + 0647 & 2014 - 06 - 29 & 2x1800 & bl lac & 0.341 & caii h&k , g , h@xmath23 , mgi + 1fgl j2133.4 + 2532 & 2014 - 06 - 29 & 2x1800 & bl lac & 0.294 & caii h&k , g , h@xmath23 , mgi + 2fgl j1848.6 + 3241 & 2014 - 07 - 02 & 2x1800 & qso & 0.981 & mgii , [ oii ] @xmath243727 + 2fgl j2021.5 + 0632 & 2014 - 07 - 02 & 1800 & qso & 0.217 & h@xmath0 , h@xmath23 , [ oiii ] @xmath244959 & @xmath245007 + 2fgl j2031.0 + 1938 & 2014 - 07 - 01 & 2x1800 & bl lac & 0.668 ? + 2fgl j1719.3 + 1744 & 2014 - 06 - 30 & 2x1800 & bl lac & ? & none + 2fgl j1801.7 + 4405 & 2014 - 06 - 30 & 2x1800 & qso & 0.663 & mgii , [ oii ] @xmath243727 + 1fgl j1942.7 + 1033 & 2014 - 06 - 29 & 2x1800 & bl lac & ? & none + 1fgl j2300.4 + 3138 & 2014 - 07 - 01 & 2x1800 & bl lac & ? & unid . systems + 1fgl j2341.6 + 8015 & 2014 - 08 - 28 & 3x1800 & bl lac & ? & none + [ tab : optical ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2300.4 + 3138 pointed during our observations is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2300.4 + 3138 pointed during our observations is indicated by the red circle.,title="fig:",width=249,height=173 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2341.6 + 8015 is indicated by the red circle.,title="fig:",width=355,height=257 ] 5 finding chart from the digitized sky survey ( red filter ) . the potential counterpart of 1fgl j2341.6 + 8015 is indicated by the red circle.,title="fig:",width=249,height=173 ] we compared our multifrequency analysis on the candidate associations with the results obtained via statistical analyses based on the classification tree ( ct ) and on the logistic regression ( lr ) procedures performed on the 1fgl sources @xcite and using the random forest algorithm , named _ sybil _ , presented by mirabal et al . ( 2012)[multiblock footnote omitted ] for the 2fgl sources . for the 1fgl sources the analysis executed with both the ct and the lr statistical methods assesses the probability of correct classification based on fitting a model form to the _ fermi _ data . the result is a psr - like or an agn - like classification , occurring when the @xmath0-ray source properties are likely more consistent with those of a pulsar or an active galaxy , respectively ; a similar prediction is also made for the 2fgl sources analyzed by the _ sybil _ algorithm . thus , all these methods provide a classification of the potential counterpart on the basis of the @xmath0-ray properties ( e.g. , spectral shape , variability , etc . ) but can not permit to locate it . consequently we can only verify if a candidate association that appear to be a blazar or a pulsar have the corresponding @xmath0-ray source classified as agn - like or psr - like by the statistical analyses cited above . there are 286 candidate associations of type bzb in the 1fglr that were analyzed in ackermann et al . ( 2012 ) . within this sample we found that there is only one bzb and one hmb , the first statistically classified as agn - like while the second as a psr - like . * then there are 24 sources classified as bcn , with 23 of them indicated as agn - like while the last is classified as psr - like . * among the galactic sources , there are : * 7 sources , classified by us as potential msps , four appear to be psr - like while 3 show @xmath0-ray properties more similar to agns ( i.e. , agn - like ) . * 25 out of 29 galactic sources classified are in our analysis as psrs are confirmed psr - like by statistical methods while 4 are indicated as agn - like ; * 2 pwn candidate associations , both resembling psr - like sources in @xmath0-rays . * 5 sources reported in our table [ tab : main ] as snrs , four are classified as psr - like objects and one has an uncertain statistical classification ( i.e. , indicated as conflict in ackermann et al . * 33 out of 47 candidate associations classified as sfrs are psr - like while only 9 appear to be classifiable as agn - like by the statistical methods ; the remaining five are unclassified . the comparison between our candidate associations in the galactic plane and the results of the statistical analyses supports the relevance of searching for the presence of sfrs within the _ fermi _ positional uncertainty region where psrs , snrs , pwne could be embedded therein or shocked regions could be potential sources of @xmath0-rays not yet confirmed . finally , in the sample of 170 unc sources , classified as candidate associations by our analysis , 125 of them appearing to be agn - like while 40 are indeed psr - like ; the remaining five are conflicts between the lr and the ct methods . in particular , four out of eight unc sources having a sfr consistent with the _ fermi _ position are classified as psr - like on the basis of their @xmath0-ray behavior . in the case of the 2fglr , we investigated 133 candidate associations with the _ sybil _ procedure , and classified them accordingly . of these , in the sample of 107 _ fermi _ sources that are classified as agn - like by _ sybil _ : 24 are bcn , 78 are unc type ; then the remaining objects are two sfrs , three psrs , seven msps ( three expected to be psr - like @xmath0-ray sources ) and five sfrs all classified as agn - like . finally , we highlight that the results of the statistical analyses are in agreement with the classification proposed in the refined association lists of both _ fermi _ catalogs . many attempts were made in the past to test for correlations between the radio and @xmath0-ray emissions of agns and in particular for blazars ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? this connection was also used before the _ fermi _ era to search for counterparts of the @xmath0-ray sources @xcite , so motivating all the past and present radio follow up campaigns for the _ fermi _ sources ( e.g. , * ? ? ? * ; * ? ? ? however , biases and selection effects have to be taken into account to prove this correlation properly , since it is important to address intrinsic source variability , biases due to redshift dependence @xcite , the `` common distance '' bias @xcite , problems related to source misidentifications and incorrect associations , to name a few . all these issues can mimic a correlation @xcite . since the launch of _ fermi _ several investigations have also been performed to search for a definitive answer regarding the existence of the radio-@xmath0-ray connection @xcite until it was proved and described accurately in ackermann et al . ( 2011b ) . recently we also suggested that a link between the radio and the @xmath0-ray emissions in blazars can be extended well below @xmath11 ghz @xcite . given the new multifrequency analysis carried out here and the candidate associations , in particular for the bcn class we illustrate the current status of the radio-@xmath0-ray connection . to this end we show both the flux - flux and the luminosity - luminosity scatter plots for the blazars in our merged refined list of _ fermi _ associations ( see figure [ fig : radiofluxes ] and figure [ fig : radioluminosities ] ) . a study of the correlations is out of the scope of this paper . our main goal is verify that sources classified as bcn are in agreement with the behavior of blazars in these parameter spaces . it is worth noting that we divided our sample of blazar - like sources in two subsamples : the northern one and the southern one . the former includes all the blazars ( i.e. , bzb , bzq , and bcn classes ) that have a radio counterpart within the nvss footprint while in the latter the subsample includes those in the area covered by the sumss . we distinguish these two samples since the radio surveys were carried out at different frequencies . finally , we remark that only blazars with a firm redshift estimate as reported in our merged list of refined associations are shown in figure [ fig : radioluminosities ] . according to this figure we could expect most of the bcn sources to be bl lac objects , although this has to be confirmed with optical spectroscopic observations . -ray flux for blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcn sources are black diamonds . sources with a radio counterpart in the nvss are in the upper panel while the sources in sumss are in the lower panel.,title="fig:",width=359,height=257 ] -ray flux for blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcn sources are black diamonds . sources with a radio counterpart in the nvss are in the upper panel while the sources in sumss are in the lower panel.,title="fig:",width=359,height=257 ] -ray luminosities for blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcns sources are black diamonds . sources with a radio counterpart in the nvss are in the upper panel while the sources in sumss are in the lower panel.,title="fig:",width=359,height=257 ] -ray luminosities for blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcns sources are black diamonds . sources with a radio counterpart in the nvss are in the upper panel while the sources in sumss are in the lower panel.,title="fig:",width=359,height=257 ] dabrusco et al . ( 2012 ) studied the _ fermi _ blazar sample listed in the 2fgl and lying in the footprint of the _ wise _ preliminary survey and reported the discovery of a correlation between their infrared and their @xmath0-ray spectral indices . this is directly related to their peculiar infrared colors and the _ wise _ gamma - ray strip @xcite . here we report an updated scatterplot for the spectral indices ( see figure [ fig : indices ] ) as originally presented in dabrusco et al . the correlation the spectral shapes in the infrared and in @xmath0-rays is still present and as occurred in the case of the radio - @xmath0-ray scatter plot ( see section [ sec : radio ] ) the locations of the bcn sources appear to be more consistent with the bl lac population . the linear correlation coefficient for the whole data set is 0.64 . -ray index for the blazar - like sources in our merged list of refined _ fermi _ associations . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcn sources are black diamonds.,width=359,height=257 ] finally , we also present the connection between infrared and @xmath0-ray fluxes and luminosities in figure [ fig : infrared ] that is again in good agreement with our previous results @xcite . -ray fluxes ( upper panel ) and luminosities ( lower panel ) for the blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcn sources are black diamonds.,title="fig:",width=359,height=257 ] -ray fluxes ( upper panel ) and luminosities ( lower panel ) for the blazar - like sources . spectroscopically confirmed bzbs and bzqs are shown in blue circles and red squares , respectively , while bcn sources are black diamonds.,title="fig:",width=359,height=257 ] two years after the release of the second _ fermi _ source catalog we present a comprehensive multifrequency investigation of all the @xmath0-ray associations listed in both the 1fgl and the 2fgl catalogs . first , we introduced a new category of @xmath0-ray source associations in addition to _ identified _ and _ associated _ sources . we label as _ candidate associations _ those @xmath0-ray sources having a potential low - energy counterpart of few specific classes of well - known @xmath0-ray emitters lying within the _ fermi _ positional uncertainty region and/or with angular separations between the _ fermi _ and the counterpart position smaller than the maximum one for all the associated sources of the same class . then we provided a new classification scheme for the low - energy counterparts of the _ fermi _ sources based on the multifrequency observations and on the optical spectroscopic information now available . we also presented here a cross - matching between the _ fermi _ catalogs and several surveys of sfrs ( see section [ sec : crossmatches ] for details ) to highlight the possibility that an unknown snr , pwn and/or psr is embedded therein , and to provide information that could be used to refine models of diffuse galactic @xmath0-ray for future releases of the _ fermi _ catalogs . the total number of @xmath0-ray sources considered in our investigation for both the 1fgl and the 2fgl comprises 2219 unique _ fermi _ objects , all listed in table [ tab : main ] with each assigned counterpart and their main multifrequency properties . overall , in the refined association list of the _ fermi _ catalogs , we found 174 _ fermi _ sources with a sfr consistent with their @xmath0-ray positions . in particular , 60 _ fermi _ objects out of these 174 do not have the `` c '' flag in at least one of the _ fermi _ names and include : 13 identified sources , 17 associated , and 30 candidate associations . their counterparts are also classified as four hmbs ; 17 psrs , among which three lie in a pwn and two in a snr ; three pwne , one with a psr ; one radio galaxy ; thirteen snrs , two with a psr included ; one binary star ; three unclassified sources ; and 19 sfrs . we found spectroscopic information for 177 extragalactic @xmath0-ray sources not reported in the previous version of the _ fermi _ catalogs . we included analyses of eight new optical spectroscopic observations performed with the 2.1-meter telescope of the oan in san pedro mrtir and with the oagh to confirm the natures of these blazar - like sources found in our multifrequency investigation . blazar candidates and counterparts with unknown origin ( i.e. , unc ) were also analyzed with the kde technique to determine the fraction having _ wise _ ir colors similar to known @xmath0-ray blazars . we compared the refined associations listed in both the _ fermi _ catalogs with the classification proposed on the basis of statistical investigations @xcite and we found a good agreement with our results . we note that in the combined list of refined associations for 1fgl and 2fgl , among 2219 sources , 394 are still ugss ( i.e. , @xmath118% of the entire sample ) with 191 of them having no @xmath0-ray analysis flags . this clean sample of 191 ugss appears to have a uniform distribution in the sky with a small excess toward the galactic plane , as shown in figure [ fig : skydistugs ] . moreover , we conclude that the fraction of _ fermi _ sources with plausible counterparts , combining identifications , associations and candidate associations , within their @xmath0-ray positional uncertainty regions , is @xmath180% and up to @xmath190% when considering sources with no @xmath0-ray analysis flags . finally , we remark that within the sample of 191 ugss that do not have any @xmath0-ray analysis flags are 40 sources ( listed in table [ tab : special ] ) that do not have any nvss and/or sumss radio source within their positional uncertainty regions at 95% level of confidence . their all - sky distribution , separated into those lying in the nvss footprint ( i.e. , declination greater than @xmath1440 ) and the others observable from the southern hemisphere is shown in figure [ fig : skynomatches ] . the excess in the subsample of those with declinations less than @xmath1440 , clearly visible in figure [ fig : skynomatches ] , could be due to shallower coverage by sumss relative to the nvss catalog . -ray analysis flag . we show those lying in the nvss footprint ( i.e. , declination greater than @xmath1440 ) as black circles while the others in the southern hemisphere are indicated as orange circles.,width=359,height=257 ] it is worth mentioning that for these 40 ugss the distributions of the @xmath0-ray spectral index and of the @xmath0-ray fluxes do not show any significant differences from those of bzqs and the bzbs ( see figure [ fig : comparing ] ) . however it is intriguing that these ugss tend to be brighter than the bzb population in the gamma - rays . it is unlikely that this small ugs subsample is constituted by blazars since given their @xmath0-ray fluxes @xmath25 and assuming the typical radio-@xmath0-ray fluxes of the blazar population , the expected radio flux densities should be greater than @xmath1 50 mjy for a large fraction , well above the nvss and the sumss flux thresholds . on the other hand we also conclude that their steep values of @xmath0-ray spectral indices are not compatible with sources emitting via dark matter annihilation ( e.g. , * ? ? ? * ; * ? ? ? -ray spectral index ( top ) and that of the energy flux ( bottom ) for the 40 ugss ( straight black line ) that do not have any nvss and/or sumss radio source within their positional uncertainty regions and without @xmath0-ray analysis flags in comparison with those of the bzbs ( dashed blue line ) and the bzqs ( straight red line ) . cumulative distributions are also reported on the right panels . the energy flux is the one reported in both the 1fgl and the 2fgl catalogs.,title="fig:",width=355,height=241 ] -ray spectral index ( top ) and that of the energy flux ( bottom ) for the 40 ugss ( straight black line ) that do not have any nvss and/or sumss radio source within their positional uncertainty regions and without @xmath0-ray analysis flags in comparison with those of the bzbs ( dashed blue line ) and the bzqs ( straight red line ) . cumulative distributions are also reported on the right panels . the energy flux is the one reported in both the 1fgl and the 2fgl catalogs.,title="fig:",width=355,height=241 ] name & name + 1fgl j0032.7 - 5519 & 2fgl j0032.7 - 5521 + 1fgl j0212.3 + 5319 & 2fgl j0212.1 + 5318 + 1fgl j0545.6 + 6022 & 2fgl j0545.6 + 6018 + 1fgl j0709.0 - 1116 & + 1fgl j0802.4 - 5622 & 2fgl j0802.7 - 5615 + 1fgl j0854.6 - 4504 & 2fgl j0854.7 - 4501 + 1fgl j0933.9 - 6228 & 2fgl j0934.0 - 6231 + 1fgl j1036.2 - 6719 & 2fgl j1036.1 - 6722 + 1fgl j1117.0 - 5339 & 2fgl j1117.2 - 5341 + 1fgl j1306.4 - 6038 & 2fgl j1306.2 - 6044 + 1fgl j1405.5 - 5846 & + 1fgl j1441.8 - 6404 & + 1fgl j1518.0 - 5233 & 2fgl j1518.4 - 5233 + 1fgl j1639.5 - 5152 & + 1fgl j1650.3 - 5410 & + 1fgl j1702.4 - 5653 & 2fgl j1702.5 - 5654 + 1fgl j1743.5 - 3314 & + 1fgl j1806.2 + 0609 & 2fgl j1805.8 + 0612 + 1fgl j1825.7 - 1410c & + 1fgl j1948.6 + 2437 & + 1fgl j2112.5 - 3044 & 2fgl j2112.5 - 3042 + 1fgl j2227.4 - 7804 & + 1fgl j2325.8 - 4043 & + 1fgl j2333.0 - 5535 & 2fgl j2332.5 - 5535 + & 2fgl j0237.9 + 5238 + & 2fgl j1016.4 - 4244 + & 2fgl j1058.7 - 6621 + & 2fgl j1208.5 - 6240 + & 2fgl j1404.0 - 5244 + & 2fgl j1410.4 + 7411 + & 2fgl j1422.3 - 6841 + & 2fgl j1507.0 - 6223 + & 2fgl j1617.3 - 5336 + & 2fgl j1626.4 - 4408 + & 2fgl j1639.8 - 5145 + & 2fgl j1641.8 - 5319 + & 2fgl j1643.3 - 4928 + & 2fgl j1741.1 - 6750 + & 2fgl j1742.5 - 3323 + & 2fgl j1753.8 - 4446 + we thank the anonymous referee for useful comments that led to improvements in the paper . f. massaro and g. tosti also thank l. costamante . this investigation is supported by the nasa grants nnx12ao97 g and nnx13ap20 g . the work by g. tosti is supported by the asi / inaf contract i/005/12/0 . h. a. smith acknowledges partial support from nasa / jpl grant rsa 1369566 . hof was funded by a postdoctoral unam grant and is currently granted by a ctedra conacyt para jvenes investigadores . v. chavushyan acknowledges funding by conacyt research grant 151494 ( mxico ) . we thank the staff at the observatorio astronmico nacional in san pedro mrtir ( mxico ) for all their help during the observation runs . part of this work is based on archival data , software or on - line services provided by the asi science data center . this research has made use of data obtained from the high - energy astrophysics science archive research center ( heasarc ) provided by nasa s goddard space flight center ; the simbad database operated at cds , strasbourg , france ; the nasa / ipac extragalactic database ( ned ) operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration . part of this work is based on the nvss ( nrao vla sky survey ) : the national radio astronomy observatory is operated by associated universities , inc . , under contract with the national science foundation and on the vla low - frequency sky survey ( vlss ) . the molonglo observatory site manager , duncan campbell - wilson , and the staff , jeff webb , michael white and john barry , are responsible for the smooth operation of molonglo observatory synthesis telescope ( most ) and the day - to - day observing programme of sumss . the sumss survey is dedicated to michael large whose expertise and vision made the project possible . the most is operated by the school of physics with the support of the australian research council and the science foundation for physics within the university of sydney . this publication makes use of data products from the wide - field infrared survey explorer , which is a joint project of the university of california , los angeles , and the jet propulsion laboratory / california institute of technology , funded by the national aeronautics and space administration . this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation . funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s . department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england . the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions . the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington . this research has made use of the usnofs image and catalogue archive operated by the united states naval observatory , flagstaff station ( http://www.nofs.navy.mil/data/fchpix/ ) . the wenss project was a collaboration between the netherlands foundation for research in astronomy and the leiden observatory . we acknowledge the wenss team consisted of ger de bruyn , yuan tang , roeland rengelink , george miley , huub rottgering , malcolm bremer , martin bremer , wim brouw , ernst raimond and david fullagar for the extensive work aimed at producing the wenss catalog . topcat[multiblock footnote omitted ] @xcite for the preparation and manipulation of the tabular data and the images . the aladin java applet[multiblock footnote omitted ] was used to create the finding charts reported in this paper @xcite . it can be started from the cds ( strasbourg - france ) , from the cfa ( harvard - usa ) , from the adac ( tokyo - japan ) , from the iucaa ( pune - india ) , from the ukadc ( cambridge - uk ) , or from the cadc ( victoria - canada ) . abdo , a. a. , ackermann , m. , ajello , m. et al . 2009 phrvl , 103 , 1101 abdo , a. a. , ackermann , m. , ajello , m. et al . 2010a apjs , 188 , 405 abdo , a. a. , ackermann , m. , ajello , m. et al . 2010b apj , 715 , 429 abdo , a. a. , ackermann , m. , ajello , m. et al . 2010c apj , 723 , 649 abdo , a. a. , ackermann , m. , ajello , m. et al . 2010d a&a , 523 , 2 abdo , a. a. , ajello , m. , allafort , a. et al . 2013 apjs , 208 , 17 acero , f. , donato , d. , ojha , r. et al . 2013 apj , 779 , 133 ackermann , m. , ajello , m. , allafort , a. et al . 2011a apj , 743 , 171 ackermann , m. , ajello , m. , allafort , a. et al . 2011b apj , 741 , 30 ackermann , m. , ajello , m. , allafort , a. et al . 2012 apj , 753 , 83 ackermann , m. , ajello , m. , allafort , a. et al . 2013 apj , 209 , 34 ahn , c. p. , alexandroff , r. , allende prieto , c. et al . 2012 , apjs , 203 , 21 anderson , l. d. , bania , t. m. , balser , dana s. et al . 2014 apjs , 212 , 1 araudo a. , et al . 2007 , a&a , 476 , 1289 atwood , w. b. , abdo , a. a. , ackermann , m. et al . 2009 apj , 697 , 1071 bade , n. , fink , h. h.,engels , d. et al . 1995 a&as , 110 , 469 bauer , f. e. , condon , j. j. , thuan , t. x. et al . 2000 apjs , 129 , 547 baker , j. c. , hunstead r. w. & brinkmann w. 1995 mnras , 277 , 553 becker , r. h. , white , r. l. , helfand , d. j. , zoonematkermani , s. 1994 apjs , 347 , 92 becker , r. h. , white , r. l. , helfand , d. j. 1995 apj , 450 , 559 belikov , a. v. , buckley , m. r. , hooper , d. 2012 phrvd , 86d3504 bennett , a. s. 1962 mnras , 125 , 75 berlin , a. & hooper , d. 2014 phrvd , 89a , 6014 beuermann k. , thomas h .- c . , reinsch k. et al . 1999 a&a , 347 , 47 bica , e. ; dutra , c. m. ; soares , j. ; barbuy , b. 2003 a&a , 404 , 223 bikmaev , i. f. , burenin , r. a. , revnivtsev , m. g. et al . 2008 astl , 34 , 653 bosch - ramon v. , et al . , 2010 a&a 511 , a8 bonnarel , f. , fernique , p. , bienaym , o. et al . 2000 a&as , 143 , 33 britzen , s. , brinkmann , w. , campbell , r. m. et al . 2007 a&a , 476 , 759 cheung , c. c. ; donato , d. ; gehrels , n. ; sokolovsky , k. v. ; giroletti , m. 2012 apj , 756 , 33 cheung , c. c. , shore , s. n. , jean , p. proceedings of the aas meeting # 223 , # 113.01 codella , c. , felli , m. , natale , v. 1994 a&a , 284 , 233 cohen , a. s. , lane , w. m. , cotton , w. d. et al . 2007 aj , 134 , 1245 condon , j. j. , cotton , w. d. , greisen , e. w. et al . 1998 , aj , 115 , 1693 cowperthwaite , philip s. et al . 2013 aj , 146 , 110 cutri et al . 2012 wise.rept , 1c culverhouse , t. , ade , p. , bock , j. et al . 2011 apjs , 195 , 8 dabrusco , r. , longo , g. , walton , n. a. 2009 mnras , 396 , 223 dabrusco , r. , massaro , f. , ajello , m. , grindlay , j. e. , smith , howard a. & tosti , g. 2012 apj , 748 , 68 dabrusco , r. , massaro , f. , paggi , a. et al . 2013 apjs , 206 , 12 de breuck , c. , tang , y. , de bruyn , a. g. et al . 2002 a&a , 394 , 59 de ruiter , h. r. , willis , a. g. , arp , h. c. 1977 a&as , 28 , 211 del valle , m. v. & romero , g. e. 2012 a&a 543 , a56 doert , m. & errando , m. 2014 apj , 782 , 41 douglas , j. n. , bash , f. n. , bozyan , f. a. , torrence , g. w. , wolfe , c. 1996 aj , 111 , 1945 draine , b. t. 2003 , ara&a , 41 , 241 drlica - wagner , a. ; gomez - vargas , g. a. ; hewitt , j. w. , linden , t. , tibaldo , l. drinkwater , m. j. , webster , r. l. , francis , p. j. et al . 1997 mnras , 284 , 85 edge , d.o . , shakeshaft , j. r. , mcadam , w. b. , baldwin , j. e. , archer , s. 1959 mmras , 68 , 37 elvis , m. , maccacaro , t. , wilson , a. s. et al . 1978 , mnras , 183 , 129 espinoza , c. m. et al . 2013 mnras , 430 , 571 evans , i. n. , primini , f. a. , glotfelty , k. j. et al . 2010 apjs , 189 , 37 evans , p. a. , osborne , j. p. , beardmore , a. p. 2014 apjs , 210 , 8 ferrand , g. , safi - harb , s. 2012 adspr , 49 , 9 , 1313 ghirlanda , g. , ghisellini , g. , tavecchio , f. & foschini , l. 2010 , mnras , 407 , 791 ghirlanda , g. , ghisellini , g. , tavecchio , f. , foschini , l. & bonnoli , g. 2011 , mnras , 413 , 852 giommi , p. et al . 2012 a&a , 541a.160 giveon , u. , becker , r. h. , helfand , d. j. et al . 2005 aj , 129 , 348 green , a. j. , cram , l. e. , large , m. i. , ye , taisheng 1999 apjs , 122 , 207 green d. a. , 2009 , bulletin of the astronomical society of india , 37 , 45 gregory , p. c. & condon , j. j. 1991 apjs , 75 , 1011 gregory , p. c. , scott , w. k. , douglas , k. , condon , j. j. 1996 apjs , 103 , 427 hanabata y. et al . 2014 apj , 786 , 145 hassan , t. , mirabal , n. , contreras , j. l. , oya , i. 2013 mnras , 428 , 220 healey , s. e. , romani , r. w. , taylor , g. b. et al . 2007 apjs , 171 , 61 healey , s. e. ; romani , r. w. , cotter , g. et al . 2008 apjs , 175 , 97 hewett , p. c. & wild , v. 2010 mnras , 405 , 2302 hewitt , a. , burbidge , g. 1989 qso , m , 0000 hobbs g. et al . 2004 mnras , 352 , 1439 hovatta , t. , lister , m. l. , aller , m. f. et al . 2012 aj , 144 , 105 hovatta , t. , aller , m. f. , aller , h. d. et al . 2014 aj , 147 , 143 hughes , v. a. & macleod 1989 aj , 97 , 786 kataoka , j. ; yatsu , y. ; kawai , n. 2012 apj , 757 , 176 kovalev , y. y. 2009 apj , 707l , 56 kuchar , t. a. & clark , f. o. 1997 apj , 488 , 224 johnston , k. j. , fey , a. l. , zacharias , n. et al . 1995 aj , 110 , 880 jones , h. d. et al . 2004 mnras , 355 , 747 jones , h. d. et al . 2009 mnras , 399 , 683 landoni , m. massaro , f. paggi , a. et al . 2014 aj submitted landt h. , padovani , p. , eric s. p. et al . 2001 mnras , 323 , 757 lane , w. m. , cotton , w. d. , van velzen , s. et al . 2014 mnras , 440 , 327 laurino , o. , dabrusco , r. , longo , g. , riccio , g. 2011 mnras , 418 , 2165 lister , m. l. , aller , m. , aller , h. et al . 2011 apj , 742 , 27 lpez - caniego , m. et al . 2013 mnras , 430 , 1566 maeda , k. ; kataoka , j. ; nakamori , t. et al . 2011 apj , 729 , 103 mahony , e. k. , sadler , e. m. , murphy , t. , et al . 2010 , apj , 718 , 587 manchester , r. n. , hobbs , g.b . , teoh , a. & hobbs , m. 2005 , aj , 129 , 1993 mao , l. s. 2011 newa , 16 , 503 marti , j. , paredes , j. m. , bloom , j. s. et al . 2004 a&a , 413 , 309 maselli , a. , cusumano , g. , massaro , e. et al . 2010 a&a , 520a , 47 masetti , n. , sbarufatti , b. , parisi , p. 2013 a&a , 559a , 58 massaro , e. , giommi , p. , leto , c. et al . 2009 a&a , 495 , 691 massaro , f. , dabrusco , r. , ajello , m. , grindlay , j. e. & smith , h. a. 2011a apj , 740l , 48 massaro , e. , giommi , p. , leto , c. et al . 2011b multifrequency catalogue of blazars ( 3rd edition ) " , aracne editrice , rome , italy massaro , f. , dabrusco , r. , tosti , g. , ajello , m. , gasparrini , d. , grindlay , j. e. & smith , howard a. 2012a apj , 750 , 138 massaro , f. , dabrusco , r. , tosti , g. , ajello , m. , paggi , a. , gasparrini , 2012b apj , 752 , 61 massaro , f. , dabrusco , r. , paggi , a. , tosti , g. , gasparrini , d. 2012c apj , 750l , 35 massaro , f. , dabrusco , r. , paggi , a. et al . 2013a apjs , 206 , 13 massaro , f. , dabrusco , r. , giroletti , m. et al . 2013b apjs , 207 , 4 massaro , f. , dabrusco , r. , paggi , a. et al . 2013c apjs , 209 , 10 massaro , f. , giroletti , m. , paggi , a. et al . 2013d apjs , 208 , 15 massaro , f. , giroletti , m. , dabrusco , r. et al . 2014a apjs , 213 , 3 massaro , f. , masetti , n. , dabrusco , r. et al . 2014b aj in press massaro , f. , masetti , n. , dabrusco , r. et al . 2014c a&a submitted mattox , j. r. , schachter , j. , molnar , l. , hartman , r. c. , patnaik , a. r. 1997 apj , 481 , 95 mattox , j. r. , hartman , r. c. , reimer , o. 2001 apjs , 135 , 155 matheson , t. et al . , 2008 , aj , 135 , 1598 mauch , t. , murphy , t. , buttery , h. j. et al . 2003 mnras , 342 , 1117 maza , j. , wischnjewsky , m. , antezana , r. & gonzlez , l. e. 1995 rmxaa , 31 , 119 mirabal , n. 2009 apj , 701 , 129 mirabal , n. ; fras - martinez , v. ; hassan , t. ; fras - martinez , e. 2012 mnras , 424l , 64 mirabal , n. 2013a mnras , 432l , 71 mirabal , n. 2013b mnras , 436 , 2461 mitton , s. , hazard , c. , and whelan , j. j. 1977 mnras , 179 , 569 monet , d. g. et al . 2003 aj , 125 , 984 montmerle , t. 1979 apj , 231 , 95 montmerle , t. 2009 conf . `` high - energy phenomena in massive stars '' , jaen ( spain ) , feb . 2 - 5 , 2009 , 14 p. [ arxiv.org/abs/0909.0222 ] murphy , t. et al . 2010 mnras , 402 , 2403 moskalenko , i. v. , digel , s. w. , porter , t. a. et al . 2007 nuphs , 173 , 44 mucke , a. , pohl , m. , reich , p. , et al . 1997 , a&a , 320 , 33 muller , c. , kadler , m. , ojha , r. et al . 2014 a&a , 562a , 4 nolan , p. l. , abdo , a. a. , ackermann , m. et al . 2012 apjs , 199 , 31 nori , m. , giroletti , m. , massaro , f. , dabrusco , r. , paggi , a. & tosti , g. et al . 2014 apjs 212 , 3 padovani , p. , ghisellini , g. , fabian , a. c. & celotti , a. 1993 , mnras , 260 , l21 paggi , a. , massaro , f. , dabrusco , r. et al . 2013 apjs , 209 , 9 paggi , a. , milisavljevic , d. , masetti , n. et al . 2014 aj , 147 , 112 paladini , r. , burigana , c. , davies , r. d. et al . 2003 a&a 397 , 213 pavlidou , v. et al . 2012 apj 751 , 149 planck collaboration 2014 a&a submitted [ arxiv:1303.0576 ] petrov , l. , phillips , c. , bertarini , a. , murphy , t. , sadler , e. m. 2011 mnras , 414 , 2528 petrov , l. , mahony , e. k. , edwards , p. g. et al . 2013 mnras , 432 , 1294 pletsch , h. j. et al . 2012a apj , 744 , 105 pletsch , h. j. et al . 2012b apj , 755l , 20 pletsch , h. j. et al . 2013 apj , 779l , 11 prestage , r. m. , & peacock , j. a. , 1983 , mnras , 204 , 355 quintana , h. & ramirez a. 1995 apjs , 96 , 343 raiteri , c. m. , villata , m. , carnerero , m. i. et al . 2014 mnras , 442 , 629 ray , p. s. et al . 2012 [ arxiv:1205.3089 ] ray , p. s. et al . 2013 apj , 763l , 13 rengelink , r. , tang , y. , de bruyn , a. g. et al . 1997 , a&a suppl . 124 , 259 richter , g. a. , 1975 , astroh . nachrichten , 296 , 65 romani , r. w. , filippenko , a. v. , cenko , s. b. 2014 apjl submitted [ arxiv:1408.2886 ] rowell , g. et al . `` proceedings of the 25th texas symposium on relativistic astrophysics . '' december 6 - 10 , 2010 . heidelberg , germany . editors : frank m. rieger ( chair ) , christopher van eldik and werner hofmann . salamon , m. h. & stecker , f. w. 1994 , apj , 430 , l21 saxton , r. d. 2008 a&a , 480 , 611 schinzel , f. k. , petrov , l. , taylor , g. b. apjs submitted [ http://arxiv.org/abs/1408.6217 ] sharpless , s. 1959 apjs , 4 , 275 shaw , m. s. , romani , r. w. ; cotter , g. et al . 2013 apj , 764 , 135 shaw , m. s. , filippenko , a. v. , romani , r. w. et al . 2013 aj , 146 , 127 schwope , a. , hasinger , g. , lehmann , i. et al . 2000 an , 321 , 1 skrutskie , m. f. et al . 2006 , aj , 131 , 1163 spinrad , h. , marr , j. , aguilar , l. , djorgovski , s. 1985 pasp , 97 , 932 stephen , j. b. ; bassani , l. , landi , r. , malizia , a. , sguera , v. , bazzano , a. , masetti , n. 2010 mnras , 408 , 422 stern , d. & assef , r. j. 2013 apj , 764l , 30 stroh , m. c. & falcone , a. d. 2013 , apjs , 207 , 28 stecker , f. w. , salamon , m. h. & malkan , m. a. 1993 , apj , 410 , l71 stickel , m. & kuehr , h. 1996a a&as , 115 , 1 stickel , m. & kuehr , h. 1996b a&as , 115 , 11 sutherland , w. & saunders , w. 1992 mnras , 259 , 413 takahashi , y. ; kataoka , j. ; nakamori , t. et al . 2012 apj , 747 , 64 takeuchi , y. , kataoka , j. , maeda , k. et al . 2013 apjs , 208 , 25 taylor a.r . , goss w.m . , coleman p.h . 1996 apjs , 107 , 239 taylor , g. b. , healey , s. e. , helmboldt , j. f. , et al . 2007 , apj , 671 , 1355 taylor , m. b. 2005 , asp conf . ser . , 347 , 29 thompson , m. a. , hatchell , j. , walsh , a. j. et al . 2006 a&a , 453 , 1003 thompson , d. j. , djorgovski , s. & de carvalho , r. 1990 pasp , 102 , 1235 tsarevsky g. , de freitas pacheco , j. a. , kardashev , n. et al . 2005 a&a , 438 , 949 titov , o. , jauncey , d. l. , johnston , h. m. et al . 2011 aj , 142 , 165 tody , d.1986 spie , 627 , 733 vandenbroucke , j. , buehler , r. , ajello , m. et al . 2010 apj , 718l , 166 vettolani , g. cappi , r. chincarini , g. voges , w. et al . 1999 a&a , 349 , 389 voges , w. et al . 2000 iauc , 7432r , 1 . warwick , r. s. ; saxton , r. d. ; read , a. m. 2012 a&a , 548a , 99 werner , m. ; reimer , o. ; reimer , a. ; egberts , k. 2013 a&a , 555a , 102 white , g. l. , jauncey , d. l. savage , a. et al . 1988 apj 327 , 561 white , r. l. , becker , r. h. helfand , d. j. , gregg , m. d. et al . 1997 apj , 475 , 479 wolstencroft , r. d. et al . 1986 , mnras , 223 , 279 wright , a. & otrupcek , r. 1990 pks , c , 0 wright , a. e. , griffith , m. r. , burke , b. f. , ekers , r. d. 1994 apjs , 91 , 111 wright , e. l. , et al . 2010 aj , 140 , 1868 zechlin , h .- s . , fernandes , m. v. , elsasser , d. , horns , d. 2012 a&a , 538a , 93 here we present the spectra of 2fgl j1719.3 + 1744 and 2fgl j1801.7 + 4405 , both with uncertain redshift , as reported in the roma - bzcat together with that of 1fgl j1942.7 + 1033 . we confirmed the redshift for 2fgl j1801.7 + 4405 and the bl lac classification of 1fgl j1942.7 + 1033 @xcite while 2fgl j1719.3 + 1744 ( alias bzbj1719 + 1745 ) has a completely featureless spectrum . @xmath243727 emission lines superimposed to the optical continuum are visible in 2fgl j1801.7 + 4405 and allowed us to classify the source as a quasar , while the lack of features in the other two spectra led toward a bl lac classification . ( lower panel ) the 5 @xmath19 5 finding chart from the digitized sky survey ( red filter ) . the potential counterparts pointed during our observations is indicated by the red circle . , title="fig:",width=241 ] @xmath243727 emission lines superimposed to the optical continuum are visible in 2fgl j1801.7 + 4405 and allowed us to classify the source as a quasar , while the lack of features in the other two spectra led toward a bl lac classification . ( lower panel ) the 5 @xmath19 5 finding chart from the digitized sky survey ( red filter ) . the potential counterparts pointed during our observations is indicated by the red circle . , title="fig:",width=241 ]	the _ fermi_-large area telescope ( lat ) first source catalog ( 1fgl ) was released in february 2010 and the _ fermi_-lat 2-year source catalog ( 2fgl ) appeared in april 2012 , based on data from 24 months of operation . since their releases , many follow up observations of unidentified gamma - ray sources ( ugss ) were performed and new procedures to associate gamma - ray sources with potential counterparts at other wavelengths were developed . here we review and characterize all the associations as published in the 1fgl and 2fgl catalog on the basis of multifrequency archival observations . in particular we located 177 spectra for the low - energy counterparts that were not listed in the previous _ fermi _ catalogs , and in addition we present new spectroscopic observations of 8 @xmath0-ray blazar candidates . based on our investigations , we introduce a new counterpart category of `` candidate associations '' and propose a refined classification for the candidate low - energy counterparts of the _ fermi _ sources . we compare the 1fgl - assigned counterparts with those listed in the 2fgl to determine which unassociated sources became associated in later releases of the _ fermi _ catalogs . we also search for potential counterparts to all the remaining unassociated _ fermi_sources . finally , we prepare a refined and merged list of all the associations of the 1fgl plus 2fgl catalogs that includes 2219 unique _ fermi _ objects . this is the most comprehensive and systematic study of all the associations collected for the @xmath0-ray sources available to date . we conclude that 80% of the _ fermi _ sources have at least one known plausible gamma - ray emitter within their positional uncertainty regions .
decision - making under risk attracts attention of economists for a long time @xcite @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , ( * ? ? ? * chapter 3 ) . the mathematical expectation of profits and losses in a game or trading is not always the main factor influencing decisions @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , ( * ? ? ? * chapter 4 ) , @xcite . understanding the nature of award in a game and construction of suitable measures for comparing awards and choosing between games is a challenging task . here , the _ certainty _ of _ sample mean _ profits and losses and _ play time _ are considered with respect to decision making . based on maurice allais s research @xcite , daniel kahneman and amos tversky @xcite found out that 80 percent of 95 students and university faculty preferred prospect b in problem 3 : choose between ( a ) 4,000 israeli pounds received with 80 percent of chance and ( b ) 3,000 for sure . 3,000 were the median net monthly income . studying these results , the author knew how to compute _ mathematical expectations _ * - 69 , chapter iv mathematical expectations ) . it is equal to @xmath0 for b and @xmath1 for a. esteeming the extra 200 , he mentally joined the majority declining the gift . kahneman and tversky label this phenomenon the _ certainty effect_. in problem 3 , 80 percent choosing b is its _ experimental measure_. can the _ probability theory _ match _ theoretically _ the experimental _ fractions of respondents _ ? [ [ two - point - distribution ] ] two - point distribution + + + + + + + + + + + + + + + + + + + + + + represents a _ random variable _ @xmath2 with two outcomes @xmath3 , @xmath4 and probabilities @xmath5 , @xmath6 . its mathematical expectation @xmath7 , _ variance _ @xmath8 , _ third central moment _ @xmath9 , _ fourth central moment _ @xmath10 , _ skewness _ @xmath11 , _ excess kurtosis _ @xmath12 , and _ entropy _ @xmath13 are @xmath14 @xmath15 ^ 2)=e(\xi^2)-\alpha_1(\xi)^2=(a_p - a_q)^2 p ( 1 - p),\ ] ] @xmath16 ^ 3 ) = ( a_p - a_q)^3p(1-p)(1 - 2p),\ ] ] @xmath17 ^ 4 ) = ( a_p - a_q)^4p(1-p)(1 - 3p + 3p^2),\ ] ] @xmath18 ^ 3)}{d(\xi)^{\frac{3}{2}}}=\frac{a_p - a_q}{\|a_p - a_q\| } \times \frac{1 - 2p}{\sqrt{p(1-p)}},\ ] ] @xmath19 ^ 4)}{d(\xi)^2 } - 3 = \frac{1 - 6p + 6p^2}{p(1 - p)},\ ] ] @xmath20 for problem 3 , in a @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 , _ standard deviation _ @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , and in b @xmath32 , @xmath33 , @xmath34 , @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 and @xmath41 are undefined , @xmath42 . computing entropy , we follow @xcite and set @xmath43 for @xmath44 . undoubtedly , declining the greater @xmath7 , voters choose a greater _ award_. the _ `` paradox of irrationality '' _ arises because in b the award coincides with @xmath36 but in a it is not @xmath25 . _ indeed , in a the award is random but the mathematical expectation is not . already because of this the award is not the mathematical expectation . _ [ [ sample - mean . ] ] sample mean . + + + + + + + + + + + + in order to see better what the award in prospect a is , the author has `` tortured '' one of the human beings and formulated four variations of problem 3 , where he might feel comfortable choosing a. a gambler may 1 . play a fixed number of times known in advance , and get the mean ; 2 . play unlimitedly , choose the stopping time , and get the known mean ; 3 . play unlimitedly , choose the stopping time , and get the last known value ; 4 . gather any number of helpers ; all choose a once ; gains are summed ; the gambler gets the mean . each variation wastes gambler s and helpers time , if all select b. for the author , choosing a , each variation intensifies his feeling to gain more following to the expectation and even more in the third variation . for him , _ to go or not to go with the fixed positive mathematical expectation _ depends on how many times @xmath45 the game can be played and the cost of each game , if the expectation does not include the latter . in @xcite , @xmath45 is represented by the _ floor function _ of the product of the playing frequency @xmath46 and time @xmath47 : @xmath48 . the _ sample mean _ @xmath49 depends on the numbers of outcomes @xmath50 and @xmath51 in @xmath52 trials @xmath53 the @xmath49 is random due to @xmath50 . if the fixed @xmath54 games represent _ independent identically distributed _ , i.i.d . , binary variables , then @xmath50 obeys a _ binomial distribution_. in prospect b of problem 3 , @xmath55 , @xmath56 make @xmath57 a _ constant award_. in prospect a of problem 3 and variations 1 , 2 , and 4 , the @xmath58 is a _ random award_. _ one cares about all properties of the award but not only one constant _ @xmath59 _ if those are responsible for getting nothing . _ variation 1 fixes @xmath45 in advance . variation 2 makes @xmath45 random and dependent on the observed @xmath49 . variation 3 makes @xmath45 random , values @xmath60 irrelevant , and the award 4,000 likely . variation 4 switches to the number of gamblers @xmath45 . for a fixed @xmath54 , @xmath61 @xcite and @xmath62 @xmath63 ^ 2 ) = e(a_1 ^ 2 ) - e^2(a_1 ) = \frac{(a_p - a_q)^2pq}{n}=\frac{d(\xi)}{\lfloor \nu t \rfloor}.\ ] ] the variance decreases with increasing @xmath45 . this formula is in agreement with the _ limit theorems _ @xcite applicable to the mean sum of i.i.d . variables with finite variance @xmath64 in _ bernoulli trials_. an analytic method for computing _ beginning moments _ @xmath65 of sums of i.i.d . random variables is suggested in @xcite . then , the _ central moments _ @xmath66 can be expressed via @xmath67 as @xmath68^k ) = e([\xi - \alpha_1]^k ) = e\left ( \sum_{j = 0}^{j = k } \frac{k!}{j!(k - j)!}\xi^j ( -\alpha_1)^{k - j } \right ) = \ ] ] @xmath69 this yields the first four @xmath66 presented in ( * ? ? ? * - 72 ) @xmath70 @xmath71 from equations [ eqsamplemean ] and [ eqmeanofsamplemean ] @xmath72^k ) = e([\frac{a_p - a_q}{n}]^k[n_p - np]^k ) = ( \frac{a_p - a_q}{n})^ke([n_p - np]^k)$ ] , where the second factor is the @xmath73th central moment of a binomial distribution . due to the factor @xmath74 the third standardized moment @xmath75 depends on the sign @xmath76 of the difference @xmath77 but not individual outcomes . the standardized moment @xmath78 does not dependent on the outcomes at all . formulas for the third and fourth beginning and central moments of a binomial distribution are found in @xcite . after accounting the first factor we get @xmath79 @xmath80}{n^3},\ ] ] @xmath81 @xmath82 the sample mean , its variance , third and fourth central moments , skewness , and excess kurtosis in equations [ eqsamplemean ] , [ eqvarianceofsamplemean ] - [ eqexcesskurtosisofsamplemean ] depend on time . in contrast , the mean of sample mean in equation [ eqmeanofsamplemean ] and its extreme values do not dependent on time . under other equal conditions , a _ @xmath45 achieved in a _ short _ time without an extra cost makes @xmath49 more symmetric , and certain for the author . it makes @xmath49 less deviating from @xmath7 and `` erases '' a difference between them . in problem 3 , @xmath83 is the least . _ the frequency of games and playing time are important for increasing confidence as long as they determine the actual number of games . _ in lotteries , the @xmath46 and @xmath47 can be compared with the life duration . mathematical expectations are at the heart of lotteries . the history of the _ big game _ began on august 31 , 1996 . in may 2002 the game was renamed to _ mega millions_. since then , the rules have been changing . currently , the initial jackpot is $ 15 million . to win , one has to guess five of 75 numbers in the _ game field _ and one of 15 numbers in the _ mega ball field _ , figure [ figlottery ] . a slip allows to play up to five $ 1 games . if no tickets win the jackpot , then it grows . table [ tbllottery ] , created using data of http://www.megamillions.com/history-of-the-game , illustrates the jackpot history . [ [ odds . ] ] odds . + + + + + let @xmath84 is the number of balls in a basket . there are @xmath85 red and @xmath86 black balls . ignoring the order , @xmath87 balls are taken out without replacement with equal chances . the probability to get @xmath88 red balls among the selected is given by the _ hypergeometric distribution _ * - 59 ) @xmath89 where @xmath90 is the binomial coefficient . here , @xmath91 and @xmath92 . + & & & & & & & & + continued from previous page + & & & & & & & & + + 1/3/2014 & 61 & 1 & 3/5/2010 & 134 & 1 & 6/16/2006 & 35 & 1 + 12/17/2013 & @xmath93 & 2 & 1/29/2010 & 144 & 1 & 6/2/2006 & 47 & 1 + 10/1/2013 & 189 & 1 & 12/22/2009 & 165 & 1 & 5/16/2006 & 94 & 1 + 7/26/2013 & 19 & 1 & 11/10/2009 & 77 & 2 & 4/18/2006 & 265 & 1 + 7/16/2013 & 20 & 1 & 10/16/2009 & 200 & 1 & 2/28/2006 & 270 & 1 + 7/5/2013 & 80 & 1 & 9/1/2009 & 12 & 1 & 1/6/2006 & 15 & 1 + 5/31/2013 & 30 & 1 & 8/28/2009 & 336 & 2 & 12/30/2005 & 88 & 1 + 5/17/2013 & 198 & 2 & 7/7/2009 & 133 & 1 & 11/29/2005 & 35 & 2 + 3/12/2013 & 41 & 1 & 5/29/2009 & 35 & 1 & 11/15/2005 & 315 & 1 + 2/19/2013 & 26 & 1 & 5/15/2009 & 38 & 2 & 9/16/2005 & 258 & 1 + 2/5/2013 & 19 & 1 & 5/1/2009 & 227 & 3 & 7/22/2005 & 170 & 1 + 1/25/2013 & 89 & 1 & 3/13/2009 & 26 & 1 & 6/3/2005 & 106 & 1 + 12/14/2012 & 35 & 1 & 3/3/2009 & 216 & 1 & 4/22/2005 & 208 & 1 + 11/27/2012 & 50 & 1 & 1/13/2009 & 22 & 1 & 3/1/2005 & 115 & 1 + 11/2/2012 & 33 & 1 & 1/2/2009 & 47 & 1 & 1/18/2005 & 131 & 1 + 10/16/2012 & 61 & 2 & 12/12/2008 & 207 & 1 & 12/3/2004 & 25 & 1 + 9/18/2012 & 14 & 1 & 10/21/2008 & 42 & 1 & 11/19/2004 & 149 & 1 + 9/11/2012 & 120 & 1 & 10/3/2008 & 42 & 1 & 10/1/2004 & 106 & 1 + 7/27/2012 & 52 & 1 & 9/16/2008 & 15 & 1 & 8/20/2004 & 52 & 1 + 7/3/2012 & 85 & 1 & 9/9/2008 & 24 & 1 & 7/27/2004 & 10 & 1 + 5/29/2012 & 32 & 1 & 8/29/2008 & 133 & 1 & 7/23/2004 & 47 & 1 + 5/15/2012 & 25 & 1 & 7/22/2008 & 126 & 1 & 7/2/2004 & @xmath94 & 1 + 5/4/2012 & 118 & 1 & 6/13/2008 & 57 & 1 & 5/7/2004 & 67 & 1 + 3/30/2012 & @xmath95 & 3 & 5/23/2008 & 17 & 2 & 4/9/2004 & 109 & 1 + 1/24/2012 & 72 & 1 & 5/16/2008 & 196 & 1 & 3/2/2004 & 21 & 1 + 12/27/2011 & 208 & 1 & 4/1/2008 & 136 & 1 & 2/20/2004 & 239 & 1 + 11/1/2011 & 78 & 1 & 2/22/2008 & 275 & 1 & 12/30/2003 & 162 & 1 + 9/30/2011 & 113 & 2 & 1/1/2008 & 33 & 1 & 11/11/2003 & 70 & 3 + 8/19/2011 & 32 & 1 & 12/18/2007 & 163 & 2 & 10/7/2003 & 12 & 1 + 8/5/2011 & 99 & 1 & 11/2/2007 & 75 & 1 & 9/30/2003 & 150 & 1 + 7/1/2011 & 107 & 1 & 10/5/2007 & 27 & 1 & 8/8/2003 & 50 & 1 + 5/27/2011 & 35 & 1 & 9/25/2007 & 12 & 1 & 7/11/2003 & 34 & 1 + 5/13/2011 & 27 & 1 & 9/21/2007 & 61 & 1 & 6/20/2003 & 183 & 1 + 5/3/2011 & 51 & 1 & 8/31/2007 & 330 & 4 & 4/25/2003 & 46 & 1 + 4/15/2011 & 72 & 1 & 7/6/2007 & 128 & 1 & 3/28/2003 & 20 & 1 + 3/25/2011 & 319 & 1 & 5/29/2007 & 44 & 1 & 3/14/2003 & 43 & 1 + 2/1/2011 & 93 & 2 & 5/11/2007 & 113 & 1 & 2/18/2003 & 12 & 1 + 1/4/2011 & @xmath96 & 2 & 4/6/2007 & 105 & 1 & 2/11/2003 & 128 & 1 + 11/9/2010 & 25 & 1 & 3/6/2007 & @xmath97 & 2 & 12/24/2002 & 68 & 1 + 10/29/2010 & 141 & 1 & 1/9/2007 & 125 & 1 & 11/19/2002 & 16 & 1 + 9/17/2010 & 54 & 1 & 12/1/2006 & 40 & 1 & 11/8/2002 & 93 & 1 + 8/27/2010 & 135 & 1 & 11/14/2006 & 75 & 1 & 9/27/2002 & 37 & 1 + 7/16/2010 & 64 & 1 & 10/17/2006 & 55 & 1 & 9/6/2002 & 17 & 1 + 6/22/2010 & 26 & 1 & 9/26/2006 & 15 & 1 & 8/27/2002 & 108 & 1 + 6/11/2010 & 36 & 1 & 9/19/2006 & 12 & 1 & 7/16/2002 & 165 & 1 + 5/28/2010 & 12 & 1 & 9/15/2006 & 163 & 1 & 5/24/2002 & 12 & 1 + 5/25/2010 & 64 & 1 & 8/1/2006 & 31 & 1 & 5/17/2002 & 28 & 1 + 5/4/2010 & 266 & 1 & 7/18/2006 & 49 & 1 & & & + 3/12/2010 & 20 & 1 & 6/27/2006 & 24 & 1 & & & if there are two baskets represented by @xmath98 , @xmath99 , @xmath100 and @xmath101 , @xmath102 , @xmath103 and @xmath104 and @xmath105 balls are taken out of each , then the probability to get @xmath106 and @xmath107 red balls distinguished between baskets is the product @xmath108 . for @xmath109 `` independent '' baskets the reciprocal @xmath110 is equal to @xmath111 for a particular case @xmath112 the formula @xmath113 describes reciprocal odds in the current mega millions and _ power ball _ lotteries . the red and black balls are winning and losing numbers . the two baskets are the game and mega ball and game and _ powerball _ fields on slips . the word _ field _ justifies the denominations @xmath114 and @xmath115 . for mega millions @xmath116 , @xmath117 , @xmath118 , @xmath119 , @xmath120 . the jackpot s @xmath121 . this coincides with the entry in `` how to win mega millions '' , figure [ figlottery ] . the values @xmath122 , @xmath123 , @xmath124 , @xmath125 , @xmath126 , @xmath127 , @xmath128 , @xmath129 match remaining entries . the sum of the nine reciprocals of @xmath130 is equal to @xmath131 . this is the probability of winning something per ticket with one chance in 14.707698 . [ [ numbers - of - tickets . ] ] numbers of tickets . + + + + + + + + + + + + + + + + + + + for drawing on january 31 , 2014 , 1,440,661 winning tickets were announced . the jackpot was not won . if each ticket would be equally probable , then their total number would be @xmath132 including @xmath133 losing tickets . [ [ expectations . ] ] expectations . + + + + + + + + + + + + + all awards except the _ effective jackpot _ @xmath134 , the odds given by equation [ eqlottery ] , and ticket price @xmath135 are fixed . the mathematical expectation of the _ profit and loss _ , @xmath136 , of the game without _ megaplier _ and tax is equal to ( see figure [ figlottery ] ) @xmath137 @xmath138 the jackpot and large prizes $ 1,000,000 and $ 5,000 are taxable . with 40 percent tax the last two numbers decrease to $ 600,000 and $ 3,000 yielding @xmath139 it would be positive for @xmath140 . for powerball @xmath116 , @xmath141 , @xmath118 , @xmath142 , @xmath120 , @xmath143 . using equation [ eqlottery ] we get @xmath144 , @xmath145 , @xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 , @xmath151 , @xmath152 . figure of a powerball slip is omitted . the expectation without _ powerplay _ and tax is @xmath153 @xmath154 decreasing $ 1,000,000 and $ 10,000 to $ 600,000 and $ 6,000 after 40% tax yields @xmath155 it would be positive for @xmath156 . the effective jackpot @xmath134 should differ from the quoted @xmath157 . the jackpot in table [ tbllottery ] assumes annuity payments . it has to be converted to a cash payout for declining annuity . the latter splits between all tickets winning jackpot and should be adjusted . finally , the adjusted cash payout is a subject for tax . the first step based on the _ ordinary annuity _ formula @xcite yields @xmath158 @xmath159 for 30 years and rates of 2 , 3 , or 5 percent , the coefficient in front of @xmath157 is equal to 0.7395 , 0.6306 , or 0.4515 . the second step has to take into consideration the number of purchased tickets @xmath160 . if each ticket wins jackpot equally likely with probability @xmath161 and they are filled independently , then the probability of @xmath162 winning coincidences is binomial @xmath163 . each winner gets @xmath164 . for @xmath165 the payoff is zero . the mathematical expectation of the non - zero payoffs is @xmath166 the probability that it is won is equal to @xmath167 for @xmath160 equal to 10,000,000 and 100,000,000 it is 0.03788983372734210 and 0.3204083454349182 . the weighted jackpot yielding the payoff expectation is @xmath168 the third step is to apply the tax percent and get @xmath169 . finally , the mathematical expectation in mega millions with @xmath170 is @xmath171 under other equal conditions @xmath172 is the function of the announced jackpot @xmath157 and the number of purchased tickets @xmath160 , figure [ figlotteryme ] . the former is known prior a drawing . the latter can be estimated multiplying by three the difference between the current and previous , if it was not won , cash jackpot amounts @xmath158 . this assumes that one third of the collected money was used for increasing the cash jackpot and the current number of tickets is not less . and purchased number of tickets @xmath160 , @xmath173 , given by equation [ eqlotteryme ] with @xmath174 and @xmath175 . two projections a and b and the plane @xmath176 help to see better the range of parameters making the expectation positive . plots are done using maple 10 from maplesoft.,width=491 ] with the initial jackpot $ 15,000,000 the mathematical expectation of profits and losses in mega millions is negative . equation [ eqlotteryme ] and figure [ figlotteryme ] indicate that it can be positive , even , after declining the annuity option , splitting jackpot between tickets winning it , and taxes . since 2002 until 2005 it was needed to guess 5 of 52 and 1 of 52 numbers . since 2005 until 2013 - 5 of 56 and 1 of 46 numbers . the eight prizes after jackpot were $ 250,000 , $ 10,000 , $ 150 , $ 150 , $ 7 , $ 10 , $ 3 , and $ 2 . equations for the mathematical expectations of profit and loss after 2002 , 2005 , and 2013 are @xmath177 , @xmath178 , and @xmath179 . with @xmath180 , the expectations are positive for @xmath181 million , @xmath182 million , and @xmath183 million . table [ tbllottery ] marks by stars five such jackpot candidates occurred in 12 years . let us assume that a gambler decides to play only , when the expectation becomes positive . _ does it mean that the game under such a favorable condition is for certain ? _ [ [ game - for - sure ] ] game for sure ? + + + + + + + + + + + + + + positive @xmath172 does not make the `` favorable '' drawings a _ game for sure_. during 80 years of an adult life a gambler may participate in such infrequent events @xmath184 times . a gambler betting twice every week can make @xmath185 attempts . both numbers are insignificant comparing with the odds 1 : 258,890,850 . the discrete nature of gains and losses implies either a chain of -$1 ( ignore smaller prizes ) or one win making previous losses unimportant . _ there are no drawings returning the favorable expected values counted by a few dozen of cents . the potential extreme gain and actual low cost are more important than the mathematical expectation for a decision to play_. let one in a favorable case with $ 552,000,000 jackpot agrees to get annuity , pays for the tickets $ 258,890,850 , and technically fills them by unique combinations of numbers . this ensures winning the jackpot . we have seen that the probability of multiple wins of the jackpot does not reduce much the mean number of winning tickets . the coefficient 0.8 - 0.9 increases 1 to @xmath186 . tax 40% takes more . but this is the mean number of winning tickets . in reality it is either 1 , 2 , or 3 ... while the tax is fixed ( not for annuity payments ) , the number of winning tickets is random . _ just only two else claim a winning ticket , the favorable attempt becomes a financial disaster because of the sharing . again , not only the mathematical expectation of the number of winning tickets but a probability that the actual number will deviate from it look important_. problem 3 was changed to 3 @xcite . now , prospect a is losing 4,000 with the probability 0.8 and prospect b is losing 3,000 for sure . in a the mathematical expectation @xmath187 is less than in b -3,000 . however , 92 percent of 95 respondents choose a. kahneman and tversky label this pattern , switching the signs of outcomes , the _ reflection effect_. respondents attempt to avoid a loss with the small probability 0.2 . again , the author `` feels '' that losing the sample mean @xmath188 characterized by the standard deviation @xmath189 in a _ big guaranteed _ number of games @xmath48 without extra cost makes a worse than b. properties of variables for problem 3 are @xmath190 , @xmath191 , @xmath192 , @xmath193 , @xmath194 , _ standard deviation _ @xmath195 , @xmath196 , @xmath197 , @xmath198 , @xmath199 , @xmath200 , and @xmath201 , @xmath202 , @xmath203 , @xmath204 , @xmath205 , @xmath206 , @xmath207 , @xmath208 , @xmath209 and @xmath210 are undefined , @xmath211 . mathematical expectations , third central moments , skewness of the random variables in problems 3a and 3a are equal by absolute value and have opposite sign . mathematical expectations of the deterministic variables in problems 3b and 3b are equal by absolute value and have opposite sign . the author has `` discovered '' @xcite that the legendary speculator jesse livermore @xcite , @xcite , @xcite , @xcite guessed the certainty and reflection effects @xcite . indeed , ( * ? ? ? * - 12 ) : _ `` ... you buy a stock at $ 30.00 . the next day it has a quick run - up to $ 32.00 or $ 32.50 . you immediately become fearful that if you do nt take the profit , the next day you may see it fade away - so out you go with a small profit , when that is very time you should entertain all the hope in the world . '' _ this is a certainty effect . further , ( * ? ? ? * - 13 ) : _ `` on the other hand , suppose you buy a stock at $ 30.00 , and the next day it goes to $ 28.00 , showing a two - point loss . you would not be fearful that the next day would possibly see a three - point loss or more . no , you would regard it merely as a temporary reaction , feeling certain that the next day it would recover its loss . ... that is when you should be fearful , because if you do not get out , you might be forced to take a much greater loss later on . '' _ this is a reflection effect . livermore believes that a human being `` injects a hope and fear into the business of speculation '' and `` is apt to get the two confused and in reverse positions '' . his 40 years trading wisdom is @xcite _ `` profits always take care of themselves , but losses never do''_. his intuition was trained by winning and losing several fortunes . he played the game many times feeling but not measuring the odds behind his advices . empirical distributions of prices , their increments , and waiting times between transactions vary @xcite . this strengthens uncertainty of profits and losses presenting a trading opportunity as the last one . under such conditions following his advices is difficult psychologically and increasing @xmath48 does not guarantee a fast convergence of @xmath49 . the author considered _ random sample means _ as universal awards in prospects and the st . petersburg game independently on khinchin @xcite and only after that found his `` forgotten '' paper @xcite . khinchin concentrates on mathematical properties of random geometric and arithmetic means , which can explain the `` paradox '' . he reviews psychology @xcite , @xcite ) : _ `` let us notice only that in this case , of course , no speech may go about any mathematical paradox but at most about that the mathematical expectation is not always adequate to those worldly - psychological representations , which it is commonly connected to . in the case of the petersburg game , it is often pointed to that petr in his expectation of winning , naturally , orients not on the mathematical expectation of winning in a particular game , which is difficult to account psychologically , but on some average winning during big number of games . such understanding of psychological prerequisites of the ' ' paradox `` puts in front of us a certain mathematical task , which can be formulated as follows : find such an estimate of the mean winning of petr during a big number of games , that its probability would go to unit with infinite increasing the number of games . however , it makes sense to say , that the task will get a quite determined sense only after a certain notion of the mean winning will be exactly defined . in the current note , we shall consider in details the set problem in two of the most simple ( and also the most important ) cases , namely in assumption that the mean winning is defined as the geometric and arithmetic mean of particular games''_. khinchin is indifferent , if @xmath45 is achievable . if `` yes '' , then his two theorems work . for the author , the sample mean @xmath49 is an award in prospects with @xmath212 . the case @xmath48 is generic . the @xmath49 is a universal random award in the st . petersburg game , prospects , and variations 1 , 2 , and 4 . for @xmath212 it coincides with underlying random variables . in a private conversation with timur misirpashaev the author discussed _ credit nuances _ of the st . petersburg game and believes that the topic `` correlates '' with samuelson s _ bankruptcy _ consideration @xcite and the following buffon s comment @xcite : _ `` ... all money on the earth is not enough to accomplish this [ vs : to pay the win ] , if the game stops on 40th trial , because it will require 1024 times more money than there exists in the entire kingdom of france''_. during the discussion , the author has proposed to pay $ 3 for the right to play one time and $ 18 to play 10 times , under a condition that in both cases the third party reserves the deposit $ 128 to pay a win , if such will be drawn . this proposal was done after experimenting with the c++ program khinchin.cpp ( * ? ? ? * appendix ) . making decisions for the four variations and prospects depends on probabilistic properties of @xmath49 . petersburg game adds the credit uncertainty of paying the win . the creditworthiness is not in scope of this article . if one choses between random variables @xmath213 and @xmath214 , then a measure affecting the decision should account their properties . the sums and mean sums of i.i.d . variables obey the laplace theorem , the law of big numbers , and the law of the iterated logarithm @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . yurii vasilevich prokhorov presents estimates for the law of big numbers @xcite @xmath215 where @xmath216 is the number of successes in @xmath217 bernoulli @xmath218 trials and @xmath219 is the probability of inequality . figure [ figprokhorov ] plots decimal logarithm of the right expression . on @xmath220 and @xmath221 and two projections . plots are done using maple 10 from maplesoft , width=340 ] in problem 3 , @xmath222 is constant and @xmath213 is random with @xmath212 . in variations 1 , 2 , and 4 with increasing @xmath45 , @xmath223 . if @xmath213 does not drop below @xmath224 , then it becomes attractive . such a drop occurs , if @xmath225 . thus , the threshold value is @xmath226 and @xmath227 . for @xmath228 and @xmath229 , we get @xmath230 and @xmath231 , if @xmath232 and @xmath233 . it is assumed that @xmath48 games can be played and there is no additional cost comparing with playing one game . in probability theory , a _ full system of events _ @xmath234 , @xmath235 , ... , @xmath236 is a group of events , where in each trial one and only one of them comes . the events , specified together with their probabilities @xmath237 , @xmath238 , ... , @xmath239 ( @xmath240 , @xmath241 ) , are named in @xcite a _ finite schema _ @xmath242 the shannon s entropy of the finite schema is the function @xmath243 where @xmath244 for @xmath245 . it is a reasonable measure of uncertainty of the finite schema . for a two - point random variable we get @xmath246 from equation [ eqbinaryvariableentropy ] . for two _ independent _ finite scheme @xmath247 the probability @xmath248 of the joint events @xmath249 and @xmath250 is equal to @xmath251 ( @xmath252 , @xmath253 ) . the set of the joint events @xmath254 form a new schema @xmath255 - the _ union of scheme _ @xmath256 and @xmath257 . in this case , @xmath258 . if for two scheme , representing two i.i.d . two - point variables , the four events @xmath259 , @xmath260 , @xmath261 , @xmath262 are distinguished , then the entropy is equal to @xmath263 for the sum of @xmath45 i.i.d . two - point variables , where @xmath264 events are distinguished , the entropy grows linearly with @xmath45 and is equal to @xmath265 . however , from profit or loss point of view , the events @xmath260 and @xmath261 are identical yielding the schema for sample mean money outcomes @xmath266 its entropy is less than or equal to @xmath267 @xmath268 because @xmath269 . from equation [ eqsamplemean ] , for @xmath54 we deal with a binomial distribution of @xmath270 outcomes . the schema for the sum of @xmath45 terms is @xmath271 the schema for the mean sum of @xmath45 terms is @xmath272 for both scheme , where random @xmath273 , @xmath274 @xmath275 is positive for @xmath276 and @xmath277 . the random sum @xmath278 and sample mean @xmath49 are linear functions of the binomial random variable @xmath50 : @xmath279 and @xmath280 . let us notice , that for a normal distribution with mean @xmath281 and variance @xmath282 the _ differential _ ( or continuous ) entropy is equal to @xmath283dx=\\ \log_2(\sqrt{2\pi\mu_2 } ) + \frac{\log_2(e)}{2 } = \frac{\log_2(2\pi e \mu_2)}{2}. \end{split}\ ] ] @xmath284 does not depend on @xmath281 and can be negative for tiny variance @xmath282 . from limit theorems , for @xmath285 binomial distribution approaches normal with mean @xmath286 and variance @xmath287 . there is temptation to approximate the binomial entropy by @xmath288 , which tends to infinity for @xmath276 and @xmath285 . similarly , a sample mean distribution approaches normal with mean @xmath289 and variance @xmath290 . however , substituting variance into the normal distribution entropy expression yields @xmath291 , which becomes negative for @xmath292 . this contradicts to non - negative value @xmath275 and emphasizes the difference between discrete and differential entropy . let us denote binomial probability @xmath293 and recollect that @xmath294 , @xmath295 , @xmath296 , @xmath297 , @xmath298 , and @xmath299 . then , @xmath300 since @xmath301 , @xmath302 . next , @xmath303 the stirling s approximation @xmath304 and @xmath305 has been applied . we get @xmath306 where evaluation of large factorials is avoided . while the number of outcomes @xmath307 increases linearly with @xmath45 , the change of corresponding binomial probabilities causes non - linear and slow growth of @xmath308 . the role of time here is in @xmath309 . the @xmath308 is common for the sample sum and mean sum . the range of the former @xmath310 $ ] is getting wider . the range of the latter @xmath311 $ ] remains intact with uniform increasing density of outcomes . the common for both sums discrete entropy , being a function of only probabilities , does not reflect these nuances . _ increasing the discrete shannon s entropy widely associated with uncertainty does not tell that the mean sum gets lower variance . however , it is the latter decreasing parameter causes author s impression that certainty of an award increases_. [ [ discrete - vs .- differential - entropy . ] ] discrete vs. differential entropy . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in contrast with discrete entropy of a binomial distribution , the differential entropy of a normal one , approached by both sums for @xmath285 , better corresponds to a common point of view that a higher entropy associates with higher uncertainty . this happens because the expression for @xmath284 contains @xmath282 under logarithm . for the sample mean sum the entropy at the gaussian limit grows by absolute value but with negative sign - decreases . the negative sign creates inconvenience . accordingly , we shall use entropy for studying fractions of responds carefully and restrictedly . let there exists a function that maps a _ random _ prospect to _ deterministic _ money . if prospects and equivalents could be exchanged at any time without cost , then respondents choosing between two prospects , would _ rationally _ select the greater value . the fractions would be zero and one and for equal values any . divergence of opinion means that the function that eliminates uncertainty was unknown . in contrast , a _ risk neutral portfolio _ of a stock and call options on it illustrates a possibility to eliminate randomness @xcite . _ we shall look for a different function that maps properties of random prospects to fractions of respondents , _ which can be verified using questionnaires like in @xcite and @xcite . selecting a profit opportunity in problem 3 , majority _ excludes a chance to get nothing_. choosing between losing scenarios in problem 3 , respondents want to _ avoid a loss completely_. if in prospects 3b and 3a the sure , @xmath312 , and uncertain , @xmath313 , gains would be 1 but not 3,000 and a mega millions jackpot but not 4,000 , then the author would favor 3a . set the uncertain , @xmath314 , loss in prospect 3a substantially greater by absolute value than -4,000 ( home , mega millions jackpot , ... _ life _ ) and 3b with the sure , -3,000 , loss looks as the lesser of evils . the fractions of voters depend on _ known properties _ of awards or losses and poorly known nature of human beings . _ can fractions of respondents be derived from properties of prospects ? _ several prospects @xcite , @xcite are based on two outcomes @xmath3 and @xmath4 and corresponding probabilities @xmath315 and @xmath6 . these are two - point variables , which the author considers as random sample means @xmath49 , mean sums of the random i.i.d . two - point variables , where the number of summands @xmath212 . this generalization includes cases with @xmath316 . the four _ independent _ quantities @xmath317 fully characterize @xmath49 . without losing generality @xmath318 . prospect 3b is the deterministic award @xmath49 with @xmath319 . prospect 3a is the random award @xmath49 with @xmath320 . prospect 3b is the deterministic loss with @xmath321 . prospect 3a is the random loss with @xmath322 . the two - point variable is deterministic , if @xmath323 and/or @xmath324 or @xmath325 . empirical fractions of respondents choosing between @xmath326 and @xmath327 summing to @xmath328 should be approximated by model functions @xmath329 and @xmath330 . ideal is a single function @xmath331 returning a fraction of respondents depending on the order of arguments . then , @xmath332 is a solution of the functional equation @xmath333 @xmath334 for some arguments , fractions of respondents make _ rational sense_. each discrete random variable with finite number of finite outcomes has a minimum and maximum value . intervals determined by the extreme values of two variables can be disjoined , adjacent , or overlapping . an interval and variable is a point and constant , if @xmath337 . the symbols @xmath338 , @xmath339 , @xmath340 , @xmath341 form @xmath342 permutations . each has three places to be filled by @xmath343 or @xmath344 like @xmath345 . this yields eight variations for each permutation and creates @xmath346 combinations . [ [ canonical - combinations . ] ] canonical combinations . + + + + + + + + + + + + + + + + + + + + + + + some combinations are equivalent : @xmath345 and @xmath347 . symbols , _ connected by equality signs _ , were sorted in the order @xmath348 . this translates @xmath347 to @xmath345 and @xmath349 to @xmath350 . the images were named _ canonical combinations _ and only one repetition was selected . the author has written a c++ program and found 75 unique canonical combinations with 26 satisfying numerical conventions @xmath351 and @xmath352 , table [ tbltheoryf1 ] . in 1 , 2 , 9 , 10 , 17 , 18 , 25 , 26 the worst outcome of one prospect is better than the best outcome of another . this advantage depends on neither probabilities nor positive numbers of trials . _ rational respondents will predictably chose a prospect with greater outcomes , if the intervals are disjoined_. in 3 , 4 , 11 , 16 , 23 , 24 one prospect is not worse than another and the fractions of respondents are either 1 or 0 . for these combinations the fractions can be undefined @xmath353 , if probabilities make both variables equal constants . in @xmath354 a two - point prospect ( 2 ) is better than ( 1 ) , if @xmath355 , and undefined otherwise . _ the fractions _ 0 , 1 , @xmath353 _ for adjacent intervals are predictable_. in 12 both prospects are equal constants : fractions are predictably undefined . in 5 , 6 , 7 , 8 , 13 , 14 , 15 , 19 , 20 , 21 , 22 the intervals overlap and fractions are not obvious . _ if the intervals of two variables determined by the maximum and minimum outcomes are disjoined , then the fractions of respondents are deduced from a common rational sense , where probabilities and positive numbers of trials are irrelevant for a decision making . if the intervals are adjacent , then the fractions are deduced from rational sense or undefined for equal deterministic variables . if the intervals overlap , then probabilities and numbers of trials matter_. + & & & & + continued from previous page + & & & & + + 1 & @xmath348 & 0 & 1 & ( 1 ) < ( 2 ) + 2 & @xmath356 & 0 & 1 & ( 1 ) < ( 2 ) ; constant @xmath327 + 3 & @xmath357 & @xmath358 & @xmath359 & @xmath360 ; adjacent + 4 & @xmath361 & @xmath358 & @xmath359 & @xmath360 ; constant @xmath327 ; adjacent + 5 & @xmath362 & ? & ? & overlapping + 6 & @xmath363 & ? & ? & overlapping + 7 & @xmath364 & ? & ? & overlapping + 8 & @xmath365 & ? & ? & constant @xmath327 ; overlapping + 9 & @xmath345 & 0 & 1 & ( 1 ) < ( 2 ) ; constant @xmath326 + 10 & @xmath366 & 0 & 1 & ( 1 ) < ( 2 ) ; constant @xmath326 , @xmath327 + 11 & @xmath350 & @xmath358 & @xmath359 & @xmath360 ; constant @xmath326 ; adjacent + 12 & @xmath367 & @xmath353 & @xmath353 & constant @xmath368 + 13 & @xmath369 & ? & ? & overlapping + 14 & @xmath370 & ? & ? & overlapping + 15 & @xmath371 & ? & ? & overlapping + 16 & @xmath372 & @xmath359 & @xmath358 & @xmath373 ; constant @xmath327 ; adjacent + 17 & @xmath374 & 1 & 0 & ( 2 ) < ( 1 ) + 18 & @xmath375 & 1 & 0 & ( 2 ) < ( 1 ) ; constant @xmath326 + 19 & @xmath376 & ? & ? & overlapping + 20 & @xmath377 & ? & ? & overlapping + 21 & @xmath378 & ? & ? & overlapping + 22 & @xmath379 & ? & ? & constant @xmath326 ; overlapping + 23 & @xmath380 & @xmath359 & @xmath358 & @xmath373 ; constant @xmath326 ; adjacent + 24 & @xmath381 & @xmath359 & @xmath358 & @xmath373 ; adjacent + 25 & @xmath382 & 1 & 0 & ( 2 ) < ( 1 ) ; constant @xmath327 + 26 & @xmath383 & 1 & 0 & ( 2 ) < ( 1 ) ; constant @xmath384 swapping indexes 1 and 2 does not change validity of a canonical combination and may require conventional sorting . following equation [ eqfunctional ] this swaps values of fractions of respondents . swapping transforms @xmath385 , @xmath386 , @xmath387 , @xmath388 , @xmath389 , @xmath390 , @xmath391 , @xmath392 , @xmath393 , @xmath394 , @xmath395 , @xmath396 . due to sorting @xmath397 , and @xmath398 . 12 and 14 represent identical intervals , where 12 is a point . due to the symmetry we have to find @xmath331 only for 1 - 14 . the cases 1 , 2 , 3 , 4 , 9 , 10 , 11 , 12 are either predictable or undefined , axiomatic . _ the cases _ 5 , 6 , 7 , 8 , 13 , 14 _ require attention_. [ [ for - two - point - variables ] ] for two - point variables , + + + + + + + + + + + + + + + + + + + + + + + + @xmath399 and @xmath400 . probabilities @xmath401 and @xmath402 are independent . their combinations are in table [ tblprobabilitiesf1 ] . a variable is constant , if @xmath403 or 0 . _ if both variables are constant because of probabilities and/or equal outcomes , then the choice is predictable_. + & & + continued from previous page + & & + + 1 & @xmath404 , @xmath405 & constant @xmath406 , @xmath407 , @xmath408 + 2 & @xmath409 & constant @xmath406 , @xmath410 + 3 & @xmath404 , @xmath411 & constant @xmath406 , @xmath412 , @xmath408 + 4 & @xmath413 , @xmath405 & constant @xmath407 , @xmath414 + 5 & @xmath415 & @xmath416 + 6 & @xmath413 , @xmath411 & constant @xmath412 , @xmath414 + 7 & @xmath417 , @xmath405 & constant @xmath418 , @xmath407 , @xmath408 + 8 & @xmath417 , @xmath419 & constant @xmath418 , @xmath410 + 9 & @xmath417 , @xmath411 & constant @xmath418 , @xmath412 , @xmath408 + 10 & @xmath420 & + 11 & @xmath421 & + combinations 1 , 3 , 7 , 9 from table [ tblprobabilitiesf1 ] make both prospects deterministic . combinations 2 , 4 , 6 , 8 make one prospect deterministic . the alternative prospect can become deterministic due to outcomes . example is @xmath365 , table [ tbltheoryf1](8 ) , combined with @xmath409 , table [ tblprobabilitiesf1](2 ) : prospect ( 2 ) is constant because @xmath422 and prospect ( 1 ) is constant because @xmath404 and @xmath423 . this is true @xmath424 . therefore , @xmath425 , @xmath426 . cartesian product table [ tbltheoryf1](5 , 6 , 7 , 8 , 13 , 14 ) @xmath427 table [ tblprobabilitiesf1](2 , 4 , 5 , 6 , 8 , 10 , 11 ) = 42 requires further attention , table [ tblcartesianf1 ] . + & & & & & & & + continued from previous page + & & & & & & & + + 1 & 5 & @xmath428 & 2 & @xmath409 & 0 & 1 & ( 1 ) < ( 2 ) + 2 & 5 & @xmath428 & 4 & @xmath413 , @xmath405 & ? & ? & + 3 & 5 & @xmath428 & 5 & @xmath415 & 0 & 1 & + 4 & 5 & @xmath428 & 6 & @xmath413 , @xmath411 & 0 & 1 & ( 1 ) < ( 2 ) + 5 & 5 & @xmath428 & 8 & @xmath417 , @xmath419 & ? & ? & + 6 & 5 & @xmath428 & 10 & @xmath420 & ? & ? & + 7 & 5 & @xmath428 & 11 & @xmath421 & ? & ? & + 8 & 6 & @xmath429 & 2 & @xmath409 & 0 & 1 & ( 1 ) < ( 2 ) + 9 & 6 & @xmath429 & 4 & @xmath413 , @xmath405 & ? & ? & + 10 & 6 & @xmath429 & 5 & @xmath415 & 0 & 1 & + 11 & 6 & @xmath429 & 6 & @xmath413 , @xmath411 & 0 & 1 & @xmath360 + 12 & 6 & @xmath429 & 8 & @xmath417 , @xmath419 & 1 & 0 & @xmath373 + 13 & 6 & @xmath429 & 10 & @xmath420 & 0 & 1 & + 14 & 6 & @xmath429 & 11 & @xmath421 & ? & ? & + 15 & 7 & @xmath430 & 2 & @xmath409 & 0 & 1 & ( 1 ) < ( 2 ) + 16 & 7 & @xmath430 & 4 & @xmath413 , @xmath405 & ? & ? & + 17 & 7 & @xmath430 & 5 & @xmath415 & ? & ? & + 18 & 7 & @xmath430 & 6 & @xmath413 , @xmath411 & ? & ? & + 19 & 7 & @xmath430 & 8 & @xmath417 , @xmath419 & 1 & 0 & @xmath431 + 20 & 7 & @xmath430 & 10 & @xmath420 & ? & ? & + 21 & 7 & @xmath430 & 11 & @xmath421 & ? & ? & + 22 & 8 & @xmath432 & 2 & @xmath409 & 0 & 1 & ( 1 ) < ( 2 ) + 23 & 8 & @xmath432 & 4 & @xmath413 , @xmath405 & ? & ? & + 24 & 8 & @xmath432 & 5 & @xmath415 & ? & ? & + 25 & 8 & @xmath432 & 6 & @xmath413 , @xmath411 & ? & ? & + 26 & 8 & @xmath432 & 8 & @xmath417 , @xmath419 & 1 & 0 & @xmath431 + 27 & 8 & @xmath432 & 10 & @xmath420 & ? & ? & + 28 & 8 & @xmath432 & 11 & @xmath421 & ? & ? & + 29 & 13 & @xmath433 & 2 & @xmath409 & 0 & 1 & @xmath360 + 30 & 13 & @xmath433 & 4 & @xmath413 , @xmath405 & 1 & 0 & @xmath373 + 31 & 13 & @xmath433 & 5 & @xmath415 & 0 & 1 & @xmath360 + 32 & 13 & @xmath433 & 6 & @xmath413 , @xmath411 & 0 & 1 & ( 1 ) < ( 2 ) + 33 & 13 & @xmath433 & 8 & @xmath417 , @xmath419 & ? & ? & + 34 & 13 & @xmath433 & 10 & @xmath420 & 0 & 1 & + 35 & 13 & @xmath433 & 11 & @xmath421 & ? & ? & + 36 & 14 & @xmath434 & 2 & @xmath409 & 0 & 1 & @xmath360 + 37 & 14 & @xmath434 & 4 & @xmath413 , @xmath405 & 1 & 0 & @xmath373 + 38 & 14 & @xmath434 & 5 & @xmath415 & @xmath353 & @xmath353 & ( 1 ) = ( 2 ) + 39 & 14 & @xmath434 & 6 & @xmath413 , @xmath411 & 0 & 1 & @xmath360 + 40 & 14 & @xmath434 & 8 & @xmath417 , @xmath419 & 1 & 0 & @xmath373 + 41 & 14 & @xmath434 & 10 & @xmath420 & 0 & 1 & + 42 & 14 & @xmath434 & 11 & @xmath421 & 1 & 0 & + if arbitrary large number of trials costs nothing and possible , then due to decreasing variance of @xmath49 following from equation [ eqvarianceofsamplemean ] the author thinks that @xmath435 @xmath436 @xmath437 the author believes that _ respondents unanimously _ select the _ sure _ 3,000.01 instead of 3,000.00 and -4,000.00 instead of -4,000.01 . for equal sure amounts the choice is undefined . these are the combinations table [ tbltheoryf1](3 ) + table [ tblprobabilitiesf1](7 ) , [ tbltheoryf1](4 ) + [ tblprobabilitiesf1](7 , 8 , or 9 ) , [ tbltheoryf1](6 ) + [ tblprobabilitiesf1](9 ) , [ tbltheoryf1](11 ) + [ tblprobabilitiesf1](1 , 4 , or 7 ) , [ tbltheoryf1](12 ) + [ tblprobabilitiesf1](any ) , [ tbltheoryf1](13 ) + [ tblprobabilitiesf1](1 ) , [ tbltheoryf1](14 ) + [ tblprobabilitiesf1](1 , or 9 ) , [ tbltheoryf1](15 ) + [ tblprobabilitiesf1](1 ) , [ tbltheoryf1](16 ) + [ tblprobabilitiesf1](1 , 2 , or 3 ) , [ tbltheoryf1](20 ) + [ tblprobabilitiesf1](9 ) , [ tbltheoryf1](23 ) + [ tblprobabilitiesf1](3 , 6 , or 9 ) , [ tbltheoryf1](24 ) + [ tblprobabilitiesf1](3 ) . in general , 0 , 1 , and @xmath353 postulated for @xmath335 and/or @xmath336 in tables [ tbltheoryf1 ] and [ tblcartesianf1 ] indicate combinations of arguments and values , which , together with equations [ eqaxiom1 ] - [ eqaxiom3 ] , are candidates to _ axioms_. kahneman and tversky present experimental @xmath438 and @xmath439 for @xmath440 @xcite . together with blue diamonds , the experimental points 2 , 10 , 12 , and 14 from table [ tblf1 ] . green lines and surface correspond to the postulates . red points indicate ambiguity . @xmath441.,width=245 ] the postulated and experimental points @xmath438 are applied to sketch @xmath335 . @xmath335 and @xmath336 are the functions of at least eight arguments . we can plot 3d projections of the eight dimensional space choosing suitable coordinates , figure [ figf1 ] . robinson crusoe took to his desert island from the sunk ship a gun instead of coins . a donation can be decided to be greater . respondents can be mentally ill . other reasons can be invented with an intention to invalidate the postulates : table [ tbltheoryf1 ] ( 1 - 4 , 9 - 12 , 16 - 18 , 23 - 26 ) ; table [ tblcartesianf1 ] ( 1 , 3 , 4 , 8 , 10 - 13 , 15 , 19 , 22 , 26 , 29 - 32 , 34 , 36 - 42 ) ; equations [ eqaxiom1 ] - [ eqaxiom3 ] . the paper omits such exercises . two - point random variables are the essence of a few prospects reviewed here . this paper suggests an extension , sample means as awards , replacing these variables with their mean sums obtained in sequences of independent trials . the properties of both are well studied theoretically and supported experimentally including computer simulations . while some properties of sequential bernoulli trials such as the _ law of arcsine _ are paradoxical and `` contradict to common sense '' * - 97 ) , the variables are fundamental `` simple '' candidates to investigate how _ randomness _ of prospects influences on the fractions of respondents selecting between the prospects . with this proposal @xmath442 depends on eight parameters . fixing two coordinates for each dimension , a rough estimate of @xmath335 , requires @xmath443 points of an experimental plan . table [ tblf1 ] collects estimates @xmath438 and _ median cash equivalents _ found in literature . + & & & & & & & & & & + continued from previous page + & & & & & & & & & & + + 1 & 2500 & 0 & 0.33 & 2400 & 0 & 0.34 & 0.83 & 72 & @xcite & 15/10/ * + 2 & 4000 & 0 & 0.80 & 3000 & 3000 & 1 & 0.20 & 95 & @xcite & 8/6/25 + 3 & 4000 & 0 & 0.20 & 3000 & 0 & 0.25 & 0.65 & 95 & @xcite & 15/10/ * + 4 & 6000 & 0 & 0.45 & 3000 & 0 & 0.90 & 0.14 & 66 & @xcite & 15/10/ * + 5 & 6000 & 0 & 0.001 & 3000 & 0 & 0.002 & 0.73 & 66 & @xcite & 15/10/ * + 6 & 0 & -4000 & 0.20 & -3000 & -3000 & 1 & 0.92 & 95 & @xcite & 8/6/25 + 7 & 0 & -4000 & 0.80 & 0 & -3000 & 0.75 & 0.42 & 95 & @xcite & 6/11/14 + 8 & 0 & -6000 & 0.55 & 0 & -3000 & 0.10 & 0.92 & 66 & @xcite & 6/11/14 + 9 & 0 & -6000 & 0.999 & 0 & -3000 & 0.998 & 0.30 & 66 & @xcite & 6/11/14 + 10 & 1000 & 0 & 0.50 & 500 & 500 & 1 & 0.16 & 70 & @xcite & 8/6/25 + 11 & 0 & -1000 & 0.50 & -500 & -500 & 1 & 0.69 & 68 & @xcite & 8/6/25 + 12 & 5000 & 0 & 0.001 & 5 & 5 & 1 & 0.72 & 72 & @xcite & 8/6/25 + 13 & 0 & -5000 & 0.999 & -5 & -5 & 1 & 0.17 & 72 & @xcite & 8/6/25 + 14 & 1260 & 1260 & 1 & 3394 & 0 & 0.50 & 0.78 & 72 & @xcite & 22/8/ * + 15 & 1260 & 0 & 0.10 & 3394 & 0 & 0.05 & 0.33 & 72 & @xcite & 13/11/35 + 16 & 9 & 9 & 1 & 50 & 0 & 0.10 & ? & 25 & @xcite & 22/8/ * + 17 & 21 & 21 & 1 & 50 & 0 & 0.50 & ? & 25 & @xcite & 22/8/ * + 18 & 37 & 37 & 1 & 50 & 0 & 0.90 & ? & 25 & @xcite & 22/8/ * + 19 & -8 & -8 & 1 & 0 & -50 & 0.90 & ? & 25 & @xcite & 22/8/ * + 20 & -21 & -21 & 1 & 0 & -50 & 0.50 & ? & 25 & @xcite & 22/8/ * + 21 & -39 & -39 & 1 & 0 & -50 & 0.10 & ? & 25 & @xcite & 22/8/ * + 22 & 14 & 14 & 1 & 100 & 0 & 0.05 & ? & 25 & @xcite & 22/8/ * + 23 & 25 & 25 & 1 & 100 & 0 & 0.25 & ? & 25 & @xcite & 22/8/ * + 24 & 36 & 36 & 1 & 100 & 0 & 0.50 & ? & 25 & @xcite & 22/8/ * + 25 & 52 & 52 & 1 & 100 & 0 & 0.75 & ? & 25 & @xcite & 22/8/ * + 26 & 78 & 78 & 1 & 100 & 0 & 0.95 & ? & 25 & @xcite & 22/8/ * + 27 & -8 & -8 & 1 & 0 & -100 & 0.95 & ? & 25 & @xcite & 22/8/ * + 28 & -23.5 & -23.5 & 1 & 0 & -100 & 0.75 & ? & 25 & @xcite & 22/8/ * + 29 & -42 & -42 & 1 & 0 & -100 & 0.50 & ? & 25 & @xcite & 22/8/ * + 30 & -63 & -63 & 1 & 0 & -100 & 0.25 & ? & 25 & @xcite & 22/8/ * + 31 & -84 & -84 & 1 & 0 & -100 & 0.05 & ? & 25 & @xcite & 22/8/ * + 32 & 10 & 10 & 1 & 200 & 0 & 0.01 & ? & 25 & @xcite & 22/8/ * + 33 & 20 & 20 & 1 & 200 & 0 & 0.10 & ? & 25 & @xcite & 22/8/ * + 34 & 76 & 76 & 1 & 200 & 0 & 0.50 & ? & 25 & @xcite & 22/8/ * + 35 & 131 & 131 & 1 & 200 & 0 & 0.90 & ? & 25 & @xcite & 22/8/ * + 36 & 188 & 188 & 1 & 200 & 0 & 0.99 & ? & 25 & @xcite & 22/8/ * + 37 & -3 & -3 & 1 & 0 & -200 & 0.99 & ? & 25 & @xcite & 22/8/ * + 38 & -23 & -23 & 1 & 0 & -200 & 0.90 & ? & 25 & @xcite & 22/8/ * + 39 & -89 & -89 & 1 & 0 & -200 & 0.50 & ? & 25 & @xcite & 22/8/ * + 40 & -155 & -155 & 1 & 0 & -200 & 0.10 & ? & 25 & @xcite & 22/8/ * + 41 & -190 & -190 & 1 & 0 & -200 & 0.01 & ? & 25 & @xcite & 22/8/ * + 42 & 12 & 12 & 1 & 400 & 0 & 0.01 & ? & 25 & @xcite & 22/8/ * + 43 & 377 & 377 & 1 & 400 & 0 & 0.99 & ? & 25 & @xcite & 22/8/ * + 44 & -14 & -14 & 1 & 0 & -400 & 0.99 & ? & 25 & @xcite & 22/8/ * + 45 & -380 & -380 & 1 & 0 & -400 & 0.01 & ? & 25 & @xcite & 22/8/ * + 46 & 59 & 59 & 1 & 100 & 50 & 0.10 & ? & 25 & @xcite & 22/8/ * + 47 & 71 & 71 & 1 & 100 & 50 & 0.50 & ? & 25 & @xcite & 22/8/ * + 48 & 83 & 83 & 1 & 100 & 50 & 0.90 & ? & 25 & @xcite & 22/8/ * + 49 & -59 & -59 & 1 & -50 & -100 & 0.90 & ? & 25 & @xcite & 22/8/ * + 50 & -71 & -71 & 1 & -50 & -100 & 0.50 & ? & 25 & @xcite & 22/8/ * + 51 & -85 & -85 & 1 & -50 & -100 & 0.10 & ? & 25 & @xcite & 22/8/ * + 52 & 64 & 64 & 1 & 150 & 50 & 0.05 & ? & 25 & @xcite & 22/8/ * + 53 & 72.5 & 72.5 & 1 & 150 & 50 & 0.25 & ? & 25 & @xcite & 22/8/ * + 54 & 86 & 86 & 1 & 150 & 50 & 0.50 & ? & 25 & @xcite & 22/8/ * + 55 & 102 & 102 & 1 & 150 & 50 & 0.75 & ? & 25 & @xcite & 22/8/ * + 56 & 128 & 128 & 1 & 150 & 50 & 0.95 & ? & 25 & @xcite & 22/8/ * + 57 & -60 & -60 & 1 & -50 & -150 & 0.95 & ? & 25 & @xcite & 22/8/ * + 58 & -71 & -71 & 1 & -50 & -150 & 0.75 & ? & 25 & @xcite & 22/8/ * + 59 & -92 & -92 & 1 & -50 & -150 & 0.50 & ? & 25 & @xcite & 22/8/ * + 60 & -113 & -113 & 1 & -50 & -150 & 0.25 & ? & 25 & @xcite & 22/8/ * + 61 & -132 & -132 & 1 & -50 & -150 & 0.05 & ? & 25 & @xcite & 22/8/ * + 62 & 118 & 118 & 1 & 200 & 100 & 0.05 & ? & 25 & @xcite & 22/8/ * + 63 & 130 & 130 & 1 & 200 & 100 & 0.25 & ? & 25 & @xcite & 22/8/ * + 64 & 141 & 141 & 1 & 200 & 100 & 0.50 & ? & 25 & @xcite & 22/8/ * + 65 & 162 & 162 & 1 & 200 & 100 & 0.75 & ? & 25 & @xcite & 22/8/ * + 66 & 178 & 178 & 1 & 200 & 100 & 0.95 & ? & 25 & @xcite & 22/8/ * + 67 & -112 & -112 & 1 & -100 & -200 & 0.95 & ? & 25 & @xcite & 22/8/ * + 68 & -121 & -121 & 1 & -100 & -200 & 0.75 & ? & 25 & @xcite & 22/8/ * + 69 & -142 & -142 & 1 & -100 & -200 & 0.50 & ? & 25 & @xcite & 22/8/ * + 70 & -158 & -158 & 1 & -100 & -200 & 0.25 & ? & 25 & @xcite & 22/8/ * + 71 & -179 & -179 & 1 & -100 & -200 & 0.05 & ? & 25 & @xcite & 22/8/ * + 72 & 100 & -100 & 0.65 & 0 & 0 & 1 & 0.50 & ? & @xcite & 8/6/25 + 73 & 0 & -200 & 0.20 & -100 & -100 & 1 & 0.50 & ? & @xcite & 8/6/25 + for points 14 and 15 the prospects were formulated with non - monetary outcomes where ( 1 ) is a one - week tour of england and ( 2 ) is a three - week tour of england , france , and italy . at the time of this writing the site http://www.budgetyourtrip.com/countrylist.php presents the following costs per day per person : united kingdom 111.67 , france 128.58 , and italy 111.65 . with the exchange rates 0.6205 /$ and 0.788 /$ we get modern dollar equivalents ( 1 ) @xmath444 and ( 2 ) @xmath445 . these modern absolute values with the ratio @xmath446 and the fractions of respondents obtained more than 35 years ago should be used carefully with low weights . the points 72 and 73 are referenced by @xcite . for points 1 - 15 the numbers of respondents selecting each prospect were counted @xcite and being divided by their total numbers give the estimates @xmath438 and complementary @xmath439 . [ [ median - cash - equivalents . ] ] median cash equivalents . + + + + + + + + + + + + + + + + + + + + + + + + in contrast , for points 16 - 71 a _ certainty equivalent _ was estimated ( * ? ? ? * - 306 ) . the latter concept , and method can be illustrated by an example . errors in the example belong to the author . let @xmath327 is @xmath447 . the certainty equivalent is the prospect @xmath448 . given @xmath327 and the expected value @xmath449 , the goal was to determine experimentally @xmath450 . the authors @xcite do not speak about sample means . such an interpretation attempts to match their results with the proposed approach . _ `` the display also [ vs : together with the original prospect and expected value ] included a descending series of seven sure outcomes ( gains or losses ) logarithmically spaced between the extreme outcomes of the prospect . the subject indicated a preference between each of the seven sure outcomes and the risky prospect '' _ @xcite . it does not say whether the seven plus two extreme values belong to one dependence , where neighboring logarithms are equidistant . in the latter case , the logarithm base is not important and the descending values are members of the geometric progression @xmath451 , where @xmath452 is the number of sure outcomes . for @xmath453 and @xmath454 we get @xmath455{\frac{a_0}{a_{n+1 } } } = \sqrt[8]{\frac{150}{50 } } \approx 1.147 $ ] and seven sure outcomes 130.75 ( + ) , 113.98 ( + ) , 99.35 ( + ) , 86.60 ( + ) , 75.49 ( - ) , 65.80 ( - ) , 57.36 ( - ) . accepted ( the sure outcome is chosen ) and rejected ( the prospect is selected ) values are marked by + and - . the signs in the example are fictional . _ `` to obtain a more refined estimate of the certainty equivalent , a new set of seven outcomes was then shown , linearly spaced between a value 25% higher than the lowest amount accepted in the first set and a value 25% lower than the highest amount rejected '' _ * - 306 ) . therefore , the lowest accepted value 86.60 would be increased by 25% to 108.25 and the highest rejected value 75.49 would be decreased by 25% to 60.39 . the interval @xmath456 $ ] would be added by new seven _ linearly spaced _ sure outcomes : 102.27 ( + ) , 96.29 ( + ) , 90.30 ( + ) , 84.32 ( + ) , 78.34 ( + ) , 72.36 ( - ) , 66.37 ( - ) . the mid point between the lowest accepted 78.34 and highest rejected 72.36 values @xmath457 was used as a certainty equivalent . inconsistent selections from the first and second sets were rejected by a computer . _ the median cash equivalents of the 25 subjects _ were reported and are cited for points 16 - 71 in columns @xmath458 . placing seven sure and two extreme outcomes on one logarithmic dependence , representing a geometric progression with @xmath459 , fails , if two extreme outcomes have a different sign or one of them is zero . the author thinks that inserting seven values under such conditions creates bias . for two extreme outcomes 0 and 50 , points 16 - 18 , one , dividing @xmath460 by 8 and gradually decrementing @xmath460 , reaches 0 , where @xmath461 imply the prospect @xmath462 but not @xmath463 . the same problem arises for points 16 - 45 , 73 . it would be also difficult to apply this method to point 72 @xmath464 provoking bias . in order to combine results of @xcite and @xcite , the method cited above @xcite being applied to prospects @xcite must reproduce the fractions of respondents . @xcite has a different goal - median cash equivalent . the number of subjects 25 in @xcite is 2 - 4 times less than the numbers of respondents in @xcite . the author can not conclude that the median cash equivalent associates with the fractions of respondents @xmath465 or other obvious fraction . @xmath438 could be obtained after requesting respondents to select between two prospects , for instance , @xmath466 and @xmath467 , point 53 , and counting votes . in axioms , _ if two variables are deterministically equivalent , then the fractions of respondents are undefined . estimates made under such conditions would fluctuate between zero and one_. with @xmath441 are on a six dimensional projection . the same is true for points 7 - 9 with @xmath468 . [ [ sobol - sequences - for - experimental - plans . ] ] sobol sequences for experimental plans . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + previous experiments , table [ tblf1 ] , had no a goal to fit @xmath335 and @xmath336 . they neither fill `` uniformly '' parts of the eight dimensional space nor the projection on figure [ figf1 ] , where a plan should fill the product - square @xmath469 . ilya meerovich sobol mentions that if dots of a tv @xmath470 raster are transfered not by rows but _ quasi - randomly _ , then recognition of large objects and their moves is possible without waiting when the image will fill all cells and independently on the object location ( * ? ? ? _ it is worth to consider sobol sequences @xcite , @xcite , @xcite , @xcite for selecting points in regions , not covered by the axioms , for experimental estimation of @xmath335 and @xmath336 . _ [ [ the - ratio ] ] the ratio + + + + + + + + + @xmath471 decreases the number of variables . the reduction can be inadequate , if absolute outcomes matched with income of respondents will result in outliers , where under other equal conditions more than one value @xmath438 will correspond to a single ratio . similar notes relate to the ratio @xmath472 . [ [ swapping - indexes-1-and-2 ] ] swapping indexes 1 and 2 + + + + + + + + + + + + + + + + + + + + + + + + in tables [ tbltheoryf1 ] - [ tblcartesianf1 ] neither eliminates nor adds uncertainty . it swaps @xmath335 and @xmath336 and transforms entries marked by * in the last column of table [ tblf1 ] and meaning `` absent in table [ tblcartesianf1 ] '' to an entry found in the latter table : @xmath473 , @xmath474 . for instance , experiment 1 from table [ tblf1 ] corresponds to a case , existing in table [ tblcartesianf1 ] , and can be applied together with other points for estimating parameters of a model function @xmath475 . from 18 unpredictable cases 2 , 5 - 7 , 9 , 14 , 16 - 18 , 20 , 21 , 23 - 25 , 27 , 28 , 33 , 35 in table [ tblcartesianf1 ] , only three 14 , 25 , 35 with the help of swapping are covered by available to the author data . the author does not see how to reuse certainty equivalents for estimating the functions of respondents . _ in later sections it will be shown that statistics of lotteries can provide additional points . _ [ [ it - is - natural ] ] it is natural + + + + + + + + + + + + + to `` aggregate '' the eight parameters to means , variances , other sample mean moments , entropy . these `` secondary '' quantities may supply a clearer meaning : variance tells about deviations from mean , entropy is a measure of uncertainty , etc . _ the formula for @xmath335 may get `` physical sense '' . _ figure [ figf1 ] and the axioms hint that @xmath335 can be a combination of continuous and step functions . we should not forget that @xmath335 can depend not only on the eight parameters fully characterizing the sample means underlying the prospects but uncovered parameters related to respondents . this can lead to several @xmath438 values collected from different groups of respondents and corresponding to the same eight values of arguments . the relationships may change in time . discrepancies can be treated as `` experimental errors '' using statistics as long as they are `` acceptably small '' . otherwise , they would promote new dimensions of the problem . it will be beneficial to sketch projections of @xmath335 dependent on the eight arguments or quantities derived from them , _ where the role of time is in the increasing @xmath476 and @xmath477 and erasing uncertainty_. if an award or loss is a sample mean , then the role of time is in the elimination of uncertainty . with time , a high frequency of games increases the number of trials @xmath45 . this decreases a sample mean variance , if the variance of a single trial is finite @xcite . in the limit @xmath285 , random sample means become deterministic mathematical expectations . the latter will dominate in decision making , if large @xmath45 is reachable and costs nothing , axioms [ eqaxiom1 ] - [ eqaxiom3 ] . estimates made with the prokhorov s inequality [ eqprokhorov ] show that required @xmath45 can be large . figure [ figf1 ] should `` converge '' to figure [ figf1inf ] . with infinite numbers of trials applied for both @xmath326 and @xmath327 . green lines and surfaces correspond to the postulates . red areas indicate ambiguity forming the `` vertical wall '' of the step function probability slices . @xmath441.,width=245 ] by the reasons explained later the following is chosen for @xmath478 @xmath479 it takes 0 or 1 , is undefined , if the mathematical expectations of two prospects given by equation [ eqmeanofsamplemean ] are equal each to other , and agrees with the axioms and equation [ eqfunctional ] : @xmath480 . in contrast , if a respondent feels that a large number of trials almost zeroing variance is not reachable in a reasonable time , then his or her choice will be influenced not only by mathematical expectations . buyers of mega million and powerball tickets `` prove '' it , when the jackpots are renewed and the expectation of the game is negative . the case @xmath212 is embedded into kahneman s and tversky s prospects @xcite , @xcite . in trading and market speculation there is a lack of control that a decision to buy or sell a security or contract is made under the same conditions . a participant focusing on a rule or pattern ignores other factors possibly affecting a next price or rate move . the majority feels a uniqueness of trading opportunity and inability to `` play '' endlessly and deviates from mathematical expectations . this tendency was described by livermore @xcite prior kahneman s and tversky s systematic research , demolishing the utility theory and adding valuable experimental points to the functions of choice . an apologist of the _ technical analysis _ @xcite can argue that formations of trading price patterns or indicator values themselves point to repetition of underlying conditions . without presenting evidences of such dependencies a healthy criticism is born because many patterns can appear as a result of _ random walk _ @xcite . less cited but not less important mathematical properties of a _ symmetric random walk _ are described in @xcite , @xcite.the review of the technical analysis @xcite leaves to a speculator `` hopes on systematic success '' , following from non - linear properties of the real financial time - series . there is a lack of published results in this area . in @xcite , @xcite , @xcite the author concludes that as long as a market creates frequent opportunities to make large profits , speculation will exist , and suggests measuring the frequencies and potential profits using the _ maximum profit strategy _ , which is an objective market property . a point of view distinguishing between economic risk and uncertainty situations is presented in @xcite . while the risk assumes probability distributions known at the outset , uncertain situations imply absence of the probability models describing them . frank knight talks about @xcite : _ ... that higher form of uncertainty not susceptible to measurement and hence to elimination_. he concludes : _ it is this true uncertainty which by presenting the theoretically perfect outworking of the tendencies of competition gives the characteristic form of `` enterprise '' to economic organizations as a whole and accounts for the peculiar income of the entrepreneur_. the author believes that inability to measure such uncertainty also deals with fundamental limits of the probability theory requiring repetition of events and trials . trading is full of _ individual random objects_. an e - mini s&p 500 future contract price time series is unique . ability to check presence of the `` same '' conditions for making decisions based on the past time series is limited . classics well understood the limits of the probability theory @xcite . the latter work summarizes three approaches and definitions of randomness of an individual sequence of 0s and 1s based on the theory of algorithms . in contrast with these difficulties , the size and frequency of market offers measured by the maximum profit strategy do not become less intensive @xcite . in the sense described here , time decreases the variance of the sample mean awards and `` evaporates '' their uncertainty . kelly @xcite introduces the _ exponential rate of growth of the gambler s capital _ as the limit @xmath481 . it is possible to show that under his conditions and for his function @xmath482 with the constant bet allocation fraction @xmath483 , where @xmath484 and @xmath485 are the numbers of wins and losses and @xmath486 and @xmath487 is the initial and after @xmath45 bets capital , this limit coincides with the mathematical expectation @xmath488 for arbitrary positive integer @xmath45 . this implies that the process @xmath489 is not a _ martingale _ , see definition in @xcite . formulas suitable for futures trading based on the allocation fraction of the capital for a next trade are considered in @xcite , @xcite , and ( * ? ? ? kelly s approach is a bright example demonstrating that people may rely not only on a mathematical expectation of a game . another famous example is the _ sharpe ratio _ @xcite . khinchin @xcite well understood the influence of the growing number of trials on the properties of the arithmetic and geometric sample means in the st . petersburg game . however , his two theorems do not allow to estimate a convergence rate of the means . such estimates based on his theorems have been obtained on a computer @xcite . prokhorov s estimate , when it can be applied , is a key to verification of influence of time linearly affecting the number of trials under a condition of constant frequency of trials . [ [ the - random - difference ] ] the random difference + + + + + + + + + + + + + + + + + + + + + of the two _ independent _ random variables @xmath490 characterizes absolute advantage , if it is positive , or disadvantage , if it is negative , of the award @xmath326 over @xmath327 . independence simplifies formulas for the moments due to known relationships for @xmath491 @xcite : @xmath492 @xmath493 it can be proved that for the difference @xmath494 two of the four formulas have to change to @xmath495 and @xmath496 . using equations from section `` problem 3 from kahneman and tversky '' we get for two - point variables @xmath497 @xmath498 @xmath499 @xmath500}{n^{(1)3 } } + \\ + 6 \frac{(a_p^{(1)}-a_q^{(1)})^2p^{(1)}(1-p^{(1 ) } ) ( a_p^{(2)}-a_q^{(2)})^2p^{(2)}(1-p^{(2 ) } ) } { n^{(1)}n^{(2 ) } } + \\ + \frac{(a_p^{(2)}-a_q^{(2)})^4p^{(2)}(1-p^{(2)})[1 + 3(n^{(2)}-2)p^{(2)}(1-p^{(2)})]}{n^{(2)3 } } , \end{split}\ ] ] @xmath501 @xmath502 for @xmath504 , the four values of @xmath220 and corresponding probabilities are @xmath505 dependencies of statistics of the random difference @xmath220 on probability of the outcome @xmath506 and number of trials @xmath476 for problem 3 ( certainty effect ) are plotted on figure [ figa1diffproblem3 ] and illustrate formulas [ eqmeandiff ] - [ eqexcesskurtosisdiff ] . dependencies of @xmath220 on probability of the outcome @xmath507 and number of trials @xmath476 for problem 3 ( reflection effect ) are presented on figure [ figa1diffproblem3b ] . how much @xmath326 _ can be _ greater than @xmath327 ? if both are deterministic , then our answer is @xmath511 . the function @xmath512 is applied to the difference in order to emphasize the adjective _ greater_. if the difference is negative , then @xmath326 is _ not greater _ than @xmath327 : the positive excess is zero . if both are random , then _ on average _ our answer is @xmath513 . but not _ on average _ it can be greater and we express it adding @xmath514 standard deviations to the mean @xmath515 , where @xmath516 . when one or both sample means are random , @xmath326 can be less than @xmath327 . how much less we express as a non - negative quantity @xmath517 , where @xmath518 . the number @xmath519 does not have to be equal to @xmath514 , if the distribution of @xmath220 is not symmetrical . how much @xmath326 _ can be _ less than @xmath327 is equal to that how much @xmath327 can be greater than @xmath326 : @xmath520 . how much @xmath326 can be greater than @xmath327 is equal to that how much @xmath327 can be less than @xmath326 : @xmath521 . _ the ratio of the fractions of respondents is the ratio of how much _ @xmath326 _ can be greater than _ @xmath327 _ to how much _ @xmath327 _ can be greater than _ @xmath326 @xmath522 @xmath523 @xmath524 _ in general , the ratio of the fractions of respondents choosing between two random or deterministic prospects is the ratio of how much the first prospect can be `` better '' than the second to how much the second can be `` better '' than the first . _ equations [ eqtheratio ] - [ eqthef2 ] express what is `` better '' for two sample means . as long as _ common rational sense _ underlies `` better '' , fractions depend on prospects only . if `` better '' is dominated by subjective preferences , then the formulas can stop working . in a private discussion with pavel grosul , it was emphasized that the same money means different things for respondents with different wealth . in @xcite and @xcite a care was taken to choose some outcomes comparable or intentionally deviating from a monthly income . accounting such effects requires a distribution of respondents by income and/or wealth . equations above can hide subjective complexity in @xmath525 and @xmath526 , but do not pretend to become _ formula amoris _ or _ equations of attractiveness ... _ if @xmath527 , then equation [ eqthef1 ] @xmath528 [ eqf1inf ] . natural constraints are @xmath529 for two - point variables @xmath530 , @xmath531 , @xmath532 , @xmath533 @xmath534 the @xmath535 and @xmath536 posses the property that swapping indexes 1 and 2 transforms @xmath537 to @xmath538 and vice versa . if the same property holds for other @xmath539 and @xmath540 , then @xmath541 and @xmath542 will have it too because the remaining @xmath543 and @xmath544 already have it . this would allow to write @xmath545 and @xmath546 and use one function . if @xmath547 , then @xmath539 and @xmath540 can be arbitrary finite numbers . in problem 3 @xmath548 , @xmath549 , @xmath550 , and @xmath551 . therefore , @xmath552 and @xmath553 . any pair @xmath554 satisfying the equation @xmath555 with @xmath556 being substituted to equations [ eqthef1 ] and [ eqthef2 ] yields @xmath557 and @xmath558 experimentally found for problem 3 . intersection with constrains gives @xmath559 and @xmath560 . in problem 3 @xmath561 , @xmath562 , @xmath563 , and @xmath551 . therefore , @xmath564 and @xmath565 . any pair @xmath554 obeying @xmath566 with @xmath567 yields @xmath568 and @xmath569 observed for problem 3. intersection with constraints gives @xmath570 and @xmath571 . the solution lines for problem 3 and 3 intersect at @xmath572 , which violates constraints : there is no one pair satisfying both problems . the @xmath537 and @xmath538 obeying constraints [ eqconstraintsab ] ensure predictable properties . if @xmath335 or @xmath336 are not equal to zero or one , then the formulated relationships [ eqtheratio ] - [ eqconstraintsab ] imply dependence between @xmath537 and @xmath538 because from two fractions of respondents only one is independent . if @xmath573 , then @xmath574 where @xmath490 . @xmath575 can be interpreted as a sharpe ratio , where @xmath326 is not a return but a sample mean over or under a benchmark sample mean @xmath327 . if @xmath326 and @xmath327 are i.i.d . , then @xmath576 , @xmath577 ( non - constant variables ) , and @xmath578 . from equations [ eqvariancediff ] and [ eqbinaryconstraintsab ] , @xmath579 . similar to equal constants , there is no rational sense to select between two identical random quantities and @xmath335 must be undefined . _ we conclude that in this case _ @xmath580 . in equation [ eqthef1 ] the first @xmath512 in the denominator is zero , if @xmath581 . @xmath582 for @xmath583 . the second max is zero , if @xmath584 . thus , @xmath585 is undefined , if @xmath583 and @xmath586 . this implies @xmath587 . if only one of the inequalities holds , then equation [ eqthef1 ] correctly returns 1 or 0 corresponding to predictable properties . + & & & & & & & & & & & + continued from previous page + & & & & & & & & & & & + + 1 & @xmath326 & 6000 & 0 & 0.45 & 2700 & 2985 & 0.201 & -1.96 & 0.993 & 0.14 & 4 + & @xmath327 & 3000 & 0 & 0.9 & 2700 & 900 & -2.67 & 5.11 & 0.469 & 0.86 & 4 + & @xmath220 & & & & 0 & 3118 & 0.24 & -1.61 & 1.46 & 1.00 & + 2 & @xmath326 & 6000 & 0 & 0.001 & 6 & 190 & 31.6 & 995 & 0.011 & 0.73 & 5 + & @xmath327 & 3000 & 0 & 0.002 & 6 & 134 & 22.3 & 495 & 0.021 & 0.27 & 5 + & @xmath220 & & & & 0 & 232 & 12.9 & 497 & 0.032 & 1.00 & + 3 & @xmath326 & 0 & -6000 & 0.55 & -2700 & 2985 & -0.201 & -1.96 & 0.993 & 0.92 & 8 + & @xmath327 & 0 & -3000 & 0.1 & -2700 & 900 & 2.67 & 5.11 & 0.469 & 0.08 & 8 + & @xmath220 & & & & 0 & 3118 & -0.24 & -1.61 & 1.46 & 1.00 & + 4 & @xmath326 & 0 & -6000 & 0.999 & -6 & 190 & -31.6 & 995 & 0.011 & 0.30 & 9 + & @xmath327 & 0 & -3000 & 0.998 & -6 & 134 & -22.3 & 495 & 0.021 & 0.70 & 9 + & @xmath220 & & & & 0 & 232 & -12.9 & 497 & 0.032 & 1.00 & + 5 & @xmath326 & 1000 & 0 & 0.5 & 500 & 500 & 0 & -2 & 1 & 0.16 & 10 + & @xmath327 & 500 & 500 & 1 & 500 & 0 & 0 & -3 & 0 & 0.84 & 10 + & @xmath220 & & & & 0 & 500 & 0 & -2 & 1 & 1.00 & + 6 & @xmath326 & 0 & -1000 & 0.5 & -500 & 500 & 0 & -2 & 1 & 0.69 & 11 + & @xmath327 & -500 & -500 & 1 & -500 & 0 & 0 & -3 & 0 & 0.31 & 11 + & @xmath220 & & & 0 & & 500 & 0 & -2 & 1 & 1.00 & + 7 & @xmath326 & 5000 & 0 & 0.001 & 5 & 158 & 31.6 & 995 & 0.011 & 0.72 & 12 + & @xmath327 & 5 & 5 & 1 & 5 & 0 & 0 & -3 & 0 & 0.28 & 12 + & @xmath220 & & & & 0 & 158 & 31.6 & 995 & 0.011 & 1.00 & + 8 & @xmath326 & 0 & -5000 & 0.999 & -5 & 158 & -31.6 & 995 & 0.011 & 0.17 & 13 + & @xmath327 & -5 & -5 & 1 & -5 & 0 & 0 & -3 & 0 & 0.83 & 13 + & @xmath220 & & & & 0 & 158 & -31.6 & 995 & 0.011 & 1.00 & + 9 & @xmath326 & 25 & 25 & 1 & 25 & 0 & 0 & -3 & 0 & ? & 23 + & @xmath327 & 100 & 0 & 0.25 & 25 & 43 & 1.15 & -0.667 & 0.81 & ? & 23 + & @xmath220 & & & & 0 & 43 & -1.15 & -0.667 & 0.81 & 1.00 & + 10 & @xmath326 & 20 & 20 & 1 & 20 & 0 & 0 & -3 & 0 & ? & 33 + & @xmath327 & 200 & 0 & 0.1 & 20 & 60 & 2.67 & 5.11 & 0.469 & ? & 33 + & @xmath220 & & & & 0 & 60 & -2.67 & 5.11 & 0.469 & 1.00 & + if profits are comparable table [ tbla1a2eta](1 , 2 , 5 ) , then the greater entropy , the less the fraction of respondents . the entropy , characterizing uncertainty , becomes a measure of the certainty effect . if one profit is significantly greater than other and the latter is not attractive , then uncertainty is less important than a possibility to become rich , table [ tbla1a2eta](7 ) . the entropy depends only on probabilities . if losses are comparable table [ tbla1a2eta](3 , 4 , 6 ) , then the greater entropy , the greater the fraction of respondents . now , entropy measures reflection effect . if one loss is significantly worse than other and the latter is not critical , then uncertainty is less important comparing with a possibility of financial ruin , table [ tbla1a2eta](8 ) . for profits table [ tbla1a2eta](5 , 7 ) , @xmath588 , @xmath592 , @xmath593 . therefore , @xmath594 , @xmath595 , @xmath596 , @xmath597 . author s attention is attracted by one _ rational _ and two experimental points @xmath598 : @xmath599 , @xmath600 , @xmath601 , which are close to a straight line . for losses table [ tbla1a2eta](6 , 8) , @xmath588 , @xmath602 , @xmath603 . hence , @xmath604 , @xmath595 , @xmath605 , @xmath606 . author s attention is attracted by one rational and two experimental points @xmath598 : @xmath607 , @xmath608 , @xmath609 , which are close to a straight line . in both cases the standard deviation can approach infinity , while @xmath610 . therefore , lines @xmath611 can not be adequate on the beam @xmath612 . the behavior on the right side is not obvious for the author . at first glance , decreasing dependence for profits could asymptotically approach the standard deviation axis @xmath613 being a mirrored s - shaped curve bound to the point ( 0 , 1 ) . however , if @xmath614 is tiny , one cent , comparing with a healthy monthly income , then large but less probable @xmath615 can become attractive . it gives a chance to `` become rich '' momentarily , while one cent would not make a principal difference anyway . we recollect people playing lotteries . two curves of fractions of respondents dependent on variances of random prospects and fitting rational and experimental points are plotted on figure [ figf1deta0 ] . they asymptotically approach zero ( profits ) or one ( losses ) . the fitting curves are @xmath617 , and @xmath618 . the latter equation is mathematically equivalent to the _ archibald hill s equation _ describing the sigmoidal oxygen binding curve of hemoglobin @xcite . in the particular case @xmath116 , the equation transforms to @xmath619 , which is mathematically equivalent to the _ leonor michaelis - maud menten equation _ describing enzyme kinetics . both can be written as straight line equations in bilogarithmic scales @xmath620 , and @xmath621 each fitting two points exactly with @xmath622 , and @xmath623 . the differences in @xmath162 , @xmath160 , and @xmath73 , @xmath109 take care about description of certainty and reflection effects . _ absence of data on the right side of dependencies leaves the question about asymptotic properties open_. equations [ eqthef1 ] and [ eqthef2 ] correctly describe rational properties where @xmath626 . under other conditions , they switch computation of fractions of respondents to evaluation of @xmath537 , @xmath538 , and possible dependences between them . let the first variable is constant @xmath627 and second - random @xmath628 , 1 \le n^{(2)}=n^{(1)}\}$ ] , where @xmath629 . compare the surface on figure [ figf1constrand1 ] with experimental and axiomatic points on figure [ figf1 ] . for large @xmath630 figure [ figf1constrand2 ] is getting closer to figure [ figf1inf ] following from equation [ eqf1inf ] . the surfaces are computed using equations [ eqmeandiff ] , [ eqvariancediff ] , and [ eqthef1 ] . fitting depends on @xmath537 , and @xmath538 . from equations [ eqvariancediff ] and [ eqbinaryconstraintsab ] , if the first variable is constant and the second is random two - point , then @xmath637 , @xmath638 , @xmath639 . swapping constant and random variables yields @xmath640 , @xmath641 , @xmath642 . usually , the greater the jackpot , the faster it grows , figure [ figmegamillionsjackpot ] . if the jackpot is won , then the curve drops to the initial jackpot @xmath645 . this creates the sequence of `` teeth '' with random height and width at the bottom . dependence after @xmath645 resembles the exponential solution @xmath646 of the equation @xmath647 with the initial condition @xmath648 . this solution can not describe random shape of the teeth and sharp dropping at random time . the jackpot , even , without dropping , could not grow to infinity . hence , the s - shaped _ logistic curve _ @xmath649 , solving the equation @xmath650 developed with other denominations by pierre francois verhulst and lamberte adolphe jacques quetelet for describing population growth , is a better fitting choice . without data close to @xmath651 , estimation of this level is inaccurate . the richards _ generalized logistic function _ @xcite is more flexible . with credit to http://www.lottostrategies.com/script/jackpot_history/draw_date/113 , the author presents the data in table [ tbljackpot ] for those who wants to try other fitting options . the exponential and logistic curves were applied minimizing the sum of square deviations ( microsoft excel , solver ) , which is not a well justified criterion in this case . @xmath652 with @xmath653 does not take into account a possibility of sharing jackpot between winners . for simplicity , we also ignore multiple winning outcomes and taxes described in section `` mega millions '' . + & & & & & & & + continued from previous page + & & & & & & & + + 3/27/2015 & 0.0000 & 15000000 & 5000000 & 15000000 & 0.0e+00 & 15000000 & 0.0e+00 + 3/31/2015 & 0.0110 & 20000000 & 5000000 & 18490306 & 2.3e+12 & 17920342 & 4.3e+12 + 4/3/2015 & 0.0192 & 25000000 & 5000000 & 21608026 & 1.2e+13 & 20478002 & 2.0e+13 + 4/7/2015 & 0.0301 & 30000000 & 9000000 & 26551757 & 1.2e+13 & 24464854 & 3.1e+13 + 4/10/2015 & 0.0384 & 39000000 & 8000000 & 30941404 & 6.5e+13 & 27956572 & 1.2e+14 + 4/14/2015 & 0.0493 & 47000000 & 8000000 & 37851574 & 8.4e+13 & 33399424 & 1.8e+14 + 4/17/2015 & 0.0575 & 55000000 & 10000000 & 43935991 & 1.2e+14 & 38166318 & 2.8e+14 + 4/21/2015 & 0.0685 & 65000000 & 9000000 & 53417755 & 1.3e+14 & 45596900 & 3.8e+14 + 4/24/2015 & 0.0767 & 74000000 & 11000000 & 61670446 & 1.5e+14 & 52104664 & 4.8e+14 + 4/28/2015 & 0.0877 & 85000000 & 11000000 & 74355473 & 1.1e+14 & 62248896 & 5.2e+14 + 5/1/2015 & 0.0959 & 96000000 & 14000000 & 85225587 & 1.2e+14 & 71133297 & 6.2e+14 + 5/5/2015 & 0.1068 & 110000000 & 16000000 & 101632292 & 7.0e+13 & 84982204 & 6.3e+14 + 5/8/2015 & 0.1151 & 126000000 & 14000000 & 115408834 & 1.1e+14 & 97111190 & 8.3e+14 + 5/12/2015 & 0.1260 & 140000000 & 19000000 & 135724730 & 1.8e+13 & 116017720 & 5.8e+14 + 5/15/2015 & 0.1342 & 159000000 & 14000000 & 152355179 & 4.4e+13 & 132576214 & 7.0e+14 + 5/19/2015 & 0.1452 & 173000000 & 21000000 & 176195256 & 1.0e+13 & 158387412 & 2.1e+14 + 5/22/2015 & 0.1534 & 194000000 & 20000000 & 195128304 & 1.3e+12 & 180993072 & 1.7e+14 + 5/26/2015 & 0.1644 & 214000000 & 19000000 & 221398922 & 5.5e+13 & 216230524 & 5.0e+12 + 5/29/2015 & 0.1726 & 233000000 & 27000000 & 241565629 & 7.3e+13 & 247091774 & 2.0e+14 + 6/2/2015 & 0.1836 & 260000000 & ? & 268580467 & 7.4e+13 & 295197950 & 1.2e+15 + the choice @xmath654 vs. @xmath655 should be made on 5/29/2015 . the jackpot has grown to 260,000,000 . we assume that 27,000,000 is one third of the number of purchased tickets 81,000,000 . since quoting is done by annuity , the number of tickets is less . based on section `` mega millions '' , we estimate the number of purchased tickets as half @xmath656 40,000,000 . mega millions http://www.megamillions.com/where-to-play is played in 44 states ar 2,966,369 , az 6,731,484 , ca 38,802,500 , co 5,355,866 , ct 3,596,677 , de 935,614 , fl 19,893,297 , ga 10,097,343 , ia 3,107,124 , i d 1,634,464 , il 12,880,580 , in 6,596,855 , ks 2,904,021 , ky 4,413,457 , la 4,649,676 , ma 6,745,408 , md 5,976,407 , me 1,330,089 , mi 9,909,877 , mn 5,458,333 , mo 6,063,589 , mt 1,023,579 , nc 9,943,964 , nd 739,482 , ne 1,881,503 , nh 87,137 2013 , nj 8,938,175 , nm 2,085,572 , ny 19,746,227 , oh 11,594,163 , ok 3,878,051 , or 3,970,239 , pa 12,787,209 , ri 1,055,173 , sc 4,832,482 , sd 853,175 , tn 6,549,352 , tx 27,695,284 , va 8,326,289 , vt 626,562 , wa 7,061,530 , wi 5,757,564 , wv 1,850,326 , wy 584,153 , the district of columbia 658,893 , and u.s . virgin islands 106,405 2010 census . the population estimates are from wikipedia for the year 2014 unless the year is cited . the sum is 302,681,519 . one must be 18 years or older to purchase lottery tickets . in accordance with the united states census bureau http://www.census.gov/population/age/ , in 2010 the population under 18 years was 24 percent . we estimate the number of potential buyers as @xmath657 . if the numbers of people from @xmath658 to @xmath659 buy the numbers of tickets from @xmath660 to @xmath661 , then the number of buyers is @xmath662 , the number of bought tickets is @xmath663 , the mean number of purchased tickets per buyer is @xmath664 , the number of not playing people is @xmath665 , and the number of potential tickets , which they did not buy , is @xmath666 . the fraction of purchased tickets to the total of purchased and potentially not purchased ones is @xmath667 . the latter is the fraction @xmath335 of those , who decided to play . with @xmath668 , @xmath669 . _ we conclude that lotteries can be used to estimate fractions of respondents selecting between random and constant variables or prospects_. for this problem , @xmath670 , @xmath671 , @xmath672 , @xmath673 , @xmath674 , @xmath675 , @xmath676=1.15e-7 . the big positive skewness @xmath75 overweights the certain loss prediction following from the tiny entropy @xmath246 . as long as @xmath677 is not won , the @xmath678 , @xmath679 , @xmath543 , and @xmath544 increase with @xmath47 . in contrast , in this hypothetical lottery , probabilities and dependent only on them @xmath75 , @xmath78 , and @xmath246 remain intact . @xmath680 , figure [ figmegamillionsjackpot ] , grows with acceleration before it reaches the second part of s - shaped curve , where the capital of _ dreamers _ exhausts . this acceleration is the increasing number of tickets from game to game and , therefore , increasing @xmath335 and @xmath681 . to estimate @xmath335 , multiply the values from column @xmath682 of table [ tbljackpot ] by @xmath683 . the previous paragraph evaluates @xmath684 . replacing @xmath685 with @xmath680 yields @xmath686 this formula implies that , if the jackpot is not won , then @xmath335 has a maximum : _ growing interest during the first phase is suppressed by a lack of the capital later_. setting @xmath687 , we get @xmath688 . substituting @xmath689 into equation [ eqlotteryf1 ] leads to @xmath690 , figure [ figtlotteryf1 t ] . the estimate @xmath651 @xmath656 500,000,000 is inaccurate because of a lack of data at the saturation of @xmath157 . twice jackpot was greater , table [ tbllottery ] . fixing @xmath543 and @xmath589 , equation [ eqthef1 ] draws a surface above the @xmath705-plane . a curve on the @xmath705-plane defines a dependence between @xmath537 and @xmath538 . this dependence is linear , equation [ eqdependenceab ] , if @xmath335 is also fixed . it is easy to prove that intersection of the surface and the plane , containing this straight line , and orthogonal to the @xmath705-plane is also a _ horizontal _ straight line . figure [ figf1abproblem3 ] illustrates these geometric properties for problem 3 , certainty effect . the surface looking curvy is a set of horizontal straight line segments , each located at its own characteristic height @xmath335 , with slope and intersect given by equation [ eqdependenceab ] . figure [ figf1abproblem3l ] presents a similar picture for problem 3 , reflection effect . dependence @xmath708 for fixed @xmath543 and @xmath589 does not have to be linear . if it is not linear or linear but has different slope and intercept than those , implied by equation [ eqdependenceab ] , then @xmath335 is not constant . let @xmath709 , and @xmath710 for @xmath711 , random two - point variable vs. constant . then , from equations [ eqmeandiff ] - [ eqexcesskurtosisdiff ] @xmath712 @xmath713 with equation [ eqpconsted ] we can design @xmath490 , change @xmath401 , @xmath476 , and keep @xmath543 , @xmath544 constant . from equations [ eqgamma12 ] this affects @xmath220th skewness and excess kurtosis . _ symmetry of the distribution can influence on respondents preferences . this leaves degrees of freedom for @xmath335 with constant @xmath543 and @xmath544 . _ from equations [ eqgamma12 ] , @xmath714 for @xmath715 , @xmath716 for @xmath717 , and @xmath718 for @xmath719 , see also figures [ figa1diffproblem3 ] , [ figa1diffproblem3b ] . let us rewrite equation for @xmath75 as @xmath720 and solve the latter @xmath721 . selection of sign and evaluation of root should correspond to the above @xmath75 inequalities and probability intervals . hence , @xmath722 from equation [ eqpconsted ] @xmath723 and @xmath724 from equations [ eqvariancediff ] and [ eqpconsted ] under the specified conditions @xmath725 two - point variable vs. constant problems matching @xmath550 , @xmath551 , @xmath726 of problem 3 are @xmath727 vs. @xmath728 , where @xmath729 , @xmath730 , @xmath731 are given by equations [ eqpgamma1 ] - [ eqap1gamma ] . for problem 3 the problems with @xmath563 , @xmath551 , @xmath732 are @xmath733 vs. @xmath734 , figure [ figedg ] . _ in both cases skewness @xmath735 can be arbitrary . it adds a degree of freedom to influence on @xmath539 and @xmath540_. keeping @xmath543 , @xmath544 , @xmath740 intact , variation of skewness @xmath735 changes @xmath741 , @xmath742 , equations [ eqap1gamma ] , [ eqaq12gamma ] , @xmath743 , @xmath744 , @xmath745 , @xmath746 , equations [ eqconstraintsab ] , [ eqbinaryconstraintsab ] . under the specified conditions these equations yield @xmath747 a random difference @xmath220 can be depicted as a horizontal line segment @xmath748 $ ] with an inner point @xmath543 . on figure [ figmmmgamma ] such segments are plotted relative each to other for different values of skewness @xmath75 for problems 3 and 3. we see that @xmath220 overlaps a greater positive area for @xmath749 in problem 3 than for @xmath750 in problem 3 . _ these diagrams support the reflection effect . _ let in a hypothetical case @xmath751 and @xmath752 . then , from equations [ eqamaxbmax ] @xmath753 is a hyperbola . this also creates a parametric dependence @xmath754 and @xmath755 and allows to express @xmath335 as a function of @xmath75 for constant @xmath543 and @xmath589 . figure [ figf1gamma ] illustrates fitting @xmath557 in problem 3 with @xmath756 and @xmath757 and fitting @xmath568 in problem 3 with @xmath758 and @xmath759 . more experimental data on fractions of respondents selecting between two - point random variables is needed to clarify further dependencies discussed so far . i would like to thank timur misirpashaev for the discussion on @xcite , the st . petersburg paradox and its credit component and pavel grosul for the discussion on relationships between fractions of respondents and wealth . 9 allais , maurice . le comportement de lhomme rationnel devant le risque , crtitique des postulats et axiomes de lecole americaine . _ econometrica _ , volume 21 , no . 4 , 1953 , pp . 503 - 546 . antonov , i. , a. , saleev , v. , m. an economic method of computing @xmath761-sequences . _ zhurnal vychsilitelnoi matematiki i matematicheskoi fiziki . _ volume 19 , no . 1 , 1979 , pp . 243 - 245 ( in russian ) ; _ ussr computational mathematics and mathematical physics _ , 1979 , volume 19 , no . 252 - 256 . arrow , kenneth , j. _ essays in the theory of risk - bearing_. chicago : markham , 1971 . cited by @xcite . bernoulli , daniel . exposition of a new theory on the measurement of risk . _ econometrica _ , volume 22 , no . 1 , january 1954 , pp . 23 - 36 . specimen theoriae novae de mensura sortis . _ commentarii academiae sceintiarum imperialis petropolitanae _ , tomus v , 1738 , pp . 175 - 192 is translated by louise sommer and includes footnotes made by karl menger editor , and translator . black , fischer , scholes , myron . the pricing of options and corporate liabilities . _ the journal of political economy _ , volume 81 , may - june 1973 , pp . 637 - 659 . buffon , georges louis leclerc comte de . _ essais darithmtique morale . in : histoire naturelle . supplment tome quatrime_. paris : de limprimerie royale , 1777 , pp . gallica bibliothque numrique http://gallica.bnf.fr/ark:/12148/bpt6k97517m . debreu , gerard . the coefficient of resource utilization , _ econometrica _ , volume 19 , n. 3 , 1951 , pp . 273 - 292 . doob , joseph , l. _ stochastic processes _ , copyright , 1953 by john wiley & sons , inc . , new york : john wiley & sons , 1990 . fabozzi , frank , j. _ bond markets , analysis and strategies . _ third edition , upper saddle river , new jersey : prentice - hall international inc . , 1996 feller , william . _ an introduction to probability theory and its applications ( vvedenie v teoriyu veroyatnostei i ee prilozheniya ) . _ volume 1 , moscow : mir , 1964 ( in russian ) . gnedenko , boris , v. , kolmogorov , andrey , n. _ limit distributions for sums of independent random variables_. moscow , leningrad : technico - theoretical literature governmental press , 1949 ( in russian ) . the book is translated to english by k.l . chung , cambridge , mass . : addison - wesley , 1954 . gnedenko , boris , v. _ the probability theory . [ kurs teorii veroyatnostei ] . _ moscow : nauka , 1988 ( in russian ) . hill , archibald , v. the possible effects of the aggregation of the molecules of hemoglobin on its dissociation curves . _ proceedings of the physiological society _ , january , 1910 , pp . joe , stephen , kuo , frances , y. constructing sobol sequences with better two - dimensional projections . _ siam j. sci . _ , volume 30 , no . 2635 - 2654 . kahneman , daniel , tversky , amos . prospect theory : an analysis of decision under risk . _ econometrica _ , volume 47 , no . 2 , march , 1979 , pp . 263 - 291 . kahneman , daniel . maps of bounded rationality : a perspective on institute judgment and choice . _ nobel lecture _ , december 8 , 2002 , pp . 449 - 489 , http://www.nobelprize.org/nobel_prizes/economics/laureates/2002/kahnemann-lecture.pdf . kelly , john , l. , jr . a new interpretation of information rate . _ bell system technical journal _ , volume 35 , no . 4 , july 1956 , pp . 917 - 926 . khinchin , aleksandr , ya . on petersburg game . _ matematicheskii sbornik _ , volume 32 , no . 2 , 1925 , pp . 330 - 341 ( in russian ) . khinchin , aleksandr , ya . _ the main laws of the probability theory . the laplace theorem , the law of big numbers , the law of iterated logarithm_. moscow , leningrad : state technical - theoretical publishing , 1932 ( in russian ) . khinchin , alexander , ya . the notion of entropy in probability theory . _ uspehi matematicheskih nauk _ , volume 8 , no . 3 , may - june 1953 , pp . 3 - 20 ( in russian ) . knight , frank , h. _ risk , uncertainty and profit_. reprints of economic classics , augustus m. kelly , bookseller , new york , n. y. 10019 : sentry press , 1964 , kolmogorov , andrey , n. ber das gesetz des iterierten logarithmus , _ mathematische annalen _ , volume 101 , 1929 , pp . 126 - 135 . kolmogorov , andrey , n. _ foundations of the theory of probability . osnovnyie ponyatiya teorii veroyatnostei . _ 2nd edition , moscow : nauka , 1974 ( in russian ) . kolmogorov , andrey , n. , zhurbenko , igor , g. , prokhorov , alexander , v. _ introduction to probability theory . _ moscow : nauka , 1982 ( in russian ) . kolmogorov , andrey , n. , uspenskii , vladimir , a. algorithms and randomness . theory of probab . volume 32 , no . 3 , 1987 , pp . 389 - 412 ( translated from russian journal by bernard seckler ) . korn , g. , korn t. , _ mathematical handbook for scientists and engineers . definitions , theorems , and formulas for reference and review _ , 2nd ed . , new york : mcgraw - hill book company , 1968 . lefvre , edwin . _ reminiscences of a stock operator_. new york : john wiley & sons , inc . , 1993 . copyright 1993 , 1994 by expert trading , ltd . originally published in 1923 by george h.doran and company . livermore , jesse , l. _ how to trade in stocks . the livermore formula for combining time element and price_. new york : duell , sloan & pearce , 1940 . malkiel , burton . g. _ a random walk down wall street . the time - tested strategy for successful investing_. 9th ed . new york : w.w.norton & company , 2007 . markowitz , harry , m. the utility of wealth . _ journal of political economy _ , volume 60 , no . 2 , april 1952 , pp . 151 - 158 . menger , karl . das unsicherheitsmoment in der wertlehre . _ nationaloeken , journal of economics _ , volume 5 , no . 4 , september 19 , 1934 , pp . 459 - 485 . murphy , john , j. _ technical analysis of the financial markets . a comprehensive guide to trading methods and applications_. new york : new york institute of finance , 1999 . neftci , salih n. naive trading rules in financial markets and wiener - kolmogorov prediction theory : a study of `` technical analysis '' . _ journal of business _ , volume 64 , no . 4 , october 1991 , pp . 549 - 571 . peters , ole . the time resolution of the st petersburg paradox , _ philosophical transactions of the royal society a _ , volume 369 , october 2011 , pp . 4913 - 4931 . packwood , daniel , m. moments of sums of independent and identically distributed random variables . _ arxiv , mathematics , statistics theory _ , version 2 , january 14 , 2012 , pp . 1 - 13 , http://arxiv.org/abs/1105.6283 . peters , ole . menger 1934 revisited , _ arxiv , quantitative finance , risk management _ , march 23 , 2011 , pp . 1 - 16 , http://arxiv.org/abs/1110.1578 . prokhorov , yurii , v. the law of a large numbers and the law of the iterated logarithm , _ uspekhi matematicheskih naul _ , volume 38 , no 4 , 1983 , pp . 281 - 286 ( in russian ) or _ russian mathematical surveys _ , volume 38 , no . 319 - 326 . richards , f. , j. a flexible growth function for empirical use , _ journal of experimental botany _ , volume 10 , no . 2 , 1959 , pp . 290 - 300 . salov , valerii . _ modeling maximum trading profits with c++ : new trading and money management concepts . _ hoboken , new jersey : john wiley and sons , 2007 . salov , valerii , v. trading system analysis : learning from perfection . _ futures magazine _ , vol . 11 , november , 2011 , pp . 34 - 39 , 43 . salov , valerii , v. high - frequency trading in live cattle futures . _ futures magazine _ , vol . 6 , may 2012 , pp . 26 - 27 , 31 . salov , valerii , v. optimal trading strategies as measures of market disequilibrium , _ arxiv , quantitative finance , general finance _ , december 6 , 2013 , pp . 1 - 222 , http://arxiv.org/abs/1312.2004 . salov , valerii . `` the gibbon of math history '' . who invented the st . petersburg paradox ? khinchin s resolution . _ arxiv , mathematics , history and overview _ , march 11 , 2014 , pp . 1 - 17 , http://arxiv-web3.library.cornell.edu/abs/1403.3001 . samuelson , paul . petersburg paradox as a divergent double limit . _ international economic review ( blackwell publishing ) _ , volume 1 , no . 1 , 1960 , pp . samuelson , paul . petersburg paradoxes : defanged , dissected , and historically described . _ journal of economic literature ( american economic association ) _ , volume 15 , no . 1 , 1977 , pp . sharpe , william , f. _ investors and markets . portfolio choices , asset prices , and investment advice _ , princeton and oxford : princeton university press , 2007 . smitten , richard . _ how to trade in stocks . jesse livermore . with updates & commentary by richard smitten . _ new york : mcgraw - hill , 2001 . thomson , jesse , h. the livermore system . _ stock & commodities _ , volume v , no . 1:4 , may - june 1983 , pp . 82 - 85 . sobol , ilya , m. uniformly distributed sequences with an additional uniform property . _ zhurnal vychsilitelnoi matematiki i matematicheskoi fiziki . _ volume 16 , no . 5 , 1976 , pp . 1332 - 1337 ( in russian ) ; _ ussr computational mathematics and mathematical physics _ , volume 16 , no.5 , 1976 , pp . 236 - 242 . sobol , ilya , m. _ points that uniformly fill a multidimensional cube . _ moscow : znanie , 1985 ( in russian ) . tversky , amos , kahneman , daniel . , advances in prospect theory : cumulative representation of uncertainty , _ journal of risk and uncertainty , vol . 5 _ , 1992 , pp . 297 - 323 . varma , jayanth , r. time resolution of the st . petersburg paradox : a rebuttal . _ the working paper series of indian institute of management , ahmedabad - 380015 , india _ , w.p . no . 2013 - 05 - 09 , may 2013 , pp . 1 - 5 , http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2270980 . vince , ralph . _ the mathematics of money management . risk analysis techniques for traders_. new york : john wiley & sons , inc . , 1992 . vince , ralph . _ the new money management . a framework for asset allocation_. new york : john wiley & sons , inc . , 1995 . williams , a. , c. attitudes toward speculative risks as an indicator of attitudes toward pure risks . _ journal of risk and insurance _ , volume 33 , 1966 , pp . 577 - 586 . * valerii salov * received his m.s . from the moscow state university , department of chemistry in 1982 and his ph.d . from the academy of sciences of the ussr , vernadski institute of geochemistry and analytical chemistry in 1987 . he is the author of the articles on analytical , computational , and physical chemistry , the book modeling maximum trading profits with c++ , _ john wiley and sons , inc . , hoboken , new jersey _ , 2007 , and papers in _ futures magazine _ and _	the prospects of kahneman and tversky , mega million and powerball lotteries , st . petersburg paradox , premature profits and growing losses criticized by livermore are reviewed under an angle of view comparing mathematical expectations with awards received . original prospects have been formulated as a one time opportunity . an award value depends on the number of times the game is played . the random sample mean is discussed as a universal award . the role of time in making a risky decision is important as long as the frequency of games and playing time affect their number . a function of choice mapping properties of two - point random variables to fractions of respondents choosing them is proposed .
a model was proposed at the vatican conference on stellar populations ( oconnell 1958 ) to explain the period - ratio dichotomy in globular cluster rr lyrae variables , studied by oosterhoff ( 1939 , 1944 ) , following its discovery by grosse ( 1932 ) and hachenberg ( 1939 ) . the postulate of the model was that the difference in the observed mean periods between the two oosterhoff groups could be understood if the absolute magnitude levels of the groups differed by @xmath90.2 mag at constant temperature , the variables in the long period group being brighter ( sandage 1958 ) . in the discussion , martin schwarzschild ( 1958 ) asked if , instead , the period ratio could be explained by a temperature difference with the long period group ( oost ii ) being cooler on average than oosterhoff i clusters , keeping the luminosities the same . i replied that a color shift would have to be at the level of 0.09 mag in b - v toward the red in the long period oosterhoff group . although a later calculation ( sandage 1993b , hereafter s93b , footnote 1 ) gave 0.06 mag for the required color shift , both 0.06 and 0.09 mag were outside the bounds permitted by the observations of clusters in the two oosterhoff period groups known at the time for m3 and m92 ( arp , baum , and sandage 1953 , sandage 1953 ) and for five other clusters ( arp 1955 ) over the relevant metallicity range of -1.5 @xmath8 [ fe / h ] @xmath8 -2.2 . a near zero color difference of the blue fundamental edge of the strip for such clusters at the level of @xmath90.02 mag in b - v was later confirmed from more precise photoelectric data of horizontal branch stars , and better determinations of the e(b - v ) ( sandage 1969 ) . it was this conclusion that led to the near canonical assumption in the late 1980s that one should use a constant mean temperature of log [email protected] for the mid point of the strip in comparing the predictions from theoretical models of the zero age horizontal branch ( zahb ) with the observations ( s82a , s90b ) . however , this assumption of constant temperature for all metallicities led to the vigorous debate in the decades of 1985 and 1995 concerning the adequacy of the extant zahb models or , in fact , whether the period shifts claimed by the observers using data on amplitudes were real ( caputo 1988 ; lee 1990 ; lee , demarque , & zinn 1990 , hereafter ldz ; brocato , castellani , & ripepi 1994 , 1996 ) . nevertheless , _ continuous _ period shifts with metallicity , not just a _ dichotomy _ of mean period ratios , had been discovered by preston ( 1959 , his fig . 4 ) for field rr lyraes with the non - zero slope of dlogp / d([fe / h ] ) @xmath9 -0.10 in the log p - metallicity relation . the continuum with this slope was later shown beyond doubt for the bright field rr lyraes from the larger sample of layden ( 1995 , his fig . 1 ) . other determinations of the slope at dlogp / d([fe / h ] ) = -0.12 @xmath7 0.02 for the period - metallicity correlation had also been derived from a variety of independent data for rrl in clusters ( sandage , katem , & sandage 1981 , hereafter sks ; s81b , table 7 , s82a , b ) . some of these determinations were from amplitude- metallicity correlations , some from rise - time - metallicity correlations , and some from period - temperature correlations within the strip as a function of metallicity . nevertheless , debate on the validity of period shifts determined using amplitudes remain . now the focus has shifted from questioning the validity of the zahb _ theoretical models _ , to the role of evolution off the initial non - evolved sequence ( eg . ldz 1990 ; sandage 1990a ; brocato et al . 1994 , 1996 ; clement & sheldon 1999a , b ; cacciari , corwin , & carney 2004 ) . the evidence for a period - amplitude - metallicity relation at various temperatures remains as it was set out in 1981 ( s81a ) ; it has been rediscovered by carney , storm , & jones [ 1992 , their eq . ( 16 ) ] , and by alcock et al . ( 1998 ; alcock 2000 ) . it is useful to recall the strong debate in the 1990s concerning the observational data on period shifts . the extant zahb models did not predict such shifts when the predictions were made in a particular way , and the theoreticians , believing their models , questioned the interpretation of the observations on amplitudes and temperatures . the reasons were these . the early pioneering models of the luminosity level of the zahb level for different metallicitites had been made by sweigart and gross ( 1976 ) followed by sweigart , renzini , & tornambe ( 1987 , hereafter srt ) , bencivenni et al . ( 1991 ) , and dorman ( 1992 ) . a more complete list of references is in srt . modern updates of these pioneering papers are , among others , those of bono et al . ( 1997a ) , vandenberg et al . ( 2000 ) , and catelan , pritzl & smith ( 2004 ) , each of which contain references to other recent models . the period shifts with metallicity could be predicted from these models in the following way . the models give calculated tracks for the zahb in hr diagrams of log l , log t@xmath11 for different metallicities . they also give mass as a function of [ fe / h ] along the tracks . imposing a constant temperature line across the tracks , generally at log t@xmath11 = 3.85 , permitted log l and mass to be read along this fixed temperature line . a pulsation equation , p(l , m , t@xmath11,[fe / h ] ) that relates period , luminosity , and mass , is then used to predict how periods along the constant temperature line should vary with metallicity . these predictions gave a near zero slope ( s93b , fig . 6b ) to a period - metallicity correlation rather than the value of dlogp / d([fe / h ] ) = -0.12 derived from the observations . the effect of the increased luminosity on the period as the metallicity is decreased was nearly cancelled by the corresponding increased mass along the zahb in these models , shown explicitly by simon and clement ( 1993 , the first and second unnumbered equations in their section 8) . and even if the effect of luminosity evolution from the zahb was included as in gratton , tornambe , & ortolani ( 1986 ) , and ldz ( 1990 , their table 2 ) , these models predicted a slope of the period - metallicity relation to be only dlogp / d([fe / h ] ) = -0.05 , but , again , _ when read at constant temperature_. this was less than half of what is observed . the solution of reconciling the model predictions with the observations , obvious now but not at the time , was to drop the condition of reading the models at constant temperature , but rather to let the temperature be cooler for more metal poor rrl at their higher luminosities . this solution had been proposed for rrc variables by simon and clement ( 1993 ) in their seminal paper on the discovery of methods to obtain luminosities , chemical abundances , masses , and temperature from various fourier components of the light curves for rrc variables . they stated , based on their theoretical expectations , that `` it is not correct to formulate a period shift at constant temperature ( eg . sandage 1990b , apj 350 , 631 ) . '' the same solution of cooler temperature for the oosterhoff ii variables was proposed independently using the pulsation equation for rrab variables at the fundamental blue edge ( s93a , b ) . it was shown there that a variation of temperature at the bfe at the rate of dlogt@xmath11/d([fe / h ] ) = -0.012 would largely reconcile the prediction from the models and the observations of period - shifts , if a mild evolution is also introduced ( s93b , fig . 6a compared to 6b ) . the purpose of the present paper is to return to the problem , demonstrating more directly the need for a temperature variation with metallicity at the fundamental blue edge of the instability strip . in making the argument we reverse the dependent and independent variables in a pulsation equation by making t@xmath11 the independent variable to be determined by assuming the observed dependencies of log p , log l , and log mass on metallicity , all of which are better known now than in s93a . the organization of the paper is as follows . section 2 shows an updated period - metallicity correlation using the data in the catalog of 302 field rrl with measured metallicities by layden ( 1994 ) on the scale of zinn and west ( 1984 ) . there are nearly three times the number of field stars in the layden sample than in the catalog by blanco ( 1992 ) that was used in s93a . section 3 shows a similar period - metallicity distribution as in 2 for rrl in globular clusters from the updated catalog of cluster variables by clement ( 2001 ) that merges the data from the third catalog of sawyer hogg ( 1973 ) with new data complete to 2000 from the literature , adding about 30% to the data base used in s93a . section 4 sets out four modern calibrations of the absolute magnitude - metallicity relation for rrl by caputo et al . ( 2000 ) , clementini et al . ( 2003 ) , catelan , pritzl , & smith ( 2004 ) , and mcnamara et al . ( 2004 ) , many of which emphasize the need for a non - linear m@xmath3([fe / h ] ) luminosity - metallicity relation both for the zahb and for the evolved hb . these four are a representative subset of many other current calibrations , reviewed by cacciari and clementini ( 2004 ) . a new calibration of the mass - metallicity relation calculated by bono et al . ( 1997a ) is given in 5 , and compared with that of dorman ( 1992 ) that was used in s93a . in 6 we combine the period - metallicity , m@xmath3-metallicity , and mass - metallicity relations from 2 - 5 with a pulsation equation to predict the temperature-([fe / h ] ) relation at the fbe . temperatures are transformed to ( b - v)@xmath2 colors in 7 using an adopted color - metallicity - temperature calibration . the predicted color - metallicity relation is compared in 8 with the observed ( b - v)@xmath2 colors for both field and cluster rrl . in 9 we turn the problem around by using the new correlations of period , temperature , and mass at the fbe to predict the rrl luminosity , again from the pulsation equation using a new temperature variation at the fbe that has been fitted empirically here to the observational data , rather than using the predicted variation in 6 & 7 . section 10 contains a discussion of the observed difference in the zero - point of the log period-([fe / h ] ) envelope locus ( figs . 1 - 4 ) between field and cluster variables , and the presence or absence of the oosterhoff period gap in the cluster and the field star samples , respectively , in terms of a proposed different formation history of the samples . eight research points made in this paper are summarized in 11 . in the 1993 study ( s93a ) , i had used the correlation of mean period with metallicity for 110 field rr lyraes for which blanco ( 1992 ) had determined improved e(b - v ) reddenings . he also had homogenized the existing [ fe / h ] metallicity values based on butler s ( 1975 ) calibration of the preston @xmath0s metallicity parameter . the blanco - butler [ fe / h ] scale averages 0.28 dex more metal rich than the globular cluster scale of zinn and west ( 1984 ) . i had used the two different metallicity scales , one for the field variables and the other for the clusters , although i had transformed the blanco - butler scale to that of zinn and west . nevertheless , the field star log p-[fe / h ] diagrams were kept on the blanco - butler system , which , at times was inconvenient . this inconvenience is overcome here , and the sample of field stars has been increased by a factor of three by using the more recent data base of layden ( 1994 , 1995 ) which is on the zinn - west metallicity scale . layden measured metallicities for his complete sample of 302 rrab lyrae field variables whose galactic latitudes are more than @xmath12 , and whose apparent magnitudes are brighter than @xmath13 = 13.5 . layden s list is from the fourth edition of the general catalog of variable stars ( kholopov 1985 ) . the log p-[fe / h ] distribution for layden s field star sample is shown in figure 1 . a linear envelope line is drawn by eye as the locus of the shortest period variables at a given [ fe / h ] value . this would define the period at the fundamental blue edge of the instability strip if there is no `` hysteresis '' within in a possible `` either - or '' transition zone for evolution tracks between the fundamental and first overtone variables . the equation of the envelope line in fig . 1 is @xmath14).\ ] ] editor : place figure 1 here the mid - point ridge line , shown as a dashed line in fig . 1 , is found by averaging the log p values in narrow intervals of [ fe / h ] . the data are listed in table 1 . the equation of this line is , @xmath15).\ ] ] also listed in table 1 are the mean ( b - v)@xmath2 colors within each metallicity bin , calculated from the photoelectric summary of individual colors listed in column 5 of nikolov , buchantsova , and frolov ( 1984 , hereafter the sophia catalog ) , using the reddening corrections for each star based on the absorptions ( divided by 3 ) measured by layden ( 1994 ) . the linear fit ( the solid line ) to the shortest period envelope data in fig . 1 is good , but not perfect . a better fit is achieved for metallicities smaller than -1.8 by bending the envelope toward longer periods . the straight lines in fig . 2 have the equations , put by eye , of , @xmath16),\ ] ] for [ fe / h ] ) more metal rich than -1.8 , and two dashed lines whose equations are @xmath17),\ ] ] and @xmath18)\eqnum{4'}\ ] ] for [ fe / h ] ) more metal poor than -1.8 . equation ( 3 ) is the same as eq . ( 1 ) but its application stops for [ fe / h ] smaller than -1.8 . editor : place figure 2 here a still better fit is the continuous parabolic envelope over the entire metallicity range from zero to -2.4 shown in fig . 3 , whose equation is , @xmath19)^2,\ ] ] used previously for a different purpose ( sandage 2004 , fig . 5 ) using a subset of the present field star data . editor : place figure 3 these various envelope lines will be used in 6 for the predictions of the temperature at the fbe . although the number of stars ( 302 ) in figs . 1 - 3 is sufficient to show the _ continuous _ variation of mean period with metallicity ( rather than a dichotomy into two period groups ) , nevertheless it is not yet sufficient to define the position of the short period envelope at the low metallicity end for [ fe / h ] @xmath20 -2 with precision . a larger sample of rr lyrae variables is needed and is available in globular clusters . the earlier sample from the third catalog of cluster variables by sawyer hogg ( 1973 ) was used in s93a for this purpose . we have repeated the log period - metallicity diagram for the globular cluster variables in fig . 4 , increasing the cluster sample from 760 stars used by s93a to 919 here by adding the new variables listed in the 2001 updated catalog by clemment ( 2001 ) cited earlier . the metallicity scale is that of zinn and west , taken from the listing in the catalogs of harris ( 1996 ) . the parabolic envelope of equation ( 5 ) is overlaid on the data to compare the cluster data with the field star data in 2 ( figs . 1 - 3 ) . editor : place figure 4 here two significant differences are evident by comparing fig . 4 for the cluster variables with figs . 1 - 3 for the field variables . the gap in the [ fe / h ] distribution ( that is , in the vertical distribution of points ) in fig . 4 between [ fe / h ] of -1.7 and -2.0 for the cluster variables is absent in figs . 1 - 3 for layden s field variables . it is this gap that makes the oosterhoff _ dichotomy _ into two discrete period groups so evident , partially masking a _ continuous _ variation of mean periods with metallicity . the parabolic envelope that fits the short period boundary of the distribution of periods for the field variables in fig . 3 fails to fit the data for the globular cluster variables . the cluster variables have longer periods at the envelope edge by @xmath0log p @xmath9 0.029 dex compared with equation ( 5 ) . the difference has a significance of about 4 sigma . there are several possibilities for explaining these two differences between field and cluster variables . point ( 1 ) . the very few variables in the cluster sample in fig . 4 between [ fe / h ] = -1.7 and -2.0 is due , as is well known , to the details of the morphology of cluster horizontal branches through the instability strip in this metallicity range . most clusters with intermediate metallicities with -1.3 @xmath8 [ fe / h ] @xmath8 -1.7 have horizontal branches well populated on both sides of the rr lyrae instability strip , whereas clusters with [ fe / h ] between -1.7 and -2.0 produce very few variables because the hb does not penetrate the instability strip . clearly , the absence of this oosterhoff period gap in the field variables shows that the morphology of the horizontal branch of the field rrl _ must _ differ fundamentally from that in globular clusters in this metallicity range . the reasons are not presently understood , but a hint may be the variation of the hb morphology with metallicity of the highly unusual globular clusters ngc 6388 and ngc 6441 ( rich 1997 et al . ; pritzl et al . 2000 , 2001 ) where the hb penetrates the instability strip despite its high metal abundance of [ fe / h ] @xmath9 -0.5 . this behavior is contrary to the horizontal branches of `` normal '' globular clusters such as 47 tuc , and the clump hbs of high metallicity galactic clusters . evidently , there is some parameter - variation ( such as the value of the core he abundance , the core mass , variations in deep mixing , etc . ) in these two highly abnormal globular clusters that causes the hb morphology to differ from that of most other of the clusters in the galaxy . point ( 2 ) . the observed period difference of 0.029 dex between cluster and field variables at the fbe can be accomplished in several ways from the pulsation eq . ( 16 ) in 5 . among them is a luminosity difference of @xmath0log l = 0.027 at constant temperature and mass , or a temperature difference of @xmath0log t@xmath11 = 0.007 at constant l and mass , or a mass difference of @xmath0log mass = 0.034 at constant l and t@xmath11 , or combinations of each . it need only be recalled that gratton ( 1998 ) has already discussed the possibility that there may be `` a real difference between the luminosity of the horizontal branch for clusters and the field '' . point ( 3 ) . a third possibility is that the metallicity scale due to layden ( 1994 ) in figs . 1 - 3 for the field variables is not the same as the used by harris ( 1996 ) for the clusters . the difference seen between figs . 1 - 3 and fig . 4 would disappear if the field points in figs.1 - 3 were moved downward by @xmath0[fe / h ] @xmath9 0.3 dex , or if the cluster data were made more metal poor by the same amount . but no systematic shifts by this amount are possible if the statements by both layden and by harris are correct that their scales are both tied to the scale of zinn and west ( 1984 ) . many calibrations , both observational and theoretical , of rr lyrae luminosities as functions of metallicity have been made since the linear calibration via the pulsation equation was made in s93a . all of the second and third generation theoretical models for the position of the _ zero - age _ horizontal branches show that the relation between m@xmath21 ( and hence closely m@xmath3 ) and [ fe / h ] must be non linear , i.e. the branch loci of the zahb tracks are more closely stacked in luminosity at the low metallicity end of the distribution than at the high metallicity end , for equal intervals of [ fe / h ] , even as the low metallicity zahbs are brighter than those of high metallicity . examples , among many , are the models of lee , demarque & zinn ( 1990 ) ; castellani , chieffi , & pulone ( 1991 ) ; bencivenni et al . ( 1991 ) ; dorman ( 1992 ) ; caputo et al . ( 1993 ) ; caloi , dantona & mazzitelli ( 1997 ) ; salaris , deglinnocenti & weiss ( 1997 ) ; cassisi et al . ( 1999 ) ; demarque et al . ( 2000 ) ; vandenberg et al . ( 2000 , figs . 2 , 3 , and 20 ) ; catelan , pritzl , & smith ( 2004 ) . a graphical summary showing several of the predicted m@xmath3(hb)-[fe / h ] relations derived from these models is given by cacciari & clementini ( 2004 ) . observational and/or semi - theoretical calibrations that also show the need for a non - linear m@xmath3([fe / h ] ) relation are , among others , the studies by caputo ( 1997 ) ; gratton et al . ( 1997 ) : de santis & cassisi ( 1999 , fig . 15 ) , caputo et al . ( 2000 ) , and mcnamara et al . ( 2004 ) . we use several of these calibrations in the pulsation equation to predict the temperature at the fbe of the instability strip in the next section , replacing the linear calibration derived in s93b of , @xmath22).\ ] ] we use the non - linear calibrations of caputo et al . ( 2000 ) , catelan et al . ( 2004 ) , and mcnamara et al . ( 2004 ) from their tie - in to delta scuti stars , and the linear calibration of clementini et al . ( 2003 ) from their calibration via the rr lyrae stars in the lmc . the equations are assumed to be as follows . we have approximated the graphical representation of the calibration of caputo et al . ( 2000 , their fig . 2a ) by a parabola , as , @xmath23 ) + 0.242 ( fe / h])^2.\ ] ] using m@xmath21 = m@xmath3 + bc , with bc = 0.06 + 0.06 ( [ fe / h ] ) , and m@xmath21(sun ) = 4.75 , it follows from ( 7 ) that @xmath24)-0.097 ( [ fe / h])^2.\ ] ] the adopted bolometric correction here is the same as was used by sandage and cacciari ( 1990 ) , justified there . the tables by bell and tripicco in sandage et al . ( 1999 , table 6 ) show , of course , that the bc varies also with temperature and atmospheric turbulent velocity as well as with metallicity . interpolation within the tables for turbulent velocity of 5 km s@xmath25 at a color of ( b - v)@xmath26 = 0.24 ( which is close to the blue edge ) gives bc = 0.05 + 0.06([fe / h ] ) . using a turbulent velocity of 1.7 km s@xmath25 gives bc = 0.06 + 0.069 ( [ fe / h ] ) , again at ( b - v)@xmath26 = 0.24 . these justify our continued use of 0.06 + 0.06 ( [ fe / h ] ) here . the calibration equations from the theoretical models of the zahb by catelan et al . ( 2004 ) using an oxygen enhancement of [ alpha / fe ] = 0.3 and [ fe / h ] = [ m / h ] - 0.213 give , @xmath27 ) + 0.108 ( [ fe / h])^2,\ ] ] from their eq . ( 12 ) after some reduction to change [ m / h ] into [ fe / h ] . this transforms in the same way as above to @xmath28 ) - 0.043 ( fe / h])^2.\ ] ] the linear calibration of clementini et al . ( 2003 ) using the observed rr lyraes in the lmc and an adopted lmc modulus of ( m - m ) = 18.54 gives , @xmath29),\ ] ] or , @xmath30).\ ] ] equations ( 11 ) and ( 12 ) refer to the mean ( evolved ) hb . to change to the level of the zahb we must make the equations fainter by about 0.1 mag in v , or 0.025 in log l ( s90a ) . the equations for the two - line calibration of mcnamara et al . ( 2004 ) are ; m@xmath3 = 0.50 independent of [ fe / h ] for metallicities more metal poor than -1.5 , and m@xmath3 = 1.13 + 0.42 ( [ fe / h ] ) for [ fe / h ] between -0.5 and -1.5 . these transform to @xmath31),\ ] ] for [ fe / h ] @xmath20 -1.5 , and , @xmath32),\ ] ] for [ fe / h ] ) @xmath8 -1.5 . to make the calculation of t@xmath11(fbe ) from the pulsation equation , we also need the mass as a function of metallicity . in s93b , i adopted the zahb calculation of rr lyrae mass at log t@xmath11 = 3.85 by dorman ( 1992 ) as log mass = -0.288 - 0.059 ( [ fe / h ] ) , where he had used enhanced oxygen abundances and modern opacity tables . more recently this theoretical calculation of rr lyrae masses has been verified and extended by bono et al . ( 1997a ) , again using enhanced oxygen abundances and the latest opacity tables cited by them . the resulting masses on the zahb at log t = 3.85 are listed in table 2 of caputo et al . ( 2000 ) , and are adopted here . the resulting linear mass equation over the relevant metallicity range is , @xmath33),\ ] ] which we adopt . this deviates by not more than 0.02 dex in mass from dorman ( 1992 ) over the metallicity range from 0 to -2.5 . confirmation that the mass of metal poor rr lyraes is larger than that for metal rich variables is available from analyses of the observations of double mode variables via the peterson ( 1973 ) diagram and the subsequent calculations by peterson ( 1978 , 1979 ) , cox et al . ( 1980 ) , cox et al . ( 1983 ) , and cox ( 1987 ) and references therein . the calculations by kovacs et al . ( 1992 ) that show the sensitivity of the results to various chemical composition mixtures are illustrated in fig . 4 of s93b . the masses from eq . ( 15 ) are for the zahb at log t@xmath11 = 3.85 . they will not be precise even at this fixed temperature if evolution brings stars from different parts of the hb into the instability strip . we neglect the detail in this reconnaissance study ; it does not change the sense of the argument that follows in figs . 5 & 6 of the next section . many new versions of the pulsation equation have been calculated since the pioneering paper by van albada and baker ( 1973 ) ( eg . chiosi et al . 1992 ; simon & clement 1993 ; bono et al . 1997b with earlier references ) , but each give nearly identical dependencies of period on temperature , luminosity , and mass . we continue to use the van albada and baker formulation here . their equation , transposed to make temperature the independent variable , is @xmath34 if the dependencies of l , p , and mass on metallicity have been expressed as relations at the fbe of the instability strip , either on the zahb or on the mean hb as evolved from the zahb , then eq . ( 16 ) will give the temperature at that edge . we make the assumption here that the hb is truly horizontal in v magnitudes and , therefore , that equations ( 8) , ( 10 ) , ( 12 ) , ( 13 ) , and ( 14 ) also define the bolometric luminosity at the blue fundamental edge as well as at the mean strip position to which many of the luminosity calibrations of the aforesaid equations apply . this is a fully adequate approximation for the restricted purpose of this paper which is only to show the _ need _ to invoke a variable temperature with metallicity at the fbe of the strip rather than to finalize a definitive determination of it . indeed , uncertainties presently exist in the temperature- color calibration ( 7 ) at the level of 0.02 mag in b - v . similar uncertainties also exist at the 0.02 mag level in the observational data for the b - v color at the fbe edge due to uncertainties in the reddening . in addition , theoretical uncertainties exist in the constants of the pulsation equation including its zero point [ i.e. 3.304 in eq . these all conspire against a definitive `` final '' calculation of t@xmath11(bfe ) . nevertheless , all log l calibrations to date , and all p-([fe / h ] ) displays such as in figs . 1 - 4 , require the same qualitative variation of t@xmath11 and color of the fundamental blue edge with metallicity which is the purpose of this paper , which we now demonstrate . to illustrate the nature of the 30 solutions for t@xmath11 in the completely filled log p([fe / h ] ) and log l([fe / h ] ) parameter space , it is sufficient to choose only a subset of the p([fe / h ] ) and l([fe / h ] ) combinations . here the subsets are divided into two groups . in the first we keep the luminosity calibration fixed using eq . ( 10 ) by catelan et al . ( 2004 ) , and use with it the four different log p - metallicity envelopes from figs . 1 - 3 . these loci are ( a ) the log p([fe / h ] ) envelope in eq . ( 1 ) as in fig . 1 ; ( b ) the three - line envelopes of eqs . ( 3 ) , ( 4 ) and ( 4 ) in fig . 2 ; and ( c ) the parabolic envelope of eq . ( 5 ) in fig . 3 . inserting the catelan et al . ( 2004 ) log l([fe / h ] ) relation from equation ( 10 ) , the bono et al . ( 1997a ) mass equation from eq . ( 15 ) , and the log p([fe / h ] ) relations from eqs . ( 1 ) , ( 3 ) , ( 4 ) , ( 4 ) and ( 5 ) into the pulsation equation ( 16 ) gives the run of predicted log t@xmath11 with [ fe / h ] shown in fig . 5 . the relation using the midpoint correlation of [ fe / h ] with period [ eq . ( 2 ) ] is also shown . editor : place figure 5 here as an example , the predicted log t@xmath11 relations at the fbe using the parabolic envelope line of eq . ( 5 ) , the bono mass eq . ( 15 ) , and the catelan et al . luminosity calibration of eq . ( 10 ) is , @xmath35)^2,\ ] ] shown in fig . 5 as the dashed line marked number 3 . alternatively , by keeping the envelope log p([fe / h ] ) relation fixed using eq . ( 5 ) for the parabolic fit , and using five different log l(fe / h ] ) calibrations give the predicted run of log t@xmath11 shown in fig . editor : place figure 6 here the characteristics of all the solutions in both diagrams is that the predicted temperature at the fundamental blue edge varies with [ fe / h ] , first becoming hotter as [ fe / h ] weakens from 0 to within the range of -1.2 to -1.5 , and then cools again for still decreasing metal abundance . at no metallicity is it constant at the fbe with varying [ fe / h ] , although its variation is less than @xmath0log t@xmath11 = 0.01 between [ fe / h ] of -0.5 and -2.0 along any given line of fixed luminosity calibration . a number of new calibrations of how b - v color varies with log t@xmath11 , metallicity , surface gravity , and atmospheric turbulent velocity have been made since the relation , taken from an unpublished calibration of bell , as quoted by butler et al . ( 1978 , their fig . 11 ) , was used by s93b ( eq . 4 there ) . six of these modern calibrations have been compared by cacciari , corwin , & carney ( 2004 , their fig . the total differences between the various calibrations in the predicted b - v colors at a given temperature near 7000 k ( log t = 3.845 ) is large at 0.08 mag . or read the other way , at a color of b - v = 0.28 , the temperatures range between 6950 k ( log t@xmath11 = 3.842 ) to 6400 k ( log t = 3.806 ) for the calibrations that are compared by cacciari et al . ( 2004 ) . the hottest calibrations for a given color are those of sandage , bell , and tripicco ( 1999 , hereafter sbt ) and castelli ( 1999 ) . the coolest is that of montegriffio et al . intermediate are the calibrations of carney , storm , & jones ( 1992 ) , and sekiguchi & fukugita ( 2000 ) . the bulk of the comparisons show a smaller spread of 0.04 mag in b - v between the calibrations of sbt , sf , and csj . we adopt the intermediate calibration of csj ( 1992 , their eq . 13 ) , which , in the temperature range of interest ( log t@xmath11 between 3.85 and 3.80 ) , can be put with sufficient accuracy into the logarithmic form rather than as function of 5040/t as they give it , to be , @xmath36 ) + 10.423,\ ] ] in the interval 0.20 @xmath20 b - v @xmath20 0.30 . comparison between the sbt and the sf scales read at the fixed color of b - v = 0.24 at [ fe / h/ ] = -1.5 ( near the mid range of the relevant parameter space ) , give log t@xmath11 = 3.848 ( t@xmath11 = 7046 k ) for csj from eq . ( 18 ) , 3.857 ( t@xmath11 = 7194 k ) for sbt , and 3.852 ( t@xmath11 = 7112 k ) from sf . hence , the temperatures on the three scales differ by @xmath0log t@xmath11 = 0.009 at b - v = 0.24 . the coefficient of [ fe / h ] in eq . ( 18 ) is , of course , a strong function of temperature , increasing due to fraunhofer blanketing and atmospheric backwarming with decreasing temperature . the value of 0.038 in eq . ( 18 ) is a compromise between 0.026 at b - v = 0.24 for the sbt models ( interpolating in their table 6 ) and 0.047 in the sf calibration , a value that is surely too large at the hot temperature corresponding to b - v = 0.24 . using eq . ( 18 ) we transform the t@xmath11 at the fbe for each of the equations that generate the family of curves in figs . 5 and 6 into the predicted run of b - v colors with metallicity at the fbe . to illustrate one equation of the family we insert the base - line parabolic equation for the log p([fe / h ] ) envelope relation of eq . ( 5 ) and the catelan et al . luminosity calibration of eq . ( 10 ) into eq . ( 18 ) to obtain the predicted run of b - v color at the fbe for that combination as , @xmath37 ) + 0.050 ( fe / h])^2,\ ] ] which is the dashed line 3 in fig . the complete sets of the families from figs . 5 and 6 , but now in b - v , are shown in figs . 7 and 8 . editor : place figures 7 and 8 here all curves in figs . 7 and 8 have the same character . each predicts a bluing as [ fe / h ] decreases from 0 to about -1.5 , after which the predicted color becomes redder . over the relevant range of metallicities from [ fe / h ] = -0.9 to -2.2 , which contains most of the galactic globular clusters , the color change is only @xmath0(b - v ) @xmath9 0.02 mag . we emphasize that the zero points of the b - v values in figs . 7 and 8 have not been adjusted to conform with observational data . they are the calculated from the theoretical zero points in the several relevant equations . given the range of errors in each of these equations , it is remarkable that the agreement of the predicted color variation with the observations , to be discussed in the next section , is so good . which definition of `` mean color '' , averaged over the light curve , best approximates the color of an equivalent static star ? three definitions are used in the literature , defined by the following operational procedures . ( 1 ) change b and v magnitudes into intensity units and average over the light curve , calculating mean values of @xmath38 and @xmath39 , changed back to magnitude units , and then subtract to form the color @xmath40 - @xmath41 . ( 2 ) change the color curve in b - v from magnitude to intensity units and average over the intensity color curve , changed back to magnitude units to form @xmath42 ( 3 ) keep the color curve in magnitude units and integrate over the cycle to obtain @xmath43 . these procedures give colors that can differ by as much as 0.05 mag for highly asymmetric light and color curves of high amplitude . which is correct to give an approximation to the color of a `` static star '' with the same average energy output ? in a semi - analytical treatment , preston ( 1961 ) showed that the best estimate of the color of such a static star is @xmath43 . i reached the same conclusion ( s90a ) in a semi - empirical discussion based on the calculated and observed slopes of the period - color relation calculated from the pulsation equation , choosing a color equivalent definition that made the predicted and observed slopes the same . corwin and carney ( 2001 ) also concluded that @xmath43 was the appropriate color to use , reversing a previous discussion by carney , storm , and jones ( 1992 ) to the contrary . a definitive theoretical discussion by bono , caputo , & stellingwerf ( 1995 ) showed that @xmath43 is within 0.01 mag of @xmath44 for all blue amplitudes smaller than 1.6 mag . a table of differences between the three types of color definitions and @xmath44 , which they calculated , is given by them . we use @xmath43 in the remainder of the paper . we have converted other listings in the literature to @xmath43 where they differ , either by using table 4 of bono et al . ( 1995 ) , or using the @xmath0c(a ) correction formulation in s90a . a detailed reading of the literature is often required to determine which mean color definition has been used by particular authors . we have attempted to assemble a data base of reddening- corrected @xmath46 colors of rr lyrae variables accurate at the 0.02 mag level for both field and clusters variables . if the observational data are from a variety of observers , they must be reduced to a common photometric standard system , taking out systematic zero - point differences between the sets . in addition , a uniform definition of what kind of mean color , as just discussed , must be used . more uncertain are the corrections for e(b - v ) reddening , that themselves are generally be accurate only to within 0.02 mag . hence , detection of the relatively small color variations with [ fe / h ] predicted in figs . 7 and 8 is near the limit of accuracy of the data now available . a large list of observed photoelectric colors for field rr lyrae variables is in the catalog by nikolov , buchantsova , & frolov ( 1984 ) , hereafter referred as the sofia catalog . nine photoelectric data sets from the literature were compared and reduced to a common system , producing a highly homogeneous list of @xmath43 colors for 210 variables in that catalog ( their table 2 , column 2 ) . the photoelectric data , so reduced , are from fitch et al . ( 1966 ) , sturch ( 1966 ) , clube et al . ( 1969 ) , paczynski ( 1965a , b , 1966 ) , stepien ( 1972 ) , epstein ( 1969 ) , lub ( 1979 ) , preston & paczynski ( 1964 ) , and kinman ( 1961 ) . i have used the listings in the sophia catalog and corrected them for reddening by using the homogeneous set of absorption values in v measured by layden ( 1994 ) , using e(b - v ) = a(v)/3.0 for all stars in common with the sophia catalog and the layden list . in this way , @xmath47 colors were obtained for 142 field rr lyraes in the layden sample , plus 10 additional from blanco s study discussed in s93a but not in layden s list , and an additional 8 from lub that are not in layden but reduced to his absorption system by cross checking the lub list as constructed in s90a with the layden list . a total of 160 field variables with @xmath47 colors are available in this way . the color values are not listed here because they can be readily obtained from the same original sources . editor : place figure 9 here the colors for these field variables are plotted in fig . 9 vs. layden s [ fe / h ] metallicities . as emphasized earlier , the [ fe / h ] values are on the system of zinn and west ( 1984 ) . the line is equation ( 19 ) , which is the prediction of t@xmath11 using the pulsation equation ( 16 ) and the transformation to b - v via eq . ( it is the predicted color - metallicity locus at the fbe using the catelan et al . ( 2004 ) luminosity calibration of the fbe in eq . ( 10 ) and the parabolic envelope locus to the period- metallicity relation of eq . ( 5 ) in the pulsation equation for t@xmath11 , transformed to b - v via eq . ( 18 ) . the drawn line is repeated from figs . 7 and 8 , which is one of the curves in the family shown there . no adjustment of the _ zero point _ of the theoretical equation ( 19 ) to fit the observations has been made in fig . 9 . hence , as stated earlier , the generally good agreement of this calculated curve with the observations of the bluest color at a given metallicity is remarkable because the various zero points in the theoretical and the observed equations used for the prediction have been left as they had been previously calculated . the predicted t@xmath11([fe / h ] ) line in fig . 9 fits the observations tolerably well except for metallicities more metal poor than [ fe / h ] = -2.0 where it does not bend sufficiently toward redder colors . a better representation is derived in the next section ( fig . 11 there ) where the color data for variables in globular clusters are combined with these field star colors . the globular cluster color data are set out in fig . they have been taken from the literature , either from modern ccd photometry , or from reliable pg photometry . the literature citations for the 21 clusters are listed in table 2 . most data have been transformed to the @xmath43 mean color system by table 4 of bono et al . ( 1995 ) . however , if the original data were originally listed as @xmath44 , they were kept at that because , according to bono et al . , the `` static star '' color is within 0.01 mag of @xmath43 for all amplitudes of interest . editor : place figure 10 here . the e(b - v ) reddening values determined by the original authors have been replaced by the homogeneous values listed by harris ( 1996 ) , except for ngc 4590 where the value proposed by walker ( 1994 ) is obviously more correct . editor : place figure 11 here . the data in figs . 9 and 10 are combined in fig . although there are outriders to the bulk of the distribution at blue colors , the trend is for the clusters with [ fe / h ] between -1.0 and -1.7 to be bluer at the short period edge than either for the more metal rich or the more metal poor variables , similar to the solid curves in figs . 9 and 10 . a stronger bend toward the red is drawn as a new curve fitted empirically to the data for [ fe / h ] @xmath20 -2.0 in fig . it is a slightly better fit to the combined data than the curve that is drawn in fig . 9 and 10 using eq . the equation of this empirical curve is , @xmath48 ) + 0.061 ( [ fe / h])^2\ ] ] this can be transformed to temperature via eq . ( 18 ) to give @xmath49 ) -0.023 ( [ fe / h])^2\ ] ] we aver that this applies at the fbe because we have assumed that the blue envelope loci in figs . 3 and 10 ( solid line ) refer to that edge . we now repeat the calculation of the luminosity calibration with metallicity made in s93a but here we use the new correlations between period , temperature , and mass relations with metallicity set out in the previous sections . the method is to insert the parabolic eq . ( 5 ) for the period - metallicity locus at the fbe , the new semi - empirical equation for the temperature at that edge from eq . ( 21 ) , fig . 11 , and the mass - metallicity relation from eq . ( 15 ) into the pulsation equation , permitting the calculation of the luminosity . using eqs . ( 21 ) , ( 5 ) , and ( 15 ) in eq . ( 16 ) gives , @xmath50 ) - 0.056 ( [ fe / h])^2.\ ] ] as before , using m@xmath21 = -2.5 log l + 4.75 and bc = 0.06 + 0.06 ( [ fe / h ] ) in m@xmath3 = m@xmath21 - bc gives , from eq . ( 22 ) , @xmath51 ) + 0.140 ( fe / h])^2,\ ] ] over the metallicity range of [ fe / h ] @xmath9 -0.5 to -2.3 . this calibration updates the linear calibration of eq . ( 6 ) given in s93b , made by the same method , but using the new assumptions here that relate period , mass , and temperature variations with metallicity . as already said several times , the zero point in equation ( 23 ) is the theoretical value based on the zero points adopted in each of the pulsation , mass , and color - temperature equations . because we can not guarantee that these combined zero points are precise , we change the theoretical zero point in eq . ( 23 ) to conform with the independent data on m@xmath3 ( rr ) determined empirically . clementini et al . ( 2003 ) have measured the mean apparent magnitude of rrls in the lmc as a function of [ fe / h ] , which can be changed to absolute magnitude by assuming a distance modulus of the lmc determined by other means . they adopt ( m - m)@xmath2 = 18.54 based on an adopted p - l relation of the lmc type i long period cepheids . however , tammann , sandage & reindl ( 2003 ) argued that the cepheid p - l relation differs between the galaxy and the lmc , and they fixed the distance modulus the lmc also at ( m - m)@xmath2 = 18.54 by means other then the long period cepheids . the data of clementini et al . give @xmath13 = 19.064 for the rr lyraes at [ fe / h ] = -1.5 in the lmc , which , with the assumed modulus of 18.54 , give m@xmath3 = 0.524 . this differs by 0.078 mag from the zero point in eq . ( 23 ) of m@xmath3 = 0.602 for this metallicity . hence , we must make the zero point in eq . ( 23 ) brighter by 0.078 mag to put it on the observed scale of the clementini et al . but the clementini data refer to the average rr lyrae apparent magnitude which already accounts for the average luminosity evolution from the zahb . a difference of about 0.1 mag is expected between the zahb and the average evolved hb level ( s93b ) . remarkably , this is near the 0.078 mag difference just quoted . it continues as a surprise that the calculated and observed zero points , corrected for evolution , are so nearly the same at the 0.10 - 0.078 = 0.02 mag level . again , this can only mean that the zero points of the various relations that go into eq . ( 23 ) are themselves fully compatible with the various observations that comprise them . therefore , our final calibration of the zahb using the theoreticians route via the pulsation equation , but zero - pointed through the lmc as in eq . ( 23 ) and made brighter by 0.078 mag , is , @xmath52 ) + 0.140 ( [ fe / h])^2,\ ] ] valid for the average evolved hb at the fbe . it remains to discuss the significant of the difference between layden s field star sample and the rrls in clusters in the absence of the oosterhoff gap in fig . 1 and its presence in fig . 4 , and , in addition , in the shift of the short period envelope lines between of figs . 1 and 4 in the period - metallicity correlations . both differences were set out earlier in 3 . the absence of a gap ( i.e. the presence of a continuum ) in particular galaxian samples of field rrls had already been seen in the original period - metallicity sample of preston ( 1959 , his fig . it is also highly manifest in the much larger sample of layden ( 1994 , 1995 ) in figs . 1 - 3 here . the gap is also absent in recent discovery surveys of faint rrl in high latitudes ( eg . vivas & zinn 2002 , 2004 ) , a summary of the relevant evidence is by catelan ( 2004 ) , in contrast to figs . 1 - 3 and to the results of vivas & zinn ( 2002 , 2004 ) . catelan ( 2004 ) remarks that `` the reason for the discrepancy between the two studies is unclear at present . '' ] even more to the point , for some samples of the dwarf spheroidal companions to the galaxy the consequence of the intermediate value of the mean period , @xmath53 , is that there is no oosterhoff separation into a _ dichotomy _ for different metallicitues ( bono , caputo , & stellingwerf 1994 ; mackey and gilmore 2003 ; mateo 1996 ; pritzl et al . 2002 ; cseresnjes 2001 ; clementini et al . ( 2004 ) ; and dallora et al . the result is that the mean periods of the rrl are _ intermediate _ between the two oosterhoff groups in the galaxy ( catalan 2004 , table 1 ; siegel & majewski 2000 , their fig . 6 ) . the general morphology - metallicity progression for horizontal branches in the globular clusters in the galaxy is this . the hbs in the high metallicity clusters for [ fe / h ] @xmath54 do not penetrate the instability strip because they are redder than the strip color boundaries . hence , such clusters contain few if any rr lyrae variables . the prototype example is 47 tuc . a counter example in the opposite sense is ngc 7006 which has an abnormally red segment to its hb for its metallicity ( sandage & wildey 1967 ) , yet it contains many rrls . also in an opposite sense ( abnormally blue hb for its metallicity ) is m13 where the branch misses the instability strip altogether because of its excessive blueness , yet its metallicity is nearly identical with that of m3 which contains many variables ( cf . cho et al . 2005 for a modern comparison of m3 and m13 ) . hence , the cluster morphology - metallicity relation that is responsible for the oosterhoff gap in the galaxy is not perfectly consistent . the most extreme examples of this second parameter effect are the metal rich ( [ fe / h ] @xmath9 -0.50 ) clusters ngc 6388 and ngc 6441 which have abnormally long periods for their amplitudes ( pritzl et al . 2000 , 2001 ) , yet both have abnormally blue extended horizontal branches never before seen in other high metallicity clusters ( rich et al . 1997 ) . the globular cluster m2 is a less extreme example ( lee & carney 1999b ) with much longer periods for a given amplitude . also , similar to m2 , the second- parameter custer ngc 5986 ( alves , bond , & onken 2001 ) has a period - amplitude diagram with abnormally long periods compared with even oosterhoff ii clusters , although the metal abundance is only -1.58 on the zinn & west system . its hb is abnormally blue , as in m13 ( van den bergh 1967 ) , for this metallicity . this is opposite the sense of the effect in ngc 7006 . although not yet conclusively proved , the reasons for such behavior is currently suspected to be different chemical composition effects that control the details of the hb morphology ( eg . variations in the helium core mass , differences in the mass loss from the agb to the hb , differences in the deep mixing etc . ) . if so , the difference between the field and the cluster rr lyraes ( figs . 1 and 4 ) would suggest that these field variables have had a different history in their chemical evolution than those in clusters . yes , but why ? suppose that the bulk of the field rrls have come from the disintegration of dwarf spheroidals that once were companions to the galaxy but have since been accreated and disrupted . this is almost certainly proved now by the discovery of the discrete rrl streams in the sloan survey for high latitude variables ( ivezic et al . 2000 ; yanny et al . 2000 ; newberg et al . 2002 ) , and the similar discovery by vivas et al . ( 2001 cf . also vivas & zinn , 2002 , 2004 ) from the venezuela quest survey . these discrete streams , widely assumed to be debris from original dwarf spheroidal companions to the galaxy , now disrupted , are the generalization of the many high velocity moving groups discovered by eggen ( 1977 ) , the first of which was the grombridge 1830 group that contains rr lyrae itself ( eggen & sandage 1959 ) . the sloan survey , and other evidence before it ( eg . freeman 1987 ; majewski 1993 ; lynden - bell & lynden bell 1995 ; freeman & bland - hawthorn 2002 ) , showed that the eggen high velocity moving groups are real . the disruption of dwarf spheroidal galaxian companions is seen directly today in the sagittarius disintegrating dwarf ( ibata , gilmore , & irwin 1994 , 1995 ) as it is being torn apart by tidal interaction with the galaxy . it is known that the chemical evolution of the individual dwarf spheroidals differ , depending on their individaual luminosity , i.e. mass ( eg . aaronson 1986 ; skillman et al . 1989 , caldwell et al . 1998 ; mateo 1998 , his fig . 7 for a summary ) . hence , the hb morphology of the products of the disruption ( among which are the halo rrls ) , should also differ by second - parameter effects . in particular , as a group , they should differ from the `` normal '' hb morphology of galactic globular clusters where the chemical evolution has gone further toward `` completion '' because the much higher mass of the galaxy means stronger gravity and therefore the chemical products of agb nucleosynthesis can be retained , unlike the situation in the dwarf spheroidals where supernovae winds can cause mass loss from the parent dwarf . because of the variable ability of the parent galaxies to retain the products of the chemical build up , depending on the mass , the chemical evolution of the parent dwarf spheroidals has differed from that history for most of the globular clusters in the galaxy that have the normal oosterhoff gap . hence , the details of the oosterhoff effect can be expected to differ . if so , one can no longer support the canonical idea that the field rr lyraes have come from the general galactic globular cluster population , but rather from the previous dwarf spheroidal companions with varying hb morphologies . the hypothesis to be tested is this . the majority of the globular clusters in the inner galactic halo have a common origin that coincided with the origin of the bulk of the galaxy itself ( van den bergh & mackey 2004 ) . the evidence supports the idea that this process is a more or less coherent collapse , albeit with noise , similar to that of eggen et al . ( 1962 , els ; sandage 1990c ) for the lower halo , the bulge , and thin , and thick disks . on the other hand , as argued above , many of the halo field variables have come from the `` outside '' in a type of searle - zinn ( 1978 ) bottom - up hierarchical fragmentary build up in some form of the cold dark matter scenario for origins . however , the most telling argument against the bottom - up build up from fragments of the _ bulk _ of the galaxy , and other similar high mass galaxies , is the recent discovery of normal e galaxies at the very large redshifts of z @xmath8 2 ( mccarthy 2004 ) which would not exist in the hierarchical build - up from small fragments . rather , there existence is the natural consequence of an els - type collapse on time scale short relative to the hubble time . said differently , normal e galaxies undergoing passive evolution , but otherwise fully formed , can not be created by slow hierarchical build up on the cold dark matter scenario , but must be formed very early , very rapidly , all components of a protogalxy having a negative total energy . this is direct els . a test of the different - origins hypothesis for the field and the cluster variables in the galaxy is to map the morphology- metallicity relation of the hb of the dwarf spheroidals and to match the differences in the period - metallicity envelope , for example , with such second parameter indicators as the period- amplitude relation , the period - rise tine correlation , and the period - color relations with the oosterhoff period shifts as functions of morphology and metallicity . such a study is beyond the scope of this paper but is in progress . there are eight principle research points in this paper . the best fitting envelope of the period - metallicity correlation ( fig . 1 ) at the shortest periods for the field rr lyraes in the layden ( 1994 ) sample is non - linear . a parabolic relation [ eq . ( 5 ) ] more adequately fits the data at low metallicity than the linear envelope adopted earlier ( s93a ) . the shape of the envelope in item 1 is also a good fit to the period - metallicity relation for rrl in globular clusters ( fig . 4 ) , except that it is displaced in zero point toward longer periods by @xmath0log p = 0.029 @xmath7 0.007 . this is another piece of evidence , joining several already in the literature , that the field and cluster variables may have different origins ( points 7 and 8 below ) . the equations for various period - metallicity envelopes in items 1 and 2 are inserted into the pulsation equation , together with a series of new calibrations of the absolute magnitude of rr lyrae variables , and with an updated calibration of rr lyrae mass with metallicity , to predict a family of temperature variations at the fundamental blue edge as function of metallicity . all combinations of the input assumptions show that t@xmath11 at the fbe varies with metallicity , changing by @xmath0log t@xmath11 = 0.04 over the range of ( [ fe / h ] ) between 0 and -2.5 . it first becomes hotter as [ fe / h ] changes from 0 to -1.5 , and then cooler for as the metallicity decreases further ( figs . 5 & 6 ) . the predicted fbe temperatures are transformed into b - v color ( figs . 7 & 8) by the adopted temperature - color relation of eq . the color variation is predicted to be at the level of 0.03 mag in ( b - v ) for [ fe / h ] between -1.0 and -2.5 for any given choice of the absolute magnitude calibration and of the variation of the period - metallicity locus at the fbe . comparison of this predicted color variation for both field and cluster variables ( figs . 9 & 10 ) shows general agreement in the shape of the variation and in the color zero point . the agreement of this theoretical color zero point with the observations is remarkable because the calculations are made adopting the zero points in the pulsation equation [ eq . ( 16 ) ] , the mass equation ( 15 ) , the color - temperature transformation eq . ( 18 ) , and the luminosity calibration equations , each of which have a range of uncertainty . \6 . an iteration of the predicted envelope locus for the color - metallicity correlations of figs . 9 and 10 to better fit the color-[fe / h ] data for metallicities smaller than -2.0 gives the semi - empirical locus of eq . ( 20 ) , shown in fig . 11 . this envelope , now made to better fit the observations , is used to make the calculation via the pulsation backward to obtain an improved absolute magnitude calibration of eq . ( 23 ) which refers to the age zero hb . this equation is then re - zero pointed to refer to the average evolved hb by adopting the clementini et al . ( 2003 ) lmc observation that @xmath55 = 19.064 at [ fe / h ] = -1.5 for the lmc rrls and using the modulus of the lmc as ( m - m)@xmath2 = 18.54 ( tammann et al . 2003 ) which does not use the lmc classical cepheids . this gives a new calibration of the average m@xmath3 of the evolved hb of galactic globular clusters [ ( eq . ( 24 ) ] . \7 . returning to the observed offset of the period- metallicity envelopes for the period distributions of the field variables relative to those in clusters ( figs . 1 & 4 ) , and the presence or absence of the oosterhoff gap in the clusters and in the field , we conclude that the origins of the two groups may be different . the presence in one and absence in the other of the oosterhoff gap proves that the variation of the morphology of the hb with [ fe / h ] differs between the two groups . this is another manifestation of the second parameter effect seen in only a handful of the galactic globular clusters such as m13 , ngc 7006 , m2 , ngc 5986 and the extreme cases of ngc 6388 and ngc 6441 . we agree with the growing consensus that the origin of at least a subset of the halo variables is the disruption by tidal friction of structures related to the dwarf spheroidals , where it is known that they have varying degrees of chemical evolution- completion depending on their mass ( luminosity ) . it is also known that the oosterhoff gap characteristics varies among them . this variation is suggested to manifest itself in the difference in the morphology of the hb between the galaxy and the dwarf spheroidals , similar to the differences seen between the bulk of the galactic globular clusters and the second parameter clusters such as ngc 6838 , ngc 6441 , ngc 7006 , m2 , and ngc 5986 . a test of the hypothesis is to study the systematics of the morphology of the dwarf spheroidal hbs and its variation with metallicity . the hb morphology for them is expected to be systematically different from that in the galaxy , and that difference is postulated to be a function of the mean metallicity . such a study of comparative hb morphology between the normal and the second parameter globular clusters in the galaxy and the local dwarf spheroidal companions is beyond the scope of this paper , but has been started . acknowledgements : an early paper ( unpublished ) , refereed by geselle clementini , has been rewritten here with a different goal and emphasis . the many suggestions by clementini , leading to the old problems being analyzed from a new direction , are gratefully acknowledged . other comments on the original paper by fillipina caputo were equally important in reconstituting it here , as were comments by carla cacciari and flavio fusi pecci . i am grateful to them all . it is again a special pleasure to thank david sandage for preparing the diagrams for press . aaronson , m. 1986 in star forming dwarf galaxies and related objects , eds . d. kunth , t.x . thaun , & j.t.t van ( paris : ed front . ) p. 125 alcock , c. et al . 1998 , apj , 492 , 190 alcock , c. 2000 , aj , 119 , 2194 alves , d.r . , bond , h.e . , & onken , c. 2001 , aj , 121 , 318 arp , h.c . 1955 , aj , 60 , 317 arp , h.c . , baum , w.a . , & sandage , a. 1953 , aj , 58 , 4 bencivenni , d. , caputo , f. , manteign , m. , & quarta , m.l . 1991 , apj , 380 , 484 bingham , e.a . , cacciari , c. , dickens , r.j . , & fusi pecci , f. 1984 , mnras , 209 , 765 blanco , v. 1992 , aj , 104 , 734 bono , g. , caputo , f. , cassisi , s. , castellani , v. , & marconi , m. 1997a , apj , 479 , 279 bono , g. , caputo , f. , cassisi , s. , incerpi , r. , & marconi , m. 1997b , apj , 483 , 811 bono , g. , caputo , f. , & stellingwerf , r.f . 1994 , apj , 423 , 294 bono , g. , caputo , f. , & stellingwerf , r.f . 1995 , apjs , 99 , 263 borissova , j. , catelan , m. , & valchev , t. 2001 , mnras , 324 , 77 brocato , e. , castellani , v. , & ripepi , v. 1994 , aj , 107 , 622 ( ngc 4590 ) brocato , e. , castellani , v. , & ripepi , v. 1996 , aj , 111 , 890 butler , d. 1975 , apj , 200 , 68 butler , d. , dickens , r.j . , & epps , e. 1978 , apj , 225 , 148 cacciari , c. 1979 , aj , 84 , 1542 cacciari , c. & clementini , g. 2004 , in standard candles for the extragalactic distance scale , ed . d. alloin , w. gieren ( springer , berlin ) , p. 105 cacciari , c. , corwin , t.m . , & carney , b.w . 2004 , astro / ph 0409567 caldwell , n. , armandroff , t.e . , dacosta , g.s . , & seitzer , p. 1998 , aj , 115 , 535 caloi , v. , dantona , f. & mazzitelli , i , ( 1997 ) , , 320 , 823 caputo , f. 1988 , , 189 , 70 caputo , f. 1997 , mnras , 284 , 994 caputo , f. , castellani , v. , marconi , m , & ripepi , v 1999 , mnras , 306 , 815 caputo , f. , castellani , v. , marconi , m. , & ripepi , v. 2000 , mnras , 316 , 819 ( ccmr ) caputo , f. , derinaldis , a. , manteiga , m. , pulone , l. , & quarta , m.l . 1993 , , 276 , 41 carney , b.w . , storm , j. , & jones , r.v . 1992 , apj , 386 , 663 ( csj ) carney , b.w . , storm , j. , trammell , s.r . , & jones , r.v . 1992 , pasp , 104 , 44 cassisi , s. , castellani , v. , deglinnocenti , s. , salaris , m. , & weiss , a. 1999 , , 134 , 103 castellani , v , chieffi , s. , & pulone , l. 1991 , apjs , 76 , 911 castelli , f. , 1999 , , 346 , 564 catalan , m. 2004 , in variable stars in the local group , eds . kurtz & k.r . pollard ; iau coll . 193 . asdp conf ser . 310 , p 113 catelan , m. , pritzl , b.j . , & smith , h.a . 2004 , apjs , 155 , 633 chiosi , c. , wood , p. , bertelli , g. , & bressan , a. 1992 , apj , 387 , 320 cho , d - h . , lee , s - g . , jeon , y - b . , & sim , k.j . 2005 , aj , 129 , 1922 clementini et al . 2004 , in variable stars in the local group , eds . kurtz & k.r . pollard ; iau col . ser 310 , p. 60 clementini , g. , gratton , r. , bragaglia , a. , carretta , e. , di fabrizio , l , & maio , m. 2003 , aj , 125 , 1309 clement , c.m . 2001 , the new cluster variable catalog accessed at http://www.astro.utoronto.ca/people.html clement , c.m . , & shelton , l. 1999a , aj , 118 , 453 clement , c.m . , & sheldon , l. 1999b , apj , 515 , l85 clube , s.v.m . , evans , d.s.e . , & jones , d.h.p . 1969 , mem . ras . , 72 , 101 corwin , t.m . , & carney , b.w . 2001 , aj , 122 , 3183 corwin , t.m . , carney , b.w . , & nifong , b.g . 1999 , aj , 118 , 2875 cox , a.n . 1987 , in second conference on faint blue stars , iau colloquim no . philip , d.s . hayes , & j.w . leibert ( davis , schenectady ) , p 161 cox , a.n . , hodson , s.w . , & clancy , s.p . 1983 , apj , 266 , 94 cox , a.n . , king , d.s . , & hodson , s.w . 1980 , apj , 236 , 219 cseresnjes , p. 2001 , aa , 375 , 909 dallora , m. et al . 2003 , aj , 126 , 197 demarque , p. , zinn , r. , lee , y - w . , & yi , s. 2000 , aj , 1398 desantis , r. , & cassisi , s. 1999 , mnras , 308 , 97 dickens , r.j . 1971 , apjs , 22 , 249 dickens , r.j . , & flinn , r. 1972 , mnras , 158 , 99 dorman , b. 1992 , apjs , 81 , 221 eggen , o.j . 1977 , apj , 215 , 812 eggen , o.j . , lynden - bell , d. , & sandage , a. 1962 , apj , 136 , 748 eggen , o.j . , & sandage , a.r . 1959 , mnras , 119 , 255 epstein , i. 1969 , aj , 74 , 1131 fitch , w. , wisniewski , w. , johnson , h.l . , 1966 , arizona comm , 5 , 71 , parts 2 , 3 freeman , k.c . 1987 , araa , 25 , 603 freeman , k.c . , & bland - hawthorn , j. 2002 , araa , 40 , 487 gratton , r.g . 1998 , mnras , 296 , 739 gratton , r.g . , tornambe , a. , & ortononi , s. 1986 , , 169 , 111 gratton , r.g . , fusi pecci , f. , carretta , e. , clementini , g. , corsi , c.e . , & lattanzi , m. 1997 , apj , 491 , 749 grosse , e. 1932 , astron . nach . , 246 , 376 hachenberg , o. 1939 , zs.f.ap , 18 , 49 harris , w.e . 1996 , aj , 112 , 1487 ibata , r. , gilmore , g. , & irwin , m.j . 1994 , nature , 370 , 194 ibata , r. , gilmore , g. , & irwin , m.j . 1995 , mnras , 277 , 781 ivezic , z. et al . 2000 , aj , 120 , 963 kaluzny , j. , olech , a. , & stanek , k.z . 2001 , aj , 121 , 1533 kholopov , p.n . 1985 , general catalog of variable stars , 4th ed . ( nauka , moscow ) kinman , t.d . 1961 , , e37 kovacs , g. , buchler , j.r . , marom , a. , iglesias , c.a . , & rogers , f.j . 1992 , , 259 , l46 layden , a.c . 1994 , aj , 108 , 1016 layden , a.c . 1995 , aj , 110 , 2312 lee , y .- w . 1990 , apj , 363 , 159 lee , j - w . , & carney , b.w . 1999a , aj , 117 , 2868 lee , j - w . , & carney , b.w . 1999b , aj , 118 , 1373 lee , y .- w . , demarque , p. , & zinn , r. 1990 , apj , 350 , 155 ( ldz ) lub , j. 1979 , aj , 84 , 383 lynden - bell , d. , & lynden - bell , r.m . 1995 , mnras , 275 , 429 mackey , a.d . , & gilmore , g.f . 2003 , mnras , 345 , 747 majewski , s.r . 1993 , araa , 31 , 575 mateo , m. 1996 , in formation of the galctic halo ... inside and out ; conf . morison , & a. sarajedini , p. 434 mateo , m. 1998 , araa , 36 , 435 mccarthy , p.j . 2004 , araa , 42 , 477 mcnamara , d.h . , rose , m.b . , brown , p.j . , ketcheson , d.i . , maxwell , j.e . , smith , k.m . , & wooley , r.c . 2004 , in variable stars in the local group , asp conf . 310 , 525 menzies , j. 1974 , mnras , 168 , 177 montegriffio , p. , ferraro , f.r . , origlia , l. , & fusi pecci , f. 1998 , mnras , 297 , 872 nemec , j.m . 2004 , aj , 127 , 2185 newberg , h.j . 2002 , apj , 569 , 245 nikolov , n. , buchantsova , & frolov , m. 1984 , `` the mean light and color of 210 field rr lyrae type stars '' , astron . council ussr acad . sofia oconnell , d. 1958 , ed . stellar populations , specola vaticana , rica , 5 olech , a. , kaluzny , j. , thompson , i.b . , pych , w. , krzeminski , w. , & schwarzenberg - czerny , a. 1999 , aj , 118 , 442 olech , a. , kaluzny , j. , thompson , i.b . , pych , w. , krzeminski , w. , & schwarzenberg - czerny , a. 2001 , mnras , 321 , 421 oosterhoff , p. th . 1939 , observatory , 62 , 104 oosterhoff , p. th . 1944 , ban , 10 , 55 paczynski , b. 1965a , acta astron . 15 , 103 paczynski , b. 1965b , acta astron . 15 , 115 paczynski , b. 1966 , acta astron . 16 , 97 peterson , j.o . 1973 , , 27 , 89 peterson , j.o . 1978 , , 62 , 205 peterson , j.o . 1979 , , 80 , 53 piersimoni , a.m. , bono , g. , & ripepi , v. 2002 , aj , 124 , 1528 preston , g.w . 1959 , apj . , 130 , 507 preston , g.w . 1961 , apj , 133 , 29 preston , g.w . , & paczynski , b. , 1964 , apj , 140 , 181 pritzl , b.j . , smith , h.a . , catelan , m. , & sweigart , a.v . 2001 , aj , 122 , 2600 pritzl , b.j . , armandroff , t.e . , jacoby , g. , and dacosta , g.s . 2002 , aj , 124 , 949 pritzl , b , smith , h.a . , catelan , m. , & sweigart , a.v . 2000 , apj , 530 , l41 pritzl , b. , smith , h.a . , catelan , m. , & sweigart , a.v . 2001 , aj , 122 , 2600 rich , r.m . 1997 , apj , 484 , l25 salaris , m , deglinnocenti , s. , & weiss , a. 1997 , apj , 479 , 665 sandage , a. 1953 , aj , 58 , 61 sandage , a. 1958 , in stellar populations , specola vaticana , ed . d. oconnell , rica , 5 , 41 sandage , a. 1969 , apj , 157 , 515 sandage , a. 1981a , apj , 244 , l23 ( s81a ) sandage , a. 1981b , apj , 248 , 161 ( s81b ) sandage , a. 1982a , apj , 252 , 553 ( s82a ) sandage , a. 1982b , apj , 252 , 574 ( s82b ) sandage , a. 1990a , apj , 350 , 603 ( s90a ) sandage , a. 1990b , apj , 350 , 631 ( s90b ) sandage , a. 1990c , jrasc , 84 , 70 sandage , a. 1993a , aj , 106 , 687 ( s93a ) sandage , a. 1993b , aj , 106 , 703 ( s93b ) sandage , a. 2004 , aj , 128 , 858 sandage , a. , bell , r.a . , & trippico , m.j . 1999 , apj , 522 , 250 sandage , a. , & cacciari , c. 1990 , apj , 350 , 645 sandage , a. , katem , b.n . , & sandage , m. 1981 , apjs , 46 , 41 ( sks ) sandage , a. , smith , l.l . , & norton , r.h . 1966 , apj , 44 , 894 sandage , a. , & wildey , r. , 1967 , apj , 150 , 469 sawyer hogg , h. 1973 , third catalogue of variable stars in globular clusters , david dunlap obs . pub 3 , no . 6 schwarzschild , m. 1958 , in stellar populations , specola vaticana , ed . d. oconnell , rica , 5 , 59 searle , l. , & zinn , r. 1978 , apj , 225 , 357 sekiguchi , m , & fukugita , m. 2000 , aj , 120 , 1072 siegel , h.m . & majewski , s.r . 2000 , aj , 120 , 284 simon , n.r . , & clement , c.m . 1993 , apj , 410 , 526 skillman , e.d . , kennicutt , r.c . , & hodge , p. w. 1989 , apj , 347 , 875 stepien , k. 1972 , acta astron . 22 , 175 storm , j. , carney , b.w . , & beck , j.a . 1991 , pasp , 103 , 1264 sturch , c.r . 1966 , apj , 143 , 774 sturch , c.r . 1977 , pasp , 89 , 349 suntzeff , n.b . , kinman , t.d . , & kraft , r.p . 1991 , apj , 367 , 528 sweigart , a.v . , & gross , p.g . 1976 , apjs , 32 , 367 sweigart , a.v . , renzini , a. , & tonambe , a. 1987 , apj , 312 , 762 tammann , g.a . , sandage , a. & reindl , b. 2003 , , 404 , 423 van albada t.s . , & baker , n. 1973 , apj , 1185 , 477 van den bergh , s. 1967 , aj , 72 , 70 van den bergh , s. , & mackey , a.d . , 2004 , mnras , 354 , 713 vandenberg , d.a . , swenson , f.j . , rogers , f.j . , iglesias , c.a . , & alexander , d.r . 2000 , apj , 532 , 430 vivas , a.k . , & zinn , r. 2002 , astro / ph 02121116 vivas , a.k . , & zinn , r. 2004 , msait , 74 , 928 vivas , a.k . 2001 , apj , 554 , l33 walker , a.r . 1994 , aj , 108 , 555 walker , a.r . 1998 , aj , 116 , 220 walker , a.r . , & nemec , j.m . 1996 , aj , 112 , 2026 wehlau , a. , slawson , r.w . , & nemec , j.m . 1999 , aj , 117 , 286 yanny , b. , et al . 2000 , aj , 120 , 963 zinn , r. , & west , m.j . 1984 , apjs , 55 , 45 distribution of period with metallicity for the 302 type ab ( fundamental mode ) rr lyrae variables studied by layden ( 1994 , 1995 ) . the full line [ eq . ( 1 ) of the text ] is an approximation to the envelope of the shortest period variables at a given metallicity . the dashed line is eq . ( 2 ) of the text for the log p values averaged over the narrow intervals of [ fe / h ] listed in table 1 . figure 2 . the data from fig . 1 for the layden field star sample but with three linear envelope lines of eqs . ( 3 ) , ( 4 ) and ( 4 ) superposed . the envelope line is the same as in fig . ( 1 ) for [ fe / h ] more metal rich than -1.8 but two steeper envelopes from eqs . ( 4 ) and ( 4 ) are shown as dashed lines for more metal poor stars . figure 4 . the period - metallicity relation for 919 rr lyrae stars in globular clusters from data in the clement ( 2001 ) update of the sawyer hogg ( 1973 ) cluster variable catalog . each cluster is represented by the horizontal line at its adopted metallicity from harris ( 1996 ) on the metallicity scale of zinn and west . the parabolic short - period locus for the field stars from eq . ( 5 ) is shown . it fails to represent the short period distribution , requiring a shift of @xmath0log p = 0.029 @xmath7 .007 toward longer periods for the cluster variables relative to those in the field . the predicted run of log t@xmath11 with [ fe / h ] keeping the luminosity fixed at the catelan et al . ( 2004 ) calibration of eq . ( 10 ) , and using the four loci for the fbe period - metallicity envelopes from eqs . ( 1 ) , ( 3 ) , ( 4 ) , ( 4 ) and ( 5 ) , together with the bono et al . mass equation from eq . ( 15 ) . these have been put into the pulsation eq . ( 16 ) to generate the curves . line number 1 is for the linear envelope to the log p([fe / h ] ) relation at the shortest periods in fig . 1 ) ; numbers 2 and 4 use the dashed lines in fig . 2 from eqs . ( 4 ) and ( 4 ) for [ fe / h ] less than -1.8 ; number 3 is for the parabolic envelope of eq . ( 5 ) from fig . 3 ; number 5 is for the midpoint log p([fe / h ] ) relation of eq . ( 2 ) which is the dashed line in fig . 1 . five predicted runs of log t@xmath11 with [ fe / h ] at the fundamental blue edge of the instability strip using a fixed parabolic envelope locus to the log p-[fe / h ] relation from eq . ( 5 ) and fig . 3 , and five different rrl luminosity calibrations . line 1 is for the linear luminosity calibration of eq . ( 6 ) which was the s93a pulsation solution ; line 2 is for the linear calibration of clementini et al . ( 2003 ) from eq . ( 12 ) ; line 3 is for the non - linear calibration of eq . ( 7 ) by caputo et al . ( 2000 ) ; line 4 is the non - linear calibration of eq . ( 10 ) by catelan et al . ( 2004 ) ; line 5 is for the two - line calibration of mcnamana et al . ( 2004 ) of eqs . ( 13 ) and ( 14 ) . figure 7 . predicted variation of b - v colors for the temperature curves in fig . 5 , using the catelan et al . ( 2004 ) luminosity calibration in eq . ( 10 ) and various assumed period - metallicity envelopes from figs . 1 - 3 . the conversion from the predicted temperatures in fig . 5 to b - v is made using the adopted temperature - color relation of eq . ( 18 ) . line 1 uses the linear envelope of equation 1 , fig . 1 ; lines 2 and 4 for [ fe / h ] @xmath20 -1.8 are for the dashed lines in fig 2 which are eqs . ( 4 ) and ( 4 ) ; line 3 is for the parabolic envelope of fig . ( 5 ) , and line 5 is for the midpoint relation of eq . ( 2 ) . figure 8 . predicted variation of b - v colors for the temperature curves in fig . 6 using the fixed period - metallicity envelope of eq . ( 5 ) , ( fig . 3 ) , and the five different luminosity calibrations of eqs . ( 6 ) , ( 8) , ( 10 ) , ( 12 ) , ( 13 ) , and ( 14 ) . the adopted temperature - color relation is eq . ( the identification of the lines is the same as in the caption to fig . the points are the de - reddened photoelectric mean ( b - v)@xmath56 colors for 160 field rr lyrae stars , plotted vs. layden s measured [ fe / h ] values , on the scale of zinn and west . the line is the predicted run of color with metallicity from eq . ( 19 ) calculated using the catelan et al . ( 2004 ) luminosity calibration in eq . ( 10 ) and the parabolic blue envelope locus to the period - metallicity relation of fig 3 , eq . the line is copied from figs . 7 and 8 . figure 10 . the de - reddened colors of rrab stars in globular clusters for which the number of rrab variables is 10 or greater . the oosterhoff period gap is more prominent than in fig . 4 because of this restriction on the number of variables . the metallicity scale is that of zinn and west , as listed by harris ( 1996 ) . the e(b - v ) reddening values of individual clusters are also from harris . the colors , not listed here , are on the system of ( b - v)@xmath57 , transformed to that system by table 4 of bono et al . ( 1995 ) if a different mean color definition was used in the original literature . the line is the prediction of eq . ( 19 ) , calculated using the luminosity calibration of catelan et al . ( 10 ) ] , and the parabolic period - metallicity envelope of eq . ( 5 ) , together with the color transformation of eq . ( 18 ) . figure 11 . combined data from fig . 9 from field variables and from fig . 10 for cluster variables . crosses are the field stars . dots are the cluster stars . the curve is from eq . it is similar to eq . 19 for ( fe / h ] ) @xmath8 -2.0 but bends more strongly toward the red for more metal poor stars . cccccc mean & -0.359 & + 0.07 to -0.70 & 21 & -0.378 & 0.340 + rms & 0.049 & & & 0.015 & 0.011 + + mean & -0.983 & -0.71 to -1.19 & 16 & -0.329 & 0.335 + rms & 0.039 & & & 0.011 & 0.013 + + mean & -1.468 & -1.20 to -1.79 & 71 & -0.268 & 0.327 + rms & 0.019 & & & 0.007 & 0.005 + + mean & -1.991 & -1.80 to -2.49 & 34 & -0.217 & 0.329 + rms & 0.031 & & & 0.012 & 0.007 + + mean & -2.278 & -2.06 to -2.49 & 7 & -0.200 & 0.331 + rms & 0.059 & & & 0.018 & 0.025 ccccl ngc 6362 & -0.95 & 0.09 & 18 & olech et al . ( 2001 ) + ngc 6712 & -1.01 & 0.45 & 7 & sandage et al.(1966 ) + ngc 6171 & -1.04 & 0.33 & 15 & dickens ( 1971 ) , sandage ( s90a ) + ngc 6723 & -1.12 & 0.05 & 23 & menzies ( 1974 ) , ( s90a ) + ngc 6121 & -1.20 & var & 31 & sturch ( 1977 ) , cacciari ( 1979 ) + ngc 1851 & -1.22 & 0.02 & 21 & walker ( 1998 ) + ngc 5904 & -1.27 & 0.03 & 91 & broccato et al . ( 1996 ) , storm et al . ( 1991 ) , + & & & & caputo et al . ( 1999 ) + ngc 6981 & -1.40 & 0.05 & 24 & dickens & flinn ( 1972 ) , ( s90a ) + ngc 6229 & -1.43 & 0.01 & 30 & borissova et al . ( 2001 ) + ngc 6934 & -1.54 & 0.10 & 68 & kaluzny et al . ( 2001 ) + ngc 5272 & -1.57 & 0.01 & 145 & corwin & carney ( 2001 ) + ngc 3201 & -1.58 & 0.23 & 72 & piersimoni et al . ( 2002 ) + ic 4499 & -1.60 & 0.23 & 63 & walker & nemec ( 1996 ) + ngc 7089 & -1.62 & 0.06 & 17 & lee & carney ( 1999a ) + ngc 7006 & -1.63 & 0.05 & 53 & wehlau et al . ( 1999 ) + ngc 6809 & -1.81 & 0.08 & 4 & olech et al . ( 1999 ) + ngc 4590 & -2.06 & 0.07 & 13 & walker ( 1994 ) + ngc 5466 & -2.22 & 0.00 & 13 & corwin et al . ( 1999 ) + ngc 7078 & -2.26 & 0.10 & 39 & bingham et al . ( 1984 ) + ngc 5053 & -2.29 & 0.04 & 5 & nemec ( 2004 ) + ngc 6341 & 2.28 & 0.02 & 11 & carney et al .	the purpose of the paper is to predict the temperature at the fundamental blue edge ( fbe ) of the instability strip for rr lyrae ( rrl ) variables from the pulsation equation that relates temperature to period , luminosity , and mass . modern data for the correlations between period , luminosity , and metallicity at the fbe for field and cluster rrl are used for the temperature calculation . the predicted temperatures are changed to b - v colors using an adopted color transformation . the predicted temperatures at the fbe become hotter as [ fe / h ] changes from 0 to -1.5 , and thereafter cooler as the metallicity decreases to -2.5 and beyond . the temperature range over this interval of metallicity is @xmath0log @xmath1 = 0.04 , or 640 k at 6900k . the predicted color variation is at the level of 0.03 mag in b - v . the predictions are compared with the observed rrl colors at the fbe for both the field and cluster variables , showing general agreement at the level of 0.02 mag in ( b - v)@xmath2 , which , however , is the uncertainty of the reddening corrections . the focus of the problem is then reversed by fitting a better envelope to the observed fbe relation between color and metallicity for metallicities smaller than -1.8 which , when inserted in the pulsation equation , gives a non - linear calibration of the absolute magnitude of the average evolved level of the hb of m@xmath3 = 1.109 + 0.600 ( [ fe / h ] ) + 0.140 ( [ fe / h])@xmath4 , where the zero point has been set by the observed rr lyrae stars in the lmc at @xmath5v@xmath6 = 19.064 for [ fe / h ] = -1.5 , and using an adopted lmc modulus of ( m - m)@xmath2 = 18.54 . the position of the envelope locus at the shortest periods for the observed period - metallicity correlation differs between the field and cluster variables by @xmath0log p = 0.029 @xmath7 0.007 , the field variables having the shorter periods at the envelope . the field and cluster variables also differ in the near absence of cluster rr lyraes in the -1.7 @xmath8 [ fe / h ] @xmath8 -2.0 metallicity interval , whereas the field variables show no such gap . we aver that these differences require different origins for the field and the cluster variables . a test is proposed by comparing the morphology of the horizontal branches in the local dwarf spheroidals with that in the galactic globular clusters in the inner halo and by relating the differences with the relevant second parameter indicators .
the present work addresses the following issue : what range of momentum components in the deuteron wave function are probed by presently available @xmath0-@xmath1 elastic scattering the question is answered within the context of a model @xcite , based on realistic interactions and currents , that predicts quite well the observed deuteron @xmath2 and @xmath3 structure functions and @xmath4 tensor polarization up to momentum transfers @xmath8 @xmath6 . to this end , the deuteron s- and d - wave components obtained in the full theory @xcite and denoted as @xmath9 , @xmath10=@xmath11 , are truncated , in momentum space , as @xmath12^{1/2 } } \ , u_l(p ) \>\>,\ ] ] where the cutoff momentum @xmath13=@xmath14 ( @xmath15 is the pion mass and @xmath16 is an integer ) and @xmath17=@xmath18 . the constants @xmath19 are fixed through the normalization condition , eq . ( [ eq : norm ] ) below . the wave functions @xmath20 corresponding to @xmath16=1 , 2 and 4 along with the reference @xmath9 are shown in fig . [ fig : deut ] . the @xmath2 , @xmath3 , and @xmath4 observables are calculated using these truncated wave functions by including one - body only and both one- and two - body currents . the results are displayed in figs . [ fig : a][fig : t20 ] . aspects of the present theory @xcite pertaining to the input potential , the boost corrections in the wave functions , the one and two - body charge and current operators , are succinctly summarized in secs . [ sec : wave][sec : ffs ] . here it suffices to note that the results shown in figs . [ fig : a][fig : t20 ] are obtained within a scheme in which these various facets are consistently treated to order @xmath21 . thus the present approach improves and extends that adopted in refs . @xcite and @xcite , in which relativistic corrections of order @xmath21 were selectively retained . for example , the @xmath21 contributions associated with the @xmath22-exchange charge operator were included , while those originating from the one - pion - exchange potential and from boosting the wave function were ignored . the consistent @xmath21 scheme employed here , however , does not alter quantitatively the predictions obtained in refs . @xcite , except for the @xmath3 structure function as discussed in sec . [ sec : res ] . a detailed analysis of the differences between the two approaches is beyond the scope of this work ; it will be presented in ref . @xcite . we assume that the three - momentum transfer @xmath23 in @xmath0-@xmath1 elastic scattering is in the @xmath24-direction . in the breit frame the deuteron has initial and final momenta @xmath25 and @xmath26 in this direction , respectively . in the presence of only single - nucleon currents , the deuteron wave function has to have components with relative momenta @xmath27=@xmath28 larger than @xmath29 in order to produce elastic scattering at momentum transfer @xmath30 . in a limiting case , for example , the nucleons have momenta of @xmath25 and @xmath31 in the initial state and @xmath26 and @xmath31 in the final , all along the @xmath24-axis . thus elastic scattering via one - body currents probes momentum distributions @xmath9 at @xmath32 . this argument , however , does not establish the maximum value of @xmath27 which the observed form factors with @xmath5 @xmath6 are sensitive to . in addition , there is no kinematical limit on initial and final relative momenta for scattering via pair currents . therefore , we use a realistic model of the deuteron to estimate the maximum value of @xmath27 probed by the available data . there is growing interest in chiral effective field theories of nuclear forces and currents @xcite . in such theories the effective lagrangian is obtained after integrating out states with momenta greater than a specified cutoff . we hope that the present results provide an estimate of the deuteron structure obtained in theories with cutoff momentum @xmath13=@xmath14 . the deuteron wave function is dominated by the one - pion exchange potential @xmath33 , where @xmath34=@xmath35 is the momentum transferred by the exchanged pion . the dependence of @xmath33 ( see eq . ( [ eq : vpai ] ) below ) on the magnitude @xmath36 is primarily given by the @xmath37 form factor @xmath38 for @xmath39 . in this limit the leading non - relativistic term of @xmath33 becomes : @xmath40 where @xmath41 is a unit vector . thus the deuteron wave function at relative momentum @xmath27 is expected to be sensitive to @xmath38 up to @xmath42 . the results shown in figs . [ fig : a][fig : t20 ] indicate that the available data on deuteron form factors at @xmath5 @xmath6 confirm the present wave functions up to @xmath43 . after including pair - current contributions , the reference wave function and that truncated at @xmath27=@xmath44 do not give significantly different form factors in this @xmath30-range . the reference and truncated wave functions , respectively , over- and under - predict the observed @xmath2 at @xmath45 @xmath6 ; both under - predict @xmath46 @xmath47 , and fail to reproduce the data point for @xmath48 at @xmath49 @xmath6 . the available data on @xmath2 at larger values of @xmath30 can be used to test the deuteron wave function at @xmath50 ; however , improved theoretical understanding and more accurate data on @xmath51 @xmath47 , and new data on @xmath52 @xmath47 are needed . the deuteron rest - frame wave function is obtained by solving the momentum - space schrdinger equation with the relativistic hamiltonian @xcite @xmath53 where @xmath54 consists of a short - range part @xmath55 parameterized as in the argonne @xmath56 potential @xcite , and of a relativistic one - pion - exchange potential ( opep ) given by @xmath57 { \bbox \tau}_1 \cdot { \bbox \tau}_2 \>\>. \label{eq : vpai}\end{aligned}\ ] ] here @xmath58 is the nucleon mass , @xmath59 is the pion - nucleon coupling constant ( @xmath60=0.075 ) , @xmath61 and @xmath62 are the initial and final momenta of particle 1 in the center - of - mass frame , @xmath34=@xmath63 is the momentum transfer , @xmath64=@xmath65 , and @xmath66=@xmath67 . the monopole form factor : @xmath68 with @xmath69=1.2 gev / c is used in this work . the @xmath70-dependent term characterizes possible off - energy - shell extensions of opep , and leads to strong non - localities in configuration space . in particular , the value @xmath70=1 ( @xmath70=1 ) is predicted by pseudoscalar ( pseudovector ) coupling of pions to nucleons , while @xmath70=0 corresponds to the so - called minimal non - localitychoice @xcite . it has been known for over two decades @xcite , and recently re - emphasized by forest @xcite that these various off - shell extensions of opep are related to each other by a unitary transformation , in the sense that @xmath71 \simeq h^\mu \>\>,\ ] ] if terms of @xmath72-range ( and shorter - range ) are neglected . the hermitian operator @xmath73 is given explicitly in ref . @xcite . this unitary equivalence implies that predictions for electromagnetic observables , such as the form factors under consideration here , are independent of the particular off - shell extension adopted for opep , provided that the electromagnetic current operator , specifically its two - body components associated with pion exchange , are constructed consistently with this off - shell parameter . this point will be further elaborated below . at this stage , however , it is important to recall that forest has constructed relativistic hamiltonians @xmath74 with @xmath70=@xmath75 , each designed to be phase - equivalent to the non - relativistic @xmath76 , based on the argonne @xmath56 potential . given this premise , the @xmath70=0 prescription is adopted for opep from now on , and the superscript is dropped from @xmath77 , for ease of presentation . the resulting momentum - space wave function in the rest frame of the deuteron is denoted with @xmath78 ( here , the argument @xmath31 indicates the rest frame in which the deuteron has velocity @xmath79=0 ) , and is written as @xmath80 \eta^0_0 \>\>,\ ] ] where @xmath81 is the relative momentum , @xmath82 is the angular - momentum projection along the @xmath24-axis , @xmath83 is the pair isospin @xmath84=0 state , and @xmath85 are standard spin - angle functions . the normalization is given by : @xmath86=1 \>\>. \label{eq : norm}\ ] ] the internal wave function in a frame moving with velocity @xmath79 with respect to the rest frame can be written to order @xmath21 as @xcite @xmath87 \psi_m({\bf p};0 ) \nonumber \\ & \simeq & \frac{1}{\sqrt{\gamma } } \left [ 1 -\frac{{\rm i}}{4 m } { \bf v } \cdot ( \bbox{\sigma}_1-\bbox{\sigma}_2)\times { \bf p } \right ] \psi_m({\bf p}_\parallel/\gamma , { \bf p}_\perp;0 ) \>\ > , \label{eq : boost}\end{aligned}\ ] ] where @xmath88 , and @xmath89 and @xmath90 denote the components of the momentum @xmath81 parallel and perpendicular to @xmath79 , respectively . only the kinematical boost corrections , associated with the lorentz contraction ( term @xmath91 ) and thomas precession of the spins ( term @xmath92 ) are retained in eq . ( [ eq : boost ] ) . there are in principle additional , interaction - dependent boost corrections . those originating from the dominant one - pion - exchange component of the interaction have been constructed explicitly in ref . @xcite to order @xmath21 . at this order , however , their contribution to the form factors in eqs . ( [ eq : g0])([eq : g2 ] ) vanishes , since it involves matrix elements of an odd operator under spin exchange between @xmath93=1 states . indeed , this same selection rule also holds for the thomas precession term . therefore the interaction - dependent boost corrections do not contribute to elastic scattering up to order @xmath21 , and are neglected in the following . it should also be noted that terms of order higher than @xmath21 due to lorentz contraction have in fact been included in the second line of eq . ( [ eq : boost ] ) . the factor @xmath94 ensures that the wave function in the moving frame is normalized to one , ignoring corrections of order @xmath95 from the thomas precession term . it is interesting to study the relation between the breit - frame matrix element @xmath96 of the point - nucleon density operator , and the rest frame @xmath97 . one finds , again ignoring thomas precession contributions : @xmath98 after rescaling of the integration variables , @xmath99 , in the last integral . here @xmath100 is the lorentz factor corresponding to @xmath101 in the breit frame . this result is consistent with the naive expectation that the density in configuration space is squeezed in the direction of motion by the lorentz factor @xmath100 or , equivalently , that its fourier transform is pushed out by @xmath100 . for @xmath30=6 @xmath6 , @xmath102 and @xmath103 , corresponding to a 5 % lorentz contraction . the electromagnetic charge and current operators include one- and two - body terms . only the isoscalar parts need to be considered , since the deuteron has @xmath84=0 . the one - body term is taken as @xmath104 u_i \>\>,\ ] ] where @xmath105=@xmath106 in the breit frame , @xmath107 and @xmath108 are the initial and final spinors of nucleon @xmath109 , respectively , with @xmath110 , and @xmath111 and @xmath112 denote the isoscalar combinations of the nucleon s dirac and pauli form factors , normalized as @xmath113=1 and @xmath114=@xmath115 ( in units of n.m . ) . the hhler parameterization @xcite of @xmath116 and @xmath117 is used in this work . the spinor @xmath118 , or rather its transpose , is given by @xmath119 where @xmath81 and @xmath64=@xmath120 are the nucleon s momentum and energy , and @xmath121 is its ( two - component ) spin state . note that @xmath122=@xmath123 . in earlier published work on the form factors of the deuteron @xcite and a=36 nuclei , most recently @xcite , boost contributions were neglected , and only terms up to order @xmath21 were included in the non - relativistic expansion of @xmath124 , namely the well known darwin - foldy and spin - orbit corrections to @xmath125 . in the calculations reported here , however , the full lorentz structure of @xmath124 is retained . the two - body terms included in @xmath126 are those associated with @xmath22- and @xmath127-meson exchanges and the @xmath128 transition mechanism . the @xmath22 and @xmath127 spatial current operators to leading order in @xmath129 are isovector , and therefore their contributions vanish in @xmath0-@xmath1 elastic scattering . the @xmath22-exchange charge operator is obtained , consistently with the off - shell parameter @xmath70 adopted for opep , as @xcite @xmath130 where @xmath131=@xmath132 , @xmath38 is the @xmath22@xmath133@xmath133 form factor defined in sec . [ sec : wave ] , and @xmath134 denotes the hhler parameterization @xcite for the nucleon s isoscalar dirac form factor . it is easy to show , to order @xmath21 , that @xmath135 \simeq \rho_{\rm 1-body}+\rho_\pi^\mu \>\>,\ ] ] where exp(@xmath136 ) is the unitary transformation of sec . [ sec : wave ] . again , we emphasize that the choice @xmath70=0 is made for @xmath137 ( as for opep in sec . [ sec : wave ] ) in the present work . at this point it is important to recall that the @xmath22-exchange charge operator corresponding to @xmath70=1 was included in all our previous studies of light nuclei form factors ( for a review , see ref . @xcite ) , based on non - relativistic hamiltonians , in which the opep part of @xmath138 was taken to be its leading local form , i.e. @xmath66=@xmath64 and @xmath139 in eq . ( [ eq : vpai ] ) . while these calculations are not strictly consistent , since relativistic corrections are selectively retained in @xmath125 but ignored in the potentials and wave functions , they nevertheless give fairly accurate results since the corrections to the wave function are rather small @xcite . vector - meson ( @xmath127 and @xmath140 ) two - body charge operators have been found to give contributions , in calculations of light nuclei form factors @xcite , that are typically an order of magnitude smaller than those associated with the @xmath141 operator . in the present study , only the @xmath127-exchange charge operator is considered : @xmath142 where @xmath143 , @xmath144 and @xmath145 are the @xmath127-meson mass , vector and tensor coupling constants , respectively . here the values @xmath146=0.84 , and @xmath145=6.1 are used from the bonn 2000 potential @xcite , while the @xmath147 monopole form factor @xmath148 has @xmath149=1.2 gev / c . finally , the @xmath128 current is obtained from the associated feynman amplitude as @xmath150 \left ( \overline{u}_2^\prime \ , \gamma_\beta k^\beta_2 \ , \gamma_5 u_2 \right ) \ , { \bbox \tau}_1 \cdot { \bbox \tau}_2 + 1 \rightleftharpoons 2 \>\ > , \label{eq : rpg}\end{aligned}\ ] ] where @xmath151 and @xmath152 are the @xmath128 coupling constant and form factor , respectively , and @xmath153=1 . the value @xmath151=0.56 is obtained from the measured width of the decay @xmath154 @xcite , while the form factor is modeled , using vector - meson dominance , as @xmath155 where @xmath156 is the @xmath140-meson mass . note that in eq . ( [ eq : rpg ] ) the term proportional to @xmath157 in the @xmath127-meson propagator has been dropped , since it vanishes if the nucleons are assumed to be on mass shell . furthermore , in both meson propagators retardation effects have been neglected . the leading terms in a non - relativistic expansion of eq . ( [ eq : rpg ] ) reduce to the familiar expressions for the @xmath128 charge and current operators , as given , for example , in ref . @xcite . however , it is known that these lowest - order expansions , particularly that for @xmath158 , are inaccurate . this fact was first demonstrated by hummel and tjon @xcite in the context of relativistic boson - exchange - model calculations of the deuteron form factors , based on the blankenbecler - sugar reduction of the bethe - salpeter equation . it was later confirmed , in a calculation of the deuteron @xmath3 structure function @xcite , that next - to - leading - order terms in the expansion for @xmath158 , proportional @xmath159 , very substantially reduce the contribution of the leading term . this issue will be returned to later in sec . [ sec : res ] . here , again we stress that the full lorentz structure of @xmath160 is retained in the present study . the deuteron structure functions @xmath161 and @xmath162 , and tensor polarization @xmath48 are expressed in terms of the charge ( @xmath163 and @xmath164 ) and magnetic ( @xmath165 ) form factors as @xmath166 where @xmath167=@xmath168 , @xmath169=@xmath170 , @xmath171=@xmath172 \left[g_1(q)/g_0(q)\right]^2 $ ] , and @xmath173 is the electron scattering angle . the expressions above are in the breit frame , and therefore @xmath30 denotes the magnitude of the three - momentum transfer . the form factors are normalized as @xmath174 where @xmath175 and @xmath176 are the deuteron magnetic moment ( in units of n.m . ) and quadrupole moment , respectively , and are related to breit - frame matrix elements of the nuclear electromagnetic charge , @xmath125 , and current , @xmath177 , operators via : @xmath178 \>\>. \label{eq : g2}\end{aligned}\ ] ] here @xmath179 denotes the standard + 1 spherical component of the current operator . the calculations are carried out in momentum space . the wave functions @xmath180 are given in the second line of eq . ( [ eq : boost ] ) , and the charge and current operators are those described in sec . [ sec : current ] . the one - body current matrix elements involve the evaluation of integrals of the type , in a schematic notation , @xmath181 with @xmath62=@xmath182 and @xmath61=@xmath183 , which are performed by standard gaussian integrations . the two - body current matrix elements , instead , require integrations of the type @xmath184 with @xmath185=@xmath186 and @xmath187=@xmath188 . these integrations are efficiently done by monte carlo techniques by sampling configurations @xmath189 according to the metropolis algorithm with a probability density @xmath190=@xmath191@xmath192 , where @xmath191=@xmath193 . the computer programs have been successfully tested by comparing , in a model calculation which ignored boost corrections and kept only the leading terms in the expansions for @xmath194 and @xmath195 , the present results with those obtained with an earlier , configuration - space version of the code @xcite . in this section we briefly discuss the contribution of the @xmath128 current to the @xmath3 structure function . in fig . [ fig : b_nr ] the results calculated with the current given in eq . ( [ eq : rpg ] ) ( curve labelled @xmath128-r ) are compared with those obtained by using the leading term in its non - relativistic expansion ( curve labelled @xmath128-nr ) , @xmath196 figure [ fig : b_nr ] demonstrates the inadequacy of the approximation ( [ eq : rpg_nr ] ) , a fact which , as mentioned already in sec . [ sec : current ] , has been known for some time @xcite . indeed a more careful analysis shows that the contributions of next - to - leading order terms are not obviously negligible , since they are proportional to the large @xmath127@xmath133@xmath133 tensor coupling constant , @xmath145=@xmath197 in ref . we sketch the derivation again here for the sake of completeness . the vector structure @xmath198 occurring in eq . ( [ eq : rpg ] ) , with @xmath199 defined as @xmath200 can be written in the breit frame as @xmath201 where the index @xmath202 and @xmath23=@xmath203 . up to order @xmath21 included , one finds , in analogy to the non - relativistic expansion of the nucleon electromagnetic current ( with point - nucleon couplings , i.e. @xmath204 and @xmath205 ) @xmath206\ , \label{eq : rpgv1 } \\ \bbox{\gamma}&\simeq&\frac{1}{2 m}\left({\bf p}_1^\prime + { \bf p}_1\right ) + \frac{\rm i}{2 m } \left(1+\kappa_{\rho nn}\right ) \bbox{\sigma}_1 \times { \bf k}_1 \ . \label{eq : rpgv2}\end{aligned}\ ] ] retaining only the leading term in @xmath207 , namely @xmath208 , leads to the expression in eq . ( [ eq : rpg_nr ] ) . some of the corrections proportional to @xmath145 were explicitly calculated in ref . @xcite , and were found to decrease substantially those from the leading term . we note , in passing , that the @xmath128 charge operator , @xmath209 , is proportional to @xmath210 where the small non - local term proportional to @xmath211 has been neglected . the standard form of the @xmath209 commonly used in studies of light nuclei form factors @xcite easily follows . in this case too , however , higher order corrections included in @xmath212 , eq . ( [ eq : rpg ] ) , reduce the contribution of the leading term , although they do not change its sign . a more detailed discussion of this issue will be given in ref . @xcite . we conclude this section with a couple of remarks . the first is that the destructive interference between the one - body and @xmath128 currents is also obtained in recent calculations of the @xmath3 structure function , carried out in the covariant framework based on the spectator equation @xcite . the second remark is that in all earlier studies of light nuclei form factors @xcite additional two - body currents , originating from the momentum - dependent terms of the two - nucleon potential , were included . work is in progress @xcite on an improved treatment of these currents , in an approach similar to that proposed in ref . the work of r.s . was supported by the u.s . department of energy contract de - ac05 - 84er40150 , under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility . the work of v.r.p . was supported by the u.s . national science foundation via grant phy 00 - 98353 . finally , most of the calculations were made possible by grants of computing time from the national energy research supercomputer center . r. schiavilla , to be published . r. schiavilla and d.o . riska , phys . c * 43 * , 437 ( 1991 ) . wiringa , v.g.j . stoks , and r. schiavilla , phys . c * 51 * , 38 ( 1995 ) . u. van kolck , prog . part . phys . * 43 * , 337 ( 1999 ) . m. walzl and u .- g . meissner , phys . * b513 * , 37 ( 2001 ) . j.l . forest , v.r . pandharipande , and a. arriaga , phys . c * 60 * , 014002 ( 1999 ) . forest , phys . c * 61 * , 034007 ( 2000 ) . friar , ann . ( n.y . ) * 104 * , 380 ( 1977 ) . g. hhler , nucl . b114 , 505 ( 1976 ) . marcucci , d.o . riska , and r. schiavilla , phys . c * 58 * , 3069 ( 1998 ) . wiringa and r. schiavilla , phys . lett . * 81 * , 4317 ( 1998 ) . j. carlson and r. schiavilla , rev . phys . * 70 * , 743 ( 1998 ) . r. machleidt , phys . c * 63 * , 024001 ( 2001 ) . d. berg _ et al . _ , phys . * 44 * , 706 ( 1980 ) . e. hummel and j.a . tjon , phys . lett . * 63 * , 1788 ( 1989 ) . van orden , n. devine , and f. gross , phys . lett . * 75 * , 4369 ( 1995 ) . r. gilman and f. gross , nucl - th/0111015 , j. phys . g in press . k. tsushima , d.o . riska , and p.g . blunden nucl . phys . * a 559 * , 543 ( 1993 ) .	what range of momentum components in the deuteron wave function are available @xmath0@xmath1 elastic scattering data sensitive to ? this question is addressed within the context of a model calculation of the deuteron form factors , based on realistic interactions and currents . it is shown that the data on the @xmath2 , @xmath3 , and @xmath4 observables at @xmath5 @xmath6 essentially probe momentum components up to @xmath7 .
the trapezium cluster is an extremely young and rich embedded cluster at the heart of the great orion nebula . discovered in early infrared images by trumpler ( 1931 ) and baade and minkowski ( 1937 ) , it is the best studied of all embedded clusters . with age of approximately 10@xmath8 years it remains one of the most important laboratories for investigations of star formation and early stellar evolution in our galaxy . in particular , the cluster has a well determined and well sampled imf ranging from o stars to substellar objects near the dueterium burning limit ( e.g. , muench et al . 2000 , 2002 ; lucas & roche 2000 ; luhman et al . 2000 , hillenbrand & carpenter 2000 ) . this enables the statistically significant examination of a number of important astrophysical questions , such as the origin and nature of the imf , the frequencies of circumstellar disks and protostars in a young population , and the origin of stellar multiplicity in the earliest stages of stellar evolution . infrared imaging observations of embedded clusters are particularly well suited for addressing such issues ( lada & lada 2003 ) . in a previous infrared imaging study lada et al . ( 2000 ) obtained extensive _ photometry of the cluster . these observations were used both to measure the frequency of infrared excess emission in order to constrain the circumstellar disk fraction in the cluster population and to identify heavily buried protostellar candidates in this young region . they found a relatively high excess fraction of @xmath9 80% for the cluster membership , indicating that the vast majority of stars in the cluster were born surrounded by protoplanetary disks . their observations also revealed a new population of heavily reddened objects deeply buried in molecular gas behind the main cluster . these sources are probably extremely young and mark the location of the most recent and active star formation in the orion nebula region . an appreciable fraction of these sources could be protostellar objects . for cluster members for which spectra were available , lada et al . ( 2000 ) found that the high excess fraction at l - band ( 3.5 @xmath0 m ) was independent of spectral type and thus stellar mass , for stars later than type a. indeed , the excess / disk fraction remained high down to the lowest mass objects near the hydrogen burning limit ( hbl ) . this raised the interesting question of whether substellar objects were also born with circumstellar disks . the answer to this question could provide important clues concerning the nature and formation of brown dwarfs ( muench et al . 2001 , reipurth & clarke 2002 ) . although the @xmath5-band observations of lada et al . were not sensitive enough to examine a statistically significant sample of objects with substellar luminosities , they did indicate that 8 of 10 objects with spectral types later than m6 exhibited strong @xmath5-band excess . since m6 ( e.g. , luhman et al . 1998 ) is the expected spectral type of a young pms star at the hbl , this result suggested the possibility that both stellar and substellar objects form surrounded by dusty disks . indeed , examining deeper @xmath10 observations of the trapezium , muench et al . ( 2001 ) found @xmath11-band ( 2.16 @xmath0 m ) excesses around a significant fraction ( @xmath9 65% ) of the cluster members with substellar luminosities . the fact that many of these brown dwarf candidates with k - band excesses were spatially coincident with optically identified _ hubble space telescope _ `` proplyds '' ( e.g. , bally , odell & mccaughrean 2000 ) indicated that these substellar candidates were indeed surrounded by disks . the excess / disk fraction implied by the muench et al . data was somewhat lower than that derived for the stellar objects in the cluster from the @xmath12 band observations , but higher than predicted by disk models for brown dwarfs which produce smaller amounts of excess emission at @xmath13-band ( e.g. , walker et al . 2004 ; natta et al . moreover , a number of the brown dwarf excess sources were found to lie well below the expected disk locus on the @xmath10 color - color diagram , also inconsistent with expectations of circumstellar disk models . because the magnitude of the infrared excess from a disk increases with wavelength and only very small amounts of dust are required to produce optically thick emission in the 3 - 5@xmath0 m range , @xmath5 band is the optimum wavelength for detecting infrared excesses from circumstellar disks using ground - based telescopes ( haisch , lada & lada 2001 , wood et al . to better constrain the excess / disk fraction for the substellar population of the trapezium cluster requires an extension of the previous l band imaging survey into the brown dwarf regime . deeper @xmath5 band observations would also be capable of improving our knowledge and understanding of the deeply embedded population behind the cluster . do these objects represent a continuation of the cluster forming process that produced the more revealed trapezium cluster ? if so , then their observation may provide interesting insights concerning the nature of the physical mechanism of cluster formation . for example , scally and clarke ( 2002 ) have argued that the trapezium cluster could have been formed from the merger of numerous small primordial subclusters . in this case we might expect to directly detect such subclustering in the embedded population . observation of structure or the lack of it in the embedded population can thus provide important constraints on models of cluster formation . to obtain a sufficiently deep 3 @xmath0 m survey of the trapezium cluster requires a large telescope , a sensitive array of @xmath5 band detectors and the best seeing possible . to achieve these goals we obtained observations of a significant portion of the trapezium cluster with the eso vlt using an @xmath1 ( 3.78@xmath0 m ) filter under conditions of good seeing . in this paper we present these data , their analysis and implications for understanding the nature of extremely young brown dwarfs and the early stages of cluster formation . we obtained deep imaging of the trapezium cluster in the @xmath14 band at 3.78 @xmath0 m using the infrared spectrometer and array camera , isaac ( moorwood et al . 1998 ) , at the vlt ut1 telescope . in order to obtain the best seeing conditions at paranal the observing program was carried out in service mode and data acquisition took place during several nights of good seeing during the southern summer of 2002 . the @xmath14-band sky coverage is shown in figure 1 . table 1 is a log of the observations and lists the positions , integration times and the measured seeing sizes ( fwhm ) for each of the fields . to remove the sky at the @xmath14-band we observed in chopping mode using the vlt chopping secondary mirror in phase with the detector readout . to obtain a difference image the telescope was also nodded 20 arcsec in right ascension in a standard abba pattern . the chopping throw was 20 arcsec in right ascension at a frequency of 0.43 s@xmath15 . each on and off image consisted of 9 coadded frames with an integration time of 0.11 seconds per frame . each nod consisted of 15 chop ( on - off ) cycles . each observation consisted of 15 abba nod cycles with a random jitter of up to 20 arcsec between abba sets . this resulted in a total integration of typically 29.7 minutes for stars observed in both the on and off chop beams . this observing technique gives the best sky subtraction in the @xmath14-band , although it generates a final image that contains many negative sources resulting from sky beam counterparts at the chopping distances . despite this very few known sources ( 15 ) fell into negative images in the reduced chopped image . because the chop throw was smaller than an individual image , most stars appeared in both the on and off images and were thus observed twice in each cycle . to take advantage of this each member of the on - off cycle was sky subtracted using the other member of the cycle as a sky to produce two positive images of a target star for each chopped cycle . these images were then flat fielded using sky flats . the flattened images were then appropriately shifted and stacked to produce the final combined image , which typically consisted of a total of 120 observations of each star . this procedure resulted in the production of a final image that was rectangular ( @xmath9 135 x 85 arcsec ) in shape . some of our final combined images contained only 118 observations since occasional problems with the chopping secondary resulted in the production elongated images on some frames and the few observations with such poor images were discarded before the final image was constructed . figure 2 displays the combined chopped @xmath1 band image for one of our fields ( l5 ) and illustrates the quality of the final images . the seeing was typically below 0.5 arcsec . sources were identified by visual inspection of the vlt chopped images and by comparison to the locations of sources known from published near - infrared catalogs ( e.g. , muench et al 2002 , hillenbrand & carpenter 2000 ) . our final catalog consisted of 424 sources with measurable @xmath1 photometry ; indeed , only 15 known sources were not detected due to negative chop images while 7 bright sources were saturated . by comparison to ground - based @xmath10 surveys we find that we detected @xmath16 of the known sources brighter than @xmath17 ( see also figure [ fig : vlt_cmds]b ) . further , 15 new @xmath1-only sources were identified with the vast majority of them projected along the omc-1 cloud core . last , @xmath18 individual sources appeared on two or more vlt images ( see figure [ fig : area ] ) , yielding quality checks on the photometric and positional accuracy of our final catalog . the iraf appphot package was used to obtain aperture photometry at a large range of radii ( @xmath19 pixels ) and with a sky annulus from @xmath20 pixels . corresponding aperture corrections were calculated using the mkapfile routine and 10 - 20 bright sources for each vlt image , although the images relatively similar psfs yielded similar corrections . for this reason , we chose the 8 pixel aperture photometry for all sources , corresponding to a beamsize of @xmath21 and a typical aperture correction of @xmath22 magnitudes . the 8 pxiel photometry of 13 sources were affected by negative chop images and/or bright stars in the wings of their psfs ; their photometry was adjusted to an appropriately smaller beamsize . zeropoint calibration was performed using standard star observations obtained on 3 of the 4 queue observing nights , and photometric zeropoint offsets between frames as revealed by sources in overlap images were removed by cross - calibration to the photometry of the l1 position . our quoted photometric accuracy as derived from overlap sources is @xmath23 . sources were placed onto the 2mass astrometric grid using plate solutions derived with iraf routine ccmap from 30 - 60 sources on each reduced image . formal rms errors of these solutions were @xmath24 ; however , sources in overlap regions reveal a smaller relative scatter of order @xmath25 . the final catalog is available in electronic format with table [ tab : catalog ] representative of its contents . to preserve positional information lost in the astrometric plate solutions , we provide both equatorial coordinates and pixel positions of each source ( on a specific vlt image ) . sources are also cross - referenced to the flwo - ntt ( muench et al 2002 ) , hillenbrand & carpenter ( 2000 ) and hillenbrand ( 1997 ) identifications . in order to describe the population of @xmath1 sources in our survey region , illustrate the sensitivity of our vlt observations , and select the candidate substellar population in the cluster we present the @xmath7 vs @xmath13 and @xmath26 vs @xmath27 color - magnitude diagrams ( cmds ) in figure [ fig : vlt_cmds ] . a total of 400 sources are detected and resolved in @xmath13 and @xmath1 passbands ; these are plotted in figure [ fig : vlt_cmds]a . most sources have @xmath28 ; however , a significant population of sources have much larger colors , with a handful of sources characterized by extremely red colors ( i.e. , 5 @xmath29 @xmath7 @xmath29 6 magnitudes ) suggesting reddenings up to @xmath30 . it is unclear if the inferred reddenings to these extreme sources are due to pure dust extinction or strong infrared excess emission of the kind that is often associated with protostellar objects , since the reddest of these are not detected at @xmath31 or @xmath27 bands . similar to lada et al . ( 2000 ) , we define the depth of our survey by our ability to detect the photospheres of cluster members . in figure [ fig : vlt_cmds]b we plot the @xmath26 vs @xmath27 color - magnitude diagram for all sources known from previous deep ir surveys within the boundaries of the vlt @xmath1 survey . presuming that the @xmath27 magnitude of these sources is primarily photospheric , we clearly detect the vast majority of sources down to @xmath32 , below which our observations are quite incomplete . alternately , we are sensitive to the photosphere of a 1 myr brown dwarf ( @xmath33 ; baraffe et al 1998 ) seen through 20 visual magnitudes of extinction . muench et al . ( 2001 ) defined a sample of trapezium sources that were postulated to be sub - stellar based on their unreddened @xmath27 band luminosity from the 1 myr isochrone of baraffe et al . the validity of that cmd analysis was bolstered by the fact that the luminosity selected sources were on the whole fainter than members with spectral types earlier than m6 . following muench et al . ( 2001 ) , we use the @xmath27 vs @xmath26 color - magnitude diagram to identify candidate brown dwarfs in the trapezium cluster by comparing their infrared luminosities to the predictions of the baraffe et al . models . we consider all sources whose dereddened luminosities are less than the the predicted luminosity of the hbl for the adopted age ( 10@xmath8 yrs ) and distance ( 400 pc ) of the cluster . by considering all sources below the the extrapolated reddening vector for the hbl we identify 38 sources as substellar candidates in our vlt fields . however we note that this sample represents an upper limit to the true number of substellar cluster members that we detected at @xmath1 since we have not accounted for background / foreground field star contamination which is predicted to be significant at these faint magnitudes . for example , muench et al . ( 2002 ) estimated , by combining observations of control fields near but off the cluster and a model of the extinction distribution produced by the molecular cloud , that field stars account for as much as 20 - 30 % of the objects in this magnitude range . in earlier studies lada et al ( 2000 ) and muench et al ( 2001 ) used @xmath12 and @xmath4 color - color ( cc ) diagrams , respectively , to diagnosis the presence of infrared excess due to disks around stellar and candidate substellar members of the trapezium cluster . we now extend their analysis to include high dynamic range vlt @xmath1 data to better assess the infrared excess fraction across the hbl and into the substellar regime . in figure [ fig : vlt_jhkl ] we present the @xmath4 and @xmath3 infrared cc diagrams for the vlt @xmath1 trapezium sources . there are a total of 345 sources with @xmath34 photometry . sources represented by filled circles are the 38 luminosity selected substellar candidates from the @xmath27 vs @xmath26 color - magnitude diagram . plotted for comparison is the intrinsic locus for dwarfs taken from a merger of optical and ir colors from winkler ( 1997 ) , kenyon & hartmann ( 1995 ) , bessell & brett ( 1988 ) and leggett ( 1992 ) . reddening vectors ( whittet 1989 ) for the giant branch and m6 and m9 dwarfs are also shown . although many sources display colors consistent with reddened photospheric colors , a large fraction ( 50% and 67% , respectively ) display infrared excess relative to the m6 reddening vector on both the @xmath10 and @xmath34 cc diagrams . as expected , the excess fraction is found to be higher in the latter cc diagram . for the candidate 10@xmath8 year old substellar population we expect most objects to have spectral types of m6 or later ( luhman et al . since most of the stars plotted in figure [ fig : vlt_jhkl ] are more luminous than the hbl , and thus likely to be earlier than m6 , those that appear in the infrared excess region of the cc diagrams ( lada & adams 1992 ) must be characterized by significant @xmath26 and @xmath7 excesses . identifying infrared excess for substellar objects , particularly in the @xmath34 cc diagram , is a much more difficult task than for it is for stars above the hbl . this is because the range of intrinsic @xmath7 colors for m6-m9 dwarfs is relatively large and , in fact , comparable to the entire range in intrinsic @xmath35 colors for all earlier spectral types and moreover larger than the corresponding @xmath26 intrinsic color spectrum . nevertheless , a significant fraction ( 42% ) of the substellar candidates fall to the right of the m9 reddening vector on the @xmath34 cc diagram and consequently have significant @xmath7 excess . a similarly high fraction ( 44% ) of substellar candidates display excesses in the @xmath10 cc diagram with respect to the m9 reddening vector . this relatively high fraction from the @xmath10 cc diagram is roughly consistent with the findings of muench et al . ( 2001 ) but inconsistent with expectations of disk models , as mentioned earlier . we also note that about 1/3 of the luminosity selected substellar objects have colors apparently consistent with reddened stars of spectral type earlier than m6 . some of these sources are likely field stars . for others their relatively blue colors could be the result of photometric uncertainties in the observed colors . moreover , the @xmath10 observations were obtained at a different epoch than the @xmath1 observations , and consequently variability , if present , could contaminate some of the @xmath7 colors . in addition , this result is also reminiscent of the recent spectroscopic study of slesnick , hillenbrand and carpenter ( 2004 ) who found a similar sized fraction of luminosity selected substellar candidates in the trapezium cluster to be characterized by spectral types earlier than the nominal m6 substellar boundary . the nature of this apparently subluminous stellar population is unknown . both our cc diagrams also show sources which fall into relatively forbidden regions of the diagrams . in figure [ fig : vlt_jhkl]a there is a population of sources with very red @xmath26 colors and relatively blue @xmath36 colors . some of these same sources also appear to be unexpectedly blue in @xmath7 color in figure [ fig : vlt_jhkl]b ; as was discussed in lada et al ( 2000 ) and muench et al ( 2001 ) many of these sources are proplyds ( odell wen & hu 1993 ) and their colors are possibly contaminated by scattered light and/or emission from the photoionized envelopes surrounding the underlying disks . to further investigate the behavior of the infrared excess / disk fraction across the hbl we divided the source distribution in the @xmath26 vs @xmath27 cmd into 7 luminosity bands , approximating equal sized steps in unreddened @xmath27 magnitude and logarithmic mass using predicted @xmath37 magnitudes from the the baraffe et al ( 1998 ) 1 myr isochrone . the magnitude bins and corresponding predicted masses are listed in table [ tab : trap_cc ] . we further constructed an extinction limited sample by including in each bin only sources with extinctions @xmath38 . the fraction of sources in each luminosity selected bin displaying infrared excess greater than their formal 1 sigma photometric errors was counted in the corresponding @xmath10 , @xmath34 and @xmath39 color - color diagrams . the m6 spectral type boundary was used to calculate the infrared excess fraction for all bins . in addition for bins 5 - 7 the infrared excess fraction was also calculated using the m9 boundary , since these bins are likely to contain brown dwarfs . the corresponding excess fractions are listed in table [ tab : trap_cc ] , while in figure [ fig : trap_cc_irx ] we plot these fractions as a function of the luminosity bin center . clearly , trapezium sources display excess at all magnitudes and in all diagrams with the @xmath34 and @xmath39 diagrams giving nearly identical results . as already shown by lada et al ( 2000 ) , the fraction of bright sources displaying excess is larger when using @xmath6 as a color diagnostic . ) sources showing excess in these diagrams appears lower than published in lada et al ( 2000 ) for two reasons . first , the @xmath40 bandpass of the vlt data yields redder photospheric colors than the @xmath41 bandpass of the flwo data making the identification of a infrared excess somewhat more difficult at @xmath1 for late spectral types . second , we are consistently counting from the m6 rather than the m5 boundary used lada et al ( 2000 ) ] yet as @xmath37 increases into the fiducial brown dwarf regime ( band 5 - 7 ) , the excess fraction traced by @xmath7 appears to decrease , although the uncertainty in the excess fraction increases due the larger range of intrinsic @xmath6 colors for m6-m9 sources . note , our results for band 7 are likely upper limits because of incompleteness , which manifests itself in our ability to detect only those faint sources with disks . this decrease in @xmath6 excess fraction is _ opposite _ the behavior traced by @xmath26 , which increases for the brown dwarf luminosity bins and recovers the @xmath42 excess fraction quoted by muench et al ( 2001 ) . the physical origin of this difference in behavior is unknown . nonetheless , the @xmath1 measurements indicate a relatively high infrared excess and likely disk fraction for the substellar population of this cluster . merging the results for luminosity bands 6 and 7 from table 2 we estimate an infrared excess fraction for the substellar population of the cluster of 50 @xmath2 20 % from the @xmath34 diagram . our measurements also hint that the disk fraction for the substellar population is somewhat lower than that of the stellar population . however , considering the uncertainties and the possible effects of field star contamination , one can not yet draw a firm conclusion about the presence of a decrease in excess fraction for the substellar population in the cluster . there are a number of limitations to our estimate of the excess / disk fraction for the substellar population from our vlt @xmath1 data . first , our measurements of the excess fraction are incomplete for the lowest luminosity bin resulting in an overestimate for the excess fraction in that bin . second , the lack of specific knowledge about the intrinsic infrared colors of the objects results in the relatively large quoted uncertainties ( @xmath43 ) for our estimate of the excess fraction of the luminosity selected substellar candidates . the lack of knowledge of source intrinsic color also presumably affects the derived excess fraction at brighter magnitudes as well since an m0 star , for example , will require 0.3 magnitudes of @xmath6 excess in order to be counted as an excess source using the m6 boundary . given that any age spread for trapezium members will result in a mix of mass ( as traced by spectral type ) as a function of luminosity , assigning a single spectral type boundary only acts to underestimate the true excess fraction . in addition we have not attempted to correct for foreground / background source contamination which is likely present for these faint sources . the presence of such contaminating sources also results in an underestimate of the true excess fraction . liu et al . ( 2003 ) have shown that a more complete assessment of infrared excess in substellar sources can be made if the spectral types and thus the intrinsic colors of the stars are individually known and taken into account in measuring infrared excess emission . combining spectra with infrared @xmath34 photometry of substellar candidates in taurus and ic 348 , liu et al . ( 2003 ) derived a @xmath7 band excess / disk fraction of 77% for stars they selected with late spectral types . this was a significantly higher fraction than that ( @xmath9 33% ) inferred from analysis of the @xmath34 cc diagram for the same sources . therefore it would be extremely valuable to examine the infrared colors of those substellar candidates in the trapezium that have been spectroscopically classified so that their intrinsic infrared colors are known . over the past few years various workers have collected spectra of faint stars in the trapezium cluster ( i.e. , hillenbrand , 1997 ; lucas et al . 2001 ; luhman et al . 2000 ; slesnick , hillenbrand and carpenter , 2004 ) . from these studies we find a total of 24 objects with spectral types of m6 or later that were also detected in our @xmath1 images . in figure [ cc_w_spectra ] we show the @xmath4 and @xmath3 cc diagrams for this spectroscopically selected sample of substellar candidates . with the assumption of dwarf colors for the spectroscopically selected sample we can convert the spectral types to intrinsic substellar colors . with knowledge of the spectroscopically derived intrinsic @xmath44 colors of the stars , we derive an excess fraction of 70% @xmath2 15% from comparison with their observed @xmath44 colors . this is in excellent agreement with the fraction derived by muench et al . ( 2001 ) from photometry using the @xmath4 cc diagram and roughly consistent with our estimate based solely on @xmath4 photometry of sources detected at @xmath1 . however this value is higher than would be expected by traditional disk models . using the @xmath45 data we derive in a similar manner an excess fraction of 52% @xmath2 20 % for the spectoscopically selected substellar candidates . although this represents a significant excess fraction , the calculated value is formally lower than both the fraction derived by liu et al . ( 2003 ) for taurus and ic 348 using @xmath7 colors and the fraction derived for the same sources from our @xmath4 colors . however given the relatively large uncertainties it is difficult to assess the significance of this difference . moreover it is difficult to compare the results between our spectroscopically selected trapezium sample and the liu et al . @xmath1 ic 348 plus taurus sample , since neither can be considered complete and may not even be fully representative of the entire substellar populations in the respective clouds . finally , an additional uncertainty may arise from our assumption of dwarf colors for these objects since they are expected to have lower surface gravities than field dwarfs . theoretical models predict that lower surface gravity stars should have somewhat bluer @xmath6 colors , leading to an underestimate of the excess fraction when assuming dwarf colors ( liu et al . we find that these same model atmospheres ( i.e. , baraffe et al . 2002 ) predict slightly bluer @xmath36 colors and no change in the @xmath26 colors for lower surface gravity atmospheres with the result that the excess fraction derived assuming dwarf colors is not significantly altered . the relatively large excess fraction derived from @xmath10 colors is unexpected and as speculated earlier by muench et al . ( 2001 ) may be the result of scattered light or emission from the photoionized envelopes of the disks , since many of the sources with these colors appear to be proplyds . the @xmath5 band imaging survey of lada et al ( 2000 ) resulted in the identification of a previously unrecognized population of heavily reddened objects toward the trapezium cluster . seventy - eight sources were found to have large @xmath6 colors ( i.e. , @xmath6 > 1.5 mag . ) suggestive of objects deeply embedded in molecular gas and dust . for example , for a normal interstellar reddening law ( e.g , mathis 1990 ) this would correspond to @xmath46 @xmath47 27 magnitudes for a typical naked field star seen through the cloud and approximately @xmath46 @xmath47 15 magnitudes for a young star with a circumstellar disk . the spatial distribution of these red objects was found to differ from that of the bulk of the ( less reddened , @xmath48 5 mag . ) cluster membership , but to be similar to that of the dense molecular ridge located behind the cluster . the molecular ridge has long been known to be a site of star formation , haboring the massive protostellar objects irc2 and bn - kl . the @xmath5 band survey indicated that the star formation activity in the ridge was considerably more active than previously suspected . many of the new and heavily buried sources were found to be characterized by the extreme infrared excess in the @xmath12 cc diagram often observed in protostars . our deeper @xmath1 observations have revealed 73 objects with @xmath7 @xmath47 1.5 magnitudes and 40 with @xmath7 @xmath47 2.0 magnitudes within the six fields we imaged . within these fields 25 objects were identified as having @xmath6 @xmath47 1.5 magnitudes in the earlier survey by lada et al . ( 2000 ) and 48 are newly identified members of the deeply embedded population ( most of which were below the @xmath5 band sensitivity limit in the previous survey ) . together with the earlier data of lada et al . , our observations indicate that that deeply embedded population consists of approximately 126 members , roughly 20% the size of the foreground and more revealed trapezium cluster . the vlt sources with @xmath7 @xmath47 1.5 magnitudes are plotted as filled squares in figure 1 . these sources form a tight distribution that more or less follows the molecular ridge ( see figure 7 of lada et al . 2000 and also johnstone & bally 1999 ) . the extremely high extinctions inferrred for these sources coupled with their relatively high surface density , clearly indicates that the vast majority are embedded in the cloud and not reddened background field stars . moreover , the reddest of these sources have @xmath7 colors suggestive of visual extinctions of up to 100 magnitudes ! such extreme extinctions typically arise from dense envelopes of deeply embedded protostellar objects and not from the more distributed dense gas of a molecular cloud core . as suggested by lada et al ( 2000 ) many such objects are likely protostellar in nature and still in the process of forming . in addition our new observations revealed 15 previously undetected @xmath1 only sources , plotted as filled diamonds in figure 1 . as seen in this figure twelve of these are found coincident with the molecular ridge and are likely additional members of the deeply embedded population . the brightest of these is the source vlt-129 ( @xmath49 ) which lies @xmath50 southwest of the bnkl and is characterized by a @xmath7 color of @xmath47 6 magnitudes . the population of deeply embedded objects ( deos ) traces the most recent epoch of star formation in the trapezium cluster region . the spatial structure of the deeply embedded population possesses an imprint of the physical processes responsible for its creation and thus can provide important insights into the earliest phases of cluster formation process . in their study of embedded clusters lada & lada ( 2003 ) found that embedded clusters could be characterized by two basic structural types . type 1 clusters exhibited hierarchical spatial structure characterized by significant sub - clustering in their surface density distributions . such highly structured systems are thought to be a signature of the important role of turbulence in the evolution of molecular gas to form stars . type 2 clusters exhibited centrally - concentrated structure typically characterized by relatively smoothly varying radial surface density profiles . such centrally condensed structures are thought to be a signature of the global dominance of gravity in the formation of the stellar system . it is not clear whether these two types of structures represent two different primordial structures or two differing evolutionary states of an embedded cluster . most embedded clusters appear to be type 2 structures , yet most progenitor molecular clouds appear to be filamentary in shape and highly structured , at least on large scales . this perhaps suggests that centrally condensed type 2 structures may form from more structured type 1-like initial configurations . indeed , in a recent numerical simulation of cluster formation from a turbulent cloud , bate , bonnell and bromm ( 2003 ) found that filamentary clouds could form stars which fall together to form small groups within a larger cluster forming system . earlier numerical simulations by scally and clarke ( 2000 ) have suggested that relaxed , centrally condensed stellar clusters can form from the merger of smaller subclusters . they suggested that the orion nebula cluster ( and by implication the trapezium cluster which forms its inner core ) could have formed from the merger of subclusters provided that these subclusters were relatively small and numerous . existing surface density maps of the trapezium cluster suggest that it is a type 2 or centrally concentrated embedded cluster ( hillenbrand & hartmann 1998 ; lada et al . 2000 ; scally & clarke 2002 ) . in figure [ proto_cluster_map ] we plot the surface density distribution of the embedded population ( contours ) along with that of the less extincted members of the trapezium cluster ( grey scale ) . our deeper observations allow us to map these distributions at higher angular resolution than previously possible . the deeply embedded population is found to be significantly structured appearing to consist of five subclusters oriented more or less along the filamentary molecular ridge . the basic parameters of the subclusters are listed in table [ tab : subclusters ] these five subclusters account for approximately 55% of the stars in the deeply embedded population . if the formation of the deeply embedded population is a continuation of the same physical process that produced the foreground trapezium cluster , then it is tempting to identify the subclusters with the fundamental units or building blocks out of which the cluster continues to be assembled . if this interpretation is correct , then it suggests that a centrally condensed cluster may be formed from the merger of smaller subclusters which themselves formed intitially in filamentary dense molecular gas in accord with the suggestion of scally and clarke ( 2002 ) . the deep infrared observations also allow us to examine the structure of the foreground trapezuim cluster in more detail than previously possible . infrared @xmath10 cc diagrams ( see figure 4 and lada et al . 2000 , figure 8) show that the bulk of the trapezium cluster is characterized by extinctions less than about 5 - 6 magnitudes . therefore to map the structure of the foreground trapezium cluster we ploted in figure [ proto_cluster_map ] the spatial surface density distribution of stars with @xmath51 1.5 magnitudes ( corresponding to @xmath46s of approximately 10 magnitudes or less ) . the surface density map of the cluster is dominated by a strong central peak at the site of the trapezium ( n@xmath52 @xmath53 6000 pc@xmath54 ) , in agreement with previous studies . however , our higher resolution map reveals additional significant structure in the cluster . three secondary peaks with peak surface densities of 3000 pc@xmath54 or greater are found to the north and east of the primary peak . together the four subclusters account for about 28% of the total population of the foreground cluster . the basic parameters of these subclusters are also listed in table [ tab : subclusters ] . these secondary peaks are reminiscent of the satellite subclusters observed in ic 348 ( lada & lada 1995 ) another otherwise highly centrally concentrated embedded cluster . comparison of the embedded source and cluster surface density distributions generally confirms earlier results in that , for the most part , the two populations have different spatial distributions which are offset from one another . however , one peak in the deo distribution is coincident with the primary peak in the trapezium cluster distribution . although significant this peak in the deo distribution may not be in fact as prominent as it appears in the map since at least 3 of the sources in it are foreground proplyds with very red colors . in addition the strongest peak in the deo distribution , which is coincident with bn - kl region , is near , but offset from the second strongest peak in the foreground cluster distribution . thus it appears that at least some of the structure in the foreground trapezium cluster reflects the more primordial structure in the background embedded population . these observations suggest that the foreground trapezium cluster is still sufficiently young that it is not yet fully relaxed into a centrally condensed , isothermal - like stellar distribution . moreover , the presence of structure in the embedded source population suggests that the cluster will need to undergo still more structural evolution in order to eventually incoporate the presently embedded sub - clusters and form a fully relaxed and centrally concentrated stellar system . we have obtained deep 3.8 @xmath0 m @xmath1 observations of the trapezium cluster using the eso vlt . we have imaged a significant fraction of the inner 5 x 5 region of the cluster . our observations extend the previous @xmath5 band survey of lada et al . ( 2000 ) to fainter cluster members and include approximately 38 objects with luminosities below that expected for cluster members at the hbl . in addition , we detected 24 objects with previously known spectral types that are later than that corresponding to the hbl ( m6 ) . combining our deep @xmath1 data with previous @xmath55 observations we have examined the frequency of infrared excess emission among the candidate substellar population . we derive an infrared fraction of 50 @xmath2 20% for the substellar population from analysis of the @xmath3 colors of both luminosity selected and spectroscopically selected samples of substellar candidates . this result suggests a significant disk fraction for substellar sources , consistent with earlier results for this and other clusters ( muench et al . 2001 ; liu et al . the presence of disks around substellar objects provides evidence that the formation mechanism for brown dwarfs and more massive stars is a physically similar process as has been previously argued ( muench et al . 2001 , liu et al . 2003 ) . due to issues of completeness and contamination of the sample by foreground / background stars we can not determine the true excess / disk fraction for the substellar population and it is therefore not clear whether our measured value of @xmath9 50% for substellar objects is significantly different from the @xmath9 65% fraction found for a complete and representative sample of the stellar members of the cluster from @xmath3 observations . our observations also confirm that a number of substellar candidate sources possess unusual colors on the @xmath4 cc diagram . this is true for both luminosity selected and spectroscopically selected substellar candidates . these sources exhibit significantly larger infrared @xmath44 excesses than predicted by conventional disk models and thus appear redder for their @xmath36 color . moreover , a few of these sources exhibit @xmath45 colors that are two blue for their @xmath36 color and do not appear as @xmath45 excess sources on the @xmath3 cc diagram . the origin of these anomalous colors is unclear . photometric errors , source variability and perhaps contamination by proplyd emission may all contribute to explain these unusal characteristics . our deep @xmath1 observations also reveal a significant number of new members of the deeply embedded population of young objects which lies buried in the molecular cloud behind the cluster . as a whole this embedded population represents the location of the most recent star forming events in the trapezium region . we find the surface density distribution of the deeply embedded population to follow that of the background molecular ridge and to be highly structured , consisting of a string of at least 5 significant sub - clusters . if the embedded population represents a continuation of the star formation process responsible for the creation of the more revealed trapezium cluster , then these embedded sub - clusters are likely representative of the primordial building blocks out of which the cluster was and perhaps is still being assembled . additional evidence for this possibility is found in high resolution map of the surface density distribution of the trapezium cluster itself which shows , in addition to a strong and centrally concentrated peak , significant substructure in the form of small satellite subclusters similar to those in the background cloud . these findings are generally consistent with the suggestion of scally and clarke ( 2002 ) that the trapezium cluster could have been assembled from the merger a large number of small ( @xmath56 50 stellar members ) subclusters over the past 10@xmath8 years . we are grateful to richard j. elston for useful advice and suggestions pertaining to the image reduction and analysis and to kevin luhman for useful discussions . eal acknowledges support from nsf grants , ast97 - 3367 and ast02 - 02976 to the university of florida . l1 & 2002 - 10 - 10 & 52557.242325 & 5:35:19.507 & -5:22:47.95 & 60 & 29.70 & 0.53 + l2 & 2002 - 10 - 10 & 52557.284110 & 5:35:19.580 & -5:23:55.45 & 59 & 29.21 & 0.34 + l3 & 2002 - 10 - 12 & 52559.242141 & 5:35:24.003 & -5:23:51.06 & 59 & 29.21 & 0.85 + l4 & 2002 - 10 - 12 & 52559.282716 & 5:35:06.943 & -5:21:45.10 & 60 & 29.70 & 0.39 + l5 & 2002 - 10 - 14 & 52561.357352 & 5:35:12.723 & -5:23:16.17 & 59 & 29.21 & 0.48 + l6 & 2002 - 12 - 19 & 52627.255228 & 5:35:12.749 & -5:24:22.58 & 60 & 29.21 & 0.42 + 1 & 9 & 05 35 04.174 & -05 22 24.52 & 0 & l4 - 033 & 274.916 & 1134.839 & 1.14 & 12.041 & 0.056 & & & 244 + 2 & 9 & 05 35 04.389 & -05 21 23.51 & 0 & l4 - 012 & 319.637 & 268.436 & 1.14 & 14.007 & 0.126 & & & + 3 & 1 & 05 35 04.956 & -05 21 09.45 & 0 & l4 - 032 & 439.672 & 68.905 & 1.14 & 10.767 & 0.019 & 00856 & & + 4 & 1 & 05 35 05.583 & -05 21 40.58 & 0 & l4 - 011 & 572.784 & 510.949 & 1.14 & 14.428 & 0.132 & 00789 & & + 5 & 1 & 05 35 06.175 & -05 22 12.51 & 0 & l4 - 001 & 698.199 & 963.988 & 1.14 & 10.811 & 0.019 & 00681 & 508 & 3064 + 6 & 1 & 05 35 06.203 & -05 21 36.06 & 0 & l4 - 013 & 703.921 & 446.854 & 1.14 & 15.031 & 0.187 & 00796 & 601 & + 7 & 1 & 05 35 06.289 & -05 22 02.75 & 0 & l4 - 004 & 722.275 & 825.489 & 1.14 & 8.593 & 0.007 & 00716 & 538 & 286 + 8 & 1 & 05 35 06.355 & -05 22 11.63 & 0 & l4 - 002 & 736.148 & 951.438 & 1.14 & 14.575 & 0.138 & 00685 & 509 & + 9 & 1 & 05 35 06.454 & -05 22 07.61 & 0 & l4 - 003 & 757.271 & 894.435 & 1.14 & 11.691 & 0.029 & 00700 & 526 & 287 + 10 & 1 & 05 35 06.472 & -05 21 18.97 & 0 & l4 - 009 & 760.816 & 204.533 & 1.14 & 11.755 & 0.030 & 00837 & 636 & 288 + 11 & 1 & 05 35 06.896 & -05 22 09.31 & 0 & l4 - 005 & 850.663 & 918.522 & 1.14 & 12.715 & 0.048 & 00692 & 521 & 298 + 12 & 0 & 05 35 7.046 & -05 22 17.06 & 0 & & & & & 99.000 & -1.000 & 00667 & 497 & + 13 & 1 & 05 35 07.223 & -05 21 34.64 & 0 & l4 - 007 & 919.862 & 426.863 & 1.14 & 13.624 & 0.077 & 00803 & 603 & + 14 & 1 & 05 35 07.242 & -05 22 03.97 & 0 & l4 - 006 & 923.860 & 842.755 & 1.14 & 13.245 & 0.063 & 00711 & 534 & + 15 & 0 & 05 35 7.505 & -05 21 45.93 & 0 & & & & & 99.000 & -1.000 & 00772 & 718 & + 1 & 7.00 & 9.66 & 5.000 & 1.200 & 83 & 0 & m6 & 38 & 67 & 66 + 2 & 9.66 & 10.50 & 1.200 & 0.570 & 49 & 0 & m6 & 22 & 71 & 73 + 3 & 10.50 & 11.53 & 0.570 & 0.250 & 63 & 0 & m6 & 27 & 57 & 60 + 4 & 11.53 & 12.44 & 0.250 & 0.130 & 70 & 0 & m6 & 40 & 63 & 69 + 5 & 12.44 & 13.41 & 0.130 & 0.080 & 43 & 1 & m6 & 52 & 73 & 76 + 6 & 13.41 & 14.50 & 0.080 & 0.045 & 34 & 7 & m6 & 57 & 52 & 44 + 7 & 14.50 & 16.50 & 0.045 & 0.020 & 45 & 28 & m6 & 39 & 73 & 71 + + tsc - a & 05:35:16.234 & -05:23:25.19 & 7200 . & 0.08 & 93 & trapezium core + tsc - b & 05:35:15.734 & -05:22:29.79 & 5600 . & 0.06 & 43 & near bnkl + tsc - c & 05:35:21.689 & -05:23:51.70 & 4400 . & < 0.05 & 16 & + tsc - d & 05:35:20.490 & -05:22:33.72 & 3600 . & 0.05 & 20 & + + desc - a & 05:35:15.672 & -05:23:24.92 & 2400 . & 0.05 & 14 & trapezium core + desc - b & 05:35:14.090 & -05:22:41.14 & 3200 . & 0.06 & 16 & bnkl + desc - c & 05:35:17.739 & -05:22:05.22 & 2400 . & 0.05 & 12 & + desc - d & 05:35:13.989 & -05:23:49.94 & 2400 . & < 0.05 & 10 & orion s. + desc - e & 05:35:12.264 & -05:24:43.50 & 2000 . & < 0.05 & 8 & +	we present deep 3.8 @xmath0 m @xmath1 imaging observations of the trapezium cluster in orion obtained with the eso vlt . we use these observations to : 1 ) search for infrared excess emission and evidence for protoplanetary disks associated with the faint , substellar population of this young cluster , and 2 ) investigate the nature and extent of a recently discovered population of deeply embedded sources located in dense molecular gas behind the cluster . we detected 38 @xmath1 sources with substellar luminosities . in addition , we detected 24 @xmath1 sources that were spectroscopically classified as substellar objects in previous studies . examining the infrared colors of all these sources we determine an infrared excess fraction of 50 @xmath2 20 % from the @xmath3 colors for both the luminosity selected and spectroscopically selected substellar samples . this finding confirms the presence of infrared excess , likely due to circumstellar disks , around a significant fraction of the cluster s substellar population , consistent with the indications of earlier observations obtained at shorter ( @xmath4 ) wavelengths . our deep @xmath1 imaging survey also provides new information concerning the deeply embedded population of young objects located in the molecular cloud behind the cluster and revealed in an earlier @xmath5 band imaging survey of the region . in particular , our present @xmath1 survey doubles the number of sources in the cluster region known to possess extremely red @xmath6 colors . these objects exhibit @xmath7 colors indicative of deeply buried , possibly protostellar , objects that likely mark the site of the most recent and ongoing star formation in the region . we find the surface density distribution of the deeply embedded population to follow that of the background molecular ridge and to be highly structured , consisting of a string of at least 5 significant subclusters . these subclusters may represent the primordial building blocks out of which the cluster was and perhaps still is being assembled . these observations may thus provide insights into the early stages of cluster formation and appear consistent with recent simulations that suggest that the trapezium cluster may have formed from numerous but small primordial subclusters .
the theory of automatic groups was introduced in @xcite . if a group has an automatic structure , there exist efficient algorithms for enumerating elements and checking two elements which differ by multiplication on the right by only one generator . in @xcite and further works , it was shown that many classes of finitely generated groups , such as hyperbolic groups , finitely generated abelian groups , braid groups , have automatic structures . it is known that part of the definition of automatic structure can be replaced with a property of paths in cayley graphs . thus automatic structures for groups are closely related to geometric structures of their cayley graphs . in this paper , we observe a discrete analogue of the growth of volumes which are known in the field of riemannian geometry . in short , for an automatic group , we consider the numbers of finite graphs isomorphic to a given finite graph in the balls of various radii centred at the identity element in the cayley graph and compute the growth series of the numbers by use of a finite state automaton which is part of an automatic structure . first , we prepare some notations to describe our concern and the main result . let @xmath1 be a finitely generated group . a generating set @xmath2 for @xmath1 is called _ symmetric _ if @xmath3 , then @xmath4 is also in @xmath2 . in the followings , we always assume that generating sets for groups are symmetric . let @xmath5 denote the cayley graph of @xmath6 . for simplicity , we usually write @xmath7 instead of @xmath5 . we consider that the length of each edge of @xmath7 is equal to @xmath8 and extend the distance to all points of @xmath7 . this makes @xmath7 into a geodesic metric space . let @xmath9 be the ball of radius @xmath0 in @xmath7 centred at the identity element of @xmath1 . let @xmath10 be a directed subgraph of @xmath7 whose edges are labelled by elements of @xmath2 . a _ morphism _ @xmath11 is defined to be a function that maps each vertex to a vertex and each labelled directed edge to a directed edge with the same label . furthermore the initial and final endpoints of a directed edge in @xmath10 are mapped to the initial and final endpoints respectively of the corresponding directed edge of @xmath7 . let @xmath12 be the number of morphisms from @xmath10 to @xmath9 . set @xmath13 for @xmath14 and @xmath15 . correspondingly we consider the formal power series @xmath16 and call it the _ growth function _ for @xmath10 . in the case where @xmath10 is a vertex , @xmath17 counts the number of elements of @xmath1 whose shortest representatives with respect to @xmath6 have exactly length @xmath0 . in @xcite , charny computed the growth functions of artin groups of finite type in that case . the problem of computing the growth functions for various subgraphs arose from attempts to generalise the ising model in quantum mechanics to more general geometries by saito(@xcite ) . in @xcite , epstein , iano - fletcher and uri zwick computed growth functions in many cases . they proved that @xmath18 is a rational function if @xmath1 has a geodesic automatic structure . furthermore they showed that one can choose a common denominator of the rational function independently of @xmath10 if @xmath1 is hyperbolic and @xmath2 is fixed . they pointed out that similar assertions seem to hold in other classes of automatic groups . the aim of this paper is to consider the problem in the case where the groups are abelian . the main theorem is the following . * the main theorem . * _ let g be a finitely generated abelian group whose rank is @xmath19 and @xmath2 be a symmetric generating set . let @xmath10 be any non - empty finite connected labelled directed subgraph of @xmath5 . then @xmath20 is a rational function with integral coefficients and the denominator is @xmath21 . _ in section @xmath22 , we recall the definition of automatic structure . also we see that finitely generated abelian groups are automatic . in section @xmath23 , we construct finite state automata which are part of automatic structures for finitely generated abelian groups . furthermore we prepare some lemmas to compute the growth functions . in section @xmath24 , we prove the main theorem . in section @xmath25 , we see some simple examples . in the followings , a finitely generating set for a group is also called an _ alphabet_. each element of an alphabet is called a _ letter_. a _ word _ over an alphabet is a finite sequence of letters of the alphabet . for an alphabet @xmath2 , we denote by @xmath26 the set of all words over @xmath2 . a subset of @xmath27 is called a _ language _ over @xmath2 . regular language _ is the set of all words accepted by a finite state automaton . see @xcite for the basic notations about finite state automata . for a finite state automaton @xmath28 , here and subsequently @xmath29 denotes the regular language consists of all words accepted by @xmath28 . there is a natural map @xmath30 since @xmath26 can be seen as the free monoid by @xmath2 . for a word @xmath31 , we denote its image in @xmath1 by @xmath32 instead of @xmath33 for simplicity . definition ( automatic structure ) . let @xmath1 be a group and @xmath2 be a symmetric generating set . @xmath1 is said to be _ automatic _ if there exist a finite state automaton @xmath34 over @xmath2 and a finite state automaton @xmath35 over @xmath36 for each @xmath37 which satisfy the following conditions . \(1 ) the map @xmath38 is surjective . \(2 ) the pair of strings @xmath39 is accepted by @xmath35 if and only if @xmath40 and @xmath41 are accepted by @xmath34 and @xmath42 . we see a relation between automatic structures for groups and geometric structures of their cayley graphs . for the purpose , we define a property of paths in geodesic metric spaces . ( fellow travellers property ) . let @xmath43 be a geodesic metric space . let @xmath44 \to x$ ] and @xmath45 \to x$ ] be two paths ( not necessarily geodesics ) . if @xmath46 , we set @xmath47 for all @xmath48 . fix a constant @xmath49 . we say @xmath50 and @xmath51 are @xmath52-_fellow travellers _ if @xmath53 for all @xmath54 such that @xmath55 . the constant @xmath52 is called a _ fellow travellers constant_. the next theorem guarantees that fellow travellers property is equivalent to the condition ( 2 ) in the definition of automatic structure . for a word @xmath56 , @xmath57 denotes the path from the identity element to @xmath32 in @xmath7 according to @xmath58 . * theorem 2.1 ( @xcite ) . * _ let @xmath34 be a finite state automaton and @xmath59 be a regular language such that the map @xmath60 is surjective . then @xmath34 is part of an automatic structure for @xmath6 if and only if there exists @xmath61 such that for @xmath62 and @xmath63 satisfying @xmath64 , @xmath65 and @xmath66 are @xmath52-fellow travellers in @xmath5_. in the followings , we say that @xmath59 satisfies @xmath52-_fellow travellers property _ if there exists a constant @xmath52 stated in theorem 2.1 . in the remain of this section , we see that finitely generated abelian groups have automatic structures . first , we define an order of words . ( shortlex order ) . let @xmath2 be an ordered alphabet . for any pair of words @xmath67 , we define @xmath68 if @xmath69 is shorter than @xmath58 , or they have the same length and @xmath69 comes before @xmath58 in lexicographical order . the order obtained in this manner is called the _ shortlex order _ of @xmath26 . a word @xmath31 is called _ shortlex _ if @xmath58 is the minimum representative in the shortlex order among all words which represent the same element @xmath32 . if there exists a word acceptor @xmath34 for @xmath6 which only accepts shortlex words , we say that @xmath1 has a _ shortlex automatic structure_. notice that if a word acceptor @xmath34 is part of a shortlex automatic structure , then the natural map @xmath70 is bijective and all paths in @xmath7 according to the words of l(w ) are geodesics in @xmath5 . further @xmath59 is _ prefix closed _ , which means that all prefixes of any word in @xmath59 are also accepted by @xmath34 . . sketch of proof . it is easily seen that there exists a finite state automaton which only accepts the shortlex words for @xmath6 with any order in @xmath2 . thus we only verify that the shortlex words satisfy the fellow travellers property . let @xmath71 be an ordered alphabet . we put every word in _ normal form _ @xmath72 such that @xmath73 are non - negative integers . further we identify @xmath72 with the @xmath0-tuple @xmath74 . for @xmath75 and @xmath76 , we say that @xmath69 is contained in @xmath58 if @xmath77 for all @xmath78 and write @xmath79 . if @xmath58 and @xmath80 are distinct words which represent the same group element , we say that we have a _ relation _ a relation @xmath82 is _ minimal _ if it is minimal for the partial order @xmath83 on pairs of strings where @xmath84 if @xmath79 and @xmath85 . let @xmath40 and @xmath41 be any pair of shortlex words which satisfies @xmath86 . then @xmath87 and @xmath41 differ only by their minimal relation . it is shown that the size of such minimal relations are bounded ( ( * ? ? ? * lemma 4.3.2 ) ) , which completes the proof . from the proof of theorem 2.2 , the next lemma follows immediately . * lemma 2.3 . * _ let @xmath1 and @xmath2 be the same as theorem 2.2 and assume that @xmath2 is ordered . let @xmath40 and @xmath41 be any pair of shortlex words which satisfies @xmath88 . let @xmath89 and @xmath90 be the normal forms of @xmath54-prefixes of @xmath40 and @xmath41 respectively . then there exists a constant @xmath61 such that @xmath91 . _ in the followings , we regard @xmath52 in lemma 2.3 as the fellow travellers constant of a shortlex automatic structure for a finitely generated abelian group with an ordered generating set . let @xmath1 be a finitely generated abelian group and @xmath2 be an ordered symmetric generating set . there exists a shortlex automatic structure for @xmath6 from theorem 2.2 . in this section , we consider a canonical construction of word acceptors which are suited to compute the growth functions . first , we see a rough example which is stated in figure 1 . the left hand side in the figure depicts the cayley graphs of @xmath92 . the right hand side depicts word acceptors both of which accept the shortlex words for @xmath6 with the order @xmath93 . arrows and states corresponding to the inverse elements and the failure states are omitted for simplicity . in the cayley graphs , the dotted lines indicate @xmath94 . the thick lines indicate paths according to words accepted by the automata stated in the right . the numbers in circles in the figure indicate the correspondence between vertices of @xmath94 and accept states of the word acceptors . notice that the vertices of @xmath94 are classified more minutely if we use the word acceptor below . now we prepare some notations to describe the rigorous construction of the word acceptors . in an automaton , an _ incoming _ ( resp . _ outgoing _ ) _ arrow _ of a state @xmath95 is an arrow whose target ( resp . source ) is @xmath95 . an arrow is called a _ loop _ if the source and target of the arrow coincide . in the followings , we use the notation incoming ( resp . outgoing ) arrows " only for arrows whose source and target states are distinct , thus loops are excluded from the set of incoming ( resp . outgoing ) arrows of any state . we fix a presentation for @xmath6 , index the relators and represent the @xmath78-th relator as @xmath96 . let @xmath97 be the word length of @xmath98 and @xmath99 . if a letter of @xmath2 may not occur infinitely in shortlex words for @xmath6 , then the number of occurrences of the letter does not exceed @xmath100 . let @xmath101 be any positive integer which is greater than @xmath100 . definition ( canonical word acceptor ) . construct a word acceptor which is part of a shortlex automatic structure for @xmath6 as follows . \(1 ) put the start state . \(2 ) put the unique failure state . add @xmath102 loops , each of which is labelled by each letter of @xmath2 , to the failure state . ( 3 - 1 ) for each state added in the previous procedure ( 1 ) , ( 3 - 3a ) or ( 3 - 3b ) , we continue the following steps . ( 3 - 2 ) let @xmath95 be the added state and @xmath50 be the label of the incoming arrow of @xmath95 . if @xmath95 is the start state , we regard @xmath50 as empty . classify each letter of @xmath2 except @xmath50 into three cases where the letter may occur infinitely , finitely or may not occur after the prefix words according to the transitions from the start state to @xmath95 . ( 3 - 3 ) let @xmath103 be one of the classified letters in ( 3 - 2 ) . in each case where the letter may occur infinitely , finitely or may not occur , do ( 3 - 3a ) , ( 3 - 3b ) or ( 3 - 3c ) respectively . ( 3 - 3a ) construct a line which is rooted at @xmath95 and which consists of arrows and states as the followings . first , we regard @xmath95 as the end of the line whose length is equal to @xmath104 . add an arrow labelled by @xmath103 to the end , put its target state and regard the state as the new end . continue these operations until the length of the line achieves @xmath101 . finally add a loop labelled by @xmath103 to the end of the line . see figure 2(a ) . ( 3 - 3b ) assume that @xmath103 may occur at most @xmath105 times . notice that @xmath106 by definition . construct a line rooted at @xmath95 whose length is equal to @xmath105 the same as ( 3 - 3a ) . finally , add an arrow , whose label is @xmath103 and whose target is the failure state , to the end of the line . see figure 2(b ) . ( 3 - 3c ) add an arrow , whose label is @xmath103 and whose target is the failure state , to @xmath95 . ( 3 - 4 ) if there remains a letter classified in ( 3 - 2 ) , we return to ( 3 - 3 ) . ( 3 - 5 ) if there remains a state listed in ( 3 - 1 ) , we return to ( 3 - 2 ) . ( 3 - 6 ) if there appear all letters of @xmath2 as labels of outgoing arrows or a loop in each state , we finish the construction . otherwise , return to ( 3 - 1 ) . we regard all states except the failure state as accept states . the automaton constructed in this manner is called the @xmath101-_canonical word acceptor _ for @xmath6 and denoted by @xmath107 . note that each state except the start state and the failure state has just one incoming arrow and at most one loop . thus @xmath107 is a tree if all loops , the failure state and all incoming arrows of the failure state are eliminated . therefore , there exists exactly one path from the start state to each accept state in @xmath107 . example 3.1 . figure 3 depicts the @xmath101-canonical automaton for some @xmath101 in the case where @xmath108 with the order @xmath109 . arrows and states corresponding to the inverse elements and the failure state are omitted . the lengths of oblique and vertical lines are equal to @xmath101 . in the followings , we often regard a finite state automaton as a labelled directed graph . a _ walk _ in a directed graph is a sequence of vertices @xmath110 such that adjacent pairs @xmath111 are consistent with the directed edges of the graph including their directions . the number of edges in a walk is called the _ length _ of the walk . notice that a path can be seen as a walk with no repeated vertices . in this paper , a walk whose length is equal to @xmath112 is called an @xmath112-_walk_. let @xmath113 be the number of accept states in an automaton and the accept states be numbered from @xmath8 to @xmath113 . a row vector in @xmath114 is called the _ characteristic vector _ of the @xmath115-th state if the @xmath115-th entry is equal to @xmath8 and the others are equal to @xmath104 . a @xmath116 matrix is called the _ transition matrix _ of the automaton if each @xmath117-entry is the number of arrows from the @xmath78-th state to the @xmath118-th state . for the states of @xmath107 , we denote the start state by @xmath119 and the other accept states by @xmath120 for @xmath121 , where @xmath113 is the number of accept states of @xmath107 . let @xmath122 be the number of @xmath118-walks from @xmath119 to @xmath120 in @xmath107 and @xmath123 . * lemma 3.2 . * _ for each state @xmath120 of @xmath107 , @xmath124 is a rational function with integral coefficients . furthermore the denominator of @xmath124 is @xmath21 with some nonnegative integer @xmath19 which does not exceed rank @xmath1 and the numerator is @xmath125 with some positive integer @xmath126 . _ proof . let @xmath28 be the transition matrix of @xmath107 and @xmath127 be the characteristic vector of @xmath128 for @xmath129 . then @xmath130 . thus the proof consists in the computation of this formula . first , we define a directed subgraph @xmath131 of @xmath107 as follows . the set of vertices of @xmath131 consists of all states which are contained in the unique path from @xmath119 to @xmath120 in @xmath107 . the set of directed edges of @xmath131 consists of all arrows which are contained in the path and all loops whose source and target states are contained in the set of vertices of @xmath131 . then the number of @xmath118-walks from @xmath119 to @xmath120 in @xmath107 coincides with that in @xmath131 . let @xmath112 be the length of the unique path from @xmath119 to @xmath120 in @xmath131 . in the case where @xmath132 has no loops , @xmath133 and it satisfies the assertion of the statement . thus it is sufficient to consider the case where @xmath132 has loops . in that case , we refine @xmath132 as follows . if some state @xmath128 except @xmath119 and @xmath120 has no loop , we put an arrow from the previous state of @xmath128 to the next state of that , and eliminate @xmath128 with the incoming and outgoing arrows . if @xmath120 has no loop , we eliminate @xmath120 with the incoming arrow . consequently , each of the remained states except @xmath119 has a loop . let @xmath134 be the number of the remained states , thus the number of the eliminated ones is @xmath135 . we denote the remained graph by @xmath136 . notice that there exists exactly one path in @xmath136 which starts at @xmath119 and whose length is equal to @xmath137 . each state of @xmath136 is renamed @xmath138 , whose index is determined as the following . the state @xmath119 of @xmath132 turns into @xmath139 . the other states are renumbered from @xmath22 to @xmath134 according to the order of vertices along the unique path , thus the end of the path turns into @xmath140 . see figure 4 . let @xmath141 be the transition matrix of @xmath136 . then @xmath142 we denote by @xmath143 the characteristic vector of @xmath138 in @xmath136 . since the number of @xmath118-walks from @xmath119 to @xmath120 in @xmath132 coincides with the number of @xmath144-walks from @xmath139 to @xmath140 in @xmath136 , we have @xmath145 @xmath146 it is obvious from the fundamental theorem of abelian groups that @xmath137 does not exceed rank @xmath1 . for the proof of the main theorem , we prepare a general formula derived from lemma 3.2 . in the next lemma , we follow the notations in the proof of lemma 3.2 . * lemma 3.3 . * _ let @xmath147 and @xmath148 be positive integers . then @xmath149 is a sum of rational functions each of whose denominators is @xmath150 for @xmath151 and each of whose numerators is @xmath152 such that each @xmath153 is a positive integer , each @xmath154 is a nonnegative integer for @xmath155 and @xmath156 . _ proof . since @xmath157 @xmath158 and the @xmath159-entry of @xmath160 for any @xmath161 is equal to @xmath8 , the statement follows immediately . in this section , we prove the main theorem . let @xmath1 , @xmath2 and @xmath10 be the same as the statement of the main theorem . let @xmath162 be the diameter of @xmath10 . fix a base point @xmath163 . let @xmath52 be the fellow travellers constant of a shortlex automatic structure for @xmath6 defined in the paragraph after lemma 2.3 . we fix a positive integer @xmath101 which is greater than @xmath164 and construct @xmath107 for @xmath6 . let @xmath69 be a vertex of @xmath165 . to simplify notation , we use the same symbol @xmath69 for the corresponding element of @xmath1 . a word @xmath58 is called an _ accepted word _ of @xmath69 if @xmath58 is accepted by @xmath107 and @xmath166 . notice that there exists exactly one accepted word for an element of @xmath1 from the uniqueness property of shortlex words . a path from the identity element to @xmath69 in @xmath7 is called the _ accepted path _ of @xmath69 if the sequence of the labels along the path coincides with the accepted word of @xmath69 . the accepted paths are geodesics in @xmath7 since @xmath107 accepts only shortest words among representatives for elements of @xmath1 . when we follow the arrows in @xmath107 according to the labels of the accepted word of @xmath69 starting at @xmath119 , we end at some state of @xmath107 . the state is called the _ final state _ of @xmath69 . in the followings , it is necessary to pay attention to the difference between walks , including paths , in @xmath107 and accepted paths in @xmath7 . let @xmath167 be the normal form of the accepted word of @xmath168 . from the assumption that @xmath0 is greater than @xmath169 , there exist indices @xmath170 such that @xmath171 are greater than or equal to @xmath101 . thus the unique path from @xmath119 to @xmath120 in @xmath107 contains exactly @xmath113 states each of which is the source and target of a loop . see figure 5 . let @xmath172 be the normal form of the accepted word of @xmath173 . since the final state of @xmath173 is @xmath120 , the walk according to @xmath172 in @xmath107 contains the path from @xmath119 to @xmath120 . thus @xmath174 for any @xmath175 and @xmath176 for @xmath177 . let @xmath126 be any vertex of @xmath10 and @xmath178 be the normal form of the accepted word of @xmath179 . by lemma 2.3 , @xmath180 , thus @xmath181 for @xmath177 . consider a word @xmath182 such that @xmath183 for any @xmath184 and @xmath185 for @xmath177 . since @xmath176 and @xmath186 , @xmath187 for @xmath177 . combining this observation with the assumption that @xmath178 is shortlex , we conclude that @xmath182 is also the normal form of a shortlex word . furthermore , @xmath182 is a representative for @xmath188 since @xmath7 is homogeneous and @xmath1 is commutative . therefore @xmath182 is the normal form of the accepted word of @xmath188 . assume that @xmath189 , it means that @xmath190 , for some integer @xmath191 . then @xmath192 from the construction of @xmath193 , and it follows that @xmath194 . thus the statement follows immediately . let @xmath0 , @xmath195 and @xmath120 be the same as the assumptions of lemma 4.1 . let @xmath196 , thus @xmath197 satisfies @xmath198 . each @xmath199-walk from @xmath119 to @xmath120 in @xmath107 determines a unique morphism @xmath200 because the natural map @xmath201 is bijective . furthermore @xmath202 satisfies @xmath203 and @xmath204 from lemma 4.1 . on the contrary , each morphism @xmath205 such that @xmath204 and that the final state of @xmath173 is @xmath120 determines a unique @xmath199-walk from @xmath119 to @xmath120 in @xmath107 according to the accepted word of @xmath173 because of the uniqueness of shortlex words . thus there is a one to one correspondence between the set of walks from @xmath119 to @xmath120 in @xmath107 whose lengths are @xmath206 and the set of morphisms from @xmath10 to @xmath9 such that the images of @xmath10 are not contained in @xmath207 and that the final states of the images of @xmath208 are @xmath120 . proof of the main theorem . let @xmath209 be the number of morphisms from @xmath10 to @xmath9 such that the images of @xmath10 are not contained in @xmath207 and that the final states of the images of @xmath208 coincide with @xmath120 . let @xmath210 , then @xmath211 . let @xmath122 be the number of walks from @xmath119 to @xmath120 in @xmath107 whose lengths are equal to @xmath118 as stated in section 3 . from the paragraph after lemma 4.1 , @xmath212 is equal to @xmath213 for @xmath214 . let @xmath28 be the transition matrix of @xmath107 and @xmath127 be the characteristic vector of @xmath128 . then we have @xmath215 @xmath216 @xmath217 such that @xmath218 is a nonnegative integer , each @xmath219 is a positive integer , each @xmath220 for @xmath221 is a nonnegative integer and @xmath222 from lemma 3.3 . further , each @xmath218 does not exceed @xmath223 for any @xmath115 and there exists an index @xmath115 such that @xmath224 from the fundamental theorem for abelian groups . let @xmath225 . then we have @xmath226 such that @xmath227 and @xmath228 are polynomials with integral coefficients and that all coefficients of @xmath228 are positive . since @xmath229 , @xmath230 is not a factor of the numerator of the last rational function . consequently , the last rational function is irreducible . in this section , we see two examples . the first one is presented in @xcite . recall that @xmath12 denotes the number of morphisms @xmath231 . 20 r. charney , _ geodesic automation and growth functions for artin groups of finite type _ , mathematische annalens . * 301 * ( 1995 ) , 307 - 324 . d. b. a. epstein , j. w. cannon , d. f. holt , s. v. f levy , m. s. paterson and w. p. thurston , _ word processing in groups _ , jones and bartlett , boston.1992 . d. b. a. epstein , a. r. iano - fletcher , and uri zwick , _ growth functions and automatic groups _ , experimental mathematics . * 5 * ( 1996 ) , no . 4 , 297 - 315 . k. saito , _ the limit element in the configuration algebra for a discrete group _ congress mathematicians ( kyoto 1990 ) , 1991 , 931 - 942 .	in this paper , we consider the formal power series whose @xmath0-th coefficient is the number of copies of a given finite graph in the ball of radius @xmath0 centred at the identity element in the cayley graph of a finitely generated group and call it the growth function . epstein , iano - fletcher and uri zwick proved that the growth function is a rational function if the group has a geodesic automatic structure . we compute the growth function in the case where the group is abelian and see that the denominator of the rational function is determined from the rank of the group .
in the paradigm of hierarchical structure formation , the evolution of the primordial density field acting under gravitational instability drives dark matter to cluster and collapse into virialised objects ( haloes ) . such haloes provide the potential wells into which baryons fall and galaxies subsequently form @xcite . haloes , galaxies and their environments also interact and merge as structure formation unfolds with time . it therefore follows that the properties of a galaxy should be correlated with the properties of its host halo , and that a galaxy s environment , its host halo s environment , and the dark matter density field are all related in some measurable way . such galaxy / halo / dark matter correlations with environment have lead to a variety of work examining the environmental dependence of the physics of galaxy formation , both theoretical and observational . measurements of the galaxy two - point correlation function and halo occupation distribution function ( hod ) have shown that more massive , brighter , redder , and passive early - type galaxies tend to be more strongly clustered and hence presumably located in denser environments , while the reverse is true for galaxies that have lower mass , are fainter , bluer and star forming ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? a more direct probe of the influence of environment is the local density field of neighbouring galaxies around each galaxy ( defined in various ways ) . these techniques are better suited to analysing targeted halo and galaxy environment correlations and have proven valuable in the current era of large galaxy survey data sets , where galaxy catalogues can be simultaneously ` sliced ' in multiple orthogonal directions to isolate the dependence of specific galaxy properties on environment ( e.g. * ? ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in undertaking any such analysis the choice of environmental indicator is important and no one standard has yet emerged . many of the above cited papers involve disparate selection criteria , research methods and goals , making direct comparisons between them difficult . the definition of environment can vary from two - point clustering and marked clustering statistics , to the number or luminosity density within a fixed spherical or cylindrical aperture , to the measured density enclosed by the @xmath0-th nearest neighbour . a further complication is that these methodologies can be performed in either two ( projected ) or three ( redshift space ) dimensions . as a consequence , some analyses have yielded irreconcilable results . all methods that attempt to quantify the environment around a galaxy require some parameter choices . those that involve a spherical or cylindrical aperture must first choose a fixed smoothing scale within which to measure the local galaxy over- or under - density . on the other hand , when environment measures involve the @xmath0-th nearest neighbour , the choice of @xmath0 instead becomes important . once @xmath0 is fixed , this statistic adapts its scale to keep the signal - to - noise constant . but how should one interpret a statistic that combines the physical processes from widely disparate scales across one smoothly varying curve ? and how should this be compared with statistics that instead fix the scale along the same curve ? further complicating comparison are the selection criteria of a dataset itself , its geometry and volume , and the redshift and magnitude uncertainties of the galaxies in it . in short , the measurement of ` environment ' used in various studies can be completely different , and environmental correlations should be interpreted and compared with caution . some environment measures can have advantages and disadvantages for particular research goals . a number of authors have tested and compared a few environment measures ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in general , while the environment measures are correlated , they often exhibit considerable scatter between them . the primary goal of this project is to compare a variety of published environment measures using a single well constrained data set . for this purpose , we take a dark matter halo catalogue and construct a mock galaxy catalogue designed to have approximately the same global statistical properties as the sloan digital sky survey ( sdss ; * ? ? ? * ) main galaxy sample . we then are able to compare the galaxy environment measures to halo mass , dark matter density , and to each other . we also attempt to answer some important questions , such as : do the different environment methodologies break nicely into different groups that optimally sample the underlying density field in particular ways ? do the statistics of various galaxy properties change dramatically in different environment bins measured in different ways ? can we find a more fundamental definition of environment that is measurable observationally ? this paper is organised as follows : in section [ sec : mod ] we outline the mock galaxy catalogue that was generated , constrained by the sdss , and used to study environment measures . in section [ sec : meas ] we review the range of environment measures available in the literature that are used as part of this study . having established the method , section [ sec : res ] explores how the different measures relate to the dark matter halo mass , galaxy colour and large - scale dark matter environment for each galaxy . we also explore how the measures relate to each other for an individual galaxy . finally in section [ sec : discussion ] we discuss and summarise our findings . we begin with the millennium simulation @xcite which is a large @xmath1-body simulation of dark matter structure in a cosmological volume . the millennium simulation uses the gadget tree - pm code @xcite to trace the evolution of 10 billion dark matter particles across cosmic time in a cubic box of @xmath2 on a side , with a halo mass resolution of @xmath3 . it adopts the concordance @xmath4cdm cosmological parameters , chosen to agree with a combined analysis of the two - degree field galaxy redshift survey ( 2dfgrs ; * ? ? ? * ) and the first - year wilkinson microwave anisotropy probe data ( wmap ; * ? ? ? * ) : @xmath5 , @xmath6 , @xmath7 , @xmath8 , and @xmath9 . the haloes are found by a two - step procedure . first , all collapsed haloes with at least 20 particles are identified using a standard friends - of - friends group - finder with linking parameter @xmath10 . then , post - processing with the substructure algorithm subfind @xcite subdivides each friends - of - friends halo into a set of self - bound subhaloes . we note that comparable halo properties are found using other structure finders ( see * ? ? ? [ ] [ ] [ 1][0]@xmath11 [ ] [ ] [ 1][0]@xmath12 [ l][][1][0]@xmath13 [ l][][1][0]@xmath14 . [ note this figure differs from the published version as this one contains all haloes , while the published version contains only haloes hosting at least one galaxy.],title="fig:",width=309 ] from the millennium simulation halo merger tree at @xmath15 , we construct a mock galaxy catalogue using the halo occupation method described in ( * ? ? ? * hereafter s06 ) and ( * ? ? ? * hereafter ss09 ) ; we refer the reader to these papers for details . other halo - model descriptions of galaxy clustering conditional luminosity functions ( e.g. * ? ? ? * ) and subhalo abundance matching ( e.g. * ? ? ? * ) would produce similar mock catalogues , although an advantage of the ss09 approach is that it includes a strongly constrained model of galaxy colours . s06 describes how the luminosities and real - space and redshift - space galaxy positions are modelled . our model distinguishes between the ` central ' galaxy in a halo and all the other galaxies ( ` satellites ' ) . we assume that central galaxies have the same positions and velocities as the haloes in the dark matter simulation . in other words , central galaxies are at the centre of the haloes , and the satellites are located around them . an important assumption in the model is that all galaxy properties their numbers , spatial distributions , velocities , luminosities , and colours are determined by halo mass alone . these galaxy properties are constrained by sdss observations , including the luminosity function @xcite , luminosity - dependent two - point clustering @xcite , and the colour - magnitude distribution and colour - dependent clustering @xcite . note that the clustering constraints result in a mock catalogue that approximately reproduces the observed environmental dependence of luminosity and colour , on scales of @xmath16 to @xmath17 . the number of satellite galaxies in the model follows a poisson distribution with a mean value that increases with halo mass . the satellites are distributed around the halo centre so that they follow a @xcite profile with the mass - concentration relation from @xcite . we assign redshift - space coordinates to the mock galaxies assuming that a galaxy s velocity is given by the sum of the velocity of its parent halo plus a virial motion contribution that is drawn from a maxwell - boltzmann distribution with dispersion that depends on halo mass ( s06 ) . we specify a minimum @xmath18-band luminosity for the galaxies in the catalogue , @xmath19 , to stay well above the resolution limit of the millennium simulation , avoiding any issues of completeness that may bias our results . we generate luminosities for the central galaxies , while accounting for the stochasticity between their luminosities and host halo mass , and then we generate the satellite luminosities so that the observed luminosity distribution is reproduced for @xmath13 ( s06 ) . we model the observed @xmath20 colour distribution at a given luminosity as the sum of two gaussian components , commonly referred to as the ` blue cloud ' and ` red sequence ' . our colour model has five constraints as a function of luminosity : the mean and scatter of the red sequence , mean and scatter of the blue cloud , and the blue fraction . we assume that the colour distribution at fixed luminosity is approximately independent of halo mass , and that the satellite colour distribution varies such that its mean increases with luminosity ( i.e. , the satellite red fraction increases with luminosity in a particular way ) . these two assumptions are tested and verified with galaxy group catalogues in @xcite . this procedure produces a mock galaxy catalogue containing 1.84 million galaxies , of which 29 percent are satellites . galaxies occupy haloes with masses ranging from @xmath21 to @xmath22 . we also construct a mock light cone from the catalogue by selecting galaxies that are within a radial distance of @xmath23 from one corner of the box . this gives an opening angle of @xmath24 degrees and a depth of @xmath23 , for which right ascension and declinations are determined . the analysis in section [ sec : res ] is carried out using a sample of galaxies that are common to both the box and the cone and are chosen so not to be affected by edges . figure [ fig : pop ] shows the mean number of galaxies as a function of halo mass , for two luminosity thresholds ( @xmath25 ) . by construction , the number of galaxies consists of the number of central galaxies plus the number of satellites , such that @xmath26 \label{fit}\ ] ] where , @xmath27 \label{ncenm}\ ] ] and @xmath28 ( see appendix a2 of ss09 for details ) . all of the free parameters depend on luminosity . the slope of the power law , @xmath29 , is nearly unity . one may define a parameter @xmath30 , which is equal to or slightly larger than @xmath31 @xcite , and is proportional to the minimum halo mass : @xmath32 . this determines the mass above which haloes typically host at least one satellite galaxy . therefore , since for @xmath33 the minimum halo mass is @xmath34 , the mean number of galaxies rises rapidly like a linear power law at masses larger than twenty times this value , or @xmath35 , as seen in figure [ fig : pop ] . at the high halo mass end , galaxy number shows a near linear relationship with dark matter halo mass , which occurs by construction in the halo occupation model . this implies that the number of galaxies per unit dark matter mass is constant , or put another way , each galaxy contributes the same mass of dark matter to the cluster . this is in agreement with the findings of @xcite and to some degree is the natural consequence of a structure built hierarchically . this also agrees with @xcite who find that galaxy distributions are only affected by the host dark matter halo , and not by the surrounding density field , for the sdss galaxy group catalogue . the @xmath36band luminosity function of galaxies in the mock catalogue is shown in the left - hand panel of figure [ fig : lumfunc ] and is compared to both the observed sdss luminosity function @xcite and a popular semi - analytic galaxy formation model @xcite . in the right - hand panel of figure [ fig : lumfunc ] we show the mock two - point correlation functions of all , red , and blue galaxies and compare them with the equivalent sdss measurements of @xcite . note that the colour - dependent two - point function measured by @xcite is slightly different from that constrained in the mock , likely due to the presence of the sloan great wall in the real data , an unusually massive supercluster at @xmath37 . we have made the mock galaxy catalogue as realistic as possible , and although the catalogue reproduces the observed environmental dependence of luminosity and colour , there are nonetheless a few limitations to the model . for example , we have assumed virialised ( dynamically relaxed ) dark matter haloes even though some haloes are not , such as those having recently experienced a merger ( e.g. * ? ? ? we have also assumed that central galaxies are always the brightest galaxy in a halo and lie at the centre of their potential well , although in a nonzero fraction of haloes , especially massive haloes , this assumption is not valid @xcite . finally , we force satellite galaxy properties to depend only on halo mass , not on halo - centric position , although there is evidence of such a dependence at fixed mass ( e.g. * ? ? ? * ; * ? ? ? while our mock galaxy catalogue resembles a spectroscopic catalogue , some environment measures used in the literature are based on photometric data ( e.g. * ? ? ? * ) ; for tests with such measures one can add scatter to the redshifted mock galaxy positions , for example . c \| c \| c num . & method & author + & * neighbours * & + 1 & 3rd nearest neighbour & muldrew + 2 & projected voronoi & podgorzec & gray + 3 & mean 4th & 5th nearest neighbour & baldry@xmath38 + 4 & 5 neighbour cylinder & li@xmath39 + 5 & 7th projected nearest neighbour & ann + 6 & 10 neighbour bayesian metric & cowan@xmath40 + 7 & 20 neighbour smooth density & choi & park@xmath41 + 8 & 64 neighbour smooth density & pearce + + & * aperture * + 9 & @xmath42 ( @xmath43 ) & grtzbauch & conselice@xmath44 + 10 & @xmath45 ( @xmath46 ) & gallazzi@xmath47 + 11 & @xmath45 ( @xmath43 ) & grtzbauch & conselice + 12 & @xmath45 ( @xmath48 ) & gallazzi@xmath47 + 13 & @xmath49 ( @xmath43 ) & grtzbauch & conselice + 14 & @xmath50 spherical & croton@xmath51 + + & * annulus * + 15 & @xmath52 ( @xmath43 ) & wilman & zibetti@xmath53 + 16 & @xmath54 ( @xmath43 ) & wilman & zibetti@xmath53 + 17 & @xmath55 ( @xmath43 ) & wilman & zibetti@xmath53 + 18 & @xmath56 ( @xmath43 ) & wilman & zibetti@xmath53 + 19 & @xmath57 ( @xmath43 ) & wilman & zibetti@xmath53 + 20 & @xmath58 ( @xmath43 ) & wilman & zibetti@xmath53 + there are many different methods of measuring galaxy environment available in the literature . most of these can be categorised into two broad groups : those which use neighbour finding and those that use a fixed aperture . an overview of the methods used in this work are presented in the following subsections and summarised in table [ tab : comp ] along with the authors who implemented them . the principle of nearest neighbour finding is that galaxies with closer neighbours are in denser environments . to create a standard measure for this , a value of @xmath0 is chosen that specifies the number of neighbours around the point of interest . in its simplest form , the projected surface density of galaxies , @xmath59 , can then be defined as @xmath60 where @xmath0 is the number of neighbours within the projected distance @xmath61 , the radius to the @xmath0-th nearest neighbour . one disadvantage of quantifying environment using projected statistics is that two galaxies can appear close together when they are in fact just a chance alignment and are actually separated by a larger distance in the third dimension . while there is no simple way to overcome this observationally , one can adopt a velocity cut about each galaxy , typically of order @xmath621000 km @xmath63 , to minimise the number of such alignments . for data where a third dimension has been measured for each galaxy ( e.g. redshift ) , the denominator of equation [ eqn : nn2d ] is replaced by the enclosed volume : @xmath64 when using three dimensions careful consideration of redshift distortions are needed and this often leads to two dimensional projected distances often being used . the nearest neighbour estimator was recently applied to the galaxy and mass assembly catalogue ( gama ; * ? ? ? * ) by @xcite using the distance to the first nearest neighbour above a given luminosity , although typically 3 - 10 neighbours are used . variations on the @xmath0-th nearest neighbour approach have been proposed in an attempt to improve the robustness of statistic as a measure of local density . one such method used by @xcite was to take the average of two different neighbour densities , in their case the 4th and 5th nearest neighbour projected surface densities . an alternative proposed by @xcite was to use the distance to every neighbour up to the tenth instead of just the distance to the tenth to calculate the density . they adopted a bayesian metric such that @xmath65 where @xmath66 is empirically determined so that the mean of @xmath67 matches the number density when the density is estimated on a regular grid for a uniform field , and @xmath68 is the distance to neighbour @xmath69 . one can also use numerical simulations to guide the nearest neighbour calibration . calculating densities using neighbours has long been used in smooth particle hydrodynamics ( sph ) and this technique can be applied to galaxies in simulations . sph calculates the density around a point by weighting each neighbour based on its distance from the point , with the smoothed galaxy density defined as @xmath70 here , @xmath0 is the number of neighbours used and @xmath71 is the weighting given by @xmath72 where @xmath18 is the distance to each neighbour and @xmath73 is the distance to the @xmath0-th nearest neighbour . this weighting corresponds to the spline kernel of @xcite and is the standard kernel of sph . this method was used with 20 neighbours in @xcite , but values of 32 and 64 are more common in sph . another way to constrain local galaxy density using neighbours was proposed by @xcite for the redshift one ldss-3 emission line survey ( roles ; * ? ? ? @xcite considered the volume element of the nearest neighbour found by constructing a three dimensional cylinder using the five nearest neighbours to define its radius and depth . in other words , this technique encloses the five nearest neighbours in a cylinder that no longer has to be centred on the galaxy being sampled , and leads to a better estimate of the relevant volume when compared with simply using a sphere of radius the fifth nearest neighbour . further consideration of the volume can be made by calculating the voronoi volumes around each galaxy as a measure of the environment ( e.g. * ? ? ? * ; * ? ? ? voronoi volumes are polyhedrons constructed by bisecting the distance vectors to the nearest neighbours . each galaxy will have a volume around it , for which it does not have to be at the centre , defining the points in space that are closer to it than any other galaxy . this gives an estimate of the local density . unlike the other neighbour - based methods , the number of neighbours used to define the shape of the volume probed is not fixed , which makes the technique fully adaptive . for this study a projected voronoi measurement is made by collapsing galaxies into two dimensional slices of @xmath74 in depth . the voronoi shapes are then constructed on these surfaces to calculate the surface density of each galaxy . [ ] [ ] [ 1][0]@xmath11 [ ] [ ] [ 1][0]percentage rank in section [ sec : res ] we apply a number of the above nearest neighbour methods to the mock galaxy catalogue described in section [ sec : mod ] and quantify their relative strengths , weaknesses and optimal applications . in contrast to nearest neighbour methods , which define environment using a varying scale around each galaxy set by the distance to a pre - determined number of galaxy neighbours , fixed aperture methods instead probe a fixed area or volume around each galaxy , within which the number of neighbours are counted . the more galaxies inside this area or volume , the denser the environment is assumed to be , and vice versa . fixed aperture measures are often expressed as a density contrast , @xmath75 , instead of a density , @xmath76 . density contrast rescales the aperture count with respect to the mean and is typically defined as @xmath77 where @xmath78 is the number of galaxies found in the aperture , and @xmath79 is the mean number of galaxies that would be expected in the aperture if galaxies were instead distributed randomly throughout the entire volume . the fixed aperture technique was used by @xcite to investigate the environments around galaxies in the 2df galaxy redshift survey @xcite . @xcite used spherical apertures of radius @xmath50 , having investigated a range of sizes from of @xmath80 to @xmath81 ( see also * ? ? ? when distance information is not of sufficient accuracy ( or absent ) , apertures in this methodology are instead projected on to the sky . where possible , authors will then impose a velocity cut of order @xmath621000 km @xmath63 to minimise interlopers ( e.g. * ? ? ? * ) , for the same reasons discussed in section [ sec : neigh ] . the magnitude of this velocity cut can vary depending on distance uncertainties . this was investigated by @xcite who found velocity cuts of @xmath626000 km @xmath63 ( @xmath82 ) represent the typical photometric redshift uncertainty and @xmath62500 km @xmath63 ( @xmath83 ) represent the typical spectroscopic redshift uncertainty . such errors can often have a detrimental effect on the measured density if not appropriately accounted for . note that when a velocity cut is imposed , an otherwise spherical aperture elongates into a cylinder in three dimensional space , within which galaxy counts are then taken . whether this distortion is important for the environment measure depends on the focus of the analysis . typical scales for the radius of an aperture range from @xmath42 to @xmath84 , probing environments spanning individual haloes to large super - structures and voids in the cosmic web . a variation on the fixed aperture method was proposed in @xcite , where counts were taken in annuli of increasing inner and outer radius , rather than within a single fixed aperture volume . this technique enables the larger scale environment to be probed and the influence of local regions around individual galaxies to be removed . in its optimal form different sized annuli are applied in combination with apertures to better constrain the halo size and changes of environment with distance from the galaxy . finally , in addition to environment being defined by galaxy positions within the volume , we also measure environment as inferred from the background dark matter distribution . to obtain the neighbourhood dark matter environment in the millennium simulation the full volume is broken into a three dimensional grid with side - length @xmath45 . at the centre of each grid element a three dimensional gaussian density is calculated using the local dark matter particles , smoothed over three different scales : @xmath85 , @xmath86 , and @xmath84 . this gaussian smoothed density is similar to the kernel smoothed density of equation [ eq : sphdens ] , but with a dark matter particle mass term in the sum . in section [ sec : res ] we apply a number of fixed aperture methods to the mock galaxy catalogue and measure local density around each galaxy . this allows us to quantify the properties that aperture measured densities best probe , and compare with the previously described nearest neighbour estimators . to investigate the different properties of each galaxy environment measure , in this section we consider how they correlate with ( 1 ) the host dark matter halo mass , ( 2 ) the underlying dark matter environment , and ( 3 ) the colour of the galaxies . to facilitate this we have converted the output of each to a ` percentage rank ' for each galaxy . this is computed by listing the galaxies in order of increasing density , then assigning them a percentage based on where they appear in that list , with zero percent being the least dense and one hundred percent the most dense . therefore , a galaxy with a percentage rank of ninety five has five percent of the galaxies in the sample denser than it and ninety five percent less dense than it . this normalisation provides a fairer comparison between environment estimators and probes their relative rather than absolute distributions across the environment spectrum , which would otherwise be definition dependent . by design , the most fundamental property for a galaxy within our model is its dark matter halo mass . halo mass determines both the spatial distribution of the galaxy population and the individual galaxy properties . therefore , each environment measure should reveal some underlying correlation . typically halo masses of @xmath87 correspond to the field , @xmath88 to groups and @xmath89 to clusters . [ ] [ ] [ 1][0]@xmath11 [ ] [ ] [ 1][0]percentage rank , this time for the voronoi method . contours are linearly spaced showing regions of constant galaxy number . vertical lines represent typical dark matter halo masses that host 1 , 10 and 100 galaxies with @xmath13 ( see figure [ fig : pop]),title="fig:",width=309 ] [ ] [ ] [ 1][0]@xmath11 [ ] [ ] [ 1][0]percentage rank figure [ fig : nhalo ] shows contours of the abundance of galaxies that have environments of a given percentage rank plotted against the host halo mass , for four different nearest neighbour - based techniques , with the number of neighbours increasing from left to right . these are : the 3rd nearest neighbour density in three dimensions , the surface density for the projected 7th nearest neighbour , the three dimensional density using a 10 neighbour bayesian metric , and the smooth kernel three dimensional density using 64 neighbours . the most noticeable feature of all panels in figure [ fig : nhalo ] is that galaxies divide into two distinct groups , with the top @xmath90 percent dense environments occupied by galaxies in haloes more massive than @xmath91 , and the remaining @xmath92 percent of environments occupied by galaxies in haloes with masses lower than @xmath91 . this bi - modality arises from the assumed association between galaxies and dark matter haloes required to fit the observed luminosity function and clustering observations , and is explored further below . looking in more detail , the lower @xmath93 percent of rank - ordered densities in figure [ fig : nhalo ] shows no trend with halo mass , and as such , the term ` local environment ' no longer applies . in terms of a characteristic halo mass for a given environment , this result leaves individual galaxies near clusters indistinguishable from isolated galaxies in voids . in contrast , the behaviour of the high density halo mass correlation depends on the neighbour method employed . in the highest @xmath94 percent environments , low @xmath0 neighbour searches smooth away any density dependence with halo mass . this can be seen by comparing the far left panel in figure [ fig : nhalo ] ( low @xmath0 ) with the far right panel ( high @xmath0 ) . as the number of neighbours used to define environment is increased , galaxies belonging to increasingly massive haloes ( which host an increasing number of satellites ) will be labelled as increasingly dense . thus , to more precisely draw out the high density halo mass environment correlations using nearest neighbour methods , a high @xmath0 is desirable . the first two panels of figure [ fig : nhalo ] provide an additional test of the importance of projection effects . here , the 3rd nearest neighbour count is performed using three dimensional redshift space distances while the 7th nearest neighbour is performed with projected galaxy positions on the two dimensional sky . both methods show the same overall trend with halo mass . we find that , in general , projecting the galaxy positions simply blurs the edges of the two clouds with the overall shape preserved . another popular neighbour - based method used for measuring environment is voronoi volumes , as discussed in section [ sec : neigh ] . figure [ fig : vor ] shows how a voronoi defined environment estimator also correlates with dark matter halo mass . we see a similar trend to that of the other neighbour - based methods , with the overall result close to the 7th nearest projected neighbour method shown in the second panel of figure [ fig : nhalo ] . a comparison of figure [ fig : pop ] with figures [ fig : nhalo ] and [ fig : vor ] reveals the origin of the bi - modality . galaxies identified to be in the upper @xmath94 percent dense environments tend to be those whose neighbour search stays within the dark matter halo due to a large satellite population . such haloes are almost always more massive than @xmath91 . in contrast , the lower @xmath93 percent density environments are identified by neighbour searches that extend beyond the halo due to a low or zero satellite population of significance . in general , haloes with few satellites almost always have masses smaller than @xmath91 , and neighbour searches will then tend to probe the inter - halo rather than inter - galaxy separations . many authors have employed fixed apertures to probe the local density around galaxies , as described in section [ sec : aperture ] . in a similar vein to figure [ fig : nhalo ] , figure [ fig : ahalo ] shows how various aperture sizes correlate with host dark matter halo mass when a projected fixed aperture is employed with a cut in velocity space around each galaxy . in addition , the central two panels show how the density halo mass correlation changes if the velocity cut is increased for the same sized aperture . this roughly corresponds to the difference one would expect with data having photometric vs. spectroscopic redshifts , as discussed in @xcite . the projected fixed aperture technique yields both similar and different trends when compared with the nearest neighbour technique shown in figure [ fig : nhalo ] . the overall shape is the same , with galaxies in haloes of mass less than @xmath91 showing little correlation of halo mass with environment . at the high mass end there is a plume of increasing density that is much better defined than found with the nearest neighbour method ( especially when compared to choices of low @xmath0 ) . this suggests that the fixed aperture methodology is a better probe of halo mass , especially for small apertures and velocity cuts . there is however contamination at a fixed density from low mass haloes due to their close proximity to the high mass halo . in particular , when there are enough galaxies to define the local large - scale structure , a fixed - scale environment probe is much more sensitive to the power - law nature of the two - point correlation function , where the abundance of close pairs falls off rapidly beyond the halo radius . this leads to the galaxy count in the fixed aperture also falling off rapidly . in contrast , nearest neighbour environment methods adapt the scale probed to keep signal - to - noise fixed . hence , the division between a halo s interior and exterior becomes much less prominent . at intermediate to low masses there is no relation between fixed aperture measured density and halo mass , and so the environment parameter breaks down , as is also the case for nearest neighbour environment parameters . from an environment point - of - view , such haloes , which usually host galaxy groups , may be difficult to distinguish from cluster outskirts and from unassociated lower - mass haloes . as the aperture is increased in size , the trend with halo mass fades when the aperture becomes much larger than the structures present . for example , a super - cluster with a collective mass of @xmath95 would have a radius of @xmath96 , smaller than the @xmath97 aperture shown in the far right panel of figure [ fig : ahalo ] . when an aperture becomes large enough the contribution of individual haloes and structures blur and the environment halo mass trend weakens or disappears . hence , aperture size should be chosen carefully from the outset and be appropriate for the science questions of interest . finally , the two central panels of figure [ fig : ahalo ] illustrate the importance of velocity ( or equivalently distance ) uncertainties on the environment measure . large velocity cuts , as is typically required with photometric data , make measuring environment with a fixed aperture ineffective . this occurs for the same reason as using large apertures . there , the aperture was wider than the structures of interest which smoothed out the signal , while here , the depth of the aperture scatters in superfluous counts from foreground and background objects , diluting any correlation . this does not apply to the highest mass clusters as they dominate the depth reducing the effect of interlopers . furthermore , any use of the angular correlation function as a probe of environment must first consider the redshift distribution of the galaxies and the uncertainties must be well understood ( e.g. * ? ? ? * ; * ? ? ? galaxy colour has been shown to correlate with local galaxy density , with galaxies in over - dense environments being redder compared with those in under - dense environments ( cf . cluster and field ) ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? the model we employ in this paper has a constrained global @xmath20 colour distribution that mimics that of local galaxies in the sdss @xcite . hence , the degree to which different environment metrics can recover this relation can be tested . figure [ fig : ncol ] shows histograms of the @xmath20 colour distribution for the 20 percent most dense and 20 percent least dense galaxies defined with the same four nearest neighbour methods used in figure [ fig : nhalo ] : the 3rd nearest neighbour density in three dimensions , the projected 7th nearest neighbour , density defined from a 10 neighbour bayesian metric , and the smooth kernel density using 64 neighbours . in the @xmath94 percent most dense environments , all nearest neighbour - based environment measures show a clear red peak and a more weakly populated blue cloud . in contrast , in the lowest @xmath94 percent of environments galaxies are split more evenly between the red and blue populations . as the neighbour number is increased ( from left to right ) , there are only small changes in the relative colour distributions in environment extremes . the kolmogorov - smirnov probability that both samples are drawn from the same distribution is shown in the upper right of each panel . [ ] [ ] [ 1][0]@xmath20 [ ] [ ] [ 1][0]fraction [ ] [ ] [ 1][0]@xmath20 [ ] [ ] [ 1][0]fraction [ ] [ ] [ 1][0]dark matter percentage rank [ ] [ ] [ 1][0]@xmath98 galaxy percentage rank figure [ fig : acol ] shows histograms of colour for the 20 percent most dense and 20 percent least dense galaxies as probed by fixed apertures of various size , as used previously in figure [ fig : ahalo ] . the central two panels show how these distributions change if the velocity cut is increased or decreased for the same sized aperture . this roughly corresponds to the difference between photometric and spectroscopic redshift uncertainties @xcite ( see section [ sec : aperture ] ) . for small apertures , the colour distributions of both density extremes look remarkably similar to that found for the nearest neighbour methods shown in figure [ fig : ncol ] . however , as the volume of the fixed aperture is increased similar trends to that found in the previous section emerge . in particular , as the aperture becomes larger ( either in radius or depth ) , the differences between the colour distributions of galaxies in environment extremes lessen . here , the individual properties of galaxies are smoothed over due to the large variety of local environments falling within the aperture . for apertures probing scales much larger than the typical cluster the distinction between environments vanishes . this suggests that environment questions relating to galaxy colour ( or properties that correlate with colour ) should avoid fixed aperture methods with large smoothing radii or depth ( e.g. * ? ? ? furthermore , the kolmogorov - smirnov probabilities indicate that nearest - neighbour based methods detect stronger colour - environment relations than all the apertures tested here . dark matter haloes are known to be biased tracers of the underlying dark matter distribution , and it is interesting to compare how haloes and the smooth background mass field correlate with respect to their environment ranking , and how this relates to the galaxy distribution . to this end , the simulation volume has been divided using a three dimensional grid of side - length @xmath45 , and the neighbourhood dark matter density field measured with a gaussian filter placed at the centre of each grid element , smoothed on three different scales : @xmath85 , @xmath99 and @xmath84 ( see section [ sec : aperture ] ) . we compare this to the environment measured directly from central galaxy counts within a fixed spherical aperture of radius @xmath50 @xcite . figure [ fig : dme ] shows how the background dark matter density , gaussian smoothed on various scales , correlates with the large - scale galaxy density , top - hat smoothed on an @xmath50 scale . the correlation is weakest for the smallest gaussian smoothing scale of @xmath100 , becomes tighter at a scale of @xmath49 , before becoming weaker again at @xmath84 . the point of tightest correlation between dark matter and galaxy measured density approximately corresponds to the same physical scale being probed by each in three dimensional space . at fixed dark matter density the scatter in density measured by galaxies is approximately @xmath101 percent . this indicates the degree of precision with which one can probe the smooth background density using galaxies as tracers of the mass distribution . we have compared the other environment measures used in this paper to the background dark matter density but omit the figures for brevity . in short , a similar trend to figure [ fig : dme ] is found for the 64 neighbour smooth density environment measure , but with the tightest correlation at a radius of @xmath100 . for the other neighbour and small aperture methods , weak correlations are found when plotted against a dark matter density smoothing scale of @xmath100 but which disappear on larger scales . environments measured in annuli and projected aperture methods that impose a photometric - type redshift velocity cut show no correlation on any scale due to only the largest clusters dominating the depth cut , while the 10 neighbour bayesian metric and 20 neighbour smooth density again show a similar correlations to the dark matter smoothing scale of @xmath100 . ) for a single central galaxy living in the fourth most massive halo in the simulation . the density methods are grouped by increasing neighbour number , increasing aperture and increasing inner radius of an annulus . ( bottom ) the same as the top panel , but this time the percentage density ranking of the outer most satellite galaxy in the same halo for each method.,width=309 ] in the previous sections we investigated how different environment parameters correlate with different galaxy properties in a statistical sense by considering the whole sample . as implied by figures [ fig : nhalo ] and [ fig : ahalo ] , when selecting the most and least dense environments different methods will potentially select different galaxy populations . an alternative and complementary way to compare the different environment methodologies is to consider individual galaxies in the mock catalogue and examine how each measure ranks them relative to the others . by considering individual galaxies a better understanding of why these galaxies were chosen can be obtained . this also highlights the consistency ( or lack thereof ) between different definitions of environment . below we present one example that is representative of the general behaviour for high mass haloes . the top panel of figure [ fig : comp ] shows how the different environment measures listed in table [ tab : comp ] compare when one focuses on the central galaxy occupying the fourth most massive halo in the simulation , with mass @xmath102 . the environment measures are separated into three groups based on the technique they use : neighbours , aperture and annulus . all environment measures place this galaxy within the top 10 percent of rank ordered densities in the simulation volume , with the majority placing it within the top 1 percent . when considering annuli to define environment , the top panel of figure [ fig : comp ] shows that the further one moves from the centre of the halo the lower the rank density measured . this simply highlights that the outer regions of a halo tend to be less dense than the core . when considering aperture methods there is less of a trend between different definitions . however , for a fixed depth , increasing the aperture size reduces the rank density measured , while for fixed aperture size , the density rank appears sensitive to the inclusion of both the halo core ( smaller velocity cut ) and full extent ( larger velocity cut ) . as mentioned in previous sections , the larger velocity cut used to represent photometric redshift uncertainties has a smaller effect on large clusters as the cluster members dominate the galaxies within the depth cut . for neighbour - based methods there is a general increase in the rank density as the neighbour number increases . this is due to the increased neighbour count contributing from within the galaxy halo . specifically , as the number of neighbours increases galaxies in smaller haloes are demoted down the rank list , and so the galaxies in large haloes are promoted . the bottom panel of figure [ fig : comp ] shows how the different density estimators rank the most distant satellite associated with the central galaxy of the same @xmath103 halo used in the top panel . this is a test of how satellites on the outskirts of cluster environments would be classified in each density scheme . the range of environment ranking is much larger between the different measures , with the apertures and annuli mostly finding higher rank densities than the neighbour estimates . this again comes back to scale , with neighbour methods probing the internal properties of the outer halo , and aperture and annulus methods being sensitive to the larger structure of the halo and its surrounds . additionally , the trend of increasing rank density with increasing neighbour number is again seen as the neighbour count reaches deeper into the halo core from the boundary . the phrase `` galaxy environment '' is a very general concept that has been used in the literature in a variety of ways . its definition what it measures and how it is measured can vary from author to author . this creates uncertainty when trying to compare results for environmental trends . in practise , galaxy environment is quantified in one of two ways : by the distance to the @xmath0-th nearest neighbour or by using a fixed aperture to probe the surrounds . over the course of time these two methods have evolved in the literature . however , both methods and their variants provide a measure of the density field surrounding a galaxy and hence can be used to answer specific environment - related questions . to fairly compare many different environment measures one would ideally like to use a common galaxy catalogue as a starting point . this was achieved in the present work by applying a halo occupation distribution model to the @xmath15 output of the millennium dark matter simulation . our model is designed to accurately reproduce the luminosity , colour and spatial distribution of galaxies in the sloan digital sky survey . the resulting data cube was also used to generate a mock light cone so that the environment measures could be applied in a more realistic geometry . the major conclusions of this paper are as follows : * * galaxy environment versus halo mass * : comparing neighbour and aperture based environment measures to the dark matter halo mass of a galaxy reveals how _ they measure different aspects of the halo_. in particular , nearest neighbour methods that use a small enough neighbour number best probe the internal properties of the halo . for haloes that contain fewer galaxies than the neighbour number , the inter - halo separation dominates the calculation and galaxy environment correlations tend to wash out . in contrast , _ aperture measures tend to better probe the halo as a whole _ and so lead to larger density values corresponding to larger haloes , which more accurately reflect their larger masses . a smaller aperture than those studied here could be used to probe cluster environments on a scale similar to the nearest neighbour - based methods , but these would be unsuitable for the field due to the distance between neighbours being too large . this is in agreement with the findings of @xcite . * * galaxy environment within haloes * : _ galaxy density internal to a halo s boundary is found to be independent of its mass when probed using the neighbour method_. while galaxies at the edge of a halo are always in less dense environments than those at the centre , the galaxy environment at the centre of intermediate mass haloes is approximately the same as that at the centre of very massive ones . by fitting the number of galaxies for a given halo mass , we find that the number of galaxies per unit dark matter mass is constant and this is in agreement with the findings of @xcite . * * environmental dependence of galaxy colour * : when comparing how the different environment measures distribute galaxy colour , almost all methods recover the observed correlation that galaxies are redder in denser environments compared to those in less dense environments . _ this relation only disappears for very large apertures , of order @xmath104_. on scales larger than this the most dense and least dense galaxies are found to have similar colour distributions . here , the aperture is large enough to encompass a statistically representative number of different haloes , resulting in a smoothing out of the colour differences over such large volumes . _ this behaviour is also expected to extend to any property that correlates strongly with colour_. * * galaxy environment versus dark matter environment * : on the other hand , _ very large fixed apertures are the most accurate at recovering the large - scale dark matter environment_. for example , an @xmath50 spherical aperture used to calculate the galaxy density correlates well with the dark matter environment measured using gaussian smoothing on @xmath49 scales . similar results are found with high number @xmath0-th neighbour estimates . the important parameter here is scale , with larger probed scales better correlated than smaller scales . * * environment measures using spectroscopic versus photometric redshifts * : in addition to the environment measures themselves , we also explored the general effects of photometric and spectroscopic redshift uncertainties by varying the velocity cut used to calculate projected environment . for a typical photometric redshift uncertainty most trends with environment disappear or become significantly weaker . this is caused by the depth of the aperture becoming much larger than the objects being probed , and the scattering of interlopers which contaminate the density probe . this effect decreases for the largest clusters as the members dominate the depth cut . we warn that _ photometric redshifts may be unsuitable for measuring certain properties when using a range of environmental scales _ , especially at high redshifts . * * environments of central and satellite galaxies * : on a galaxy - by - galaxy basis , most environment methods agree on the relative environment rank of central galaxies in massive haloes ( to within a few percent ) . _ there is less agreement with the satellite population in clusters _ , where the result becomes more sensitive to the method employed . dark matter haloes are often broadly categorised as residing in ` field ' , ` group ' , and ` cluster ' environments based solely on their mass . in figure [ fig : pop ] , for example , haloes of mass @xmath105 , @xmath106 , and @xmath107 approximately correspond to these environment bins , respectively . many environment analyses use this categorisation , although as we have seen , the distinctions between them can often be blurred in detail . some studies also attempt to explicitly identify galaxies that are isolated or in groups or rich clusters , for example using friends - of - friends group - finding algorithms ( e.g. * ? ? ? * ; * ? ? ? analyses using group catalogues are complementary to studies with nearest neighbour or fixed aperture measures , or with galaxy clustering ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? work focused on galaxy clusters have also yielded complementary results ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? importantly , the way a galaxy forms and evolves is clearly related to its environment . some galaxy properties , such as luminosity , colour , and stellar mass , are directly correlated with the large - scale environment through the host dark matter halo ( e.g. * ? ? ? * ; * ? ? ? other galaxy properties , such as structure , are to some extent only indirectly correlated with the environment ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? indeed , a number of authors report that , for many aspects of the galaxy population , environmental dependence is often weak once stellar mass is fixed @xcite . in any case , these studies highlight the fact that it is important to carefully determine how a galaxy s environment is characterised , and to identify and navigate the potential aspects of the environment analysis that may bias the results . the key consideration when picking an environment measure is the scale that is being probed . the term environment is very general but in fact breaks down into two main regions , and we argue that the community should agree on a standard terminology for clarity and to avoid future confusion . the first region is the ` local environment ' which corresponds to scales internal to a halo . these are best probed using nearest neighbour methods , but the value of @xmath0 is important . when @xmath0 is larger than the number of galaxies likely to reside within the halo the usefulness of this environment measure can weaken . the second region lies external to the halo , the ` large - scale environment ' . the large - scale environment is best probed using aperture based methods . in general , there is no simple way to probe all environments with a single method , and one should consider carefully the best tool to answer the questions at hand . this paper marks the first in a series exploring the meaning and methods of galaxy environment , as measured in the current literature . in the present work we have focused on using a clean sample of mock galaxies to quantify how selected properties of the galaxy population correlate with different environment methods , and how these methods themselves compare . future work will include investigating the detrimental effects of survey geometry , edges and holes ( such as those caused by stars ) on environment and techniques that can be applied to successfully overcome them . furthermore , the relationship between galaxy , halo and dark matter environment warrants additional exploration , as does the redshift dependence of a galaxy s environment ( defined in various ways ) , what the different environment methods tell us about galaxy evolution , and how these can best be applied to the noisy data of the high redshift universe . the authors wish to thank hanni lux and idit zehavi for useful discussions and the referee , christian wolf , for improving the quality of the paper . djc acknowledges receipt of a qeii fellowship by the australian research council . hba thanks nrf for the research grant 2010 - 0023319 . nbc is supported by a ciera postdoctoral fellowship . meg is supported by an stfc advanced fellowship . the dark cosmology centre is funded by the danish national research foundation . the millennium simulation used in this paper was carried out by the virgo supercomputing consortium at the computing centre of the max - planck society in garching . the halo merger trees used in the paper are publicly available through the gavo interface , found at http://www.mpa - garching.mpg.de / millennium/. the mock galaxy catalogue generated from the millennium simulation trees can be downloaded from the theoretical astrophysical observatory ( tao ) repository : http://tao.it.swin.edu.au / mock - galaxy - factory/. 199 abbas u. , sheth r.k . , 2006 , mnras , 372 , 1749 baldry i.k . , balogh m.l . , bower r.g . , glazebrook k. , nichol r.c . , bamford s.p . , budavari t. , 2006 , mnras , 373 , 469 ball n.m . , loveday j. , brunner r.j . , 2008 , mnras , 383 , 907 balogh m.l . et al . , 2009 , mnras , 398 , 754 bamford s.p . et al . , 2009 , mnras , 393 , 1324 berlind a.a . et al . , 2006 , apjs , 167 , 1 blanton m.r . et al . , 2003 , apj , 592 , 819 blanton m.r . , eisenstein d. , hogg d.w . , schlegel d.j . , brinkmann j. , 2005 , apj , 629 , 143 blanton m.r . , berlind a.a . , 2007 , apj , 664 , 791 bolzonella m. et al . , 2010 , a&a , 524 , 76 brough s. et al . , 2011 , mnras , 413 , 1236 cassata p. et al . , 2007 , apjs , 172 , 270 coil a.l . , newman j.a . , kaiser n. , davis m. , ma c .- p . , kocevski d.d . , koo d.c . , 2004 , apj , 617 , 765 colless m. et al . , 2001 , mnras , 328 , 1039 colless m. et al . , 2003 , astro - ph/0306581 cooper m.c . , newman j.a . , madgwick d.s . , gerke b.f . , yan r. , davis m. , 2005 , apj , 634 , 833 cooper m.c . et al . , 2006 , mnras , 370 , 198 cowan n.b . , ivezi . , 2008 , apj , 674 , l13 croton d.j . et al . , 2005 , mnras , 356 , 1155 de la torre s. et al . , 2011 , mnras , 412 , 825 de lucia g. , blaizot j. , 2007 , mnras , 375 , 2 driver s.p . et al . , 2011 , mnras , 413 , 971 elbaz d. et al . , 2007 , a&a , 468 , 33 ellison s.l . , simard l. , cowan n.b . , baldry i.k . , patton d.r . , mcconnachie a.w . , 2009 , mnras , 396 , 1257 gallazzi a. et al . , 2009 , apj , 690 , 1883 gilbank d.g . et al . , 2010 , mnras , 405 , 2419 grtzbauch r. , conselice c.j . , varela j. , bundy k. , cooper m.c . , skibba r. , willmer c.n.a . , 2011 , mnras , 411 , 929 haas m.r . , schaye j. , jeeson - daniel a. , 2012 , mnras , 419 , 2133 hansen s.m . , sheldon e.s . , wechsler r.h . , koester b.p . , 2009 , apj , 699 , 1333 kauffmann g. , white s.d.m . , heckman t.m . , mnard b. , brinchmann j. , charlot s. , tremonti c. , brinkmann j. , 2004 , mnras , 353 , 713 knebe a. et al . , 2011 , mnras , 415 , 2293 kova k. et al . , 2010 , apj , 708 , 505 kravtsov a.v . , berlind a.a . , wechsler r.h . , klypin a.a . , gottlber s. , allgood b. , primack j.r . , 2004 , apj , 609 , 35 lewis i. , 2002 , mnras , 334 , 673 li c. , kauffmann g. , jing y.p . , white s.d.m . , brner g. , cheng f.z . , 2006 , mnras , 368 , 21 li i.h . et al . , 2011 , mnras , 411 , 1869 macci a.v . , dutton a.a . , van den bosch f.c . , moore b. , potter d. , stadel j. , 2007 , mnras , 378 , 55 macci a.v . , dutton a.a . , van den bosch f.c . , 2008 , mnras , 391 , 1940 marinoni c. , davis m. , newman j.a . , coil a.l . , 2002 , apj , 580 , 122 martnez h.j . , muriel h. , 2006 , mnras , 370 , 1003 monaghan j.j . , lattanzio j.c . , 1985 , a&a , 149 , 135 navarro j.f . , frenk c.s . , white s.d.m . , 1996 , apj , 462 , 563 norberg p. et al . , 2002 , mnras , 332 , 827 omill a.l . , padilla n. , garca lambas d. , 2008 , mnras , 389 , 1763 park c. , choi y .- y . , vogeley m.s . , gott iii j.r . , blanton m.r . , 2007 , apj , 658 , 898 poggianti b.m . et al . , 2008 , apj , 684 , 888 poggianti b.m . , de lucia g. , varela j. , aragn - salamanca a. , finn r. , desai v. , von der linden a. , white s.d.m . , 2010 , mnras , 405 , 995 quadri r.f . , williams r.j . , lee k .- s . , franx m. , van dokkum p. , brammer g.b . , 2008 , apj , 685 , l1 rudnick g. et al . , 2009 , apj , 700 , 1559 scodeggio m. et al . , 2009 , a&a , 501 , 21 sheth r.k . , jimenez r. , panter b. , heavens a.f . , 2006 , apj , 650 , l25 skibba r. , sheth r.k . , connolly a.j . , scranton r. , 2006 , mnras , 369 , 68 ( s06 ) skibba r.a . , 2009 , mnras , 392 , 1467 skibba r.a . , sheth r.k . , 2009 , mnras , 392 , 1080 ( ss09 ) skibba r.a . et al . , 2009 , mnras , 399 , 966 skibba r.a . , van den bosch f.c . , yang x. , more s. , mo h. , fontanot f. , 2011 , mnras , 410 , 417 spergel d.n . et al . , 2003 , apjs , 148 , 175 springel v. , white s.d.m . , tormen g. , kauffmann g. , 2001 , mnras , 328 , 726 springel v. , 2005 , mnras , 364 , 1105 springel v. et al . , 2005 , nat , 435 , 629 tasca l.a.m . et al . , 2009 , a&a , 503 , 379 tinker j.l . , conroy c. , norberg p. , patiri s.g . , weinberg d.h . , warren m.s . , 2008 , apj , 686 , 53 van den bosch f.c . , aquino d. , yang x. , mo h.j . , pasquali a. , mcintosh d.h . , weinmann s.m . , kang x. , 2008a , mnras , 387 , 79 van den bosch f.c . , pasquali a. , yang x. , mo h.j . , weinmann s. , mcintosh d.h . , aquino d. , 2008b , mnras , submitted , preprint ( astro - ph/0805.0002 ) vulcani b. et al . , 2011 , mnras , 412 , 246 weinmann s.m . , van den bosch f.c . , yang x. , mo h.j . , 2006 , mnras , 366 , 2 white s.d.m . , rees m.j . , 1978 , mnras , 183 , 341 wilman d.j . , zibetti s. , budavri t. , 2010 , mnras , 406 , 1701 wolf c. et al . , 2009 , mnras , 393 , 1302 yang x. , mo h.j . , van den bosch f.c . , 2003 , mnras , 339 , 1057 yang x. , mo h.j . , van den bosch f.c . , weinmann s.m . , li c. , jing y.p . , 2005 , mnras , 362 , 711 yang x. , mo h.j . , van den bosch f.c . , pasquali a. , li c. , barden m. , 2007 , apj , 671 , 153 yang x. , mo h.j . , van den bosch f.c . , 2009 , apj , 695 , 900 york d.g . et al . , 2000 , aj , 120 , 1579 zehavi i. et al . , 2005 , apj , 630 , 1 zehavi i. et al . , 2011 , apj , 736 , 59 zheng z. , coil a.l . , zehavi i. , 2007 , apj , 667 , 760	the influence of a galaxy s environment on its evolution has been studied and compared extensively in the literature , although differing techniques are often used to define environment . most methods fall into two broad groups : those that use nearest neighbours to probe the underlying density field and those that use fixed apertures . the differences between the two inhibit a clean comparison between analyses and leave open the possibility that , even with the same data , different properties are actually being measured . in this work we apply twenty published environment definitions to a common mock galaxy catalogue constrained to look like the local universe . we find that nearest neighbour - based measures best probe the internal densities of high - mass haloes , while at low masses the inter - halo separation dominates and acts to smooth out local density variations . the resulting correlation also shows that nearest neighbour galaxy environment is largely independent of dark matter halo mass . conversely , aperture - based methods that probe super - halo scales accurately identify high - density regions corresponding to high mass haloes . both methods show how galaxies in dense environments tend to be redder , with the exception of the largest apertures , but these are the strongest at recovering the background dark matter environment . we also warn against using photometric redshifts to define environment in all but the densest regions . when considering environment there are two regimes : the ` local environment ' internal to a halo best measured with nearest neighbour and ` large - scale environment ' external to a halo best measured with apertures . this leads to the conclusion that there is no universal environment measure and the most suitable method depends on the scale being probed . [ firstpage ] methods : numerical methods : statistical galaxies : evolution galaxies : haloes dark matter large - scale structure of universe
a relativistic charge crossing a dielectric interface in a rectilinear and uniform motion can originate a highly directional and broad wavelength band radiation emission propagating backward and forward from the boundary surface within a small angle scaling down with the energy of the charge , the so called transition radiation @xcite . thanks to the instantaneous , highly directional and charge - energy dependent features , transition radiation is a precious tool in the beam diagnostics of a particle accelerator . in the vacuum chamber of a particle accelerator the relativistic charged beam can be indeed intercepted by a movable thin metallic screen and the resulting light pulse can be imaged by a ccd - camera ( charge - coupled - device ) , for instance , to measure the transverse profile of the charged beam . under the key - word otr beam diagnostics ( optical transition radiation ) , the reader can find an enormous specialized literature dedicated to this topic . for most part of the practical detection conditions ( from the extreme visible to the thz region ) , the metallic screen can be assimilated to an ideal conductor surface and the transition radiation emission can be easily schematized as the result of the dipolar oscillation of the conduction electrons of the metallic surface induced by the electromagnetic field of the incident relativistic charge . the dipolar oscillation induced in the double layer of charge by the relativistic charge explains indeed how an electromagnetic radiative mechanism , originated by a charge in a rectilinear and uniform motion , can also propagate in the backward direction . such a model of the double layer of charge , describing the transition radiation emission as the result of the dipolar oscillation of the conduction electrons induced by the relativistic charge , is also enlightening about the kinematics of this radiative mechanism and about the common relativistic nature that it shares with other electromagnetic radiative mechanisms by relativistic charged beams . the kinematics of the transition radiative mechanism can be indeed schematized as the head - on collision of two distributions of charge as observed in the reference frame of rest of one of the two colliding charged distributions . the backward and forward double conical transition radiation emission can be thus interpreted as the photon bremsstrahlung emission that two head - on colliding electron beams can originate . taking into consideration that the transition radiation mechanism shares with other electromagnetic radiative mechanisms , such as the synchrotron or the bremsstrahlung radiation , the same kinematical and relativistic nature , it is thus reasonable to expect that , even at a very short wavelength , some spectral modifications of the radiation intensity due to transverse density of the beam should also affect the transition radiative mechanism by an electron beam in a similar way as , in other electromagnetic radiative mechanism by charged beams , the beam transverse size contributes to determine the so called brilliance or luminosity properties of the radiation source . in the case of an electron beam at a normal angle of incidence onto a metallic screen with arbitrary size and shape , it can be demonstrated that a covariant and temporal causal formulation of the transition radiation energy spectrum of an electron beam necessarily implies , even at a very short wavelength , a dependence of the radiation spectral intensity on the distribution function of the particle density in the transverse plane @xcite . in fact , in the transition radiation emission of an electron beam at a normal angle of incidence onto a metallic screen , the longitudinal and transverse coordinates of the electrons play different roles . the distribution function of the longitudinal coordinates of the @xmath0 electrons determines indeed the sequence of the particle collisions onto the metallic screen and , consequently , on the basis of the temporal causality principle , also the distribution function of the relative emission phases - from the metallic surface - of the @xmath0 single electron amplitudes composing the radiation field . the distribution function of the @xmath0 electron transverse coordinates contributes as well in determining the relative phase delay of the @xmath0 single electron field amplitudes at the observation point as a function of the transverse displacement of the @xmath0 electrons with respect to the beam axis where the radiation field is supposed to be observed . but , in addition , because of the relativistic invariance under a lorentz transformation in the direction of motion of the beam , the @xmath0 electron transverse density manifests itself as a covariant feature of the radiation field whose observability can transform but not disappear under a lorentz transformation . in the present work , the covariant and temporal causal formula of the transition radiation energy spectrum of a @xmath0 electron bunch , already derived in an implicit form in @xcite in the most general case of a radiator surface with an arbitrary size and shape , will be here rendered into an explicit form in the particular case of a round radiator with an arbitrary radius . the so obtained formula of the transition radiation energy spectrum of a @xmath0 electron bunch at a normal angle of incidence onto a round metallic screen is demonstrated to be temporal causal and covariant @xcite and to reproduce , as a limit , some results already well known in literature , such as : the frank - ginzburg formula of the single particle transition radiation energy spectrum in the ideal case of an infinite radiator or the single electron transition energy spectrum radiated by a round metallic screen with a finite radius . finally , because of the covariant dependence of the radiation field on the transverse density of the @xmath0 electrons , it results that the distribution function of the @xmath0 electron transverse coordinates affects both the temporal coherent and incoherent parts of the transition radiation energy spectrum . even at a very short wavelength , the effect of the @xmath0 electron transverse density on the spectral distribution of the radiation intensity manifests itself as an increase of the radiated energy with the decrease of the beam transverse size and a spectral narrowing of the angular distribution of the radiation intensity in comparison with the ideal case of an electron beam having a point - like transverse extension . in the following , in the case of a bunch of @xmath0 electrons hitting the metallic screen at a normal angle of incidence , details on the formula of the transition radiation energy spectrum and numerical results will be presented . a bunch of @xmath0 electrons in a rectilinear and uniform motion along the @xmath1-axis of the laboratory reference frame is colliding , at a normal angle of incidence , onto a flat ideal conductor surface @xmath2 placed in the plane @xmath3 . the @xmath0 electrons are supposed to fly in vacuum with a common velocity @xmath4 . all the electrons are supposed to hit the metallic screen at a normal angle of incidence . effects of the angular divergence of the electron beam on the radiation energy spectrum are not considered in the present work . the radiator surface @xmath2 has a round shape with a finite radius @xmath5 . the observation point of the radiation emission is on the @xmath1-axis at a distance @xmath6 from the screen much larger than the observed wavelength @xmath7 . the spatial coordinates of the @xmath0 electrons being @xmath8 $ ] with @xmath9 at a given reference time @xmath10 when the center of mass of the charged distribution is crossing the boundary surface , the spectral component of the radiation field resulting from the collision of the @xmath0 relativistic electrons onto the metallic screen reads , see @xcite , @xmath11 where , under the far - field approximation @xcite , the single electron contribution to the radiation field can be calculated in the most general case of a radiator surface @xmath2 with an arbitrary shape and size ( either infinite @xmath12 or finite @xmath13 ) as follows @xcite : @xmath14 in previous equation , @xmath15 is the wave number , @xmath16 is the transverse component of the wave - vector , @xmath17 ( @xmath18 being the relativistic lorentz factor ) and the vector @xmath19 represents the integration variable on the surface @xmath2 of the screen . with reference to eqs.([uno],[due ] ) , the transition radiation energy spectrum by a @xmath0 electron beam can be finally calculated as the flux of the poynting vector , see also @xcite : @xmath20 where @xmath21 with @xmath22 , as defined in eq.([due ] ) . the formal expression of the radiation field and of the radiation energy spectrum of the @xmath0 electron bunch , as represented in eqs.([uno],[due ] ) and eq.([tre ] ) in the most general case of a flat radiator surface @xmath2 with an arbitrary size and shape , meets the constraints of the temporal causality and of the covariance . the structure of the emission phases from the screen of the @xmath0 single electron field amplitudes composing the radiation field - see eqs.([uno],[due ] ) - is indeed causally related to the temporal sequence of the @xmath0 particle collision onto the metallic screen . furthermore , with reference to @xcite , covariant or covariance consistent are all the formal steps which , from the expression of the electromagnetic field of the @xmath0 electron bunch in the laboratory reference frame , leads to the radiation field - see eqs.([uno],[due ] ) - and to the radiation energy spectrum , see eq([tre ] ) . for further details about the covariant and the temporal causal formulation of the transition radiation emission from a @xmath0 electron bunch - as described in eqs.([uno],[due],[tre ] ) - the reader is addressed to @xcite . for the sake of ease , in order to proceed with the formal explicit derivation of the transition radiation energy spectrum in the case of a round radiator with a radius @xmath5 , a polar angle representation is considered for the following vectors @xmath23 where @xmath24 @xmath25 , @xmath26 and @xmath19 with @xmath27 being the vectors of the @xmath0 electron transverse coordinates and the vector of the screen coordinates , respectively . taking into account the following integral representation of the bessel function of the first kind @xcite ( see also eq.(@xmath28 ) in @xcite ) : @xmath29 in the limit of a round metallic surface with a radius @xmath5 much longer than the average transverse size of the beam @xmath30 the integral that , in eq.([due ] ) , determine the single electron field amplitude can be explicitly calculated : @xmath31\nonumber\end{aligned}\ ] ] with @xmath32 and @xmath33k_1[\alpha(r+\rho_{0j})]+\frac{\alpha^2}{\kappa } j_1[\kappa(r+\rho_{0j})]k_0[\alpha(r+\rho_{0j})]\label{sette}\end{aligned}\ ] ] where eqs.([cinque],[sei],[sette ] ) follows from eq.([quattro ] ) and from formulas @xmath34 , @xmath35 , @xmath36 and @xmath37 in @xcite . in conclusion , with reference to eqs.([cinque],[sei],[sette ] ) , the harmonic component of the radiation field produced by a @xmath0 electron bunch hitting a round radiator surface with a finite radius @xmath5 reads , see eqs.([uno],[due ] ) , @xmath38}\times\nonumber\\\label{settebis } & & \times\left(\begin{array}{c } \cos\phi\\ \sin\phi\\ \end{array}\right)[\rho_{0j}\phi(\kappa,\alpha,\rho_{0j})-(r+\rho_{0j})\phi(\kappa,\alpha , r+\rho_{0j})],\end{aligned}\ ] ] where @xmath39 is the transverse component of the wave - vector ( @xmath40 ) . in some special and relevant cases , the formula of the transition radiation field given in eqs.([sei],[sette],[settebis ] ) reproduces as a limit some results already well known in literature . single electron @xmath42 and ( @xmath43 , @xmath44 ) : i.e. , a single electron moving on the @xmath1-axis ( @xmath45 ) , where the observation point is also located , in collision onto a radiator having an infinite surface . in the limit @xmath46 and @xmath47 , the functions defined in eqs.([sei],[sette ] ) tend to the following limit values : @xmath48 taking into account the limits above , the following expression of the single electron radiation field , see eqs.([sei],[sette],[settebis ] ) , follows @xmath49 where @xmath50 . finally , from the equation above and eq.([tre ] ) , the well known result of the frank - ginzburg formula of the transition radiation energy spectrum in the ideal case of a single electron hitting an infinite ideal conductor surface can be obtained @xcite : @xmath51 case @xmath52 . single electron @xmath42 and ( @xmath43 , @xmath53 ) : similar experimental situation as in previous case but finite radius of the radiator . with reference to the above reported limit of the expression in eq.([sei ] ) for @xmath47 , the radiation field reads , see eqs.([uno],[due],[cinque],[sei],[sette ] ) , @xmath54\end{aligned}\ ] ] which , in the far - field approximation , represents the radiation field of a single electron hitting a finite round metallic screen . in the case of a single electron hitting a round metallic screen with a finite radius , as a function of the transverse extension ( @xmath55 ) of the single electron virtual quanta field compared to the finite radius @xmath5 of the radiator , the transition radiation energy spectrum experiences a low frequency diffractive cut - off . this problem was already tackled in several works @xcite . compare eq.([dieci ] ) in the present work with eqs.(@xmath56 ) in @xcite , for instance . finally , from eqs.([tre],[settebis ] ) , the explicit expression of the transition radiation energy spectrum of a @xmath0 electron bunch colliding onto a round ideal conductor surface with a finite radius @xmath5 can be obtained @xmath57}\right)\label{setteter}\end{aligned}\ ] ] where @xmath58 is the single electron radiation energy spectrum already defined in eq.([nove ] ) and @xmath59 the approximation in eq.([settequater ] ) being valid for most of the experimental situations where diffractive modifications of the spectral distribution of the radiated energy due to the finite size of the screen can be neglected . in the formula of the transition radiation energy spectrum of a @xmath0 electron beam , as explicitly derived in eqs.([setteter],[settequater ] ) in the case of a normal collision onto a round metallic screen with a finite radius @xmath5 , the double role played by the transverse coordinates of the @xmath0 electrons appears evident . on the one hand , as a function of the displacement of the electrons with respect to the beam axis where the radiation field is supposed to be observed , they contribute to determine the relative phase delay of the @xmath0 electron field amplitudes at the observation point as the presence of the well known three - dimensional phase factor in the temporal coherent part of the formula given in eq.([setteter ] ) clearly indicates . on the other hand , because of the invariance of the transverse density of the @xmath0 electrons under a lorentz transformation in the direction of motion of the beam , the @xmath0 single electron radiation field amplitudes show a covariant dependence on the transverse density of the @xmath0 electron beam . as a result , see eqs.([setteter],[settequater ] ) , both the temporal coherent and the temporal incoherent components of the transition radiation energy spectrum of a @xmath0 electron beam differ from the ideal scenario of a single electron hitting an infinite metallic screen . it results indeed that , even at a very short wavelength , the radiation spectral intensity increases with the decrease of the beam transverse size towards an asymptotic limit as well as the spectral angular distribution of the radiation experiences a broadening towards the asymptotic limit given by the well known ideal case of a single electron hitting an infinite metallic surface . in figs.([fig1],[fig2],[fig3],[fig4 ] ) , results of the numerical simulation of the temporal incoherent part of the transition radiation spectrum produced by a gaussian electron bunch with a transverse size much shorter than the finite radius of the radiator are shown , see eqs.([setteter],[settequater ] ) . in fig.([fig1 ] ) , for a beam energy of @xmath60 and a bunch of @xmath61 electrons , the angular distribution of the transition radiation intensity in the visible ( green , @xmath62 ) for different values of the beam standard deviation @xmath63 @xmath64 is calculated and compared with the ideal result obtainable from a bunch with the same charge but radiating according to the formula given in eq.([nove ] ) . in figs.([fig2],[fig3],[fig4 ] ) , for different beam energies @xmath65 @xmath66 and a constant beam charge of @xmath61 electrons , the angular distributions of the transition radiation intensity radiated at different wavelengths in the visible ( red , green , blue @xmath67 @xmath68 ) by a beam with standard deviation @xmath69 @xmath64 are compared with the analogous result obtained in the ideal case from the formula in eq.([nove ] ) . temporal causality and covariance constrain the role that the distributions of the longitudinal and transverse coordinates of an electron bunch plays in determining the physical features of an electromagnetic radiative mechanism , in particular , the transition radiation mechanism . in the case of a @xmath0 electron bunch hitting at a normal angle of incidence a metallic screen , the distribution of the electron longitudinal coordinates defines the causal correlation occurring between the temporal sequence of the electron collisions onto the metallic screen and the structure of the relative emission phases of the @xmath0 single electron radiation field amplitudes propagating from the metallic screen to the observation point , located on the beam longitudinal axis . the distribution of the transverse coordinates of the @xmath0 electrons contributes as well to determine the relative phase delay of the @xmath0 single electron radiation field amplitudes at the observation point as a function of the @xmath0 electron displacements with respect to the beam axis . the transverse density of the @xmath0 electrons , being an invariant under a lorentz transformation with respect to the direction of motion of the beam , constitutes for the radiation field a covariant feature that affects both the spectral intensity and the angular distributions of the transition radiation under observation conditions of both temporal incoherence and temporal coherence . compared to the reference asymptotic results provided by the well known ideal formalism of a single electron hitting an infinite metallic screen , an increase of the radiated spectral intensity with the decrease of the beam transverse size and a broadening of the spectral angular distribution constitute the relevant phenomenological aspects of the covariant beam - transverse - size effects on the transition radiation energy spectrum originated by an electron beam hitting a round metallic screen at a normal angle of incidence . 00 v. l. ginzburg and i. m. frank , soviet phys . jetp * 16 * ( 1946 ) 15 . garibian , sov . jetp , * 6 * ( 33 ) , 6 ( 1958 ) 1079 - 1085 garibyan , sov . jetp , * 37 * ( 10 ) , 2 ( 1960 ) 372 - 376 . frank , sov . , * 4 * , 5 , ( 1962 ) 740 - 746 . bass , v. m. yakovenko , sov . , * 8 * , 3 ( 1965 ) 420 - 444 . frank , sov . usp . , * 8 * , 5 ( 1966 ) 729 - 741 ter - mikaelian , _ high - energy electromagnetic processes in condensed media _ , wiley , new york ( 1972 ) . ginzburg , v.n . tsytovich , _ transition radiation and transition scattering _ , adam hilger , bristol ( 1990 ) . jackson , _ classical electrodynamics _ , wiley , new york ( 1975 ) . orlandi , _ on the covariance of the charge form factor in the transition radiation energy spectrum of a beam at normal incidence onto a metallic screen _ , paul scherrer institut preprint psi - pr-12 - 03 . orlandi , opt . commun . * 211 * ( 2002 ) 109 - 119 . orlandi , opt . commun . * 267 * ( 2006 ) 322 - 334 . orlandi , proc . fel 2005 , stanford university ( 2005 ) , jacow / econf c0508213 , 576 - 579 orlandi , proc . epac 08 , genoa , italy , ( 2008 ) 1221 - 1223 . orlandi , b. beutner , r. ischebeck , v. schlott , b. steffen , proc . fel 2009 , liverpool , uk , 153 ( 2009 ) . orlandi , r. ischebeck , v. schlott , b. steffen , proc . fel 2010 , malm , sweden , ( 2010 ) . m. born , e. wolf , _ principles of optics _ , pergamon press , oxford ( 1965 ) . gradshteyn and i.m . ryzhik , _ table of integrals , series , and products _ , @xmath70 ed . ( 1998 ) academic press . m. castellano and v.a . verzilov , phys . st - ab * 1 * , 062801 ( 1998 ) . shulga , s.n . dobrovolsky , jetp lett . , vol . * 65 * , no . 8 , ( 1997 ) 611 - 614 . potylitsyn , nucl . instr . and meth . b * 145 * ( 1998 ) 169 - 179 . m. castellano , a. cianchi , g. orlandi , v.a . verzilov , nucl . instr . and meth . a * 435 * ( 1999 ) 297 - 307 . s. casalbuoni , b. schmidt and p. schmser , tesla report 2005 - 15 ( 2005 ) . d. stterlin , d. erni , m. dehler , h. jckel , h. sigg , v. schlott , nucl . instr . and meth . b * 264 * ( 2007 ) 361 - 370 . s. casalbuoni , b. schmidt , p . schmser , v. arsov , s. wesch , phys . st - ab * 12 * , 030705 ( 2009 ) .	in the transition radiation emission from a @xmath0 electron bunch hitting at a normal angle of incidence a metallic screen , the transverse and the longitudinal spatial coordinates of the electron bunch play different roles in determining the @xmath0 single electron radiation field amplitudes and their relative phases in relation to the different physical constraints which an electromagnetic radiative mechanism by a charged beam must meet : i.e. , temporal causality and covariance . the distribution of the @xmath0 electron longitudinal coordinates determines indeed the sequence of the @xmath0 electron collisions onto the metallic screen and , on the basis of the temporal causality principle , it also determines the distribution function of the relative emission phases of the @xmath0 single electron field amplitudes from the metallic surface . the distribution of the transverse coordinates of the @xmath0 electrons contributes as well to determine the relative phase distribution of the @xmath0 electron field amplitudes at the observation point - located on the longitudinal axis of the reference frame - providing a further phase information that accounts for the transverse displacement of the @xmath0 electrons with respect to the beam axis . the distribution of the transverse coordinates of the @xmath0 electrons is a relativistic invariant under a lorentz transformation with respect to the direction of motion of the beam and , consequently , it is expected to leave a covariant mark on the @xmath0 single electron amplitudes composing the radiation field . the covariant imprinting of the @xmath0 electron transverse density on the radiation field affects both the temporal coherent and incoherent parts of the transition radiation energy spectrum . such a dependence of the @xmath0 single electron radiation field amplitudes on the electron density in the transverse plane manifests itself as an increase - towards an asymptotic limit - of the radiated spectral energy with the decrease of the beam transverse size and as a spectral narrowing of the angular distribution of the radiation intensity with respect to the ideal case of a single electron hitting an infinite metallic screen . in the case of a round metallic screen with an arbitrary radius , the formal expression of the transition radiation energy spectrum will be derived and numerical results will be presented . form factor , coherence , fourier transform , collective effects 41.60.-m , 41.75.-i , 42.25.kb , 42.30.kq
the fundamental parameters of the gaugino / higgsino sector in supersymmetric theories , the u(1 ) and su(2 ) gaugino masses @xmath7 and @xmath8 , and the higgsino mass @xmath9 , can be determined very accurately in experiments at prospective @xmath0 linear colliders . this has been demonstrated in the elaborate analysis of ref.@xcite . in the initial phase of the colliders , a total energy of @xmath10 , raised later to @xmath11 tev , is planned to be reached with a high integrated luminosity of @xmath12 within a few years @xcite ; the electron and positron beams are planned to be polarized with a degree of 80 and 60% , respectively @xcite . + in many scenarios it is expected @xcite that the light charginos @xmath1 and the two lightest neutralinos @xmath2 and @xmath3 can be accessed kinematically are realized in the chargino / neutralino sector . ] in the initial phase of the colliders . in this addendum to ref.@xcite it will be shown that in the minimal supersymmetric standard model ( mssm ) the gaugino and higgsino mass parameters can be reconstructed completely in this case , even in cp non invariant versions of the model , by measuring the properties of the light particle set @xmath13 . + in @xmath0 collisions , charginos and neutralinos can be produced in diagonal and mixed pairs among which the reactions , giving rise to visible final states , @xmath14 & & e^+e^-\,\rightarrow\,\,\tilde{\chi}^0_1\,\tilde{\chi}^0_2 \end{aligned}\ ] ] are of particular experimental interest in the present context . + in standard definition @xcite , the diagonalization of the chargino matrix in the mssm @xmath15 \sqrt{2}m_w s_\beta & \mu \end{array}\right)\end{aligned}\ ] ] generates the light and heavy states @xmath16 ( @xmath17 ) , while diagonalizing the neutralino mass matrix @xmath18 0 & m_2 & m_z c_\beta c_w & -m_z s_\beta c_w\\[2 mm ] -m_z c_\beta s_w & m_z c_\beta c_w & 0 & -\mu \\[2 mm ] m_z s_\beta s_w & -m_z s_\beta c_w & -\mu & 0 \end{array}\right)\ \label{eq : massmatrix}\end{aligned}\ ] ] leads to four neutralino states @xmath19 ( @xmath20 ) , ordered sequentially with rising mass . the coefficients @xmath21 , @xmath22 are given by the ratio of the vacuum expectation values of the higgs fields , @xmath23 , and @xmath24 are the sine and cosine of the electroweak mixing angle . in cp noninvariant theories , the mass parameters are complex . by reparametrization of the field basis , the su(2 ) mass parameter @xmath8 can be set real and positive , while the u(1 ) mass parameter @xmath7 is assigned the phase @xmath25 , and the higgsino mass parameter @xmath9 the phase @xmath26 . + the examples presented later , will be based on a single reference point for a cp non invariant extension of the mssm , compatible with all experimental constraints @xcite , + @xmath27 these fundamental parameters generate the following light chargino and neutralino masses , @xmath28 while the heavy masses are given by @xmath29 the cross sections depend on the sneutrino and selectron masses which we assume , for the sake of simplicity , to be measured in threshold scans : @xmath30 [ angular correlations in the production of chargino / neutralino states can be exploited otherwise to determine the slepton masses @xcite . ] the cross sections for chargino and neutralino pair production with polarized beams are big at @xmath31 gev , @xmath32 & & \sigma_l\{\tilde{\chi}^0_1\tilde{\chi}^0_2\ } = 327.9\ , { \rm fb } \qquad \ \ \sigma_r\{\tilde{\chi}^0_1\tilde{\chi}^0_2\ } = 16.4\ , { \rm fb } \label{eq : n12 x - section}\end{aligned}\ ] ] so that sufficiently large ensembles of events , between @xmath33 and @xmath34 events for @xmath35 and @xmath36 , will be generated , etc , can be used to refine the analysis . ] , allowing the analysis of the properties of the chargino @xmath1 and the neutralinos @xmath37 at great detail . defining the mixing angles in the unitary matrices diagonalizing the chargino mass matrix @xmath38 by @xmath39 and @xmath40 for the left and right chiral fields , the fundamental susy parameters @xmath8 , @xmath41 , @xmath42 and @xmath43 can be derived from the chargino masses and the cosines @xmath44 of the mixing angles , @xmath45 \left[\sigma^2-\delta^2(c_{2l}+c_{2r})^2\right ] } } \label{eq : cosphi}\\ \tan\beta&=&\sqrt{\frac{1-\delta ( c_{2l}-c_{2r } ) } { 1+\delta ( c_{2l}-c_{2r})}}\label{eq : tanb}\end{aligned}\ ] ] where the dimensionless quantities @xmath46 /2 m^2_{_w}\qquad { \rm and } \qquad \delta = \left[m^2_{\tilde{\chi}^\pm_2}-m^2_{\tilde{\chi}^\pm_1}\right]/ 4 m^2_{_w}\end{aligned}\ ] ] have been introduced for notational convenience . + if only the light charginos @xmath1 can be produced , besides the mass @xmath47 , both the mixing parameters @xmath48 can be measured nevertheless @xcite . the @xmath48 can be determined uniquely if the polarized cross sections are measured at one energy including transverse beam polarization , or else if the longitudinally polarized cross sections are measured at two different energies . + it is apparent from eq.([eq : tanb ] ) that the heavy chargino mass is bounded from above after @xmath47 and @xmath48 are measured experimentally . at the same time , it is bounded from below by not observing the heavy chargino in mixed light@xmath49heavy pair production . the ensuing constraint on the heavy chargino mass @xmath50 is quite restrictive ; the upper bound can still be improved by exploiting the slightly more restrictive , but algebraically more complicated bound derived from @xmath51 in eq.([eq : cosphi ] ) . for the example introduced above , a narrow window of @xmath52 is predicted after initial experimentation at the energy @xmath31 gev . + if both the light chargino mass @xmath47 and the heavy chargino mass @xmath53 can be measured , the fundamental parameters @xmath54 can be extracted unambiguously . however , if @xmath55 is not accessible , it depends on the cp properties of the higgsino sector whether they can be determined or not in the chargino system alone . + * ( a ) * if the higgsino sector is , eq.([eq : cosphi ] ) can be exploited to determine @xmath56 from @xmath57 , up to at most a two fold ambiguity , refs.@xcite . this ambiguity can be resolved if other observables can be evaluated , _ notabene _ the mixed pair @xmath58 production cross sections . + * ( b ) * if @xmath55 is not accessible , the parameters in eqs.([eq : m2][eq : tanb ] ) can not be determined in a theory in the chargino sector alone . they remain dependent on the unknown heavy chargino mass @xmath53 . two trajectories are generated in @xmath59 space , parametrized by @xmath53 and classified by the two values @xmath26 and @xmath60 for the phase of the higgsino mass parameter , _ i.e. _ the sign of @xmath61 . it will be shown in the next section that the analysis of the two light neutralino states @xmath2 and @xmath3 can be used to predict the heavy chargino mass @xmath53 in the mssm . the phase ambiguity can be resolved with the sign of @xmath62 , coupled to the sign of @xmath63 , remaining undetermined , _ i.e. _ the sign of @xmath64 coupled to the sign of @xmath61 . ] by measuring the sign of cp odd observables associated with normal @xmath3 polarization in @xmath65 pair production @xcite . subsequently the entire set of fundamental gaugino and higgsino parameters can be determined uniquely . the symmetric neutralino mass matrix @xmath66 is diagonalized by a unitary matrix , defined such that the mass eigenvalues @xmath67 of the four majorana fields @xmath19 are positive . + the squared mass eigenvalues of @xmath68 are solutions of the characteristic equations @xcite @xmath69 with the invariants @xmath70 , @xmath71 , @xmath72 and @xmath73 given by the fundamental su(2 ) and u(1 ) gaugino mass parameters @xmath8 and @xmath7 , and the higgsino mass parameter @xmath9 , _ i.e. _ the moduli @xmath8 , @xmath74 , @xmath41 and the phases @xmath25 , @xmath26 . each of the four invariants @xmath70 , @xmath71 , @xmath72 and @xmath73 is a binomial of @xmath75 and @xmath76 . therefore , each of the characteristic equations in the set ( [ eq : characteristic ] ) for the neutralino mass squared @xmath77 can be rewritten in the form @xmath78 the coefficients @xmath79 , @xmath80 and @xmath81 are functions of the parameters @xmath8 , @xmath41 , @xmath26 , @xmath43 and the mass eigenvalue @xmath77 for fixed @xmath82 . the coefficient @xmath80 is necessarily proportional to @xmath61 because physical neutralino masses are cp even ; the sign ambiguity for @xmath61 , a result of the two fold cos solution @xmath83 , transfers to the associated sign ambiguity in the cp odd quantity @xmath62 , _ i.e. _ in @xmath64 the characteristic equation ( [ eq : mphase ] ) defines a circle in the @xmath84 plane for each neutralino mass @xmath67 . with only two light neutralino masses @xmath85 and @xmath86 measured , we are left with a two fold ambiguity marked by the two , black and open , dots in fig . [ fig : circle ] . + * ( a ) * in the version of the mssm , the measurements of @xmath47 and @xmath48 lead to at most a two fold ambiguity in @xmath59 . inserting the two neutralino masses @xmath87 in the set eq.([eq : mphase ] ) for @xmath88 , this induces at most a two fold ambiguity in @xmath7 . this ambiguity can finally be resolved by measuring the cross sections for @xmath36 pair production . + * ( b ) * however , in scenarios with , the _ loci _ of the two crossing points depend on the unknown heavy chargino mass @xmath53 . the two values of @xmath89 and @xmath62 are depicted for the window of the allowed values @xmath53 in the two upper panels fig . 2a / b assuming that the sign of @xmath61 will have been determined [ cf . footnote # 4 ] . + by measuring the pair production cross sections @xmath90 and @xmath91 , a unique solution , for both the parameters @xmath53 and @xmath92 can be found at the same time as demonstrated in the lower panels fig . the second solution @xmath93 gev for @xmath90 can be excluded by the measurement of @xmath91 because the predicted value of @xmath94 fb is far away from its experimental value of @xmath95 fb . the cross sections depend on the neutralino mixing parameters which are given by the fundamental u(1 ) and su(2 ) gaugino and higgsino parameters . they are parametrized therefore solely by @xmath53 after the chargino system and the two neutralino masses @xmath87 are evaluated as elaborated before . as a result , the additional measurement of the cross sections leads to a unique solution for @xmath53 and subsequently to a unique solution for @xmath96 [ assuming that the discrete cp ambiguity in the associated signs of @xmath61 and @xmath64 has been resolved by measuring the normal @xmath3 polarization ] . + + this procedure can nicely be summarized in a single figure : fig . the two crossing points of the masses @xmath85 and @xmath86 in fig.[fig : circle ] define the two trajectories in the complex @xmath7 plane , parametrized by the heavy chargino mass @xmath53 . the @xmath97 and @xmath98 cross sections vary along the two trajectories ; comparing the predicted values with the measured values leads to a unique solution on the trajectories marked by a black dot , _ i.e. _ to unique values for @xmath7 and @xmath53 , along with finally unique solutions for @xmath99 and @xmath43 . + _ to summarize . _ if only the light chargino @xmath1 and the two light neutralinos @xmath2 and @xmath3 can be accessed kinematically in the initial phase of @xmath0 linear colliders , measurements of the masses @xmath85 and @xmath86 , and the neutralino production cross section and @xmath3 polarization in the process @xmath100 , combined with the light chargino mass @xmath47 and the chargino production cross section of the process @xmath101 for polarized beams , allow us to perform a complete and precise analysis of the basic mssm parameters in the gaugino / higgsino sector : @xmath102 . + + syc is supported by the korea science and engineering foundation ( kosef ) through the kosef dfg collaboration project , project no . 20015111022 , jk by the kbn grant no . 2p03b 060 18 . the work is supported in part by the european commission 5-th framework contract hprn - ct-2000 - 00149 . + g. moortgat pick and h. steiner , epjdirect c * 6 * ( 2001 ) 1 ; tesla technical design report , part ii : the accelerator , _ eds . _ r. brinkmann , k. flttmann , j. rossbach , p. schmser , n. walker and h. weise , desy 2001011 [ hep ph/0106315 ] . s.y . choi , a. djouadi , h. dreiner , j. kalinowski and p.m. zerwas , eur . phys . j. c * 7 * ( 1999 ) 123 ; s.y . choi , a. djouadi , h.s . song and p.m. zerwas , eur . j. c * 8 * ( 1999 ) 669 ; s.y . choi , m. guchait , j. kalinowski and p.m. zerwas , phys . lett . * b479 * ( 2000 ) 235 . y. kizukuri and n. oshimo , phys . lett . * b249 * ( 1990 ) 449 ; t. ibrahim and p. nath , phys . d * 58 * ( 1998 ) 111301 [ err . * 60 * ( 1998 ) 099902 ] and hep ph/0107325 ; t. ibrahim and p. nath , phys . d * 61 * ( 2000 ) 093004 ( 2000 ) ; m. brhlik , g.j . good and g.l . kane , phys . d * 59 * ( 1999 ) 115004 ; r. arnowitt , b. dutta and y. santoso , phys . d * 64 * ( 2001 ) ; s. abel , s. khalil and o. lebedev , nucl . phys . * b606 * ( 2001 ) 151 and references therein .	in the preceding reference @xcite we have shown how the fundamental gaugino and higgsino parameters of the chargino and neutralino system in supersymmetric theories can be determined in high precision experiments at @xmath0 linear colliders . within the minimal supersymmetric standard model these parameters can be reconstructed completely even if only the light charginos @xmath1 and the light neutralinos @xmath2 and @xmath3 are kinematically accessible in the initial phase of these machines , as demonstrated in this addendum . + desy 02020 + ift02/03 + hep ph/0202039 + + + s.y . choi@xmath4 , j. kalinowski@xmath5 , g. moortgat pick@xmath6 and p.m. zerwas@xmath6 1 . @xmath4 _ department of physics , chonbuk national university , chonju 561756 , korea _ + 2 . @xmath5 _ instytut fizyki teoretycznej , uniwersytet warszawski , pl00681 warsaw , poland _ + 3 . @xmath6 _ deutsches elektronen - synchrotron desy , d-22603 hamburg , germany _
cavity optomechanics @xcite explores the interaction between a mechanical resonator and photons in a driven electromagnetic cavity via radiation pressure force . in the past decade , remarkable progress has been made in this emerging field , including quantum ground state cooling of the nanomechanical resonators @xcite , optomechanically induced transparency ( omit ) @xcite , as well as coherent photon - phonon conversion @xcite and quantum state transfer @xcite . the force exerted by a single photon on a macroscopic mechanical resonator is typically weak and intrinsically nonlinear . experiments to date have focused on the regime of strong optical driving , where the optomechanical coupling can by enhanced by a factor @xmath0 , where @xmath1 is the mean photon number in the cavity @xcite . but such enhancement comes at the expense of making the effective interaction linear . recently , it has been reported that single - photon strong - coupling regime , where the single - photon optomechanical coupling rate @xmath2 exceeds the cavity decay rate @xmath3 , can be realized in single - mode @xcite or two - mode optomechanical systems @xcite . in this regime , the inherently nonlinear optomechanical interaction is significant at the level of single photons and phonons . among all the nonlinear phenomenon in a cavity optomechanical system , optical bistability is one of the focuses of research interest . recently , the bistable behavior of the mean intracavity photon number in optomechanical systems with a bose - einstein condensate ( bec ) @xcite , ultracold atoms @xcite and a quantum well @xcite has been extensively studied . the photon number in the optical cavity with a bec or ultracold atoms to allow bistable behavior is usually low and even below unity due to the collective atomic motion . however , in the generic optomechanical system consisted of an empty optical cavity with one movable end mirror , optical bistability typically occurs at high photon numbers . producing such phenomena at very low photons number is desirable for applications from optical communication to quantum computation . in the present paper , we theoretically investigate the bistable behavior of the intracavity photon number in a two - mode optomechanical system in the simultaneous presence of two strong pump laser beams and a weak probe laser beam . by adjusting the frequency and power of the pump beams , we can efficiently control the optical bistability in each cavity . in particular , one can control the bistable behavior in the right cavity by adjusting the pump beam driving the left cavity . most importantly , such phenomena can appear at low photon numbers below unity because the two cavities are coupled to a common mechanical resonator based on photons and phonons . the paper is organized as follows . section gives the theoretical model and method . results and discussion are shown in sec . . a summary is presented in sec . . and @xmath4 , are coupled to the same mechanical mode @xmath5 . the left cavity is driven by a strong pump beam @xmath6 in the simultaneous presence of a weak probe beam @xmath7 while the right cavity is only driven by a pump beam @xmath8.,width=302 ] the system under consideration is shown in fig . 1 . two optical cavity modes @xmath9(@xmath10 ) are coupled to a common mechanical mode @xmath5 via an interaction hamiltonian @xmath11 , where @xmath9 and @xmath5 are the annihilation operators of the cavity and mechanical mode , respectively , and @xmath12 is the single - photon coupling rate between the mechanical mode and the @xmath13th cavity mode . physically , @xmath12 represents the frequency shift of cavity mode @xmath13 due to the zero - point motion of the mechanical resonator . the left cavity is driven by a strong pump laser beam @xmath6 of frequency @xmath14 and a weak probe laser beam @xmath7 of frequency @xmath15 simultaneously , and the right cavity is only driven by a strong pump laser beam @xmath8 of frequency @xmath16 . in a rotating frame at the pump frequency @xmath17 and @xmath16 , the hamiltonian of the two - mode optomechanical system reads @xcite @xmath18 the first term represents the energy of the two optical cavity modes with resonance frequency @xmath19 , where @xmath20 and @xmath21 are the corresponding cavity - pump field detunings . the second term gives the energy of the mechanical mode resonance frequency @xmath22 and effective mass @xmath23 . the last three terms describe the interaction between the input fields and the cavity fields , where @xmath6 , @xmath8 , and @xmath7 are related to the power of the applied laser fields by @xmath24 , @xmath25 , and @xmath26 ( @xmath27 the linewidth of the _ _ k__th cavity mode ) , respectively . each optical cavity is coupled not only to a shared mechanical mode , but also to an optical bath at rate @xmath28 and to an external photonic waveguide at rate @xmath29 . therefore , the total cavity linewidth @xmath30 . here , @xmath31 is the detuning between the probe laser beam and the left pump laser beam . according to the heisenberg equations of motion and the commutation relation @xmath32=1 $ ] , and @xmath33=1 $ ] , the temporal evolutions of operators @xmath34 , @xmath4 , and @xmath35 [ which is defined as @xmath36 can be obtained . introducing the corresponding damping and noise terms for the mechanical and cavity modes , we derive the quantum langevin equations as follows : @xmath37 the cavity modes decay at the rate @xmath38 and are affected by the input vacuum noise operator @xmath39 with zero mean value , which obey the correlation function in the time domain @xcite , @xmath40 the mechanical mode is affected by a vicious force with damping rate @xmath41 and by a brownian stochastic force with zero mean value @xmath42 that has the following correlation function @xcite @xmath43,\end{aligned}\ ] ] where @xmath44 is the boltzmann constant and @xmath45 is the temperature of the reservoir of the mechanical resonator . setting all the time derivatives to zero , we derive the steady - state solution to eqs . ( 2)-(4 ) @xmath46 where @xmath47 and @xmath48 are the effective cavity detunings including radiation pressure effects . therefore , the mean intracavity photon numbers @xmath49 can be determined by the following coupled equations @xmath50 ^ 2},\\ n_{p2}=\frac{\kappa_{e,2}e_r^2}{\kappa_2 ^ 2+\left[\delta_2 - 2g_2/\omega_m(g_1n_{1}-g_2n_{2})\right]^2}.\end{aligned}\ ] ] this form of coupled cubic equations are characteristic of the optical multistability @xcite . recently , optical bistability of mean intracavity photon numbers in optomechanical systems with ultracold atoms @xcite and a bec @xcite have been experimentally observed in the transmission of the probe laser beam through the cavity . here we theoretically propose a flexible scheme to control the bistable behavior in the two - mode optomechanical system according to eqs . ( 9)-(10 ) . in essence it is the nonlinearity owing to optomechanical coupling that is responsible for the bistable behavior in this scheme . we consider for illustration an experimentally realized two - mode optomechanical system . the parameters used are @xcite : @xmath51 thz , @xmath52 thz , @xmath53 mhz , @xmath54 ghz , @xmath55 , @xmath56 , @xmath57 khz , @xmath58 khz , @xmath59 ghz , @xmath60 , where @xmath61 is the quality factor of the nanomechanical resonator , and the damping rate @xmath62 is given by @xmath63 we can see that @xmath64 and @xmath65 , therefore the system would be in the good - cavity limit , a prerequisite for ground - state cooling of nanomechanical resonators @xcite . for left pump power @xmath66 equals to 0.1 , 2 , and 3 @xmath67w(from bottom to top ) ( b ) the left pump power @xmath66 for @xmath68 the right pump power is kept equal to 0.1 @xmath67w . other parameters used are @xmath51 thz , @xmath52 thz , @xmath53 mhz , @xmath54 ghz , @xmath55 , @xmath56 , @xmath59 ghz , @xmath60 , @xmath57 khz , and @xmath58 khz.,width=302 ] the two - mode optomechanical system we consider here enables more controllability in the bistable behavior of the intracavity photon number . characterization of the optomechanical cavity can be performed by using two strong pump beams combined with a weak probe beam . here , we mainly investigate the optical bistability in each cavity by adjusting the power and frequency of the left pump beam . figure 2(a ) plots the mean intracavity photon number in the left cavity as a function of the left cavity - pump detuning @xmath69 for various pump powers . when the power of the the left pump beam is @xmath70w , the curve is nearly lorentzian . however , when the power increases above a critical value , the system exhibits bistable behavior as shown in the curves for @xmath71w and @xmath72w , where the initially lorentzian resonance curve becomes asymmetric . in this case , the coupled cubic equations ( 9)-(10 ) for the mean intracavity photon number yields three real roots . the largest and smallest roots are stable , and the middle one is unstable , which is represented by the dashed lines in fig . furthermore , we can see that larger cavity - pump detuning is necessary to observe the optical bistable behavior with the increasing pump beam power . the bistable behavior can also be seen from the hysteresis loop for the mean intracavity photon number versus the pump power curve shown in fig . 2(b ) . here , both the cavities are pumped on their respective red sidebands , i.e. , @xmath73 , which is beneficial for resolved sideband cooling of the nanomechanical resonator @xcite . consider the left pump power increases from zero gradually , the mean intracavity photon number @xmath74 initially lies in the lower stable branch ( corresponding to the smallest root ) . when the pump power @xmath66 increases to a critical value , about @xmath75w in our case , @xmath74 approaches the end of this branch . the hysteresis then follows the arrow and jumps to the upper branch . if the @xmath66 is increased further , @xmath74 remains on the upper branch . if @xmath66 is decreased , @xmath74 gradually approaches the beginning of the second stable branch . if @xmath66 is decreased even further , the hysteresis follows the arrow and switches back to the lower stable branch . based on the above discussions , one can easily realize an optical switch by modulating the power and frequency of the pump beam . with @xmath76 and @xmath77 , respectively . the left pump power @xmath66 equals to @xmath78 and the right pump power @xmath79 equals to @xmath80 . when the coupling strength between the left cavity and the mechanical resonator @xmath81 is equal to zero and @xmath76 , the intracavity photon number keeps constant . other parameters are the same as in figure 2.,width=302 ] in the following , we mainly investigate the optical bistability in the right cavity by controlling the frequency and power of the left pump beam . mean intracavity photon number @xmath82 in the right cavity as a function of the left cavity - pump beam detuning @xmath83 is plotted in fig . 3 . the left pump power @xmath71w and the right pump power @xmath84w . when the coupling between the left cavity and the mechanical resonator turns off , i.e. , @xmath85 , the two - mode optomechanical system becomes the generic single - mode optomechanics , and the pump beam driving the left cavity ca nt have an impact on the photon numbers in the right cavity via the mechanical mode . in this case , if the the right cavity is pumped on its red sideband ( @xmath76 ) , it can be seen clearly from the middle straight line in fig . 3 that the mean intracavity photon number @xmath82 keeps constant when the left cavity - pump detuning @xmath83 changes . however , if the coupling between the left cavity and the nanomechanical resonator turns on , bistable behavior of the mean intracavity photon number in the right cavity will appear . when @xmath76 , the average photon number is larger than the constant value obtained before . however , if the right cavity is pumped on its blue sideband , i.e. , @xmath77 , the average photon number is smaller than the above constant value . the underling physical mechanism for this phenomenon can be explained as follows . when @xmath85 and @xmath76 , the hybrid system turns to the typical single mode optomechanical system , and the intracavity photon number is directly related to the pump power @xmath79 and the square of the cavity - pump detuning ( @xmath86 ) @xcite . w with ( a ) @xmath76 ( b ) @xmath77 , respectively . the left cavity - pump detuning @xmath83 is kept equal to @xmath87 . other parameters are the same as in figure 2.,width=302 ] therefore , the photon number keeps constant when @xmath88w , and @xmath89 . however , when @xmath90 , the intracavity photons in the left cavity will have an effect on the common nanomechanical resonator and subsequently the photon number in the right cavity . the simultaneous presence of the left pump and probe beam induces a radiation pressure force at the beat frequency @xmath31 , which drives the mechanical mode to oscillate coherently . when @xmath91 , the highly off - resonant stokes scattering at the frequency @xmath92 is strongly suppressed and only the anti - stokes scattering at the frequency @xmath93 builds up within the right cavity , leading to the up - conversion of the pump photons to the cavity photons at the frequency @xmath94 . therefore , the average photon number in the right cavity is lager than the constant value without of the effect of the left cavity . on the other hand , when @xmath77 , the anti - stokes scattering is strongly suppressed while the stokes scattering at the frequency @xmath92 builds up within the cavity , leading to the down - conversion of cavity photons at the frequency @xmath94 to the pump photons . in this case , the average photon number in the right cavity is smaller than the constant value when @xmath85 . therefore , by adjusting the left cavity - pump beam detuning @xmath83 , one can observe the bistable behavior of the intracavity photon number in the right cavity . optical bistability in the right cavity can also be seen from the hysteresis loop for the mean intracavity photon number versus the left pump power when @xmath76 and @xmath77 , as shown in fig . we have taken left cavity - pump beam detuning to be @xmath95 and @xmath84w . similarly , mean intracavity photon number when @xmath76 is larger than the situation when @xmath77 . different from the curve for @xmath76 , @xmath82 initially lies in the upper stable branch ( corresponding to the largest root ) if @xmath77 . with the increase of the pump power @xmath66 , @xmath82 decreases gradually and jumps to the lower branch when @xmath66 approaches a critical value , about @xmath75w in our case . we can also realize a switch between the lower and upper stable branch by controlling the power of the left pump beam . for @xmath71w ( b ) the left pump power @xmath66 for @xmath95 . other parameters used are @xmath96pw , @xmath76 , @xmath51 thz , @xmath52 thz , @xmath53 mhz , @xmath54 ghz , @xmath55 , @xmath56 , @xmath59 ghz , @xmath60 , @xmath57 khz , and @xmath58 khz.,width=302 ] in our previous discussions , we have demonstrated optical bistability in both cavities , and the mean intracavity photon number is usually very large , at least thousands of photons in the right cavity ( see fig . 3 and 4 ) . for the single - mode optomechanical system ( @xmath97 ) , it would need much more photons in the cavity in order to reach the bistable regime ( see fig . 2 for illustration ) @xcite . in what follows , we will show that two - mode optomechanical system allows for optical bistability at extremely low cavity photon numbers . 5 plots the mean photon number in the right cavity as a function of ( a ) left cavity - pump beam detuning @xmath83 for @xmath71w and ( b ) left pump power @xmath66 for @xmath95 . parameters of the right pump beam are @xmath96pw and @xmath76 . due to the low pump power , intracavity photon number in the right cavity is very small , i.e. , @xmath98 . generally , such low photon numbers can not exhibit bistable behavior in the empty cavity optomechanical systems . however , in the two - mode optomechanics we consider here , when the left cavity is driven by a strong pump power , because the two cavities are coupled to a common nanomechanical resonator , optical bistability can still exist in the right cavity at the extremely low intracavity photon numbers . this phenomenon signifies the strong nonlinear effects in the weak coupling regime @xcite , which is enabled by the long lifetime of the mechanical mode and the strong pump on the left cavity . recently , two related works by l _ @xcite and kuzyk _ et al . _ @xcite have also shown that strong nonlinearities can be obtained in two - mode optomechanical systems in the weak coupling regime . as pointed out by previous works , the bistable behavior of intracavity photon numbers in the two - mode optomechanical system under consideration also provides a candidate of realizing a controllable optical switch . for this , the two stable branches of photon numbers in the right cavity act as the optical switch . when the frequency and power of the left pump beam are fixed , the switch between the lower stable branch and the upper stable branch can be easily realized by controlling the frequency and power of the right pump beam . furthermore , the left pump beam can be used as a control parameter to enable or disable this switch . when the photon number in the right cavity is less than one , if we turn off the left pump beam , the bistable behavior disappears and only one stable branch exists . in conclusion , we have investigated the optical bistability in a two - mode optomechanical system where two optical cavities are coupled to a common mechanical resonator via radiation pressure force . compared with the generic single - mode cavity optomechanics , such a two - mode optomechanical system allows one to control the optical bistability in a much more flexible way . the bistable behavior of the mean intracavity photon number in one cavity can be tuned by the power and frequency of the pump laser beam driving another cavity . furthermore , bistability at low photon numbers below unity should be possible in such a coupled system . the authors gratefully acknowledge support from national natural science foundation of china ( grant nos.11074088 and 11174101 ) and jiangsu natural science foundation ( grant no . bk2011411 ) .	we explore theoretically the bistable behavior of intracavity photon number in a two - mode cavity optomechanical system , where two cavity modes are coupled to a common mechanical resonator . when the two cavity modes are driven by two pump laser beams , respectively , we find that the optical bistability can be controlled by tuning the power and frequency of the pump beams . the common interaction to a mechanical mode enables one to control the bistable behavior in one cavity by adjusting the pump laser beam driving another cavity . we also show that both branches of optical bistability at photon numbers below unity can be observed in this two - mode optomechanical system . this phenomenon can find potential applications in controllable optical switch .
timing irregularities in a rotation - powered pulsar , such as discontinuous glitches @xcite and stochastic timing noise @xcite , provide an indirect probe of the internal structure of the star . the physical processes usually invoked to explain these phenomena are ( un)pinning of feynman - onsager vortices in the crystalline inner crust @xcite , starquakes @xcite , and thermally driven vortex creep @xcite . less attention has been directed at the _ global hydrodynamics _ of the superfluid , except within the context of the spin - up problem in cylindrical geometry @xcite . the importance of the global hydrodynamics was demonstrated by @xcite , who simulated pulsar rotational irregularities in the laboratory by impulsively accelerating rotating containers of he ii , obtaining qualitative agreement with radio timing data ( e.g. glitch amplitudes and post - glitch relaxation times ) . in this paper , we examine how the global flow pattern of superfluid in the outer core of a neutron star affects the rotation of the star . we focus on the outer core for simplicity : vortex pinning is thought to be weak or non - existent there @xcite , and the fluid is mainly isotropic ( @xmath2 cooper pairing ) @xcite , reducing the problem to a hydrodynamic one in a spherical shell . even with this simplification , the calculation remains numerically challenging : the spherical couette problem for a superfluid was solved for the first time only recently @xcite , generalizing previous work on the cylindrical taylor - couette problem for a superfluid @xcite and the spherical couette problem for a classical viscous fluid @xcite . an isotropic ( @xmath2-paired ) neutron superfluid is described by the two - fluid hall - vinen - bekarevich - khalatnikov ( hvbk ) model . in this model , the viscous normal fluid and inviscid superfluid components feel a mutual friction force whose magnitude and direction depends on the distribution of feynman - onsager vortices @xcite . the vortices are organized in a rectilinear array if the flow is strictly toroidal , but they evolve into a tangle of reconnecting loops when the counterflow along the rotation axis exceeds a threshold , exciting the donnelly - glaberson instability ( dgi ) @xcite . @xcite showed that the dgi is excited in a neutron star under a wide range of conditions , driven by the meridional component of the spherical couette flow ( scf ) in the interior . the mutual friction force changes dramatically during transitions between a vortex array and a vortex tangle @xcite , affecting the rotational evolution of the star . in this paper , we propose a phenomenological model for timing irregularities in radio pulsars based on the creation and destruction of a vortex tangle _ superfluid turbulence _ @xcite in the outer core of a rotating neutron star . in our scenario , a glitch comprises the following sequence of events . ( i ) differential rotation between the outer core and crust of the star , built up over time through electromagnetic spin down , generates a meridional ekman counterflow in the outer core . we show that the axial counterflow exceeds the dgi threshold , creating a vortex tangle throughout the outer core . the mutual friction in this turbulent state takes the isotropic gorter - mellink ( gm ) form and is much weaker than the mutual friction associated with a rectilinear vortex array . ( ii ) when the glitch occurs , triggered by an unknown mechanism , the outer core and inner crust suddenly come into corotation and the vortex tangle decays , ultimately converting into a rectilinear vortex array . the decay process lasts @xmath0 s , depending on the drag force acting on the vortex rings , after which the mutual friction takes the anisotropic hall - vinen ( hv ) form @xcite and increases by @xmath3 orders of magnitude , precipitating a torque crisis " . ( iii ) after the glitch , differential rotation builds up again between the outer core and the crust due to electromagnetic spin down . when the axial counterflow exceeds the dgi threshold , after @xmath1 yr , the vortex array breaks up again into a tangle and the mutual friction drops sharply . similar transitions from turbulent to laminar flow in a superfluid have been observed in laboratory experiments where he ii , cooled to a few mk , flows around an oscillating microsphere @xcite . the paper is organized as follows . hvbk theory is briefly reviewed in [ sec : hvbkt ] , together with the pseudospectral numerical method which we use to solve the hvbk equations in a differentially rotating spherical shell . the physics of the turbulent - laminar transition in a generic glitch scenario is elaborated in [ sec : vtmodel ] . the response of the stellar crust to a turbulent - laminar transition in the outer core is calculated numerically in [ sec : rotevol ] . finally , the results are summarized and applied to observational data in [ sec : summary ] ; a fuller observational analysis will be carried out in a future paper . the neutron superfluid in the outer core exists in an isotropic ( @xmath2 ) phase for densities @xmath4 in the range @xmath5 ( where @xmath6 is the nuclear saturation density ) and an anisotropic ( @xmath7 ) phase for @xmath8 @xcite ; the depth at which the @xmath2-@xmath7 transition occurs is not known precisely @xcite . as the hydrodynamic equations for the @xmath7 phase @xcite are complicated and hard to handle numerically , we construct our model around the @xmath2 phase , described by hvbk theory in this paper . we plan to extend the model to the @xmath7 phase in the future . the protons in the outer core are probably in a type ii superconducting state , where the magnetic field is quantized into fluxoids @xcite . protons inside magnetic fluxoids interact with neutrons in feynman - onsager vortices @xcite . if vortex pinning occurs , and if the core magnetic field has comparable poloidal and toroidal components @xcite , the geometry of the neutron - fluxoid interaction can be complicated @xcite . to avoid these difficulties , we incorporate the protons ( and other charged species ) into the normal component of the superfluid , a common approximation @xcite we neglect proton - neutron entrainment @xcite , which is observed in terrestrial @xmath9he-@xmath10he mixtures @xcite and is an important mechanism modifying the mutual friction for rectilinear vortices in a neutron star @xcite . we neglect hydrodynamic forces arising from entropy gradients and angular momentum textures @xcite . finally , we neglect vortex pinning for simplicity , except at the inner and outer boundaries , even though it seems likely that pinning plays an important role in glitch dynamics @xcite . the hvbk model for a rotating superfluid is a generalization of the landau - tisza two - fluid model that includes the hydrodynamic forces exerted by quantized vortices in he ii @xcite and @xmath2-paired neutron matter @xcite . fluid particles in the theory are assumed to be threaded by many coaligned vortices , a valid assumption over length - scales longer than the average inter - vortex separation ( and shorter than the average radius of curvature ) . in the continuum limit , the vorticity of the superfluid component satisfies @xmath11 macroscopically ( cf . @xmath12 microscopically ) . the normal component comprises charged species ( protons and electrons , locked together by the external magnetic field ) plus thermal excitations and behaves like a classical , viscous , navier - stokes fluid ( kinematic viscosity @xmath13 ) . the isothermal , incompressible ( @xmath14 ) hvbk equations of motion take the form @xcite @xmath15 @xmath16 with @xmath17 , where @xmath18 ( @xmath19 ) and @xmath20 ( @xmath21 ) are the normal fluid ( superfluid ) velocities and densities respectively . effective pressures @xmath22 and @xmath23 are defined by @xmath24 and @xmath25 , with @xmath26 . note that , in neutron star applications , the gravitational potential can be subsumed into @xmath27 without loss of generality , in the incompressible limit . the vortex tension force per unit mass , which arises from local circulation around quantized vortices @xcite , is defined as @xmath28 with @xmath29 . here , @xmath30 is the stiffness parameter , @xmath31 is the quantum of circulation , @xmath32 is the mass of the neutron , @xmath33 is the radius of the vortex core , and @xmath34 is the intervortex spacing . note that @xmath35 , which has the dimensions of a kinematic viscosity , controls the oscillation frequency of kelvin waves excited on vortex lines @xcite . the mutual friction force per unit mass , @xmath36 , arises from the interaction between the quantized vortex lines and the normal fluid ( via roton scattering in he ii and electron scattering in a neutron star ) @xcite . the structure of this force depends on the global configuration of the vortices . if the vortices form a rectilinear array , the friction takes the anisotropic hv form @xcite @xmath37 where @xmath38 and @xmath39 are temperature - dependent mutual friction coefficients @xcite . if the vortices form a vortex tangle , the friction takes the isotropic gm form @xcite @xmath40 where @xmath41 is a dimensionless temperature - dependent coefficient , related to the original gm constant ( usually denoted by @xmath42 in the literature ) by @xmath43 . here , @xmath44 and @xmath45 are dimensionless constants of order unity @xcite . the hv and gm forces are in the ratio @xmath46 under the physical conditions prevailing in the outer core of a neutron star @xcite . this implies that the normal and superfluid components of the star ( and hence the crust and core , through viscous torques ) are effectively uncoupled most of the time ( when enough differential rotation has built up ) , but become tightly coupled in the immediate aftermath of a glitch when a turbulent - to - laminar transition occurs @xcite . it is important to bear this in mind when reading [ sec : vtmodel ] . we model the global hydrodynamics of the outer core by considering an hvbk superfluid enclosed in a differentially rotating spherical shell , i.e. superfluid spherical couette flow . the inner ( radius @xmath47 ) and outer ( radius @xmath48 ) surfaces of the shell rotate at angular velocities @xmath49 and @xmath50 , respectively , about a common axis . all quantities are expressed in dimensionless form using @xmath48 as the unit of length and @xmath51 as the unit of time . we adopt spherical polar coordinates @xmath52 , with the @xmath53 direction along the axis of rotation . the boundary conditions satisfied by the normal and superfluid components are subtle . neither component can penetrate the boundaries , implying @xmath54 . the tangential normal fluid velocity satisfies the no - slip dirichlet condition , viz . @xmath55 , @xmath56 . the behavior of the superfluid at the boundaries is influenced by the quantized vortices , which can either slide past , or pin to , irregularities on the boundaries . the former scenario implies @xmath57 , while the latter implies @xmath58 and requires a knowledge of the vortex line velocity @xmath59 locally . as the hvbk model provides no information about @xmath59 , we adopt the widely used no - slip compromise @xmath60 , @xmath61 , which in the spherical case does not restrict the orientation of the vortex lines at the surface @xcite . note that , initially , @xmath18 and @xmath19 must be divergence - free , with @xmath62 ; the stokes solution with @xmath63 satisfies this constraint @xcite . however , the numerical results presented in [ sec : vtmodel ] and [ sec : rotevol ] are obtained after several ekman times ( and hence rotation periods ) have elapsed , by which time the flow has forgotten its initial conditions , as in most astrophysical applications . many existing glitch models are based on cylindrical geometries @xcite , except for a few experiments with spheres @xcite . however , the global flow pattern in spherical couette flow differs from its cylindrical counterpart in two important ways . first , in a cylinder , the principal flow is toroidal , whereas , in a sphere , the principal flow is an axisymmetric combination of toroidal flow and meridional circulation for all @xmath64 @xcite . second , it is misleading to approximate the equatorial region of a sphere by a cylinder , even for @xmath65 ; the curvature , although slight , causes significant differences in the critical taylor number at which vortices appear @xcite . the differences between cylindrical and spherical couette flow are equally prominent in classical fluids and superfluids @xcite . spherical couette flow of a viscous fluid is controlled by three parameters : the reynolds number @xmath66 , the dimensionless gap width @xmath67 , and the angular shear @xmath68 . if a superfluid is also introduced , @xmath35 emerges as an additional parameter . in the laminar regime ( @xmath69 ) , the global flow can be classified according to the number of cells of meridional circulation in each hemisphere . in a narrow gap ( @xmath70 ) , the meridional circulation is slow and can be approximated by @xmath71 @xcite . as @xmath64 increases , a centrifugal instability generates toroidal taylor vortices near the equator until , at the onset of turbulence ( @xmath72 ) , a helical traveling wave develops . in a wide gap ( @xmath73 ) , by contrast , all secondary flows are nonaxisymmetric , and they oscillate when @xmath64 exceeds a threshold @xcite . in classical fluids , these flows have been thoroughly investigated numerically and experimentally @xcite . recently , high - resolution numerical simulations of superfluid spherical couette flow were performed successfully for the first time @xcite . figures [ fig : fig1]a and [ fig : fig1]b display streamlines of the normal fluid and superfluid components for @xmath74 , @xmath75 , @xmath76 , and hv mutual friction . both components exhibit a similar structure in each hemisphere : a square " ekman cell near the pole , three to four secondary meridional cells , a number of smaller cells , and two equatorial vortices near the outer boundary which are created and destroyed intermittently . we emphasize that the normal component is not assumed to be in uniform ( solid body ) rotation ; it is affected by the superfluid ( through mutual friction ) and the boundary conditions ( differential rotation ) . figures [ fig : fig1]c and [ fig : fig1]d display streamlines for a run with identical parameters and gm mutual friction , for which the inter - fluid coupling is @xmath46 times smaller , as noted in [ sec : hvbkt1 ] . we observe that the normal fluid exhibits a similar number of vortical structures , which are better developed and more regular than in figure [ fig : fig1]a , and the superfluid flow pattern is nearly unchanged from figure [ fig : fig1]b . meridional circulation is apparent in all the results ; we find axial velocity components in the range @xmath77 . the flow pattern in figure [ fig : fig1 ] changes with temperature just like in cylindrical couette flow . the superfluid behaves like a classical , viscous fluid near the critical temperature . the taylor vortices elongate axially , with a more complex pattern of eddies and counter - eddies emerging , as the temperature is lowered @xcite . for @xmath78 , anomalous modes emerge in the normal component , as in classical axisymmetric flow @xcite , characterized by ekman cells rotating in the opposite sense to the classical flow @xcite except near the critical temperature . for @xmath79 , the tension becomes less important than the mutual friction and the superfluid component comes to resemble the normal component @xcite . in this section , we investigate how the vorticity state of the outer core changes before , during , and after a glitch , in the context of a generic glitch scenario where the trigger mechanism of the glitch is unspecified . laboratory experiments on the attenuation of second sound in narrow channels @xcite , and numerical simulations based on the vortex filament method @xcite , show that , in a rotating container of he ii , an axial ( along @xmath53 ) counterflow @xmath80 excites growing kelvin waves which destabilize a rectilinear vortex array @xcite . the kelvin waves grow exponentially until reconnection between adjacent vortices occurs , generating a dense tangle . for typical neutron star parameters , one finds @xmath81 taking @xmath82 . the axial component of the ekman counterflow in the outer core of a typical neutron star generally exceeds the instability threshold ( [ eq : dgi ] ) . this is illustrated by figure [ fig : fig2 ] , a greyscale plot of @xmath83 ; the dgi is active in regions with @xmath84 . figure [ fig : fig2]a corresponds to hv mutual friction ( initially rectilinear vortex array ) ; figure [ fig : fig2]b corresponds to gm mutual friction ( after a tangle forms ) . in both figures , the dgi is active in most of the computational domain , implying that _ an initially rectilinear array is disrupted and , once disrupted , stays that way_. this result is extracted empirically from the simulations . the inclusion of compressibility and hence stratification @xcite restricts the dgi region to a thin boundary layer , but the overall conclusion is the same ( see [ sec : strat ] ) . the axial counterflow that excites the dgi is driven by ekman pumping . as the crust spins down electromagnetically at a rate @xmath85 , differential rotation builds up between the crust and outer core , with @xmath86 where @xmath87 is the time elapsed since the last glitch @xcite . the differential rotation induces meridional circulation @xcite : an ekman boundary layer , with @xmath88 , develops on a time - scale @xmath89 , and grows radially to a thickness @xmath90 cm on a time scale @xmath91 , spinning up the interior fluid @xcite . only the normal fluid is ekman pumped directly ; the superfluid is spun up by the normal fluid through the mutual friction force @xcite . in order to estimate the counterflow velocity , we compute @xmath92 empirically from numerical experiments and scale up the results to neutron star parameters . for example , in a typical set of runs with @xmath75 ( hv mutual friction ) , @xmath93 ( hv mutual friction ) , and @xmath94 ( gm mutual friction ) , and @xmath95 , we find @xmath96 , @xmath97 and @xmath98 , respectively , on average over the grid , implying @xmath99 , as a general rule and hence @xmath100 in a typical neutron star , the reynolds number in the outer core ( temperature @xmath101 ) is very large , viz . @xmath102 given the standard viscosity @xmath13 resulting from electron - electron scattering @xcite . consequently , the flow is likely to be turbulent @xcite . our numerical simulations can not access this regime due to computational limitations @xcite . however , we know from laboratory experiments with he ii that the flow at @xmath103 closely resembles classical , navier - stokes turbulence : the superfluid and normal fluid vorticity evolve in concert , because the vortex lines are locked to the normal fluid eddies by mutual friction @xcite . @xmath104 @xcite , comparable to that in a neutron star , viz . @xmath105 @xmath104 @xcite . ] for the dgi to be triggered in superfluid turbulence , two conditions must be met . first , there must be vortex segments directed parallel to the counterflow , i.e. @xmath106 @xcite . as the superfluid is locked to the normal fluid , and the vortex density is high , the condition @xmath106 is satisfied locally throughout the core . second , a vortex segment is locally unstable to ( helical ) kelvin wave perturbations of wavelength @xmath107 if the local counterflow exceeds the critical speed @xmath108 @xcite . large wavelengths ( of the order of the size of the star ) are most unstable , with @xmath109 cm giving @xmath110 cm s@xmath111 . in this regard , the dgi threshold ( [ eq : dgi ] ) is conservative ( @xmath112 ) . grows if the vacillation frequency @xmath113 is mantained over a time - scale @xmath114 @xcite . ] during a glitch event , the differential rotation instantaneously shuts off and the tangle starts to decay . in a self - sustaining vortex tangle , kelvin waves grow until their amplitude approaches the average vortex separation , whereupon the vortex lines continuously reconnect to form loops @xcite . reconnection stops once the counterflow ceases , as happens immediately after a glitch , when the outer core and inner crust come into corotation ( @xmath115 ) . the decay time - scale @xmath116 equals , to a good approximation , the reconnection time - scale just before the counterflow ceases . to estimate @xmath116 , we approximate the vortex loops by rings of radius @xmath117 , whose characteristic lifetime is given by @xmath118 , with @xmath119 @xcite . the radius of a ring can be estimated from the vortex line density @xmath120 ( length per unit volume ) , with @xmath121 in a steady - state tangle . the evolution of @xmath120 is given by vinen s equation @xcite , generalized by @xcite to include rotation : @xmath122 x^{3/2}\ ] ] @xmath123 x,\ ] ] with @xmath124 , where @xmath125 , @xmath126 are dimensionless friction coefficients , whose values depend on the temperature @xcite . by fitting to experimental data , one finds @xmath127 , @xmath128 , @xmath129 , and @xmath130 @xcite . from experiments , we also have @xmath131 and @xmath132 @xcite , so we take @xmath133 below . setting @xmath134 in ( [ eq : vinen_jou ] ) and solving for @xmath120 in the steady state , we find that the tangle decays on a time - scale @xmath135 where @xmath136 is the counterflow speed immediately before the differential rotation shuts off , given by ( [ eq : vns_ep ] ) . estimates for the friction parameter @xmath137 are uncertain @xcite , but calculations based on electron scattering by proton vortex clusters and magnetized vortex cores give @xmath138 @xcite . , calculated from electron scattering off vortex clusters in the high - density regime , where the relaxation time - scale equals the age of the pulsar [ @xcite ; a. sedrakian 2006 , private communication ] . ] this implies @xmath139 for @xmath140 in the outer core @xcite . therefore @xmath116 is greater than the glitch trigger time - scale , as assumed a priori . note that the @xmath38 and @xmath141 parameters were also calculated by @xcite and @xcite , who found @xmath142 and @xmath143 where @xmath144 is the proton fraction , and @xmath145 and @xmath146 are the bare and effective masses of the proton respectively @xcite . equation ( [ eq : bbprime1 ] ) includes modifications from entrainment @xcite . in the outer core , with @xmath147 g @xmath148 and @xmath149 @xcite , we obtain @xmath150 , approximately the value quoted by @xcite and @xcite . in laboratory and numerical experiments , it is observed that the vortex tangle is polarized ; the average vorticity projected along the rotation axis is not zero @xcite . this is because , under these experimental conditions , one has @xmath151 in equation ( [ eq : vinen_jou ] ) , where @xmath152 is the polarization - inducing term . naively , one might expect the tangle to be even more polarized in a rapidly rotating neutron star , but in fact the opposite is true : the tangle is less polarized , because we have @xmath153 from figure [ fig : fig2 ] . a thorough study of this issue , including whether the large-@xmath50 limit considered by @xcite in deriving equation ( [ eq : vinen_jou ] ) is applicable for @xmath154 , lies outside the scope of this paper . after the tangle decays to a rectilinear array , differential rotation due to electromagnetic spin down between the inner and outer shells begins to grow until @xmath92 exceeds @xmath155 . from equations ( [ eq : dgi ] ) and ( [ eq : vns_ep ] ) , this occurs after a time @xmath156 whereupon a vortex tangle develops again in the outer core , via the dgi . this triggers a switch from hv to gm mutual friction , weakening the coupling between the normal fluid and superfluid components . the characteristic time - scale for the tangle to develop equals the growth rate of helical vortex perturbations ( kelvin waves ) via the dgi . by linearizing schwarz s equation @xcite , in the vortex filament model , one finds that the fastest growth occurs at a wavenumber @xmath157 , with @xcite @xmath158 where ( [ eq : growth_tangle2 ] ) follows from ( [ eq : growth_tangle1 ] ) by substituting ( [ eq : dgi ] ) . hence , for typical neutron star parameters , the tangle is reestablished over @xmath159 d after @xmath159 yr elapses following a glitch , less than the inter - glitch interval observed in most pulsars @xcite . note that @xmath160 does not equal @xmath116 ; the time required for a tangle to grow from a rectilinear array is shorter than the time required for an existing tangle to decay back to a rectilinear array , provided that @xmath161 ( and longer otherwise ) . note also that @xmath136 just before the glitch ( @xmath162 ) , which appears in ( [ eq : decay_tangle1 ] ) , is typically greater than @xmath136 at @xmath163 ( @xmath164 ) , which appears in ( [ eq : growth_tangle1 ] ) , provided that @xmath165 . equation ( [ eq : growth_tangle2 ] ) gives the minimum time for a tangle to reform , assuming it does so simultaneously everywhere in the outer core of the star . in practice , @xmath136 does not exceed @xmath166 everywhere simultaneously . regions where the dgi is activated are interspersed with regions where it is not , even at @xmath167 and @xmath48 , so that the transition from hv to gm friction is more gradual than equation ( [ eq : growth_tangle2 ] ) suggests . this important issue , which is also relevant to the interpretation of the laboratory experiments performed by @xcite , is explored briefly in [ sec : summary ] . however , a thorough numerical study lies outside the scope of this paper . in this section , we calculate numerically the torque acting on the stellar crust during a transition from a state of turbulent vorticity ( tangle ) to laminar vorticity ( rectilinear array ) , by solving the hydrodynamic hvbk equations in a differentially rotating spherical shell . global simulations of this type are necessary to calculate the observable response of a neutron star , e.g. @xmath168 and @xmath169 , to the internal physics elaborated in [ sec : vtmodel ] . we solve equations ( [ eq : supeq1 ] ) and ( [ eq : supeq2 ] ) using a pseudospectral collocation method to discretize the problem @xcite and a time - split algorithm to step forward in time @xcite , as described by @xcite . the velocity fields are expanded in a restricted fourier series in @xmath170 and @xmath171 and a chebyshev series in @xmath172 @xcite , with @xmath173 modes required to fully resolve the flow . the initial time - step @xmath174 is lowered to @xmath175 during the gm @xmath176 hv transition to prevent spurious oscillations in the torque @xcite . instabilities arising from the sensitivity of spectral methods to the boundary conditions @xcite , and oscillations due to the gibbs phenomenon @xcite , are smoothed using a low - pass spectral filter @xcite , a common practice @xcite . we adopt parameters as close to those of a realistic neutron star as our computational resources permit . in the outer core , we take @xmath177 and @xmath178 @xcite , and hence @xmath179 , @xmath180 , and @xmath181 at the corresponding scaled temperature @xmath182 in he ii @xcite . these friction coefficients are @xmath183 times greater than those used in the analytic estimates in section [ sec : vtmodel ] . we adopt the higher values deliberately . otherwise , the effects introduced by @xmath38 and @xmath141 would take too long to build up ; computational limitations prevent us from simulating more than @xmath184 rotation periods . ( for the same reason , our @xmath185 is unrealistically large . ) realistic reynolds numbers , estimated from equation ( [ eq : rey ] ) , are too high to be simulated directly , so we restrict ourselves to the range @xmath186 ; the most stable evolution occurs for @xmath187 , corresponding to an ekman number @xmath188 that is two orders of magnitude smaller than in a typical neutron star @xcite . the tension force is dominated by mutual friction , except in relatively old ( @xmath189 yr ) neutron stars @xcite , with @xmath190 in our dimensionless units . for @xmath191 , we find empirically that the torque depends weakly on the gm force , so we use @xmath192 to obtain identical results at lower computational cost . we consider a turbulent initial state with @xmath193 and @xmath194 everywhere , such that the normal and superfluid components are coupled by gm mutual friction . the transition from a vortex tangle to a rectilinear vortex array is simulated by changing the mutual friction suddenly , from gm to hv , and simultaneously spinning up the outer shell , such that @xmath195 . this occurs at @xmath196 in all the figures to follow . the angular shear before the transition lies in the range @xmath197 , for gaps in the range @xmath198 . note that we are forced to choose @xmath199 unrealistically large , compared to typical neutron star values , in order to allow the inner and outer surfaces of the shell to lap " each other within a reasonable run time .. ] various combinations of @xmath199 and @xmath200 affect the viscous torque on the inner and outer surfaces during steady differential rotation in different ways , as discussed below . streamline snapshots for a run with @xmath201 and @xmath202 are presented in figures [ fig : fig3]a[fig : fig3]f , with the turbulent - laminar transition gm @xmath176 hv , @xmath203 occurring at @xmath196 . the hv mutual friction , with @xmath204 , couples the normal and superfluid components strongly . is supported by order - of - magnitude estimates @xcite but should be treated with caution , because the microphysics of the gm friction force ( and hence the value of @xmath205 ) has not yet been worked out rigorously in a neutron superfluid . ] both components exhibit a quasi - periodic motion , with a secondary circulation cell filling most of the shell , and two or three smaller vortices emerging intermittently . an oscillatory polar ekman cell also exists initially , disappearing at @xmath206 . we can compare this behaviour to the flow without a turbulent - laminar transition . figures [ fig : fig4]a[fig : fig4]f display a sequence of streamline snapshots for @xmath207 and @xmath208 , with gm mutual friction at all times . at @xmath209 , three polar cells are observed in both components , together with a secondary cell at mid - latitudes . the latter cell elongates after @xmath196 and is present only in the superfluid component at later times . at the equator , the flow switches between one and three cells , but this behavior persists after @xmath210 only in the superfluid component , while the normal fluid decouples increasingly . the axial counterflow is significant at all times in figures [ fig : fig4 ] and [ fig : fig5 ] . scaling to neutron star parameters , we find @xmath211 near the equator , and @xmath212 near the poles , at @xmath213 . figures [ fig : fig5]a[fig : fig5]d display @xmath214 as a function of time after the turbulent - laminar transition for @xmath215 , @xmath216 , @xmath93 , and @xmath94 , respectively . is not calculated self - consistently : it is obtained by numerically integrating the viscous torque on the outer sphere , after solving ( [ eq : supeq1 ] ) and ( [ eq : supeq2 ] ) subject to the boundary condition @xmath217 , i.e. we do not include the ( slight ) evolution of @xmath218 when computing the fluid flow . ] the evolution is qualitatively similar in all the cases considered . there is an initial transient , in which the flow adjusts to the initial spin up on a time - scale @xmath219 , just like in the classical ( viscous ) spin - up problem @xcite . the fractional torque deviation , @xmath220 , is plotted as a function of time in figures [ fig : fig6]a[fig : fig6]d . an initial jump , coinciding with the sudden spin up , is observed in all the panels , followed by a rapid exponential decay and persistent , small - amplitude oscillations . we find an @xmath221 relaxation time of @xmath222 , @xmath223 , and @xmath223 for @xmath76 , @xmath224 , and @xmath216 respectively , with @xmath225 essentially independent of @xmath200 . in contrast , in the inverse experiment , where the mutual friction changes from hv to gm , @xcite observed that @xmath226 is almost constant after the post - glitch transient . the period of the oscillations does not change with @xmath87 for @xmath227 . the effect of the superfluid on the evolution can be analyzed with the help of figure [ fig : fig8]a , which plots the torque on the outer sphere as a function of time ( solid curve ) in the absence of spin up ( @xmath207 , @xmath74 , @xmath76 , gm friction ) . for comparison , the dashed curve plots the torque for a classical viscous fluid with the same parameters mentioned above . the evolution is qualitatively similar in both cases , displaying oscillations with the same periodicity and similar amplitude , but the curves diverge for @xmath228 as the viscous fluid evolves more rapidly to a stationary state , driven by classical ekman pumping . classical theory predicts a time - scale @xmath229 in the limit where the rossby number @xmath185 vanishes ; an exponential fit to the dashed curve gives @xmath230 , which agrees well given that we have @xmath231 in our numerical experiments . in a superfluid , on the other hand , the torque decays faster initially , with time constant @xmath232 , then starts to increase again for @xmath233 , as part of a long - period oscillation whose ultimate fate is unknown as it occurs over a time - scale beyond our computational limit . figure [ fig : fig8]b shows the evolution of the fractional change in angular velocity after spin up ( @xmath234 at @xmath196 ) in three scenarios : instantaneous change in mutual friction , gm @xmath176 hv ( solid curve ) ; constant gm friction ( dashed curve ) ; and a classical viscous fluid ( dashed dotted curve ) . the classical viscous fluid and constant gm friction evolve similarly . the torque decays exponentially on a time - scale @xmath235 , then decays almost linearly with @xmath87 . by contrast , in the gm @xmath176 hv transition , the torque decays exponentially , on a time - scale @xmath236 , then oscillates with peak - to - peak amplitude @xmath237 ( in units of @xmath238 ) and period @xmath239 . in other words , the oscillations are sustained by the hv friction , which is much weaker than its gm counterpart . note that the oscillations do not correspond to tkachenko waves , since the hvbk equations assume that the free energy of the vortex array depends only on the vortex line density , not on vortex lattice deformations @xcite . the meridional streamlines for the transitions described in the previous paragraph evolve as in figure [ fig : figc ] . figures [ fig : figc]a[fig : figc]b show the streamlines for a classic viscous fluid before ( @xmath196 ) and after ( @xmath240 ) the glitch . there is a rapid redistribution of vorticity near the outer sphere , with two circulation cells replaced by one elongated meridional shell , while the flow near the inner sphere hardly changes , with one circulation cell near the equator and one near the pole . figures [ fig : figc]c[fig : figc]d show the meridional streamlines for the normal component of the superfluid . the flow resembles a classical viscous fluid ( cf . figure [ fig : figc]a ) , although it does become more complex after the spin up . as shown in figure [ fig : figc]d , the large cell near the outer sphere is replaced by one primary circulation cell and three secondary cells ( including one near the poles ) . note that the greatest contributions to the torque come from the mid - latitude regions where the structure of eddies and counter - eddies is richer , as in figure [ fig : figc]d . finally , in figures [ fig : figc]e[fig : figc]f , we see the meridional streamlines for the inviscid component of the superfluid . the pattern before the jump ( figure [ fig : figc]e ) differs from the normal fluid ; the gm friction creates additional eddies near the outer sphere and closer to the poles . however , after the transition to hv friction , the inviscid component comes to resemble the normal component , as can be seen by comparing figure [ fig : figc]f with figure [ fig : figc]d . this occurs because the two components are coupled more strongly by hv friction ( @xmath204 ) , so that the superfluid is dragged along by the normal fluid . gravitational stratification can strongly suppress ekman pumping , as the ekman layer is squashed close to the outer sphere @xcite . nevertheless , the ekman layer , however thin , always exists , in order to satisfy the boundary conditions at @xmath241 . it contains a meridional counterflow given by equation ( [ eq : vns_ep ] ) . consequently , the transition from turbulent to laminar vorticity ( and vice versa ) , which relies on such a meridional counterflow , still occurs in a stratified star , and its dramatic effect on the torque is the same . the ekman layer , no matter how thin , acts like a film of oil " between two sliding surfaces ( here , the outer core and inner crust ) , whose coupling strength ( stickiness " ) changes abruptly when the meridional speed of the oil " exceeds a threshold . in other words , equations ( [ eq : vns_ep ] ) , ( [ eq : decay_tangle2 ] ) , ( [ eq : reform ] ) , and ( [ eq : growth_tangle2 ] ) , and the conclusions that follow from them , are unaltered by stratification ; even though the volume of the outer core occupied by the vortex tangle ( which does not affect the torque on the crust directly ) does change . the ekman layer thickness decreases as @xmath242 , where @xmath243 is the thickness without stratification and @xmath244 is the compressibility of the fluid . the ekman time - scale increases as @xmath245 , where @xmath246 is the time - scale without stratification . the fluid is restricted to move on concentric spherical shells @xcite . importantly , @xmath245 is the time for the ekman layer to extend throughout the outer core , not the time required to establish the meridional flow at @xmath247 ( @xmath248 ) , which controls the onset of the dgi . it is conceivable that the dgi is excited in shells at certain radii where @xmath136 peaks , so that we end up with a sequence of shells containing alternating laminar and turbulent superfluid vorticity . in this scenario , the detailed study of which lies outside the scope of this paper , the stability of the vorticity configuration is restricted by the richardson criterion , which states that the configuration is kelvin - helmholtz unstable for @xmath249 @xcite . in a neutron star , the brunt - visla frequency is @xmath250 @xcite . the effects of stratification are not considered in detail in this paper because they can not be studied properly with our numerical method ; the pressure projection step only works with divergence - free velocity fields @xcite . however , a crude approach to get a feel for the effects is to numerically suppress @xmath251 using a low - pass exponential filter @xcite , viz . @xmath252 v_r $ ] , with @xmath253 , where @xmath254 is the machine zero , and @xmath255 is the order of the filter . if we ramp up @xmath255 with @xmath172 as @xmath256 , with @xmath257 and @xmath258 , then @xmath251 is weakly suppressed near @xmath48 ( @xmath259 ) and strongly suppressed near @xmath47 ( @xmath260 ) , so that the flow is confined approximately to concentric shells . preliminary results , for a viscous navier - stokes fluid , show that the filtering reduces the meridional circulation , flattening the streamlines radially . in this paper , we investigate how transitions between turbulent and laminar states of superfluid vorticity alter the standard theoretical picture of pulsar rotational irregularities like glitches and timing noise . ( i ) most of the time , except in the immediate aftermath of a glitch , differential rotation in the outer core drives a nonzero , poloidal counterflow which continuously excites the dgi . a vortex tangle is thereby maintained in the outer core . the mutual friction in this regime , which is of gm form , couples the normal and superfluid components loosely . ( ii ) immediately after a glitch , the differential rotation ceases , as does the poloidal counterflow . the vortex tangle decays over the mean life - time of its constituent vortex rings , @xmath261 s. a rectilinear vortex array develops , and the mutual friction switches to hv form , coupling the normal and superfluid components much more strongly . ( iii ) after @xmath262 yr , electromagnetic spin down builds up the differential rotation sufficiently to drive a poloidal counterflow that exceeds the dgi threshold . a vortex tangle forms again in a time @xmath263 s , and the mutual friction reverts to gm form . note that vortex pinning provides the boundary conditions for the superfluid scf but does not occur within the outer core itself @xcite . _ therefore our new phenomenological picture is not a complete model for glitches_. it merely clarifies the vorticity state of the outer core before and after a glitch as an input into future models than incorporate the full glitch dynamics , including trigger mechanisms related to pinning in the inner crust . we draw together the strands of the model in figure [ fig : fig9 ] , which displays the evolution of the torque and the regions where the dgi is active during the following numerical experiment : we fix @xmath76 until @xmath196 , accelerate the outer sphere instantaneously to corotation at @xmath196 , then decelerate the outer sphere according to @xmath264 for @xmath265 . this mimics the situation in a real pulsar , where we have @xmath266 , i.e. ekman pumping brings the fluid into corotation _ before _ the dgi gradually starts being reexcited throughout the outer core . to make the experiment as realistic as possible , we do not assume that the mutual friction takes the same form everywhere in the outer core , but rather choose gm or hv friction at each point according to whether @xmath267 is greater or less than @xmath166 locally . in this comparison , we approximate @xmath268 by @xmath269 , as in equation ( [ eq : vns_ep ] ) , because @xmath270 can become very complicated ( e.g. figure [ fig : fig1 ] ) , creating numerical difficulties . in order to satisfy @xmath266 while keeping @xmath199 large enough so that the spheres lap " each other several times , we are forced by computational exigencies to adopt a relatively low reynolds number @xmath271 , shortening the ekman time ( @xmath272 ) , and to artificially boost @xmath166 , so that @xmath273 does not exceed @xmath274 throughout the computational domain . the results of the above numerical experiment are presented in figure [ fig : fig9 ] . contours of @xmath273 , before and after the spin up at @xmath196 , are plotted in figures [ fig : fig9]a[fig : fig9]e . shaded regions indicate where the dgi is active , i.e. @xmath275 . just before the glitch ( figure [ fig : fig9]a ) , @xmath276 % of the superfluid is in a turbulent state , with the dgi active close to the inner sphere and at intermediate latitudes where meridional circulation is significant . after the differential rotation shuts off at @xmath196 , the dgi initially spreads through the shell as transient axial flows increase , occupying @xmath277 % of the volume at @xmath240 . however , the flow quickly settles down to a state of near - corotation during the time interval @xmath278 , hv friction dominates , and the torque decays exponentially , with time constant @xmath184 ( followed by a linear decay ) . at @xmath279 , the dgi slowly begins to assert itself again , starting from the inner sphere . as for a classical viscous fluid , when the outer sphere spins down , it pumps fluid radially inward and along the axis of rotation @xcite , so the axial flow speed is greatest near the inner sphere . by @xmath280 , when @xmath76 , the vorticity state is similar to that at @xmath196 , before spin up . one might wonder if , in a realistic neutron star , the superfluid ever exits the turbulent state and becomes laminar . for the simulations in this paper , which have @xmath281 , the answer is yes . figure [ fig : fig9 ] shows that , when the glitch occurs and the spheres come into corotation , @xmath282 falls below the dgi threshold after a time @xmath283 . the vortex tangle is then guaranteed to decay on a time - scale given by ( [ eq : decay_tangle1 ] ) and ( [ eq : decay_tangle2 ] ) , as observed in terrestrial experiments . however , for more realistic neutron star reynolds numbers ( @xmath284 ) , which are too challenging to simulate at present , the turbulent eddies in the normal fluid decay more slowly when the spheres come into corotation , so that @xmath282 remains above the dgi threshold for longer . if this happens , the vortex tangle may persist until the next glitch occurs , so that the superfluid never exits the turbulent state . what are the implications of figure [ fig : fig9 ] and the results in [ sec : vtmodel ] and [ sec : rotevol ] for observations of glitches ? before considering this question , we emphasize again that the results in this paper do not constitute a theory of glitches , because important questions regarding the glitch trigger remain unresolved . nevertheless , some general remarks can be made . first of all , it is clear that transitions between flow ( and vorticity ) states in superfluid scf are caused by changes in @xmath64 , and that such transitions become more frequent and complicated as @xmath64 increases @xcite . this is compatible with the observation that adolescent pulsars ( @xmath183 yr old , like vela ) glitch most actively @xcite . in younger pulsars ( age @xmath285 yr ) , @xmath101 and hence @xmath13 are relatively high , so @xmath64 is low . in older pulsars ( age @xmath286 yr ) , @xmath287 and hence @xmath64 are low following electromagnetic spin down [ although this trend is not straightforward and can be masked by localized heating from differential rotation between the superfluid and the crust @xcite or crust cracking @xcite ] . a systematic statistical study of glitch activity versus @xmath64 will be published elsewhere @xcite , but preliminary estimates of @xmath101 and hence @xmath64 from cooling curves @xcite including superfluidity @xcite give reynolds numbers in the range @xmath288 for glitching pulsars . two of the most active glitchers , the crab and vela , have @xmath289 and @xmath290 respectively . one expects that , at such high @xmath64 , the fluid is turbulent , with the kinetic energy concentrated at large scales @xcite , as for a classical viscous fluid @xcite . this suggests that superfluid turbulence in pulsar interiors is an important factor in glitch dynamics . if it is true that the vorticity in the outer core exists in a turbulent state before a glitch , as postulated in our model , then @xmath163 represents a lower bound on the time between glitches . in testing whether this bound is respected by the glitching pulsars currently known , we are hampered by the fact that most of these objects have only glitched once . nevertheless , for all the @xmath291 pulsars that have glitched repeteadly , we find that the minimum inter - glitch time interval @xmath292 is greater than @xmath163 , as the theory predicts @xcite . the object psr @xmath293 approaches the bound most closely , with @xmath294 yr and @xmath295 yr . this is encouraging , because the @xmath291 objects cover five decades in @xmath163 and three decades in @xmath292 , and the theoretical expression ( [ eq : reform ] ) for @xmath163 contains zero free parameters . note that @xmath163 is proportional to the characteristic age ( @xmath296 ) divided by @xmath297 . note also that the activity parameter defined by @xcite involves glitch amplitudes ( which are highly variable ) as well as mean recurrence times , so we do not predict a correlation between the activity parameter and @xmath163 . it is harder to test the theoretical decay time - scale of the vortex tangle , as predicted by ( [ eq : decay_tangle2 ] ) , because it remains unclear what observable features are engendered by the decay process . the observed exponential post - glitch relaxation is of viscous origin and occurs on a time - scale much larger than @xmath116 . on the other hand , the decay of the tangle is accompanied by a large increase in mutual friction ( gm @xmath176 hv ) , which may be connected with the rapid jump in @xmath287 during a glitch . the jump in @xmath287 has never been resolved in time , in pulsars which are nearly constantly monitored , consistent with the predictions of ( [ eq : decay_tangle2 ] ) for the crab ( @xmath298 s ) and vela ( @xmath299 s ) . however , equation ( [ eq : decay_tangle2 ] ) predicts that it may be possible to resolve the @xmath287 jump in older pulsars , provided that the time between glitches does not increase faster than @xmath85 . in making these estimates we assume the canonical value @xmath300 for every object , in the absence of a microscopic theory , yet this is clearly an oversimplification because @xmath137 is sensitively temperature dependent . oscillations in @xmath85 were observed before ( period @xmath184 d ) and after ( period @xmath301 d ) the vela christmas glitch , with @xmath302 @xcite . in our numerical simulations , persistent torque oscillations are always observed when the outer core rotates differentially , as occurs before a glitch . they are also observed after a switch from gm to hv friction , as occurs after a glitch . by comparing the dashed and solid curves in figure [ fig : fig8]b , we see that the oscillations are sustained by hv mutual friction . the oscillation period in our simulations is much shorter than in pulsar data , because we are restricted to @xmath303 . an alternative explanation is that vortices in the inner crust oscillate relative to the normal fluid in the core @xcite . several of the effects explored in this paper have been studied in terrestrial laboratories . our results will motivate new experiments of this sort , cf . @xcite and @xcite . although it is hard to access the neutron star regime @xmath304 in he ii , where interatomic forces are appreciable , transitions between turbulent and laminar superfluid vorticiy have been observed in experiments with microspheres immersed in @xmath10he at mk temperatures @xcite . promising results on the relaxation of rotating he ii were obtained by @xcite , but again these results are for @xmath305 and hollow spheres rather than a differentially rotating shell . we propose to extend these experiments in two directions : ( i ) by investigating low - rossby - number ( @xmath306 ) , high - reynolds - number ( @xmath307 ) scf with he ii at the temperature which minimizes @xmath308 ; and ( ii ) by repeating ( i ) with a nonideal dilute - gas bose - einstein condensate confined in a differentially rotating magneto - optical trap , in order to probe the stability of a vortex lattice to kelvin wave excitations in the regime @xmath304 @xcite . the presence of a vortex tangle in he ii can be detected by standard second - sound absorption techniques @xcite , and the torque in experiment ( i ) can be monitored to look for oscillations when a change from hv to gm friction ( or vice versa ) is triggered by the dgi . in classical navier - stokes fluids , injection of vorticity into a metastable laminar state can trigger turbulence , e.g. seed vortices injected into a cylindrical vessel containing @xmath9he - b ( with @xmath309 ) generate a vortex tangle that eventually decays into a rectilinear vortex array @xcite . unlike he ii , the normal component in @xmath9he - b is laminar in these experiments and does not participate in the turbulent dynamics , due to its comparatively high viscosity [ @xmath310 @xmath311 s@xmath111 @xmath312 , cf . @xmath313 in he ii ; see @xcite ] . standard glitch theories assume that the normal component is tightly coupled to the crust by the external magnetic field @xcite . this suggest a second possible scf experiment with @xmath9he - b ( at @xmath309 ) in a hollow spherical container in which the normal fluid corotates with the container , and therefore does not participate in the dgi , while the superfluid is free to become turbulent . such an experiment can probe what aspects of the turbulent - laminar transition are caused by the normal and superfluid components respectively . transitions between a vortex tangle and a rectilinear array can be detected using non - invasive nuclear magnetic resonance techniques @xcite . we gratefully acknowledge the computer time supplied by the australian partnership for advanced computation ( apac ) and the victorian partnership for advanced computation ( vpac ) . we also thank s. balachandar , from the university of illinois at urbana - champaign , for supplying us with his original pseudo - spectral solver , designed for a single navier - stokes fluid , from which our two - fluid , hvbk solver was developed . this research was supported by a postgraduate scholarship from the university of melbourne .	we investigate the global transition from a turbulent state of superfluid vorticity ( quasi - isotropic vortex tangle ) to a laminar state ( rectilinear vortex array ) , and vice versa , in the outer core of a neutron star . by solving numerically the hydrodynamic hall - vinen - bekarevich - khalatnikov equations for a rotating superfluid in a differentially rotating spherical shell , we find that the meridional counterflow driven by ekman pumping exceeds the donnelly - glaberson threshold throughout most of the outer core , exciting unstable kelvin waves which disrupt the rectilinear vortex array , creating a vortex tangle . in the turbulent state , the torque exerted on the crust oscillates , and the crust - core coupling is weaker than in the laminar state . this leads to a new scenario for the rotational glitches observed in radio pulsars : a vortex tangle is sustained in the differentially rotating outer core by the meridional counterflow , a sudden spin - up event ( triggered by an unknown process ) brings the crust and core into corotation , the vortex tangle relaxes back to a rectilinear vortex array ( in @xmath0 s ) , then the crust spins down electromagnetically until enough meridional counterflow builds up ( after @xmath1 yr ) to reform a vortex tangle . the turbulent - laminar transition can occur uniformly or in patches ; the associated time - scales are estimated from vortex filament theory . we calculate numerically the global structure of the flow with and without an inviscid superfluid component , for hall - vinen ( laminar ) and gorter - mellink ( turbulent ) forms of the mutual friction . we also calculate the post - glitch evolution of the angular velocity of the crust and its time derivative , and compare the results with radio pulse timing data , predicting a correlation between glitch activity and reynolds number . terrestrial laboratory experiments are proposed to test some of these ideas .
there are two versions of the asymptotic plateau problem . the first version asks the existence of a least area plane @xmath7 in @xmath4 asymptotic to a given simple closed @xmath0 in @xmath1 , i.e. @xmath8 . in this version , there is a topological restriction on the surface @xmath7 to be in the type of disk . the other version asks the existence of an area minimizing surface @xmath9 in @xmath4 asymptotic to a given collection of jordan curves @xmath10 in @xmath1 , i.e. @xmath11 . in the latter version , there is no a priori topological restriction on the surface @xmath9 , hence @xmath9 can have positive genus depending on given @xmath10 . anderson gave positive answers to both questions 3 decades ago @xcite . constant mean curvature ( cmc ) surfaces are natural generalizations of minimal surfaces , and in many cases , the results related to minimal surfaces are studied to see whether they can be generalized to cmc setting . in our case , we will call this natural generalization as _ asymptotic @xmath2-plateau problem_. a decade after anderson s result , the second version of the asymptotic plateau problem was generalized to cmc case by tonegawa @xcite . tonegawa showed that for any given collection of jordan curves @xmath10 in @xmath1 , there exists a minimizing @xmath2-surface @xmath12 in @xmath4 with @xmath13 where @xmath14 . indeed , both anderson and tonegawa used geometric measure theory methods , and the solutions are automatically smoothly embedded surfaces by the regularity results of gmt . the survey @xcite gives a fairly complete account of the old and new results on the problem . on the other hand , for the generalization of the first ( plane ) version to the cmc case , the only result came out a few years ago by cuschieri @xcite . he showed the existence of _ immersed _ @xmath2-planes asymptotic to given smooth jordan curve in @xmath1 by using pde techniques . in this paper , we give positive answer to the asymptotic @xmath2-plateau problem for a larger family of curves . furthermore , we show that these solutions are indeed _ embedded_. let @xmath0 be a simple closed curve @xmath0 in @xmath1 with at least one @xmath15-smooth point . then , for any @xmath14 , there exists a properly embedded @xmath2-plane @xmath3 in @xmath4 with @xmath16 . our techniques are also valid for @xmath17 case , and we are able to reprove the existence of least area planes in @xmath4 . hence , with this result , we filled the gap in anderson s result in @xcite , too ( see remark [ gap ] ) . on the other hand , our proof is indeed for @xmath18 by considering the orientation . hence , if you forget the sign of the mean curvature , the above theorem shows that for a given jordan curve @xmath0 in @xmath1 and @xmath19 , there exist _ a pair of _ complete , embedded @xmath2-planes @xmath20 and @xmath21 with @xmath22 ( see remark [ pair ] ) . recently , meeks , tinaglia and the author constructed nonproperly embedded @xmath2-planes in @xmath4 for any @xmath14 @xcite , where the asymptotic boundary is a pair of infinite lines in @xmath1 . here , we also show that if the jordan curve @xmath0 in @xmath1 is smooth enough , then the minimizing @xmath2-planes @xmath3 with @xmath16 are _ properly embedded _ in @xmath4 ( corollary [ proper ] ) . in the final section , we discuss the questions of the _ generic uniqueness _ of the @xmath2-planes in @xmath4 , and the _ foliations of _ @xmath4 with @xmath2-planes , and give an outline to solve them . the organization of the paper is as follows : in the next section , we go over the basic notions , and the related results . in section 3 , we prove the main theorem for least area planes ( @xmath17 case ) . in section 4 , we show the existence of embedded @xmath2-planes in @xmath4 . finally in section 5 , we give some concluding remarks . in this section , we will go over the related results which will be used in the following sections . for further details of the notions , and the results used in this paper , one can see the survey @xcite . let @xmath23 be a @xmath24-manifold . * _ minimal surface : _ a surface @xmath25 in @xmath23 is minimal if the mean curvature vanishes everywhere on @xmath25 . * _ least area disk : _ a compact disk @xmath26 in @xmath23 with @xmath27 is the least area disk in @xmath23 if it has the smallest area among the disks in @xmath23 with the boundary @xmath0 . note that minimal surfaces are the critical points of the area functional . least area disks and area minimizing surfaces are the minimum of the area functional in the corresponding spaces . a natural generalization of minimal surfaces are cmc surfaces ( @xmath2-surfaces ) . they can be defined as the critical points of the area functional with a volume constraint as follows . for an immersion @xmath28 , the critical points of the following variational problem are immersed disks with constant mean curvature @xmath2 @xcite . @xmath29 \ dxdy\ ] ] here , the second summand in the integral represents the volume constraint . we can reformulate this variational problem in a different way so that it will be independent of the parametrization of the surface @xcite . let @xmath30 be a surface in @xmath23 with boundary @xmath31 . we fix a surface @xmath32 in @xmath23 with @xmath33 , and define @xmath34 to be the domain bounded by @xmath32 and @xmath30 . again , let @xmath35 if @xmath30 is a critical point of the functional @xmath36 for any variation @xmath37 , then this will imply @xmath30 has constant mean curvature @xmath2 . note also that critical point of the functional @xmath36 is independent of the choice of the surface @xmath32 since if @xmath38 is the functional which is defined with a different surface @xmath39 , then @xmath40 for some constant @xmath41 . note that to keep the solution surface away from @xmath32 , one needs a convexity condition on @xmath32 ( e.g. @xmath42-convex for @xmath43 ) to employ the maximum principle @xcite . let @xmath23 be a @xmath24-manifold . * _ @xmath2-surface : _ a surface @xmath25 in @xmath23 is @xmath2-surface if the mean curvature is equal to @xmath2 everywhere on @xmath25 . * _ minimizing @xmath2-disk : _ a compact disk @xmath26 in @xmath23 with @xmath27 is a minimizing @xmath2-disk in @xmath23 if @xmath44 ( or equivalently @xmath45 ) has the smallest value among the disks in @xmath23 with the boundary @xmath0 . here , we will quote the generalization of meeks - yau s embeddedness result @xcite to @xmath2-disks . [ @xmath42-convex domains ] let @xmath34 be a compact @xmath24-manifold with piecewise smooth boundary . we call @xmath34 an _ @xmath42-convex domain _ if * the mean curvature vector @xmath46 always points towards inside @xmath34 along the smooth parts of @xmath47 , * the mean curvature @xmath48 for any smooth point @xmath49 , * along the nonsmooth parts of @xmath47 , the inner dihedral angle is less than @xmath50 . the following lemma is on the embeddedness of the solutions of @xmath2-plateau problem for @xmath2-extreme curves . [ embed ] @xcite let @xmath23 be a compact @xmath42-convex ball . let @xmath0 be a jordan curve in @xmath51 . then , for any @xmath52 , there exists a minimizing @xmath2-disk @xmath53 in @xmath23 with @xmath54 , and any such @xmath53 is embedded . the following lemma is know as maximum principle . @xcite [ maximum ] [ maximum principle ] let @xmath55 and @xmath56 be two surfaces in a riemannian @xmath24-manifold which intersect at a common point tangentially . if @xmath56 lies in positive side ( mean curvature vector direction ) of @xmath55 around the common point , then @xmath57 is strictly less than @xmath58 ( @xmath59 ) where @xmath60 is the mean curvature of @xmath61 at the common point . now , we restrict ourselves to @xmath4 . let @xmath0 be a simple closed curve in @xmath1 . @xmath0 separates @xmath1 into two open disks , say @xmath62 and @xmath63 . we fix an orientation , and consider the mean curvature @xmath2 with sign depending on the direction of the mean curvature vector , i.e. @xmath18 such that @xmath64 and @xmath65 @xcite . now , by using the definitions above , we define the least area planes and the minimizing @xmath2-planes in @xmath4 . [ least area plane ] let @xmath7 be a complete surface in @xmath4 which is topologically in the type of a disk . we call @xmath7 a least area plane in @xmath4 , if any compact subdisk @xmath26 in @xmath7 is a least area disk . [ minimizing @xmath2-plane ] fix @xmath66 . let @xmath3 be a complete surface in @xmath4 which is topologically in the type of a disk . we call @xmath3 a minimizing @xmath2-plane in @xmath4 , if any compact subdisk @xmath26 in @xmath3 is a minimizing @xmath2-disk . for a given surface @xmath25 in @xmath4 , we define the _ asymptotic boundary _ of @xmath25 as follows . if @xmath67 is the natural ( geodesic ) compactification of @xmath4 , and @xmath68 is the closure of @xmath25 in @xmath69 , then the asymptotic boundary @xmath70 of @xmath25 defined as @xmath71 . now , we define the shifted convex hulls @xmath72 as generalizations of the convex hulls in @xmath4 @xcite . fix @xmath66 . let @xmath0 be a jordan curve in @xmath1 . let @xmath31 be a a round circle in @xmath1 with @xmath73 . let @xmath74 be the unique @xmath2-plane in @xmath4 with @xmath75 . @xmath31 separates @xmath1 into two open disks @xmath76 and @xmath77 . similarly , @xmath74 divides @xmath4 into two domains @xmath78 and @xmath79 where @xmath80 . we will call these regions as _ @xmath2-shifted halfspaces_. if @xmath81 , then we will call @xmath82 a _ supporting @xmath2-shifted halfspace_. similarly , if @xmath83 , then we will call @xmath84 a _ supporting @xmath2-shifted halfspace_. [ shifted convex hull ] let @xmath0 be a simple closed curve in @xmath1 . fix @xmath66 . then the _ @xmath2-shifted convex hull _ of @xmath0 , @xmath72 is defined as the intersection of all supporting closed @xmath2-shifted halfspaces @xmath85 of @xmath4 . for @xmath17 , this is the usual convex hull definition in @xmath4 , i.e. @xmath86 . now , the generalization of convex hull property of minimal surfaces in @xmath4 to @xmath2-surfaces in @xmath4 is as follows @xcite . [ convex ] @xcite , @xcite let @xmath30 be a @xmath2-surface in @xmath4 where @xmath87 and @xmath88 . then @xmath30 is in the @xmath2-shifted convex hull of @xmath0 , i.e. @xmath89 . [ convex2 ] notice that the result is true for any @xmath2-surface . this is a straightforward generalization of the convex hull property for minimal surfaces . in particular , if @xmath30 is an @xmath2-surface in @xmath4 with @xmath90 , then @xmath30 can not go into a nonsupporting @xmath2-shifted halfspaces of @xmath0 , as we can foliate these halfspaces with @xmath2-planes , and the first point of touch gives a contradiction with the maximum principle . in this section , we will focus on @xmath17 case . in other words , we will consider the original asymptotic plateau problem , and show the existence of smoothly embedded least area planes @xmath7 in @xmath4 with @xmath8 for a given jordan curve @xmath0 in @xmath1 . note that gabai showed the existence of least area planes in @xmath4 in @xcite by using hass and scott s techniques . recently , ripoll and tomi also showed the existence of complete embedded minimal planes in hadamard manifolds @xcite . in this paper , we will adapt anderson s techniques in @xcite to construct minimizing @xmath2-planes . to generalize his techniques , we need to fill a gap in the proof for least area plane case . following remark explains the problem . [ gap ] ( gap in ( * ? ? ? * theorem 4.1 ) ] ) anderson showed the existence of least area planes in ( * ? ? ? * theorem 4.1 ) . he basically generalized the techniques he used for absolutely area minimizing surface case to the plane case . in particular , let @xmath0 be a jordan curve in @xmath1 , and @xmath91 be a sequence of least area disks in @xmath92 with @xmath93 where @xmath94 . then , the idea is to show the existence of a subsequence of @xmath91 converging to a least area plane @xmath7 in @xmath4 with @xmath8 . in particular , for every fixed compact domain @xmath95 in @xmath4 , he showed the sequence @xmath96 has a subsequence converging to a smooth disk in @xmath95 by using the compactness and regularity results of gmt . then , by using the diagonal sequence argument , he obtained a limit least area plane @xmath7 with @xmath97 . however , to use the compactness result in this approach , one needs a uniform area bound on the disks @xmath98 . let @xmath99 be the collection of simple closed curves @xmath100 . if @xmath101 was a sequence of area minimizing surfaces , then the area of @xmath102 would be a uniform area bound for @xmath103 as @xmath99 bounds a surface in @xmath102 and @xmath104 is absolutely area minimizing surface with boundary @xmath99 . hence , for any @xmath105 , @xmath106 . however , in the disk case @xmath107 may contain many disks or planar surfaces in @xmath95 , and the area of @xmath102 can not give an upper bound for @xmath108 . in particular , the estimate ( 4.2 ) ( @xmath109 ) in ( * ? ? ? * theorem 4.1 ) is not valid in general . for example , if @xmath110 is @xmath111 disjoint disks close to the equator disk in @xmath112 , and the area of @xmath110 would be close to the area of @xmath113 equator disks , which is much larger than the area of @xmath114 . the main difference with the area minimizing case is that if we have @xmath113 annuli @xmath115 in @xmath102 bounding @xmath99 , we can not compare the sum of the areas of the disks with the sum of the areas of the annuli . because if we replace two disks with an annulus in @xmath116 we would get a genus @xmath117 surface , which is no longer a disk so , the area of an annulus can not be compared with the area of the two disks , because of the restriction of the topology on @xmath91 . since there is no restriction on the topology of the surface for area minimizing surfaces @xmath118 gives a uniform bound , but in the least area disk case , @xmath118 does not give a uniform bound for @xmath98 . now , we will show the existence of least area plane in @xmath4 by using anderson s techniques . in order to get the uniform area bound on @xmath98 for fixed compact set @xmath95 , we will use the ideas in @xcite . [ laplane ] let @xmath0 be a simple closed curve with at least one smooth ( @xmath15 ) point in @xmath1 . then , there exists a properly embedded least area plane @xmath7 in @xmath4 with @xmath8 . we will define a special sequence of least area disks @xmath91 where their restriction to a compact subset @xmath95 has a uniform area bound . then , by following @xcite , we get a least area plane . _ notation and the setting : _ let @xmath0 be a jordan curve with at least one smooth ( @xmath15 ) point in @xmath1 . let @xmath119 be the convex hull of @xmath0 in @xmath4 . near the smooth point @xmath120 , we can have two sufficiently small and close round circles @xmath121 and @xmath122 in the opposite sides of @xmath123 such that @xmath124 bounds a least area annulus @xmath125 in @xmath4 @xcite . hence , @xmath125 goes through @xmath119 . fix a point @xmath126 in @xmath119 . let @xmath127 be the closed ball in @xmath4 of radius @xmath105 with center @xmath126 . then , for sufficiently large @xmath128 , @xmath129 is in @xmath130 . let @xmath131 . for @xmath132 , let @xmath133 and @xmath134 . then , since @xmath125 is a least area annulus , by @xcite , the pair @xmath135 bounds a least area annulus @xmath136 in @xmath119 . let @xmath137 be the compact region which @xmath136 separates from @xmath119 . let @xmath138 be an essential , smooth , simple closed curve in @xmath136 . then by construction , @xmath139 as @xmath140 . let @xmath116 be the least area disk in @xmath137 with @xmath141 @xcite . by construction , @xmath116 is also a least area disk in @xmath4 . now , we claim that there exists a subsequence of @xmath91 which converges to a least area plane @xmath7 in @xmath4 with @xmath8 . we will follow the proof of ( * ? ? ? * theorem 4.1 ) . hence , if we show the estimate ( 4.2 ) ( * ? ? * theorem 4.1 ) ( @xmath142 ) is valid for our sequence , we are done . now , let @xmath143 be a transversal arc in @xmath119 connecting @xmath144 and @xmath145 through the point @xmath126 ( centers of the balls @xmath127 ) . let @xmath146 be the length of @xmath143 . consider the following lemma from @xcite . let @xmath147 be a least disk in @xmath148 with @xmath149 . then , we call @xmath147 _ nonseparating with respect to @xmath0 in @xmath148 _ ( say wrt-@xmath0 ) , if @xmath150 is not an essential curve in the annulus @xmath151 . [ key ] ( * ? ? ? * lemma 4.1 ) let @xmath0 be a jordan curve in @xmath1 with at least one smooth ( @xmath15 ) point . let @xmath147 be a least area disk , nonseparating wrt-@xmath0 in @xmath148 for @xmath152 . then there is a monotone increasing function @xmath153 such that @xmath154 as @xmath155 , and @xmath156 where @xmath157 is the distance . let @xmath158 be sufficiently large so that @xmath159 . now , we will prove the uniform bound @xmath160 for @xmath161 where @xmath162 represents the area . by ( * ? ? ? * lemma 4.2 ) , for a given least area disk @xmath116 , @xmath163 is a collection of disjoint embedded disks for any generic @xmath164 . this is simply because @xmath163 is a surface for generic @xmath165 , and by the convexity of @xmath148 , any component in @xmath163 must be a disk . we claim that for @xmath166 , @xmath163 is just a disk ( only one component ) for any generic @xmath161 . assume on the contrary , and let @xmath167 be two such disks in @xmath168 . since @xmath116 is an embedded disk in @xmath127 , there is a path @xmath31 connecting @xmath169 and @xmath170 in @xmath116 . let @xmath171 be the smallest radius such that @xmath169 and @xmath170 are in the same component @xmath172 of @xmath173 . hence , @xmath172 is a least area disk , nonseparating wrt-@xmath0 in @xmath174 for sufficiently small @xmath175 . hence by the lemma [ key ] , @xmath176 . hence , we get @xmath177 which gives a contradiction . this proves that for @xmath166 and for any @xmath161 , @xmath163 is a disk ( only one component ) . hence , for any fixed @xmath161 , @xmath178 , which gives the desired uniform bound . this proves the estimate ( 4.2 ) of ( * ? ? ? * theorem 4.1 ) is valid for our sequence @xmath91 . the proof follows . in this section , we will show the existence of the solutions of asymptotic @xmath2-plateau problem in @xmath4 for @xmath66 . in particular , we will generalize the techniques in the previous section to the cmc case , and show that for any simple closed curve @xmath0 in @xmath1 with one smooth point , there exists an embedded @xmath2-plane @xmath3 with @xmath16 . first , we need to generalize lemma [ key ] proven in @xcite to the minimizing @xmath2-planes in @xmath4 . we will use the same notation , i.e. let @xmath179 be as in the previous section . fix @xmath66 . let @xmath180 be an annulus in @xmath181 . again , let @xmath147 be a minimizing @xmath2-disk in @xmath148 with @xmath182 . then , we call @xmath147 _ nonseparating wrt-@xmath0 in @xmath148 _ , if @xmath150 is an essential curve in the annulus @xmath183 . [ key2 ] fix @xmath66 . let @xmath0 be a jordan curve with at least one smooth ( @xmath15 ) point in @xmath1 . for @xmath152 , let @xmath147 be a minimizing @xmath2-disk , and nonseparating wrt-@xmath0 in @xmath148 . then there is a monotone increasing function @xmath153 such that @xmath154 as @xmath155 , and @xmath156 where @xmath157 is the distance . we will adapt the proof of lemma [ key ] to this case . lemma [ key ] finishes the @xmath17 case . hence , we can take @xmath184 . for @xmath185 , the same proof works by changing the orientation . fix @xmath184 . recall that @xmath186 is the least area annulus in @xmath148 with @xmath187 for generic @xmath188 . let @xmath189 be the annulus in @xmath181 with @xmath190 . let @xmath191 . we claim that for any minimizing @xmath2-disk @xmath147 which is nonseparating wrt-@xmath0 in @xmath148 , @xmath192 , i.e. @xmath147 stays in the solid torus @xmath193 in @xmath148 with @xmath194 ; see figure [ key3 ] . in particular , this shows that @xmath156 , and the proof will follow . now , let @xmath147 be a minimizing @xmath2-disk and nonseparating wrt-@xmath0 in @xmath148 with @xmath195 . since @xmath147 is nonseparating wrt-@xmath0 , @xmath196 is a not an essential curve in @xmath189 . in other words , @xmath196 bounds a disk @xmath197 in @xmath189 ; see figure [ key3 ] . let @xmath198 be the region in @xmath148 with @xmath199 . since @xmath147 is a minimizing @xmath2-disk in @xmath148 with @xmath184 , @xmath200 is the smallest among all the disks in @xmath148 with boundary @xmath196 . here , @xmath162 represents the area or the volume of the corresponding region . assume that @xmath201 . recall that @xmath193 is the region in @xmath148 with @xmath194 . let @xmath202 . then as @xmath201 , @xmath203 . furthermore , let @xmath204 be the planar region in @xmath186 with @xmath205 . let @xmath32 be the planar region in @xmath147 with @xmath206 . since @xmath147 is a nonseparating disk , and @xmath186 is an annulus , @xmath207 is also a disk in @xmath148 with @xmath208 . furthermore , as @xmath186 is a least area annulus , @xmath209 and hence @xmath210 . this implies @xmath211 . however , @xmath147 is a minimizing @xmath2-disk in @xmath148 . this is a contradiction . this proves @xmath192 , and @xmath212 . hence , this shows that @xmath213 . the lemma follows . now , we prove the main result of the paper . [ h - plane ] [ existence of @xmath2-planes ] let @xmath0 be a jordan curve in @xmath1 with at least one smooth ( @xmath15 ) point . let @xmath66 . then , there exists a properly embedded minimizing @xmath2-plane @xmath3 in @xmath4 with @xmath16 . fix @xmath66 . given @xmath0 in @xmath1 , we will use the same setup as before , i.e. let @xmath214 be as in the previous lemma . let @xmath215 for @xmath132 . let @xmath138 be an essential smooth curve in @xmath216 . then by construction , @xmath139 as @xmath217 . recall that the mean curvature of the geodesic sphere of radius @xmath218 is @xmath219 . hence , @xmath127 is @xmath117-convex for any @xmath105 . then , by lemma [ embed ] , there exists an embedded minimizing @xmath2-disk @xmath116 in @xmath127 with @xmath141 . here , the sign of @xmath2 determines the minimizing @xmath2-disk as there are two minimizing @xmath2-disks in @xmath127 facing each other for @xmath220 @xcite . furthermore , since @xmath221 , by lemma [ convex ] ( see also remark [ convex2 ] ) , @xmath222 . hence , this implies @xmath223 . we claim that the sequence of embedded minimizing @xmath2-disks @xmath91 converges to a minimizing @xmath2-plane @xmath3 . first , we claim that the sequence of minimizing @xmath2-disks @xmath91 has a convergent subsequence in any compact set @xmath95 in @xmath4 . then , by using the diagonal sequence argument as before , we will get a limit minimizing @xmath2-plane in @xmath4 . consider the closed balls @xmath224 with center @xmath126 . by using convexity of @xmath225 , for generic @xmath226 , we can assume @xmath227 is a collection of disks for any @xmath105 by ( * ? ? ? * lemma 4.2 ) . now , let @xmath143 be a transversal arc in @xmath72 connecting @xmath228 and @xmath229 through the point @xmath126 . let @xmath146 be the length of @xmath143 . fix a generic @xmath230 such that @xmath231 . then by lemma [ key2 ] , and the proof of theorem [ laplane ] , @xmath232 is a closed disk ( only one component ) for any @xmath105 . hence , the sequence of minimizing @xmath2-disks @xmath233 has a uniform area bound , say @xmath234 . then , by following the proof of ( * ? ? ? * theorem 4.1 ) , the existence of smoothly embedded minimizing @xmath2-plane @xmath53 can be shown as follows : with the uniform area bound , the sequence of minimizing @xmath2-disks @xmath235 has a convergent subsequence in @xmath236 by the compactness theorem for integral currents . hence by considering them as integral currents , a subsequence of @xmath235 converges to a properly embedded minimizing @xmath2-disk @xmath237 in @xmath238 . let @xmath239 be a monotone increasing sequence with @xmath240 . for @xmath241 , by starting with this subsequence , get another subsequence converging on @xmath242 . by iterating this process and diagonal sequence argument , we get a sequence of integral currents @xmath91 converges on compacts to the integral current @xmath53 in @xmath4 . by allard s regularity , the convergence is smooth on compact sets . also , the asymptotic boundary of the support of @xmath53 is @xmath0 by the convex hull property ( lemma [ convex ] ) , i.e. @xmath90 as @xmath217 . the limit of minimizing @xmath2-disks @xmath53 is a minimizing @xmath2-surface . hence , by @xcite , for any point @xmath243 in the support of @xmath53 , there exists @xmath175 with @xmath244 is a smooth embedded disk . hence , the support of @xmath53 is smoothly embedded surface . finally , since the convergence is smooth in compact sets , and @xmath91 is a sequence of embedded disks , @xmath53 is a complete minimizing @xmath2-plane in @xmath4 with @xmath90 . the proof follows . [ pair ] [ pair of @xmath2-planes ] notice that if you forget the sign of the mean curvature @xmath2 , the theorem above shows that for a given jordan curve @xmath0 in @xmath1 , there exist two minimizing @xmath2-planes @xmath20 and @xmath21 with @xmath245 for @xmath19 . furthermore , these @xmath2-planes are disjoint , and the convex sides are facing each other . the above result was shown for special families of curves in @xmath1 like star - shaped curves @xcite , `` mean convex '' curves @xcite , and graph over a line @xcite , where they showed that the area minimizing surface @xmath53 is indeed a graph over a geodesic subspace in @xmath4 . [ proper ] [ properly embeddedness ] let @xmath0 be a @xmath246 smooth jordan curve in @xmath1 . let @xmath3 be a minimizing @xmath2-plane in @xmath4 with @xmath16 . then , @xmath3 is properly embedded in @xmath4 . by @xcite , since @xmath0 is @xmath246 smooth , @xmath3 is regular near infinity . in particular , there exists a @xmath247 such that in the upper half space model of @xmath4 , @xmath248 is a graph over @xmath249 . then , for sufficiently large @xmath250 , @xmath251 is a jordan curve @xmath252 in @xmath253 hence , @xmath254 is a minimizing @xmath2-disk in @xmath255 with @xmath256 by the definition of minimizing @xmath2-plane . by lemma [ embed ] , @xmath257 is properly embedded . since @xmath258 is a graph over @xmath259 in the upper half space model @xcite , the proof follows . indeed , it can be showed that the above result is true for far more generality . by using the techniques in @xcite , one can naturally generalize the above result to jordan curves in @xmath1 with at least one smooth ( @xmath15 ) point . even though there is no regularity near infinity in that case , by using lemma [ key2 ] , the arguments in @xcite can easily be adapted . note also that , meeks , tinaglia and the author recently showed that there exists a _ nonproperly _ embedded complete @xmath2-plane @xmath53 in @xmath4 for any @xmath14 @xcite . in particular , @xmath53 is an @xmath2-plane between two rotationally invariant @xmath2-catenoids @xmath260 and @xmath261 where @xmath53 spirals into @xmath260 in one end , and spirals into @xmath261 in the other end . hence , @xmath262 is a pair of infinite lines @xmath263 and @xmath264 in @xmath1 . here , if @xmath265 , and @xmath266 is the annuli in @xmath1 with @xmath267 , then @xmath268 and @xmath269 where @xmath270 spirals into @xmath271 in one end , and spirals into @xmath272 in the other end . the generic uniqueness results for minimizing @xmath2-surfaces in @xmath4 @xcite can naturally be generalized to our context , i.e minimizing @xmath2-planes . in particular , for fixed @xmath66 , let @xmath55 and @xmath56 be minimizing @xmath2-planes with @xmath273 where @xmath274 and @xmath275 are disjoint simple closed curves in @xmath1 . then , by using meeks - yau exchange roundoff trick , it can be showed @xmath55 and @xmath56 are disjoint , too . by using this , and similar ideas to @xcite , it can be showed that any simple closed curve @xmath0 bounds either a unique minimizing @xmath2-plane @xmath30 in @xmath4 , or there are two canonical disjoint minimizing @xmath2-planes @xmath276 and @xmath277 with @xmath278 . hence , foliating an annular neighborhood of @xmath0 in @xmath1 with simple closed curves , and considering the canonical @xmath2-planes constructed , one can get generic uniqueness result as in @xcite . in particular , this shows that _ for fixed @xmath66 , a generic jordan curve @xmath0 in @xmath1 bounds a unique minimizing @xmath2-plane @xmath30 in @xmath4 with @xmath279_. similar to the previous part , it is also possible to show that two @xmath2-planes in @xmath4 with different @xmath2 values , which are asymptotic to the same asymptotic curve in @xmath1 are disjoint . in particular , for a given jordan curve @xmath0 in @xmath1 , and given @xmath280 , it can be showed that the minimizing @xmath2-planes @xmath281 and @xmath282 in @xmath4 with @xmath283 are disjoint by using the ideas in @xcite . hence , if @xmath0 bounds a unique minimizing @xmath2-plane for any @xmath66 , it can be showed that the family of planes @xmath284 foliates @xmath4 . by the disjointness of the planes @xmath285 for different @xmath2 s , in order to get the foliation , all one needs to show is that there is no gap between the planes @xmath285 by using the uniqueness . by the arguments in @xcite , a gap between the planes , say between @xmath286 and @xmath287 , implies the nonuniqueness for @xmath42-planes with @xmath288 . this gives a contradiction , and shows that @xmath289 foliates @xmath4 . for example , if @xmath0 is a star - shaped curve in @xmath1 , then the family @xmath289 foliates @xmath4 @xcite .	we show that for any jordan curve @xmath0 in @xmath1 with at least one smooth point , there exists an embedded @xmath2-plane @xmath3 in @xmath4 with @xmath5 for any @xmath6 . [ section ] [ thm]lemma [ thm]corollary [ thm]definition [ thm]remark
the decay of metastable states is a basic phenomenon of great generality@xcite with numerous applications in a large number of contexts , ranging from the decay of false vacua in field theory@xcite , e.g. in cosmology@xcite , to the creep type motion of topological defects in solids@xcite . at a given temperature @xmath11 the inverse lifetime @xmath12 of a metastable system can be written in the form @xmath13 with @xmath14 the euclidean action of the saddle - point configuration and @xmath15 the prefactor determined by the associated fluctuations . in this paper we present a calculation of the full temperature dependence of the euclidean action of a metastable @xmath0-dimensional elastic manifold pinned by an isotropic potential @xmath16 in the presence of a small external force . typical examples of physical realizations of this model system are a string pinned by a linear or columnar defect ( ( 1 + 1)- , and ( 1 + 2)-dimensional problems respectively ) or a membrane pinned by a plane ( ( 2 + 1)-dimensional problem ) . with increasing temperature , the decay changes its nature from quantum to classical . the function @xmath14 might be either a smooth function of temperature or exhibit a sharp kink with a discontinuity in its first derivative at the crossover temperature @xmath9 . the former case is called a `` second - order '' transition from quantum to classical behavior@xcite , while in the latter case a first - order transition takes place . the wording `` transition '' is appropriate as the crossovers have all the features of mean - field phase transitions upon identifying the euclidean action with the free energy . the possibility of a first - order transition in a tunneling problem has first been discussed by lifshitz and kagan@xcite in the context of quantum phase transitions in @xmath17 systems . the general theory and an appropriate criterion for a first - order transition in 1d hamiltonian systems has been developed by meshkov@xcite and by ioselevich and rashba@xcite , see also the paper of chudnovsky@xcite . for the case of non - hamiltonian systems this problem has been considered in ref .. recently , the possibility of observing a first - order transition from quantum to classical behavior in spin systems has been discussed in ref .. systems with many degrees of freedom have been studied by garriga@xcite ( see also ref.@xcite ) and he finds that , for the case of tunneling of massive @xmath18 and @xmath19 elastic manifolds between two minima disbalanced by the action of a small external force , the euclidean action exhibits a discontinuous derivative , while for a @xmath20 dimensional string the problem exhibits a smooth crossover@xcite . a natural counterpart of the bistability problem studied by coleman , garriga , and others is the generic problem of weak metastability , see fig . 1 . in this paper we then study the problem where an elastic manifold is pinned by an isotropic potential well and is driven by a weak external force . after the tunneling process ( or thermal activation at high temperatures ) the manifold becomes free , while in the previous case it is repinned again in the second valley . the approximation introduced by coleman@xcite and used by garriga is usually termed the `` thin wall approximation '' , while in the present problem it is more appropriate to denote the technique used as a `` thick wall approximation '' . we find that at least for the case @xmath4 the problem exhibits a sharp transition from quantum to classical behavior : the function @xmath21 is a constant ( up to exponentially small corrections ) at temperatures lower than the critical temperature @xmath22 while for @xmath23 @xmath24 with @xmath25 the activation energy . the ( 1 + 1)-dimensional problem has been considered by skvortsov@xcite . in the present paper we generalize the results of this work for an arbitrary number of internal degrees of freedom @xmath0 and improve on the calculation of the preexponential factor for the @xmath10-problem as the approximations made in ref . turn out to be too rough to reach the correct result . 0.5 cm -0.7 cm the paper is organized as follows : in sec . [ ea ] we formulate the problem . in sec . [ quantuminst ] and [ thth ] the zero temperature and thermal instantons are found and the corresponding euclidean actions are calculated . in sec . [ instability ] the instability temperature of the thermal instanton is calculated and the full temperature dependence of the euclidean action is found . [ preexponential ] is devoted to the calculation of the high - temperature asymptotics of the preexponential factor for the @xmath10-problem . we summarize the results in sec . [ conclusion ] and discuss their possible applications to the thermal depinning of vortices in high-@xmath9 superconductors and nucleation phenomena in phase transitions . consider a massive manifold with @xmath0 internal dimensions embedded in a @xmath1-dimensional space . the ( real time ) lagrangian of the manifold can be written in the form @xmath26= \int\limits_{\omega } \thinspace d{\bf x } \left \ { \frac{\rho}{2 } { \left ( \frac{\partial { \bf u}}{\partial t}\right ) } ^{2}- \frac{\epsilon}{2 } { \left ( \frac{\partial { \bf u}}{\partial { \bf x}}\right ) } ^{2 } -v(\|{\bf u}\|)+{\bf f}\cdot { \bf u}\right \}. \label{lagrangian}\ ] ] here , @xmath27 is the ( transverse ) displacement field of the manifold , @xmath28 is the vector characterizing the internal degrees of freedom , @xmath29 and @xmath30 are the mass density and the elasticity of the manifold , respectively , and @xmath31 is an @xmath32-invariant trapping potential . the function @xmath33 is supposed to be monotonously increasing , @xmath34 the integration in eq . ( [ lagrangian ] ) goes over the volume @xmath35 of the manifold . finally , @xmath36 is the external driving force which we assume to be small , @xmath37 with @xmath38 the characteristic radius of the pinning potential @xmath39 specific realizations of this model system are strings moving in a plane ( @xmath10-problem ) or in space ( @xmath40-problem ) , or membranes pinned by planar boundaries or interfaces in 3d space ( @xmath41-problems ) . after the transformation @xmath42 and @xmath43 the euclidean action of the manifold reads @xmath44= \int\limits_{-{\hbar}/{2t}}^{+{\hbar}/{2 t } } d\tau\int\limits_{\omega } \thinspace d{\bf x } \left \ { \frac{\rho}{2 } { \left ( \frac{\partial { \bf u}}{\partial \tau}\right ) } ^{2}+ \frac{\epsilon}{2 } { \left ( \frac{\partial { \bf u}}{\partial { \bf x}}\right ) } ^{2 } + v(\|{\bf u}\|)-{\bf f}\cdot { \bf u}\right \}. \label{euclidinitial}\ ] ] after introducing the new variables @xmath45 and @xmath46 , the euclidean action ( [ euclidinitial ] ) can be rewritten in the form @xmath47= { \rho}^{{1}/{2}}{\epsilon}^{{d}/{2 } } \int\limits_{-{\hbar}/{2\sqrt{\rho}t}}^{+{\hbar}/{2\sqrt{\rho}t } } { d\tau}^{\prime}\int\limits_{{\omega}^{\prime}}\thinspace d{\bf x}^{\prime } \left \ { \frac{1}{2 } { \left ( \frac{\partial { \bf u}}{\partial { \tau}^{\prime}}\right ) } ^{2}+ \frac{1}{2 } { \left ( \frac{\partial { \bf u}}{\partial { \bf x}^{\prime}}\right ) } ^{2 } + v(\|{\bf u}\|)-{\bf f}\cdot { \bf u}\right \}. \label{euclideanfinal}\ ] ] below we shall work with this euclidean action and find its zero- and finite temperature instantons . let us calculate the euclidean action corresponding to the extremal trajectories at a given temperature @xmath48 the variation of the action ( [ euclideanfinal ] ) yields the saddle - point equation @xmath49 we need to find the solution satisfying the boundary condition @xmath50 the solution of eq . ( [ saddlepoint ] ) can be written in the form @xmath51 at zero temperature the instanton is spherically symmetric and introducing the new variable @xmath52 we obtain an ordinary differential equation for the function @xmath53 @xmath54 the function @xmath55 must be non - zero and continuous , have a continuous derivative @xmath56 and satisfy the boundary condition @xmath57 the corresponding solution consists of two parts : a `` macroscopic '' solution for @xmath58 ( @xmath59 is a parameter which will be determined later ) and a `` microscopic '' solution for @xmath60 the former corresponds to displacements @xmath61 ( we remind the reader that @xmath38 is the characteristic radius of the pinning potential ) , while the latter describes the bounce in the region @xmath62 where the potential is relevant . consider the `` microscopic '' solution where we can neglect the ( small ) force @xmath63 in the equation of motion ( [ saddlenew ] ) as compared to the potential term @xmath64 . furthermore , in the limit of small @xmath63 the radius @xmath59 of the nucleus is large ( we will check this assumption at the end ) , such that we can neglect the term @xmath65 in eq . ( [ saddlenew ] ) , and the remaining equation takes the form @xmath66 eq . ( [ newton ] ) is identical to newton s second law . accounting for the boundary condition ( [ boundarycondition ] ) we obtain @xmath67 which is equivalent to the law of energy conservation for the 1d conservative motion in classical mechanics . as @xmath68 , we can obtain the boundary condition for the function @xmath55 @xmath69 a simple integration then provides the microscopic part of the solution , once the parameter @xmath59 is known . next , we solve eq . ( [ saddlenew ] ) in the region @xmath70 with the boundary conditions given by eq . ( [ boundary ] ) and @xmath71 as we will see below , @xmath72 allowing us to use the condition @xmath73 without changing the main term in the asymptotics of the euclidean action in the limit @xmath74 in the region @xmath70 we can neglect the term @xmath75 in eq . ( [ saddlenew ] ) and the general solution can be written in the form @xmath76 where @xmath77 the term @xmath78 is always singular at the point @xmath79 , hence @xmath80 taking into account the boundary condition ( [ boundary ] ) we obtain @xmath81 and the quantum instanton can be written in the form @xmath82 substituting the function @xmath55 into eq . ( [ euclideanfinal ] ) and taking into account that @xmath83 we obtain the zero temperature euclidean action @xmath84 where @xmath85 is the surface area of a unit sphere in @xmath0-dimensional space . neglecting the terms @xmath63 or @xmath75 in eq . ( [ saddlenew ] ) might seem questionable as there is an interval with the length of order @xmath86 where @xmath87 however , if this interval is much smaller than @xmath59 the approximation used above is valid : assume that the potential @xmath33 at large @xmath88 behaves as @xmath89 the equation @xmath90 can be easily solved and yields the result @xmath91 i.e. , in the limit @xmath92 @xmath93 and the approximation made above is asymptotically correct . next , let us calculate the activation energy . the saddle - point solution is time - independent in this case , i.e. , we can eliminate the term @xmath94 in eq . ( [ saddlepoint ] ) . the remaining equation is identical to the @xmath95-dimensional zero temperature problem considered above and we can use the instanton ( [ instanton ] ) with @xmath96 we substitute this solution into eq . ( [ euclideanfinal ] ) and obtain the expression for the activation energy , @xmath97 we remind the reader that for the thermal instanton @xmath98 next , let us calculate the two characteristic temperatures @xmath99 and @xmath100 the former corresponds to the value of @xmath101 where the quantum and thermal exponents @xmath102 and @xmath103 match up , while the latter defines the limit of applicability of the zero temperature instanton ( see fig . 2 ) : at temperatures @xmath104 the instanton ( [ instanton ] ) does not satisfy the periodic boundary condition @xmath105 note that the thermal instanton is always a valid solution as it is time - independent and satisfies the periodic boundary condition identically . the exponents @xmath102 and @xmath103 are equal at the temperature @xmath99 given by the expression ( we make use of eqs . ( [ sq ] ) and ( [ u ] ) ) @xmath106 formally , the quantum instanton found in section [ quantuminst ] is valid only at zero temperature , however , we can use it also at @xmath107 as long as the periodic boundary conditions are satisfied asymptotically . the zero temperature bounce solution then can be still applied if the periodicity in imaginary time @xmath108 @xmath109 , is larger than the diameter of the nucleus @xmath110 i.e. , if @xmath111 ( see fig . 2 ) . -3.2 cm nucleus corresponding to the low temperature decay process . as long as the diameter @xmath112 remains smaller than the periodicity in time @xmath113 ( solid lines ) the instanton ( [ instanton ] ) asymptotically satisfies the boundary conditions . at temperatures @xmath114 the boundary conditions are not fulfilled and the distortion of the nucleus walls becomes relevant . 0.5 cm making use of eq . ( [ rd ] ) we obtain the following expression for the temperature @xmath115 at which one needs to take into account the distortion of the quantum instanton due to the periodic boundary conditions , @xmath116 in this section we calculate the instability temperature @xmath117 of the thermal instanton . at this temperature the saddle - point solution becomes time - dependent . then we show that @xmath118 for @xmath4 and , consequently , the problem exhibits a first - order transition from quantum to classical behavior . let us expand the euclidean action around the thermal saddle - point solution . as the bounce satisfies the euler - lagrange equations , the term proportional to @xmath119 ( @xmath119 describes the distortion of the saddle - point configuration ) is equal to zero . to second order in @xmath119 we obtain @xmath120 & \simeq & s_{\rm eucl}[{\bf u}_{th}({\bf x}^{\prime})]+ \frac{1}{2 } \int\limits_{-{\hbar}/{2\sqrt{\rho}t}}^{+{\hbar}/{2\sqrt{\rho}t}}\thinspace d{\tau}^{\prime } \int\limits_{\omega^{\prime}}\thinspace d{\bf x^{\prime } } \delta u_{1 } \left ( { \hat h}_{1}-\rho \frac{\partial^{2}}{\partial{\tau^{\prime}}^{2}}\right ) \delta u_{1 } \nonumber\\ & + & \frac{1}{2 } \int\limits_{-{\hbar}/{2\sqrt{\rho}t}}^{+{\hbar}/{2\sqrt{\rho}t}}\thinspace d{\tau}^{\prime } \int\limits_{\omega^{\prime}}\thinspace d{\bf x^{\prime } } \delta { { \bf u}_{\bot } } \left ( { { \hat h}_{\bot}}-\rho\frac{\partial^{2}}{\partial{\tau^{\prime}}^{2 } } \right ) \delta { { \bf u}_{\bot}},\end{aligned}\ ] ] where @xmath121 and @xmath122 the operator @xmath123 can be written in the form @xmath124 the instability temperature is determined by its only negative eigenvalue . the operator @xmath125 has only positive eigenvalues . let us calculate the lowest eigenvalue @xmath126 of the operator ( [ eigenvalue ] ) . here , we follow the procedure described in detail in refs .. first , we show that the operator ( [ eigenvalue ] ) has @xmath0 zero eigenvalues : the @xmath0 functions @xmath127 @xmath128 satisfy the equation @xmath129 due to the spherical symmetry of the instanton and , consequently , of the `` potential '' in eq . ( [ eigenvalue ] ) , the eigenfunctions of the @xmath0-dimensional schrdinger operator ( [ eigenvalue ] ) can be represented in the form @xmath130 with @xmath131 and @xmath132 the radial and angular components , respectively , @xmath133 for the @xmath0 eigenfunctions corresponding to the zero eigenvalue @xmath134 for @xmath135 ( see eq . ( [ instanton ] ) and the first paragraph in section [ thth ] ) . in the region @xmath136 the eigenfunctions of the lowest levels are asymptotically ( @xmath137 ) equal , allowing us to find the boundary condition for the lowest eigenvalue @xmath126 from that of the zero eigenvalue wavefunction . the condition for binding the radial parts of the wavefunctions at the point @xmath138 takes the form @xmath139 the general solution of the equation @xmath140 in the region @xmath141 can be written in the form @xmath142,\ ] ] where @xmath143 and @xmath144 are bessel functions of imaginary argument . the function @xmath145 is singular at the point @xmath146 and hence @xmath80 ( for the case @xmath20 this analysis is not valid , however it can be easily shown that the equation for the lowest eigenvalue ( see eq . ( [ mu ] ) ) remains the same ) . using eq . ( [ binding ] ) we obtain a transcendental equation for the negative eigenvalue @xmath147 , @xmath148 and @xmath149 using eqs . ( [ tc ] ) and ( [ t * ] ) and the definition of the parameter @xmath150 see eq . ( [ mu ] ) , we find @xmath151 and @xmath152 0.5 cm .ratios @xmath153 and @xmath154 for @xmath155 for @xmath156 the system exhibits a sharp transition from quantum to classical behavior . for @xmath157 both ratios exceed unity and no conclusion can be drawn regarding the order of the transition . [ cols= " < , < , < , < " , ] 0.5 cm [ temperatures ] in table [ temperatures ] we summarize the numerical values of the parameters @xmath158 and @xmath159 we see that for @xmath4 the relation @xmath160 holds and , consequently , the euclidean action has the form plotted in fig . 3 : it is equal to @xmath161 ( see eq . ( [ sq ] ) ) for @xmath162 ( see eq . ( [ tc ] ) ) , whereas for @xmath163 we have @xmath164 . the jump from the zero - temperature to the high - temperature instanton takes place at @xmath165 : the system exhibits a sharp transition from quantum to classical behavior , see fig . 0.5 cm the euclidean action as a function of temperature for @xmath155 for @xmath5 the euclidean action is constant and given by eq . ( [ sq ] ) ( see eq . ( [ tc ] ) for an expression for @xmath9 ) . for @xmath166 the action is equal to @xmath167 , with @xmath168 given by eq . ( [ u ] ) . 0.5 cm in this section we calculate the preexponential factor for the @xmath10-problem in the high temperature region , improving on the results of ref .. we start from langer s expression for the decay rate@xcite ( see also ref . ) @xmath169 where @xmath170 see eq . ( [ temp0 ] ) and table [ temperatures ] , and @xmath171 is the partition function of the system . for a ( 1 + 1)-string we can write@xcite @xmath172}^{{1}/{2 } } { \left \| \frac{{\rm det}\left ( { \delta^{2}s}/{\delta u^{2}}\|_{u=0}\right ) } { { \rm det}^{\prime}\left ( { \delta^{2}s}/{\delta u^{2}}\|_{u = u_{th}(r)}\right ) } \right \|}^{{1}/{2}}\exp{\left ( -\frac{u(1)}{t}\right ) } , \label{decay}\ ] ] where @xmath173 is the length of the string and the prime indicates that we exclude the zero eigenvalue of the operator @xmath174 . the activation energy @xmath175 is given by the expression @xmath176 and the eigenvalues of the operators @xmath177 and @xmath178 take the form @xmath179 and @xmath180 respectively . carrying out the product over @xmath181 we obtain the high temperature result @xmath172}^{{1}/{2 } } { \left \| \frac{{\rm det}{\hat h}_{0 } } { { \rm det}^{\prime}{\hat h}_{1 } } \right \|}^{{1}/{2}}\exp{\left ( -\frac{u^{}(1)}{t}\right ) } , \label{decay1}\ ] ] where @xmath182 \label{renorm}\ ] ] is the quantum renormalized activation energy , see ref .. in eq . ( [ renorm ] ) @xmath183 @xmath184 is the difference in the continuous part of the spectral densities of the operators @xmath185 and @xmath123 , and the sum is taken over the positive discrete eigenvalues of the operator @xmath123 ( note that the operator @xmath185 has no discrete eigenvalues ) . at large @xmath186 the correction to @xmath175 diverges as @xmath187 , i.e. , we have to introduce a cutoff @xmath186 into the problem . the determinant ratio in eq . ( [ decay1 ] ) has been calculated by krmer and kulic@xcite using the gelfand - yaglom formula@xcite with the result @xmath188 the coefficient of proportionality depends on the detailed form of the pinning potential @xmath39 the final answer for the decay rate then takes the form @xmath189 accounting for gaussian fluctuations we obtain an exponential enhancement of the decay rate . note that @xmath190 , see eqs . ( [ renorm ] ) and ( [ acten ] ) . the result ( [ otwet ] ) is valid for temperatures @xmath11 below the kink activation energy @xmath191 @xmath192 ( here we assumed that the quantum correction to the activation energy @xmath175 is small and @xmath193 ) . at higher temperatures we need to take into account the renormalization of the free energy due to thermal fluctuations : the free energy per unit length of an elastic string trapped in a potential well coincides with the ground state energy of a massive particle in the equivalent quantum mechanical problem@xcite , where the temperature plays the role of planck s constant . if the temperature is small , the ground state energy is close to zero . at high temperatures , satisfying the condition @xmath194 ( which is the condition for a weak potential well , see ref . ) , the ground state energy increases substantially and becomes comparable with the potential well depth . in this case the thermal instanton found in sec . [ euclid ] is not applicable and the problem is more conveniently solved in the quantum mechanical formulation : since in the 1d quantum particle trapped in an attractive potential always exhibits a bound state , the @xmath10-dimensional string problem at @xmath195 does not undergo a depinning transition at any temperature . it is instructive to compare the result ( [ otwet ] ) to that obtained for the case of thermally activated motion of a string between two nearly degenerate minima@xcite , see fig . 1(a ) , where the prefactor shows a power - like dependence on the external force @xmath63 . the different dependences of the prefactors follow from the different low energy spectra : in the present case we have @xmath196 low - lying eigenvalues of order @xmath197 ( see ref.@xcite ) , the product @xmath198 of which behaves as @xmath199 , while for the problem studied in refs . only one small eigenvalue exists@xcite . finally , let us compare the result ( [ otwet ] ) with that obtained previously by skvortsov@xcite where the semiclassical approximation for the eigenvalues of the operators @xmath185 and @xmath123 has been used . the derivation in ref . differs from the present one in two respects : _ i _ ) the calculation has been based on the low temperature expression @xmath200 for the decay rate , such that the final high temperature result lacks the correct classical limit ( @xmath201 , see eq . ( 57 ) in ref . ) . _ ii _ ) correcting for the factor @xmath202 , the remaining difference can be traced back to the use of the quasiclassical approximation in the calculation of the ratio ( [ otwet ] ) , which produces a prefactor @xmath203 rather than the correct result @xmath204 . note that at zero temperature eq . ( [ nultemperatur ] ) is applicable , however , the semiclassical approach is not accurate enough to produce the correct expression for the decay rate ( though it is still possible to obtain the correct exponential enhancement of the preexponential factor and the correct quantum renormalized euclidean action @xmath205 , see ref . , similar to the high temperature case discussed above ) . in principle , the complete prefactor in the low temperature quantum regime can be caclulated using the procedure suggested in refs . : the calculation of the determinant ratios can be reduced to the calculation of an infinite product of determinants of 1d schrdinger operators , which then can be calculated numerically . as possible applications of the present problem we discuss _ i ) _ the thermal depinning of vortices in high-@xmath9 superconductors and _ ii ) _ the nucleation of phase transitions in the vicinity of a boundary or an interface . _ i ) _ the dynamics of vortices in high - temperature superconductors is dominated either by the dissipative or the hall term . the vortex mass can be neglected . however at high temperatures the activation energy is independent of the dynamics ( see eq . ( [ u ] ) ) and we can use eq . ( [ u ] ) for the description of the thermal depinning of a single vortex line ( @xmath20 ) from a columnar defect ( @xmath206 ) , or of a vortex sheet ( @xmath18 ) from an interface between superconducting grains@xcite ( @xmath207 ) . we wish to point out that eq . ( [ u ] ) can also be used for the activation energy of a single vortex depinning from a twin boundary . note that a vortex pinned by a columnar defect ( @xmath40-dimensional problem ) again does not undergo a depinning transition at any finite temperature : the corresponding 2d quantum mechanical problem always has a bound state and hence the vortex remains localized at all temperatures provided that @xmath195 , see ref .. as our result is restricted to temperatures below the kink energy , our analysis is not in contradiction with this result . _ ii ) _ consider the nucleation of a new phase@xcite in the vicinity of a boundary@xciteor interface . let us assume that there exist two competing ( @xmath208-dimensional ) phases , the stability of which are depending on the given external conditions . initially , phase 1 is prepared , after which the external conditions are changed adiabatically , rendering phase 1 metastable . the nucleation of phase 2 takes a finite time and might be of two types : usual bulk nucleation or nucleation in the vicinity of the boundary . the latter can be described as the problem of depinning of a @xmath209-dimensional manifold . the boundary between the two phases plays the role of the elastic manifold , while the difference in the chemical potentials plays the role of the small external force . it is interesting to study the competition between these two possible types of nucleation . the free energy @xmath210 of the @xmath208-dimensional spherical nucleus of radius @xmath211 is given by the expression @xmath212 where @xmath213 and @xmath214 are the surface and the volume of a unit sphere in @xmath208-dimensional space . minimizing this energy with respect to @xmath211 we obtain @xmath215 and @xmath216 this expression should be compared to @xmath217 as given by eq . ( [ u ] ) . after some algebra we obtain @xmath218 it is reasonable to assume that @xmath219 , i.e. , the boundary tension @xmath30 roughly matches up with the pinning potential @xmath220 . in this case @xmath221 for @xmath222 and 3 , i.e. , nucleation at the boundary is more favorable than that in the bulk ( the same result applies also for the weak pinning case @xmath223 ) . the analysis for the quantum case is more complicated : the kinetic energy of the nucleus can be written in the form @xmath224 with @xmath225 a function of the nucleus radius @xmath211 . the process of nucleation of the new phase leads to a redistribution of mass in space as the phases have different densities in general . for the case @xmath226 the function @xmath225 has been calculated in ref . , where it has been assumed that both phases are incompressible quantum liquids . for the case @xmath222 this approach gives a divergent kinetic energy and , consequently , we need a more detailed description of the phases 1 and 2 to reach a sensible result . similarly , the effective mass per area of the boundary between the two phases can not be calculated assuming incompressible phases . briefly summarizing , we have analyzed the problem of depinning of a @xmath1-dimensional massive elastic manifold from an @xmath32-invariant trapping potential @xmath16 in the presence of a small external force @xmath3 for the case @xmath4 the euclidean action has been calculated in the whole temperature range , see eq . ( [ sq ] ) for @xmath5 ( @xmath9 is given by eq . ( [ tc ] ) ) and eq . ( [ u ] ) for @xmath166 , and we have found a sharp transition from quantum to classical behavior , see fig . the high - temperature asymptotics for the preexponential factor of the @xmath10-problem has been calculated . possible applications to the thermal depinning of vortices in high-@xmath9 superconductors and the nucleation problem in the vicinity of a boundary in @xmath208-dimensional phase transitions have been discussed and we have shown that for @xmath227 nucleation at the boundary is more favorable than from the bulk . 99 for a general review concerning the decay of metastable states , see p. hnggi , p. talkner , and b. borkovec , rev . phys . 62 * , 251 ( 1990 ) . s. coleman , phys . rev . d * 15 * , 2929 ( 1977 ) ; c. g. callan and s. coleman , phys . d * 16 * , 1762 ( 1977 ) . a. d. linde , nucl phys . b * 216 * , 421 ( 1983 ) and references therein . for a review of quantum and classical creep of vortices in high-@xmath9 superconductors see g. blatter , m. v. feigelman , v. b. geshkenbein , a. i. larkin , and v. m. vinokur , rev . phys . * 66 * , 1125 ( 1994 ) ; b.v . petukhov and v.l . pokrovsky , zh . teor . fiz . * 63 * , 634 ( 1972 ) [ sov . jetp * 36 * , 336 ( 1973 ) ] . larkin and yu.n . ovchinnikov , pisma zh . * 37 * , 322 ( 1983 ) [ jetp lett . * 37 * , 382 ( 1983 ) ] . i.m . lifshitz and yu . kagan , zh . 62 * , 1 ( 1972 ) [ sov . jetp * 35 * , 206 ( 1972 ) ] . s. v. meshkov , zh . fiz . * 89 * , 1734 ( 1985 ) [ sov . jetp * 62 * , 1000 ( 1985 ) ] ; a. s. ioselevich and e. i. rashba , zh . . fiz . * 91 * , 1917 ( 1986 ) [ sov . jetp * 64 * , 1137 ( 1986 ) ] . e. m. chudnovsky , phys . a * 46 * , 8011 ( 1992 ) . d. a. gorokhov and g. blatter , phys . b * 56 * , 3130 ( 1997 ) . e. m. chudnovsky and d. a. garanin , phys . lett * 79 * , 4469 ( 1997 ) . j. garriga , phys . d * 49 * , 5497 ( 1994 ) . a. ferrera , phys . d * 52 * , 6717 ( 1995 ) . b.i . ivlev and v.i . melnikov , phys . b * 36 * , 6889 ( 1987 ) . m.a . skvortsov , phys . b * 55 * , 515 ( 1997 ) . a. krmer and m. l. kulic , phys . b * 48 * , 9673 ( 1993 ) . j. s. langer , ann . * 54 * , 258 ( 1969 ) . i. affleck , phys . lett . * 46 * , 388 ( 1981 ) . r. rajaraman , _ solitons and instantons _ ( north - holland publishing company , amsterdam , new york , oxford , 1982 ) . i. m. gelfand and a. m. yaglom , j. math . phys . * 1 * , 48 ( 1960 ) . h. kleinert , _ path integrals in quantum mechanics , statistics , and polymer physics _ ( world scientific , singapore , 1990 ) . d. r. nelson and v. m. vinokur , phys . b * 48 * , 13060 ( 1993 ) . l. d. landau and e. m. lifshitz , _ quantum mechanics , course in theoretical physics , vol . 3 ( pergamon , oxford , 1977 ) . p. hnggi , f. marchesoni , and p. sodano , phys . lett . * 60 * , 2563 ( 1988 ) . f. marchesoni , c. cattuto , and g. costantini , phys . b * 57 * , 7930 ( 1998 ) . d. a. gorokhov and g. blatter , phys . rev . b * 57 * , 3577 ( 1998 ) . k. g. selivanov , zh . . fiz . * 94 * , 57 ( 1988 ) [ sov . jetp * 67 * , 1548 ( 1988 ) ] . j. baacke and v. g. kiselev , phys . d * 48 * , 5648 ( 1993 ) . b. i. ivlev and n. b. kopnin , phys . lett * 64 * , 1828 ( 1990 ) .	we investigate the depinning of a massive elastic manifold with @xmath0 internal dimensions , embedded in a @xmath1-dimensional space , and subject to an isotropic pinning potential @xmath2 the tunneling process is driven by a small external force @xmath3 we find the zero temperature and high temperature instantons and show that for the case @xmath4 the problem exhibits a sharp transition from quantum to classical behavior : at low temperatures @xmath5 the euclidean action is constant up to exponentially small corrections , while for @xmath6 @xmath7 the results are universal and do not depend on the detailed shape of the trapping potential @xmath8 possible applications of the problem to the depinning of vortices in high-@xmath9 superconductors and nucleation in @xmath0-dimensional phase transitions are discussed . in addition , we determine the high - temperature asymptotics of the preexponential factor for the @xmath10-dimensional problem . 0.5 cm 1.6 cm preprint eth - th/98 - 10 ; accepted for publication in phys . rev .
in papers @xcite , the algebraic formalization of zwiebach invariants gives a purely algebraic construction of genus expansion of solutions of the wdvv equation in terms of ch - algebras . since we know that zwiebach invarinats induce gromov - witten invariants on subbicomlexes with zero differential , then it is a natural question to check different relation coming from geometry of the moduli space of curves . in gromov - witten theory , we represent a solution of the wdvv equation as a generating function for genus zero gromov - witten invariants of a siutable algebraic variety ( without @xmath1-classes ) . then we consider gromov - witten invariants with @xmath1-classes and in arbitrary genus . in this approach , any relation among natural strata in cohomology of the moduli space of curves gives us a differential equation for the gromov - witten potential . this property is just a corollary of the splitting axiom . in our construction , we do not have a definition that can be compared with full gromov - witten potential . we just define genus expansion with descendants ( @xmath1-classes ) only at one point . but some relations from gromov - witten theory are already nontrivial even for this reduced genus expansion . in particular , it is enough to have descendants at one point to pose a question on belorousski - pandharipande relation among codimension @xmath2 strata in @xmath0 @xcite . in this paper , we prove that our genus expansion satisfy the differential equation defined by the belorousski - pandharipande relation . in fact , it is the unique relation in genera @xmath3 that makes sense in our construction and that we have not yet checked in @xcite . since this paper is just a sequel to @xcite , we refer to these papers for the origin , motivation , and more detailed exposition of the new construction that we study here . also we mention that in this paper we keep all ealier mathematical problems of this theory such as the lack of examples and the problem of convergence . this paper is organized as follows . in section 2 , we recall our construction of genus expansion ( in fact , only its parts used in this paper ) . in section 3 , we recall the belorousski - pandharipande relation and formulate our main theorem . in the rest of the paper , we prove ( or rather outline the proof of ) our theorem . we are very grateful to a. s. losev for the numerous of helpful remarks and discussions . in our case , the belorousski - pandharipande relation is a differential equation for four different formal power series : generating functions for the correlators in genera @xmath4 , @xmath5 , and @xmath2 without descendants ( @xmath6 , @xmath7 , and @xmath8 , respectively ) , and generating function for the correlators in genus @xmath2 with one descendant at one point ( @xmath9 ) . our goal in this section is to define these four formal power series . in subsection [ ch - algebras ] we explain what is a ch - algebra and fix notations for all necessary operators in it . in subsection [ tensors ] we explain how we use graph to encode tensor expressions . in subsection [ usage ] we fix notation for all tensors in ch - algebras that we use in this paper . also we discuss there a subtlety related to the signs . then in subsection [ def - graphs ] we define @xmath6 , @xmath7 , @xmath8 , and @xmath9 , and give precise formulas ( in terms of graphs ) for the first several terms of these power series . in fact , it can also be useful to study the definition of the full potential with descendants only at one point given in @xcite . but the full definition given there is rather involved and appears to be natural only in the course of calculations in @xcite or in the framework of zwiebach invariants @xcite . in this section , we recall the definition of ch - algebras @xcite . a supercommutative associative @xmath10-algebra @xmath11 is called ch - algebra , if there are two odd linear operators @xmath12 and an integral @xmath13 satisfying the following axioms : 1 . @xmath14 ; 2 . @xmath15 , where @xmath16 and @xmath17 is represented as a direct sum of subspaces of dimension @xmath18 generated by @xmath19 for some vectors @xmath20 , i. e. @xmath21 ( hodge decomposition ) ; 3 . @xmath22 is an operator of the first order , it satisfies the leibniz rule : @xmath23 ( here and below we denote by @xmath24 the parity of @xmath25 ) ; 4 . @xmath26 is an operator of the second order , it satisfies the @xmath27-term relation : @xmath28 5 . @xmath26 satisfies the property called @xmath29-axiom : @xmath30 ( here @xmath31 and @xmath32 are the operators of multiplication by @xmath33 and @xmath34 respectively ) . we define an operator @xmath35 . we put @xmath36 ; on each subspace @xmath37 , we define @xmath38 as @xmath39 , @xmath40 , and @xmath41 . we see that @xmath42=0 $ ] ; @xmath43 $ ] is the projection to @xmath17 along @xmath44 ; @xmath45 is the projection to @xmath44 along @xmath17 . an integral on @xmath11 is an even linear function @xmath13 . we require @xmath46 , @xmath47 , and @xmath48 . these properties imply that @xmath49 , @xmath50 , and @xmath51 . we can define a scalar product on @xmath11 : @xmath52 we suppose that this scalar product is non - degenerate . using scalar product we may turn an operator @xmath53 into bivector that we denote by @xmath54 $ ] . here we explain a way to encode some tensor expressions over an arbitrary vector space in terms of graphs . consider an arbitrary graph ( we allow graphs to have leaves and we require vertices to be at least of degree @xmath55 ) . we associate a symmetric @xmath56-form to each internal vertex of degree @xmath56 , a symmetric bivector to each egde , and a vector to each leaf . then we can substitute the tensor product of all vectors in leaves and bivectors in edges into the product of @xmath56-forms in vertices , distributing the components of tensors in the same way as the corresponding edges and leaves are attached to vertices in the graph . this way we get a number . let us study an example : @xmath57 we assign a @xmath58-form @xmath59 to the left vertex of this graph and a @xmath55-form @xmath60 to the right vertex . then the number that we get from this graph is @xmath61 . note that vectors , bivectors and @xmath56-forms used in this construction can depend on some variables . then what we get is not a number , but a function . consider a ch - algebra @xmath11 . there are some standard tensors over @xmath11 , which we associate to elements of graphs below . here we introduce the notations for these tensors . we always assign the form @xmath62 to a vertex of degree @xmath56 . there is a collection of bivectors that will arise below at edges : @xmath63 $ ] , @xmath64 $ ] , @xmath65 $ ] , @xmath66 $ ] , @xmath67 $ ] , @xmath68 $ ] , and @xmath69 $ ] . in pictures , edges with these bivectors will be denoted by @xmath70 respectively . note that an empty edge corresponding to the bivector @xmath65 $ ] can usually be contracted ( if it is not a loop ) . the vectors that we will put at leaves depend on some variables . let @xmath71 be a homogeneous basis of @xmath44 . to each vector @xmath72 we associate two formal variables , @xmath73 and @xmath74 , of the same parity as @xmath72 . then we will put at a leaf either the vector @xmath75 ( denoted by an empty leaf ) or the vector @xmath76 ( denoted by an arrow at the leaf ) . there is a subtlety related to the fact that @xmath11 is a @xmath77-graded space . in order to give an honest definition we must do the following . suppose we consider a graph of genus @xmath78 . we can choose @xmath78 edges in such a way that the graph being cut at these edge turns into a tree . to each of these edges we have already assigned a bivector @xmath54 $ ] for some operator @xmath79 . now we have to put the bivector @xmath80 $ ] instead of the bivector @xmath54 $ ] , where @xmath81 is an operator defined by the formula @xmath82 . in particular , consider the following graph ( this is also an example to the notations given above ) : @xmath83 an empty loop corresponds to the bivector @xmath65 $ ] . an empty leaf corresponds to the vector @xmath84 . a trivalent vertex corresponds to the @xmath55-form given by the formula @xmath85 . if we ignore this remark , then what we get is just the trace of the operator @xmath86 . but using this remark we get the supertrace of this operator . in fact , this subtlety will play no role in this paper . it affects only some signs in calculations and all these signs will be hidden in lemmas shared from @xcite . so , one can just ignore this remark . now we describe @xmath6 , @xmath7 , @xmath8 , and @xmath9 using the notations given above . the formal power series @xmath6 ( @xmath7 , @xmath8 ) is just the sum over all trivalent graphs of genus @xmath4 ( @xmath5 , @xmath2 , respectively ) with empty leaves and edges with thick black dots . at each graph we put a coefficient equals to the inverse of the number of its automorphisms . @xmath87 the formal power series @xmath9 is the some over graphs of genus @xmath2 with edges with thick black dots satisfying some additional conditions . first , there is exactly one vertex of degree @xmath18 and all othe vertices are trivalent . second , at this vertex of degree @xmath18 , there is a leaf with arrow . third , all other leaves are empty . each graph is weighted with the inverse of the number of its automorphisms fixing the leaf with arrow . @xmath88 in fact , in order to obtain an expression for @xmath9 one can just take the expression for @xmath8 and add an additional leaf with arrow in all possible ways to each graph . also it is obvious that @xmath9 is linear in @xmath74 , @xmath89 . the belorousski - pandharipande relation is a relation in ( co)homology of @xmath0 between the cycles of natural strata of complex codimension @xmath2 in @xmath0 . below , we list the strata participating in the belorousski - pandharipande relation : @xmath90 we explain our notations . note that graphs here have completely different meaning then all other graphs in this paper . we use the language of dual graphs , that is , vertices correspond to irreducible curves , edges corespond to points of intersection , leaves correspond to marked points . a thick vertex labeled by @xmath2 corresponds to a genus two curve ; a thick vertex labeled by @xmath5 corresponds to a genus one curve . a simple vertex corresponds to a genus zero curve . an arrow on an edge or a leaf means that we take the @xmath1-class at the destination of the arrow . this way to describe strata in the moduli space of curves was introduced by e. getzler , see @xcite . for example , consider the picture of @xmath91 . a generic point of this stratum is represented by a three component curve such that one component has genus @xmath4 ; there is one marked point on these curve and two other curves intersect this one . one of the other curves has genus @xmath4 and two marked point ; another curve has genus @xmath2 and no marked points . another example . we consider the picture of @xmath92 . a generic point of this stratum is represented by a two - component curve ; one component has genus zero ; there are three marked point and one point of intersection with another curve . another curve has genus @xmath2 ; there are no marked points , but we take the @xmath1-class on this curve at the point of intersection . one more example is the picture of @xmath93 . a generic point of this stratum is represented by a two - component curve ; one curve has genus @xmath4 , two marked points , and one point of intersection with another curve . another curve has genus @xmath2 , one point of intersection with the first curve , and one marked point with @xmath1-class . we recall the belorousski - pandharipande relation : @xmath94 one can notice that the coefficients in equation ( [ bp - relation ] ) do not coinside with the coefficients of the initial relation in @xcite . this is because of two reasons . first , we do not weight the strata in the formula with the inverse order of the authomorphism group of their generic point . second , we consider each possible enumeration of marked points only once , without multiplicities . we refer to @xcite for the explanation of the conventions that we do not keep here . as we have already explained in the introduction , the belorousski - pandharipande relation gives us a differential equation for @xmath6 , @xmath7 , @xmath8 , and @xmath9 . we illustrate this correspondence with examples . if all variables are even , we have : @xmath95 @xmath96 and so on . the metric @xmath97 used here is given by the scalar product on @xmath44 . we have : @xmath98 , @xmath99 $ ] . note that @xmath92 is defined with the help of one @xmath1-class on @xmath100 . let @xmath101 be the projection forgetting all marked points except for the first one . then there is a formula relating @xmath102 and @xmath103 on @xmath104 . so , the differential expressions corresponding to these strata rely on this formula , which is exactly the first factor in expression . the same remark concerns @xmath92 , @xmath93 , and the pull - back of @xmath102 from @xmath105 . in this case , the required formula is the first factor of expression . we state our theorem . [ belorousski - pandharipande ] @xmath6 , @xmath7 , @xmath8 , and @xmath9 satisfy the belorousski - pandharipande relation . in this section we prove theorem [ belorousski - pandharipande ] . the proof is organized in two steps . first we consider the differential equation determined by the belorousski - pandharipande relation at zero point . it is proved by a straightforward calculation with tensors . then we can use the universal technique developed in @xcite . that is , for any differential equation proved at zero point by the same type of calculation as given below , we immediately obtain its proof at any point . this was done very carefully for the wdvv equation and less carefully for the getzler relation in the last section of @xcite , and the argument for the belorousski - pandharipande relation is literally the same . so , in subsections [ relation - graphs]-[other ] we discuss only the simplest case of the belorousski - pandharipande relation . in subsection [ relation - graphs ] we rewrite it in terms of graphs ; in subsection [ outline ] we explain what kind of calculation is to be done ; in subsections [ delta3 ] and [ other ] we give an example of such calculation for one stratum and discuss it for the other strata . finally , in subsection [ reduction ] we recall the basic idea from the last section of @xcite that completes the proof of any relation of this type . consider the degree zero term of the power series obtained from @xmath6 , @xmath7 , @xmath8 , and @xmath9 by the differential operator determined by @xmath106 . slightly abusing notations , we denote it also by @xmath106 . then we have : @xmath107 @xmath108 and so on . we recall that a thick white point on an edge denotes the bivector @xmath64 $ ] ( or @xmath109 $ ] ) . we explain the proof of the simplest case of theorem 1 . we have already expressed each @xmath106 at zero point in terms of graphs with one or two edges marked by @xmath64 $ ] . using the leibniz rule for @xmath22 and the @xmath27-term relation and @xmath29-axiom for @xmath26 , we get out of @xmath64 $ ] in our expressions . this way we obtain an expression of each @xmath106 in terms of @xmath110 _ final graphs_. then we substitute these expressions in belorousski - pandharipande relation ( [ bp - relation ] ) , and we see that the coefficient of each final graph in this relation is equal to @xmath4 . this proves the simplest case of our theorem . final graphs are listed in appendix a ; final expressions of @xmath106 are given in appendix b. below , we explain how to get out of @xmath64 $ ] in our graphs by the way of example ( we give detailed calculations for @xmath93 ) . we consider the right hand side of equation ( [ del3 ] ) . our goal is to get out of thick white points in these graphs . finally , we must obtain an expression in terms of graphs from appendix a. we carry out our calculations in two step . at the first step we consider each graph the right hand side of equation ( [ del3 ] ) separately . at the second step we arrange the results of the first step and obtain an expression in final graphs . we recall that @xmath111 . also we note that if we have an edge ( not a loop ) marked by @xmath112 $ ] , then we can contract this edge . we have : @xmath113 we recall that @xmath114=-g_-$ ] and @xmath115 for any @xmath116 . using these properties , the leibniz rule for @xmath22 , and taking into account the symmetries of our graphs , we have : @xmath117 therefore , @xmath118 the same calculations for all other pictures give us : @xmath119 @xmath120 in this section , we prove that some graphs in equations ( [ d3 - 1])-([d3 - 7 ] ) are equal to @xmath4 . vector @xmath121 is equal to @xmath4 . indeed , from @xmath29-axiom , it follows that @xmath122 since @xmath123 , the last vector is equal to zero . from this lemma , it is obvious that @xmath124 for any @xmath116 vector @xmath125 is equal to @xmath4 . first , we apply @xmath29-axiom , and then we apply the auxiliary lemma from @xcite . we have : @xmath126 since @xmath123 and @xmath127 , the last two vectors are equal to zero . from this lemma , it is obvious that @xmath128 in this section , we list some corollaries of the @xmath27-term relation . we have : @xmath129 we prove this formula . for convenience , we split all these graphs in the same tensor pieces . we list the notations for these tensor pieces : @xmath130 note that @xmath131 are even vectors , but @xmath132 is an odd one . equation ( [ 17 t ] ) is equivalent to @xmath133 also note that @xmath134 . then from the @xmath27-term relation , it follows that @xmath135 substituting equations ( [ lhs1 ] ) and ( [ lhs2 ] ) in the left hand side of equation ( [ xyz - f ] ) , we get : @xmath136 substituting equation ( [ rhs ] ) in the right hand side of equation ( [ xyz - f ] ) , we get exactly the same . this proves equation ( [ xyz - f ] ) and therefore equation ( [ 17 t ] ) . we prove in the same way that @xmath137 @xmath138 @xmath139 using equations ( [ del3])-([fdel3 ] ) and notations for the final graphs from appendix a , we get : @xmath140 we carry out the same calculation for all other @xmath106 . if there are two thick white points in graphs for @xmath106 , then we get out of them successively . the results of these calculations are arranged in tables in appendix b. if we substitute all these expressions for @xmath106 in belorousski - pandharipande relation , we identically get zero . this proves our theorem . for a much more detailed exposition of our calculations , see @xcite . now we explain what to do when parameters are not set to zero . in terms of graphs , this means that for any @xmath106 we are to consider graphs with the same structure as before but with an arbitrary number of additional leaves . in @xcite the authors notice that these additional leaves can be gathered in some special operators . that is , instead of considering graphs with an arbitrary number of additional leaves , we can consider the same graphs as in the simplest case , but we substitute the vectors @xmath84 and @xmath141 on leaves and bivectors @xmath63 $ ] and @xmath64 $ ] on edges with new complicated vectors and bivectors . these new vectors and bivectors depend on parameters and can be written down explicitely in terms of the barannikov - kontsevich solution of the maurer - cartan equation as it is done in @xcite . here is one subtlety related to the strata with one @xmath1-class . at this step we simultaneously have switched from @xmath1-classes to a kind of pull - backs of @xmath1-classes . but is was proved in @xcite that these pull - backs are related to @xmath1-classes via exactly the same formulas as in gromov - witten theory ! so , we take the same graphs as in the simplest case , we put new vectors on leaves and bivectors on edges , and we must prove _ exactly the same _ relation as in the simplest case . the main feature of this approach is that the properties of these new vectors and bivectors are almost the same as the properties of @xmath84 , @xmath141 , @xmath63 $ ] , and @xmath64 $ ] . so we can just repeat our argument with getting out of thick white points in graphs . we refer to the last section of @xcite for the precise formulas for these new vectors and bivectors and lemmas describing their properties . in fact , this reconstruction of the full proof works for a rather large class of differential equations in ch - algebras , in particular all possible relations coming from geometry of the moduli space of curves are definitely in this class . @xmath142 @xmath143 @xmath144 [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] 00 p. belorousski , r. pandharipande , a descendent relation in genus 2 , ann . scuola norm . ( 4 ) * 29 * ( 2000 ) , no . 1 , 171191 . e. getzler , topological recursion relations in genus @xmath2 , integrable systems and algebraic geometry ( kobe / kyoto , 1997 ) , world scientific , river edge , nj , 1998 , pp . 73106 . a. losev , s. shadrin , from zwiebach invariants to getzler relation , arxiv : math.qa/0506039 . s. shadrin , a definition of descendants at one point in graph calculus , arxiv : math.qa/0507106 . i. shneiberg , in preparation .	we prove that the genus expansion of solutions of the wdvv equation constructed from dgbv algebras satisfy the differential equation determined by the belorousski - pandharipande relation in cohomology of the moduli space of curves @xmath0 .
investigating compounds containing light elements has been a simple and powerful guiding principle for discovery of high - temperature superconductors . according to the bcs theory , @xcite the superconducting transition temperature ( @xmath0 ) is scaled by the phonon frequency and therefore light atoms are advantageous for achieving high @xmath0 . despite its simplicity , this principle has been surprisingly successful as represented by the discoveries of superconductivity in doped fullerene solids , @xcite magnesium diboride , @xcite lithium under pressure @xcite and boron - doped diamond . @xcite along this principle , possible superconductivity in compressed hydrogen and hydrogen compounds has been explored as an extreme case . @xcite recently , it has been discovered that h@xmath1s exhibits superconductivity under high pressures at 190k ( ref . ) . being the new record of the superconducting transition temperature ( @xmath0 ) , this report has immediately aroused intense debate.@xcite several facts imply that this superconducting phase is induced by the conventional mechanism due to the vibrations of hydrogen atoms : the observed @xmath0 is subject to the hydrogen isotope effect @xcite ; in prior to the experimental discovery , there was an _ ab initio _ calculation which predicted strong electron - phonon coupling @xcite ; the electronic bandwidth is so large that the migdal approximation seems valid . @xcite however , some puzzling results have also been exposed . first , the crystal structure realized in the experimental situation has not been specified . if we estimate @xmath0 of h@xmath1s using the conventional mcmillan formula @xcite with the empirical coulomb parameter @xmath7=0.13 , the calculated value is too low compared with the experimentally observed value . it has also been proposed that h@xmath2s phase instead emerges under high pressures,@xcite where the electron - phonon coupling is thought to be stronger than in h@xmath1s . @xcite second , anomalously large hydrogen isotope effect coefficient @[email protected] has been observed . although it has been hypothesized that the unharmonic effect on the lattice dynamics has some role @xcite or that different structures emerge in h@xmath1s and d@xmath1s ( sulfur di - deuteride ) , @xcite this anomaly remains an open question . to further investigate the above points , we need to address not only the electron - phonon interaction but also the electron - electron coulomb interaction in the h@xmath10s systems . accurate evaluation of the impact of the pair - breaking coulomb repulsion is vital because this governs the absolute value of @xmath0 , as well as @xmath8 . @xcite in addition , experimentally realized pressure range is rather out of that in the previous _ ab initio _ studies and therefore more thorough investigations of the pressure dependence of the superconducting properties are desired . in this article , we present an _ ab initio _ study on the superconductivity in solid h@xmath1s and h@xmath2s covering the experimental pressure range . in the standard migdal - eliashberg theory , @xcite the effect of the electron - electron coulomb interaction is practically treated with an empirical parameter @xmath7 . to incorporate this effect non - empirically , we utilize the density functional theory for superconductors ( scdft @xcite ) . with this theory , we can calculate @xmath0 and @xmath8 without any empirical parameter , which can be directly compared with the experimental data . to calculate @xmath0 from first principles , we employed the scdft gap equation given by @xmath11}{e_{n'{\bf k'}}}\!\delta_{n'\!{\bf k'}}. \label{eq : gap - eq}\end{aligned}\ ] ] here , @xmath12 and @xmath13 denote the band index and crystal momentum , respectively , @xmath14 is the gap function , and @xmath15 is the inverse temperature . the energy @xmath16 is defined as @xmath16=@xmath17 and @xmath18 is the one - electron energy with respect to the fermi level calculated with the normal kohn - sham equation . the functions @xmath19 and @xmath20 are called exchange - correlation kernels , which describe the effects of the interactions . the nondiagonal kernel @xmath20 consists of two parts @xmath20@xmath21@xmath22@xmath23@xmath24 representing the electron - phonon and electron - electron interactions , respectively . the diagonal kernel @xmath19@xmath21@xmath25 represents the mass renormalization of the normal - state band structure due to the phonon exchange . using these kernels , @xcite the conventional strong - coupling superconductivity can be treated with the level of the migdal - eliashberg theory . @xcite in particular , the electronic nondiagonal kernel @xmath24 describes the screened electron - electron coulomb interaction , where the dynamical screening effects are incorporated within the random - phase approximation . @xcite we can therefore evaluate effects of the static coulomb repulsion suppressing the pairing , as well as the plasmon superconducting mechanism . @xcite .pressure settings and the corresponding input structures for the calculations . we observed that it is difficult to achieve the numerical convergence in the phonon calculations for the calculations for h@xmath2s at 190 gpa since it is near the second - order structural transition point . @xcite [ cols="<,^,^,^,^,^,^,^",options="header " , ] 999 j. bardeen , l. n. cooper , and j. r. schrieffer , phys . rev . * 108 * , 1175 ( 1957 ) . a. f. hebard , m. j. rosseinsky , r. c. haddon , d. w. murphy , s. h. glarum , t. t. m. palstra , a. p. ramirez , and a. r. kortan , nature ( london ) * 350 * , 600 ( 1991 ) . j. nagamatsu , n. nakagawa , t. muranaka , y. zenitani , and j. akimitsu , nature ( london ) * 410 * , 63 ( 2001 ) . k. shimizu , h. ishikawa , d. takao , t. yagi , and k. amaya , nature ( london ) 419 , 597 ( 2002 ) . struzhkin , m. i. eremets , w. gan , h. k. mao , and r. j. hemley , science 298 , 1213 ( 2002 ) . e. a. ekimov , v. a. sidorov , e. d. bauer , n. n. melnik , n. j. curro , j. d. thompson , and s. m. stishov , nature ( london ) * 428 * , 542 ( 2004 ) . h. okazaki , t. wakita , t. muro , t. nakamura , y. muraoka , t. yokoya , s. kurihara , h. kawarada , t. oguchi , and y. takano , arxiv:1411.7752 . n. w. ashcroft , phys . * 21 * , 1748 ( 1968 ) . p. cudazzo , g. profeta , a. sanna , a. floris , a. continenza , s. massidda , and e. k. u. gross , phys . * 100 * , 257001 ( 2008 ) . n. w. ashcroft , phys . 92 * , 187002 ( 2004 ) . j. feng , w. grochala , t. jaro , r. hoffmann , a. bergara , and n. w. ashcroft , phys . rev . lett . * 96 * , 017006 ( 2006 ) . y. yao , j. s. tse , y. ma , k. tanaka , europhys . lett . * 78 * , 37003 ( 2007 ) . m. i. eremets , i. a. trojan , s. a. medvedev , j. s. tse , and y. yao , science * 319 * , 1506 ( 2008 ) . o. degtyareva , j. e. proctor , c. l. guillaume , e. gregoryanz , and m. hanfland , solid state commun . * 149 * , 1583 ( 2009 ) . x. jin , x. meng , z. he , y. ma , b. liu , t. cui , g. zou , and h. mao , proc . usa * 107 * , 9969 ( 2010 ) . y. li , g. gao , y. xie , y. ma , t. cui , and g. zou , proc . usa * 107 * , 15708 ( 2010 ) . j. a. flores - livas , m. amsler , t. j. lenosky , l. lehtovaara , s. botti , m. a. l. marques , and s. goedecker , phys . lett . * 108 * , 117004 ( 2012 ) . t. scheler , o. degtyareva , m. marqus , c. l. guillaume , j. e. proctor , s. evans , and e. gregoryanz , phys . b * 83 * , 214106 ( 2011 ) . x. f. zhou , a. r. oganov , x. dong , l. zhang , y. tian , and h. t. wang , phys . b * 84 * , 054543(2011 ) . d. y. kim , r. h. scheicher , c. j. pickard , r. j. needs , and r. ahuja , phys . lett . * 107 * , 117002 ( 2011 ) . j. s. tse , y. yao , and k. tanaka , phys . * 98 * , 117004 ( 2007 ) . g. gao , a. r. oganov , p. li , z. li , h. wang , t. cui , y. ma , a. bergara , a. o. lyakhov , t. iitaka , and g. zou , proc . usa * 107 * , 1317 ( 2010 ) i. goncharenko , m. i. eremets , m. hanfland , j. s. tse , m. amboage , y. yao , and i. a. trojan , phys . lett . * 100 * , 045504 ( 2008 ) . g. gao , a. r. oganov , a. bergara , m. martinez - canales , t. cui , t. iitaka , y. ma , and g. zou , phys . * 101 * , 107002 ( 2008 ) . r. szczniak , a. p. durajski , and d. szczniak , solid state commun . * 165 * , 39 ( 2013 ) . g. gao , h. wang , a. bergara , y. li , g. liu , and y. ma , phys . rev . b * 84 * , 064118 ( 2011 ) . r. szczniak and a. p. durajski , supercond . * 27 * , 015003 ( 2014 ) . h. wang , j. s. tse , k. tanaka , t. iitaka , and y. ma , proc . natl . acad . usa * 109 * , 6463 ( 2012 ) . d. c. lonie , j. hooper , b. altintas , and e. zurek , phys . b * 87 * , 054107 ( 2013 ) . z. wang , y. yao , l. zhu , h. liu , t. iitaka , h. wang , and y. ma , j. chem . phys . * 140 * , 124707 ( 2014 ) . d. zhou , x. jin , x. meng , g. bao , y. ma , b. liu , and t. cui , phys . b * 86 * , 014118 ( 2012 ) . d. y. kim . r. h. scheicher , h. mao , t. w. kang , and r. ahuja , proc . usa * 107 * , 2793 ( 2010 ) . g. gao , r. hoffmann , n. w. ashcroft , h. liu , a. bergara , and y. ma , phys.rev . b * 88 * , 184104 ( 2013 ) . y. li , j. hao , h. liu , y. li , and y. ma , j. chem . phys . * 140 * , 174712 ( 2014 ) . d. duan , y. liu , f. tian , d. li , x. huang , z. zhao , h. yu , b. liu , w. tian , and t. cui , sci . reports * 4 * , 6968 ( 2014 ) . drozdov , m. i. eremets , and i. a. troyan , arxiv:1412.0460 . j. e. hirsch and f. marsiglio , physica c * 511 * , 45 ( 2015 ) . a. p. durajski , r. szczniak , and y. li , physica c * 515 * , 1 ( 2015 ) . n. bernstein , c. s. hellberg , m. d. johannes , i. i. mazin , and m. j. mehl , phys . b * 91 * , 060511(r ) ( 2015 ) . d. duan , x. huang , f. tian , d. li , h. yu , y. liu , y. ma , b. liu , and t. cui , phys . b * 91 * , 180502(r ) ( 2015 ) . d. a. papaconstantopoulos , b. m. klein , m. j. mehl , and w. e. pickett , phys . b * 91 * , 184511 ( 2015 ) . a. b. migdal , sov . jetp * 7 * , 996 ( 1958 ) . p. b. allen and r. c. dynes , phys . b * 12 * , 905 ( 1975 ) . p. morel and p. w. anderson , phys 125 * , 1263 ( 1962 ) . j. w. garland , phys . 11 * , 114 ( 1963 ) j. p. carbotte , rev . * 62 * , 1027 ( 1990 ) . g. m. eliashberg , sov . jetp * 11 * , 696 ( 1960 ) ; d. j. scalapino , in _ superconductivity _ edited by r. d. parks , ( marcel dekker , new york , 1969 ) volume 1 ; j. r. schrieffer,_theory of superconductivity ; revised printing _ , ( westview press , colorado , 1971 ) . m. lders , m. a. l. marques , n. n. lathiotakis , a. floris , g. profeta , l. fast , a. continenza , s. massidda , and e. k. u. gross , phys . rev . b * 72 * , 024545 ( 2005 ) . m. a. l. marques , m. lders , n. n. lathiotakis , g. profeta , a. floris , l. fast , a. continenza , e. k. u. gross , and s. massidda , phys . rev . b * 72 * , 024546 ( 2005 ) . r. akashi and r. arita , phys . * 111 * , 057006 ( 2013 ) . r. akashi and r. arita , j. phys . soc . jpn . * 83 * , 061016 ( 2014 ) . y. takada , j. phys . . jpn . * 45 * , 786 ( 1978 ) . j. p. perdew , k. burke , and m. ernzerhof , phys . 77 * , 3865 ( 1996 ) . p. giannozzi , s. baroni , n. bonini , m. calandra , r. car , c. cavazzoni , d. ceresoli , g. l. chiarotti , m. cococcioni , i. dabo , a. dal corso , s. fabris , g. fratesi , s. de gironcoli , r. gebauer , u. gerstmann , c. gougoussis , a. kokalj , m. lazzeri , l. martin - samos , n. marzari , f. mauri , r. mazzarello , s. paolini , a. pasquarello , l. paulatto , c. sbraccia , s. scandolo , g. sclauzero , a. p. seitsonen , a. smogunov , p. umari , and r. m. wentzcovitch , j. phys . : condens . matter 21 , 395502 ( 2009 ) ; http://www.quantum-espresso.org/ s. baroni , s. de gironcoli , a. dal corso , and p. giannozzi , rev . mod . phys . * 73 * , 515(2001 ) . r. akashi and r. arita , phys . b * 88 * , 014514 ( 2013 ) . r. akashi , k. nakamura , r. arita , and m. imada , phys . b * 86 * , 054513 ( 2012 ) . m. kawamura , y. gohda , and s. tsuneyuki , phys . b * 89 * , 094515 ( 2014 ) . we employed the approximation @xmath26@xmath27@xmath28 for the numerical @xmath29-integral with @xmath30@xmath2170 ev . the high - frequency contribution @xmath31 was estimated analytically with the approximation @xmath32@xmath27@xmath33 . m. methfessel and a. t. paxton , phys . b * 40 * , 3616 ( 1989 ) . a. floris , a. sanna , s. massidda , and e. k. u. gross , phys . b * 75 * , 054508 ( 2007 ) . a. floris , g. profeta , n. n. lathiotakis , m. lders , m. a. l. marques , c. franchini , e. k. u. gross , a. continenza , and s. massidda , phys . rev . lett . * 94 * , 037004 ( 2005 ) ; a. floris , a. sanna , m. lders , g. profeta , n. n. lathiotakis , m. a. l. marques , c. franchini , e. k. u. gross , a. continenza , and s. massidda , physica c * 456 * , 45 ( 2007 ) . a. sanna , g. profeta , a. floris , a. marini , e. k. u. gross , and s. massidda , phys . b * 75 * , 020511(r ) ( 2007 ) . g. profeta , c. franchini , n. n. lathiotakis , a. floris , a. sanna , m. a. l. marques , m. lders , s. massidda , e. k. u. gross , and a. continenza , phys . rev . lett . * 96 * , 047003 ( 2006 ) . j. a. flores - livas , a. sanna , and e. k. u. gross , arxiv:1501.06336 . in ref . ( i ) they used modified forms for the phonon part of the scdft kernels , though the detail of the modification has been unpublished yet ( see ref . 56 in ref . ) , ( ii ) they did not include the plasmon - induced dynamical effect , and ( iii ) their calculation was based on the local - density approximation , @xcite which yielded a @xmath5-@xmath6 structural transition point slightly different from that with the generalized - gradient approximation . @xcite d. m. ceperley and b. j. alder , phys . rev . lett . * 45 * , 566 ( 1980 ) . j. p. perdew and a. zunger , phys . b * 23 * , 5048 ( 1981 ) . i. errea , m. calandra , c. j. pickard , j. nelson , r. j. needs , y. li , h. liu , y. zhang , y. ma , f. mauri , phys . * 114 * , 157004 ( 2015 ) . n. troullier and j. l. martins , phys . b * 43 * , 1993 ( 1991 ) ; http://www.abinit.org/downloads/psp-links/psp-links/gga_fhi . j. rath and a. j. freeman , phys . b * 11 * , 2109 ( 1975 ) .	we calculate superconducting transition temperatures ( @xmath0 ) in sulfur hydrides h@xmath1s and h@xmath2s from first principles using the density functional theory for superconductors . at pressures of @xmath3 150 gpa , the high values of @xmath0 ( @xmath4130 k ) observed in the recent experiment [ a. p. drozdov , m. i. eremets , and i. a. troyan , arxiv:1412.0460 ] are accurately reproduced by assuming that h@xmath1s decomposes into @xmath5-h@xmath2s and s. for the higher pressures , the calculated @xmath0s for @xmath6-h@xmath2s are systematically higher than those for @xmath5-h@xmath2s and the experimentally observed maximum value ( 190 k ) , which suggests the possibility of another higher-@xmath0 phase . we also quantify the isotope effect from first principles and demonstrate that the isotope effect coefficient can be larger than the conventional value ( 0.5 ) when multiple structural phases energetically compete .
our ability to measure the star - formation rate density ( sfrd ) at @xmath19 relies almost entirely on either narrow - band surveys which detect strong ly@xmath20 emitting galaxies @xcite or deep imaging surveys of uv bright lyman - break galaxies @xcite . these surveys , by virtue of being flux limited , trace the bright end of the galaxy luminosity function down to @xmath21 l@xmath22 . the various observations have revealed a decline in the sfrd by about a factor of 3 between @xmath23 and @xmath16 @xcite , with much of this decline being due to the evolution of the bright end of the galaxy luminosity function . more than @xmath24 of the estimated sfrd at these redshifts takes place in sub - l@xmath22 galaxies . in addition , spectroscopic confirmation of high redshift galaxies relies on ly@xmath20 emission , which is easily obscured by dust . there is now increasing evidence for rapid dust production within @xmath25 gyr of the big bang @xcite . as a result , quantifying possible dust extinction corrections and measuring the faint - end slope of the galaxy luminosity function is essential for minimizing uncertainties in the high redshift sfrd . measurements of the metallicity in typical star - forming galaxies at @xmath19 is beyond the technological capability of the current generation of instrumentation . the relevant rest - frame optical emission lines are very weak and are redshifted to the mid - ir . the alternative approach of studying chemical enrichment through damped ly@xmath20 absorbers ( dlas ) detected against background quasars appears to be limited to @xmath26 @xcite , and is biased towards tracing the properties of extended halo gas , which at lower redshift , significantly underestimates the disk metallicity . as a result , the apparent evolution in the mass - metallicity and luminosity - metallicity relations ( @xmath27-@xmath28 and @xmath29-@xmath28 ) from @xmath30 to @xmath31 ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) can not be traced to @xmath32 , where such information should shed light on the initial stages of mass build - up and metal enrichment in galaxies . long duration grbs , by virtue of being associated with the deaths of massive stars , provide a complementary technique for measuring the sfrd and the chemical enrichment history . has revolutionized this study by detecting grbs out to @xmath16 @xcite . prompt spectroscopy of the bright afterglows has now provided a sample of @xmath33 grbs over a wide range of redshifts with a wealth of metal absorption features arising in the host galaxy ( e.g * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? these observations provide a unique window into the metallicity and gas column density in star forming environments at high redshifts @xcite . once the afterglows fade away , deep observations of the field can also reveal the stellar mass and star - formation rate of the host galaxies , which can then be correlated with the inferred metallicities . in order to study the host galaxies of high redshift grbs and take advantage of the diagnostics afforded by grbs , we present _ spitzer space telescope _ 3.6 observations of the hosts of three grbs at @xmath34 . building on the constraints provided by @xcite on the host galaxy of grb050904 at @xmath35 , we discuss the nature of the host galaxies , redshift evolution of the luminosity - metallicity relation and provide an independent measure of the high redshift sfrd for comparison with estimates from lyman - break galaxy samples . throughout this paper , we adopt a @xmath36 , @xmath37 , h@xmath38=71 km s@xmath13 mpc@xmath13 cosmology . as part of program go20000 ( pi : berger ) we observed the fields of grbs 060223a ( @xmath39 ; berger et al . , in prep . ) , 060510b ( @xmath1 ; * ? ? ? * ) , and 060522 ( @xmath40 ; berger et al . , in prep . ) with the infrared array camera ( irac ; * ? ? ? * ) in the bandpasses centered at 3.6 and 5.8 @xmath3 m ( table 1 ) . the observations were undertaken between september and november 2006 , after the afterglows associated with the grbs had faded below the detectability threshold . as shown in table 1 , the grb fields lie in regions with low " to `` medium''-level zodiacal background and cirrus of 13@xmath4128 mjy / sr at 24 @xmath3 m on the date of the observations . we used 100 s integrations with about 130 medium scale dithers from the random cycling pattern for total on - source integration times of @xmath42 s at each passband . the nominal @xmath43 point source sensitivity limits are 0.26 and 2.4 @xmath3jy , respectively . starting with the s14.4.0 pipeline - processed basic calibrated data ( bcd ) sets we corrected the individual frames for muxbleed and column pull down using software developed for the great observatories origins deep survey ( goods ) . due to the presence of bright stars in the field , many of the frames at 3.6 @xmath3 m also showed evidence for `` muxstriping '' . this was removed using an additive correction on a column by column basis . the processed bcd frames were then mosaiced together using the mopex routine @xcite and drizzled onto a @xmath44 grid . astrometry was performed with respect to the brightest 2mass stars in the field which showed a peak - to - peak astrometric uncertainty of @xmath45 at 3.6 @xmath3 m . the location of the grb hosts was determined by aligning the _ spitzer _ images against images of the afterglow from the uv / optical telescope ( 060223a and 060522 ) and the gemini multi - object spectrograph on the gemini - north 8-m telescope ( 060510b ) . for the latter , the astrometric uncertainty is @xmath46 in each coordinate , while using the uvot images we obtain an astrometric uncertainty of about @xmath44 . all three grb locations show the presence of nearby ( @xmath47 ) brighter galaxies : grb060223a has two sources with flux densities of 7.2 and 9.1 @xmath3jy at distances of @xmath48 and @xmath49 from the grb position ; grb060510b has a source with a flux of 6.1 @xmath3jy about 3.1@xmath50 from the grb position ; and grb060522 has a source with 0.72 @xmath3jy located 1.6@xmath50 away from the burst position . this is not unexpected given the high source density in deep irac images . we subtracted the contribution of these sources , in order to obtain the strongest possible constraints on the flux from the grb host galaxies . photometry at the position of the host galaxies was performed in fixed circular apertures of @xmath51 radius with appropriate beam size corrections applied as stated in the observer s manual . we clearly detect a galaxy coincident with the position of grb060510b with a flux density of @xmath2 @xmath3jy at 3.6 @xmath3 m , and a @xmath43 upper limit of 2.4 @xmath3jy at 5.8 @xmath3 m ( figure 1 ) . for grbs 060223a and 060522 , due to blending from nearby brighter sources and the residual effects from muxbleed , we are only able to provide @xmath43 upper limits to the flux of the host galaxy ( table 1 ) . of the 4 grb host galaxies at @xmath0 observed in this program at 3.6 and 5.8 , ( including grb050904 ) , only grb060510b is clearly detected . the observed 3.6 flux densities / limits for these galaxies correspond to rest - frame @xmath4band luminosities of @xmath52 l@xmath53 , where l@xmath53 is about @xmath54 l@xmath6 @xcite . it is illustrative to compare the properties of grb hosts with the field galaxy population at similar redshifts . the goods fields have extensive spectroscopy of galaxies at high redshift @xcite . there are 275 lyman - break galaxies in both the goods fields which are classified as @xmath4band `` dropouts '' i.e. @xmath17 . the magnitude limit of the goods optical observations imply that they are brighter than @xmath55 l@xmath22 @xcite , of these , @xmath5620% have spectroscopic redshifts while @xmath5630% are individually detected with irac . at higher redshifts , @xmath57 , it has been shown that galaxies which are individually undetected with irac appear to harbor a younger stellar population and have a factor of 10 lower stellar mass than irac detected galaxies @xcite . as shown in figure 2 , grb host galaxies are factors of @xmath58 times fainter than the median @xmath4band luminosity of galaxies which have spectroscopic redshifts of @xmath59 in the goods field . furthermore , the luminosities are comparable to the rest - frame @xmath4band luminosity of grb hosts studied at lower redshifts ( e.g. ; * ? ? ? * ; * ? ? ? this suggests that grb host galaxies are unlike the luminous end of the star - forming , lyman - break galaxy population which have had about a factor of 10 increase in their stellar mass between @xmath17 and @xmath60 . they are more typical of the blue , faint end of the galaxy @xmath4band luminosity function , a population for which it is difficult to measure redshifts or metallicities , in the absence of grbs , due to their inherent faintness . grb host galaxies at @xmath61 , which have extensive multi - wavelength data , show clear evidence for very high specific star formation rates indicating an on - going starburst @xcite . we do not yet have constraints on the star formation rates in the @xmath0 host galaxies presented here , due to their intrinsic faintness in the rest - frame uv ( see e.g. * ? ? ? * ; * ? ? ? * ) . however , spectroscopy of the afterglows by @xcite and berger et al . ( in prep . ) has revealed a wealth of absorption lines which have been used to derive the metallicity and gas column density in the vicinity of the burst . absorption spectroscopy of the three bursts presented here have yielded neutral hydrogen gas densities in their host galaxies of : @xmath62=21.6\pm 0.1 $ ] ( 060223a ) , @xmath63=21.3\pm 0.1 $ ] ( 060510b ) , and @xmath63=21.0\pm 0.3 $ ] ( 060522 ) . thus , all three systems are clearly dlas , with column densities near the median of the distribution for grb - dlas @xcite . in addition , the metallicities of the grb 060223a and 060510b systems have been determined from the detection of weak metal lines . for grb060522 the signal - to - noise of the spectrum is too low to clearly identify any metal lines and an estimate of the metallicity is thus not possible . in the case of grb060223a , we find an upper limit on the column density of of @xmath64<15.3 $ ] , leading to a metallicity of @xmath65}<-1.45 $ ] . the non - detection of @xmath66 leads to a limit of @xmath67}<-2.65 $ ] , but we stress that iron can be heavily depleted onto dust grains . from the detection of the @xmath68 line we find @xmath64 \approx 15.3 $ ] , and hence @xmath69}\approx -1.8 $ ] . as in the case of iron , silicon is also strongly depleted , so we conclude that the metallicity of the grb060223a dla is in the range of @xmath70 to @xmath71 . for grb060510b , we use the @xmath72 lines to measure @xmath73=15.6\pm 0.1 $ ] , and hence a metallicity , @xmath74}=-0.85\pm 0.15 $ ] ( see also * ? ? ? * ) . the metallicity estimates of the grb hosts along with their rest - frame @xmath75band luminosities ( assuming a @xmath76 color of 0 , typical of star - forming galaxies ) are shown in figure 3 . also shown for comparison are the metallicity - luminosity relationships for different samples of field galaxies . despite the one detection and two limits for the luminosity of the host galaxies , the figure shows that the redshift evolution of metallicity at a fixed @xmath75band luminosity that is seen between @xmath77 , clearly extends out to @xmath17 . the chemical enrichment of galaxies is directly related to their past history of star - formation since supernovae and stellar winds are responsible for recycling the products of nucleosynthesis back into the interstellar medium . the stellar mass density is the time integral of the star - formation history . by comparing the redshift evolution of the stellar mass density ( @xmath78 ) ( e.g * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , with the redshift evolution of the metallicity ( @xmath28 ) , we can search for evolution of the stellar initial mass function and assess the role of feedback in the build - up of galaxies . the _ spitzer _ observations of the hosts are crucial , since they enable metallicity comparisons to be made at a fixed rest - frame @xmath4band luminosity , over a wide range of redshifts . due to the fact that we have constraints on the @xmath4band luminosity and metallicity of only one @xmath17 grb host , we make the assumption that the median metallicity at each redshift , is that of a galaxy which has a similar luminosity as the grb host . this is not an unreasonable assumption . within the observational uncertainties , the slope of the mass - metallicity relation appears to be invariant between @xmath18 and @xmath79 @xcite . the metallicity values are obtained by effectively making a vertical cut at @xmath80 mag in figure 3 and are determined to be @xmath810.15 , @xmath820.1 and @xmath830.1 dex at redshifts of 5 , 2.3 and 0 respectively . the average estimated @xmath78 at these redshifts are 1.4 , 6 and 56 in units of 10@xmath84 m@xmath6 mpc@xmath14 ( see references above ) . we performed a monte - carlo analysis to obtain the best fit between @xmath85 and @xmath86 . star - forming galaxies which fall on the local mass - metallicity relationship , show a scatter of @xmath560.1 dex at bright luminosities and @xmath560.2 dex at faint luminosities @xcite . we use a random number generator to offset the stellar mass density and metallicity by the observed scatter from the mean values quoted above ( see * ? ? ? * table 3 for the range in stellar mass density ) . we fit for the relation between @xmath87 and repeat the process 10000 times . we find that @xmath88 appears to be invariant between @xmath89 and that @xmath90 . this suggests that the chemical enrichment of star - forming galaxies takes place at a slower rate than the build up of stellar mass . this is presumably due to the loss of metals from low - mass galaxies by outflows and stellar winds , an effect which is primarily responsible for the mass - metallicity relation seen in the local universe @xcite and @xmath79 lyman - break galaxies @xcite . however , alternate mechanisms such as depletion of metals onto dust grains can not be ruled out at this time . there is the possibility of a selection effect in this analysis . if long duration grbs arise in collapsars , they might preferentially be in low - metallicity galaxies . as a result , it is possible that grb hosts have a lower metallicity than the average field galaxy of the same rest - frame optical luminosity . although grb hosts appear to be have low luminosities in the rest - frame uv and @xmath4band , the observational evidence does not indicate that the hosts have an unusually low - metallicity for their luminosity . metallicity of grb host galaxies appear to span the range 0.1@xmath411 z@xmath91 ( e.g * ? ? ? * ; * ? ? ? * ) and some of the hosts have even been found to be associated with dusty , infrared luminous galaxies ( e.g * ? ? ? nevertheless , we assess the reliability of our derived @xmath87 relation by considering a bias in the metallicity of grb environments . if we assume that the metallicity of the grb environment is higher by @xmath920.3 dex compared to the mean metallicity of a galaxy at its luminosity , it implies that the mean metallicity at @xmath17 for a field galaxy at the luminosity of the grb host is [email protected] . the best fit relation to the three points is then consistent with an exponent of unity i.e. @xmath94 but has a worse @xmath95 . the corollary is that if grb hosts were biased by 0.3 dex towards lower metallicities , compared to the mean metallicity of a galaxy at its luminosity , the best fit relation is @xmath96 which is a larger deviation from unity . furthermore , if there were a bias in grb host metallicities , the slope of the @xmath87 relation derived above at a fixed @xmath75band luminosity , would have a different value between @xmath97 and @xmath77 due to the fact that the @xmath77 relation is determined from star - forming galaxies while the @xmath97 relation is derived from grb hosts and lyman - break galaxies . this is inconsistent with our fits , although larger samples of grb hosts are needed to eliminate suggestions of bias . detection of individual grb hosts at high redshifts is likely to remain difficult , due to their intrinsic faintness . there is a clear need for homogeneous infrared surveys of grb host galaxies which will enable stacking to be performed as a function of metallicity , gas density and rest - frame ultraviolet properties . within our sample , grb050904 is a marginal irac detection @xcite , while grb060223a is dominated by detector systematics . as a result , we are unable to provide additional constraints using stacking . observations of a larger sample of grb hosts , such as those currently being targeted in program go4 - 40599 ( pi : chary ) , will allow the luminosity - metallicity relation to be measured at high redshift and lead to a better understanding of the faint end of the galaxy luminosity function , a regime which is currently inaccessible even through ultradeep surveys like goods and the udf . it is now well known from various mid - infrared , far - infrared and submillimeter surveys , that the star - formation rate density at @xmath98 is dominated by infrared luminous galaxies with l@xmath99=l@xmath100 l@xmath6 and l@xmath99/l@xmath101 ( e.g * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . at @xmath102 , current long - wavelength surveys , due to their limited sensitivity , are unable to detect galaxies which harbor the bulk of the star - formation . thus , rest - frame ultraviolet observations of galaxies are the only avenue for probing star - formation at high redshifts . the primary uncertainties associated with quantifying the sfrd at @xmath103 , are the contribution from galaxies at the faint end of the uv - luminosity function and dust corrections . since sub - l@xmath104 galaxies contribute @xmath5690% of the sfrd , measurement of the faint - end slope of the ultraviolet luminosity function , where completeness corrections and surface brightness dimming issues are significant , needs to be undertaken carefully @xcite . similarly , if extinction were a significant issue , the galaxies that dominate the star - formation rate density would be uv - faint or undetected in magnitude limited rest - frame ultraviolet surveys . grbs are relatively insensitive to these limitations . if the grb rate density were correlated with the co - moving star - formation rate density at lower redshifts , where cross - calibration between the uv and ir are in broad agreement , measurement of the grb rate density at @xmath103 could provide an independent pathway to quantifying the sfrd ( see also e.g. ; * ? ? ? the three parameters which are most likely to dominate the calibration between grbs and the star - formation rate density are the evolution of metallicity with redshift , evolution of the initial mass function of stars and identification and spectroscopic follow up of the grb afterglow . if long duration grbs were to preferentially occur in low metallicity environments , the increase in the average metallicity of the universe with decreasing redshift would result in a higher sfr / grb - rate ratio at low redshift . similarly , evolution of the stellar initial mass function from a top - heavy " to a salpeter mass function with decreasing redshift would increase the sfr / grb - rate ratio at low redshift . on the other hand , the detection efficiency and spectroscopic completeness of grbs should be increasing with decreasing redshift , implying a lower sfr / grb - rate ratio at low redshift . calibrating each of these parameters individually is challenging at the present time , partly because the relationship between grb rate and environment is not well known and due to the fact that observational selection effects can not be quantified . therefore , we need to rely on empirical comparisons between known star - formation rate estimates and grb rate densities to assess grbs as a star - formation rate indicator . this empirical comparison can be optimally done at @xmath105 since in this redshift range the star - formation rate , including the dust obscured component , has been accurately determined from deep mid - infrared and submillimeter surveys . we use the star - formation rate density at @xmath105 from @xcite . we distribute the 52 grbs with spectroscopic redshifts into redshift bins and divide by the co - moving volume in each redshift bin . we also correct for the time dilation to estimate the comoving grb rate density over the @xmath106 year lifetime . the redshift bin at @xmath107 is omitted since the grb rate density appears to be anomalously high compared to the rapidly evolving star - formation rate density . we find that within the uncertainties , the rate density of grbs with spectroscopic redshifts in the range @xmath108 is constant at a value of @xmath109 mpc@xmath14 yr@xmath13 . this can be compared with the extinction - corrected comoving star - formation rate density in the same redshift range which is in the range [email protected]@xmath6 yr@xmath13 mpc@xmath14 and has an average value of @xmath55 m@xmath6 yr@xmath13 mpc@xmath14 @xcite . since these two independent rate densities are relatively constant in the @xmath108 range , we can tentatively make the assumption that the sfr / grb - rate is constant ( figure 4 ) . the ratio of these two rates implies : @xmath110 where sfrd is the extinction - corrected star - formation rate density in m@xmath6 yr@xmath13 mpc@xmath14 and grb rate is in units of mpc@xmath14 yr@xmath13 . using our derived calibration , and the measured grb rate densities at @xmath111 and @xmath112 of @xmath113 and @xmath114 mpc@xmath14 yr@xmath13 , respectively , we infer a net star - formation rate , corrected for extinction , of @xmath115 and @xmath12 m@xmath6 yr@xmath13 mpc@xmath14 at @xmath15 and @xmath16 respectively . these estimates are systematically higher than those derived by @xcite by factors of @xmath116 . the @xcite estimates were calibrated at @xmath117 where neither the completeness correction factor for the faint end of the uv luminosity function nor the dust extinction correction are reliably known while deep _ spitzer _ mid - infrared surveys have confirmed the dominant contribution of infrared luminous galaxies to the star - formation rate density at @xmath108 @xcite . the fact that the grb rate density is almost flat between @xmath118 , while parameters such as the detection efficiency and spectroscopic completeness should be decreasing with increasing redshift , implies that the measured grb rate density provides at least a lower limit to the star - formation rate density . it is illustrative to compare this star formation rate estimate with those from deep rest - frame ultraviolet surveys at @xmath19 . @xcite derive a star - formation rate density at @xmath119 of 0.02 m@xmath6 yr@xmath13 mpc@xmath14 when integrating to 0.2 @xmath120 . after application of an extinction correction of @xmath121 mag , based on the extinction properties in local starburst galaxies , they estimate the total star - formation rate density at @xmath122 to be 0.15 m@xmath6 yr@xmath13 mpc@xmath14 . similarly , @xcite derive a star - formation rate density at @xmath16 by integrating the luminosity function of lyman - break galaxies in the udf and other deep fields . integrating the uv luminosity function down to 0.2@xmath123 results in a value of @xmath124 m@xmath6 yr@xmath13 mpc@xmath14 while the integral to @xmath125 @xmath126 yields a sfrd of 0.04 m@xmath6 yr@xmath13 mpc@xmath14 . application of an extinction correction , inferred to be about @xmath127=0.45 mag at @xmath57 , to this latter number , implies a sfrd of 0.06m@xmath6 yr@xmath13 mpc@xmath14 . the agreement between the sfrd values estimated from ultraviolet surveys and the grb rate density is reassuring , considering that there have been only 8 grbs that have been spectroscopically confirmed to be at @xmath19 ( figure 4 ) . however , the sfrd from grbs primarily traces the faint end of the galaxy luminosity function while the surveys are measuring the contribution from the bright end . as a result , a more reasonable sfrd estimate requires adding the sfrd contribution estimated from the faint end of the galaxy luminosity function , from grbs , to that from bright lbgs . grb hosts are fainter than 0.2 @xmath128 . based on the uv to @xmath4band flux ratios of star - forming galaxies at @xmath117 , it implies that grb hosts must be fainter than 0.2@xmath129 adding the sfrd from @xmath130 galaxies to that inferred from the grb rate density results in an extinction corrected sfrd of @xmath131 and @xmath132 m@xmath6 yr@xmath13 mpc@xmath14 at @xmath15 and @xmath16 , respectively . if confirmed through a larger statistical sample , this is a substantial upward revision suggesting that @xmath133 galaxies contribute at least four times as much to the star - formation rate density at @xmath57 as the bright end ( @xmath130 ) of the uv luminosity function . indirectly , this implies that the faint end slope of the uv luminosity function at @xmath57 must be @xmath134 , compared to the value of @xmath135 that was derived by @xcite . grbs are a powerful tool for measuring the high redshift star - formation rate density . in particular , deep _ spitzer _ observations of grb hosts can reveal the contribution to the star - formation rate density from the faint end of the galaxy luminosity function , a regime which is inaccessible to deep , rest - frame ultraviolet / near - infrared surveys . increasing the sample of high redshift grbs will reduce the uncertainties in the star - formation rate density unaffected by extinction and through stacking analysis on the host galaxies will help estimate the contribution to the stellar mass density from sub-@xmath136 galaxies . comparison between star - formation rate estimates from grbs with those from deep uv surveys will provide better constraints on the evolution of dust extinction at high redshift and provide tremendous insights into the chemical enrichment of the early universe . we wish to thank mark dickinson for his comments which strengthened the arguments in this paper . we also acknowledge the extensive resources that are invested by the entire grb community , not all of whom can be cited here , which enables prompt imaging and spectroscopic follow - up of the bursts . this work is based on observations made with the _ spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa . support for this work was provided by nasa through an award issued by jpl / caltech . eb acknowledges support by nasa through hubble fellowship grant hst-01171.01 awarded by stsci , which is operated by aura , inc . , for nasa under contract nas5 - 26555 . , e. , chary , r. , cowie , l. l. , price , p. a. , schmidt , b. p. , fox , d. b. , cenko , s. b. , djorgovski , s. g. , soderberg , a. m. , kulkarni , s. r. , mccarthy , p. j. , gladders , m. d. , peterson , b. a. , & barger , a. j. 2006 , arxiv astrophysics e - prints , d. , prez - gonzlez , p. g. , tyler , k. d. , rieke , g. h. , buat , v. , takeuchi , t. t. , lauger , s. , arnouts , s. , ilbert , o. , barlow , t. a. , bianchi , l. , lee , y .- w . , madore , b. f. , malina , r. f. , szalay , a. s. , & yi , s. k. 2006 , , 450 , 69 , e. , dickinson , m. , morrison , g. , chary , r. , cimatti , a. , elbaz , d. , frayer , d. , renzini , a. , pope , a. , alexander , d. m. , bauer , f. e. , giavalisco , m. , huynh , m. , kurk , j. , & mignoli , m. 2007 , arxiv e - prints , 705 , a. s. , levan , a. j. , strolger , l. , vreeswijk , p. m. , thorsett , s. e. , bersier , d. , burud , i. , castro cern , j. m. , castro - tirado , a. j. , conselice , c. , dahlen , t. , ferguson , h. c. , fynbo , j. p. u. , garnavich , p. m. , gibbons , r. a. , gorosabel , j. , gull , t. r. , hjorth , j. , holland , s. t. , kouveliotou , c. , levay , z. , livio , m. , metzger , m. r. , nugent , p. e. , petro , l. , pian , e. , rhoads , j. e. , riess , a. g. , sahu , k. c. , smette , a. , tanvir , n. r. , wijers , r. a. m. j. , & woosley , s. e. 2006 , , 441 , 463 , j. p. u. , starling , r. l. c. , ledoux , c. , wiersema , k. , thne , c. c. , sollerman , j. , jakobsson , p. , hjorth , j. , watson , d. , vreeswijk , p. m. , mller , p. , rol , e. , gorosabel , j. , nrnen , j. , wijers , r. a. m. j. , bjrnsson , g. , castro cern , j. m. , curran , p. , hartmann , d. h. , holland , s. t. , jensen , b. l. , levan , a. j. , limousin , m. , kouveliotou , c. , nelemans , g. , pedersen , k. , priddey , r. s. , & tanvir , n. r. 2006 , , 451 , l47 , m. , dickinson , m. , ferguson , h. c. , ravindranath , s. , kretchmer , c. , moustakas , l. a. , madau , p. , fall , s. m. , gardner , j. p. , livio , m. , papovich , c. , renzini , a. , spinrad , h. , stern , d. , & riess , a. 2004 , , 600 , l103 , j. , mller , p. , gorosabel , j. , fynbo , j. p. u. , toft , s. , jaunsen , a. o. , kaas , a. a. , pursimo , t. , torii , k. , kato , t. , yamaoka , h. , yoshida , a. , thomsen , b. , andersen , m. i. , burud , i. , castro cern , j. m. , castro - tirado , a. j. , fruchter , a. s. , kaper , l. , kouveliotou , c. , masetti , n. , palazzi , e. , pedersen , h. , pian , e. , rhoads , j. , rol , e. , tanvir , n. r. , vreeswijk , p. m. , wijers , r. a. m. j. , & van den heuvel , e. p. j. 2003 , , 597 , 699 , p. , fynbo , j. p. u. , ledoux , c. , vreeswijk , p. , kann , d. a. , hjorth , j. , priddey , r. s. , tanvir , n. r. , reichart , d. , gorosabel , j. , klose , s. , watson , d. , sollerman , j. , fruchter , a. s. , de ugarte postigo , a. , wiersema , k. , bjrnsson , g. , chapman , r. , thne , c. c. , pedersen , k. , & jensen , b. l. 2006 , , 460 , l13 , b. l. , fynbo , j. u. , gorosabel , j. , hjorth , j. , holland , s. , mller , p. , thomsen , b. , bjrnsson , g. , pedersen , h. , burud , i. , henden , a. , tanvir , n. r. , davis , c. j. , vreeswijk , p. , rol , e. , hurley , k. , cline , t. , trombka , j. , mcclanahan , t. , starr , r. , goldsten , j. , castro - tirado , a. j. , greiner , j. , bailer - jones , c. a. l. , kmmel , m. , & mundt , r. 2001 , , 370 , 909 , n. , kosugi , g. , aoki , k. , yamada , t. , totani , t. , ohta , k. , iye , m. , hattori , t. , aoki , w. , furusawa , h. , hurley , k. , kawabata , k. s. , kobayashi , n. , komiyama , y. , mizumoto , y. , nomoto , k. , noumaru , j. , ogasawara , r. , sato , r. , sekiguchi , k. , shirasaki , y. , suzuki , m. , takata , t. , tamagawa , t. , terada , h. , watanabe , j. , yatsu , y. , & yoshida , a. 2006 , , 440 , 184 , e. , duc , p .- a . , mirabel , i. f. , sanders , d. b. , bosch , g. , diaz , r. j. , donzelli , c. j. , rodrigues , i. , courvoisier , t. j .- l . , greiner , j. , mereghetti , s. , melnick , j. , maza , j. , & minniti , d. 2003 , , 400 , 499 , t. , murayama , t. , maiolino , r. , marconi , a. , kashikawa , n. , ajiki , m. , hattori , t. , ly , c. , malkan , m. a. , motohara , k. , ohta , k. , sasaki , s. s. , shioya , y. , & taniguchi , y. 2007 , arxiv astrophysics e - prints , e. , cristiani , s. , dickinson , m. , kuntschner , h. , nonino , m. , rettura , a. , rosati , p. , vernet , j. , cesarsky , c. , ferguson , h. c. , fosbury , r. a. e. , giavalisco , m. , grazian , a. , haase , j. , moustakas , l. a. , popesso , p. , renzini , a. , stern , d. , & the goods team . 2006 , , 454 , 423 , p. m. , ellison , s. l. , ledoux , c. , wijers , r. a. m. j. , fynbo , j. p. u. , mller , p. , henden , a. , hjorth , j. , masi , g. , rol , e. , jensen , b. l. , tanvir , n. , levan , a. , castro cern , j. m. , gorosabel , j. , castro - tirado , a. j. , fruchter , a. s. , kouveliotou , c. , burud , i. , rhoads , j. , masetti , n. , palazzi , e. , pian , e. , pedersen , h. , kaper , l. , gilmore , a. , kilmartin , p. , buckle , j. v. , seigar , m. s. , hartmann , d. h. , lindsay , k. , & van den heuvel , e. p. j. 2004 , , 419 , 927 grb060223a & , & 4.406 & 2006 sep 23 & 24.8 & 130@xmath137100s & @xmath1380.3 & @xmath1382.4 + grb060510b & , & 4.942 & 2006 oct 26 & 13.4 & 130@xmath137100s & [email protected] & @xmath1382.1 + grb060522 & , & 5.110 & 2006 nov 23 & 28.4 & 138@xmath137100s & @xmath139 & @xmath1382.4 +	we present deep 3.6 observations of three @xmath0 grb host galaxies with the _ spitzer space telescope_. the host of grb060510b , at @xmath1 , is detected with a flux density of @xmath2 @xmath3jy , corresponding to a rest - frame @xmath4band luminosity of @xmath5 l@xmath6 , or @xmath7 l@xmath8 . we do not detect the hosts of grbs060223a and 060522 and constrain their rest - frame @xmath4band luminosity to @xmath9 l@xmath8 . our observations reveal that @xmath0 grb host galaxies are a factor of @xmath10 less luminous than the median luminosity of spectroscopically - confirmed @xmath0 galaxies in the great observatories origins deep survey ( goods ) and the _ hubble _ ultra deep field ( udf ) . the strong connection between grbs and massive star formation implies that not all star - forming galaxies at these redshifts are currently being accounted for in deep surveys and grbs provide a unique way to measure the contribution to the star - formation rate density from galaxies at the faint end of the galaxy luminosity function . by correlating the co - moving star - formation rate density with co - moving grb rates at lower redshifts , we estimate a lower limit to the star - formation rate density of @xmath11 and @xmath12 m@xmath6 yr@xmath13 mpc@xmath14 at @xmath15 and @xmath16 , respectively . this is in excellent agreement with extinction corrected estimates from lyman - break galaxy samples . finally , our observations provide initial evidence that the metallicity of star - forming galaxies evolve more slowly than the stellar mass density between @xmath17 and @xmath18 , probably indicative of the loss of a significant fraction of metals to the intergalactic medium , especially in low - mass galaxies .
attempts to determine the statistics of intensity and polarization fluctuations of cosmic microwave background ( cmb ) renewed interest to the fluctuations of galactic foreground radiation ( see tegmark et al . 2000 ) . spectra of intensity of synchrotron emission and synchrotron polarization ( see papers in de oliveira - costa & tegmark 1999 ) as well as starlight polarization ( fosalba et al . 2002 ; henceforth flpt ) have been measured . those measurements revealed a range of power - laws , the origin of which has not been addressed as far as we know . in tegmark et al . ( 2000 ) there is an allusion that the spectra may be relevant to kolmogorov turbulence , but the issue of how those different spectra may arise has not been addressed . interstellar medium is turbulent and kolmogorov - type spectra were reported on the scales from several au to several kpc ( see armstrong et al . 1995 ; lazarian & pogosyan 2000 ; stanimirovic & lazarian 2001 ) . therefore it is natural to think of the turbulence as the origin of the fluctuations of the diffuse foreground radiation . interstellar medium is magnetized with magnetic field making turbulence anisotropic . it may be argued that although the spectrum of mhd turbulence exhibits scale - dependent anisotropy if studied in the system of reference defined by the local magnetic field ( goldreich & sridhar 1995 ; lithwick & goldreich 2001 ; cho & lazarian 2002 ) , in the observer s system of reference the spectrum will show only moderate scale - independent anisotropy . thus from the observer s point of view kolmogorov description of interstellar turbulence is acceptable in spite of the fact that magnetic field is dynamically important and even dominant ( see discussion in lazarian & pogosyan 2000 ; cho , lazarian & vishniac 2002 ) . it is customary for cmb studies to expand the foreground intensity over spherical harmonics @xmath0 , @xmath1 , and write the spectrum in terms of @xmath2 . the measurements indicate that angular power spectra ( @xmath3 ) of the galactic emission follows power law ( @xmath4 ) ( see flpt and references in 3 ) . this paper tests whether the measured spectra are compatible with the theoretical predictions of spatial statistics that is expected in the presence of mhd turbulence . analytical studies in this direction were done in lazarian ( 1992 , 1994 , 1995 ) and here we supplement them with numerical simulations of the synthetic spectra . in this section we show that , when the angle between the lines of sight is small ( i.e. @xmath5 ) , the angular spectrum @xmath3 has the same slope as the 3-dimensional energy spectrum of turbulence . here @xmath6 is the typical size of the largest energy containing eddies , which is sometime called as outer scale of turbulence or energy injection scale , and @xmath7 is the distance to the farthest eddies ( see fig . [ fig_all]a ) . to illustrate the problem consider the observations with lines of sight being parallel . the observed intensity is the intensity summed along the _ line of sight _ , @xmath8 . @xmath9 rearranging the order of summation and using @xmath10 , we get @xmath11 which means fourier component of @xmath12 is @xmath13 . when the angular size of the observed region ( @xmath14 in radian ) on the sky is small , @xmath3 is approximately the ` energy ' spectrum of @xmath15 ( bond & efstathiou 1987 ; hobson & majueijo 1996 ; seljak 1997 ) , which means @xmath16 with @xmath17 and @xmath18 . the analysis of the geometry of crossing lines of sight is more involved , but for power - law statistics it follows from lazarian & shutenkov ( 1990 ) that if @xmath19 , then the ` energy ' spectrum of @xmath15 is also @xmath20 . therefore , we have @xmath21 in the small @xmath22 limit from the observation of the correlation function @xmath23 in real space ( or @xmath24 on the sky ) . when the three - dimensional spectrum of turbulence follows a power law ( @xmath25 ) , @xmath26 , where @xmath27 is a constant . however , when the slope of the turbulence spectrum is steeper than @xmath28 , this relation breaks down and @xmath29 regardless of the turbulence slope . when we infer @xmath3 from this @xmath23 , we obtain @xmath30 regardless of the true slope , when the slope is steeper than @xmath31 . ] . for kolmogorov turbulence ( @xmath32 ) , we expect @xmath33 note that @xmath34 . following lazarian & shutenkov ( 1990 ) , we can show that the correlation function for @xmath35 , @xmath36 where we change variables : @xmath37 is clear from fig . 1b ; we accounted for the jacobian of which is @xmath38 . we can understand @xmath39 behavior qualitatively from fig . when the angle is large , points along of the lines - of - sight near the observer are still correlated . these points extend from the observer over the distance @xmath40 . in the limit of @xmath41 we get the angular power spectrum @xmath3 using fourier transform : @xmath42 where @xmath43 , @xmath44 is the bessel function , and we use @xmath45 . in summary , for kolmogorov turbulence , we expect from equations ( [ eq_5 ] ) and ( [ eq_7 ] ) that @xmath46 which means that the power index @xmath47 of @xmath3 is . ] the critical number @xmath49 depends on the size of the size of the large turbulent eddies and on the direction of the observation . if in the naive model we assume that turbulence is homogeneous along the lines of sight and has @xmath50 corresponding to a typical size of the supernova remnant for disk with @xmath51 kpc , we get @xmath52 . for the synchrotron halo , @xmath53 kpc ( see smoot 1999 ) and we get @xmath54 . recent statistical studies of _ total _ synchrotron intensity include haslam all - sky map at 408 mhz that shows that the power index @xmath47 is in the range between 2.5 and 3 ( tegmark & efstathiou 1996 ; bouchet , gispert , & puget 1996 ) . parkes southern galactic plane survey at 2.4 ghz suggests shallower slope : giardino et al . ( 2002 ) obtained @xmath55 after point source removal and baccigalupi et al . ( 2001 ) obtained @xmath56 to @xmath57 . using rhodes / hartrao data at 2326 mhz , giardino et al . ( 2001 ) obtained @xmath58 for all - sky data and @xmath59 for high galactic latitude regions with @xmath60 . the rough tendency that follow from these data is that @xmath47 which is close to @xmath61 for the galactic plane gets steeper ( to @xmath62 ) for higher latitudes . in other words @xmath63 which differs from naive expectations given by equation ( [ eq_1_11_3 ] ) . can the difference be due to non - linear law of synchrotron emissivity ? for synchrotron radiation , emissivity at a point @xmath64 , @xmath65 , where @xmath66 is the electron number density , @xmath67 is the component of magnetic field perpendicular to the line of sight . the index @xmath68 is approximately @xmath61 for radio synchrotron frequencies ( see smoot 1999 ) . if electrons are uniformly distributed over the scales of magnetic field inhomogeneities , the spectrum of intensity reflects the statistics of magnetic field . for small amplitude perturbations ( @xmath69 ; this is true for scales several times smaller than @xmath6 when we interpret @xmath70 as local mean magnetic field strength ) , if @xmath71 has a power - law behavior , the statistics of intensity will have the same power - law behavior ( see deshpande et al . therefore emissivity non - linearity does not account for the difference between the observed spectra and the theoretical expectations . to address the problem , we perform simple numerical calculations for galactic disk and halo . we obtain @xmath3 using the relation @xmath72 we use gauss - legendre quadrature integration ( see szapudi et al . 2001 for its application to cmb ) to obtain @xmath3 . we numerically calculate the angular correlation function from @xmath73 where @xmath74 and assume that the spatial correlation function @xmath23 follow kolmogorov statistics : @xmath75 where @xmath76 is a constant . [ fig_all]c and fig . [ fig_all]d illustrate the agreement of our calculations with the theoretical expectations within the naive model of the disk and the halo from the previous section . to make the spectrum closer to observations we need to consider more realistic models . first , synchrotron emission is stronger in spiral arms , and therefore we have more synchrotron emission coming from the nearby regions . second , if synchrotron disk component is sufficiently thin , then lines of sight are not equivalent and effectively nearby disk component contributes more . indeed , when we observe regions with low galactic latitude , the effective line - of - sight varies with galactic latitude . suppose we observe a region with @xmath77 . then emission from @xmath78 is substantially weaker than that from @xmath79 , because the region with @xmath78 is @xmath80 above the galactic plane and , therefore , has weak emission . to incorporate this effect , we use the weighting function @xmath81 , which gives more weight on closer distance . the resulting angular power spectrum ( fig . [ fig_all]e ) shows a slope similar to -2 . for halo , the simple model predicts that @xmath82 , but observations provide somewhat less steep spectra . is this discrepancy very significant ? the spectrum of magnetic field is expected to be shallower than @xmath83 in the vicinity of the energy injection scale and at the vicinity of the magnetic equipartition scale . the observed spectrum also gets shallower if @xmath7 gets larger . for instance beuermann , kanbach , & berkhuijsen ( 1985 ) reported the existence of thick radio halo that extends to more than @xmath84 near the sun . finally , filamentary structures and point sources can make the spectrum shallower as well . further research should establish the true reason for the discrepancy . polarized radiation from dust is an important component of galactic foreground that strongly interferes with intended cmb polarization measurements ( see lazarian & prunet 2001 ) . flpt attempted to predict the spectrum of the polarized foreground from dust and obtained @xmath85 for starlight polarization . this spectrum is different from those discussed in the previous sections . to relate this spectrum to the underlying turbulence we should account for the following facts : a ) the observations are done for the disk component of the galaxy , b ) the sampled stars are at different distances from the observer with most of the stars close - by . to deal with this problem we use numerical simulations again . we first generate a three ( i.e. _ x , y _ , and _ z _ ) components of magnetic field on a two - dimensional plane ( @xmath86 grid points representing @xmath87 ) , using the following kolmogorov three - dimensional spectrum : @xmath88 if @xmath89 , where @xmath90 . our results show that how we continuously extend the spectrum for @xmath91 does not change our results . to simulate the actual distribution of stars within the sample used in flpt , we scatter our emission sources using the following probability distribution function : @xmath92 the starlight polarization is due to the difference in absorption cross section of non - spherical grains aligned with their longer axes perpendicular to magnetic field ( see review by lazarian 2000 ) . in numerical calculations we approximate the actual turbulent magnetic field by a superposition of the slabs with locally uniform magnetic field in each slap and assume that the difference in grain absorption parallel and perpendicular to magnetic field results in the 10% difference in the optical depths @xmath93 and @xmath94 for a slab . we calculate evolution of stocks parameters of the starlight within the slab and use the standard transforms of stocks parameters from one slab to another ( dolginov , gnedin , & silantev 1996 ; see similar expressions in martin 1972 ) . we show the result in fig . [ fig_all]f . for comparison , we also calculate the degree of polarization assuming all stars are at the same distance of @xmath95 . the result shows that , for a mixture of nearby and faraway stars , the slope steepens and gets very close to the observed one , i.e. @xmath96 . our conclusions may be tested if stars at particular distance only are correlated . if those stars are nearby , we would expect to have steeper index and it will become shallow if only distant stars are chosen . choosing only nearby stars we expect to get a steeper index . a systematic study of this change can provide an estimate of the energy injection scale @xmath6 . from the point of view of foreground studies , we conclude that a naive use of the polarization template produced with the random sample of stars may be misleading . in this paper we have addressed the origin of spatial fluctuations of emissivity and polarization of galactic diffuse emission . we have shown that mhd turbulence can qualitatively explain observed properties of total synchrotron emission and starlight polarization . the variety of measured spatial spectra of synchrotron emission can be explained by the inhomogeneous distribution of emissivity , structure of galactic disk and halo , and/or various energy injection scales . similarly , mhd turbulence plus inhomogeneous distribution of stars can explain the reported scaling of starlight polarization statistics . evidently more systematic studies are required . those studies will not only give insight into how to separate cmb from foregrounds , but also would provide valuable information on the structure of interstellar medium and the sources / energy injection scales of interstellar turbulence . bouchet , f.r . , gispert , r. , & puget , j.l . 1996 , in unveiling the cosmic infrared background , aip conf . 348 , ed . e. dwek ( baltimore : aip ) , 225 beuermann , k. , kanbach , g. , & berkhuijsen , e. 1985 , a&a , 153 , 17 cho , j. & lazarian , a. 2002 , phy . , submitted cho , j. , lazarian , a. , & vishniac , e. t. 2002 , in _ simulations of magnetohydrodynamic turbulence in astrophysics _ , eds . by t. passot & e. falgarone ( springer lecture notes in physics ; 2002 ) , submitted dolginov , a. z. , gnedin , iu . n. , & silantev , n. a. 1996 , _ propagation and polarization of radiation in cosmic media _ , ( gordon & breech ) fosalba , p. , lazarian , a. , prunet , s. , & tauber , j.a . 2002 apj , 564 , 762 ( flpt ) lazarian , a. 1992 , astron . & astrophys . trans . , 3 , 33 lazarian , a. 1994 , ph . d. thesis ( univ . of cambridge , uk ) lazarian , a. 1995 , a&a , 293 , 507 lazarian , a. 2000 , in _ cosmic evolution and galaxy formation _ , asp v.215 , ed . by j. franco , e. terlevich , o. lopez - cruz , i. aretxaga ( astron . pacific ) p.69 ( astro - ph/0003414 ) lazarian , a. & pogosyan , d. 2000 , apj , 537 , 720l lazarian , a. & prunet , 2001 , in _ microwave foregrounds _ , asp conf . 181 , ( san francisco : asp ) , p32 lazarian , a. & shutenkov , v. p. 1990 , sov . astron . lett . , 16 , 297	measurements of intensity and polarization of diffuse galactic synchrotron emission as well as starlight polarization reveal power law spectra of fluctuations . we show that these fluctuations can arise from magnetohydrodynamic ( mhd ) turbulence in the galactic disk and halo . to do so we take into account the converging geometry of lines of sight for the observations when the observer is within the turbulent volume . assuming that the intensity of turbulence changes along the line of sight , we get a reasonable fit to the observed synchrotron data . as for the spectra of polarized starlight we get a good fit to the observations taking into account the fact that the observational sample is biased toward nearby stars .
nowadays , the coherent state has found widespread application in many branches of physics such as non linear optics , laser , nuclear and particle physics . in the non - relativistic limit , the coherent states are well known and one can produce them using three different ways @xcite : * _ they can be constructed as eigenstates of the annihilation operator . _ * _ they can be defined as quantum states with minimum uncertainty relationship . _ * _ they can be obtained by operating the glauber s displacement operator on the vacuum state . _ the above definitions are known to be equivalent for the schrdinger equation with the harmonic - oscillator potential . in references [ 2 - 12 ] coherent states for systems other than the harmonic - oscillator are also investigated . to extend the _ coherence _ to relativistic region one may use manifestly covariant equations @xcite . of course in this way , it is difficult to describe the wave functions in terms of four vectors . thus , in many areas of physics , one may use the schrdinger equation with relativistic corrections or the klein - gordon ( kg ) or dirac equation with specific potentials to describe a relativistic particle or even bound state in a non - covariance manner . to describe the relativistic coherent states using the kg or dirac equation one encounters particle pair creation that makes one particle interpretation nonsense . to avoid this problem one should consider the theory under conditions that the pair creation is impossible . furthermore in relativistic quantum mechanics the consistency of the one particle description is limited and in this case the only valid operators are _ even _ operators which do not mix different charge states . in this direction authors of reference @xcite have studied a spin 0 charged particle in two cases : free particle and a particle in a constant homogeneous magnetic field . in this paper we examine the possibility of coherence for kg equation with a pure scalar potential and adopt the definition @xmath0 to extend the coherence to the relativistic region . such a potential describes a position dependent valence and conduction - band edge of semiconductors near the @xmath1 or @xmath2 point in the brillouin zone . in fact the position dependent mass characterizes a position dependent band gap . examples for such materials are given in ref . + in section 2 we obtain @xmath3 and @xmath4 operators in the feshbach - villars representation for a general scalar potential . subsequently we derive eigensolutions of a linear scalar potential in the schrdinger representation to show that pair creation does not occur . consequently we construct the eigenstates of the annihilation operator for this potential . the time evolution of the states is numerically discussed in section 3 . in section 4 we introduce a potential with equally spaced energy levels and examine their coherence with time . the resolution of unity for the obtained states is also discussed in this section . in order to explore the coherence of a relativistic particle in the simplest case we consider the kg equation in one dimension with a vector and a scalar potential ( @xmath5)[15,16 ] @xmath6\psi(x , t)=0,\ ] ] where @xmath7 is the mass of the particle , @xmath8 is the scalar potential and @xmath9 is the time component of a 4-vector potential . in non - relativistic limit , eq.(1 ) leads to a schrdinger equation with an effective potential as @xmath10 . in fact this equation gives all relativistic corrections to the schrdinger equation with such an effective potential . for a pure scalar potential we have @xmath11 in which @xmath12 now by introducing two functions @xmath13 and @xmath14 and the ansatz @xmath15 we transform eq.([kg1 ] ) to the schrdinger form as follows @xmath16\psi , \label{schkg}\ ] ] where @xmath17 , @xmath18 , are the pauli matrices and @xmath19 is a two components column vector as @xmath20 it is easy to show that each component of @xmath19 individually satisfies the kg equation i.e. eq.([kg1 ] ) . if we define @xmath21 as eigenstate of the hamiltonian @xmath22 with corresponding eigenvalue @xmath23 with the ansatz @xmath24 } , \label{ansa}\ ] ] one obtains the positive and negative energy solutions as follows @xmath25 } , \label{un}\ ] ] where @xmath26 are the appropriate normalization constants and @xmath27 now an operator @xmath28 in which @xmath29 can transform the schrdinger representation to the feshbach - villars representation therefore it can be shown that the hamiltonian in this representation is @xmath30 which is even and the even parts of operators @xmath3 and @xmath4 are also the same as the corresponding operators in the non - relativistic region : @xmath31 for x - operator in momentum space and @xmath32 for p - operator in configuration space . now we are ready to study the coherent states for potentials without pair creation . first , we choose a linear potential as follows @xmath33 where @xmath34 is a coupling constant . introducing parameters @xmath35 and @xmath36 , by using the ansatz of eq.([ansa ] ) , eq.([kg1 ] ) can be written in a familiar form -the schrdinger equation with a harmonic - oscillator potential . therefore one can easily find @xmath37},\ ] ] in which @xmath38 is the hermite polynomial of order @xmath39 and @xmath40 where @xmath41 . one can easily see from eq.([hoe ] ) that pair creation in this case never occurs . for a pure scalar potential the kg equation is independent of the sign of @xmath42 . in fact the scalar interaction is independent of the charge of the particle and has the same effect on particles and anti - particles . thus , naively speaking , this kind of potential can describe the confining part of the potential of a quarkonium . since the even part of an operator is a measurable quantity therefore eq.([xfv ] ) and eq.([pfv ] ) allow us to consider the usual annihilation operator for a non - relativistic harmonic oscillator potential to construct the coherent state for our linear potential . therefore the coherent state @xmath43 at t=0 in the feshbach - villars representation in terms of the energy eigenstate for a non - relativistic harmonic oscillator @xmath44 , can be obtained as @xmath45 for positive energy state and @xmath46 for negative energy state . thus the state given in eq.([alfa ] ) , when we restrict ourselves to the positive energy states , evolves as @xmath47 where @xmath48 . although @xmath43 is coherent , its time evolution is not the eigenstate of the annihilation operator . indeed , this is due to the non - equally - spaced of the energy levels given in eq.([hoe ] ) . + one of the significant properties of the coherent states is an over completeness relation or resolution of unity as follows @xmath49 where the measure @xmath50 for the obtained states in this section ( i.e. eq.([alfa ] ) ) can be easily determined as @xmath51d[im \alpha]\over \pi } .\label{meas}\ ] ] to explore the behavior of the solutions given in eq.([alfat ] ) we first , examine the expectation values of the appropriate quantities as follows @xmath52 } { \|\alpha\|^{2n}\over n!},\ ] ] @xmath53 { \|\alpha\|^{2n}\over n!},\ ] ] @xmath54 @xmath55 { \|\alpha\|^{2n}\over n!},\ ] ] @xmath56 @xmath57 { \|\alpha\|^{2n}\over n!},\ ] ] where @xmath58 time variations of @xmath59 , @xmath60 and @xmath61 , for various values of @xmath62 , are illustrated in figs.(1)-(11 ) . + for @xmath63 the value of @xmath64 oscillates between 0.50 and 0.51 and the state is nearly coherent , fig.(1 ) . in contrast @xmath64 for @xmath65 increases with time though its value for long time is bounded and oscillates about a certain value , figs.(2 - 3 ) . + the numerical values of @xmath59 and @xmath60 in both cases are not constant as shown in figs.(4 - 7 ) . when @xmath66 they oscillate almost periodically about their minimum values . furthermore , their oscillations are exactly out of phase and the states are squeezed , figs.(4 - 5 ) . when @xmath62 increases to @xmath67 , the values of @xmath59 and @xmath60 increase with time and finally oscillate about a finite value , figs.(6 - 7 ) . + the expectation values of @xmath3 and @xmath4 for @xmath66 and @xmath68 are plotted in figs.(8 - 9 ) and figs.(10 - 11 ) , respectively . in the case @xmath66 the values @xmath69 and @xmath70 oscillate with an amplitude and a period nearly constant which is very similar to a classical harmonic oscillator . again for @xmath68 the situation is exactly different and the values of @xmath71 and @xmath70 oscillate with a variable period and amplitude . + in the evaluation of the various quantities we have summed terms up to @xmath72 and assumed @xmath73 , though the general behaviors do not depend on @xmath34 . in the preceding section we saw that a relativistic scalar particle with a linear scalar potential , in general , is not coherent . the main reason for deviation from coherence is the fact that the eigenenergies are not equally spaced . now , we would like to construct potentials with equally - spaced energy levels . for this purpose we rewrite @xmath22 as follows @xmath74 where @xmath75 . we now introduce a potential as follows @xmath76 in the region @xmath77 and is infinite elsewhere , or in the same region @xmath78 where @xmath79 is a constant . substituting this potential in eq.([kgpt ] ) and using ansatz ( [ ansa ] ) equation ( [ kg1 ] ) leads to the non - relativistic poschl - teller ( pt ) equation @xcite and consequently @xmath21 in eq.([un ] ) can be obtained as @xmath80^{1\over 2 } ( \cos\omega x)^{1\over 2}p_{n+\lambda -{1\over 2}}^{{1\over 2}-\lambda}(\sin\omega x)\ ] ] and for the energy eigenvalues one has @xmath81 where @xmath82 it should be noted that although in the relativistic pt potential , eq.([pt1 ] ) , the energy levels are equally - spaced but the pt potential in the schrdinger equation has non - equally - spaced levels [ 3,4 ] . here again pair creation does not occur and one - particle sector of the theory is applicable . now we need the annihilation operators to construct the coherent states . to this end , one can use the schrdinger method to determine the ladder operators as follows @xcite : @xmath83^{1\over2}\mp { 1\over \omega}(\cos\omega x){d\over dx}\ ] ] where @xmath84 is the hamiltonian of the schrdinger equation for the pt potential . therefore in the feshbach - villars representation we have @xmath85 where @xmath86 is the wave function in the feshbach - villars representation @xmath87},\ ] ] in which @xmath88 stands for positive and negative energy solutions , respectively , and @xmath89^{1\over 2}.\ ] ] let us now determine @xmath90 as the eigenstate of the annihilation operator @xmath91 , with eigenvalue @xmath62 , by using an expansion on the states @xmath92 : @xmath93 we then have @xmath94 therefore @xmath95^{1\over 2}c_n,\ ] ] and , finally @xmath96 can be obtained as @xmath97^{1\over 2}{\psi}^{fv}_n(t=0),\label{alpt}\ ] ] where @xmath98^{-{1\over 2}},\ ] ] and @xmath99 now we examine the evolution of @xmath96 at every instant . for this purpose we restrict ourselves to the positive energy states which implies @xmath100^{1\over 2}e^{-ie_nt}{\psi}^{fv}_n(t=0),\label{alptt}\ ] ] where @xmath101 is given in eq.([pten ] ) . in this case since the eigenenergies are equally spaced one has @xmath102 where @xmath103 therefore , the coherent state wave packet remains an eigenvector of @xmath91 with an eigenvalue @xmath104 . now we investigate the resolution of unity for the obtained states . as ( [ alpt ] ) shows the measure in this case is different from the standard form of ( [ meas ] ) but one can follow the method of references @xcite and @xcite to find @xmath105d[im \alpha]\over \pi},\label{measpt}\ ] ] where @xmath106 can be determined by solving the following equation @xmath107 to summarize , in the feshbach - villars representation we obtained @xmath3 and @xmath4 operators for a general scalar potential . the even parts of these operators coincide with their counterparts in the schrdinger representation , see eqs.([xfv],[pfv ] ) . consequently , we constructed the annihilation operator coherent state for a purely linear scalar potential for a relativistic spinless particle . the properties of these states in @xmath108 dimension are : * the eigenfunctions for the small eigenvalues eq.([eigen ] ) are quasi - coherent and squeezed . furthermore @xmath71 and @xmath70 oscillate like a classical harmonic oscillator . in the other words for the small values of @xmath62 , which is not the non - relativistic limit of eq.([kg1 ] ) , the states regain their coherence though the relativistic corrections destroy the coherence of harmonic potential in the non - relativistic schrdinger equation . * for both small and large eigenvalues the minimum uncertainties do not increase unbounded with time . in fact this means that wave packet does not spread unbounded . * the expectation values of @xmath3 and @xmath4 for large @xmath62 oscillate with decreasing amplitude and approach zero which is equal to the values of @xmath71 and @xmath70 in the energy basis . we also introduced the relativistic poschl - teller potential and showed that its energy levels are equally - spaced , eqs.([pt1],[pten ] ) . we obtained the eigenfunctions of the annihilation operator for this potential , eq.([alpt ] ) . it is consequently shown that the time evolution of the obtained functions are still eigenstates of the annihilation operator , eq.([alptt ] ) . + * acknowledgment * the authors would like to thank b. ferdowsian for his contribution to this work at preliminary stage . financial support of isfahan university of technology and institute for studies in theoretical physics and mathematics ( ipm ) is gratefully acknowledged . r. j. glauber , phys . 131 ( 1963 ) 2766 . m. martin nieto , phys . a17 ( 1978 ) 1273 , phys . a20 ( 1979 ) 700 . m. martin nieto and d. rodney truax , new j. phys . 2 ( 2000 ) 18.1 . m. martin nieto and l. m. simmons , phys . d20 ( 1979 ) 1332 . v. p. gutschick and m. martin nieto , phys . d22 ( 1980 ) 403 . c. g. christopher , j. phys . a17 ( 1984 ) 737 . t. fukui and n. aizawa , phys . a180 ( 1978 ) 308 . w. m. zhang , d. h. feng and r. gilmore , rev . 62 ( 1990 ) 867 . i. l. cooper , j. phys . a25 ( 1992 ) 1671 . p. majumdar and h. s. sharatchandra , quant - ph/9708010 . b. molnar and m. g. benedict , quant - ph/9812041 . s. nouri , phys . a60 ( 1999 ) 1702 . m. mathur and d. sen , j. math 42 ( 2001 ) 4181 . v. aldaya and gucrrcro , j. math . 36 ( 1995 ) 3191 . b. i. lev , a. a. semenov , c. v. usenko and j. r. klauder , phys . a66 ( 2002 ) 022115 . r. dornhaus , g. nimtz and b. schlicht , narrow - gap semiconductors , springer tracts in modern physics vol.98 , ( springer - verlag , 1983 ) . g. junker , supersymmetric methods in quantum and statistical physics , ( springer - verlag , 1997 ) . f. dominguez - adame , phys . lett . a136 ( 1989 ) 175 . w. greiner , relativistic quantum mechanics ( springer - verlag , 1990 ) . g. poschl and e. teller , z. phys . 83 ( 1933 ) 143 . v. i. manko , g. marmo , e. c. g. sudarshan and f. zaccaria , phys . 55 ( 1997 ) 528 .	the klein - gordon equation with scalar potential is considered . in the feshbach - villars representation the annihilation operator for a linear potential is defined and its eigenstates are obtained . although the energy levels in this case are not equally - spaced , depending on the eigenvalues of the annihilation operator , the states are nearly coherent and squeezed . the relativistic poschl - teller potential is introduced . it is shown that its energy levels are equally - spaced . the coherence of time evolution of the eigenstates of the annihilation operator for this potential is evaluated . epsf.tex -2.0 cm = cmbx10 = cmr10 = cmti10 = cmbx10 scaled1 = cmr10 scaled1 = cmti10 scaled1 = cmbx9 = cmr9 = cmti9 = cmbx8 = cmr8 = cmti8 = cmr7
the thompson group @xmath0 is a very fascinating mathematical object @xcite . despite the many various attempts , after many years the question whether it is amenable remains unsettled ( see e.g. @xcite ) . it is well known that @xmath0 admits some nice pictorial representation , as one may describe its elements by pairs of rooted planar binary trees with the same number of leaves . in a very recent paper ( @xcite ) , devoted to the search of connections between ( the planar algebra description of ) subfactors and the ( vertex algebra description of ) conformal field theory , v. jones gave some new insight on the mathematical structure of @xmath0 by considering it as a suitable replacement of the group of diffeomorphisms of the circle , and examining its role in the possible constructions of conformal models . proceeding along these lines , and analyzing carefully the mathematical setup , perhaps the most surprising conclusion in @xcite is that it is possible to take @xmath0 as a replacement of the familiar braid groups in a description of knots and links , thus opening a new line of research on the interplay between @xmath0 and low - dimensional topology . indeed , to each element of @xmath0 one can associate a suitable graph and a related link diagram in a way that we briefly recall below . especially , by exploiting this construction , it is provided evidence that certain well - known polynomial invariants , suitably normalized , define positive definite functions of @xmath0 and thus give rise to unitary representations . this is further illustrated with a concrete example by looking at the chromatic polynomial ( * ? ? ? * proposition 5.2.1 ) . in the present paper we take up this point of view . among other examples , we show that , up to a normalization , the evaluation of the kauffman bracket at some specific roots of unity provides again a positive definite function on @xmath0 . we also discuss similar statements for the tutte polynomial and the @xmath2-colouring of links . therefore , the specializations of these functions also yield unitary representations of the thompson group @xmath0 . in this short note we only consider the case of unoriented link diagrams . however , a similar treatment involving oriented link diagrams should also be possible . the thompson group @xmath0 can be defined by the following finite presentation @xmath3 where @xmath4 for @xmath5 . in an alternative picture , @xmath0 can be seen as a particular subgroup of the group of homeomorphisms of the interval @xmath6 $ ] . indeed , it is generated by the following homeomorphisms ( see @xcite for further details ) @xmath7 t+\frac{1}{4 } & \frac{1}{4 } \leq t\leq \frac{1}{2 } \\[0.5ex ] \frac{t}{2}+\frac{1}{2 } & \frac{1}{2 } \leq t\leq 1 \end{array}\right . \quad x_1=\left\{\begin{array}{ll } t & 0\leq t\leq \frac{1}{2 } \\[0.5ex ] 2t-\frac{1}{2 } & \frac{1}{2 } \leq t\leq \frac{5}{8 } \\[0.5ex ] t+\frac{1}{8 } & \frac{5}{8 } \leq t\leq \frac{3}{4 } \\[0.5ex ] \frac{t}{2}+\frac{1}{2 } & \frac{3}{4 } \leq t\leq 1 \end{array}\right.\end{aligned}\ ] ] an equivalent description is the following , @xcite . one can define standard dyadic intervals , namely those whose endpoints are @xmath8 and @xmath9 for @xmath10 . any finite partition of the interval @xmath6 $ ] made with standard dyadic intervals is called a dyadic subdivision . given two dyadic subdivision @xmath11 and @xmath12 with the same cardinality , it is possible to define a homeomorphism @xmath13\to [ 0,1]$ ] which maps linearly each interval of @xmath11 onto the corresponding interval of @xmath12 . the maps @xmath14 form the group @xmath0 . this characterization of the thompson group has the following graphical description . set @xmath15 the space of rooted planar binary trees . with @xmath16 , we denote by @xmath17 the set of leaves of @xmath18 . of course , denote by @xmath20 the set of matched pairs of trees @xmath21 , i.e. such that @xmath22 . we also say that any such pair @xmath21 is bifurcating . to any leaf of a binary tree it is associated a standard dyadic interval ( see @xcite ) , thus a pair of trees can be used to determine an element of @xmath0 . therefore , there is a map @xmath23 , @xmath24 . indeed , it is surjective but not injective . a cheap way to see this is to realize that any such pair with @xmath25 gives rise to the identity element of @xmath0 . moreover , it holds @xmath26 and if @xmath27 it is possible to connect the two pairs by a sequence of addition / deletion of opposite carets . in this section we review jones procedure that associates links to elements of @xmath0 , @xcite . this can be described by the composition of four maps . denote by @xmath28 the set of signed planar oriented finite graphs . first of all , there is a specific function @xmath29 , @xmath30 . for any @xmath31 one has the associated medial graph @xmath32 ( see @xcite , p.398 ) , that is a @xmath33-regular finite graph . finally , there is a map @xmath34 , where @xmath35 denotes the set of unoriented link diagrams . by composition , get a map @xmath36 given by @xmath37 also denoted @xmath38 for simplicity . every element in the group @xmath0 has a unique reduced , that is without opposing carets , tree diagram ( see @xcite , p. 6 ) . by this result , there is a canonical section @xmath39 and by the above this result provides unambiguously a map @xmath40 we denote it simply by @xmath41 . [ ese - omega2 ] we now show an example of the above procedure with the trefoil ( cf . _ _ 5.3.3 , @xcite ) . let @xmath42 be the element described by the following pair of trees , with associated oriented graph @xmath43 , and its associated medial graph @xmath44 @xmath45{1}\ ] ] the associated link diagram is obtained from @xmath44 by putting a crossing of type @xmath46 over the four vertices corresponding to the upper tree , and a crossing of type @xmath47 on the remaining vertices , as follows : @xmath45{2}\ ] ] it is easy to see that the diagram @xmath48 obtained in this way represents a trefoil . the graph @xmath49 is obtained by gluing along the common boundary the two graphs @xmath50 and @xmath51 . ( @xcite , p. 19 - 20 ) the graphs @xmath50 and @xmath51 are rooted trees ( not bifurcating in general ) . let @xmath52 be a rooted tree , then there exists a bifurcating tree @xmath53 such that @xmath54 ( see _ loc.cit._ , _ lemma _ 4.1.1 . for a proof ) . thus , giving a pair of matched trees @xmath21 is equivalent to giving the graph @xmath49 . actually @xmath50 is nothing but a subgraph of the dual of the tree @xmath55 considered as a planar graph in the upper half - plane with the leaves on the boundary . it is the subgraph generated by the edges corresponding to the north - east edges of @xmath55 . similarly , @xmath51 is a subtree of the dual of @xmath56 in the lower half - plane . [ link - inverse - element ] for any @xmath57 , it holds @xmath58 where @xmath59 denotes the mirror image link . if we associate to @xmath60 the pair of trees @xmath21 , then we associate to @xmath61 the pair @xmath62 . when we consider the medial graph of @xmath51 ( for @xmath60 ) we used the crossing @xmath47 . when we consider the medial graph of @xmath63 ( for @xmath61 ) we use the crossing @xmath46 . if we rotate the link @xmath64 about the x - axis we recognise the mirror image of the original link . we record a simple but useful result about cancelling carets for later reference . since this fact is already mentioned in @xcite we leave the proof to the reader . [ add - opp - car ] let @xmath21 be a pair of bifurcating trees , and consider another such pair @xmath65 obtained by adding an opposing pair of carets . then 1 . @xmath66 differs from @xmath49 by the addition of a new @xmath67valent vertex only connected to a vertex of @xmath49 on its left ; 2 . @xmath68 ( addition of a distant unknot ) . the aim of this paper is to provide some answer to the following question : for which ( polynomial ) invariants of unoriented links ( or graph polynomial ) @xmath69 is the function @xmath70 ( or @xmath71 ) of positive type ? we recall the definition of positive type function on a discrete group @xmath72 . a function @xmath73 is said _ of positive type _ ( or _ positive definite _ ) if one of the following equivalent conditions hold * @xmath74 for any @xmath75 and @xmath76 , @xmath77 ; * the matrix @xmath78 for any @xmath75 and @xmath76 , @xmath77 ; [ fpt ] the polynomial invariant must satisfy the following conditions ( see @xcite for more details ) : 1 . @xmath79 : it is enough consider the definition with @xmath80 . @xmath81 @xmath82 : it is enough to consider the definition with @xmath83 , @xmath84 and @xmath85 . the property follows because the matrix is positive definite and the determinant is positive 3 . @xmath86 : it is enough to consider the definition with @xmath87 , the property follows becuase the matrix is selfadjoint . we give an example of a graph polynomial that is positive definite function on @xmath0 , namely the ( normalized ) chromatic polynomial . as a consequence of ( * proposition 5.2.1 . ) , jones shows that @xmath88 is a function of positive type on @xmath0 , for all values of @xmath89 , using the theory of planar algebras . we recall the definition of the chromatic polynomial and present a detailed proof of this fact when @xmath90 . the method employed here will be useful later , when considering other invariants . + let @xmath72 be a graph and @xmath90 . the chromatic polynomial is a polynomial @xmath91 such that , for any @xmath92 , @xmath93 is the number of proper colourings of the vertices of @xmath72 with @xmath89 colours . moreover , if @xmath94 , it satisfies the following condition @xmath95 consider the following @xmath96graphs @xmath45{3}\ ] ] when @xmath97 , the chromatic polynomial takes the following values on the above graphs : * @xmath98 , * @xmath99 , * @xmath100 * @xmath101 * @xmath102 * @xmath103 the associated matrix is @xmath104 let @xmath105 and suppose that @xmath106 for some @xmath107 . then we can consider the well defined function on @xmath0 : @xmath108 . let @xmath109 and let @xmath110 ( not necessarily in reduced form ) . without loss of generality , inserting pairs of opposing carets whenever necessary , we can assume that @xmath111 , for all @xmath112 . then @xmath113 and in order to prove that @xmath114 is of positive type , we have to consider the symmetric matrix @xmath115 recall that both @xmath50 and @xmath51 are rooted trees with @xmath116 vertices , and that for any such tree the chromatic polynomial evaluated at @xmath89 takes the value @xmath117 . it is clear that , for the diagonal terms , @xmath118 . before proceeding with the proof , we introduce some notations and preliminary results . consider two trees @xmath119 . we denote the set of colouring of @xmath120 and @xmath121 by @xmath122 and @xmath123 , respectively . any colouring @xmath124 can be described by an array , which we denote by the same symbol , @xmath125 , with @xmath126 . consider the hilbert space @xmath127 . to any @xmath124 we can associate the vector @xmath128 . we also define the vector @xmath129 . let @xmath119 . as consequence of the orthogonality of the vectors @xmath130 , then the following equality holds @xmath131 in fact , the right hand side of the formula counts the number of colourings that are valid for both @xmath120 and @xmath132 , i.e. the colourings of @xmath121 . ( @xcite , p. 2 , 3)[inner - prod - matrix ] @xmath133 is positive if and only if there exist a hilbert space @xmath134 and vectors @xmath135 such that @xmath136 by the above discussion we have that the function @xmath114 is a function of positive type on @xmath0 . we briefly recall some definitions , the interested reader is referred to @xcite for further details . let @xmath72 be a graph . we denote the vertices by @xmath137 and the edges by @xmath138 , or for simplicity by @xmath139 and @xmath140 , respectively . the tutte polynomial may be defined as @xmath141 where @xmath142 denotes the number of connected components of the graph generated by @xmath143 . the tutte polynomial satisfies the following deletion and contraction rule @xmath144 we want to define a function on the thompson group @xmath0 using the tutte polynomial . first of all we need the following lemma consider @xmath145 and let @xmath146 be the threes obtained by the addition of a pair of opposite carets . then the following equality holds @xmath147 by _ proposition _ [ add - opp - car ] , the graphs @xmath148 and @xmath149 differs by a pair of parallel edges attached only to a vertex . we denote these two edges by @xmath150 and @xmath151 . using the rule of deletion and contraction we have that @xmath152 we define the tutte function as @xmath153 where @xmath145 and @xmath154 . the above lemma shows that it is well defined . consider the partition function of the potts model @xmath155 where @xmath156 is an edge and @xmath157 is the spin at site @xmath158 . it may be proved that ( up to a constant ) the partition function is equal to the tutte polynomial for certain values of the variables . more precisely @xmath159 for @xmath160 , @xmath161 . for more details we refer to @xcite , p. 1127 , 1132 . the function @xmath162 is of positive type on @xmath0 for @xmath160 , @xmath161 and @xmath163 . without loss of generality we can suppose that @xmath164 with @xmath165 for @xmath166 , and @xmath167 . therefore , we need to consider the matrix @xmath168 and show that it is positive semi - definite . our aim is to use the potts partition function and prove the claim by actually showing that @xmath169 is positive semi - definite . indeed , the partition function and the tutte function differ by the four different factors : @xmath89 , @xmath170 , @xmath171 and @xmath172 . we notice that all the factors do not depend on the pair @xmath173 . moreover , @xmath174 and @xmath175 so we may neglect them . the remaining factor , @xmath176 , is positive as long as @xmath163 . suppose that @xmath177 . we note that the partition function may be rewritten as @xmath178 for any @xmath179 , the expression @xmath180 defines the @xmath181th component of a vector in @xmath182 , i.e the component corresponding to @xmath183 . thus , we may define a vector @xmath184 and similarly a vector @xmath185 such that @xmath186 it follows by _ lemma _ [ inner - prod - matrix ] that the matrix @xmath187 is positive definite for any @xmath75 , i.e. the function @xmath162 is of positive type . one might wonder whether there are other values of the parameters for which the tutte function is of positive type . we want to observe that the result on the chromatic polynomial seen in the previous section can be obtained by results in the present section . in fact , consider the equality @xmath188 where @xmath160 , @xmath161 . for @xmath189 , the right hand - side converges to @xmath190 it is well known that this limit is equal to @xmath93 . we now consider the left hand - side . first of all , we observe that the integer @xmath191 can only take values between @xmath192 and @xmath116 . if the sum corresponding to @xmath193 is smaller than @xmath116 , then @xmath194 vanishes in the limit when @xmath189 . if the sum is equal to @xmath116 , then the exponential is equal to @xmath195 . it may easily be seen that the vectors considered for the chromatic polynomial correspond to those chosen for the tutte polynomial ( up to a factor @xmath196 ) in the limit @xmath189 . we wish to define a function on the thompson group @xmath0 using the kauffman bracket . the kauffman bracket ( @xcite ) is defined by the following skein - relation @xmath197 this polynomial is invariant under reidemeister moves of type 2 and 3 . the first reidemeister move produces a factor @xmath198 ( see @xcite , p. 365 ) . first of all , we prove a lemma that will allow us to define the kauffman bracket function . consider @xmath145 and let @xmath146 be the trees obtained by the addition of a pair of opposite carets . then the following equality holds @xmath199 by _ proposition _ [ add - opp - car ] , the graphs @xmath148 and @xmath149 differs by a pair of parallel edges attached only to a vertex , this means that a new distant unknot has been added . the properties of the kauffman bracket imply the claim . consider the kauffman bracket function @xmath200 where @xmath145 and @xmath201 . the above lemma shows that the function is well defined . we now recall some results proved in @xcite . consider a signed graph @xmath72 . denote by @xmath202 and @xmath203 the subgraphs whose edges are the positive and the negative edges , respectively . for any @xmath204 , define the function @xmath205 where @xmath157 is the spin at site @xmath206 . set @xmath207 and @xmath208 . consider the partition function defined by @xmath209 where the sum over @xmath193 runs over all the spin configurations @xmath210 . given a link @xmath211 with link diagram @xmath212 , we denote by @xmath213 its face graph ( for a definition see @xcite , p. 379 ) . we notice that the face graph and @xmath96graphs coincides , namely @xmath214 let @xmath215 be an integer and let @xmath216 be such that @xmath217 . consider a link @xmath211 with link diagram @xmath212 . then the following identity holds @xmath218 the claim follows from the following identities @xmath219 where @xmath220 denotes the writhe . before stating the main theorem of this section we formulate the following simple , but useful , lemma : consider the equation @xmath221 then * if @xmath222 , the complex solutions have modulus equal to @xmath195 , namely they are @xmath223 , @xmath224 ; * if @xmath225 , the complex solutions have modulus equal to @xmath195 , namely they are @xmath226 , @xmath227 ; * if @xmath97 , the complex solutions have modulus equal to @xmath195 , namely they are @xmath228 , each one with with multiplicity two ; * if @xmath229 , the complex solutions are purely imaginary with modulus different from 1 . all the statements are easily verified . we only make one comment for the case @xmath229 . in this case , we have that @xmath230 the solutions corresponding to the negative sign clearly have modulus different from @xmath195 . the solutions corresponding to the positive sign may be rewritten as @xmath231 and the condition @xmath232 require @xmath233 . however , this happens only when @xmath97 . [ pos - def - kauff ] the function @xmath234 , where @xmath216 is any solution of @xmath217 for @xmath235 , is of positive type on @xmath0 . as we have done before , without loss of generality we can suppose that the @xmath164 with @xmath165 for @xmath166 and @xmath167 . therefore , it is enough to consider @xmath236 and prove that @xmath237 is positive semi - definite . our aim is to use the partition function in order to prove our claim . the partition function and the kauffman polynomial are equal by the above lemma . therefore , recalling the equality between the face graph and the @xmath96graph , @xmath238 for any @xmath179 , the expression @xmath239 defines the @xmath181th component of a vector in @xmath182 , i.e the component corresponding to @xmath183 . thus , we may define a vector @xmath240 and similarly a vector @xmath241 such that @xmath242 it follows by _ lemma _ [ inner - prod - matrix ] that the matrix @xmath243 is positive definite for any @xmath75 , i.e. the function @xmath244 is of positive type . we observe that the assumption of @xmath235 was made in order to recognize the summands of the partition function as the scalar product of two vectors . in fact , in these cases @xmath216 is a complex root of unit and @xmath245 when one considers a signed planar graph , a knot can be obtained drawing the medial graph and using two different types of crossing depending on the sign ( see @xcite , chapters 16 , 17 ) . for simplicity , we have decided to follow the same notation as in jones paper @xcite . however , we warn the reader that the correspondence berween signs and crossing differs from the one adopted in standard textbooks ( for example see p. 380 in @xcite and p. 33 in @xcite ) . consider the following elements of the thompson group @xmath0 : @xmath246 , @xmath247 , @xmath248 , @xmath249 . + now we compute @xmath250 . the element @xmath251 can be described by the trees @xmath252 . moreover , @xmath253 is equivalent to the unknot . however , in order to obtain the unknot we have to use four times the first reidemeister move , two producing a factor @xmath254 and the others its inverse . therefore @xmath255 + now we compute @xmath256 . the element @xmath257 is equal to @xmath258 . in the standard ( i.e. minimal ) representation @xmath258 can be described by a pair trees @xmath259 . the link associated to @xmath258 is equivalent to the disjoint union of two unknots . in order to obtain this trivial link one has to perform two reidemeister moves of type @xmath195 , one producing the factor @xmath260 and the other one the factor @xmath261 . therefore , @xmath262 + now we compute @xmath263 . the element @xmath264 is equal to @xmath265 and it is associated to a pair of trees @xmath266 . after an easy calculation one obtains the _ hopf link_. actually , we have to perform the resolution of two curls ( one positive and one negative ) . therefore , we have that @xmath267 + now we compute @xmath268 . the element @xmath269 is equal to @xmath270 and it is associated to a pair of trees @xmath271 . the associated link is equivalent to the unknot . however , in order to obtain this trivial knot one has to perform four reidemeister moves of type @xmath195 , two producing the factor @xmath272 and the other two the factor @xmath254 . therefore @xmath273 + now we compute @xmath274 . the element @xmath275 is equal to @xmath276 . in the standard ( i.e. minimal ) representation @xmath276 can be described by a pair trees @xmath277 . the link associated to @xmath276 is equivalent to the disjoint union of two unknot . moreover , in order to obtain this trivial link one has to perform two reidemeister moves of type @xmath195 , one producing the factor @xmath260 and the other one the factor @xmath261 . therefore @xmath278 + now we compute @xmath279 . as we have seen in _ example _ [ ese - omega2 ] there exist two trees @xmath280 such that @xmath281 . after the application of two reidemeister moves of type @xmath282 , one move of type @xmath283 and one of type @xmath195 we get the trefoil . we have that @xmath284 so far we have computed the entries over the diagonal . the entries under the diagonal differ from the corresponding entries over the diagonal by the change of variable @xmath285 , cf . _ proposition _ [ link - inverse - element ] . the only different entry is the one in position @xmath286 : @xmath287 summing up , the matrix associated to these elements is @xmath288 one can check that for @xmath289 and @xmath290 we get the following matrices , respectively @xmath291 the above example also shows that the obtained matrix is not self - adjoint for @xmath229 and thus certainly not positive semi - definite . therefore we have the following result the function @xmath292 , where @xmath216 is any solution of @xmath217 for @xmath229 , is not positive definite on the thompson group @xmath0 . the @xmath293index of a link is defined as the smallest number of leaves required for an element of @xmath0 to give that link ( @xcite , p. 41 ) . in the previous example we showed that the hopf link and the trefoil can be obtained with pairs of trees in @xmath294 . one might wonder whether their @xmath293index is @xmath295 . the following result allow us to give a positive answer to this question . every non - trivial knot / link has @xmath293index at least equal to @xmath295 . by symmetry ( i.e. taking into account _ propositions _ [ link - inverse - element ] and [ add - opp - car ] ) , one has to consider only the following @xmath296-graphs @xmath45{4}\ ] ] it can be easily checked that the associated links are trivial . our aim is to define a function of positive type on @xmath0 using the number of colourings of a link . first of all we introduce the preliminary definitions . let @xmath89 be a positive odd integer . the definition @xmath297colouring is a generalization of tricoloring of link ( see @xcite , p. 162 , and @xcite ) . let @xmath211 be a link and @xmath212 one of its diagrams . a link is said to be @xmath297colourable if every arc of the diagram is associated to an element of @xmath298 ( called colour ) , such that at each crossing the sum of the colours of the undercrossing is equal to the sum of the overcrossing . the number of @xmath297colourings is a link invariant and is denoted by @xmath299 . in order to prove that @xmath297colourings defines a positive definite function on @xmath0 we introduce the following definition . let @xmath201 . the associated link may be decomposed in the two upper and lower halves , that is @xmath300 . we may call these halves as _ semi - links_. we may represent these semi - links associated to @xmath21 by the following tangles @xmath45{5}\ ] ] consider a pair of bifurcating trees @xmath21 with @xmath1leaves . the link @xmath38 may be seen as the _ n - sum of the semi - links _ which is defined by joining the corresponding strings in the rectangles . we denote the @xmath1sum of semi - links by the symbol @xmath301 . we want to introduce a partition function for the @xmath297colorings and we will recall some results of de la harpe and jones ( see @xcite , p. 219 ) . a link diagram is @xmath309regular graph , where the vertices are the crossings . for each vertex @xmath310 we denote by @xmath311 the upper string and by @xmath312 the lower string . we set @xmath313 consider the partition function defined by @xmath314 where the sum over the functions @xmath315 . any such function is called _ state _ of the graph @xmath72 . we have that @xmath316 ( @xcite , _ example _ 4.1 , p. 22 ) . we can suppose that the @xmath164 with @xmath165 for @xmath166 and @xmath167 . therefore , it is enough to consider @xmath318 and prove that @xmath319 is positive semi - definite . our idea is to use the partition function in order to prove the claim . therefore @xmath320 for any @xmath321 , the expression @xmath322 defines the @xmath323th component of a vector in @xmath324 , i.e the component corresponding to @xmath325 . thus , we may define a vector @xmath240 and similarly a vector @xmath241 such that @xmath326 it follows by _ lemma _ [ inner - prod - matrix ] that the matrix @xmath319 is positive definite for any @xmath75 , i.e. the function @xmath317 is of positive type . it is well - known that a positive type function on a discrete group @xmath72 gives rise to a completely positive multiplier on the reduced @xmath327-algebra @xmath328 . one might investigate whether these or similar methods can be exploited to say anything of interest for other kind of multipliers on @xmath0 . j. belk , _ thompson s group f _ , phd thesis , cornell university , ( 2004 ) , arxiv preprint arxiv:0708.3609 ( 2007 ) . r. bhatia , _ positive definite matrices _ , princeton university press , 2009 . b. bollobas , _ modern graph theory _ , vol . springer science and business media , 1998 . j. w. cannon , w. j. floyd , w. r. parry , _ introductory notes on richard thompson s groups _ , enseign . 42 ( 1996 ) , 215 - 256 . g. b. folland , _ a course in abstract harmonic analysis _ , crc press , 1994 . c. godsil , f. g. royle , _ algebraic graph theory _ , vol . 207 . springer science and business media , 2013 . g. golan , m. sapir , _ on jones subgroup of r. thompson group @xmath0 _ , arxiv preprint arxiv:1501.00724 ( 2015 ) , to appear in the proceedings of the msj - si , hyperbolic geometry and geometric group theory . s. haagerup , u. haagerup , m. ramirez - solano , _ a computational approach to the thompson group f _ , internat . j. algebra comput . 25 ( 2015 ) , 381 - 432 . p. de la harpe , v. f. r. jones , _ graph invariants related to statistical mechanical models : examples and problems _ , j. combin . theory , series b 57 ( 1993 ) , 207 - 227 . m. hazewinkel ( ed . ) , _ encyclopaedia of mathematics _ , 3 ( d fey ) , kluwer , 1989 ( isbn 1 - 55608 - 002 - 6 ) . v. f. r. jones , _ some unitary representations of thompson s groups f and t _ , arxiv preprint arxiv:1412.7740 ( 2014 ) . l. h. kauffman , _ state models and the jones polynomial _ , topology 26 ( 1987 ) , 395 - 407 . j. h. przytycki , _ 3-coloring and other elementary invariants of knots _ , in knot theory ( warsaw , 1995 ) , 275 - 295 , banach center publ . , 42 , polish acad . warsaw , 1998 . d. j. a. welsh , c. merino . _ the potts model and the tutte polynomial _ , j. math . 41 ( 2000 ) , 1127 - 1152 .	using jones correspondence between elements in the thompson group @xmath0 and certain graphs / links we establish , for certain specializations of the variables , the positive definiteness of some familiar link invariants and graph polynomials , namely the kauffman bracket , the number of @xmath1colourings and the tutte polynomial , when viewed as functions on @xmath0 . 0.9 cm * msc 2010 * : 43a35 , 57m27 , 05c31 . * keywords * : thompson group , function of positive type , trees , knots , links , chromatic polynomial , tutte polynomial , rank polynomial , kauffman bracket , link colourings , fox colourings . [ section ] [ theorem]corollary [ theorem]lemma [ theorem]proposition [ theorem]conjecture [ theorem]definition [ theorem]example [ theorem]remark h l o p s
the angular correlation function of galaxies has been used to characterize the large - scale distribution of galaxies for over 2 decades . if the number density on the sky at angular position @xmath4 is @xmath5 , then @xmath0 is defined as @xmath6 the 3-dimensional unit vectors @xmath4 and @xmath7 are used to define the angular separation @xmath8 as @xmath9 , and @xmath10 is the mean galaxy number density on the sky . the observed galaxy distribution is well described by a power law @xmath11 , with slope @xmath12 . since @xmath0 is a projection on the sky of the 3-dimensional auto - correlation function @xmath13 , the above power law for @xmath0 has been associated with a power law for @xmath14 with slope @xmath15 which is close to the observed value in the nearby galaxy distribution . measurements of @xmath0 are difficult to interpret as it involves a projection of the galaxy distribution out to redshifts of order 1 ( e.g. for a sample with limiting @xmath3 magnitude of about 24.5 ) . thus the effects of evolution of large scale structure due to gravitational clustering as well as galaxy evolution need to be understood to interpret @xmath0 . recently villumsen ( 1996 ) has considered the effect on @xmath0 of gravitational lensing by large - scale structure along the line of sight ( gunn 1967 ) . lensing increases the area of a given patch on the sky , thus diluting the number density . on the other hand , galaxies too faint to be included in a sample of given limiting magnitude are brightened due to lensing and may therefore be included in the sample . the net effect , known as magnification bias , can go either way : it can lead to an enhancement or suppression of the observed number density of galaxies , depending on the slope of the number - magnitude relation . variations in the number density which are correlated over some angular separation alter @xmath0 . villumsen ( 1996 ) showed how the linear evolution of density fluctuations along the line of sight can be used to compute the change in @xmath0 due to magnification bias . the deflections of neighbouring photon trajectories due to lensing by large scale structure are very small , hence the calculation can be done in the limit of weak lensing . in this paper we compute @xmath0 for different cosmological models taking into account the effects of nonlinear gravitational evolution and gravitational lensing . since @xmath0 is a 2-point statistic , even in the fully nonlinear regime it is determined completely by the 3-dimensional power spectrum . we use extensions of the proposal of hamilton et al . ( 1991 ) to include the nonlinear evolution of the power spectrum into the calculation of @xmath0 . we also include the dependence on the cosmological parameters @xmath16 and @xmath17 . in section 2 the formalism for computing @xmath0 is presented . section 2.1 discusses the effects of the cosmological model via the growth of density perturbations and the distance - redshift relation . results for cdm - like power spectra for different values of @xmath16 and @xmath17 are presented in section 3 . we explore ways to isolate the effect of the lensing contribution to @xmath0 in section 4 and conclude in section 5 . the effect of magnification bias due to weak gravitatinal lensing on the galaxy angular correlation function @xmath0 has been derived in villumsen ( 1996 ) . written in terms of the time - dependent power spectrum instead of the present day one , which makes it applicable to nonlinear power spectra , @xmath0 is given by @xmath18 ^ 2 \nonumber \\ & & \times \int_0^\infty dk\ , k\ , { p(\chi , k)\over a^2(\chi)}\ , j_0\left[k r(\chi)\theta\right]\ , . \label{wfull}\end{aligned}\ ] ] in this section the physical meaning of the various terms in the above equation and the notation used will be clarified . we have used the notation of jain & seljak ( 1997 ) , where @xmath19 is the radial comoving distance , @xmath20 that to the horizon , @xmath21 is the comoving angular diameter distance , @xmath22 the normalized radial distribution of galaxies in the sample , and @xmath23 from the expression for the unperturbed line element @xmath24 @xmath25 being conformal time , it follows that @xmath26 where @xmath27 is the spatial curvature . note that for a delta - function distribution of galaxies , @xmath28 , @xmath29 reduces to @xmath30 . this notation is related to the one of villumsen ( 1996 ) through @xmath31 , @xmath32 , @xmath33 and @xmath34 . first we shall briefly describe the effect of magnification bias on @xmath0 . gravitational lensing of a galaxy by dark matter concentrations between us and the galaxy increases the area of the galaxy image while conserving surface brightness . this results in a magnification @xmath35 which is given by the ratio of lensed to unlensed area of the image . the amplitude of @xmath35 depends on the convergence @xmath36 , which is the surface mass density divided by the critical density , and the shear @xmath37 through ( young 1981 ) = [ mugen ] in the limit of weak lensing , @xmath38 , applicable to lensing by large - scale structure , the above expression reduces to = 1 + 2[mu ] this magnification has two effects . since the lensed area is increased due to deflection of the light rays , the number density of galaxies decreases inversely proportional to @xmath35 . there is a competing effect , however . in a flux limited survey , magnification brings some faint galaxies above the flux limit , which would not otherwise have been detectable , thus increasing the number density of galaxies . which of the effects wins depends on the reservoir of faint galaxies available , which can be quantified by the slope @xmath39 of the true number counts @xmath40 for a magnitude limit @xmath41 , s= . the two effects change the galaxy numbers by ( e.g. broadhurst , taylor & peacock 1995 ) n^(m)=n_0(m ) ^2.5s-1 which reduces to n^(m)=n_0(m)[np ] in the weak lensing limit . in addition , the number density of galaxies is changed by @xmath42 due to intrinsic clustering . let @xmath10 denote the average number density of galaxies . at a given position in the sky @xmath43 , the number density is thus changed to n()=\|n ( 1 + n ( ) + 5(s-0.4 ) ( ) ) . [ nth ] now we further assume that galaxies trace the underlying dark matter distribution so that the 3-dimensional galaxy overdensity is , _ g(x ) = b ( x ) , where @xmath44 is the dark matter overdensity and the bias @xmath45 is the proportionality factor . in this linear bias model , the perturbed , projected number density is given by ( villumsen 1996 ) n()= b _ 0^_h d w ( ) ( r ( ) ) . [ ndelta ] whereas the convergence @xmath36 is given by ( villumsen 1996 ) ( ) = _ 0^_h d g ( ) . [ kappa ] the angular correlation function @xmath0 as defined in equation [ wthav ] involves the expectation value @xmath46 . since both @xmath42 due to intrinsic clustering , and @xmath36 are related to @xmath47 according to eqs . [ ndelta ] and [ kappa ] , intrinsic dark matter correlations will induce correlations among galaxies . there are three distinct effects : correlations among the galaxies tracing correlated dark matter densities , described by @xmath48 ; among background galaxies being lensed by correlated dark matter concentrations , described by @xmath49 ; and finally among background galaxies and the foreground galaxies tracing the correlated dark matter concentrations , which are responsible for the lensing of the background galaxies , described by the term @xmath50 . the expressions for @xmath51 , @xmath52 and @xmath53 from the above equations are inserted into eq . [ wthav ] . expressing @xmath44 in terms of its fourier transform @xmath54 and performing the ensemble average @xmath55 then introduces the power spectrum @xmath56 . one finally obtains the result of eq . [ wfull ] ( villumsen 1996 ) by evaluating the integrals , under the further assumption of @xmath57 , and using the plane - parallel approximation that only fourier modes with wave vectors nearly perpendicular to the line of sight @xmath43 contribute to the integral . in the previous paragraphs we discussed the origin of the three terms in the angular correlation function , obtained by multiplying out the squared bracket in eq . [ wfull ] . we can write them schematically as , @xmath58 the third term @xmath59 is proportional to the correlation function @xmath60 of galaxy image ellipticities ( villumsen 1996 ) , _ ll ( ) = ( 2.5s-1)^2 c_pp ( ) , the dependence of this term on @xmath61 and @xmath62 is discussed in detail in jain & seljak ( 1997 ) ( see also bernardeau et al . 1997 ; kaiser 1996 ) . with the assumption of constant linear bias , we can factor out the constants involving bias , @xmath39 and @xmath16 , and write @xmath0 as @xmath63 \ ; . \label{wtilde}\end{aligned}\ ] ] besides the explicit @xmath61 dependence , @xmath0 depends on @xmath61 and @xmath62 mainly through the evolution of the power spectrum and the dependence of @xmath21 on @xmath16 and @xmath17 . in the linear regime the power spectrum depends on the linear growing mode of density perturbations @xmath64 and the normalization @xmath65 as , p(,k ) ^2 . the linear growing mode is well approximated by ( carroll , press & turner 1992 ) @xmath66^{-1 } \ ; , \end{aligned}\ ] ] where we have defined , following mo , jing & boerner ( 1997 ) , the time dependent fractions of density in matter and vacuum energy , @xmath67 and @xmath68 in terms of the present - day values @xmath61 and @xmath62 , ( a)= and ( a)= . the spatial geometry also differs in different models , leading to a dependence of the angular distance @xmath21 on @xmath61 and @xmath62 according to eq . these effects are contained in the terms @xmath69 , @xmath70 and @xmath71 in eq . [ wtilde ] . on small scales ( high-@xmath72 ) the growth of the power spectrum is significantly affected by nonlinear gravitational clustering . the nonlinear power spectrum is obtained from the linear one through the fitting formulae of jain , mo & white ( 1995 ) for @xmath73 , and from those of peacock & dodds ( 1996 ) for the open and @xmath74 dominated models . these fitting formulae are based on the idea of relating the nonlinear power spectrum at scale @xmath72 to the linear power spectrum at a larger scale @xmath75 , where the relation @xmath76 depends on the power spectrum itself ( hamilton et al . 1991 ) . they have been calibrated from and tested extensively against n - body simulations . for the linear power spectrum we take a cdm - like spectrum ( bardeen et al . 1986 ) with shape parameter @xmath77 , which provides a reasonable fit to the observed galaxy power spectrum . with this choice of the linear power spectrum , we use the fitting formulae for the fully nonlinear power spectrum as a function of @xmath78 and @xmath72 to evaluate equation [ wfull ] for @xmath0 . having specified the shape of the mass power spectrum , we need to set the global cosmological parameters @xmath16 and @xmath17 . the normalization of the power spectrum in turn will depend on these parameters . the four different cosmological models we consider include : a flat universe with @xmath79 , an open model with @xmath80 , and a flat @xmath74 dominated model , with @xmath80 and @xmath81 . these three models are normalized to cluster abundances , which give the approximate relation ( white et al . 1993 ; viana & liddle 1996 ; eke , cole & frenk 1996 ; pen 1996 ) : @xmath82 ( for @xmath80 , we use @xmath83 , which is slightly lower than given by the above relation . ) our fourth model is a flat universe , @xmath73 , with a higher normalization of @xmath84 . while the cosmological model is specified above , the galaxy population still needs to be described in terms of its redshift distribution , the number count slope @xmath39 and the bias parameter @xmath45 . we model the redshift distribution of galaxies by @xmath85 } \exp(-(z / z_0)^{\beta})\ ; , \label{nz}\ ] ] for @xmath86 , which agrees reasonably well with the redshift distribution estimated for the hubble deep field from photometric redshifts ( mobasher et al . , 1996 ) . @xmath22 is related to the redshift distribution by @xmath87 , where the expressions for the hubble constant @xmath88 and the radial comoving distance @xmath89 depend on the background cosmology . the mean redshift is given by @xmath90 for a flat model with @xmath79 , we obtain the @xmath3 magnitude limit corresponding to a given mean redshift by interpolation of a set of estimated values from charlot ( 1996 ) . in this section we consider red and blue galaxy populations , with number count slopes @xmath91 and @xmath92 respectively , and a fixed bias @xmath93 for both samples . in the next section we will consider the more realistic case of a relative bias for the different colour samples , since red galaxies are more strongly clustered . figure [ wtheta1 ] shows @xmath0 for the @xmath73 , @xmath84 model . the linear and nonlinear predictions , shown by the solid and dashed curve , are compared with a power law shape with slope @xmath94 . these curves are for @xmath95 , for which the lensing term drops out , and therefore show the contribution of the clustering term to @xmath0 . the effect of nonlinear evolution enhances @xmath0 for @xmath96 , while it slightly suppresses @xmath0 for @xmath97 . on larger scales the linear and nonlinear curves coincide . it is interesting that the nonlinear @xmath0 is much closer to the observed power law than the linear curve which reflects the curvature of the cdm power spectrum . at large angles , @xmath98 the @xmath0 curve starts to deviate from the power law shape . for the open and @xmath2 dominated models , the results are similar , with the nonlinear contribution being larger than the @xmath73 model . in fig . [ wtheta ] we show how lensing affects @xmath0 in the flat @xmath79 cosmological model , for three values of the number count slope @xmath99 , @xmath100 , and @xmath95 . the mean redshift is @xmath101 , corresponding to an @xmath3 magnitude limit of about @xmath102 . the @xmath95 curve represents the intrinsic clustering as in the previous figure . for @xmath103 , the amplitude of @xmath0 is increased due to lensing , and for @xmath104 it is decreased . magnification bias hardly changes the shape of @xmath0 since the three curves are nearly parallel to each other . in fig . [ wintrinsic ] we show the dependence of @xmath0 on mean redshift of the sample . the contribution due to intrinsic clustering is shown for the four cosmological models , i.e. setting @xmath95 . three values of @xmath8 are shown , ranging from @xmath105 , where nonlinear effects are important , to @xmath106 where nonlinear effects become unimportant . at @xmath105 , the models with low @xmath61 have higher amplitudes than the models with @xmath73 . this is because in the former models the growth of perturbations is slowed down at late times ; when normalizing to present - day cluster abundances , this implies a higher normalization at earlier times compared to the @xmath73 models . this in turn means that nonlinear effects become important earlier on , and lead to a greater effect of nonlinear clustering by today . at large mean redshifts , however , the amplitude for the @xmath74model dips below the @xmath73 , @xmath84 model . here a competing geometrical effect starts to come in : at a given redshift a physical volume element is larger if @xmath107 than in the absence of a cosmological constant , and therefore the intrinsic clustering is decreased . for @xmath106 , this geometrical effect in the @xmath74model is more noticeable and starts to be seen at lower mean redshifts because competing nonlinear effects are unimportant . in fig . [ wauto2 ] we plot @xmath108 as a function of mean redshift for three values of @xmath39 , as in fig . [ wtheta ] , and for all four cosmological models . for the @xmath79 models , we also give the @xmath3 magnitude limits inferred from charlot ( 1996 ) on the top axis of the plot , corresponding to the mean redshifts shown on the bottom axis . these values are given as an orientation only , since the relation between @xmath3 magnitude and mean redshift is not exactly linear , and depends on the model of galaxy evolution used to derive it . we can understand these results better by looking at the three terms of @xmath0 in eqs . [ wterms ] and [ wtilde ] separately . at low mean redshifts , the intrinsic clustering term @xmath109 dominates , but it decreases with redshift . at low @xmath110 the cross - term @xmath111 is much smaller than the intrinsic term , but it is approximately constant with redshift ; @xmath112 is even smaller than the cross - term due to the small lensing depth at small mean redshift , but it increases with @xmath113 . as the mean redshift of the sample increases , @xmath109 drops , and the cross - term starts to become noticeable . the cross - term contains the factor @xmath114 , so that for @xmath115 , @xmath0 is increased with respect to the intrinsic contribution due to lensing , and for @xmath116 it is decreased . also , the redshift at which this happens is larger for smaller @xmath16 and @xmath117 . at still larger @xmath110 , @xmath112 has increased sufficiently to become noticeable as well . since @xmath118 is always positive , its effect goes in the same direction as that of the cross - term for @xmath119 . but for @xmath116 , its effect is in the opposite direction to that of the cross - term , so that the decrease due to @xmath111 is partly compensated . the flat model with the higher normalization @xmath84 shows a stronger lensing signal than the one with the lower @xmath120 , because the bias factor @xmath93 is smaller , which increases the lensing contribution from the cross - term @xmath121 relative to the intrinsic clustering term @xmath122 . in the open or @xmath74 model , as compared to the flat model , the change in growth rate of perturbations and the spatial geometry affect the intrinsic clustering part of @xmath0 as discussed above . the geometric effect for @xmath107 , which decreases the intrinsic clustering , is the opposite of the lensing part . since the volume element is larger , the optical depth to lensing is larger , and lensing is more effective . this explains why the amplitudes for red and blue samples deviate more strongly than in the @xmath73 case in spite of the extra factor of @xmath61 in front of the cross - term @xmath123 . for the open model also this geometric effect is present , and compensates for some of the decrease due to the @xmath61-prefactor , but it is weaker than in the @xmath2 model . the results presented so far for @xmath0 include the contribution from intrinsic clustering and from lensing . it is interesting to consider measurable statistics that are more sensitive to the lensing contribution ; in this section we show that the ratio of @xmath0 for red and blue samples is one such example . we assume the presence of a population of blue galaxies with a number count slope @xmath92 out to @xmath124 , and also of a population of red galaxies with a slope of @xmath91 . a sample with such a slope may be obtained by defining color selected subsamples @xcite , using the fact that the number count slope is a decreasing function of @xmath125 color @xcite . since red and blue galaxy samples are observed to have a relative bias at present , with red galaxies being more strongly clustered ( they are more likely to be found in the centers of clusters ) than blue ones , in this section we consider a bias factor for the red sample @xmath126 which is larger than the bias @xmath127 of the blue sample . we assume the individual biases to be constant in time , and we neglect possible physical evolution of the spatial correlation function ( besides the one due to gravitational evolution ) . in the absence of lensing , and assuming one has selected samples with the same redshift distribution , the ratio of @xmath0 of the blue and red samples would be equal to the ratio of their relative biases squared , @xmath128 however , lensing changes this relationship . in the regime where the cross - term dominates the third term , the lensing contribution increases @xmath0 for the blue , and decreases it for the red sample . in fig . [ wratio ] we plot @xmath129 , which ought to be equal to @xmath101 in the absence of lensing , versus mean redshift for @xmath130 , and a relative bias of two of the red and blue samples . the four curves show the results for the different cosmological models considered ( see section 2 ) . we can see that there is only little variation with angle as expected from the results of fig . [ wtheta ] . for intermediate mean redshifts @xmath131 we can neglect the @xmath112term , and expand in the small quantity @xmath132 to see the dependence on bias factors and number count slope , @xmath133 \frac{\tilde{w}_{gl}(\theta)}{\tilde{w}_{gg}(\theta ) } \right)\nonumber \\ & & \times \left ( \frac{b_b}{b_r}\right)^2 \ ; .\end{aligned}\ ] ] for @xmath134 , @xmath112 can no longer be neglected . as discussed in the previous section , this term has the effect of compensating some of the decrease in amplitude of @xmath0 due to the cross - term for the red sample , and the exact redshift at which this happens depends on @xmath16 and @xmath135 . this effect is responsible for the flattening off of the ratio seen in fig . [ wratio ] . at still larger redshifts , ( not plotted here ) , the ratio would decrease again . as discussed in section 3 , for the open and @xmath74dominated models , both the growth rate of perturbations and the spatial geometry are different from the @xmath73 model . by taking the ratio of two correlation functions , the effect of the growth rate cancels out , so that we are left with the effect of the different geometries on the three terms . for the @xmath74dominated model the geometrical effect is largest , so that in spite of the factor of @xmath61 in the cross - term , the ratio is larger than for the flat model normalized to @xmath120 . it even attains a value comparable to the @xmath73 model with the higher normalization of @xmath84 . recall that a larger @xmath65 favors the lensing contribution over the intrinsic clustering since the bias factor @xmath93 is smaller , and @xmath121 whereas @xmath122 . for the open model the geometrical effect is not as important , and the ratio is below that for the flat case . we have quantified the effect of gravitational lensing by large - scale structure on the angular correlation function of galaxies @xmath0 , for different cosmological models , on angular scales ranging from @xmath136 to @xmath137 . we have taken into account nonlinear gravitational clustering which affects both the intrinsic clustering and lensing contributions to @xmath0 . we find that the ratio of the angular correlation function for red and blue galaxy samples , normalized by the inverse of the relative bias of the two samples , deviates from the value of @xmath101 expected in the absence of lensing at sufficiently large mean redshifts of the sample . for a mean redshift of 1.3 of the sample , this ratio rises to about @xmath138 for the model with @xmath73 and @xmath84 , and continues to rise , flattening off , until it reaches @xmath139 by a mean redshift of 2 . at @xmath140 it rises to about @xmath141 for the model with @xmath73 and @xmath120 , and to about @xmath142 for the open model . in the @xmath74dominated model , the ratio reaches a value of about @xmath143 at @xmath140 , and continues to rise to a value of about @xmath144 at a mean redshift of 2 . these values are at @xmath136 angular separation ; for larger angles they differ somewhat , but not to a great extent as shown in fig . [ wratio ] . the largest uncertainty in the application of this result is due to the assumption of two populations with different number count slopes but the same redshift distribution with @xmath145 . the effect of magnification bias on the angular correlation function could be used in future surveys , such as the eso imaging survey ( renzini et al . 1996 ) , to detect weak lensing by large - scale structure and constrain the cosmological parameters @xmath61 and @xmath62 . the eso imaging survey aims to detect about 200 - 300 galaxies with redshifts between @xmath101 and @xmath146 in a field of 25 arcminutes squared , and a similar number with redshifts larger than @xmath146 in a field ten times larger . one problem with trying to detect the effect of magnification bias on @xmath0 is that unknown physical evolution of the galaxies can modify the intrinsic clustering and introduce uncertainties which may be larger than the lensing signal . the presence of bias evolution would introduce further uncertainties . in future work , we plan to study the cross - correlation of two galaxy samples with different mean redshifts and non - overlapping redshift distributions ( moessner & jain 1997 ) . in this way we hope to minimize the importance of uncertainties due to intrinsic clustering and the exact form of the redshift distribution , and propose a measurable statistic that is dominated by the lensing contribution . we would like to thank matthias bartelmann , teresa brainerd , peter schneider , uro seljak and simon white for useful discussions .	the angular correlation function @xmath0 of faint galaxies is affected both by nonlinear gravitational evolution and by magnification bias due to gravitational lensing . we compute the resulting @xmath0 for different cosmological models and show how its shape and redshift evolution depend on @xmath1 and @xmath2 . for galaxies at redshift greater than 1 ( @xmath3 magnitude fainter than about 24 ) , magnification bias can significantly enhance or suppress @xmath0 , depending on the slope of the number - magnitude relation . we show for example how it changes the ratio of @xmath0 for two galaxy samples with different number - count slopes . galaxies : clustering - cosmology : observations - gravitational lensing - large scale structure of the universe
in the standard model ( sm ) , with three families of quarks , the unique cp violating phase of the cabibbo - kobayashi - maskawa ( ckm ) matrix can explain some of the observed cp violating phenomena in @xmath8 and @xmath9 systems . however , the failure of the km phase in explaining the matter - antimatter asymmetry and some recent measurements of cp violating observations in @xmath9 meson mixings and decays motivates the search for new source of cp violation ( cpv ) . therefore , it is an important issue to explore and to find new cp violating effects in various systems , such as cosmos , large hadron collider ( lhc ) , tevatron , @xmath9 factories etc . recently , several hints for the existence of new cp violating sources are revealed in experiments . the first hint is observed in the cp asymmetries ( cpas ) of @xmath10 decays where by naive sm estimation , one expects that @xmath11 and @xmath12 decays have similar cpas . however , it is surprising that the world average difference between the two cpas contradicts the expectation as the experimental result is @xcite a_cp= a_cp(^+ k^- ) - a_cp(^0 k^-)= -(14.8_-1.4^ + 1.3)% , whereas the sm prediction is @xmath13 @xcite . the large deviation from the sm prediction indicates a puzzle in the asymmetries and it is introduced in the literature as @xmath14 puzzle @xcite . the second hint is observed in the time - dependent cpa of @xmath15 system , where cdf and d have shown an unexpected large cp phase in the mixing - induced cpa for @xmath16 and the two possible solutions are given by @xcite 2^j/_s=2_s + 2^np_s= -0.75^+ 0.32_-0.21 ` or ` -2.38^+0.25_-0.34 at @xmath17 confidence level ( cl ) . here , @xmath18 @xcite is the sm cp violating phase and @xmath19 is the cp violating phase of new physics . the significant deviation from the sm prediction could be speculated by the contributions of new physics . the third hint is observed in the like - sign charge asymmetry which is defined as @xcite ^b_s & = & , [ eq : absl_exp ] where @xmath20 denotes the number of events that @xmath21- and @xmath22-hadron semileptonically decay into two positive ( negative ) muons . recently , d collaboration has announced the measurement on @xmath5 in the dimuon events @xcite with ^b_s = ( -9.57 2.51(stat ) 1.46 ( syst ) ) 10 ^ -3 . the sm prediction is @xmath23 @xcite . if the semileptonic b - hadron decays do not involve cp violating phase , then the charge asymmetry is directly related to the mixing - induced cpas in @xmath24- and @xmath15-meson oscillations . although the errors of the data are still large , however the @xmath25 standard deviations from the sm prediction can be attributed to new cp violating phases in @xmath4 and @xmath26 transitions @xcite . in order to explore the new physics and to avoid the uncontrollable qcd uncertainties , we will concentrate our study on the mixing parameter @xmath1 , the charge asymmetry @xmath5 and the time - dependent cpa in @xmath27 oscillation , where qcd effects can be controlled well by lattice qcd . in the literature , many extensions of the sm such as chiral color models @xcite , @xmath28 models @xcite etc have been proposed . the flavor non - universal axigluon in the generalized chiral color models @xcite has been studied for solving the anomalous forward - backward asymmetry ( fba ) in the top - quark pair production at the tevatron @xcite . although other models such as z , diquarks models @xcite etc may have significant contributions to the fba , however , large gauge couplings and flavor changing effects should be introduced in which chiral color model does not need . inspired by the effects of the axigluon on the top - quark fba , we study the axigluon - mediated phenomena in @xmath9-meson system . a flavor universal axigluon has flavor - conserving effects only . for non - universal axigluon which has different couplings to different quarks , flavor changing neutral currents ( fcncs ) can be generated at tree level . this is achieved after transforming the weak eigenstates of the quarks into their physical eigenstates . as a consequence , many phenomena will be affected by these fcnc effects . in this paper , we analyze in detail the non - universal axigluon contributions to the time - dependent cpas in @xmath27 oscillation after taking into account the constraint from the mixing parameter @xmath1 . this paper is organized as follows . in sec . [ sec : formalism ] , we formulate the interactions of @xmath2 transitions which are induced by flavor non - universal axigluon exchange . accordingly , we derive the corresponding effective hamiltonian for @xmath29 processes . furthermore , we discuss the contributions of the axigluon to the charge asymmetry @xmath5 and the time - dependent cpas for @xmath30 , @xmath31 , @xmath16 decays . the detailed numerical analysis is presented in sec . [ sec : na ] . we give the conclusion in sec . [ sec : conclusion ] . in order to study the contributions of the non - universal axigluon to the fcnc processes , we start by writing the interactions of the massive color - octet gauge boson with quarks as _ a & = & g_v \|q _ t^b q g^b_a + g_a \|q _ _ 5 * z * t^b q g^b_a , where we have suppressed the flavor and color indices , @xmath32 are the gauge couplings of the new gauge group @xmath33 , @xmath34 are the gell - mann matrices which are normalized by @xmath35 and @xmath36 is @xmath37 diagonalized matrix with diag(z)=(1 , 1 , @xmath38 ) . here @xmath39 where @xmath40 denotes the gauge coupling of the third - generation quark and its value depends on a specific model , e.g. @xmath41 in ref . @xcite . for simplicity , we assume that the new exotic quarks which are required for anomaly free are very heavy and their effects are negligible . hence , we still focus on three flavors for each up and down type quarks . following the scenario in refs . @xcite for solving the large top - quark fba , we assume that the axigluon couplings to the third generation are different from their couplings to the first two generations . the left- and right - handed quarks are @xmath42 doublet and singlet respectively . thus , after spontaneous symmetry breaking , the interacting and physical eigenstates can be related by unitary matrices as @xmath43 with @xmath44 being the chiralities @xmath45 and @xmath46 and @xmath47 being up or down type quarks . since @xmath36 is not a unit matrix , the fcncs are arisen from the axial - vector currents and the corresponding lagrangian is given by _ fcnc&= & g_a \|q _ ( v^q_r * z * v^q_r p_r - v^q_l * z * v^q_l p_l ) t^b q g^b_a [ eq : fcnc ] with @xmath48 . since @xmath49 are unknown matrices , the fcncs are associated with left and right - handed currents generally . nevertheless , if @xmath50 , from eq . ( [ eq : fcnc ] ) we see that the fcncs are only associated with axial - vector currents . in terms of the flavor indices , the matrix @xmath51 can be decomposed as ( v^q _ v^q _ ) _ ij & = & _ ij + ( v^q_(z- 1 ) v^q _ ) _ ij = _ ij + ( -1 ) ( v^q_)_i 3 ( v^q * _ ) _ j3 . therefore , the lagrangian of @xmath2 transition can be written as _ b q&= & g_a \|q _ ( f^qr_qb p_r - f^ql_qb p_l ) t^b b g^b_a [ eq : bq ] with @xmath52 where @xmath53 denotes the family order of the same type @xmath47 quark . based on eq . ( [ eq : bq ] ) , we study the impacts of non - universal axigluon exchange on @xmath54 processes and the time - dependent cpas in @xmath27 system . by eq . ( [ eq : bq ] ) , the effective hamiltonian for @xmath54 transitions which is generated by the tree - level axigluon mediation can be written as ^a_b=2 & = & , where @xmath55 denotes the number of colors and we have used the identity t^b_ij t^b_k = - _ ij _ k + _ i _ jk . [ eq : su3 ] in order to calculate the @xmath56 mixing , we write the relevant hadronic matrix elements to be @xmath57 to estimate the new physics effects , we employ the vacuum insertion method to calculate the above matrix elements , i.e. @xmath58 @xcite . additionally , in the heavy quark limit , we take @xmath59 . as a result , the transition matrix element for @xmath56 oscillation mediated by axigluon exchange becomes m^a , q_12 & = & b_q \| h^a_b=2\| \|b_q = m_b_q f^2_b_q u^d_qb , + u^d_qb & = & ( f^dr_qb)^2 + ( f^dl_qb)^2 + 4 f^dr_qb f^dl_qb . [ eq : uqb ] for reducing the number of free parameters , we will take the approximation @xmath60 in our analysis , i.e. @xmath61 , then @xmath62 . we note that the approximation @xmath63 can be realized in hermitian yukawa matrices @xcite . by combining the contributions of sm and axigluon , the transition matrix element for @xmath64 can be formulated as m^b_q_12 & = & \|m^sm , q_12\| r^q_a e^i2(_q + ^np_q ) , where the new parameters are defined by r^q_a & = & ( 1+(r^q_a)^2 + 2r^q_a 2(^np_q-_q))^1/2 , + 2^np_q & = & arg(m^a , q_12 ) , + r^q_a & = & , + 2^np_q & = & , [ eq : mbq ] and @xmath65 is given by @xcite m^sm , q_12= _ b m_b_qf^2_b_q_q ( v^_tqv_tb)^2 s_0(x_t ) with @xmath66 , @xmath67 and @xmath68 is the qcd correction to @xmath69 hence , the mass difference between heavy and light @xmath27 is @xmath70 . after obtaining @xmath71 , the time - dependent cpa through inclusive semileptonic decays can be defined as @xcite a^q_s&= & , + & = & with ( ) ^2 & = & , where @xmath72 denotes the absorptive part of @xmath73 transition . due to @xmath74 , the wrong - sign charge asymmetry can be simplified as a^q_s & = & i m ( ) ( 2_q + 2^np_q - ^_q ) . [ eq : aqsl ] here , @xmath75 stands for the phase of @xmath72 . since the absorptive part is dominated by the sm contribution , we will assume that @xmath76 in our numerical analysis . a detailed discussions about new physics effects on @xmath72 can be found in refs . @xcite . since @xmath77 is associated with the cp phases directly , a non - zero charge asymmetry will be an indication of cp violation . accordingly , the like - sign charge asymmetry defined in eq . ( [ eq : absl_exp ] ) can be written as @xcite ^b_s & = & , + & = & , [ eq : absl_1 ] where @xmath78 is the production fraction of @xmath27 and z_q & = & - , + y_q & = & , x_q = . using @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 and @xmath84 , the asymmetry can be rewritten as ^b_s = c_d a^d_s + c_s a^s_s [ eq : absl_2 ] with @xmath85 and @xmath86 @xcite . another important time dependent cpa can be defined by @xcite a_f_cp(t)&= & , + & = & s_f_cp m_b_q t - c_f_cp m_b_q t , + s_f_cp & = & , c_f_cp = [ eq : sf ] with _ f_cp & = & -()^1/2 = -e^-2i(_q + ^np_q ) , [ eq : lambdaf ] where @xmath87 denotes the final cp eigenstate , @xmath88 and @xmath89 are the so - called mixing - induced and direct cpas , @xmath90 and @xmath91 are the amplitudes of @xmath9 and @xmath92 mesons decaying to @xmath87 and @xmath93 with @xmath94 and @xmath95 are the cp eigenvalue of @xmath87 and the weak cp phase respectively . clearly , besides @xmath54 effects , the mixing - induced cpa is also related to the @xmath96 process . in this paper , we will concentrate on @xmath97 and @xmath98 for @xmath99 and on @xmath100 for @xmath101 . to calculate the decay amplitude of @xmath102 , we need to discuss the interactions of @xmath96 processes . with the approximation @xmath103 , the effective hamiltonian of @xmath104 can be expressed as _ bqqq & = & f^d_qb \|q _ _ 5 t^b b _ q=u , d , s , c \|q ^(g_+ p_r + g_- p_l ) t^b q with @xmath105 . using eq . ( [ eq : su3 ] ) , we can rewrite the last equation as ^a_bqqq & = & v^_tqv_tb [ eq : bqint ] in which the new wilson coefficients are expressed by c_q3 & = & , c_q4=-n_c c_q3 , + c^l_q5&=&-c_q3 , c^l_q6= -n_c c^l_q5 , + c_q5 & = & , c_q6 = - n_c c_q5 , + c^r_q3 & = & - c_q5 , c^r_q4 = -n_c c^r_q3 [ eq : coeff ] and the associated operators are o^q_3 & = & ( \|q b)_v - a _ q ( \|q q)_v - a , o^q_4 = ( \|q_b_)_v - a _ q ( \|q_q_)_v - a , + o^q_5 & = & ( \|q b)_v - a _ q ( \|q q)_v+a , o^q_6 = ( \|q_b_)_v - a _ q ( \|q_q_)_v+a , + o^qr_3 & = & ( \|q b)_v+a _ q ( \|q q)_v+a , o^qr_4 = ( \|q_b_)_v+a _ q ( \|q_q_)_v+a , + o^ql_5 & = & ( \|q b)_v+a _ q ( \|q q)_v - a , o^qr_6 = ( \|q_b_)_v+a _ q ( \|q_q_)_v - a with @xmath106 . besides the new free parameters that are introduced earlier , the non - leptonic @xmath9 decays suffer from large uncertain qcd effects such as @xmath107 . for estimating the new physics effects , we employ the naive factorization approach ( nfa ) . under the nfa , we find that the related effective wilson coefficients for @xmath108 and @xmath109 are c_s3 + + c^l_s5 + + c_s5 + + c^r_s3 + . [ eq : zerocs ] with the results in eq . ( [ eq : coeff ] ) , we clearly see that the influence of axigluon - mediated effects on @xmath110 modes vanishes . in our analysis we neglect the nonfactorizable contributions as they are subleading and difficult to estimate . now , only @xmath111 can display the axigluon - mediated effects . using nfa and the interactions of eq . ( [ eq : bqint ] ) , the total decay amplitude of @xmath111 is written as \|a_\|k^0&= & \|k^0 \| h_bss\|s\|\|b^0 , + & = & v^_ts v_tb ( a^sm+a_s4 + a^r_s4 ) \|\|s _ s\|0\|k^0\| \|s ^b\|\|b where @xmath112 is the sum of the sm and axigluon effective hamiltonian and @xmath113 with a_3&=&c_3 + , a_4 = c_4 + , a_5 = c_5 + , + & & a_s4=c_s4 + c_s3/n_c , a^r_s4= c^r_s4 + c^r_s3/n_c . here , @xmath114 are the effective wilson coefficients from the gluon penguin in the sm @xcite . we note that the electroweak penguin contributions are very small and thus we neglect them . using @xmath115 @xcite , we can write = -e^2i_s = -e^2i(_s + ^np_s ) with ^np_s & = & . by eqs . ( [ eq : sf ] ) and ( [ eq : lambdaf ] ) , the mixing - induced cpa via @xmath116 decay is obtained as s_k^0 & & 2_k^0 = 2(_d + ^np_d-_s -^np_s ) , [ eq : s_phik ] while the cpas through @xmath117 decays are given by s_j / k^0 & & 2_j / k^0 = 2(_d + ^np_d ) , + s_j/ & & 2^j/_s = 2(_s + ^np_s ) . [ eq : sjpsi_phi ] although the measurement of @xmath3 has approached to the precision level , however , it might be difficult to tell if there exists new physics by measuring @xmath3 only . nevertheless , one can investigate a new asymmetry defined by @xcite _ _ d = 2_j / k^0 -2_k^0 [ eq : dbeta ] which is less than @xmath118 in the sm @xcite . if a large value of @xmath119 is measured , it will be a strong hint for new physics beyond sm . so far , we have introduced seven new free parameters in the general chiral color models and they are : two gauge couplings @xmath120 , four parameters in the two complex quantities @xmath121 and @xmath122 . in order to display the dependence of @xmath119 on @xmath122 , we use the results in ref . @xcite and take @xmath123 and @xmath124 with @xmath125 . thus , the five remaining parameters are @xmath126 , @xmath127 for q = d , s and @xmath122 . we list the input values used for numerical calculations in table [ tab : inputs ] , where the relevant ckm matrix elements @xmath128 are obtained from the utfit collaboration @xcite , the decay constant of @xmath27 is referred to the result given by hpqcd collaboration @xcite , the cdf and d@xmath129 average value of @xmath130 is from ref . @xcite and the sm wilson coefficients of @xmath131 are obtained from ref . other inputs are obtained from particle data group ( pdg ) @xcite . .numerical inputs for the parameters in the sm . [ cols="^,^,^,^,^,^",options="header " , ] after setting up the inputs , we study the contributions of the axigluon to fcnc processes and their associated cpas that are defined earlier . we start by exploring the allowed parameter space . since the non - universal axigluon induces fcncs at tree level , the observed @xmath56 mixing parameter @xmath132 will give a strict constraint on the parameter space . in fig . [ fig : constraints](a)[(b ) ] , the allowed range for @xmath133}$ ] and @xmath134 ( in units of @xmath135 ) is drawn by the down - left hatched lines where we have taken the sm contributions ( @xmath136 , @xmath137 ) to be @xmath138 ps@xmath139 . furthermore , since the observed @xmath140 has been a precise measurement , it is plausible that the current data can further exclude the values of the parameter space which are allowed by @xmath141 . taking @xmath142 errors of @xmath143 as the experimental bound , the allowed region for @xmath144 and @xmath145 sketched by down - right hatched lines is plotted in fig . [ fig : constraints](a ) . clearly , @xmath146 gives a strong constraint on the parameters that contribute to @xmath147 . from fig . [ fig : constraints ] , we see that , except the two narrow regions correspond to @xmath148 gev@xmath139 , the allowed values of @xmath149 are limited to be @xmath150 gev@xmath139 , whereas the allowed values of @xmath151 can be one order of magnitude larger than those of @xmath149 . in general , the range of the cp violating phase @xmath127 is @xmath152 $ ] , for illustration , we just show the results within @xmath153 $ ] . the pattern of the constraint in @xmath154 $ ] is similar to that in @xmath155 $ ] . in order to illustrate the influence of the uncertainties of the sm on the free parameters , in fig . [ fig : tot ] we plot the allowed values of @xmath151 and @xmath156 by including the errors of @xmath157 and @xmath158 . comparing with fig . [ fig : constraints](b ) , we see that the allowed range is extended slightly . we note that due to the strict constraint of @xmath146 , the bounds on the parameters for @xmath4 transition are not changed significantly , therefore , we do nt show the corresponding diagram for @xmath4 transition . constraints on @xmath133}$ ] and @xmath159b}\|/m_v$ ] ( in units of @xmath135 ) obtained from @xmath160}-\bar b_{d[s]}$ ] mixing ( down - left hatched lines ) and @xmath161 ( down - right hatched lines ) . ] ( b ) , but the errors of @xmath162 in the sm are included.,width=4 ] according to eq . ( [ eq : absl_1 ] ) , if we assume no new cp violating phase in semi - leptonic decays , we will see that the charge asymmetry @xmath163 depends on two kinds of cp violating phases . one of the two phases is originated from @xmath164 mixing which is a @xmath4 transition , and the other phase is originated from @xmath165 which is associated with @xmath166 transition . in other words , we have to consider four parameters @xmath167 and @xmath168 simultaneously . however , if we consider @xmath169 transitions at the same time , we may induce a large effect on @xmath170 because the @xmath171 process is associated with @xmath172 , i.e. @xmath164 , @xmath165 and @xmath173 mixings have strong correlations . in order to avoid inducing a large @xmath173 mixing , we set a small value for @xmath174 . this is consistent with the results shown in fig . [ fig : constraints](a ) where @xmath141 and @xmath175 strongly constrain @xmath145 . hence , we assume that @xmath176 is dominated by the sm contribution where @xmath177 @xcite . consequently , the enhanced @xmath178 can be attributed to @xmath166 transition . with eqs . ( [ eq : mbq ] ) , ( [ eq : aqsl ] ) and ( [ eq : absl_2 ] ) and the values given in table [ tab : inputs ] , the contours of @xmath5 as a function of @xmath179 and @xmath180 are shown in fig . [ fig : absl](a ) where the values of the contours are in units of @xmath181 . as can be seen from the figure , not only the sign of @xmath5 can fit the data , but also its magnitude can be enhanced by axigluon - mediated effects . by combining with the constraint of @xmath162 , the region of @xmath179 for large @xmath178 is limited . in fig . [ fig : absl](b ) , we display @xmath182 as a function of @xmath179 where the solid , dashed and dash - dotted line represents @xmath183 gev@xmath139 , respectively . as shown in the figure , negative and positive values of @xmath179 can enhance @xmath5 . it should be noted that , although the axigluon - mediated effect can not enhance the like - sign charge asymmetry to be as large as the central value of d data , however , @xmath178 is enhanced by one order of magnitude larger than the sm prediction . as a function of @xmath179 and @xmath151 ( in units of @xmath135 ) . ( b ) @xmath163 as a function of @xmath156 , where the solid , dashed and dash - dotted line stands for @xmath184 , respectively . the values on the plot ( a ) are @xmath5 in units of @xmath181 . , width=5 ] unlike the case of the charge asymmetry , the time - dependent cpa of @xmath16 decay depends only on the cp phase in @xmath185 transition . as a consequence , when the new cp violating effects are small in @xmath147 , @xmath5 and @xmath186 defined in eq . ( [ eq : sf ] ) can have a strong correlation . by using eq . ( [ eq : sjpsi_phi ] ) , the contours of @xmath187 are plotted as a function of @xmath179 and @xmath180 in fig . [ fig : sjpsi_phi](a ) . from the figure , we find that large @xmath187 can be archived when @xmath5 is one order of magnitude larger than the sm prediction . moreover , we also plot @xmath187 as a function of @xmath188 in fig . [ fig : sjpsi_phi](b ) , where the solid , dashed and dash - dotted line denotes @xmath184 gev@xmath139 , respectively . clearly , a large @xmath5 indicates a large @xmath187 . although the measured values of @xmath5 and @xmath189 contain large errors , however , a few sigma deviations from the sm prediction can be considered as a hint for new physics effect . as a function of @xmath156 and @xmath151 ( in units of @xmath135 ) . ( b ) @xmath187 as a function of @xmath156 , where the solid , dashed and dash - dotted line represents @xmath190 , respectively . , width=5 ] it is well known that the golden process to measure the angle @xmath191 in the sm is @xmath192 which is dominated by tree diagram . although new physics can also affect this decay mode via @xmath193 transition , however , as discussed in eq . ( [ eq : zerocs ] ) , the axigluon contributions to @xmath194 vanish . hence , the source of the time - dependent cpa in @xmath195 decay is only originated from the @xmath24 oscillation . since @xmath191 is also a parameter in the sm , a single measurement of @xmath140 or @xmath196 is hard to uncover the new physics . to probe the new physics , the best way is to compare the cpa of @xmath197 with that of @xmath98 . therefore , we do not discuss each of @xmath140 and @xmath198 separately . instead , we focus on the cpa difference @xmath119 which is defined in eq . ( [ eq : dbeta ] ) and it is only few percent in the sm . by eqs . ( [ eq : s_phik ] ) and ( [ eq : dbeta ] ) , we see that although @xmath119 is insensitive to @xmath199 however it is strongly dependent on @xmath200 . to see the contributions of the axigluon to @xmath119 , we present the contours of @xmath119 as a function of @xmath179 and @xmath180 in fig . [ fig : delta_jpsi_phi](a)[(b ) ] , where we have set @xmath201 and figure ( a)[(b ) ] corresponds to @xmath202 $ ] tev . since the decay amplitude of @xmath111 depends on @xmath203 while @xmath162 is @xmath204 , thus a specific value for @xmath122 has to be given when calculating the contours of @xmath119 . for further understanding the @xmath156-dependence , we display @xmath119 as a function of @xmath179 in fig . [ fig : delta_jpsi_phi_sin ] , where figure ( a)[(b ) ] is for @xmath202 $ ] tev and the solid , dashed and dash - dotted line stands for @xmath205 gev@xmath139 , respectively . it is clear that the axigluon contributions to @xmath119 are larger than that of the sm . as a function of @xmath156 and @xmath151 ( in units of @xmath135 ) with ( a ) @xmath206 tev and ( b ) @xmath207 tev.,width=5 ] as a function of @xmath156 with ( a ) @xmath206 tev and ( b ) @xmath207 tev , where the solid , dashed and dash - dotted line represents @xmath190 , respectively . , width=5 ] in order to comprehend further the correlations among various physical observables under the influence of the axigluon , we display the scatter plots of @xmath163 , @xmath187 and @xmath119 with @xmath208 tev versus @xmath162 in fig . [ fig : correl1 ] , where we have chosen the range of @xmath179 to be @xmath153 $ ] . as an illustration , we also show the scatter plots of ( @xmath209 , @xmath187 ) and ( @xmath163 , @xmath119 ) with @xmath207 tev in fig . [ fig : correl2 ] , in which the constraint of @xmath162 has been included and @xmath179 belongs to @xmath153 $ ] . by fig . [ fig : correl2](a ) , we see that the correlation between @xmath163 and @xmath187 is linear , where this behavior can be understood by the linear dependence between the like - sign charge asymmetry and the mixing - induced cpa of @xmath15 . due to the linearity , we expect that the correlation between @xmath187 and @xmath119 should be similar to that between @xmath163 and @xmath119 . therefore , we just show the latter case in fig . [ fig : correl2](b ) . and ( a ) @xmath209 , ( b ) @xmath187 , ( c)[(d ) ] @xmath119 with @xmath210 $ ] tev , where the angle @xmath156 belongs to @xmath153$].,width=6 ] and @xmath187 and ( b ) correlation between @xmath163 and @xmath119 with @xmath207 tev , where the constraint of @xmath162 has been included and the angle @xmath188 belongs to @xmath153 $ ] . ] in general , a flavor non - universal axigluon in generalized chiral color models can induce fcncs at tree level . we study phenomenologically the axigluon - mediated effects on @xmath54 fcnc processes and the associated cpas . we find that although @xmath1 strongly constrain the free parameters , the precise measurement of @xmath140 can further exclude the parameter space of @xmath4 transition . furthermore , for avoiding inducing large @xmath173 mixing , the parameter @xmath174 is chosen to be small so that the like - sign charge asymmetry @xmath182 and @xmath119 are insensitive to the parameters of @xmath4 transition . as a result , the cp violating observables @xmath5 , @xmath211 and @xmath119 are strongly correlated and are only sensitive to the parameters of @xmath26 transition . by the study , we find that the axigluon effects do not only preserve the negative sign in @xmath5 , but also enhance its magnitude by one order of magnitude larger than the sm prediction . subsequently , the associated values of the parameters can also enhance the cpa @xmath187 and the cpa difference @xmath212 largely although they are only few percent in the sm . this work is supported by the national science council of r.o.c . under grant nsc-97 - 2112-m-006 - 001-my3 . the author c.h.c would like to thank prof . young - chung hsue for his help on using plot tool . g. faisel would like thank the national center for theoretical sciences ( ncts ) at cheng kung university for the hospitality where this work has been done . the heavy flavor averaging group _ et al . _ , arxiv:1010.1589 [ hep - ex ] . m. beneke and m. neubert , nucl . phys . b 675 * , 333 ( 2003 ) [ arxiv : hep - ph/0308039 ] . a. j. buras , r. fleischer , s. recksiegel and f. schwab , phys . * 92 * ( 2004 ) 101804 [ hep - ph/0312259 ] ; nucl . b * 697 * ( 2004 ) 133 [ hep - ph/0402112 ] . c. h. chen , c. q. geng and l. li , phys . b * 670 * , 374 ( 2009 ) [ arxiv:0808.0127 [ hep - ph ] ] . v. m. abazov _ et al . _ [ d0 collaboration ] , phys . d * 82 * , 032001 ( 2010 ) [ arxiv:1005.2757 [ hep - ex ] ] . a. lenz and u. nierste , jhep * 0706 * , 072 ( 2007 ) [ arxiv : hep - ph/0612167 ] ; a. j. lenz , aip conf . proc . * 1026 * , 36 ( 2008 ) [ arxiv:0802.0977 [ hep - ph ] ] . l. randall and s. f. su , nucl . b * 540 * , 37 ( 1999 ) [ arxiv : hep - ph/9807377 ] . a. dighe , a. kundu and s. nandi , phys . d * 82 * , 031502 ( 2010 ) [ arxiv:1005.4051 [ hep - ph ] ] . b. a. dobrescu , p. j. fox and a. martin , phys . lett . * 105 * , 041801 ( 2010 ) [ arxiv:1005.4238 [ hep - ph ] ] . d. choudhury and d. k. ghosh , arxiv:1006.2171 [ hep - ph ] . j. c. pati and a. salam , phys . b 58 , 333 ( 1975 ) . l. j. hall and a. e. nelson , phys . b 153 , 430 ( 1985 ) . p. h. frampton and s. l. glashow , phys . b 190 , 157 ( 1987 ) ; phys 58 , 2168 ( 1987 ) . j. bagger , c. schmidt and s. king , phys . d 37 , 1188 ( 1988 ) . arxiv : 0909.0307 [ hep - ph ] ; phys . 69 , 2889 ( 1992 ) ; p.h . frampton and b.h.lee , phys . lett.64 , 619 ( 1990 ) ; p. h. frampton and t. w. kephart , phys . rev . d 42 , 3892 ( 1990 ) ; p. h. frampton , p. i. krastev and j. t. liu , mod . lett . a 9 , 761 ( 1994 ) . l. m. sehgal and m. wanninger , phys . lett . b * 200 * , 211 ( 1988 ) . m. a. doncheski and r. w. robinett , phys . b * 412 * , 91 ( 1997 ) [ arxiv : hep - ph/9706490 ] . m. p. giordani [ cdf and d0 collaborations ] , eur . j. c * 33 * , s785 ( 2004 ) . d. choudhury , r. m. godbole , r. k. singh and k. wagh , phys . b * 657 * , 69 ( 2007 ) [ arxiv:0705.1499 [ hep - ph ] ] . v. barger _ et al . _ , phys . rev . d80 , 055008 ( 2009 ) [ arxiv:0902.4507 [ hep - ph ] ] ; v. barger , l. l. everett , j. jiang , p. langacker , t. liu and c. e. m. wagner , jhep * 0912 * , 048 ( 2009 ) [ arxiv:0906.3745 [ hep - ph ] ] . c. h. chen and h. hatanaka , phys . rev . d73 , 075003 ( 2006 ) [ arxiv : hep - ph/0602140 ] ; q. chang , x. q. li and y. d. yang , jhep * 0905 * , 056 ( 2009 ) [ arxiv:0903.0275 [ hep - ph ] ] ; c. w. chiang _ et al . _ , arxiv:0910.2929 [ hep - ph ] ; c. h. chen , phys . b * 683 * ( 2010 ) 160 [ arxiv:0911.3479 [ hep - ph ] ] ; j. hua , c. s. kim and y. li , arxiv:1002.2531 [ hep - ph ] j. hua , c. s. kim and y. li , arxiv:1002.2532 [ hep - ph ] ; q. chang , x. q. li and y. d. yang , jhep * 1004 * , 052 ( 2010 ) [ arxiv:1002.2758 [ hep - ph ] ] ; q. chang , x. q. li and y. d. yang , arxiv:1003.6051 [ hep - ph ] . v. m. abazov _ et al . _ [ d collaboration ] , phys . lett . * 100 * , 142002 ( 2008 ) [ arxiv:0712.0851 [ hep - ex ] ] . t. aaltonen _ et al . _ [ cdf collaboration ] , phys . * 101 * , 202001 ( 2008 ) [ arxiv:0806.2472 [ hep - ex ] ] ; cdf note at http://www - cdf.fnal.gov / physics / new / top/2009/tprop / afb/. a. arhrib , r. benbrik and c. h. chen , phys . d * 82 * , 034034 ( 2010 ) [ arxiv:0911.4875 [ hep - ph ] ] . f. gabbiani , e. gabrielli , a. masiero and l. silvestrini , nucl . phys . b * 477 * , 321 ( 1996 ) [ arxiv : hep - ph/9604387 ] . a. badin , f. gabbiani and a. a. petrov , phys . b * 653 * , 230 ( 2007 ) [ arxiv:0707.0294 [ hep - ph ] ] . c. h. chen , phys . b * 521 * , 315 ( 2001 ) [ arxiv : hep - ph/0110098 ] ; c. h. chen , phys . b * 541 * , 155 ( 2002 ) [ arxiv : hep - ph/0206143 ] . g. buchalla , a. j. buras and m. e. lautenbacher , rev . phys * 68 * , 1230 ( 1996 ) [ arxiv : hep - ph/9512380 ] . y. grossman and m. p. worah , phys . b * 395 * , 241 ( 1997 ) [ arxiv : hep - ph/9612269 ] ; c. h. chen and c. q. geng , phys . d * 71 * , 054012 ( 2005 ) [ arxiv : hep - ph/0403188 ] . m. bona _ et al . _ , arxiv:0906.0953 [ hep - ph ] .	a non - universal axigluon in generalized chiral color models leads to flavor changing neutral currents ( fcncs ) at tree level . we analyze phenomenologically the new contributions to @xmath0 ( q = d , s ) mixing and the related cp asymmetries ( cpas ) that are generated by axigluon exchange . we find that although @xmath1 can give a strict constraint on the parameters of @xmath2 transition , the precise measurement of @xmath3 can further exclude the parameter space of @xmath4 transition . the axigluon - mediated effects can enhance the like - sign dimuon charge asymmetry @xmath5 by one order of magnitude larger than the standard model prediction . accordingly , large cpa @xmath6 and cpa difference @xmath7 are achieved .
among the various modes of star formation , triggered star formation attracts a special interest due to an increasing availability of multi - wavelength observations of expanding shells in the milky way @xcite . the idea of self - stimulating star formation is quite old @xcite . however , there are still only a few examples of star - forming regions , where triggering can be inferred more or less reliably @xcite . the reason is that expanding shells often have irregular shapes , so it is hard to deduce an evolutionary sequence based on morphology alone . in our previous paper ( * hereafter paper i ) we presented arguments in favour of triggered star formation around an extended h region , sh2 - 235 ( hereafter s235 ) , using both morphological and kinematical data . s235 is located in the perseus spiral arm toward the anticentre of the galaxy at @xmath28 and @xmath3 . the region surrounding s235 is a site of active star formation . young star clusters in its vicinity were recently investigated , e.g. by @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite . in paper i we argued that the formation of three clusters , located close to the edge of s235 , could have been triggered by an expansion of the h region via a `` collect - and - collapse '' ( c&c ) process due to gravitational instability in a dense material accumulated between the ionisation front and the shock wave . in the present paper we refer to these clusters as s235east1 , s235east2 , and s235central , following the notation introduced in paper i. their cross - identification can be found in @xcite . the inhomogeneously distributed molecular gas around the s235 h region @xcite complicates an analysis of the triggering scenario . in paper i arguments in favour of triggering were found on the basis of 1 ) relative positions of the three embedded star clusters , dense molecular clumps , and the h region itself , and 2 ) kinematics of molecular gas emitting @xmath4co(10 ) and cs(21 ) lines . it was shown that the young star clusters s235east1 , s235east2 , and s235central are embedded in dense molecular gas which forms a shell - like structure at the south - eastern side of s235 . each cluster is embedded in a separate dense clump , which is clearly visible on a cs(21 ) map . the morphology and location of the sources and the distribution of the molecular gas are shown in fig . [ fig : general_view ] . note that s235 is not completely surrounded by the dense shell . there is some gas emitting in @xmath5co(10 ) and @xmath4co(10 ) lines to the north and west from s235 @xcite . however , the @xmath4co(10 ) brightness significantly decreases in these directions , so we had not observed that area during observing sessions described in paper i. also , the column density of molecular gas located in front of the h region decreases to the south of s235 as this gas becomes transparent to the optical emission . in paper i we proposed that expansion of s235 triggered the formation of a dense shell and subsequent formation of three young embedded stellar clusters . three kinematic components were distinguished in paper i that may represent basic stages of interaction of young stellar objects with the surrounding medium . the first component that we refer to as primordial arguably comprises quiescent clumpy gas of the parent giant molecular cloud g174 + 2.5 that has not yet been disturbed by the expansion of s235 . the second kinematic component represents gas that was entrained in motion and compressed by the expansion of the h region . this is the gas out of which the young star clusters have been formed . this gas still partially surrounds them . the third kinematic component supposedly corresponds to gas blown away by stellar winds from s235east2 and s235central ( see fig . [ fig : general_view ] , right panel ) . it was proposed in paper i that these two clusters are , probably , more evolved than s235east1 . the star forming region s235a - b encompasses both s235a and s235b nebulae and is located almost 10 to the south of s235central . the s235a nebula is an expanding h region @xcite . @xcite have shown that s235b is a reflection nebula . a water maser was found toward the region between s235a and s235b @xcite suggesting that massive star formation takes place there . stellar mass estimates by @xcite indicate that this water maser is associated with a young stellar object of high mass . the s235a - b region is not located in the compressed shell around the s235 h region . in paper i we suggested that star formation in s235a - b is not related to the expansion of s235 . this paper presents new observational data related to the star forming regions s235east1 , s235east2 , s235central in the immediate vicinity of s235 , and to regions around the s235a and s235b nebulae . the aim of this work is to determine physical conditions in these star forming regions . ammonia emission was observed to estimate gas temperature and number density as this molecule is known to be both a thermometer and a densitometer for dense molecular gas @xcite . the derived physical structure turns out to be consistent with our star formation scenario in this region and the inferred evolutionary stages of the young embedded star clusters . ammonia emission was mapped toward the s235east1 , s235east2 , and s235central star clusters as well as toward s235a - b and primordial molecular gas at other locations . we first searched for ( @xmath6 and ( 2,2 ) emission ( @xmath7 = 23.694496 and 23.722631ghz ) toward three selected positions with the pushchino 22-m telescope ( rt-22 , astro space centre of lebedev physical institute , pushchino , russian academy of sciences ) in april 2006 . rest frequencies of observed ammonia transitions were taken from the catalogue of @xcite . we used an on - on observing technique based on beam switching @xcite with a beam throw of 10 at 23ghz . the main beam efficiency was 0.38 and the calibration uncertainty is estimated to be 10@xmath815@xmath8 . we used the 50 mhz band of the pushchino autocorrelator with 2048 channels . the spectral resolution was about 24 khz . the full width at half power ( fwhp ) beam size of the 22-m telescope is about 2.6 which is larger than the size of the individual sources . therefore these observations are suitable to estimate average parameters in the region . the ammonia lines were detected toward the three selected positions , shown in fig . [ fig : general_view ] with dashed circles . the typical system temperature was about 150 k on an antenna temperature ( @xmath9 ) scale . pointing was checked twice a day . the values of introduced corrections ranged from 0 to 20 . emission from the ( @xmath10 , ( 2,2 ) , ( 3,3 ) , and ( 4,4 ) ammonia transitions was mapped with the 100-m effelsberg telescope in april and october 2007 . we observed areas in star forming regions around s235 that are bright in the cs(21 ) emission and host young embedded star clusters . long integrations were also performed toward locations of primordial gas . the regions observed in ammonia are shown in fig . [ fig : general_view ] ( left panel ) . the fwhp of the effelsberg telescope beam is about 40 at 23ghz . we used an 8192 channel correlator ( ak90 ) split into 8 separate bands with a bandwidth of 10 mhz and 1024 channels . thus we were able to obtain data on all four ammonia lines in two orthogonal linear polarizations simultaneously . the chosen backend configuration provided a channel spacing of 0.12 kms@xmath11 ( 9.77 khz ) . observations were done in a position switching mode . the off - position was chosen to be located outside the giant molecular cloud g174 + 2.5 @xcite and has offset coordinates ( @xmath12)= ( 15,0 ) with respect to our reference position of @xmath138 and @xmath3 . we checked the telescope pointing every hour . corrections for azimuth and elevation were typically @xmath14 , but sometimes @xmath15 . the focus was checked after sunrise and sunset . corrections for focusing were of the order of 1 mm . data calibration was performed by observing the calibrators ngc7027 , 3c123 and 3c286 . a standard 22 ghz gain curve @xcite was used to account for a change of the telescope gain with elevation . absolute and relative calibration uncertainties were estimated to be 15% and 5% , respectively . we spent about 12 minutes of on - time toward the majority of the positions . positions of particular interest were measured twice . the map spacing was chosen to be 40 . the system temperature was 200300k on a @xmath16 scale . the average noise level was 0.118 k on a @xmath16 scale . observations of the star forming region s235a - b were carried out in october 2007 . frontend and backend configurations were the same as for the previous observations , but the map spacing was reduced to 20 implying full sampling . the position of reference for the s235a - b spectra is @xmath174 and @xmath180 , the location of the hco@xmath19(1 - 0 ) emission peak @xcite . it is situated to the south of the s235 . the mapped area is also shown in fig . [ fig : general_view ] . data obtained with the rt-22 and the effelsberg telescope were reduced with the gildas software . we used gildas routine `` nh3(1,1 ) '' to estimate the optical depth @xmath20 of the ammonia @xmath21 line . spectra of other ammonia transitions were fitted with the `` gauss '' routine to determine their intensity and width . the standard interpretation of ammonia observations , described by @xcite , was utilised to determine ammonia column density @xmath22 and rotational temperature @xmath23 . we calculated the gas number density ( @xmath24 , essentially , number density of molecular hydrogen ) , using eq . 2 from @xcite , toward positions with determined values of @xmath20 and nh@xmath25 excitation temperature @xmath26 . rate coefficients for rotational excitation from @xcite were used to determine the kinetic temperature @xmath27 at positions with known @xmath23 . if data quality at a certain position was insufficient to determine @xmath20 , i. e. the uncertainty of the value was higher than the value itself , we did not use corresponding @xmath20 and @xmath26 estimates for further analysis . ammonia spectra , measured with the pushchino 22-m telescope , are shown in fig . [ fig : rt22_spectra ] . satellites of the nh@xmath25(1,1 ) line are seen toward all observed positions , while the hyperfine structure of the nh@xmath25(2,2 ) line is not pronounced . the quality of our spectra allows estimating @xmath27 toward all young clusters under study . the values are presented in table [ tab : rt22_res ] . the gas number density could be determined toward s235central because we could determine @xmath26 toward this position . numbers in parentheses in all tables show the values of uncertainties . table [ tab : rt22_res ] indicates that @xmath27 in the direction of s235east1 is considerably lower than @xmath27 toward s235east2 and s235central . as it was mentioned already , results of observations with rt-22 represent physical parameters averaged over a large area . ( 1,1 ) and nh@xmath25(2,2 ) emission toward selected positions , obtained with the rt-22 telescope . the two lines were observed simultaneously within a single band . solid lines show spectra fitted with the gildas `` nh3(1,1 ) '' and `` gauss '' routines for nh@xmath25(1,1 ) and nh@xmath25(2,2 ) lines , respectively . all spectra are hanning - smoothed . for the reference position , see fig [ fig : general_view ] . ] [ cols="^,^,^,^,^,^,^,^ " , ] [ tab : sizeclusters ] we describe ammonia observations toward star forming regions around s235 and present maps of nh@xmath25(1,1 ) and nh@xmath25(2,2 ) emission . the dense gas content in young embedded clusters s235east1 , s235east2 , s235central , as well as in the star forming region s235a - b is investigated . we determine temperature , ammonia column density and gas number density toward the clusters . gas kinematic structure and physical parameters suggest that at least two ( s235east2 and s235central ) of the three clusters nearest to s235 have been formed via triggering by a collect - and - collapse process if the density of the surrounding gas exceeds @xmath0 @xmath1 . the embedded cluster s235east1 could have been formed as a result of an interaction of the shock front from s235 with a pre - existing dense clump . the new observations are consistent with our previous conclusion that s235east1 is at an earlier evolutionary stage than s235east2 and s235central . parameters of the dense gas toward s235a imply that this h region expands into a medium with increasing density . the s235a h region is too young to trigger star formation around itself by a collect - and - collapse process . however , it can trigger formation of stars in pre - existing clumps . external large - scale shocks could have been responsible for shaping the s235a - b star forming region . co , cs , and nh@xmath25 lines are shifted relative to each other in s235east1 . a shift between co and nh@xmath25 lines confirm our conclusion from paper i that there is a relative motion between dense and diffuse gas in this object . we note also that star formation in s235a - b is not coeval . massive stars in the region are considerably younger than the lower mass stars in the region . this work was supported by the rfbr ( grants 13 - 02 - 00642 , 11 - 02 - 01332 ) and by the president of the rf grant nsh-3602.2012.2 . msk and dsw are grateful to the ofn-17 program of the russian academy of sciences . msk and ams thank the staff of the effelsberg observatory for their hospitality and assistance during observations . we thank aurora sicilia - aguilar for help in observations of s235a . allen l. e. , hora j. l. , megeath s. t. , deutsch l. k. , fazio g. g. , chavarria l. , dell r. w. , 2005 , in cesaroni r. , felli m. , churchwell e. , walmsley m. , eds , proc . 227 , massive star birth : a crossroads of astrophysics . cambridge university press , cambridge , p. 352 dyson j. e. , williams d. a. , 1997 , the physics of the interstellar medium . edition : 2nd ed . publisher : bristol : institute of physics publishing , 1997 . edited by j. e. dyson and d. a. williams . series : the graduate series in astronomy . isbn : 0750303069	gas density and temperature in star forming regions around sh2 - 235 are derived from ammonia line observations . this information is used to evaluate formation scenarios and to determine evolutionary stages of the young embedded clusters s235east1 , s235east2 , and s235central . we also estimate the gas mass in the embedded clusters and its ratio to the stellar mass . s235east1 appears to be less evolved than s235east2 and s235central . in s235east1 the molecular gas mass exceeds that in the other clusters . also , this cluster is more embedded in the parent gas cloud than the other two . comparison with a theoretical model shows that the formation of these three clusters could have been stimulated by the expansion of the sh2 - 235 h region ( hereafter s235 ) via a collect - and - collapse process , provided the density in the surrounding gas exceeds @xmath0 @xmath1 , or via collapse of pre - existing clumps . the expansion of s235 can not be responsible for star formation in the southern s235a - b region . however , formation of the massive stars in this region might have been triggered by a large - scale supernova shock . thus , triggered star formation in the studied region may come in three varieties , namely collect - and - collapse and collapse of pre - existing clumps , both initiated by expansion of the local h regions , and triggering by an external large - scale shock . we argue that the s235a h region expands into a highly non - uniform medium with increasing density . it is too young to trigger star formation in its vicinity by a collect - and - collapse process . there is an age spread inside the s235a - b region . massive stars in the s235a - b region are considerably younger than lower mass stars in the same area . this follows from the estimates of their ages and the ages of associated h regions . [ firstpage ] stars : formation hii regions ism : molecules open clusters and associations : general
we first recall a definition from [ an ] . [ de1 ] a solvable group @xmath0 has finite rank , if there is a decreasing sequence @xmath1 of subgroups , each normal in its predecessor , such that @xmath2 is abelian and @xmath3 is finite dimensional for all @xmath4 . in what follows @xmath5 denotes a free group with the number of generators @xmath6 . our main result is [ te1 ] let @xmath7 be a compact khler manifold . assume that the fundamental group @xmath8 is defined by the sequence @xmath9 where @xmath10 is a solvable group of finite rank of the form @xmath11 with non - trivial abelian groups @xmath12 so that @xmath13 and @xmath14 . assume also that @xmath15 does not admit a surjective homomorphism onto @xmath16 . then all eigen - characters of the conjugate action of @xmath17 on the vector space @xmath18 are torsion . in lemma [ le2 ] we will show that the condition for @xmath19 holds if @xmath20 does not admit a surjective homomorphism onto @xmath16 . using theorem [ te1 ] we prove a result on solvable quotients of khler groups . [ te2 ] assume that a khler group @xmath21 is defined by the sequence @xmath22 where @xmath20 does not admit a surjective homomorphism onto @xmath16 and @xmath10 is a solvable group of finite rank . then there exist normal subgroups @xmath23 of @xmath10 so that ( a ) : : @xmath24 has finite index in @xmath10 ; ( b ) : : @xmath25 is nilpotent , and ( c ) : : @xmath26 is torsion . [ re1 ] ( 1 ) clearly the conclusion of theorem [ te1 ] is valid for @xmath10 being an extension of @xmath27 by @xmath28 , @xmath29 , and @xmath20 being a finitely generated group . assume that not all eigen - characters of the action of @xmath27 on @xmath30 are torsion . then a semidirect product of such @xmath20 and @xmath10 ( where @xmath20 is a normal subgroup of this product ) is not a khler group . + ( 2 ) let @xmath21 be a khler group . by @xmath31 we denote the derived subgroup of @xmath21 , and set @xmath32 . assume that @xmath33 , @xmath34 , is a solvable group of finite rank . then it was proved in [ an , th . 4.9 ] and [ ca , th . 2.2 ] that @xmath10 satisfies conditions ( a)-(c ) of theorem [ te2 ] . it is a consequence of the fact that @xmath21 does not admit a surjective homomorphism onto @xmath5 with @xmath35 . in what follows @xmath36 denotes the lie group of upper triangular matrices . let @xmath37 be the lie group of matrices of the form @xmath38 @xmath39 and @xmath40 be the groups of diagonal and unipotent matrices . let @xmath7 be a compact khler manifold . for a homomorphism @xmath41 we let @xmath42 denote the upper diagonal character of @xmath43 . the main result used in our proofs is the following [ pro1 ] assume that @xmath8 is defined by the sequence @xmath44 where the normal subgroup @xmath20 does not admit a surjective homomorphism onto @xmath16 . assume that @xmath41 satisfies @xmath45 but @xmath46 . then @xmath47 is a torsion character . * proof . * given a character @xmath48 , let @xmath49 denote the associated @xmath8-module . we define @xmath50 to be the set of characters @xmath51 such that @xmath52 is nonzero . the structure of @xmath50 was described in the consequent papers of beauville [ be ] , simpson [ s ] , campana [ ca ] . + * bsc theorem . * _ there is a finite number of surjective holomorphic maps with connected fibres @xmath53 onto smooth compact complex curves of genus @xmath54 and torsion characters @xmath55 such that _ @xmath56 further , the group @xmath57 acts on @xmath58 by conjugation . any two homomorphisms from @xmath59 belonging to the orbit of this action will be called _ equivalent_. let @xmath41 satisfy the conditions of proposition [ pro1 ] . then it is well known that the class of equivalence of @xmath43 is uniquely defined by an element @xmath60 ( see e.g. [ a , prop . 2 ] ) . in particular , if @xmath61 , @xmath43 is equivalent to a representation into @xmath62 . in our case , @xmath63 because @xmath46 . thus @xmath47 satisfies the conditions of bsc theorem . if @xmath47 coincides with one of @xmath64 then it is torsion by the above theorem . so assume that @xmath65 for some torsion character @xmath66 and @xmath67 . let @xmath68 and @xmath69 be the galois covering of @xmath7 corresponding to the finite abelian galois group @xmath70 . let @xmath71 be the stein factorization of @xmath72 . here @xmath73 is a morphism with connected fibres onto a smooth curve @xmath74 and @xmath75 is a finite morphism . assume , to the contrary , that @xmath47 is not torsion . then we prove [ le1 ] @xmath76 . * proof . * set @xmath77 . according to the assumptions of proposition [ pro1 ] we have either ( a ) @xmath21 is a subgroup of finite index in @xmath78 , or ( b ) @xmath21 is isomorphic to @xmath5 with @xmath79 . let us consider ( a ) . since by definition @xmath45 and @xmath66 is torsion , @xmath80 is a finite abelian group . then @xmath81 is a subgroup of finite index in @xmath21 . in particular , @xmath82 is a subgroup of finite index in @xmath78 and so it is not free . but by our assumption , @xmath83 is not torsion and so @xmath84 . thus @xmath82 is also free as a subgroup of @xmath16 . this shows that ( a ) is never happen . consider now ( b ) . using the fact that @xmath85 is a surjection , we conclude that @xmath21 is a normal subgroup of @xmath78 . let @xmath86 be a regular covering corresponding to @xmath87 . if @xmath88 then the group @xmath89 of isometries of @xmath90 ( with respect to the hyperbolic metric ) is finite and since @xmath91 is infinite we have @xmath92 . if @xmath93 , any discrete subgroup of @xmath89 is virtually cyclic and in particular does not act cocompactly on @xmath90 . thus @xmath94 which means that @xmath95 and @xmath96 . then the assumption of the proposition implies that @xmath97 . now consider @xmath98 with the upper diagonal character @xmath99 . since @xmath100 is a subgroup of finite index , @xmath101 is also not torsion . set @xmath102 . then @xmath103 . let @xmath104 . then the same argument for @xmath82 as above for @xmath21 ( with @xmath101 and @xmath105 instead of @xmath47 and @xmath83 ) yields @xmath76 . @xmath106 according to lemma [ le1 ] and the assumptions of proposition [ pro1 ] we have that @xmath107 is non - trivial and @xmath101 is the pullback of a character from @xmath108 . then from [ br , proposition 3.6 ] it follows that @xmath101 is torsion . therefore @xmath47 is torsion , as well . this contradiction proves the proposition . @xmath106 we are ready to prove theorem [ te1 ] . according to the assumptions of the theorem there is a homomorphism @xmath4 of @xmath10 into the lie group @xmath109 of the form @xmath110 whose kernel is @xmath111 . here we idenitify @xmath112 with @xmath113 . consider the action @xmath114 by conjugation . since , by the definition of @xmath8 , @xmath17 is a finitely generated abelian group , @xmath115 where @xmath116 is equivalent to a nilpotent representation @xmath117 with a diagonal character @xmath118 . here @xmath119 . from this decomposition it follows that there is an invariant @xmath17-submodule @xmath120 of @xmath121 such that @xmath122 is a one - dimensional @xmath17-module and the action of @xmath17 on @xmath123 is defined as multiplication by the character @xmath118 . by definition , @xmath124 is a normal subgroup of @xmath109 and the quotient group @xmath125 is defined by the sequence @xmath126 here the action of @xmath17 on @xmath127 is multiplication by the character @xmath118 . ( as before the associated @xmath17-module is denoted by @xmath128 . ) let us denote by @xmath129 the composite homomorphism @xmath130 . further , the equivalence class of extensions of @xmath17 by @xmath127 isomorphic to @xmath131 is defined by an element @xmath132 . we assume that the character @xmath118 is non - trivial ( for otherwise , @xmath118 is clearly torsion ) . then @xmath133 ( for the proof see e.g. [ an , lemma 4.2 ] ) . this shows that @xmath131 is isomorphic to the semidirect product of @xmath127 and @xmath17 , i.e. , @xmath134 with multiplication @xmath135 let us determine a map @xmath136 of @xmath131 to @xmath58 by the formula @xmath137 obviously , @xmath136 is a correctly defined homomorphism with upper diagonal character @xmath118 . hence @xmath138 is a homomorphism non - trivial on @xmath15 by its definition . also @xmath139 . since by our assumptions @xmath19 does not admit a surjective homomorphism onto @xmath16 , proposition [ pro1 ] applied to @xmath140 implies that @xmath118 is torsion . this completes the proof of the theorem . @xmath106 we prove now the following result . [ le2 ] assume that a group @xmath21 is defined by the sequence @xmath141 where @xmath142 do not admit surjective homomorphisms onto @xmath16 . then @xmath21 satisfies the similar property . * proof of lemma [ le2 ] . * assume , to the contrary , that there is a surjective homomorphism @xmath143 . then @xmath144 is a normal subgroup of @xmath16 and @xmath145 is a quotient of @xmath146 . now the assumption of the lemma implies that @xmath147 with @xmath79 . let @xmath148 be a complex hyperbolic surface with @xmath149 and @xmath150 be the regular covering corresponding to @xmath151 . assume first that @xmath152 . since @xmath153 , any subgroup of the group @xmath89 of isometries of @xmath90 ( with respect to the hyperbolic metric ) is finitely generated . in particular @xmath151 is finitely generated ( as well as @xmath154 ) . this implies that @xmath16 should be finitely generated which is wrong . thus @xmath94 and @xmath155 . this contradicts to our assumption and shows that there is no such @xmath83 . for a group @xmath156 set @xmath157 . we say that an @xmath156-module @xmath158 is _ quasi - unipotent _ if there is a subgroup @xmath159 of finite index whose elements act unipotently on @xmath158 . to prove the theorem we will check the following condition from [ an , lemma 4.8 ] . * proof . * we set @xmath163 , @xmath164 , @xmath165 , and @xmath166 . here @xmath167 is a finitely generated abelian group . indeed , @xmath161 is finitely generated as a subgroup of finite index of the finitely generated group @xmath10 ( @xmath168 the image of the finitely generated group @xmath21 ) . thus @xmath167 is finitely generated as the image of @xmath161 . the fact that @xmath167 is abelian follows directly from the definition . further , since @xmath161 is a subgroup of finite index in @xmath10 , @xmath169 is a subgroup of finite index in @xmath21 . in particular , it is khler . moreover , we have @xmath170 where @xmath156 is a solvable group of finite rank defined by the sequence @xmath171 now the statement of the lemma is equivalent to the fact that all eigen - characters of the conjugate action of @xmath167 on @xmath162 are torsion . to prove that it suffices to show that @xmath172 does not admit a surjective homomorphism onto @xmath16 , and then to apply lemma [ le2 ] and theorem [ te1 ] . note that @xmath172 is defined by the sequence @xmath173 where @xmath90 is a solvable group of finite rank . by our assumption @xmath20 does not admit a surjective homomorphism onto @xmath16 . thus by lemma [ le2 ] , @xmath172 satisfies the same property . @xmath106 m.f.atiyah , complex analytic connections in fibre bundles , trans . * 85 * ( 1957 ) , 181 - 207 . d.arapura and m.nori , solvable fundamental groups of algebraic varieties and khler manifolds , comp . math . * 116 * ( 1999 ) , 173 - 188 . a.beauville , annulation du @xmath174 pour les fibrs en droites plats , lecture notes in math . , * 1507 * , springer verlag ( 1992 ) , 1 - 15 . a.brudnyi , solvable matrix representations of khler groups , preprint ( 2001 ) , 27 pp . f.campana , ensembles de green - lazarsfeld et quotients resolubles des groupes de khler , j. alg . geom . * 10 * ( 2001 ) , no . 4 , 599 - 622 . c.t.simpson , subspaces of moduli spaces of rank one local systems , ann . ens , * 26 * ( 1993 ) , 361 - 401 .	we prove several results on the structure of solvable quotients of fundamental groups of compact khler manifolds ( _ khler groups _ ) . -10 mm = msbm10 scaled 1 = msbm7 scaled 1 = msbm5 scaled 1 = = = p [ section ] [ th]lemma [ th]corollary [ th]definition [ th]proposition [ th]remark [ th]example
the scattering process of electromagnetic radiation by an atom is regarded as a second - order quantum mechanical process involving the annihilation of an incident photon and the creation of a scattered photon . therefore , from the second - order time dependent perturbation theory , we can obtain the scattering cross section , which is known as the kramers - heisenberg formula ( e.g. sakurai 1969 ) . the exact scattering cross section can be obtained after summing all bound state contributions and integrating over all continuum state contributions related with the dipole processes of annihilation and creation of photons . hydrogen being a single electron atom , the energy eigenstate wavefunctions of hydrogen are analytically known and hence each matrix element appearing in the kramers - heisenberg formula is given in a closed form . near resonance , the scattering cross section is excellently approximated by the lorentzian characterized by the narrow radiation damping term that can be interpreted as the life time of the excited state . the lorentzian dependence is obtained since the cross section is dominated by the single contribution from the excited state relevant to the resonance whereas the contributions from the other states are almost completely negligible . this lorentzian approximation is quite generally used in the astronomical community for analyses of resonance lines . far from resonance , the radiation damping term is negligibly important and the scattering cross section is approximately proportional to @xmath13 , where @xmath3 is the difference of the angular frequency of the incident radiation and that of the ly@xmath0 line center , @xmath2 . this cross section given by the lorentzian is symmetric with respect to the line center in frequency space . however , the true scattering cross section deviates from @xmath13 very far from the line center as the contributions increase from the other energy eigenstates including both bound and continuum states . the zeroth order term @xmath14 being symmetric with respect to the line center in frequency space , the first order term @xmath4 for this deviation will contribute to the asymmetry of the scattering cross section . therefore , a very careful quantitative consideration of this effect should be taken in order to obtain an accurate estimation of the true line center of high h i column density absorption systems . h i absorption systems are easily found in typical quasar spectra and are very useful to investigate the baryonic matter content and the large scale structure in the universe . the quasar h i absorption systems include the ly@xmath0 forest , lyman limit systems and damped ly@xmath0 systems , which are classified according to the h i column density @xmath15 . a number of damped ly@xmath0 systems ( dla s ) are known to exhibit h i column density in excess of @xmath16 ( e.g. turnshek & rao , 1998 ) . very high h i column density media may be found in the early universe , when the reionization is not complete and the universe is only partially ionized by the first objects ( gunn & peterson 1965 , scheuer 1965 ) . in these high column h i density media , the correct redshift of the absorbing system can be obtained after proper considerations of the asymmetric scattering cross section in frequency space . in this paper , we calculate the asymmetric deviation of the scattering cross section of radiation around ly@xmath0 by atomic hydrogen and apply our results to high h i column density scattering media that may be important in observational cosmology . we start with the kramers - heisenberg formula that is obtained from the fully quantum mechanical second - order time dependent theory ( e.g. sakurai 1969 , merzbacher 1970 ) . in terms of the matrix elements of the dipole operator @xmath17 , the differential cross section @xmath18 for incident radiation with angular frequency @xmath19 is given by the kramers - heisenberg formula , which can be written as @xmath20 where @xmath21 is the electron mass , @xmath22 is the classical electron radius , @xmath23 represents the matrix element of the position operator between the ground @xmath24 state and excited @xmath25 state , and @xmath26 are polarization vectors of incident and scattered radiation , respectively . here , @xmath27 is the angular frequency between the bound @xmath25 state and the ground @xmath24 state , and @xmath28 is the angular frequency of the transition between @xmath24 and the continuum @xmath29 state . in the atomic units adopted in this work , the bound @xmath25 state has an energy eigenvalue @xmath30 and correspondingly @xmath31 . similarly , @xmath32 and @xmath33 . in the limit @xmath34 , with the use of the closure relation it can be shown that the kramers - heisenberg formula yields a constant scattering cross section independent of the wavelength of incident radiation , which is the thomson scattering cross section @xmath35 . in the opposite limit where @xmath36 , another use of the completeness of the intermediate states gives @xmath37 \right\|^2\end{aligned}\ ] ] which shows the well - known @xmath38 dependence of the rayleigh scattering cross section in the low energy limit . in order to investigate the behavior of the cross section around ly@xmath0 , we introduce @xmath39 and expand the kramers - heisenberg formula in terms of @xmath40 . in the first term inside the summation of eq . 1 with @xmath41 , we can write @xmath42 in the same term for @xmath43 , we may write @xmath44.\ ] ] in a similar way , for @xmath45 , the second term in the summation in eq . 1 can be written as @xmath46.\ ] ] similar expressions for continuum states are obtained in a straightforward manner . making use of the wigner - eckart theorem we may separate the matrix elements of the position operator into the angular part and the radial part , where the radial part is given by the reduced matrix elements @xmath47 . ( e.g. merzbacher 1970 ) . the angular part gives the same dipole type angular distribution as that for thomson scattering or classical rayleigh scattering ( e.g. ahn , lee & lee 2000 , lee 1997 , schmid 1989 ) . substituting eqs 5,6 into eq . 1 , we obtain @xmath48 < x>_{n1}^2 \nonumber \\ & + & \sum_{n=2}^\infty { \delta\omega\omega_{n1}\over \omega_{n1}+\omega_{21 } } \left [ 1 -{\omega_{n1}\over \omega_{21 } } \sum_{k=1}^\infty \left ( { -\delta\omega\over\omega_{n1}+\omega_{21 } } \right)^k \right ] < x>_{n1}^2 \nonumber \\ & - & \int_0^\infty dn ' { \delta\omega\omega_{n'1}\over\omega_{n'1}-\omega_{21 } } \left [ 1 + { \omega_{n'1}\over \omega_{21 } } \sum_{k=1}^\infty \left ( { \delta\omega\over \omega_{n'1}-\omega_{21 } } \right)^k \right ] < x>_{n'1}^2 \nonumber \\ & + & \left . \int_0^\infty dn ' { \delta\omega\omega_{n'1}\over \omega_{n'1}+\omega_{21 } } \left [ 1 -{\omega_{n'1}\over \omega_{21 } } \sum_{k=1}^\infty \left ( { -\delta\omega\over\omega_{n'1}+\omega_{21 } } \right)^k \right ] < x>_{n'1}^2 \right\|^2\end{aligned}\ ] ] where the atomic units have been used and angular integration has been performed . here , the thomson scattering cross section @xmath49 . the matrix elements of the dipole operators @xmath50 are easily found in textbooks on quantum mechanics ( e.g. berestetski , lifshitz & pitaevskii 1971 , saslow & mills 1969 , bethe & salpeter 1967 ) . the matrix elements are given by @xmath51^{1\over2 } a_0\ ] ] for the bound states . here , the bohr radius @xmath52 is set equal to 1 in atomic units . for the continuum states , the corresponding values are given by @xmath53}\over { 3[(n')^2 + 1]^{5}[1-\exp(-2\pi n ' ) ] } \right]^{1\over2}\ ] ] where we also adopt the atomic units . . 7 can be written as @xmath54 where the coefficients @xmath55 are obtained through the following relations @xmath56<x>_{n1}^2 \nonumber \\ & -&\int_{0}^\infty dn ' { \omega_{n'1}^2\over\omega_{21 } } \left [ \left({\omega_{21}\over\omega_{n'1}-\omega_{21}}\right)^k -\left({-\omega_{21}\over\omega_{n'1}+\omega_{21}}\right)^k \right]<x>_{n'1}^2 . \end{aligned}\ ] ] here , @xmath7 is the oscillator strength for the ly@xmath0 transition . the computation of the coefficients can be done in a straightforward way , from which we obtain a list of these coefficients up to @xmath57 @xmath58 therefore , up to the first order in @xmath40 , the scattering cross section of radiation around ly@xmath0 @xmath59 in fig . 1 , we show the scattering cross section multiplied by a column density @xmath60 , that is the scattering optical depth of a neutral medium with this column density . the solid line represents the result obtained from the kramers - heisenberg formula , the dotted line depicts the first order approximation to it obtained in eq . 13 and the dashed line shows the zeroth approximation proportional to @xmath13 obtained from the lorentzian by ignoring the damping part . the inclusion of the first order term implies the introduction of asymmetry of the scattering cross section with respect to the line center in frequency space . as is apparent in fig . 1 , the scattering cross section is smaller than the lorentzian blueward of the line center , whereas it is larger in the opposite part . in fig . 2 we show the absorption profile @xmath61 that is expected from a damped ly@xmath0 system with a column density @xmath62 . the thick solid line represents the result obtained using the first - order corrected formula and the dotted line is the absorption profile obtained using the voigt profile by shifting the center wavelength by an amount @xmath63 to the blue part . the wavelength shift by this amount is determined using the least square method that best fits the thick solid line . because in the observational data it is difficult to obtain the exact continuum level , we minimize the difference @xmath64 in the wavelength interval between 1180 and 1250 , where the suppression of the continuum is larger than 20 percent . therefore , if we apply the voigt profile fitting process to observational data for an h i scattering system with @xmath62 , we may underestimate the redshift of the h i system by an amount of @xmath65 or in terms of the velocity scale @xmath66 . rao & turnshek ( 2000 ) investigated the damped ly@xmath0 absorption system along the line of sight to the quasar pks 1127 - 145 , which was best fitted with the voigt profile with @xmath67 ( see also rao et al . hence , the first order correction from our quantum mechanical consideration may affect the fourth decimal place of @xmath68 , which is not critically important considering the accuracy of @xmath69 quoted by the authors . among dla s so far investigated , the highest h i column density is @xmath70 ( rao & turnshek 2000 ) . according to braun ( 1997 ) , gas rich galaxies may contain a medium with @xmath71 . in fig . 3 we show the scattering optical depth for this h i column density . for a higher column density system with @xmath72 , we obtain @xmath73 , which may result in underestimation of the redshift by @xmath74 . this amounts to a velocity shift of @xmath75 , and may affect significantly the exact redshift determination of the absorbing system . gavrila ( 1967 ) provided a quantum mechanical computation of the rayleigh scattering cross section using the green function in momentum space . ferland ( 2001 ) obtained the power fit to the result of gavrila ( 1967 ) and incorporated it in his photoionization code ` cloudy , ' in which the formula is useful for radiation with @xmath76 and for radiation nearer the ly@xmath0 line center voigt function is proposed to be used . in the low energy limit where @xmath77 , the scattering cross section tends to be proportional to @xmath38 , which is also expected from classical physics . this approximation is quite useful in dealing with basic radiative transfer , because the voigt function ( or equivalently the lorentzian function ) well approximates the true scattering cross section . therefore , in this work , we emphasize the asymmetric deviation of the true scattering cross section from the lorentzian in frequency space , which may result in underestimation of the exact wavelength of the ly@xmath0 line center . careful comparisons of redshifts determined using other lines may provide important clues to the kinematics of the absorbing systems . in particular , prochaska & wolfe ( 1997 , 1998 ) proposed that the absorption profiles observed in many dla s are consistent with the hypothesis that they are formed in a rotating disk . the apparent line shift due to the asymmetric deviation around ly@xmath0 may affect the kinematic analysis of rotating galaxies . scattering of ly@xmath0 photons is also important in search for the first objects that are responsible for the reionization of the universe ( gunn & peterson 1965 , scheuer 1965 ) . in several high red shift quasars with @xmath78 the presence of the gunn - peterson trough has been reported blueward of ly@xmath0 ( becker et al . 2001 , fan et al . 2003 ) . the gunn - peterson optical depth will be important in the red wing part of ly@xmath0 ( e.g. miralda - escud 1998 ) . in eq . ( 13 ) the correction from the lorentzian cross section is about 1 % when we consider a width of the damped profile of about @xmath79 , for the characteristic expected value of the intergalactic medium ( igm ) density at around @xmath80 . hence , the change in the cross section is equivalent to a change in the redshift of the gunn - peterson trough of about 1 % of th ewidth of the damped profile or @xmath81 , when only the first order term in @xmath82 is considered . another high column density media may be found in the atmospheres of supergiants , in which classical rayleigh scattering is important ( e.g. isliker nussbaumer & vogel 1989 ) . also about a half of the symbiotic stars believed to consist of a heavily mass losing giant and a white dwarf exhibit raman scattered o vi features around 6830 and 7088 ( schmid 1989 , nussbaumer , schmid , & vogel 1989 ) . these features are originated from the resonance doublet o vi 1032 , 1038 line photons that are scattered in a neutral region with @xmath83 . however , due to resonant scattering by neutral hydrogen in interstellar space , it will be difficult to accurately determine the asymmetry of the scattering cross section . the author is very grateful to an anonymous referee for useful comments . this work is a result of research activities of the astrophysical research center for the structure and evolution of the cosmos ( arcsec ) funded by the korea science and engineering foundation . ahn , s. -h . , lee , h. -w . & lee , h. m. , journal of the korean astronomical society , 2000 , 33 , 29 becker , r. h. et al . 2001 , aj , 122 , 2850 berestetskii , v.b . , lifshitz , e.m . , & pitaevskii , l.p . , 1971 , relativistic quantum mechanics , pergamon press bethe , h. a. & salpeter , e. e. 1967 , quantum mechanics of one and two electron atoms , academic press inc . , new york braun , r. , 1997 , apj , 484 , 637 fan , x. et al . 2003 , aj , 125 , 1649 ferland , g. , 2001 , _ hazy , a brief introduction to cloudy 94.00 _ gavrila , m. , 1967 , physical review , 163 , 147 gunn , j. e. , peterson , b. a. , 1965 , apj , 142 , 1633 isliker , h. , nussbaumer , h. , & vogel , m. , 1989 , a & a , 219 , 271 lee , h. -w . , & lee , k. w. 1997 , mnras , 287 , 211 merzbacher , e. 1970 , quantum mechanics , wiley , new york miralda - escud , j. , 1998 , , 501 , 15 nussbaumer , h. , schmid , h. m. & vogel , m.,1989,a & a , 221 , l27 peebles , p. j. e. , 1993 , principles of physical cosmology , princeton university press , princeton prochaska , j. x. & wolfe , a. m. , 1997 , apj , 487 , 73 prochaska , j. x. & wolfe , a. m. , 1998 , apj , 507 , 113 rao , s. m. , nestor , d. b. , turnshek , d. a. , lane , w. m. , monier , e. m. , & bergeron , j. , 2003 , apj , rao , s. m. , turnshek , d. a. , 2000 , apjs , 130 , 1 sakurai , j. j. , 1967 , advanced quantum mechanics , addison - wesley publishing company , reading , massachusetts saslow , w. m. , mills , d. l. 1969 , physical review , 187 , 1025 scheuer , p. a. g. 1965 , nature , 207 , 963 schmid , h. m. 1989,a & a , 211 , l31 turnshek , d. a. , & rao , s. , 1998 , apj , 500 , l115	we investigate the asymmetry of the scattering cross section of radiation around ly@xmath0 by atomic hydrogen , which may be applied to analyses of scattering media with high column neutral hydrogen densities including damped ly@xmath0 absorption systems of quasars . the exact scattering cross section is given by the kramers - heisenberg formula obtained from the fully quantum mechanical second - order time dependent theory , where , in the case of hydrogen , each matrix element is given in a closed analytical form . the asymmetric deviation of the scattering cross section from the lorentzian near the line center is computed by expanding the kramers - heisenberg formula in terms of @xmath1 , where @xmath2 is the angular frequency of the ly@xmath0 transition and @xmath3 is the deviation of incident radiation from @xmath2 . to the first order of @xmath4 , we obtain @xmath5 , where @xmath6 is the thomson scattering cross section and @xmath7 is the oscillator strength for the ly@xmath0 transition . with this deviation , the line center of the damped wing profile apparently shifts blueward of the true ly@xmath0 line center . in the case of a damped ly@xmath0 system with a h i column density @xmath8 , the apparent line center shift relative to the true center amounts to @xmath9 resulting in an underestimation of redshift by @xmath10 . a measurable underestimation by an amount of @xmath11 is expected for absorbing systems with @xmath12 .
ultra high energy cosmic rays ( uhecr ) showers ( from pevs up to eevs and above , mainly of hadronic nature ) born at the high altitude in the atmosphere , may blaze ( from the far edge ) _ above the horizon _ toward telescopes such as magic one . the earliest gamma and cerenkov lights produced while they propagate through the atmosphere are severely absorbed because of the deep horizontal atmosphere column depth ( @xmath16 @xmath17 ) , must anyway survive and also revive : indeed additional diluted but penetrating muon bundles ( from the same by c.r . shower ) are decaying not far from the telescope into electrons which are source themselves of small cerenkov lights . also the same muon while hitting the telescope may blaze a ring or arc of cherenkov lights . these suppressed muon bundle secondaries , ( about @xmath18 times less abundant than the peak of the gamma shower photons ) are arising at high altitude , at an horizontal distances @xmath19 km far from the observer ( for a zenith angle @xmath20 while at 2.2 km . height ) ; therefore these hard ( tens - hundred gev ) muon shower bundles ( from ten to millions muons from tevs - eevs c.r . primary ) might spread in huge areas ( tens- hundred @xmath21 ) ; they are partially bent by geo - magnetic fields and they are randomly scattered , often decaying at tens - hundred gev energies , into electrons and consequent mini electromagnetic - showers traced by their optical cerenkov flashes . these diluted ( but spread and therefore better detectable ) brief ( nanosecond - microsecond ) optical signals may be captured as a cluster by largest telescope on ground as recent stereoscopic magic or hess , veritas arrays . their cerenkov flashes , single or clustered , must take place , at detection threshold , at least tens or hundreds times a night for magic - like telescope facing toward horizons @xmath22 . their `` guaranteed '' discover may offer a new tool in cr and uhecr detection . their primary hadronic signature might be hidden by the distance but its tail may arise in a new form by its secondary muon - electron - cerenkov of electromagnetic nature . on the same time _ below the horizons _ a more rare ( three - four order of magnitude ) but more exciting pev - eevs neutrino @xmath23 astronomy may arise by the earth - skimming horizontal tau air - showers ( hortaus ) ; these uhe taus are produced inside the earth crust by the primary uhe incoming neutrino @xmath6 , @xmath7 , generated mainly by their muon - tau neutrino oscillations from galactic or cosmic sources,@xcite,@xcite,@xcite . finally just above or below the horizon edge , within a few hundred of km distances , it might also be observable the guaranteed and well tuned @xmath24-@xmath25 air - showers at @xmath26 glashow resonant peak energy ; the w main hadronic ( @xmath27 ) or leptonic and electromagnetic ( @xmath28 ) signatures may be well observed and their rate might calibrate a new horizontal neutrino - multi - flavour astronomy @xcite . the @xmath24-@xmath25 of nearby nature ( respect to most far away ones at same zenith angle of hadronic nature ) would be better revealed by a stereoscopic magic twin telescope or a telescope array like hess , veritas . additional horizontal flashes might arise by cosmic uhe @xmath29 electromagnetic showers within most susy models , if uhecr are born in topological defect decay or in their annihilation , containing a relevant component of susy particles . the uhe @xmath29 behaves ( for light @xmath30 masses around z boson ones ) as the glashow resonant case @xcite . finally similar signals might be abundantly and better observed if uhe neutrinos share new extra - dimension ( tev gravity ) interactions : in this case also neutrino - nucleons interaction may be an abundant source of pevs - eevs horizontal showers originated in air @xcite . the total amount of air inspected within the solid angle @xmath31 by magic height at horizons ( @xmath32 km . ) exceed @xmath33 but their consequent detectable beamed volume are corresponding to an isotropic narrower volume : v@xmath34 @xmath35 , nevertheless comparable ( for pevs @xmath24-@xmath25 and eevs @xmath6 , @xmath36 showers ) to the present amanda confident volume . the ultrahigh energy cosmic rays ( uhecr ) have been studied in the past mainly versus their secondaries ( @xmath1 , @xmath4 , @xmath3 ) collected vertically in large array detectors on the ground . this is due to the rare event rate of the uhecr in the atmosphere and due to the high altitude where the shower takes place , expand and amplify downward . on the contrary at the horizons the uhecr are hardly observable ( but also rarely looked for ) . they are diluted both by the larger distances as well as by the exponential atmosphere opacity suppressing the electromagnetic ( electron pairs and gamma ) secondaries ; also their rich optical cerenkov signal is partially suppressed by the horizontal air opacity . however this suppression acts also as a useful filter leading to the choose of higher cr events ; their cerenkov lights will be scatter and partially transmitted ( @xmath37 at @xmath38 nm . , @xmath39 at @xmath40 nm . ) depending on the exact zenith angle and seeing : assuming a suppression on average @xmath41 and the nominal magic threshold at @xmath42 gev gamma energy , it does corresponds to a hadronic shower at far horizons ( diluted by nearly three order of magnitude by larger distances ) at an energy above @xmath43 pev . their primary flux maybe estimated considering the known cosmic ray fluxes at same energy on the top of the atmosphere ( both protons or helium ) ( see dice experiment referred in@xcite ) : @xmath44 ; the consequent event rate spread within a shower cerenkov angle @xmath45 at a distance @xmath46 ( zenith angle @xmath47 ) corresponding to a wide shower area @xmath48 $ ] , observed by a opening angle @xmath49 $ ] is for a night of record ( @xmath50 $ ] ) : @xmath51 therefore one may foresee nearly every two minutes a far hadronic cerenkov lightening shower in magic facing at the far horizons at zenith angle @xmath52 . increasing the observer altitude h , the horizon zenith angle also grows : @xmath53 $ ] in analogy at a more distant horizontal edges ( standing at height @xmath54 as for magic , while observing at zenith angle @xmath55 still above the horizons ) the observation range @xmath56 increases : @xmath57 ; the consequent shower area widen by more than an order of magnitude ( and more than three order respect to vertical showers ) and the foreseen event number , now for much harder c.r . at @xmath58 , becomes : @xmath59 therefore at the far edges of the horizons @xmath60 , once a night , an uhecr around eev energies , blazes to the magic ( or hess , veritas , telescopes ) . at each of these far primary cherenkov flash is associated a long trail of secondary muons in a very huge area ; these muons eventually are also hitting inside the telescope disk ; their nearby showering , while decaying into electrons in flight , ( source of tens - hundred gevs mini - gamma showers ) is also detectable at a rate discussed below . as already noted the main shower blazing photons from a cr may be also reborn or overlap with its secondary tens - hundred gevs muons , either decaying in flight as a gamma flashes , or by direct cerenkov muons lights painting arcs or rings while hitting the telescope . indeed these secondary very penetrating muon bundles may reach hundreds km far distances ( @xmath62 ) away from the shower origin . to be more precise a part of the muon primary energy will dissipate along @xmath32 km air - flight ( nearly a hundred gev energy ) , but a primary @xmath63 gev . muon will reach a final @xmath64 gev energy , just at minimal magic threshold value . let us remind the characteristic secondary abundance in a shower : @xmath65 these multiplicity are just at a minimal ( gev ) energies @xcite ; for the harder ( a hundred gev ) muons their number is ( almost inversely proportionally to energy ) reduced : @xmath66 these values must be compared with the larger peak multiplicity ( but much lower energy ) of electro - magnetic shower nature : @xmath67 . as mentioned most of these electromagnetic tail is lost ( exponentially ) at horizons ( above slant depth of a few hundreds of @xmath68)(out of the case of re - born , upgoing @xmath69 air - showers @xcite,@xcite ) ; therefore gamma - electron pairs are only partially regenerated by the penetrating muon decay in flight , @xmath61 as a parasite electromagnetic showering @xcite . indeed @xmath70 may decay in flight ( let say at @xmath71 gev energy , at @xmath72 level within a @xmath73 km distances ) and they may inject more and more lights , to their primary ( far born ) shower beam . these tens - hundred gevs horizontal muons and their associated mini - cerenkov @xmath1 showers have two main origin : ( 1 ) either a single muon mostly produced at hundreds km distance by a single ( hundreds gev - tev parental ) c.r . hadron primary ( a very dominant component ) or ( 2 ) rarer muon , part of a wider and spread horizontal muon bundle of large multiplicity born at tevs - pev or higher energies , as secondary of horizontal shower . between the two cases there is a smooth link . a whole continuous spectrum of multiplicity begins from an unique muon up to a multi muon shower production . the dominant noisy `` single '' muons at hundred - gev energies will loose memory of their primary low energy and hidden mini - shower , ( a hundreds gev or tevs hadrons ) ; a single muon will blaze just alone . the muon `` single '' rings or arcs frequency is larger ( than muon bundles ones ) and it is based on solid observational data ( @xcite ; @xcite , as shown in fig.2 and references on mutron experiment therein ) ; these `` noise '' event number is : @xmath74 the additional gamma mini - showers around the telescope due to a decay ( at a probability @xmath75 ) of those muons in flight , recorded within a larger collecting area @xmath76 is even a more frequent ( by a factor @xmath77 ) noisy signal : @xmath78 these single background gamma - showers must take place nearly once a minute ( in an silent hadronic background ) and they are an useful tool to be used as a prompt meter of the horizontal c.r . verification . on the contrary pevs ( or higher energy ) cr shower cerenkov lights maybe observed , more rarely , in coincidence both by their primary and by their later secondary arc and gamma mini - shower . their @xmath79 gev energetic muons are flying nearly undeflected @xmath80 for a characteristic horizons distances d , partially bent by geo - magnetic @xmath81 gauss fields ; as mentioned , to flight through the whole horizontal air column depth ( @xmath32 km equivalent to @xmath32 water depth ) the muon lose nearly @xmath71 gev ; consequently the origination muon energy should be a little above this threshold to be observed by magic : ( at least @xmath82 gev along most of the flights ) . the deflection angle is therefore a small one : @xmath83 ) . magic telescope area ( @xmath84 ) may record at first approximation the following event number of direct muon hitting the telescope , flashing as rings and arcs , each night : @xmath85 to be correlated ( at @xmath86 probability ) with the above results of @xmath87 primary cerenkov flashes at the far distances . as already mentioned before , in addition the same muons are decaying in flight at a minimal probability @xmath88 leading to a mini - gamma - shower event number in a quite wider area ( @xmath89 ) : @xmath90 therefore , in conclusion , at @xmath52 zenith angle , there are a flow of primary @xmath91 c.r . whose earliest showers and consequent secondary muon - arcs as well as nearby muon - electron mini - shower take place at comparable ( one every @xmath92 s. ) rate . these @xmath93 clustered signals offer an unique tool for gauging and calibrating magic ( as well as hess , cangaroo , veritas cerenkov telescope arrays ) for horizontal high energy cosmic ray showers . some more rare event may contain at once both rings , arcs and tail of gamma shower and cerenkov of far primary shower . it is possible to estimate also the observable muons - electron - cerenkov photons from up - going albedo muons observed by recent ground experiments @xcite @xcite : their flux is already suppressed at zenith angle @xmath94 by at least two order of magnitude and by four order for up - going zenith angles @xmath95 . pairs or bundles are nevertheless more rare ( up to @xmath96 @xcite @xcite ) . they are never associated to up - going shower out of the case of tau air - showers or by nearby glashow @xmath97 and comparable @xmath98 detectable by stereoscopic magic or hess array telescopes , selecting and evaluating their column depth origination , just discussed below . the appearance of horizontal uhe @xmath99 @xmath100 air - showers ( hortaus or earth - skimming neutrinos ) has been widely studied @xcite,@xcite , @xcite,@xcite,@xcite,@xcite,@xcite , @xcite,@xcite,@xcite ; their rise from the earth is source of rare clear signals for neutrino uhe astronomy ( see fig.3 ) . however also horizontal events by uhe @xmath10 pev , glashow @xmath97 and a possible comparable susy @xmath98 @xcite hitting and showering in air have non negligible event number : @xmath101 assuming the minimal gzk neutrino flux : @xmath102 ev @xmath103 . therefore during a year of night records and such a minimal gzk flux , a crown array of a @xmath104 magic - like telescopes on @xmath105 circle facing the horizons , would discover an event number comparable to a @xmath106 detector , ( nearly a dozen events a year ) . indeed magic facing at the horizons as it is , offer a detection comparable to present amanda @xmath107 effective volume . in conclusion while magic looking up see gamma gev astronomy , magic looking at horizons may well see uhe ( pevs - eevs ) cr , and rarely along the edge , gzk @xmath97 neutrinos , @xmath108 air - showers and , surprisingly even susy @xmath109 lights in the sky ( with showers ) .	high energy cosmic rays ( c.r . ) versus neutrino and neutralino induced air - shower maybe tested at horizons by their muons , gamma and cerenkov blazing signals . inclined and horizontal c.r . showers ( @xmath0 zenith angle ) produce secondary ( @xmath1,@xmath2 ) mostly suppressed by high column atmosphere depth . the air column depth suppresses low energy showers ( tev - pev ) and it dilutes higher energy ( pev - eevs ) ones . indeed also earliest shower cherenkov photons are diluted by large distances and by air opacity , while secondary penetrating , @xmath3 and their successive decay into @xmath4,@xmath1 , may revive additional cerenkov lights . the larger horizontal distances widen the shower s cone while the geo - magnetic field open it in a very characteristic fan - like shape polarized by local field vector , making these elongated showers spread , fork - shaped diluted and more frequent , up to three order of magnitude ( respect to vertical showers ) . gevs @xmath1 telescopes at the top of the mountains or in space may detect at horizons pevs up to eev and more energetic hadronic cosmic rays secondaries . details on arrival angle and column depth , shower shape , timing signature of photon flash intensity , may inform us on the altitude interaction and primary uhecr composition . below the horizons , at zenith angle ( @xmath5 ) among copious single albedo muons , rare up - going showers traced by muon ( @xmath4,@xmath1 ) bundles would give evidence of rare earth - skimming neutrinos , @xmath6 , @xmath7 , at eevs energies . they are arising by tau air - showers ( hortaus ) ( @xmath8 , @xmath9 hadrons and/or electromagnetic shower ) . their rate may be comparable with @xmath10 pevs @xmath11 neutrino induced air - shower ( mostly hadronic ) originated above and also below horizons , in interposed atmosphere by @xmath12 resonance at glashow peak . additional and complementary uhe susy @xmath13 at tens pevs - eev energy may blaze , as @xmath14 shower , by its characteristic electromagnetic signature . ( these uhe @xmath15 are expected in topological defect scenarios for uhecrs ) . their secondary shower blazing cerenkov lights and distances might be disentangled from uhecr by stereoscopic telescopes such as magic ones or hess array experiment . the horizontal detection sensitivity of magic in the present set up ( if totally devoted to the horizons shower search ) maybe already be comparable to amanda underground neutrino detector at pevs energies .
this work has been supported by the polish mnii under the contract pbz - min - p03/2003 . 99 l. i. glazman , m. e. raikh , jetp lett . * 47 * , 452 ( 1988 ) ; t. k. ng , p. a. lee , phys . rev . lett . * 61 * , 1768 ( 1988 ) . d. goldhaber - gordon _ et al . _ , nature * 391 * , 158 ( 1998 ) ; s. m. cronenwett _ et al . _ , science * 281 * , 540 ( 1998 ) . kouwenhoven , l. glazman , physics world , * jan . * , 33 ( 2001 ) . m. krawiec , k. i. wysokiski , solid state commun . * 115 * , 141 ( 2000 ) ; supercond . * 17 * , 103 ( 2004 ) . l. v. keldysh , sov . jetp * 20 * , 10 108 ( 1965 ) ; h. haug , a. p. yauho , _ quantum kinetics in transport and optics of semiconductors _ , springer , berlin ( 1996 ) . d. boese , r. fazio , europhys . lett . * 56 * , 576 ( 2001 ) ; b. dong , x. l. lei , j. phys . : condens . matter * 14 * , 11747 ( 2002 ) . r. scheibner _ _ cond - mat_/0410671 ( 2004 ) . a. m. lunde and k. flensberg , j. phys . : condens . matter * 17 * 3879 ( 2005 ) .	we study thermoelectric effects in kondo correlated quantum dot coupled to ferromagnetic electrodes by calculating thermopower s in the kondo regime as function of on - dot energy level and temperature . the system is represented by the anderson model and the results agree well with those recently measured for a quantum dot coupled to nonmagnetic leads . for magnetic electrodes one observes marked dependence of s on the degree of their polarization . , , quantum dot , thermopower , kondo effect 75.20.hr , 72.15.qm , 72.25.-b , 73.23.hk transport properties of artificial nanostructures with highly correlated electrons present fruitful research area . in particular the many body effects have been predicted @xcite and later observed @xcite in quantum dots coupled to external leads . such structures allow study of various physical phenomena in well controlled conditions and geometries . number of possibilities have already been tested experimentally resulting in discoveries of new effects @xcite . in this contribution we are interested in the study of the voltage @xmath0 across quantum dot induced by the temperature gradient characterised by @xmath1 . the ratio @xmath2 is a measure of the thermopower @xmath3 . we consider single quantum dot with only one energy level coupled to two external ferromagnetic leads . the second electron on the dot experiences strong repulsion @xmath4 due to charging energy @xmath5 , where @xmath6 is the capacitance of the dot . we model the system by the @xmath7 single impurity anderson hamiltonian in the slave boson representation and adopt the calculation scheme and the approximations presented previously in @xcite . @xmath8 where @xmath9 ( @xmath10 ) denotes left ( right ) lead , @xmath11 ( @xmath12 ) is the creation ( annihilation ) operator for a conduction electron with the wave vector @xmath13 , spin @xmath14 in the lead @xmath15 and @xmath16 is the hybridization matrix element between localized electron on the dot with the energy @xmath17 and conduction electron of energy @xmath18 in the lead @xmath15 . the non - equilibrium green s function technique @xcite allows one to study the general situation with arbitrary value of the @xmath1 . here , however , we limit our consideration to the linear response . this means that the response coefficients @xmath19 between charge and heat fluxes and chemical potential and temperature gradients are evaluated for an equilibrium ( @xmath20 ) system with constant temperature ( @xmath21 ) . electrical conductance @xmath22 is given by the coefficient @xmath23 as @xmath24 , while thermopower requires knowledge of both @xmath23 and @xmath25 and is given by @xmath26 . the linear response coefficients read @xmath27 @xmath28 ): @xmath29 , @xmath30 and @xmath31 . polarisation of the leads @xmath32 ( upper panel ) and @xmath33 ( lower panel ) . splitting of the kondo resonance results in a very small thermopower.,title="fig:",scaledwidth=38.0% ] ): @xmath29 , @xmath30 and @xmath31 . polarisation of the leads @xmath32 ( upper panel ) and @xmath33 ( lower panel ) . splitting of the kondo resonance results in a very small thermopower.,title="fig:",scaledwidth=38.0% ] as it has already been noted in @xcite in the context of n - qd - n system , the thermopower s is a powerful characteristic to identify the appearance of the kondo effect as it changes sign when the system is cooled down to temperatures @xmath34 below @xmath35 . this is particularly clear in fig . [ fig1 ] and in fig . [ fig2 ] ( solid curve ) , where s vanishes for temperature equal actual kondo temperature @xmath35 . the behaviour is more complicated in the fm - qd - fm system with nonzero polarisation ( lower panel of fig . [ fig1 ] and fig . [ fig2 ] ) . it is worth to note that the thermopower of the quantum dot defined in the two dimensional electron gas has recently been measured experimentally @xcite and qualitatively agrees with that presented in figure [ fig1 ] for @xmath36 . to understand their results the authors @xcite use the mott formula , which states that thermopower is proportional to the logarithmic derivative of the conductance with respect to the energy evaluated at the actual fermi energy . the kondo effect shows up as additional maximum appearing in conductance @xmath37 quantum dot when the temperature decreases . it is connected with the abrikosov - suhl resonance ( resonances ) in the density of states , which appears ( appear ) at low temperatures and is ( are ) located at the position(s ) of the fermi level of the leads in ( non)equilibrium situation . the slope of conductance thus changes and the thermopower changes sign . thermopower measured directly and obtained from the mott formula differ in line shapes @xcite and this calls for the discussion of the validity of mott formula @xcite in the interacting quantum dot systems . this as well as detailed comparison between experimental data and calculations will be the subject of future work . in magnetically polarised systems the abrikosov - suhl resonance splits ( as in external b field @xcite ) and this results in decrease of low temperature s ( fig . [ fig1 ] ) .
basing on the performed numerical analysis of the considered explicit examples for input amplitudes we can claim that * quasieikonal unitarization model under suitable choice of @xmath44 allows to describe scattering amplitude beyond the black disk limit . * both the qe and qu unitarization schemes can describe a possible ( hypothetic ) regime of hadron interaction where @xmath115 at @xmath18 and @xmath116 . * it is important to check out carefully ( at least numerically ) how the unitarity bounds for @xmath3 are satisfied when specific model for input amplitude @xmath13 is taken . * input simple pomeron pole must be modified before unitarization is applied in order to avoid an oscillation of @xmath117 in a region where @xmath118 . author thanks s.m . troshin for many challenging and fruitful discussions . the work is supported by the department of nuclear physics and power engineering of the national academy of sciences of ukraine ( grant no co-2 - 1/2014 ) . 99 a. alkin , o. kovalenko , e. martynov , s.m . troshin , phys.rev . * d89 * ( 2014 ) 091501(r ) . g. antchev _ et al . _ , totem collaboration , europhys . * 101 * ( 2013 ) 21002 . gribov , the theory of complex angular momentum , cambridge university press , 2003 . troshin , n.e . tyurin , phys . lett . * b 316 * ( 1993 ) 175 ; int . phys . * a 22 * ( 2007 ) 4437 . ter - martirosyan , pisma zh . exp theor phys . * 15 * ( 1972 ) 734 . glushko , n.a . kobylinsky , e.s . martynov , v.p.shelest , yad . fiz . * 38 * ( 1983 ) 180 . p. desgrolard , m. giffon , e. martynov , e. predazzi , z. phys . * c 76 * ( 1997 ) 155 . o. v. selyugin , j .- cudell , e. predazzi , eur.phys.j.st * 162 * ( 2008 ) 37 . cudell , e. predazzi , o.v . selyugin , phys.rev . * d79 * ( 2009 ) 034033 . kancheli , arxiv : hep - ph/1309.5860v2 ( 2014 ) . anisovich , v.a . nikonov , j. nyiri , arxiv : hep - ph/1408.0692 ( 2014 ) . troshin , arxiv : hep - ph/0712.3359v2 ( 2008 ) .	quasi - eikonal and quasi - u - matrix unitarization of the standard regge - pole amplitude for @xmath0 have been considered . we show that some violation of unitarity even at high energy exists in both models . we have found in quasi - eikonal model a bump - oscillation structure of @xmath1 at large values of impact parameter @xmath2 but where @xmath1 is closed to the maximal value . we argue that it is possible to choose the parameter regulating deviation of generalized models from pure eikonal or u - matrix modes in order to restore unitarity . it was shown in the recent paper @xcite that the impact - parameter amplitude @xmath3 extracted from @xmath4 elastic scattering data of the totem experiment at @xmath5 tev @xcite exceeds the black disk ( bdl ) limit @xmath6 . we define @xmath3 as the following transformation of standard scattering amplitude ( at high @xmath7 ) @xmath8 the extracted data for @xmath1 at @xmath5 tev are shown in fig . [ fig : imhdata ] ( the figure is taken from the @xcite ) . extracted from the totem @xmath9 data at @xmath107 tev , title="fig : " ] + this result , provided that it will be confirmed at higher energies ( 8 , 13 , 14 tev at lhc ) , leads to the important consequences for many phenomenological models constructed within a hypothesis that the bdl regime is realized in hadron elastic scattering at high energy . first of all it concerns with a widely explored eikonal model @xmath11 where usually and in accordance with regge approach an input amplitudes @xmath12 and @xmath13 are assumed to have the following properties . * amplitude @xmath12 is presumably imaginary at least at small @xmath14 . * amplitude @xmath15 where @xmath16 and @xmath17 gev at fixed impact parameter @xmath2 and at @xmath18 . * amplitude @xmath19 at @xmath20 where @xmath21 and @xmath22 . ) the impact - parameter amplitude must have an exponential , @xmath23 , where @xmath24 is a constant , rather than gauss behavior caused by contribution of regge pole with linear trajectory . ] then it can be concluded that the unitarized amplitude @xmath25 at @xmath18 is coming close to 1/2 at @xmath26 and going to zero at @xmath20 as shown in fig . [ fig : imh ] . + in another unitarization method , @xmath27-matrix unitarization @xcite , the output amplitude @xmath28 is expressed through @xmath13 by the following equation @xmath29 at @xmath30 the impact - parameter amplitude @xmath31 because @xmath32 . the height of a step in @xmath3 is not equal to 1 at any value of @xmath7 but it goes to 1 when @xmath18 . thus the @xmath27-matrix unitarization does not lead to any contradiction with above mentioned impact - parameter analysis of the totem data while in the eikonal method the maximal value of amplitude @xmath33 conflicts with the above mentioned analysis of the totem data . is it possible to fix the problem in eikonal or eikonal - type approach ? several modifications and generalizations of the eikonal as well as of the u - matrix methods have been suggested and explored long ago and recently @xcite . in what follows we will concentrate on two models , quasi - eikonal @xcite and quasi-@xmath27-matrix . they possess some interest and importance because a choice of additional parameter in these models allows to obtain asymptotical regime with the maximal value of the impact amplitude step between 1/2 and 1 , i.e @xmath34 . it is worthwhile to remind that the both models can be treated on the unique basis by summing multireggeon exchanges in @xmath7-channel or multiple rescatterings . namely , elastic scattering amplitude at high @xmath7 can be written as sum of @xmath35-pomeron exchanges . @xmath36 where @xmath37 the function @xmath38 is the vertex function that describes an interaction of two hadrons with @xmath35 pomerons , each of them being characterized by the impact parameter @xmath39 . the simplest assumption is that @xmath38 depends only on number of pomerons , @xmath35 . then we define @xmath40 and obtain @xmath41 let us consider two explicit forms of @xmath42 . assuming the first form @xmath43 one can immediately have the quasi - eikonal ( qe ) model @xcite ( in the pure eikonal model @xmath44 equals to 1 ) @xmath45 the second assumption that @xmath46 leads to quasi-@xmath27-matrix ( qu ) model ( in the original pure @xmath27-matrix model @xcite @xmath47 equals to 1/2 ) . then @xmath48 it follows from eqs . ( [ eq : qe],[eq : qu ] ) that @xmath49 ) at @xmath18 and @xmath50 . then from the unitarity inequality @xmath51 we have @xmath52 . thus , if @xmath53 then @xmath54 at @xmath18 in the region @xmath55 . many pro and contra arguments concerning qe - model were discussed in @xcite . in the present paper we give some additional arguments that such a model can be consistent with unitarity restrictions on @xmath3 . we would like to notice that the result of unitarization depends on both ingredients : input amplitude @xmath12 ( or @xmath13 ) and scheme that determines a dependence of output amplitude @xmath3 on input @xmath13 . we would like to notice that it is not necessary to require ( as is mentioned in @xcite ) that the whole upper semi - plane ( @xmath56 ) must be mapped under unitarization to the unitarity circle in ( @xmath57 ) . such a requirement can be valid only for a part of the semi - plane ( @xmath56 ) depending on the specific model for input @xmath13 . any scheme does not guaranty the correct properties of @xmath3 for an arbitrary @xmath58 . however , it is necessary for output amplitude to satisfy the unitarity requirements . from the unitarity equation @xmath59 where @xmath60 takes into account a contribution of inelastic processes , which is positive at @xmath61 ( for the sake of simplicity we consider identical hadrons with mass @xmath24 ) . unitarity imposes some restrictions on the both ingredients @xmath3 and @xmath13 of the unitarization scheme . it follows from a positivity of @xmath62 that @xmath63 let us consider now a quasi - eikonal model ( [ eq : qe ] ) with @xmath64 and take a contribution of simple regge pole defined in eq . ( [ eq : inputp ] ) as input amplitude . inequality @xmath65 makes no sense because in this case @xmath66 at @xmath18 , that is not supported by data . of course , other contributions ( @xmath67 reggeons , for example ) are important at not very high energy , but at energies @xmath68 exceeding or approximately equal to few hundred gev they are very small and can be neglected in the given qualitative analyses . let rewrite @xmath3 as follows @xmath69=-i(-1+\mu c+i\mu s)/2\lambda\ ] ] where @xmath70 . the inequality ( [ eq : unitar - ineq ] ) in terms of @xmath71 and @xmath72 has the form @xmath73 \1 . if @xmath74 and @xmath75 then @xmath76 and @xmath77 , inequality ( [ eq : ineq-0 ] ) is read as @xmath78 . \2 . if @xmath20 then @xmath79 and @xmath80 independently of energy and @xmath81 . obviously that in this case @xmath82 at any @xmath81 if @xmath83 . one can show that the above two conclusions 1 . and 2 . are still valid for the qu - model ( [ eq : qu ] ) as well . \3 . it is impossible to analyze analytically the unitarity inequalities at nonasymptotic energy and in a region of @xmath2 where @xmath3 is rapidly decreasing . that is why we calculate numerically @xmath1 and @xmath60 at some typical parameters of input amplitude @xmath13 , eq . ( [ eq : inputp ] ) . to this end we have fixed parameters @xmath84 and @xmath85 ( given in gev@xmath86 ) . their values and results obtained for @xmath1 and @xmath60 with various parameters @xmath87 as well as energy @xmath68 ( gev ) are shown in figs . [ fig : h - violation],[fig : h - e - u - s],[fig : h(b - b0)],[fig : h - evsqe ] . * we have found that unitarity inequalities ( [ eq : unitarity ] ) and ( [ eq : unitar - ineq ] ) are violated at `` low '' energy in both models ( qe at @xmath88 and qu at @xmath89 ) . it is well pronounced if the bare pole intercept @xmath90 in qe - model and @xmath91 ( see fig . [ fig : h - violation ] ) . it is not clear how to fix the problem . maybe a nonlinear trajectory of pomeron or an additional ( non regge - pole ? ) term in @xmath13 can help . * the violation is saving in qe - model while it is absent in qu - model at increasing energy . ( fig . [ fig : h - e - u - s ] ) . the violation is small but it exists . let us consider a region @xmath92 , where @xmath93 is determined from the following equation @xmath94=0.\ ] ] + in qe- and u - models , ] + and @xmath60 in qe - model ( left ) and u - model ( right ) at various energies ( values of @xmath95 are indicated near the curves @xmath1 ) . ] + ) at various energies with @xmath96 , @xmath97-values are given in figure , corresponding curves are pointed by arrows . @xmath98 . calculations were made with parameters : @xmath99 . the red line illustrates @xmath1 in pure eikonal model ( @xmath100 ) at @xmath101 ] * one can see ( fig . [ fig : h(b - b0 ) ] ) a bump followed by small oscillations ( they can be seen well on the larger scale of the figure ) above @xmath102 in qe - model at @xmath103 . it is important to notice that such a bump appears as well in the pure eikonal model with @xmath100 but above @xmath33 . it is illustrated in fig . [ fig : h(b - b0 ) ] on the right panel . thus we conclude that the black disk bound @xmath104 is violated in the pure eikonal unitarization of a simple pole with a linear trajectory and @xmath0 . we would like to notice that the same oscillation - bump structure exists in another generalization of e- and u - models considered in @xcite @xmath105.\ ] ] it interpolates between qu - model ( at @xmath106 ) and qe - model ( at @xmath107 ) . this structure appears at @xmath108 if input amplitude @xmath13 is the contribution of simple pomeron pole with line trajectory and @xmath0 . * the second important observation consists in a constant height of bump in a huge energy interval from ( @xmath109 up to @xmath110 ) and most likely at higher @xmath7 ) . it can be verified that bump height depends on @xmath111 while it does not depend on energy at very high @xmath7 . the same structure is appeared as well if pomeron trajectory is nonlinear ( for example if @xmath112 ) . however there is no such a structure if a term @xmath113 ( where @xmath93 is constant ) is added to @xmath13 . * it follows from these listed properties that unitarity violation in qe - model can be eliminated by an appropriate choice of @xmath44 , that is demonstrated in fig . [ fig : h - evsqe ] . the minimal value of @xmath44 is larger than 1/2 and depends on the input pomeron intercept @xmath114 . and @xmath60 by changing @xmath44 in qe - model ]
wall bounded turbulence is highly relevant to a variety of engineering systems and naturally occurring flows . as such , the importance of the fundamental understanding of the generation of turbulent fluctuations by fluid - solid interactions in the boundary layers and their transport and interaction with the bulk mean flow @xcite is evident . in many geophysical and industrial flows , shear flows and wall turbulence are subjected to either stable or unstable thermal stratifications . stable stratification is known to inhibit the bursting phenomenon , i.e. depleting the transport of turbulent fluctuations from the wall to the bulk , thus reducing the turbulent drag and increasing the mean flow velocity . many studies have been devoted to this situation when the stratified scalar field is passive @xcite and active @xcite and even in presence of non - oberbeck - boussinesq effects @xcite . the unstable configuration is also relevant in a variety of instances , such as the physics of the atmospheric surface layer superposed to an over - heated ground ( as , e.g. , in summer days , determining the generation of thermo - convective storms @xcite ) ; when buoyancy forces are strong enough , thermal convection can occur , altering significantly the unstable boundary layer dynamics . early theoretical results on turbulent boundary layers under unstable thermal stratification date back to @xcite who pioneered a mixing - length based approach which inspired the later work of @xcite anticipating the monin - obukhov similarity theory @xcite , and subsequently revised and compared with experimental data by @xcite . @xcite proposed an eddy - damped quasi - gaussian closure able to predict the inversion regions of heat flux profiles in strongly buoyant sheared boundary layers . numerical results @xcite and experimental flume measurements ( combined with a spectral equation model ) @xcite showed that natural thermal convection affects the mechanisms of momentum and heat transport from the wall and tends to flatten the velocity profile in the bulk ; these observation were confirmed by large - eddy simulations of a variable density fluid @xcite , although with non - boussinesq effects , as , e.g. profile asymmetries emerging at large stratifications . in this work we perform direct numerical simulations based on the lattice boltzmann ( hereafter lb ) method of an unstably ( thermally ) stratified turbulent channel flow . our focus is on identifying the effects of buoyancy on the channel flow structure by comparing profiles of mean fields and fluctuations over a wide range of parameters with a pure ( unstratified ) channel flow . our results show a decreased fluid throughput due to a strongly flattened velocity profile , which could , however , be fitted by a log - law with coefficients depending on the input controlling parameters ( friction reynolds number @xmath2 and rayleigh number @xmath1 , defined below ) , as also derived theoretically . we simulated a fluid enclosed between two walls , kept at fixed temperatures @xmath3 at @xmath4 and @xmath5 at @xmath6 , and driven by a constant pressure gradient @xmath7 along the streamwise direction @xmath8 ( see figure [ sketch ] ) . the equations of motion are the incompressible navier - stokes equation for the fluid velocity field @xmath9 in the boussinesq approximation ( namely , the density is assumed to be constant @xmath10 , but for the linearised buoyancy term ) @xmath11 ( @xmath12 being the mean temperature ) coupled with the advection - diffusion equation for the temperature field @xmath13 @xmath14 in the above equations , @xmath15 is the fluctuating pressure field ( rescaled by @xmath16 ) , @xmath17 the acceleration of gravity , @xmath18 a constant acceleration due to the imposed pressure head @xmath7 ; @xmath19 , @xmath20 and @xmath21 are the kinematic viscosity , the thermal diffusivity and thermal expansion coefficient , respectively . as a numerical scheme , we adopted a 3d lb algorithm @xcite with two probability densities ( for density / momentum and for temperature , respectively ) @xcite . the method has been extensively used to study both thermal convection @xcite and turbulent channel flow @xcite ; in particular @xcite validated the code and tested grid resolutions against previous studies with different numerical methods . we used a computational grid @xmath22 of @xmath23 lattice points , with each run longer than @xmath24 time steps ( in lb units ) , in such a way to achieve statistically steady states of @xmath25 ( @xmath26 being the large - scale eddy turnover time ) . the domain under consideration is sketched in figure [ sketch ] , where a snapshot of an instantaneous @xmath27 field is shown . $ ] from blue to red ) for @xmath28 and @xmath29 ( left ) and @xmath30 ( right ) . the physical domain is also displaced along with the coordinate system adopted in the text . ] from equations ( [ eq : ns ] ) and ( [ eq : temp ] ) two dimensionless groups can be identified giving rise to two parameters which control the dynamics , namely the shear ( or friction ) reynolds number , @xmath31 ( @xmath32 being the friction velocity ) and the rayleigh number @xmath33 quantifying , respectively , the strength of the pressure induced shear and of the buoyancy with respect to viscous dissipation . we performed turbulent channel flow simulations with @xmath34 $ ] ; for each @xmath0 we tuned the gravity ( at fixed temperature jump ) , whence the buoyancy , spanning the range @xmath35 $ ] . mean profiles of the streamwise velocity @xmath36 ( the overline indicates here and henceforth averaging over planes parallel to the walls @xmath37 ) in the channel are shown in figure [ 02u_mean_norm ] . for each reynolds number ( @xmath38 and @xmath39 ) we show data from simulations at three different rayleigh numbers ( @xmath40 , @xmath41 and @xmath42 ) , besides the unstratified channel flow ( @xmath29 ) . an evident effect of thermal stratification is a decrease of the centreline velocity and a flattening of the profiles at increasing @xmath1 ; indeed , mixing between the bulk and the boundary layer regions due to wall - normal thermal fluctuations ( or `` plumes '' ) results in an increase of the effective wall drag @xcite . such effects are more pronounced for lower @xmath0 , as it clearly appears by comparison of left and right panels of figure [ 02u_mean_norm ] . these observations will be discussed more quantitatively under the light of the modelling in the next section . , normalized with the friction velocity @xmath43 , for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . left : @xmath44 , right : @xmath45.,title="fig : " ] , normalized with the friction velocity @xmath43 , for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . left : @xmath44 , right : @xmath45.,title="fig : " ] figure [ 03t_mean_norm ] shows the mean temperature profile with respect to @xmath46 , for the same cases displayed in figure [ 02u_mean_norm ] . here , we note the bending of the thermal profile in the bulk with increased reynolds number , in contrast to the thermal shortcut observed in pure rayleigh - bnard ( at sufficiently high rayleigh numbers ) @xcite , owing to the destruction or sweeping of the coherent plumes rising from the wall surfaces . as a function of the height @xmath46 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . left : @xmath44 , right : @xmath45.,title="fig : " ] as a function of the height @xmath46 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . left : @xmath44 , right : @xmath45.,title="fig : " ] longitudinal and transverse mean squared components of the fluctuating velocity field ( hereafter @xmath47 ) are shown in figure [ 04re_stress ] for the same cases as in figure [ 02u_mean_norm ] ( normalized with the friction velocity ) . when the fluctuating quantities are concerned , most notably the lateral component is increased in the bulk region , and as the rayleigh number is increased , the magnitude becomes comparable or greater than the stream - wise component . the near wall regions are also affected , primarily in the streamwise component , where an increase and a subsequent decrease in magnitude is observed as the rayleigh number is increased . and transverse @xmath48 mean squared components of the fluctuating velocity field , normalized by the friction velocity @xmath43 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . ] and transverse @xmath48 mean squared components of the fluctuating velocity field , normalized by the friction velocity @xmath43 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . ] and transverse @xmath48 mean squared components of the fluctuating velocity field , normalized by the friction velocity @xmath43 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . ] and transverse @xmath48 mean squared components of the fluctuating velocity field , normalized by the friction velocity @xmath43 for various rayleigh numbers ( @xmath29 , @xmath40 , @xmath41 and @xmath42 ) at fixed shear reynolds number @xmath0 . ] in order to quantify the overall effect of the thermal forcing on the channel flow , figure [ 06c_f ] shows the skin - friction coefficient @xmath49 as a function of shear reynolds number @xmath0 . the increased drag due to the thermal field is evident , resulting in higher than usual friction coefficient . the effect is reduced asymptotically , as the reynolds number is increased for a given rayleigh number , emphasizing that shear becomes the dominant source of turbulence at sufficiently high reynolds numbers . as a function of the shear reynolds number @xmath2 for various rayleigh numbers @xmath1 . the dashed lines are predictions obtained by approximating the mean velocity by a log - law over the whole channel ; for the stratified cases we made use of equations ( [ eq : tlog - law])-([eq : kc ] ) with the same fitting parameter @xmath50 for all @xmath1 ( see section [ modelling].),scaledwidth=60.0% ] inspection of the profiles of the mean quantities and the fluctuations showed that buoyancy alters the channel flow structure . we now focus on the mean velocity profiles . in figure [ fig : vel - log - law ] we report the lin - log plot of profiles , in wall units , for @xmath51 and for various @xmath1 . convective motion , realised in the form of thermal plumes rising from the walls to the bulk , disturbs the coherence of the channel flow , resulting in a source of drag which decreases the mean velocity in the channel and hence the mass throughput , as well as flattening the velocity profile . ( left panel ) and rayleigh numbers @xmath29 ( @xmath52 ) , @xmath53 ( ) and @xmath54 ( ) . the dashed lines indicate the linear law , valid in the viscous layer , @xmath55 , while the dash - dotted ones correspond to the prediction ( [ eq : tlog - law])-([eq : kc ] ) with @xmath56 , @xmath57 and @xmath58 $ ] ( @xmath59 ) . ] in order to quantify these observations , in the following we attempt to generalize von krmn s law of the wall accounting for buoyancy effects . to this aim , we propose a simple model constructed from conservation laws and phenomenological arguments , which predicts that the viscous buffer is insensitive to buoyancy ( i.e. @xmath55 ) , while in the log - law region the following relation @xmath60 holds ( quantities have been expressed in wall units @xmath61 and @xmath62 , with @xmath63 ) . in equation ( [ eq : tlog - law ] ) , @xmath64 is the von krmn constant , @xmath65 is an integration constant and @xmath66 is given by @xmath67 @xmath68 is a phenomenological parameter . all profiles in figure [ fig : vel - log - law ] agree with our predictions : all curves collapse on the function @xmath69 for @xmath70 ( dashed lines ) while in the logarithmic layer they can be fitted with equation ( [ eq : tlog - law ] ) ( dash - dotted lines ) , whose derivation is detailed in what follows . + applying the reynolds decomposition @xmath71 for the velocity field and averaging the navier - stokes equation for the @xmath8 velocity component , we get , upon integration in @xmath46 , an exact relation @xcite between the mean shear @xmath72 and the reynolds stress @xmath73 namely : @xmath74 or , in wall units : @xmath75 notice that equations ( [ eq : avens1])-([eq : avens2 ] ) remain valid also in the presence of buoyancy , since the forcing term @xmath76 does not act along the streamwise direction . for very large @xmath2 ( in principle @xmath77 ) , equation ( [ eq : avens2 ] ) gives @xmath78 ; close to the wall ( for @xmath70 ) , then , where viscous terms dominate , @xmath79 whence @xmath80 which implies @xmath81 , as anticipated above . conversely , away from the wall ( typically for @xmath82 ) , turbulent fluctuations dominate over viscous processes and @xmath83 , i.e. @xcite @xmath84 such result alone , however , does not give any further insight on the behaviour of the velocity profile in the corresponding region of the channel . to this aim , we need to consider the budget equation of turbulent kinetic energy @xmath85 @xcite . in the log - law layer the energy production @xmath86 is balanced by dissipation @xmath87 @xcite ; the latter can not be calculated exactly but only inferred by means of phenomenological arguments : for large reynolds number and outside the viscous boundary layer , the dissipation equals the turbulent energy flux , which can be estimated as @xmath88 , where @xmath89 is the typical eddy turn - over time at @xmath46 @xcite . thus , the energy balance equation reads : @xmath90 where @xmath91 is a non - dimensional parameter of order unity . equation ( [ eq : clos1 ] ) is not yet closed : a relation between @xmath92 and @xmath93 is required . dimensional analysis suggests that the two quantities should be proportional to each other , i.e. @xmath94 @xcite ; such assumption can be readily tested in the numerical simulations : in the main panel of figure [ fig : checkdns ] we show that the ratio @xmath95 , indeed , approaches a constant value for @xmath96 . over the reynolds stress @xmath97 : the plateau attained for @xmath98 suggests that the two quantities are proportional to each other in the log - law region . inset : ratio of the turbulent heat flux over the mean shear ; the solid line indicates a linear dependence on @xmath99 and represents a numerical check of the validity of the assumptions leading to equation ( [ eq : ml3]).,scaledwidth=60.0% ] equation ( [ eq : clos1 ] ) can be then rewritten as ( in view also of ( [ eq : tauconst ] ) ) @xmath100 with @xmath101 yet another dimensionless number . the energy production by buoyancy term , viz . the heat flux , needs also to be modelled ; although more refined closures can be employed @xcite involving tensorial eddy diffusivities and coupling with gradients of the temperature field in all directions , we adopt a simple standard mixing length _ ansatz _ , i.e. @xmath102 a first order closure like ( [ eq : ml ] ) must not be expected to work well for second order quantities like temperature fluctuations , reynolds stresses , etc , but , as we will show , mean streamwise velocity profiles are satisfactorily reproduced through such model ; according to prandtl s hypothesis the mixing length @xmath103 is proportional to the distance from the wall , i.e. @xmath104 , whence @xmath105 where @xmath106 is a numerical constant . the mean temperature gradient will , in principle , depend itself on the shear profile ; however , in a perturbative spirit , we postulate here , for simplicity , a logarithmic form , such that @xmath107 where @xmath108 is a scale for the dynamic temperature . the form ( [ eq : meantprof ] ) somehow interpolates between the passive scalar case @xcite and the natural convection @xcite ; inserted in the expression for the heat flux ( [ eq : ml2 ] ) it gives @xmath109 for the sake of validation of our arguments , we check equation ( [ eq : ml3 ] ) ( which predicts @xmath110 ) against the numerics in the inset of figure [ fig : checkdns ] , finding a reasonably good agreement . inserting ( [ eq : ml3 ] ) into ( [ eq : clos2 ] ) provides @xmath111 which can be recast , introducing the wall units and the definitions of @xmath0 and @xmath1 , in the following form @xmath112 it must be noticed that the two parameters @xmath64 and @xmath68 are combinations of dimensionless quantities emerging in the derivation , which can not be , however , derived from first principles ; fits of the numerical data indicate that good estimates are the values @xmath56 and @xmath113 . explicitating the shear term in ( [ eq : clos3 ] ) , we get @xmath114 whose integration finally yields expressions ( [ eq : tlog - law ] ) and ( [ eq : kc ] ) for the velocity profile . the robustness of the model is confirmed in figure [ fig : tloglawconstant ] where we plot the fitted values of the parameter @xmath66 as function of @xmath1 for fixed @xmath0 ( main panel ) and as function of @xmath0 for fixed @xmath1 ( inset ) , together with the predictions of equation ( [ eq : kc ] ) ( dashed lines ) . of the model as function of @xmath1 for fixed @xmath0 ( main panel ) and as function of @xmath0 for fixed @xmath1 ( inset ) . the dashed lines represent equation ( [ eq : kc ] ) with constant @xmath68 fixed at @xmath57 for all cases.,scaledwidth=60.0% ] we have studied , by means of direct numerical simulations based on a thermal lattice boltzmann algorithm , the dynamics of a turbulent channel flow under a gravity field orthogonal to the streamwise direction coupled to an imposed temperature difference between the top ( cold ) wall and the bottom ( hot ) wall . the resulting unstably stratified configuration flattens the velocity profile and decreases the centreline value when the buoyancy strength is increased . this effective drag shows up in an enhancement of the friction coefficient . the action of buoyancy on the boundary layer structure has also been probed looking at other relevant statistical quantities in wall bounded turbulent system , such as reynolds stress ; we have found that , as the rayleigh number is increased , the squared wall normal velocity grows in the bulk becoming comparable or even larger than the streamwise component ( which , in turn , is depleted close to the wall ) . to provide a quantitative interpretation of the numerical findings , we have proposed a phenomenological model resulting in a modified logarithmic law of the boundary layer ; such model could successfully capture the various velocity profiles at changing the shear reynolds and rayleigh number , with just one adjustable parameter . + as , he and ag acknowledge financial support from the icelandic research fund . as acknowledges funding from the european research council under the eu seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement no[279004 ] . this work was partially supported by the foundation for fundamental research on matter ( fom ) , a part of the netherlands organisation for scientific research ( nwo ) , and by the cost action mp1305 .	we perform direct numerical simulations of an unstably stratified turbulent channel flow to address the effects of buoyancy on the boundary layer dynamics and mean field quantities . we systematically span a range of parameters in the space of friction reynolds number ( @xmath0 ) and rayleigh number ( @xmath1 ) . our focus is on deviations from the logarithmic law of the wall due to buoyant motion . the effects of convection in the relevant ranges are discussed providing measurements of mean profiles of velocity , temperature and reynolds stresses as well as of the friction coefficient . a phenomenological model is proposed and shown to capture the observed deviations of the velocity profile in the log - law region from the non - convective case .
let @xmath0 be an oriented @xmath1-component link in @xmath2 with @xmath3 . for an oriented knot @xmath4 , we denote the orientation - reversed knot by @xmath5 . if @xmath6 is an orientation - reversing ( orientation - preserving , respectively ) homeomorphism of @xmath2 so that @xmath7 for all @xmath8 where @xmath9 or @xmath10 , and @xmath11 is a permutation of @xmath12 , then @xmath13 is said an _ @xmath14-amphicheiral link _ ( an _ @xmath14-invertible link _ , respectively ) . a term amphicheiral link " is used as a general term for an @xmath14-amphicheiral link . a link is said an _ interchangeable link _ if it is an @xmath14-invertible link such that @xmath11 is not the identity . an @xmath14-invertible link is said an _ invertible link _ simply if there exists @xmath15 such that @xmath16 . if @xmath11 is the identity , then an amphicheiral link is said a _ component - preservingly amphicheiral link _ , and @xmath11 may be omitted from the notation . we mainly deal with component - preservingly amphicheiral link in the present paper . if every @xmath17 is identical for all @xmath8 ( including the case that @xmath11 is not the identity ) , then an @xmath14-amphicheiral link ( an @xmath14-invertible link , respectively ) is said an @xmath18-amphicheiral link ( an @xmath18-invertible link , respectively ) . we use the notations @xmath19 and @xmath20 . a link @xmath13 with at least @xmath21-component is said an _ algebraically split link _ if the linking number of every @xmath21-component sublink of @xmath13 is zero . we note that a component - preservingly @xmath18-amphicheiral link is an algebraically split link . necessary conditions for the alexander polynomials of amphicheiral knots are studied by r. hartley @xcite , r. hartley and a. kawauchi @xcite , and a. kawauchi @xcite . in @xcite , non - invertibility of @xmath22 is firstly proved by the conditions . on the other hand , t. sakai @xcite proved that any one - variable laurent polynomial @xmath23 over @xmath24 such that @xmath25 and @xmath26 is realized by the alexander polynomial of a strongly invertible knot in @xmath2 . b. jiang , x. lin , shicheng wang and y. wu @xcite showed that ( 1 ) a twisted whitehead doubled knot is amphicheiral if and only if it is the unknot or the figure eight knot , and ( 2 ) a prime link with at least @xmath21 components and up to @xmath27 crossings is component - preservingly @xmath28-amphicheiral if and only if it is the borromean rings . they used s. kojima and m. yamasaki s @xmath29-function @xcite . shida wang @xcite determined prime component - preservingly @xmath28-amphicheiral links with at least @xmath21 components and up to @xmath30 crossings by the same method as @xcite . there are four such links . for geometric studies of symmetries of arborescent knots , see f. bonahon and l. c. siebenmann @xcite . in @xcite , we determined symmetries such as invertibility , amphicheirality and interchangeability of @xmath21-bridge links by the parameters such as schubert s normal form , conway s normal form and conway s normal form whose entries are even integers . in the present paper , we study necessary conditions for the alexander polynomials of algebraically split component - preservingly amphicheiral links by computing the reidemeister torsions of surgered manifolds along the link . the results and the techniques of the present paper have been already applied to some directions . the author and a. kawauchi @xcite obtained a necessary condition by invariants deduced from the quadratic form of a link @xcite , showed a partial affirmative answer for the conjecture stated below , and determined prime amphicheiral links with up to @xmath27 crossings by combining with the present results . the author @xcite developed methods to detect component - preservingly amphicheiral links by the results in @xcite and @xcite , and determined prime amphicheiral links with up to @xmath30 crossings by techniques including the present results . there are 27 prime amphicheiral links with up to @xmath30 crossings . the techniques of the present paper are based on v. g. turaev @xcite . by the same techniques , the following two results on dehn surgeries are shown : in @xcite , the author , n. maruyama and m. shimozawa determined all dehn surgeries yieding lens spaces ( i.e. lens surgeries ) . in @xcite , we will show that the @xmath31-component milnor link with @xmath32 does not have a lens surgery by the reidemeister torsion and some geometric techniques . let @xmath33 be the alexander polynomial of @xmath13 which is an element of an @xmath1-variable laurent polynomial ring @xmath34 $ ] over @xmath24 where @xmath35 ( @xmath36 ) is a variable corresponding to a meridian of @xmath37 . we raise a conjecture : [ cj : cj1 ] for an even - component algebraically split component - preservingly amphicheiral link @xmath13 , we have @xmath38 . our results stated below support the conjecture . in section [ sec : rem ] , we explain about other supporting results in @xcite . we remark that the similar statement for the odd - component case does not hold . for example , a link which is connected sums of copies of the borromean rings is a component - preservingly amphicheiral link with odd components , and has the non - zero alexander polynomial . for any odd number , there are such examples . for an algebraically split component - preservingly amphicheiral link @xmath13 , if @xmath39 is obtained from @xmath13 by taking untwisted parallels of some components , and the number of components of @xmath39 is strictly greater than that of @xmath13 , then @xmath39 is also an algebraically split component - preservingly amphicheiral link , and we have @xmath40 . therefore we can not find a counterexample for the conjecture by the construction . we raise supporting examples in section [ sec : rem ] . if @xmath13 is an @xmath1-component algebraically split link with @xmath41 , then the alexander polynomial of @xmath13 is of the form : @xmath42 where we can take @xmath43 satisfying @xmath44 . note that @xmath45 is uniquely determined up to multiplication of @xmath46 . we set @xmath47 . if we take @xmath48 , then we denote @xmath49 , @xmath50 , and @xmath51 \cong { \mit \lambda}_s$ ] . if @xmath52 , then @xmath53 where we can take @xmath54 satisfying @xmath55 , and the sign of @xmath56 is uniquely determined by ( [ eq : torres1 ] ) and lemma [ lm : torres2 ] . in particular , @xmath57 . we set @xmath58 where @xmath59 . for @xmath60 , we use the following notations : @xmath61 we set @xmath62 and @xmath63 the following is our first main theorem : [ th : mt1 ] let @xmath0 be an @xmath1-component algebraically split component - preservingly amphicheiral link where @xmath41 , and @xmath64 . then for any @xmath65 , we have the following : 1 . if @xmath66 , then @xmath67 . 2 . if @xmath68 , then @xmath69 or @xmath67 . to prove the theorem , we compute the reidemeister torsions of the manifolds surgered along @xmath13 with associated coefficients to @xmath65 . we can deduce some corollaries . [ co : f=0 ] under the same assumption as in theorem [ th : mt1 ] , we have the following : 1 . if @xmath1 is even and @xmath66 , then @xmath70 . 2 . if @xmath71 ( i.e. @xmath72 ) and @xmath73 , then @xmath74 is divisible by @xmath75 . 3 . if @xmath71 ( i.e. @xmath72 ) and @xmath76 , then @xmath77 . in particular , if @xmath78 , then we have the following : [ co:2-comp1 ] if @xmath79 is an algebraically split component - preservingly amphicheiral link , then @xmath80 is divisible by @xmath81 . the following is our second main theorem : [ th : mt2 ] if @xmath0 is an @xmath1-component algebraically split component - preservingly @xmath18-amphicheiral link with @xmath1 even , and @xmath82 or @xmath10 , then the alexander polynomial of @xmath13 satisfies @xmath83 where @xmath84 @xmath85 . to prove the theorem , we span a seifert surface on @xmath13 . this method is a slightly extended argument in ( * ? ? ? * theorem 2.1 ) . in particular , if @xmath78 , then we have the following : [ co:2-comp2 ] if @xmath79 is an algebraically split @xmath86-amphicheiral link where @xmath82 or @xmath10 , then @xmath80 is divisible by @xmath87 . we remark that after proving the results above , the author and a. kawauchi @xcite showed that conjecture [ cj : cj1 ] is afffirmative for even - component algebraically split component - preservingly @xmath18-amphicheiral links by invariants deduced from the quadratic form of a link @xcite . in section [ sec : r - tor ] , we prepare facts on the alexander polynomials . in section [ sec : ampinv ] , we discuss about basic properties on amphicheiral links and invertible links . we give an almost purely algebraic proof for a lemma due to hartley @xcite . in section [ sec : mt1 ] , we prove theorem [ th : mt1 ] and its corollaries . in section [ sec : mt2 ] , we prove theorem [ th : mt2 ] . in section [ sec : rem ] , we raise some examples which support conjecture [ cj : cj1 ] . let @xmath88 be a finite cw complex , and @xmath89\to r$ ] is a ring homomorphism where @xmath90 is an integral domain . then we denote the reidemeister torsion of @xmath88 related to @xmath91 by @xmath92 where @xmath93 is the quotient field of @xmath90 ( see @xcite ) . the value @xmath94 is determined up to multiplication of @xmath95 . an equation between two values @xmath96 and @xmath97 is denoted by @xmath98 if @xmath99 for some @xmath100 . when @xmath91 is the identity , we denote @xmath94 by @xmath101 . the alexander polynomial is a kind of the reidemeister torsion . [ lm : alexander](@xcite ) let @xmath0 be an @xmath1-component link , and @xmath102 the complement of @xmath13 . then we have @xmath103 we will use the surgery formula to show theorem 1.1 . [ lm : surgery](surgery formula ) let @xmath104 be a compact @xmath105-manifold whose boundary consists of tori , @xmath106 a solid torus whose core is @xmath107 , and @xmath108 is the result of dehn filling where @xmath109 is an attaching map . let @xmath110\to r$ ] be a ring homomorphism where @xmath90 is an integral domain , and @xmath111\to r$ ] the induced map from @xmath91 . if @xmath112)\ne 1 $ ] , then we have @xmath113)-1)^{-1}.\ ] ] we raise some properties on the alexander polynomials . [ lm : dual](duality @xcite ) let @xmath0 be an @xmath1-component link . then we have @xmath114 where @xmath115 is even , and @xmath116 where @xmath117 @xmath85 . the torres condition is a special case of the surgery formula . [ lm : torres](torres condition @xcite ) let @xmath118 be an oriented @xmath119-compnent link , and @xmath120 an @xmath1-component sublink . then we have @xmath121 where @xmath122 is the linking number of @xmath123 and @xmath124 , and @xmath125 where @xmath126 is the linking number of @xmath37 and @xmath127 @xmath85 . one necessary condition for the alexander polynomial of an amphicheiral or invertible link is the following : [ lm : amp ] let @xmath0 be an @xmath1-component @xmath128-amphicheiral or invertible link where @xmath9 or @xmath10 @xmath85 . then we have @xmath129 we raise basic properties of amphicheiral links and invertible links . [ lm : sub ] let @xmath0 be an @xmath1-component link . 1 . if @xmath13 is an @xmath128-amphicheiral link , then a sublink @xmath130 @xmath131 is an @xmath132-amphicheiral link . if @xmath130 @xmath133 is an @xmath134-component sublink of @xmath13 such that @xmath135 is odd , and @xmath136 where @xmath137 is the linking number of @xmath138 and @xmath139 , then @xmath13 is not component - preservingly amphicheiral . if @xmath13 is an @xmath128-invertible link , then a sublink @xmath130 @xmath131 is an @xmath132-invertible link . 4 . let @xmath79 be a @xmath21-component link with non - zero linking number . if @xmath13 is an invertible link , then @xmath13 is a @xmath140-invertible link . the linking numbers are the first information to detect both amphicheirality and invertibility as in lemma [ lm : sub ] ( 2 ) and ( 4 ) . by lemma [ lm : sub ] ( 1 ) and ( 3 ) , to study sublinks is also important for the problems . hartley @xcite showed the following by the jsj ( jaco - shalen - johanson ) decomposition . the result is also about relation between amphicheirality and the linking number . we reprove it by another way . [ lm : even](hartley @xcite ) let @xmath79 be a @xmath21-component link with non - zero even linking number . then @xmath13 is not component - preservingly amphicheiral . * proof * suppose that @xmath13 is component - preservingly amphicheiral , and the linking number of @xmath123 and @xmath124 is non - zero and even . by lemma [ lm : dual ] , we may assume @xmath141 by lemma [ lm : amp ] and lemma [ lm : sub ] ( 4 ) , we may assume @xmath142 where @xmath143 or @xmath10 , and @xmath144 . by substituting @xmath145 to ( [ eq : dual ] ) , we have @xmath143 and @xmath146 . by substituting @xmath147 to ( [ eq : dual ] ) , we have @xmath148 , and hence we have @xmath149 by substituting @xmath150 to the equation above , we have @xmath151 . in the similar way , we have @xmath152 , and hence @xmath153 is divisible by @xmath154 . we set @xmath155 where @xmath156 . by substituting @xmath145 to @xmath153 , and lemma [ lm : torres ] , we have @xmath157 since the righthand side is not divisible by @xmath21 , we have a contradiction . both lemma [ lm : sub ] and lemma [ lm : even ] motivate us to study amphicheirality of algebraically split links . we remark that our proof of lemma [ lm : even ] works for the case that @xmath13 is in an integral homology sphere with an orientation - reversing autohomeomorphism . to prove theorem [ th : mt1 ] , we study the form of the alexander polynomial of an algebraically split link , and compute the reidemeister torsions of surgered manifolds along the link . let @xmath0 be an oriented @xmath1-component algebraically split link where @xmath41 . we add one component @xmath127 to @xmath13 such that @xmath158 @xmath85 is the connected sum of @xmath37 and the hopf link , where the linking number of @xmath37 and @xmath127 is @xmath159 . then we have @xmath160 we set @xmath161 . by lemma [ lm : torres ] , we have @xmath162 where @xmath163 and @xmath164 . by lemma [ lm : dual ] , we may assume that @xmath165 and @xmath166 where @xmath167 . for @xmath168 , we set @xmath49 , @xmath169 , and @xmath170 is obtained by substituting @xmath171 for all @xmath172 to @xmath173 . by ( [ eq : alex1 ] ) and ( [ eq : dual2 ] ) , if @xmath174 @xmath175 , then we may take @xmath176 where @xmath177 . if @xmath178 @xmath179 , then we may take @xmath180 where @xmath181 @xmath182 and @xmath183 we set @xmath184 and @xmath185 . we remark that @xmath186 and @xmath187 . [ lm : torres2 ] under the situation above , for @xmath188 , we have @xmath189 * proof * by applying lemma [ lm : torres ] repeatedly , we have the result . we exapand @xmath173-part in ( [ eq : torres1 ] ) as follows : [ lm : form ] if @xmath41 , then we have @xmath190 where @xmath191 , @xmath192 , and @xmath193 . * proof * we show by induction on @xmath1 . ( i ) the case @xmath78 . by lemma [ lm : torres2 ] and ( [ eq : torres2 ] ) , @xmath194 is divisible by @xmath195 . hence we have the result . ( ii ) the case @xmath196 . suppose the case @xmath197 . by ( [ eq : torres3 ] ) , lemma [ lm : torres2 ] and the assumption , we have @xmath198 where @xmath199 , @xmath191 , @xmath200 , @xmath201 , and @xmath202 . we set @xmath203 then we have @xmath204 in the similar way , if we substitute @xmath205 for any @xmath15 to @xmath206 , then we have @xmath207 , and hence @xmath206 is divisible by @xmath208 . let @xmath209 be the complement of @xmath210 , and @xmath211 the result of @xmath212-surgery along @xmath210 where @xmath213 . for @xmath48 , we suppose that @xmath214 if @xmath215 , @xmath216 if @xmath217 , and @xmath218 . we set @xmath58 and @xmath219 . if @xmath220 , then we set @xmath221 , and if @xmath222 , then we set @xmath223 . we set the natural inclusion @xmath224 . from now on , we consider only the case @xmath59 @xmath225 . let @xmath226 and @xmath227 @xmath228 be a meridian and a longitude of @xmath37 . we denote the homology class of a loop @xmath229 by @xmath230 $ ] . in @xmath231 , we have @xmath232=[m_{r+1}]\hfill & ( i=1 , \ldots , r)\hfill\medskip\\ [ l_{r+1}]=[m_1]\cdots [ m_r]\hfill & \end{matrix}\ ] ] in @xmath233 , we have @xmath234=[m_i]^{r_i}[l_i]=[m_i]^{r_i}[m_{r+1}]=1,\quad [ l_i']=[m_i]\qquad ( i\in i_r\setminus i)\ ] ] where @xmath235 and @xmath236 are a meridian and a longitude of the attaching solid torus for @xmath37 @xmath225 . in @xmath237 , we have @xmath238=[m_{r+1}]=1,\quad [ l_{r+1}']=[l_{r+1}]\ ] ] we note that @xmath239 @xmath240 and @xmath241 where we set @xmath242 $ ] @xmath228 . for @xmath243 and @xmath244 , we use the following notations : @xmath245 the reidemeister torsions of @xmath246 is the following : [ lm : r - tor1 ] suppose that @xmath41 and @xmath247 . 1 . if @xmath248 @xmath175 , then we have : @xmath249 where @xmath250 . 2 . if @xmath251 , then we have : @xmath252 where @xmath250 . * proof * by lemma [ lm : alexander ] , lemma [ lm : surgery ] and ( [ eq : relation2 ] ) , we have @xmath253 by ( [ eq : torres1 ] ) , we have @xmath254 by lemma [ lm : form ] , we have @xmath255 where @xmath256 and @xmath257 are the same as in lemma [ lm : form ] . in ( [ eq : tor3 ] ) , we note that @xmath258 and @xmath259 if and only if @xmath260 . by the fact , and ( [ eq : tor1 ] ) , ( [ eq : tor2 ] ) and ( [ eq : tor3 ] ) , we have the result . the reidemeister torsions of @xmath261 are the following : [ lm : r - tor2 ] suppose that @xmath41 and @xmath262 . 1 . if @xmath248 @xmath175 , then we have : @xmath263 2 . if @xmath251 , then we have : @xmath264 * proof * by ( [ eq : relation1 ] ) and ( [ eq : relation3 ] ) , we have @xmath265=t_{r+1}=1 $ ] and @xmath266=\prod_{i\in i}t_i$ ] in @xmath237 . hence by lemma [ lm : surgery ] , we have @xmath267 by combining with lemma [ lm : r - tor1 ] , we have the result . for @xmath64 with @xmath268 , and any @xmath58 , we set @xmath62 @xmath269 and @xmath270 is obtained from @xmath65 by replacing @xmath271 into @xmath272 for all @xmath217 . [ lm : minus ] @xmath273 * proof * for @xmath260 , since @xmath274 , we have @xmath275 and the results . * proof of theorem [ th : mt1 ] * suppose that @xmath13 is component - preservingly amphicheiral . then @xmath276 is homeomorphic to @xmath277 . ( 1 ) by lemma [ lm : r - tor2 ] ( 1 ) and lemma [ lm : minus ] , we have @xmath278 and @xmath279 since @xmath280 and @xmath281 , we have @xmath67 . ( 2 ) by lemma [ lm : r - tor2 ] ( 2 ) and lemma [ lm : minus ] , we have @xmath282 and @xmath283 since @xmath280 , we have @xmath69 or @xmath67 . * proof of corollary [ co : f=0 ] * ( 1 ) we set @xmath284 where @xmath285 and @xmath286 . we prove by induction on @xmath1 . suppose @xmath78 ( @xmath287 ) . by theorem [ th : mt1 ] ( 1 ) , we have @xmath288 . suppose @xmath289 . by the assumption of induction , we have @xmath290 for every @xmath291 such that @xmath292 is even , and @xmath293 . by theorem [ th : mt1 ] ( 1 ) , we have @xmath288 . ( 2 ) since @xmath294 , we have @xmath288 by theorem [ th : mt1 ] ( 2 ) . the equation @xmath70 holds if and only if @xmath74 is divisible by @xmath75 from lemma [ lm : dual ] . ( 3 ) since @xmath295 , we have @xmath296 by theorem [ th : mt1 ] ( 2 ) . by corollary [ co : f=0 ] ( 3 ) and lemma [ lm : sub ] ( 1 ) , for an algebraically split component - preservingly amphicheiral link @xmath13 , if we can add one component @xmath297 to @xmath13 satisfying that @xmath298 is also an algebraically split component - preservingly amphicheiral link such that @xmath299 is not divisible by @xmath300 where @xmath301 corresponds to a meridian of @xmath297 , then we have @xmath38 . we hope that it is possible for the case that @xmath13 is an algebraically split component - preservingly amphicheiral link with even components . if it is true , then conjecture [ cj : cj1 ] is affirmative . however , if @xmath13 is the borromean rings ( @xmath105-component link ) , then there does not exist a knot like @xmath297 . * proof of corollary [ co:2-comp1 ] * we take @xmath302 . by corollary [ co : f=0 ] ( 2 ) ( or corollary [ co : f=0 ] ( 1 ) ) , @xmath303 is divisible by @xmath304 . we can argue similarly for the case @xmath305 . therefore we have the result . we prove theorem [ th : mt2 ] by a slightly generalized argument of hartley ( * ? ? ? * theorem 2.1 ) . * proof of theorem [ th : mt2 ] * suppose that @xmath13 is oriented . we span a seifert surface @xmath306 corresponding the orientation . we set a seifert matrix from @xmath306 as @xmath307 . we can compute the one variable alexander polynomial of @xmath13 from @xmath307 as : @xmath308 where @xmath309 is the transposed matrix of @xmath307 . let @xmath6 be an orientation - reversing homeomorphism of @xmath2 . since @xmath13 is @xmath310-amphicheiral , @xmath311 is still a seifert surface of @xmath13 , and the corresponding seifert matrix changed into @xmath312 . since the s - equivalences do not change the one variable alexander polynomial , we have @xmath313 since the size of @xmath307 is odd , we have @xmath314 . we can argue similarly if @xmath37 is changed into @xmath315 . therefore we have the result . * proof of corollary [ co:2-comp2 ] * by corollary [ co:2-comp1 ] and theorem [ th : mt2 ] , we have the result . we raise examples which support conjecture [ cj : cj1 ] . ( 1 ) in figure 1 , let @xmath316 be the @xmath31-component milnor link @xcite where @xmath317 . in particular , @xmath318 is the borromean rings . the alexander polynomial of @xmath316 is @xmath319 and @xmath320 @xmath321 . the borromean rings @xmath318 is @xmath322-amphicheiral , but it is not @xmath323-amphicheiral ( cf . @xcite ) . ( 2 ) in figure 2 , let @xmath324 be a @xmath21-component @xmath21-bridge link where the number in a rectangle implies the number of half twists . a @xmath21-component amphicheiral @xmath21-bridge link is not algebraically split ( see @xcite ) , and the alexander polynomial of every non - trivial @xmath21-bridge link is not zero . as a special case , the alexander polynomial of @xmath325 @xmath326 ( @xmath327 is the whitehead link ) is @xmath328 by kanenobu s formula @xcite . we can see that @xmath13 is not amphicheiral by corollary [ co:2-comp1 ] . ( 3 ) for links with up to @xmath30 crossings , we use slightly modified notations in a web site maintaied by d. bar - natan and s. morrison @xcite . in the class , only @xmath330 and @xmath331 are algebraically split component - preservingly amphicheiral links with even components . moreover they are algebraically split component - preservingly @xmath28-amphicheiral links . we can confirm that the alexander polynomials of them are @xmath207 by direct computations or ( * ? ? ? * theorem 1.3 ) . we also remark that the condition component - preservingly " is needed . @xmath332 and @xmath333 are algebraically split amphicheiral links with even components which are not component - preservingly amphicheiral ( cf . the alexander polynomials of them are @xmath334 we note that @xmath335 satisfies the condition @xmath336 in theorem [ th : mt2 ] , and both @xmath332 and @xmath333 are @xmath337-amphicheiral where @xmath338 is the nontrivial permutation of @xmath339 . d. bar - natan and s. morrison , _ knot atlas _ , http://katlas.math.toronto.edu/wiki/main@xmath340page f. bonahon and l. c. siebenmann , _ new geometric splittings of classical knots and the classification and symmetries of arborescent knots _ , http://www-bcf.usc.edu/ fbonahon / research / preprints / bonsieb.pdf r. hartley , _ invertible amphicheiral knots _ , math . * 252 * ( 1980 ) , 103109 . r. hartley and a. kawauchi , _ polynomials of amphicheiral knots _ , math . * 243 * ( 1979 ) , 6370 . j. hillman , _ symmetries of knots and links , and invariants of abelian coverings ( part i ) _ , kobe j. math . , * 3 * ( 1986 ) , 727 . b. jiang , x. lin , shicheng wang and y. wu , _ achirality of knots and links _ , * 119 * ( 2002 ) , 185208 . t. kadokami , _ the link - symmetric groups of @xmath21-bridge links _ , to appear in jounal of knot theory and its ramifications . t. kadokami , _ amphicheiral links with special properties , ii _ , to appear in jounal of knot theory and its ramifications , arxiv math.gt/1107.0378 t. kadokami , _ finite slope surgeries along the milnor links _ , in preparation . t. kadokami and a. kawauchi , _ amphicheirality of links and alexander invariants _ , to appear in science china mathematics . t. kadokami , n. maruyama and m. shimozawa , _ lens surgeries along the @xmath341-twisted whitehead link _ , preprint . t. kanenobu , _ alexander polynomials of two - bridge links _ , j. austral . a ) , * 36 * ( 1984 ) , 5968 . a. kawauchi , _ on quadratic forms of @xmath105-manifolds _ , invent . , * 43 * ( 1977 ) , 177198 . a. kawauchi , _ the invertibility problem on amphicheiral excellent knots _ , proc . japan acad . , ser . a , math . , * 55 * ( 1979 ) , 399402 . a. kawauchi , _ a survey of knot theory _ , birkhuser verlag , ( 1996 ) . a. kawauchi , _ the quadratic form of a link _ , in : proc . of low dimensional topology , contemp . , * 233 * ( 1999 ) , 97116 . s. kojima and m. yamasaki , _ some new invariants of links _ , invent . math . , * 54 * ( 1979 ) , 213228 . j. milnor , _ link groups _ , ann . of math . ( 2 ) * 59 * ( 1954 ) , 177195 . j. milnor , _ a duality theorem for reidemeister torsion _ , ann . of math . ( 2 ) , * 76 * ( 1962 ) , 137147 . t. sakai , _ polynomials of invertible knots _ , math . ann . , * 266 * ( 1983 ) , 229232 . g. torres , _ on the alexander polynomial _ , ann . of math . ( 2 ) , * 57 * ( 1953 ) , 5789 . v. turaev , _ reidemeister torsion in knot theory _ , russian math . surveys , * 41 - 1 * ( 1986 ) , 119182 . shida wang , _ strict achirality of prime links up to @xmath30-crossings _ , acta math . sinica , * 24 * ( 2008 ) , 9971004 . w. whitten , _ symmetries of links _ , * 135 * ( 1969 ) , 213222 .	we provide necessary conditions for the alexander polynomials of algebraically split component - preservingly amphicheiral links . we raise a conjecture that the alexander polynomial of an algebraically split component - preservingly amphicheiral link with even components is zero . our necessary conditions and some examples support the conjecture . [ section ] [ df]lemma [ df]theorem [ df]remark [ df]proposition [ df]example [ df]corollary [ df]claim [ df]question [ df]problem [ df]conjecture addtoresetequationsection
the catalan numbers @xmath4 are ubiquitous in combinatorics . among other things they count dyck paths , which are lattice paths from @xmath5 to @xmath6 staying above the main diagonal , and rooted binary trees with @xmath7 leaves ( where each internal node has exactly 2 children ) . in this paper we present a perviously undiscovered connection between catalan numbers and certain schubert polynomials . our result also has an interpretation in terms of the geometry of certain schubert varieties . let @xmath8 be a permutation . the schubert polynomial @xmath0 is a polynomial representative for the class of the schubert variety @xmath9 in the flag manifold @xmath10 . a combinatorial formula for schubert polynomials , which for our purposes can be taken as their definition , was proven independently in @xcite and @xcite . diagrams illustrating this formula , called rc - graphs , were introduced in @xcite and extensively studied in @xcite . for the remainder of this paper , let @xmath11 denote the permutation @xmath12 in @xmath13 . using rc - graphs , we will show that that the principal specialization of the schubert polynomial for this permutation , @xmath14 , is equal to @xmath15 , where @xmath16 is the carlitz - riordan @xmath3-analogue of the catalan numbers originally introduced in @xcite . section [ sect : rc - graphs ] gives the definition of rc - graphs and proves our main theorem via a recurrence counting rc - graphs for @xmath11 . section [ sect : elem - bij ] gives a bijection to dyck paths , and section [ sect : edel - green ] shows that this bijection can also be described using the edelman - greene correspondence . since @xmath11 is its own inverse in @xmath13 , transposition is a natural involution on its rc - graphs . in section [ sect : tree - inv ] , we describe our bijection in terms of binary trees , giving a correspondence between features of the rc - graph known as elbow joints and the internal nodes of a binary tree . from this description it will be evident that transposing an rc - graph corresponds to flipping a binary tree around its vertical axis . finally we discuss in section [ sect : geom ] the geometric example which originally motivated this study . let @xmath17 be the permutation @xmath18 in @xmath19 . we show that the multiplicity of the schubert variety @xmath20 at its most singular point is given by @xmath1 . experimental evidence @xcite suggests that this is higher than the multiplicity of any point on any other schubert variety of @xmath21 . let @xmath8 be a permutation . an _ rc - graph _ for @xmath22 is a filling of the upper - left half of an @xmath23 array with cross pieces ( @xmath24 ) and elbow joints ( @xmath25 ) such that the strand entering the left in row @xmath26 exits the top in column @xmath27 , with the additional condition that no two strands cross more than once . this second condition can alternatively be stated as there being exactly @xmath28 cross pieces . for example , the 5 rc - graphs for @xmath29 are shown in figure [ fig : rc - graphs ] . it was shown independently in @xcite and @xcite that rc - graphs are related to schubert polynomials by the formula @xmath30 where @xmath31 is the set of rc - graphs for @xmath22 , and @xmath32 are the locations of the cross pieces in @xmath33 ( indexed so that @xmath34 would be in @xmath32 if @xmath33 has a cross in the top row and the third column ) . for example , we have @xmath35 , with the 5 terms corresponding to the 5 rc - graphs in the figure [ fig : rc - graphs ] from left to right . 4 & 1&2&3&4 + 1 & & & & + 4 & & & & + 3 & & & & + 2 & & & & + 4 & 1&2&3&4 + 1 & & & & + 4 & & & & + 3 & & & & + 2 & & & & + 4 & 1&2&3&4 + 1 & & & & + 4 & & & & + 3 & & & & + 2 & & & & + 4 & 1&2&3&4 + 1 & & & & + 4 & & & & + 3 & & & & + 2 & & & & + 4 & 1&2&3&4 + 1 & & & & + 4 & & & & + 3 & & & & + 2 & & & & + first we will show that there are in fact catalan many rc - graphs for @xmath11 ; this will allow us to prove that the combinatorial maps we give from the set of rc - graphs for @xmath11 to other catalan objects are in fact bijections by only showing that they are injections or surjections . the carlitz - riordan @xmath3-catalan numbers @xmath16 are defined by the recurrence @xmath36 , with @xmath37 ; under the interpretation of catalan numbers as counting partitions @xmath38 whose young diagrams fit inside the young diagram of the staircase partition @xmath39 , we have @xmath40 . @xmath41 [ prop : count ] we proceed by induction on @xmath7 . the proposition is clear for @xmath42 . given a rc - graph @xmath33 , the strand @xmath43 entering the left in row @xmath44 and exiting the top in column @xmath45 travels through only the first and second columns . there is a unique @xmath46 , @xmath47 , such that @xmath43 goes through both the square @xmath48 and @xmath49 . fixing @xmath46 , there must be cross pieces at @xmath50 for all @xmath51 with @xmath52 , @xmath53 ; this is true by definition for @xmath54 , and each strand @xmath55 for @xmath56 crosses @xmath43 in row @xmath57 and never travels to the right of column @xmath58 , so they must each go straight through the topmost @xmath59 places in column @xmath60 , which therefore must all be cross pieces . note each column of these crosses contributes @xmath61 to the weight , for a total of @xmath62 . we must also have cross pieces at @xmath63 for @xmath64 , and these contribute @xmath65 to the weight . now let @xmath66 be the portion of @xmath33 consisting of squares @xmath50 with @xmath67 and @xmath53 ; this is an rc - graph for @xmath68 . also , let @xmath69 be the portion of @xmath33 consisting of the squares @xmath50 , where @xmath70 and either @xmath71 or @xmath72 , ignoring the intervening cross pieces in columns @xmath73 through @xmath74 ; this is an rc - graph for @xmath75 . figure [ fig : rc - split ] illustrates the situation ; the elbow joints have been left out of the diagram to make it smaller and more readable . the numbers on the left denote in this figure the row number ; note that @xmath76 . for convenience , let @xmath77 . note @xmath78 is the exponent of @xmath3 in the term of @xmath79 corresponding to the rc - graph @xmath33 . the rc - graph @xmath69 contributes exactly @xmath80 to @xmath78 . the cross pieces in @xmath66 , however , are each shifted @xmath59 rows down , and there are @xmath81 cross pieces in @xmath66 , so @xmath66 contributes @xmath82 to the weight of @xmath33 . therefore , for an rc - graph @xmath33 for @xmath11 , we have @xmath83 let @xmath84 denote @xmath85 . then we have @xmath86 which proves the proposition . given proposition [ prop : count ] , we would like bijections between rc - graphs for @xmath11 and other sets of objects counted by catalan numbers . one such set is the set of partitions @xmath38 whose young diagrams fit inside the young diagram for the staircase partition @xmath39 , or , equivalently , such that @xmath87 for all @xmath46 . we denote this set @xmath88 . let @xmath33 be an rc - graph for @xmath11 , and let @xmath89 be the set of locations of its elbow joints . we can then associate a partition @xmath90 to @xmath33 by requiring that the parts of its conjugate @xmath91 be , as a multiset , @xmath92 ; in other words , each elbow joint in the @xmath26-th row of @xmath33 should contribute a part of @xmath93 to @xmath91 . the proof will involve generalized inverse chute moves , which are local moves first given in @xcite that , given an rc - graph for some permutation @xmath22 , allows one to generate new rc - graphs for @xmath22 . let @xmath33 be any rc - graph for some permutation @xmath22 . suppose @xmath33 has an elbow joint at @xmath50 , and that the following all hold for some @xmath94 and some @xmath95 : 1 . for each @xmath46 , @xmath96 , @xmath97 is a cross piece , and @xmath98 is an elbow joint . 2 . for each @xmath46 , @xmath99 , @xmath100 is a cross piece , and @xmath101 is an elbow joint . 3 . for each @xmath46 , @xmath102 , @xmath103 is a cross piece . 4 . for each @xmath46 , @xmath104 , @xmath105 is a cross piece . then the diagram @xmath66 obtained from @xmath33 by changing the elbow joint at @xmath50 to a cross piece and the cross piece at @xmath106 to an elbow joint is also an rc - graph for @xmath22 , since in both cases the strand entering this area from the left at @xmath106 exits @xmath50 on the right , the strand entering from the bottom at @xmath106 exits @xmath50 on top , all other strands are unchanged , and the number of cross pieces remains the same . [ prop : elem - bij ] the above described map from @xmath33 to @xmath90 gives a bijection from @xmath107 to @xmath88 . we will show that this map is surjective and then appeal to proposition [ prop : count ] . 6 & 1 & 2 & & & & + 1 & & & & & & + & & & & & & + & & & & & & + & & & & & & + & & & & & & + 2 & & & & & & + begin with the bottom rc - graph for @xmath11 with elbow joints in the first row and cross pieces everywhere else as shown in figure [ fig : rc - bottom ] ; we denote this rc - graph @xmath108 . note that @xmath109 . now we construct the rc - graph @xmath33 with @xmath110 . let @xmath111 denote the @xmath26-th part of @xmath112 , with the parts in decreasing order . now , for each @xmath26 starting from 1 , if @xmath113 , take the rightmost cross piece in row @xmath114 that is not under a cross piece in the top row , and let @xmath57 be the column it is in . there exists such a cross piece since , as @xmath115 , @xmath112 has at most @xmath116 parts of size greater than or equal to @xmath117 , and @xmath108 has @xmath116 cross pieces in row @xmath118 . now let @xmath119 be the leftmost column to the right of @xmath57 such that @xmath120 has an elbow joint . then moving the cross piece at @xmath121 to @xmath122 is a generalized chute move as follows . condition 2 follows from the definition of @xmath119 , and conditions 1 and 3 follow from @xmath108 having only cross pieces in rows 2 through @xmath46 , none of which have been moved . condition 4 holds because the leftmost elbow joint in row @xmath114 to the right of column @xmath57 must either be at the end of the row , or the result of an immediately preceding move from row @xmath114 , and therefore in column @xmath119 . performing these corresponding generalized chute moves for all the parts of @xmath112 in decreasing order constructs an rc - graph for @xmath11 that goes to @xmath38 under the given map . therefore the map is a surjection and hence by our earlier count of rc - graphs a bijection . our bijection has a second description in terms of the edelman - greene correspondence , a variant of the usual rsk correspondence , originally introduced in @xcite and extended to the semi - standard case used here in @xcite . this correspondence associates to each rc - graph a pair @xmath123 of column - strict young tableaux of the same shape in such a way that if two rc - graphs have the same @xmath124-tableau , they must be rc - graphs for the same permutation . the edelman - greene correspondence works as follows . first convert the rc - graph into a sequence @xmath125 of pairs of numbers to put into the tableaux as follows . reading each row of the rc - graph from left to right and starting with the top row , if the @xmath46-th cross piece is encountered at @xmath126 , let @xmath127 and @xmath128 . then we insert the @xmath129 one by one to create the @xmath124 tableau using edelman - greene insertion ; this is identical to rsk insertion except that , when inserting the letter @xmath26 into a row with both an @xmath26 and an @xmath130 , that row remains unchanged and an @xmath130 is bumped into the next row . as in rsk insertion , after all the bumping associated with inserting a single letter @xmath129 is completed , @xmath131 is added to the @xmath132 tableau so that it has the same shape as @xmath124 . when we are finished , @xmath124 will be both row and column - strict , but @xmath132 will only be row - strict . since column - strict tableaux are customary , we transpose both tableaux . @xmath133 @xmath134 edelman and greene showed that the @xmath124 tableau for the long word @xmath135 is always the one in figure [ fig : longwordeg ] . as a trivial corollary , the @xmath124 tableau for any rc - graph for @xmath11 must be the one in figure [ fig : ourwordeg ] . in the case of the long word , they also gave an inverse to their insertion procedure similar to _ evacuation _ , an operation on tableaux originally due to schtzenberger@xcite . take the @xmath132-tableau , and find the box on the outer boundary with the biggest label , breaking ties by preferring the southernmost such box . let @xmath136 be the label of this box , and @xmath137 the row this box is in . now remove the box and do jeu de taquin to fill the space , eventually leaving a `` hole '' in the northwest corner of the tableau . repeat to recover @xmath138 and @xmath139 and so on until the tableau is empty . to adjust this so that it works for @xmath11 rather than the long word , we simply increase each @xmath129 by 1 . the edelman - greene correspondence sends an rc - graph for @xmath11 to a @xmath132-tableau in which the label @xmath26 occurs only in rows @xmath93 and @xmath26 . since @xmath132 is column - strict , the label @xmath26 can not occur in any row strictly below than row @xmath26 . now suppose the label @xmath26 occurs in row @xmath117 for some @xmath140 . then the rightmost entry in row @xmath117 must have label @xmath46 for some @xmath141 . this entry will never be moved by the jeu de taquin during the evacuation procedure , so , eventually , we will have an element @xmath142 . but , since @xmath143 , the @xmath118-st anti - diagonal does not meet the @xmath46-th row , so we could not have started with an rc - graph for @xmath11 in the first place . now we can define a bijection from rc - graphs for @xmath11 to partitions fitting inside @xmath144 by letting the partition associated to an rc - graph be the boxes whose label matches the row number in the @xmath132-tableau corresponding to the rc - graph . this is an injection since the edelman - greene correspondence is injective , and this is therefore a bijection due to proposition [ prop : count ] . the bijection given in section [ sect : elem - bij ] is the same as the one given by the edelman - greene correspondence . for an rc - graph with @xmath26 cross pieces in the @xmath46-th row , the edelman - greene corrspondence produces a tableau with @xmath26 occurrences of the letter @xmath46 , or , equivalently , a partition whose conjugate has exactly @xmath145 parts of size @xmath59 . an rc - graph for any permutation of @xmath13 has @xmath146 places , and therefore @xmath147 elbow joints , in the @xmath46-th row , so the two bijections are the same . since the proof of proposition [ prop : elem - bij ] gives a surjection and the edelman - greene correspondence is known to be an injection , a purely bijective proof omitting the counting lemma is possible . since @xmath11 is its own inverse in @xmath148 , transposing an rc - graph for @xmath11 gives another rc - graph for @xmath11 . it is a natural question to ask what this involution translates to on partitions . unfortunately , the description of this involution on partitions is not immediately evident . however , described on bracketings of a string of length @xmath44 subject to a binary nonassociative operation , it turns out to be simply reversing the string along with the brackets . equivalently , under the obvious bijection to binary trees , this corresponds to flipping the tree along its vertical axis . we describe the map from rc - graphs for @xmath11 to bracketings as follows . for convenience , let the `` letters '' of the string be the numbers from @xmath149 to @xmath44 . now , let @xmath33 be an rc - graph for @xmath11 , and for each @xmath150 ( the set of locations of ( nontrivial ) elbow joints in @xmath33 ) , place a left bracket before the letter @xmath117 and a right bracket after the letter @xmath151 . it is clear such a map is injective and sends the transposition involution on rc - graphs to reversal of order on parenthesizations . what remains to be shown is that this actually gives a proper full bracketing for a binary associative operation . actually , more than this is true ; the pair of brackets aassociated with each elbow joint is in fact a matching pair . we prove this by induction on the generalized inverse chute moves in the proof of proposition [ prop : elem - bij ] . @xmath108 corresponds to the bracketing @xmath152 , and each elbow joint clearly corresponds to a matching pair of brackets . now suppose there is valid generalized inverse chute move moving an elbow joint at @xmath50 to @xmath106 . the elbow joint at @xmath50 corresponds to a matching pair of brackets with the left bracket before the letter @xmath117 and the right bracket after the letter @xmath151 . the second condition for a valid generalized inverse chute move forces the next right bracket to also occur between the letters @xmath151 and @xmath153 ; the first condition forces us to have another left bracket to the left of the letter @xmath117 matching a right bracket after the letter @xmath154 , although this could be an imaginary pair of brackets around the letter @xmath117 ( corresponding to a trivial required elbow joint at @xmath155 ) . the remaining conditions merely state that the original pair of brackets is a matching pair , which for us is true by induction . we can draw the situation as follows : @xmath156 the generalized inverse chute move shifts the parentheses to the following configuration : @xmath157 clearly , the new pair of brackets is a matching pair whenever we start with a proper full bracketing . translated into the language of binary trees , this operation is ( left ) rotation , an operation used in many schemes for keeping binary search trees balanced . a _ ( complete ) flag _ @xmath158 in @xmath159 is a sequence of subspaces @xmath160 such that the subspace @xmath161 has dimension @xmath26 . fixing a basis for @xmath159 , we can represent @xmath158 non - uniquely by an invertible matrix @xmath162 , where the first @xmath26 columns of @xmath119 form a basis for @xmath161 . two matrices @xmath119 and @xmath163 represent the same flag precisely when @xmath164 for some @xmath165 , the group of upper - triangular matrices ; as a result the _ flag variety _ which parameterizes set of all flags is the quotient @xmath10 . note that @xmath166 and its subgroups @xmath167 and @xmath168 , the lower triangular matrices , act on @xmath10 on the left . given our choice of basis , @xmath10 has a distinguished flag @xmath169 called the _ standard flag _ represented by the identity matrix . for each permutation @xmath8 there is a subvariety @xmath170 of @xmath10 known as the _ schubert variety _ ; @xmath170 is the closure of the left @xmath167-orbit of the flag @xmath171 . @xmath172 is the flag whose @xmath26-th vector space @xmath161 is spanned by the vectors @xmath173 , or , alternatively , @xmath172 is the flag represented by @xmath22 as a permutation matrix . given two permutations @xmath174 and @xmath22 , @xmath175 iff @xmath176 in the bruhat order on @xmath148 ; in particular , for @xmath177 the identity permutation , @xmath178 for every @xmath8 . note that @xmath179 consists of a single point , namely the flag @xmath169 . a dense open neighborhood of @xmath169 is given by @xmath180 , the orbit of @xmath169 under the left action of @xmath168 . a detailed treatment of flag and schubert varieties can be found in , for example , @xcite . the _ multiplicity _ of a variety @xmath181 at a point @xmath182 is the degree of the projective tangent cone @xmath183 , considered as a subvariety of the projective tangent space @xmath184 . the multiplicity is one measure of `` how singular '' @xmath181 is at @xmath182 ; in particular it is always 1 if @xmath181 is smooth at @xmath182 . general semi - continuity theorems imply that the multiplicity of @xmath170 at @xmath179 is at least the multiplicity of @xmath170 at any other point . the multiplicity of @xmath170 at @xmath179 can be calculated using local equations for @xmath170 on @xmath180 . in general , these equations will be a specialization of the equations for the matrix schubert varieties given in @xcite . however , if @xmath22 satisfies the condition that , for every @xmath50 with @xmath185 , either @xmath186 or @xmath187 , the local equations are exactly the equations for the matrix schubert varieties . therefore , in this special case , the multiplicity of @xmath170 at @xmath179 is exactly the degree of the matrix schubert variety . let @xmath17 be the permutation @xmath188 . these permutations satisfy the condition stated above , so the multiplicity of @xmath20 at @xmath179 is given by the degree of the matrix schubert variety . knutson and miller @xcite relate rc - graphs to degenerations of matrix schubert varieties so that , in particular , the degree of a matrix schubert variety is given by @xmath189 . note that @xmath190 , and schubert polynomials are unchanged under the inclusion of @xmath191 into @xmath192 fixing the last element , as can easily be seen by adding an anti - diagonal of elbow joints to every rc - graph . therefore , the multiplicity of @xmath20 at @xmath179 is given by @xmath193 . thanks to ezra miller for suggesting this project and getting me started , and bernd sturmfels , brian rothbach , and mark haiman for useful suggestions along the way . thanks also to alexander yong for convincing me to finally write up these results , and again to ezra miller for the latex macros for drawing rc - graphs .	we show that the schubert polynomial @xmath0 specializes to the catalan number @xmath1 when @xmath2 . several proofs of this result as well as a @xmath3-analog are given . an application to the singularities of schubert varieties is given .
cosmic strings are topological defects that may have been created by symmetry breaking phase transitions in the early universe . ( for reviews see@xcite ) . witten@xcite discovered that in many theories , cosmic strings could acquire a condensate of zero - mass particles , and thus could behave as superconducting wires carrying charge or current . the existence of charges and currents on the string can substantially modify its dynamics . if the charge and current are large enough , the string may have a stable static configuration known as a vorton @xcite . for a non - superconducting string , as long as the string thickness ( typically a tiny microphysical quantity ) is small as compared to its radius of curvature ( typically of cosmological size ) , the motion of the string is given by the nambu - goto equations of motion , which are easily solved . with an appropriate choice of parameter ( gauge ) , the string position can be given by a function @xmath0 , where @xmath1 parameterizes the energy on the string , and @xmath2 is the ordinary time coordinate . the general motion of the string can then be written . ] [ eqn : motion ] ( , t ) = where the functions @xmath3 and @xmath4 are arbitrary except that they must obey the constraints \| ( ) \|^2 & = & 1 + in general , a nambu - goto string loop will have two or more points at which @xmath5 . such a point is called a cusp . it has @xmath6 and @xmath7 . thus @xmath8 ; a point moves at the speed of light . for a superconducting string , the equations of motion are much more complicated . however , if we can neglect long - range electromagnetic interactions ( either because the condensate is not coupled to the electromagnetic field , or because such corrections are small ) , the situation is much simpler . if furthermore the current is `` chiral '' ( charge and current equal in magnitude ) , then the equations can be solved exactly@xcite . this situation can occur both in strings which can carry currents in only one direction@xcite , but also in more usual superconducting strings . as the string approaches a cusp , all charge - carriers will be accelerated , and left - moving and right - moving charge carriers can scatter with each other and be ejected from the string . this will lead to a predominance of whichever type of charge - carrier was in larger numbers originally , and thus an approach to a chiral state . in the chiral case , the motion is still given by eq.([eqn : motion ] ) , but now the constraints are [ eqn : newconstraints]\| ( ) \|^2 & = & 1 + @xmath4 exchanged , depending on the direction of current flow ) , where @xmath9 is a measure of the magnitude of the current , to be discussed later . from eqs . ( [ eqn : motion ] ) and ( [ eqn : newconstraints ] ) we can see that such a string can not have a cusp , per se . since @xmath10 and @xmath11 have different magnitudes , they can not cancel exactly , and since @xmath12 , we can not have @xmath13 . however , if @xmath14 , then we can have an event that looks very similar to a nambu - goto cusp at large distances , but has a different structure near the place that the cusp would have occurred . thus we will consider a point with @xmath10 and @xmath11 pointing in opposite directions , even though their magnitudes are different , so @xmath15 . this means that @xmath16 has some small but nonzero magnitude . we will show that this can lead to a self - intersection near the cusp and the emission of a vorton . because the two - dimensional world sheet current is conserved , we can describe it as the curl of an auxiliary scalar field @xmath17 . the world sheet current is then [ eqn : current ] j^a = q 1 ^ab _ , b where @xmath18 is the electric charge multiplied by the integral of the field density across the string core , and renormalized if necessary , and @xmath19 is the induced metric in the world sheet . the physical current is j^(x ) = ddt j^ax^ _ , a^(4)(x - x ( , t ) ) = qd^ab _ , a x^ _ , b^(3 ) ( - ( , t ) ) in the chiral case , we can assume without loss of generality that the charge carriers move in the negative @xmath1 direction , so that @xmath17 is a function only of @xmath20 , and thus @xmath21 . we will define @xmath22 , which gives the constraints of eqs . ( [ eqn : newconstraints ] ) . the current @xmath9 is dimensionless and gives a measure of the ratio of the energy in the current to the energy in the string . with @xmath23 we would have @xmath24 which gives a static vorton solution . we will consider a cusp in a string loop with @xmath25 , and we will define @xmath26 and @xmath27 , and let subscript @xmath28 denotes quantities at the time and place of the cusp . we have @xmath29 , @xmath30 , and so _0 & = & ( _0 + _0)/2 = ( /2)_0 + _ 0 & = & ( _0-_0)/2 = - 1 + 2_0 in@xcite we showed that , for an ordinary string cusp , one can choose a lorentz frame in which @xmath31 and @xmath32 are parallel . in the appendix , we extend this result to chiral superconducting strings , and show that we can work in a frame with @xmath33 and @xmath32 parallel . in that frame , we can expand the position of the string at the time of the cusp . we take the cusp to occur at @xmath34 and its position to be @xmath35 . thus ( ) = _0+_0 ^ 22 + _0 ^ 36 + if @xmath33 and @xmath32 point in opposite directions , which one would expect roughly half the time , then the string will intersect itself at @xmath36 , where _ 1 = this self - intersection will lead to an intercommutation and the emission of a loop containing the original string from @xmath37 to @xmath38 . after the intercommutation , the main string will have kinks where the cusp would have been , and there will be a small loop ( also with kinks ) that has been produced by intercommutation . the effect on the main string is very similar to emission of string due to overlap of the core , as discussed in @xcite . here we will study the small loop that is emitted . first of all , what amount of the string is emitted ? the magnitude of @xmath33 is @xmath39 , while @xmath32 could be expected to be of order @xmath40 , where @xmath41 is the size of the original loop , or the typical feature size in the case of a long string . thus we expect @xmath42 . if we start out , for example , with 1% of the vorton current , then roughly 1% of the original loop will be chopped off . we would like to compute the velocity and rest - frame properties of the emitted loop . the energy - momentum tensor is given by@xcite t^ = dd(^ab + ^ab ) x^_,a x^_,b^4 [ x - x ( , ) ] = d^ab x^ _ , b x^_,a^3 [ -(,t ) ] and thus the total energy - momentum in a region is p^=d^3xt^0 = d^ab x^ _ , bx^0_,a = dx^since @xmath43 , we have @xmath44 , and so the emitted energy is just @xmath45 , which is just the energy in a segment of string of length @xmath46 . we can expand @xmath47 in a taylor series , = _ 0+_0+^2_0 and integrate to find the momentum , = 2_1_0+_1 ^ 33_0 = e ( _ 0+_1 ^ 26_0)= e(-_0 + 2_0+_1 ^ 26_0 ) . the magnitude of @xmath48 is 1 , whereas the magnitude of the second term is @xmath49 , and the third is likewise @xmath49 , since @xmath50 and @xmath51 . furthermore , the second and third terms can not cancel , because @xmath52 is p = e [ 1-o ( ) ] . the rest - frame energy of the loop is @xmath53 , so the length of the loop in its rest frame is _ 1 = ^-1 = ^-1 e o ( ) = 2_1 o ( j ) = o ( l ) since @xmath17 is a scalar , it is unaffected by the boost . the current @xmath9 is proportional to @xmath54 and thus is enhanced by the ratio of @xmath1 in the moving frame to @xmath1 in the rest frame , which is @xmath55 . the new value is @xmath56 , so we have essentially a vorton . what is the boost vector @xmath57 that brings the loop to rest ? it must point in the direction of @xmath58 and should give a lorentz factor @xmath59 . thus the boost should have magnitude @xmath60 , so @xmath61 . then from eq . ( [ eqn : fbdef ] ) , @xmath62 . from eq ( [ eqn : jtransform ] ) the current at the cusp transforms as _ 0 = f_bj_0~1 , in accord with the above , and from eq . ( [ eqn : j2transform ] ) the derivative of the current is _0 = f_b^2j_0+f_bf_b , b j_0 ~j_0j_0 ^ 2 + _ 0j_0 ^ 3 . now @xmath63 , so the second term is of the same order as the first , and @xmath64 . this means that the change in current over the loop is ( _ 1 ) -_0~(1/(j_0l))j_0 ^ 2l = j_0 _ 0~1 and thus the current is essentially constant over the loop . we have shown that cusps ( or , perhaps more properly , `` cusp - like events '' ) in chiral strings come in two forms . if @xmath32 points in the same direction as @xmath33 , in the frame in which these are parallel , then the cusp is merely smoothed out , as shown by the dashed line in fig . [ fig : cusps ] . however , if @xmath32 points in the opposite direction to @xmath33 , then there is a self intersection , as shown by the solid line in fig . [ fig : cusps ] . we thus expect that half of all cusps in strings with chiral currents will have self intersections . when the string does intersect itself , intercommutation emits a loop of string with large lorentz factor which is approximately in a vorton state already , and could be expected to become a stable vorton . what about currents which are not chiral ? we do not have an exact solution in this case , but we expect similar behavior . the effect of the current is still to prevent the string from doubling back on itself , so there will be a finite value of @xmath33 where the point of the cusp would have been . this vector controls the shape very close to this point . further away , however , the effect of @xmath32 becomes more important , and if these vectors point in opposite directions , we might expect a self intersection . if a system of strings can produce vortons , then it is very likely that the vortons will contribute to the matter density of the universe in amounts excess of observation . thus most such theories are ruled out@xcite . for a string to become a vorton , the current must somehow rise to the necessary level to oppose the tension in the string . we have shown here a new mechanism by which strings with small currents can nevertheless produce stable vortons . this mechanism may lead to vorton production in more theories than would otherwise have this problem , and thus restrict further the range of allowable superconducting string theories . it should be noted that , in the case of the string with chiral current and no coupling to the electromagnetic field , the motion will be strictly periodic@xcite . after a single cusp has formed at a particular point in the oscillation of the string loop , and a vorton has been emitted , there will be a kink , rather than a cusp , and no further emission of vortons . however , long strings , currents that are not exactly chiral , and coupling to the electromagnetic field produce motion which is not exactly periodic , and thus perhaps lead to ongoing vorton emission . smoothing of the kink may also produce future cusps , as discussed in@xcite . in this appendix we extend our results@xcite for lorentz boosting of ordinary string cusps to superconducting strings with chiral currents . the motion of a chiral superconducting string is given by eq.([eqn : motion ] ) with the constraints of eqs.([eqn : newconstraints ] ) . following the technique of @xcite , we can treat the current in the same way as the physical position by considering a 4 + 1-dimensional spacetime where the fourth spatial component of @xmath4 is @xmath65 , so that the fourth spatial component of @xmath66 is @xmath9 . since the current travels only in one direction , the new components of @xmath67 and its derivatives are all 0 . in this appendix , we will let @xmath68 and @xmath69 denote the 3-space vectors previously called @xmath67 and @xmath4 , let @xmath67 and @xmath4 denote the new 4-space vectors , and introduce 5-dimensional spacetime vectors @xmath70 and @xmath71 . at the cusp , we can take @xmath72 , which is parallel to @xmath73 , as the `` direction of the cusp '' . the vector @xmath74 is perpendicular to this vector , but @xmath75 will not be perpendicular . instead we have , at the cusp , 0 = = - +jj so _ 3_3 = = jj. we can define a new vector @xmath76 , so that = 0 . now we have two vectors , @xmath74 and @xmath77 , in general independent , which are both perpendicular to @xmath78 . both have vanishing component in the current direction . thus if we can find a frame where @xmath79 is perpendicular to both @xmath74 and @xmath77 , then @xmath80 must be parallel to @xmath78 . as in@xcite , we have two null spacetime vectors , here 5-dimensional , a^ & = & ( 1 , - ) + b ^ & = & ( 1, ) . we define b_2^ & = & b ^_b ^= ( 0, + ) = ( 0,2 ) + a_2^ & = & a ^_a ^= ( 0,- + ) = ( 0 , 2 ) + a_3 ^ & = & a ^_(a ^_a ^)=(0 , -4 ) + b_3 ^ & = & b ^_(b ^_b ^)=(0 , 4 ) and we have inner products at the cusp , g ( a , b ) & = & + g ( a_2 , a ) & = & g ( b_2 , b ) = 0 + g ( a_2 , b ) & = & 0 + g ( b_2 , a ) & = & 2jj + g ( a_3 , a ) & = & -4 \| \|^2 + g ( b_3 , b ) & = & -4 \| \|^2 + g ( b_3 , a ) & = & 4 \| \|^2 + 4jj + g ( a_3 , b ) & = & 4 \| \|^2 . as before , a coordinate transformation yields a ^ & = & ( a ^ , * a * ) + b ^ & = & ( b ^ , * b * ) . and ^&= & ( 1,- ) = a / a ^ + ^&= & ( 1, ) = b / b ^ .[eqn : btransform ] we will consider only boosts which do not involve the fifth direction . the above then says that @xmath72 is boosted and rescaled by the transformation , while @xmath81 is merely rescaled , because there is no boost in that direction . we will let our new coordinate system move with velocity @xmath82 , so that a particle at rest in the original system is moving with velocity @xmath57 with respect to the new coordinates . the vector @xmath57 is a space 4-vector , but has zero current component . the lorentz transformation then gives @xmath83 and @xmath84 , where @xmath85 . we will define f_a & = & 1a ^ = 1(1- ) [ eqn : fadef ] + f_b & = & 1b ^ = 1(1+ ) [ eqn : fbdef ] exactly as before . from eq . ( [ eqn : fbdef ] ) and ( [ eqn : btransform ] ) we see that the current transforms as [ eqn : jtransform ] = f_bj . we find the transformation laws _ 2 & = & f_a ^ 2a_2 + f_af_a , a a + _ 2 & = & f_b ^ 2b_2 + f_bf_b , b b[eqn : b2transform ] + _ 3 & = & f_a ^ 3 a_3 + 3f_a ^ 2f_a , a a_2 + ( f_af_a , a ^2 + f_a^2 f_a , aa ) a + _ 3 & = & f_b ^ 3 b_3 + 3f_b ^ 2f_b , b b_2 + ( f_bf_b , b ^ 2 + f_b^2 f_b , bb ) b with @xmath86 and @xmath87 , @xmath88 and @xmath89 . we also have f_a , a & = & -2f_a ^ 2 + f_b , b & = & -2f_b ^ 2 . applying eq . ( [ eqn : b2transform ] ) to the derivative of the current , we find in particular , [ eqn : j2transform ] = f_b^2j+f_bf_b , b j. & = & f_b ^ 5 + f_b ^ 4f_b , b \|\| ^ 2 = f_b ^ 5 ( -2f_b \|\| ^ 2 ) for cross terms , we have -8 & = & g ( a_3,b_2 ) = f_a^3f_b[f_bg ( a_3 , b_2 ) + f_b , b g ( a_3 , b ) ] + 3f_a^2f_a , a f_b[f_bg ( a_2 , b_2 ) + & & + f_b , b g ( a_2 , b ) ] + ( f_af_a , a ^2 + f_a^2 f_a , aa ) f_b[f_bg ( a , b_2 ) + f_b , b g ( a , b ) ] + & = & f_a^3f_b[-8f_b+4f_b , b \| \|^2 ] + 12f_a^2f_a , a f_b^2 + & & + ( f_af_a , a ^2 + f_a^2 f_a , aa ) f_b[2f_bjj^-1 + f_b , b ] + 8 & = & g ( b_3,a_2 ) = f_b^3f_a[f_ag ( b_3 , a_2 ) + f_a , a g ( b_3 , a ) ] + 3f_b^2f_b , b f_a[f_ag ( b_2 , a_2 ) + & & + f_a , a g ( b_2 , a ) ] + ( f_bf_b , b ^2 + f_b^2 f_b , bb ) f_a[f_ag ( b , a_2 ) + f_a , a g ( b , a ) ] + & = & f_b^3f_a[8f_a + 4 ^ -1 f_a , a \| \|^2 + 4 ^ -1 jj ] + & & + 3f_b^2f_b , b f_a[4f_a + 2f_a , a^-1 jj ] + ( f_bf_b , b ^2 + f_b^2 f_b , bb ) f_a f_a , a now we consider the case that @xmath92 , and so @xmath93 and @xmath94 . we will look for a lorentz transformation where @xmath95 , so the current is unchanged by the boost . this means that @xmath96 , and so f_a = 1 -1- = 1 + 1-= 1+o ( ) . there are three parameters in @xmath57 , but we have fixed one of them by demanding @xmath95 . this leaves two parameters of the lorentz transformation , which we can vary in an effort to solve the two simultaneous equations @xmath97 and @xmath98 . these are the same equations that we solved in@xcite for the ordinary string case , with additional perturbations of order @xmath99 . for generic values of the cusp parameters , the solution will be stable and thus the perturbed equations will still be solvable and the effect of the perturbation will be to change the solution by terms of order @xmath100 . thus in the present case we will be able to find a frame in which @xmath80 and @xmath78 are parallel , and , as in@xcite , the required boost will not be particularly large .	in about half of all near - cusp events in superconducting cosmic strings with chiral currents ( and probably with general currents as well ) , the string intersects itself near the cusp . intercommutation causes the conversion of the string near the cusp into a vorton ( in the chiral case ) with very high lorentz boost . we demonstrate how to analyze the cusp shape in a lorentz frame that makes the motion simple , by the use of a 5-dimensional procedure , and analyze the resulting production of vortons . # 1
this paper aims at calculating the associate norm of the generalized classical lorentz spaces . we have used the characterization of the weighted reverse hardy inequality to calculate the associate norm of the spaces @xmath0 . let @xmath1 be a measurable subset of @xmath2 . we denote by @xmath3 the class of all measurable functions @xmath4 defined on @xmath1 for which @xmath5 @xmath6 and denote by @xmath7 the weak @xmath8 space such that @xmath9 where @xmath10 denotes the distribution function of @xmath11 given by @xmath12 let for @xmath13 and @xmath14 @xmath15 where @xmath16 is the decreasing rearrangement of @xmath4 defined by @xmath17 we denote by @xmath18 be the set of all extended real valued @xmath19-measurable functions on @xmath2 and @xmath20 the set of all non - negative measurable functions on @xmath21 , @xmath22 the set of all non - decreasing functions from @xmath23 . now we recall definitions of lorentz , classical lorentz and generalized classical lorentz spaces . [ cay2 ] the lorentz space @xmath24 , @xmath13 is the collection of all measurable functions @xmath4 on @xmath25 such the quantity @xmath26 is finite . note that @xmath27 ( see , for example , @xcite ) . note that , @xmath28 for @xmath29 . if @xmath30 or @xmath31 , then the functional @xmath32 is a norm . for @xmath33 we have , with continuous embeddings , that @xmath34 the function @xmath35 $ ] is defined as @xmath36 in the case @xmath37 , we give a functional @xmath38 by @xmath39 ( with the usual modification if @xmath40 , @xmath41 ) which is a norm on @xmath42 for @xmath43 , @xmath44 or @xmath31 . if @xmath45 @xmath46 , then @xmath47 about @xmath48 lorentz spaces see @xcite . let @xmath49 and @xmath50 . we denote by @xmath51 the classical lorentz spaces , the spaces of all measurable functions with finite quasinorm @xmath52 therefore the following statement @xmath53 is valid . the spaces @xmath54 were introduced by lorentz in 1951 in @xcite . spaces whose norms involve @xmath55 appeared explicitly for the first time in calderon s paper @xcite . in @xcite sawyer give description of the dual of @xmath56 . lorentz @xcite proved that , for @xmath57 , @xmath58 is a norm if and only if @xmath59 is nonincreasing . the class of weights for which @xmath58 is merely equivalent to a banach norm is however considerably larger . in fact it consists of all those weights w which , for some @xmath60 and all @xmath61 , satisfy @xmath62 ( @xcite , theorem 4 ] , see also @xcite ) , or @xmath63 ( @xcite , theorem 2.3 ) . in @xcite , theorem 1.1 ( see also @xcite , corollary 2.2 , @xcite , p. 6 ) it was observed that the functional @xmath58 ; @xmath64 , does not have to be a quasinorm . it was shown that it is a quasinorm if and only if the function @xmath65 satisfies the @xmath66-condition , i.e. , @xmath67 in @xcite were given necessary and sufficient conditions for @xmath56 to be a linear space . about historical developments classical lorentz spaces see @xcite . let @xmath49 and let @xmath68 . we denote by @xmath69 the generalized lorentz spaces , the spaces of all measurable functions with finite quasinorm @xmath70 let @xmath40 and let @xmath68 . we denote by @xmath71 the generalized classical lorentz spaces , the spaces of all measurable functions with finite quasinorm @xmath72 therefore the following statement @xmath73 @xmath74 is valid . note that , the space @xmath75 we called generalized lebesgue spaces . recall the definition of generalized marcinkiewicz space @xmath76 where the supremum is taken for all measurable subset of @xmath2 . it easy to see that @xmath77 therefore , we have @xmath78 and @xmath79 . [ geyd1 ] note that , the classical lorentz space @xmath51 is not a linear space , therefore the generalized classical lorentz spaces @xmath71 are also not linear spaces in general . it is easy to see that the condition @xmath80 is sufficient for the generalized classical lorentz spaces @xmath71 to be a quasinorm space . also the condition @xmath81 or @xmath82 is sufficient for the generalized classical lorentz spaces @xmath71 to be a norm space . the problems mentioned in the remark [ geyd1 ] are not as easy as the problems in the case of the classical lorentz spaces . at this point it s required to have the characterizations of the embedding between these generalized classical lorentz spaces and the boundedness of the maximal operator in these spaces . we havent got the solution of these problems yet . to solve these problems it will be useful to find the characterizations of the associate spaces of the generalized classical lorentz spaces . in this paper we give the characterizations of the associate spaces of these spaces . note that , in the special case @xmath83 the generalized lorentz space @xmath69 was introduced and investigated in @xcite . let @xmath84 be a set of functions from @xmath85 endowed with a positively homogenous functional @xmath86 defined for every @xmath87 and such that @xmath88 if and only if @xmath89 , we define the associate space @xmath90 of @xmath84 as the set of all functions @xmath87 such that @xmath91 , where @xmath92 in what follows we assume @xmath93 and @xmath94 . [ prp1](@xcite , p. 58 ) let @xmath95 be a rearrangement - invariant function norm over a resonant measure space @xmath96 . then the associate norm @xmath97 is also rearrangement invariant . furthermore , @xmath98 holds . throughout the paper , we write @xmath99 if there exists a positive constant @xmath60 , independent of appropriate quantities such as functions , satisfying @xmath100 . we write @xmath101 when @xmath99 and @xmath102 . let us recall some results from @xcite . let @xmath103 and @xmath59 will denote weights , that is , locally integrable non - negative functions on @xmath21 . we set , once and for all @xmath104 we assume that @xmath105 for every @xmath106 . we then denote @xmath107 when @xmath108 ( hence @xmath109 ) , we will omit the subscript u. furthermore , for given @xmath110 and every @xmath87 , the necessary and sufficient conditions for the inequality @xmath111 were established in @xcite . let @xmath112 be a continuous strictly increasing function on @xmath113 such that @xmath114 and @xmath115 . then we say @xmath112 is admissible . let @xmath112 be an admissible function . we say that a function @xmath116 is @xmath112-quasiconcave if @xmath116 is equivalent to a non - decreasing function on @xmath113 and @xmath117 is equivalent to a non - increasing function on @xmath21 . we say that a @xmath112-quasiconcave function @xmath116 is non - degenerate if @xmath118 the family of non - degenerate @xmath112-quasiconcave functions will be denoted by @xmath119 . let @xmath103 be weights above and let @xmath120 defined by @xmath121 then , @xmath122 let @xmath120 be an admissible function and let @xmath123 be a non - negative borel measure on @xmath113 . we say that the function @xmath116 defined as @xmath124 fundamental function of the measure @xmath123 with respect to @xmath120 . we will also say that the function @xmath123 is a representation measure of @xmath116 with respect to @xmath120 . we say that @xmath123 is non - degenerate if the following conditions are satisfied for every @xmath106 : @xmath125}\frac{d\nu(s)}{\sigma(s)}= \int_{[1,\infty)}d\nu(s)=\infty.\ ] ] let @xmath120 be an admissible function and let @xmath123 be a non - negative non - degenerate borel measure on @xmath113 . let @xmath116 be the fundamental function of @xmath123 with respect to @xmath120 . then @xmath126 and also @xmath127}d\nu(s)+\sigma(t)\int_{[t,\infty]}\frac{d\nu(s)}{\sigma(s ) } , t\in(0,\infty).\ ] ] [ mak1 ] @xcite let @xmath128 and let @xmath129 be weights . assume that @xmath130 is such that @xmath131 is admissible . let @xmath120 defined by , be non - degenerate with respect to @xmath131 . let @xmath123 be the representation measure of @xmath132 with respect to @xmath133 . \(i ) if @xmath134 , then holds for all @xmath4 if and only if @xmath135 moreover , the optimal constant @xmath60 in satisfies @xmath136 . \(ii ) if @xmath137 , then holds for all @xmath4 if and only if @xmath138 where @xmath139 moreover , the optimal constant in satisfies @xmath140 . the associated spaces of classical lorentz spaces @xmath141 was calculated in @xcite . [ abdfk ] let @xmath142 , @xmath143 , @xmath149 and @xmath150 . then the associate spaces of @xmath0 are described as follows : + ( i ) if @xmath144 , then @xmath151 where @xmath123 is the representation measure of @xmath152 with respect to @xmath153 . + ( ii ) if @xmath147 , then @xmath154 where @xmath123 is the representation measure of @xmath152 with respect to @xmath155 . from proposition [ prp1 ] we have @xmath156 if we take @xmath157 , then we can write @xmath158^{1/p}.\end{aligned}\ ] ] if we define the function @xmath159 , then we get @xmath160 therefore @xmath161^{1/p } & \approx\left[\sup\limits_{h\geq 0}\frac{\left(\int_{0}^{\infty}f^{}(t)h^{}(t)^{\frac{1}{p}}dt\right)^{p } } { \sup\limits_{r>0}\varphi(r)^{p}\psi_p^p(r)h_{u}^{**}(r)}\right]^{1/p}.\end{aligned}\ ] ] \i ) @xmath182 @xmath183 where @xmath123 is the representation measure of @xmath152 with respect to @xmath184 . + ( ii ) if @xmath185 , then @xmath186 where @xmath123 is the representation measure of @xmath152 with respect to @xmath187 . using a simple observation function , it is obvious that @xmath4 is decreasing if and only if , @xmath188 is decreasing for all @xmath189 . we see that the embedding holds if and only if the following embedding holds @xmath190 one can get the required result by using theorem [ abdfk ] . this paper has been written during the visit of a. gogatishvili to ankara university in ankara and to ahi - evran university in kirsehir . a special gratitude he gives to these universities for their warm hospitality .	in this paper we have calculated the associate norms of the @xmath0 generalized classical lorentz spaces .
in @xcite , hariya obtained an integration by parts formula on a subset of the pinned path space on @xmath2 , which is a partial generalization of the work by zambotti @xcite . to state it more precisely , let @xmath3 be a bounded domain in @xmath2 . we assume that the boundary of @xmath3 is sufficiently smooth . take @xmath4 and define the path spaces as follows : @xmath5\to{{\mathbb r}}^d)\mid w(0)=a,\ w(1)=b\},\\ w_{a , b}^{\omega}&=\{w\in c([0,1]\to { \omega})\mid w(0)=a,\ w(1)=b\},\\ h_0&=\left\{h\in c([0,1]\to{{\mathbb r}}^d)\left\|\ , \begin{array}{l}h(0)=h(1)=0,\ \text{$h$ is absolutely continuous}\\ \text{and } \int_0 ^ 1 \|\dot h(s)\|_{{{\mathbb r}}^d}^2\,ds<\infty\end{array}\right.\right\}.\end{aligned}\ ] ] we regard @xmath6 as a subset of @xmath7 . the topological boundary @xmath8 of @xmath6 with respect to the uniform topology is given by @xmath9 $ and}\\ w(s)\in\partial { \omega}\text { for some $ s\in(0,1)$ } \end{array } \right.\right\},\ ] ] where @xmath10 and @xmath11 denote the closure and the boundary of @xmath3 in @xmath2 , respectively . we define a subset @xmath12 of @xmath8 by @xmath13 let @xmath14 be the pinned wiener measure on @xmath7 . for a smooth cylindrical function @xmath15 on @xmath7 and @xmath16 , hariya @xcite proved the identity @xmath17 where @xmath18 is the partial derivative along the direction of @xmath19 , @xmath20 denotes the wiener integral @xmath21 , and ( bc ) is the `` boundary contribution , '' which is expressed as an integral over @xmath8 . the explicit expression of ( bc ) is provided in @xcite . in this study , we provide only the following remarks on ( bc ) . 1 . the mass of the measure on @xmath8 appearing in the integral representation of ( bc ) concentrates on @xmath12 . 2 . the integrand in ( bc ) contains the normal derivatives of the heat kernel density on @xmath3 at @xmath11 with the dirichlet boundary condition . the integration by parts formula implies that the indicator function @xmath22 of @xmath6 is a bv function , and we can construct the reflecting ornstein - uhlenbeck process on @xmath6 with the skorokhod representation ( cf . section 2 below ) . on the other hand , property ( b ) above imposes on the strong regularity of @xmath3 since we can not expect the normal derivatives of the heat kernel density to exist at the boundary if @xmath11 is not very smooth . if we are only interested in the probabilistic aspect , it is sufficient to prove that @xmath22 is a bv function ; in other words , even if we do not know the explicit expression of ( bc ) , only proving the validity of the integration by parts is sufficient . this is expected to be done under a milder assumption on @xmath3 since such a claim can be proved only by a series of inequalities and not by equalities . this is the objective of this paper . in this paper , we introduce the concept of the _ uniform exterior ball condition _ for @xmath3 , which allows some singularity at @xmath23 , and prove that @xmath22 is a bv function under such a condition . based on this , we can construct the reflecting ornstein - uhlenbeck process on the closure of @xmath6 and prove its skorokhod representation . further , we prove that the mass of the measure on @xmath8 appearing in the skorokhod representation concentrates on @xmath24 , which is consistent with property ( a ) above . the proof is based on the quantitative estimates of brownian motion on @xmath2 , and the method is different from that used in @xcite . we expect that our method is sufficiently flexible to discuss more general situations . this paper is organized as follows . in section 2 , we provide a framework and state the main theorem . some key estimates for the brownian motion on @xmath2 are proved in section 3 . these estimates are obtained by reducing them to a few detailed estimates of a one - dimensional brownian motion with a constant drift . the main theorem is proved in section 4 . in the last section , we provide a few remarks . first , we recall the concept of the bv functions on the wiener space , according to @xcite . let @xmath25 be an abstract wiener space , that is , @xmath26 is a separable banach space , @xmath27 is a separable hilbert space densely and continuously embedded in @xmath26 , and @xmath28 is a gaussian measure on @xmath26 that satisfies the condition @xmath29 here , @xmath30 denotes the topological dual and we use natural inclusions and an identification @xmath31 . when @xmath32 is a separable hilbert space , @xmath33 denotes the @xmath34-space on @xmath26 with respect to @xmath28 which consists of @xmath32-valued functions . when @xmath35 , we omit @xmath32 from the notation . let @xmath36 be the set of all bounded continuous functions @xmath37 on @xmath38 such that all the first - order partial derivatives of @xmath37 are bounded and continuous . define @xmath39 for @xmath40 , an @xmath27-valued function @xmath41 on @xmath26 is given by the following identity : @xmath42 let @xmath43 be a ( formal ) adjoint operator of @xmath44 , which is defined by the following relation : @xmath45 we define the set of bv functions on @xmath26 as @xmath46 we shall now revisit several properties of bv functions on @xmath26 . [ th:1 ] for @xmath47 , the following are equivalent conditions . 1 . @xmath48 . 2 . there exists a sequence @xmath49 in @xmath50 such that @xmath51 converges to @xmath52 in @xmath53 and @xmath54 is bounded in @xmath55 . ( integration by parts formula ) there exist , a finite borel measure @xmath56 and an @xmath27-valued function @xmath57 on @xmath26 such that @xmath58 @xmath56-a.e . and @xmath59 [ th:2 ] let @xmath48 and assume @xmath60 @xmath28-a.e . let @xmath61 be the support of the measure @xmath62 . define a bilinear form on @xmath63 by @xmath64 assume that @xmath65 is closable on @xmath63 . then , its closure @xmath66 is a quasi - regular , local , and conservative dirichlet form on @xmath63 , and the following skorokhod representation holds : @xmath67 here , @xmath68 is a diffusion process on @xmath61 associated with @xmath66 , @xmath69 is an @xmath26-valued brownian motion starting at @xmath70 , @xmath71 is an additive functional in revuz correspondence with @xmath56 , and @xmath56 and @xmath57 are provided in theorem @xmath72 @xmath73 . note that @xmath56 above is smooth with respect to the @xmath66 from ( * ? ? ? * theorem 3.9 ) , which justifies the consideration of the revuz correspondence of @xmath56 . when @xmath52 is an indicator function , we term @xmath74 a reflecting ornstein - uhlenbeck process on @xmath61 . in such a case , the measure @xmath56 can be regarded as a surface measure of @xmath61 . [ th:3 ] under the conditions described in theorem @xmath75 , if moreover @xmath52 is an indicator function of a set @xmath76 , then the mass of @xmath56 concentrates on the topological boundary of @xmath76 . we remark that the original assertion of theorem 3.15 in @xcite provides more detailed information on the support of @xmath56 . for @xmath77 , @xmath78 , we define @xmath79 fix @xmath80 , and let @xmath7 and @xmath81 as defined in section 1 . the pinned wiener measure @xmath14 on @xmath7 is a borel probability measure such that for @xmath82 and borel sets @xmath83 of @xmath2 , @xmath84\\ & = p_1(a , b)^{-1}\int\!\cdots\!\int_{a_1\times \dots\times a_n}\prod_{i=1}^{n+1}p_{t_i - t_{i-1}}(x_{i-1},x_i)\,dx_1\cdots dx_n,\end{aligned}\ ] ] where @xmath85 and @xmath86 . then , @xmath87 is an abstract wiener space . when @xmath88 , @xmath7 is not a linear space . however , @xmath7 is isomorphic to @xmath89 as an affine space according to the shift map @xmath90 where @xmath91 , @xmath92 $ ] , and @xmath93 is isomorphic to @xmath94 as a measure space according to the map @xmath95 . therefore , by pushing everything forward to @xmath94 , we can define the concepts of @xmath96 , @xmath44 , the bv space @xmath97 etc . , on @xmath98 . furthermore , theorems [ th:1 ] , [ th:2 ] , and [ th:3 ] are valid on this space by appropriate modification . let @xmath3 be a domain of @xmath2 . we do not assume that @xmath3 is bounded , but assume @xmath99 . for @xmath100 and @xmath101 , @xmath102 denotes the closed ball in @xmath2 with center @xmath103 and radius @xmath104 . [ def : uebc ] we state that @xmath3 satisfies the _ uniform exterior ball condition _ if there exists @xmath105 such that for every @xmath106 , there exists @xmath107 satisfying @xmath108 . for example , bounded domains with boundaries in the @xmath109-class and convex domains satisfy the uniform exterior ball conditions . it may be said that this condition allows outward cusps , but not inward cusps . we consider @xmath6 , @xmath8 , and @xmath24 as defined in section 1 . let @xmath110 . the main theorem in this paper is as follows . [ th : main ] assume that @xmath3 satisfies the uniform exterior ball condition . then , @xmath111 . further , the bilinear form @xmath112 on @xmath113 defined by @xmath114 is closable , and its closure @xmath115 is a quasi - regular , local , and conservative dirichlet form . moreover , when @xmath116 denotes the diffusion process on @xmath117 associated with @xmath115 , @xmath118 satisfies the skorokhod representation with @xmath119 and @xmath120 . furthermore , @xmath121 has a null capacity that is associated with @xmath115 . in particular , the mass of the measure @xmath56 that corresponds to @xmath52 in theorem @xmath72 @xmath73 concentrates on @xmath122 . subsequently , @xmath123 denotes an insignificant positive constant and a domain @xmath3 in @xmath2 is assumed to satisfy the uniform exterior ball condition . we define a lipschitz function @xmath124 on @xmath2 by @xmath125 for @xmath126 , set @xmath127 . note that @xmath128 and @xmath129 . let @xmath130 . let @xmath131 be the probability measures on @xmath132 such that the coordinate process @xmath133 is a @xmath134-dimensional brownian motion starting at @xmath103 under @xmath135 for each @xmath136 . for @xmath137 , let @xmath138 be a @xmath57-field generated by @xmath139,\ d \text { is a borel set of } { { \mathbb r}}^d\}$ ] . then , @xmath140 is a minimal filtration to which @xmath141 is adapted on the canonical measurable space @xmath142 . for an @xmath140-stopping time @xmath143 , define @xmath144 . we denote the integral with respect to @xmath135 by @xmath145 . the shift operator @xmath146 is defined by @xmath147 , @xmath137 . [ lem : ito ] let @xmath148 . choose @xmath149 and @xmath150 such that @xmath151 and @xmath152 , where @xmath153 is provided in definition @xmath154 . let @xmath155 and @xmath156 for @xmath157 . then , for each @xmath158 , @xmath159\}\subset\{r_t\le q(x)+{\delta}+kt+s_t \text { for all } t\in[0,u]\}\ ] ] up to a @xmath135-null set . here , @xmath160 is a one - dimensional brownian motion under @xmath135 starting at @xmath70 that is defined by @xmath161 define an @xmath140-stopping time @xmath57 by @xmath162 . note that @xmath163 @xmath135-a.e . by virtue of it s formula , @xmath164 therefore , the assertion holds . [ prop:1 ] there exists @xmath165 such that for every @xmath158 and @xmath148 , @xmath166}q({\omega}_t)\ge0\right]\le c_1(1+u^{-1/2})q(x).\ ] ] we retain the notations in lemma [ lem : ito ] , from which @xmath167\right ] \le \hat p_x\left[q(x)+{\delta}+k t+s_t\ge{\delta}\text { for all } t\in[0,u]\right].\ ] ] let @xmath168 and define @xmath169 . the law of @xmath170 under @xmath135 is given by @xmath171=1_{(0,\infty)}(t)\frac{r}{\sqrt{2\pi t^3}}\exp\left(-\frac{(r+kt)^2}{2t}\right)\,dt+(1-\exp(-2kr)){\delta}_{\infty}(dt),\ ] ] where @xmath172 is a delta measure at @xmath173 . ( see , e.g. , @xcite . ) then , we have @xmath174}q({\omega}_t)\ge0\right ] & \le \hat p_x\left [ r_t\ge{\delta}\text { for all } t\in[0,u]\right ] \le \hat p_x[\eta > u]\\ & = \int_u^\infty \frac{r}{\sqrt{2\pi t^3}}\exp\left(-\frac{(r+kt)^2}{2t}\right)\,dt + 1-\exp(-2kr)\\ & \le \int_u^\infty \frac{r}{\sqrt{2\pi t^3}}\,dt + 2kr = \sqrt{\frac2\pi}\frac{r}{\sqrt u}+2kr.\end{aligned}\ ] ] letting @xmath175 , we obtain the desired inequality . for @xmath101 , define an @xmath140-stopping time @xmath176 by @xmath177 . let @xmath178 be the law of @xmath176 under @xmath135 . [ lem : differentiable ] @xmath179)$ ] is differentiable in @xmath180 on @xmath181 and there exists a constant @xmath182 such that @xmath183)\le c_2 t^{-1 } $ ] . the constant @xmath184 is taken independently of @xmath103 , @xmath104 and @xmath180 . it is sufficient to consider the case that @xmath185 . let @xmath186 be the transition density of the brownian motion of @xmath187 killed at @xmath188 . then , @xmath189)&=\hat p_x[\tau_r\le t ] = 1-\int_{{\omega}_r}p^r_t(x , z)\,dz\\ & = 1-\iint_{{\omega}_r\times{\omega}_r}p^r_s(x , y)p^r_{t - s}(y , z)\,dy\,dz \end{aligned}\ ] ] for @xmath190 . from ( * ? ? ? * theorem 6.17 ) , @xmath191 is differentiable in @xmath180 on @xmath192 for a.e . @xmath193 and the following estimate holds : @xmath194 ) & \le \iint_{{\omega}_r\times{\omega}_r } p^r_s(x , y)\left\|\frac{\partial}{\partial t}p^r_{t - s}(y , z)\right\|\,dy\,dz\\ & \le c_3\iint_{{\omega}_r\times{\omega}_r } p^r_s(x , y)(t - s)^{-d/2 - 1}\exp\left(-\frac{c_4\|y - z\|_{{{\mathbb r}}^d}^2}{t - s}\right)\,dy\,dz\\ & \le c_5 ( t - s)^{-1}\int_{{\omega}_r } p^r_s(x , y)\,dy \le c_5 ( t - s)^{-1 } , \end{aligned}\ ] ] where @xmath195 , @xmath196 and @xmath197 are taken independently of @xmath103 , @xmath104 , @xmath198 and @xmath180 . by letting @xmath199 , we complete the proof . [ prop : path ] there exists @xmath200 such that for all @xmath158 , @xmath101 , and @xmath201 , @xmath202}q({\omega}_t)\le r\right]\le c_6(1+u^{-1/2})r.\ ] ] first , let @xmath203 . from proposition [ prop:1 ] and the fact that @xmath204 , @xmath202}q({\omega}_t)\le r\right ] \le c_1(1+u^{-1/2})q(x ) \le c_1(1+u^{-1/2})r.\ ] ] next , let @xmath185 . then , @xmath205}q({\omega}_t)\le r\right ] = \hat p_x\left[\tau_r\le u,\ 0\le\inf_{t\in[\tau_r , u]}q({\omega}_t)\right]\\ & = \hat p_x\left[\tau_r\le u,\ 0\le\inf_{t\in[0,u-\tau_r]}q((\theta_{\tau_r}{\omega})_t)\right]\\ & = \sum_{k=1}^\infty \hat p_x\left[2^{-k}u < u-\tau_r\le 2^{-k+1}u,\ 0\le\inf_{t\in[0,u-\tau_r]}q((\theta_{\tau_r}{\omega})_t)\right]\\ & \le \sum_{k=1}^\infty \hat p_x\left[2^{-k}u < u-\tau_r\le 2^{-k+1}u,\ 0\le\inf_{t\in[0,2^{-k}u]}q((\theta_{\tau_r}{\omega})_t)\right].\end{aligned}\]]here , we used @xmath206=0 $ ] in the third line , which follows from lemma [ lem : differentiable ] . from the strong markov property and proposition [ prop:1 ] , @xmath207}q((\theta_{\tau_r}{\omega})_t)\right\|\hat{{\mathcal f}}_{\tau_r}\right]\\ & = 1_{\{2^{-k}u < u-\tau_r\le 2^{-k+1}u\}}\cdot \hat p_{{\omega}_{\tau_r}}\!\left[0\le\inf_{t\in[0,2^{-k}u]}q({\omega}_t)\right]\\ & \le c_1 1_{\{2^{-k}u < u-\tau_r\le 2^{-k+1}u\}}\cdot ( 1+(2^{-k}u)^{-1/2})r\\ & \le c_1 1_{\{2^{-k}u < u-\tau_r\le 2^{-k+1}u\}}\cdot ( 1+((u-\tau_r)/2)^{-1/2})r.\end{aligned}\ ] ] therefore , @xmath205}q({\omega}_t)\le r\right]\\ & \le c_1 r\hat e_x[1+((u-\tau_r)/2)^{-1/2};\ \tau_r\le u]\\ & \le c_1 r\int_0^u \left(1+\left(\frac{u - s}2\right)^{-1/2}\right)\,\hat p_x^r(ds)\\ & \le c_1 r(1+(u/4)^{-1/2})\hat p_x[\tau_r\le u/2 ] + c_1 r\int_{u/2}^u \left(1+\left(\frac{u - s}2\right)^{-1/2}\right)c_2s^{-1}\,ds\\ & \le c_1 r(1+(u/4)^{-1/2})+\frac{2c_1c_2 r}{u}\int_{u/2}^u\left(1+\left(\frac{u - s}2\right)^{-1/2}\right)\,ds\\ & \le c_6(1+u^{-1/2})r.\end{aligned}\]]here , we used lemma [ lem : differentiable ] in the third line . this completes the proof . let @xmath208 be a probability measure on @xmath132 such that @xmath209}$ ] is a pinned brownian motion under @xmath208 with @xmath210 and @xmath211 . the following lemma is proved by the definition of @xmath208 and the monotone class theorem . [ lem : a ] for @xmath212 , @xmath213 , and a borel set @xmath214 of @xmath2 , @xmath215 \le \hat p_{a}[a\cap \{{\omega}_t\in d\ } ] \cdot \sup_{y\in d}\frac{p_{1-t}(y , b)}{p_1(a , b)}.\ ] ] [ lem : b ] let @xmath143 be an @xmath140-stopping time and @xmath216 . let @xmath214 be an open set of @xmath2 . then , @xmath217 \le \hat p_a[\{\tau<1\}\cap a\cap\{{\omega}_\tau\in \overline{d}\ } ] \cdot \sup_{t\in(0,1],\,y\in d}\frac{p_t(y , b)}{p_1(a , b)}.\ ] ] here , @xmath218 is a closure of @xmath214 in @xmath2 . consider a sequence of @xmath140-stopping times @xmath219 such that each @xmath220 takes only finite values of @xmath221 and @xmath222 . here , @xmath223 is an index set consisting of finite elements . then , @xmath224 for @xmath225 , @xmath226 . therefore , from lemma [ lem : a ] , @xmath227\\ & = \sum_{k\in \lambda_n,\,t_n^{(k)}<1}\hat p_{a , b}[\{\tau_n = t_n^{(k)}\}\cap a\cap\{{\omega}_{\tau_n}\in d\}]\\ & \le\sum_{k\in \lambda_n,\,t_n^{(k)}<1}\hat p_{a}[\{\tau_n = t_n^{(k)}\}\cap a\cap\{{\omega}_{\tau_n}\in d\ } ] \cdot\sup_{y\in d}\frac{p_{1-t_n^{(k)}}(y , b)}{p_1(a , b)}\\ & \le\sum_{k\in \lambda_n,\,t_n^{(k)}<1}\hat p_{a}[\{\tau_n = t_n^{(k)}\}\cap a\cap\{{\omega}_{\tau_n}\in d\ } ] \cdot\sup_{t\in(0,1],\,y\in d}\frac{p_t(y , b)}{p_1(a , b)}\\ & = \hat p_a[\{\tau_n<1\}\cap a\cap\{{\omega}_{\tau_n}\in d\ } ] \cdot\sup_{t\in(0,1],\,y\in d}\frac{p_t(y , b)}{p_1(a , b)}.\end{aligned}\ ] ] by letting @xmath228 , we complete the proof from fatou s lemma . denote the borel @xmath57-field on @xmath229 by @xmath230 . [ lem : c ] let @xmath143 be an @xmath140-stopping time such that @xmath231 and @xmath232 an element in @xmath233 . assume that @xmath234 then , for a borel set @xmath214 of @xmath2 , @xmath235\\ & \le \sup_{x\in d}\hat e_x\left[\sup_{s\in(0,1/2]}1_{\gamma}(s,{\omega})\cdot s^{-d/2}\right]\exp(\|a - b\|_{{{\mathbb r}}^d}^2/2).\end{aligned}\ ] ] let @xmath236 . then , @xmath235\\ & = \sum_{k=0}^\infty\hat p_{a , b}[\{c^{k+1}<1-\tau\le c^k\}\cap\{{\omega}_\tau\in d\}\cap \{((1-\tau)/2,\theta_\tau{\omega})\in{\gamma}\}].\end{aligned}\ ] ] since @xmath237 implies that @xmath238 and @xmath239 , by combining the assumption , the set in @xmath240 $ ] belongs to @xmath241 . from lemma [ lem : a ] and the strong markov property , the above equation is dominated by @xmath242\\ & \phantom{\sum_{k=0}^\infty}\cdot p_1(a , b)^{-1}(\pi c^{k+1})^{-d/2}\\ & \le \hat e_a[1_{\{\tau<1\}\cap\{{\omega}_\tau\in d\}\cap \{((1-\tau)/2,\theta_\tau{\omega})\in{\gamma}\}}\cdot ( \pi(1-\tau)c)^{-d/2}]\cdot p_1(a , b)^{-1}\\ & \le \hat e_a\!\left[1_{\{\tau<1\}\cap\{{\omega}_\tau\in d\ } } \sup_{s\in(0,1/2]}\left(1_{\gamma}(s,\theta_\tau{\omega})\cdot ( 2\pi sc)^{-d/2}\right)\right]\cdot p_1(a , b)^{-1}\\ & = \hat e_a\!\left[1_{\{\tau<1\}\cap\{{\omega}_\tau\in d\ } } \hat e_{{\omega}_\tau}\!\left[\sup_{s\in(0,1/2]}1_{\gamma}(s,{\omega})\cdot ( 2\pi sc)^{-d/2}\right]\right]\cdot p_1(a , b)^{-1}\\ & \le \sup_{x\in d } \hat e_x\!\left[\sup_{s\in(0,1/2]}1_{\gamma}(s,{\omega})\cdot s^{-d/2}\right]\cdot c^{-d/2}\exp(\|a - b\|_{{{\mathbb r}}^d}^2/2).\end{aligned}\ ] ] by letting @xmath243 , we reach the conclusion . [ prop:2 ] there exists @xmath244 such that for every @xmath101 , @xmath245}q(w(t))\le r\right]\right . \le c_7 r.\ ] ] let @xmath246 . it is sufficient to prove that there exists @xmath244 such that holds for all @xmath247 . choose @xmath247 and let @xmath248 and @xmath249 . then , @xmath250}q(w(t))\le r\right.\right]\\ & \le \hat p_{a , b}[\tau_r<1\text { and } { \omega}_t\in\overline{{\omega}}\setminus v \text { for all } t\in[\tau_r,(\tau_r+1)/2]]\\ & \phantom{{}\le{}}{}+ \hat p_{a , b}\left[\begin{array}{l}\tau_r<1,\ { \omega}_t\in v \text { for some } t\in[\tau_r,(\tau_r+1)/2],\\ \text{and } { \omega}_t\in\overline{{\omega } } \text { for all } t\in[\tau_r,(\tau_r+1)/2 ] \end{array } \right]\\ & = : i_1+i_2.\end{aligned}\ ] ] for @xmath251 , lemma [ lem : b ] with @xmath252 implies @xmath253,\ \text{and } { \omega}_\tau\in{{\mathbb r}}^d\setminus v']\\ & \le\hat p_a[\tau_r<1,\ { \omega}_t\in\overline{{\omega}}\setminus v \text { for all } t\in[\tau_r,\tau]]\cdot \sup_{t\in(0,1],\ y\in{{\mathbb r}}^d\setminus v'}\frac{p_t(y , b)}{p_1(a , b)}.\end{aligned}\ ] ] now , @xmath254,\ y\in{{\mathbb r}}^d\setminus v'}\frac{p_t(y , b)}{p_1(a , b ) } \le p_1(a , b)^{-1}\sup_{t\in(0,1]}(2\pi t)^{-d/2}\exp\left(-\frac{(\alpha/2)^2}{2t}\right ) \le c_8\ ] ] and @xmath255\\ & \le \hat p_a[\tau_r<1,\ ( \theta_{\tau_r } { \omega})_t\in\overline{{\omega } } \text { for all } t\in[0,(1-\tau_r)/2]]\\ & \le \sum_{k=1}^\infty \hat p_a[2^{-k}<1-\tau_r\le 2^{-k+1},\ ( \theta_{\tau_r } { \omega})_t\in\overline{{\omega } } \text { for all } t\in[0,2^{-k-1 } ] ] . \end{aligned}\ ] ] since @xmath256\mid \hat{{\mathcal f}}_{\tau_r}]\\ & = 1_{\{2^{-k}<1-\tau_r\le 2^{-k+1}\}}\cdot\hat p_{{\omega}_{\tau_r}}[{\omega}_t\in\overline{{\omega } } \text { for all } t\in[0,2^{-k-1}]]\\ & \le 1_{\{2^{-k}<1-\tau_r\le 2^{-k+1}\}}\cdot c_1(1 + 2^{(k+1)/2})r\\ & \le 1_{\{2^{-k}<1-\tau_r\le 2^{-k+1}\}}\cdot c_1(1 + 2(1-\tau_r)^{-1/2})r \end{aligned}\ ] ] from the strong markov property and proposition [ prop:1 ] , @xmath257\\ & \le c_8c_1 r(1 + 2\sqrt2)\hat p_a[\tau_r\le 1/2 ] + c_8c_1 r \int_{1/2}^1 ( 1 + 2(1-s)^{-1/2})\cdot c_2 s^{-1}\,ds\\ & \le c_9 r , \end{aligned}\ ] ] by virtue of lemma [ lem : differentiable ] . we will estimate a value for @xmath258 . from lemma [ lem : c ] with @xmath259 , @xmath260 , and @xmath261\text { and}\\ { \omega}_t\in\overline{{\omega } } \text { for all } t\in[0,s ] \end{array}\right.\right\},\ ] ] we obtain @xmath262}1_{\gamma}(s,{\omega})\cdot s^{-d/2}\right].\ ] ] by letting @xmath263}1_{\gamma}(s,{\omega})\cdot s^{-d/2}$ ] , we have @xmath264 = c_{10 } \sup_{x\in\partial{\omega}_r}\int_0^\infty \hat p_x[f > u]\,du.\ ] ] let @xmath265 , and define @xmath266 , @xmath267 , @xmath268 , @xmath269 , and @xmath160 as in lemma [ lem : ito ] . we have @xmath270 and @xmath271 . define the stopping times with respect to the canonical augmentation of @xmath140 by @xmath131 as follows : @xmath272 since @xmath273 implies @xmath274 , @xmath275\\ & \le\hat p_x[(s,{\omega})\in{\gamma}\text { for some } s < u^{-2/d}]\\ & \le \hat p_x[\rho < u^{-2/d}\text { and } r_\rho={\delta}+\alpha]\\ & \le \hat p_x[\rho ' < u^{-2/d}\text { and } r+{\delta}+k\rho'+s_{\rho'}={\delta}+\alpha ] \quad\text{(from lemma~\ref{lem : ito})}\\ & \le \hat e_x[\exp(1-u^{2/d}\rho');\ , r+k\rho'+s_{\rho'}=\alpha]\\ & = e^{1+k(\alpha - r)}{\sinh\left(r\sqrt{2u^{2/d}+k^2}\right)}\left/{\sinh\left(\alpha\sqrt{2u^{2/d}+k^2}\right)}\right .. \quad \text{(cf.\ \cite[p.~309]{bs } ) } \end{aligned}\ ] ] since @xmath276 for @xmath277 , the above term is dominated by @xmath278 substituting this estimate into , we obtain @xmath279 . this completes the proof . in this section , we prove theorem [ th : main ] . we retain the notations in the previous sections . we will utilize the following theorem . [ th : es ] let @xmath15 be a measurable function on @xmath7 and @xmath81-lipschitz ; in other words , there exists @xmath280 such that @xmath281 then , if @xmath282 , @xmath15 belongs to @xmath283 . here , @xmath283 is a first order @xmath284-sobolev space on @xmath7 that is defined as the completion of @xmath96 with respect to the norm @xmath285 . moreover , @xmath286 @xmath14-a.e . from proposition [ prop:2 ] , for any @xmath101 , @xmath287 = \mu_{a , b}\left [ \inf_{s\in[0,1 ] } q(w(s))=0\right ] \le c_7 r.\ ] ] therefore , @xmath288=0 $ ] . by combining this with the remark in @xcite , the bilinear form @xmath112 is closable on @xmath113 , and the closure @xmath115 is a quasi - regular , local , and conservative dirichlet form . in particular , we obtain the diffusion process @xmath289 on @xmath117 associated with @xmath115 . next , we prove that @xmath290 . define @xmath291}q(w(t))$ ] for @xmath292 . for @xmath293 , we define @xmath294 for @xmath295 and @xmath296 for @xmath292 . then , since @xmath297 , we obtain @xmath298 @xmath14-a.e . and in @xmath299 . therefore , from theorem [ th:1 ] , it is sufficient to prove @xmath300 . we note that @xmath301 is a lipschitz function on @xmath2 with lipschitz constant 1 ; thus , we obtain the following estimate for @xmath292 and @xmath16 @xmath302}q(w(t)+h(t))-\inf_{t\in[0,1]}q(w(t))\right\|\\ & \le \sup_{t\in[0,1]}\|q(w(t)+h(t))-q(w(t))\|\\ & \le \sup_{t\in[0,1]}\|h(t)\| \le \|h\|_{h_0}.\end{aligned}\ ] ] thus , @xmath15 is @xmath81-lipschitz continuous . from theorem [ th : es ] , we deduce that @xmath303 and @xmath304 @xmath14-a.e . now , we use the chain rule of @xmath81-derivative to obtain @xmath305}q(w(t ) ) \le \frac1n\right].\end{aligned}\ ] ] according to proposition [ prop:2 ] , @xmath300 . by virtue of theorem [ th:2 ] , the process @xmath306 satisfies the skorokhod equation . next , we will prove @xmath307 , where @xmath308 denotes the capacity associated with @xmath115 . when @xmath309 , there exist at least two points @xmath310 such that @xmath311 . therefore , @xmath312}q(w(t))=0,\ \inf_{t\in[s_1,s_2]}q(w(t))=0 \right\}\right .. \nonumber\end{aligned}\ ] ] for @xmath313 and @xmath314 with @xmath315 , we define @xmath316}q(w(t))=\alpha,\ \inf_{t\in[s_1,s_2]}q(w(t))=\beta\right\}\right .. \ ] ] the right - hand side of is rewritten as @xmath317 . for a subset @xmath318 of @xmath319 , we denote @xmath320 by @xmath321 . fix @xmath314 with @xmath315 . we define a map @xmath322 by @xmath323}q(w(t)),\ \inf_{t\in[s_1,s_2]}q(w(t))\bigr).\ ] ] we denote an open ball in @xmath319 with its center at @xmath70 and radius @xmath104 by @xmath324 . by the continuity of @xmath37 , @xmath325 is an open neighborhood of @xmath326 in @xmath7 . take @xmath327 and @xmath328 . we choose a smooth function @xmath329 on @xmath229 such that @xmath330 and @xmath331 for all @xmath332 . we define a function @xmath333 by @xmath334 and set @xmath335 . since @xmath336 is a bounded @xmath81-lipschitz continuous function , it belongs to @xmath337 to @xmath338 in particular by virtue of theorem [ th : es ] . moreover , @xmath339 on @xmath340 . denoting the gradient operator on @xmath319 by @xmath341 , we have @xmath342here , @xmath343 denotes the image measure of @xmath344 by @xmath37 . in the second line , @xmath345 denotes a pairing between the elements in @xmath346 and in @xmath319 and has values in @xmath81 . the inequality from the second line to the third follows from the fact that @xmath37 is @xmath81-lipschitz continuous . now , @xmath347 by considering @xmath348 , we obtain @xmath349 from lemma [ lem : a ] , the markov property of @xmath350 and proposition [ prop : path ] , @xmath351 ) = ( f_(\mu_{a , b}\|_{\overline{w_{a , b}^{\omega}}}))(\zeta^{-1}([0,r]))\\ & = \mu_{a , b}\left [ w\in \overline{w_{a , b}^{\omega } } \left\|\ \inf_{t\in[0,s_1]}q(w(t))^2+\inf_{t\in[s_1,s_2]}q(w(t))^2 \le r^2\right . \right]\\ & \le\mu_{a , b}\left [ w\in \overline{w_{a , b}^{\omega } } \left\|\ 0\le\inf_{t\in[0,s_1]}q(w(t))\le r,\ 0\le\inf_{t\in[s_1,s_2]}q(w(t ) ) \le r\right . \right]\\ & \le \hat p_a \left[\ 0\le \inf_{t\in[0,s_1]}q({\omega}_t)\le r,\ 0\le \inf_{t\in[s_1,s_2]}q({\omega}_t)\le r \right ] \cdot \frac{p_{1-s_2}(b , b)}{p_1(a , b)}\\ & = c_{13}\hat e_a\left[1_{\{0\le \inf_{t\in[0,s_1]}q({\omega}_t)\le r\}}\hat p_{{\omega}_{s_1}}\left [ 0\le \inf_{t\in[0,s_2-s_1]}q({\omega}_t)\le r \right]\right]\\ & \le c_{14}(1+(s_2-s_1)^{-1/2})r \hat p_a\left [ 0 \le \inf_{t\in[0,s_1]}q({\omega}_t)\le r\right]\\ & \le c_{15}(1+(s_2-s_1)^{-1/2})(1+s_1^{-1/2})r^2 = c_{16}r^2 . \end{aligned}\]]thus , we obtain @xmath352_{\lambda{\varepsilon}}^{{\varepsilon}}+\int_{\lambda{\varepsilon}}^{{\varepsilon}}\frac{2\xi(r)}{r^3 } \,dr\right\}\\ & \le c_{17}(\log{\lambda})^{-2 } \bigl(1+\int_{\lambda{\varepsilon}}^{{\varepsilon}}\frac{1}{r } \,dr\bigr)\\ & = c_{17}(\log{\lambda})^{-2 } ( 1-\log { \lambda}).\end{aligned}\ ] ] therefore , @xmath353 by letting @xmath354 and @xmath355 , we obtain @xmath356 . therefore , @xmath357 the last claim follows from the above result and theorem [ th:3 ] , and the fact that @xmath56 is a smooth measure . this completes the proof of theorem [ th : main ] . 1 . in a similar and simpler way , we can prove the counterpart of theorem [ th : main ] that concerns the one - sided pinned path space on @xmath3 ; this theorem was proved in @xcite . more precisely , we define the path spaces as follows : @xmath358\to{{\mathbb r}}^d)\mid w(0)=a\},\\ w_{a}^{\omega}&=\{w\in c([0,1]\to { \omega})\mid w(0)=a\},\\ h&=\left\{h\in c([0,1]\to{{\mathbb r}}^d)\left\|\ , \begin{array}{l } h(0)=0,\ \text{$h$ is abslutely continuous}\\ \text{and } \int_0 ^ 1 \|\dot h(s)\|_{{{\mathbb r}}^d}^2\,ds<\infty \end{array}\right.\right\},\\ \partial w_{a}^{\omega}&=\left\{w\in w_{a}\left\|\ , \begin{array}{l } w(t)\in\overline{{\omega}}\text { for every $ t\in[0,1]$ and}\\ w(s)\in\partial { \omega}\text { for some $ s\in(0,1]$ } \end{array } \right.\right\},\\ \partial ' w_{a}^{\omega}&=\left\{w\in \partial w_{a}^{\omega}\left\|\ , \begin{array}{l } \text{there exists a unique } s\in(0,1]\\ \text{such that } w(s)\in\partial{\omega}\end{array}\right.\right\},\\ \overline{w_{a}^{\omega}}&=w_{a}^{\omega}\cup \partial w_{a}^{\omega}.\end{aligned}\ ] ] let @xmath359 be the probability measure on @xmath360 that is the law of the @xmath134-dimensional brownian motion starting at @xmath361 . then , we can prove the claim that is modified by replacing @xmath7 , @xmath6 , @xmath14 , @xmath89 , @xmath362 , @xmath81 , and @xmath95 in theorem [ th : main ] by @xmath360 , @xmath363 , @xmath359 , @xmath364 , @xmath365 , @xmath27 , and @xmath366 , respectively . also , from ( ? ? ? * theorem 4.4 ) , @xmath367 if @xmath368 and @xmath369 , where @xmath370 is a sobolev space on @xmath364 with differentiability index @xmath371 and integrability index @xmath372 according to the malliavin calculus . therefore , we obtain the following theorem , which is a generalization of a part of the results of @xcite . + assume that @xmath373 and @xmath3 satisfies the uniform exterior ball condition . then , @xmath374 and @xmath375 . moreover , for any real numbers @xmath371 and @xmath372 such that @xmath368 and @xmath369 , the function @xmath376 belongs to @xmath370 . 2 . precisely speaking , the diffusion process associated with @xmath115 should be called the _ modified _ reflecting ornstein - uhlenbeck process as in @xcite , since @xmath377 is defined as the completion of @xmath96 and may be strictly smaller than the `` canonical '' first order @xmath284-sobolev space @xmath378 . when @xmath377 is equal to @xmath378 remains an open problem ; a partial answer is provided in @xcite . hino , m. , on dirichlet spaces over convex sets in infinite dimensions , in _ finite and infinite dimensional analysis in honor of leonard gross ( new orleans , la , 2001 ) _ , 143156 , contemp . , * 317 * , amer . math . soc . , providence , ri , 2003 . hino , m. , integral representation of linear functionals on vector lattices and its application to bv functions on wiener space , in _ stochastic analysis and related topics in kyoto in honour of kiyosi it _ , 121140 , advanced studies in pure mathematics , * 41 * , 2004 .	consider a set of continuous maps from the interval @xmath0 $ ] to a domain in @xmath1 . although the topological boundary of this set in the path space is not smooth in general , by using the theory of functions of bounded variation ( bv functions ) on the wiener space and the theory of dirichlet forms , we can discuss the existence of the surface measure and the skorokhod representation of the reflecting ornstein - uhlenbeck process associated with the canonical dirichlet form on this set .
the inflationary paradigm , promoted by guth ( 1981 ) , linde ( 1982 ) and others , has achieved spectacular success in explaining the acoustic peak structure seen in the cmb ( see bennett 2003 and references therein ) . nevertheless , very little is known about the mechanism of inflation and how it is related to fundamental physics . the simplest mechanisms predict an adiabatic spectrum of nearly scale - invariant , gaussian , primordial fluctuations ( for reviews see linde 1990 , lyth and riotto 1999 , liddle and lyth 2000 ) . however , as peiris ( 2003 , hereafter p03 ) and kinney ( 2004 ) demonstrate , a wide class of phenomenological inflationary models are compatible with the cmb and other cosmological data . given our ignorance of the underlying physics , it is worth asking what general statements can be made about inflation from particular types of observation . for example , it is well known ( _ e.g. _ lyth 1984 ) that the amplitude of the tensor mode cmb anisotropy fixes the energy scale of slow - roll inflation @xmath5 where @xmath0 is the relative amplitude of the tensor and scalar modes defined as in p03 . , where @xmath6 and @xmath7 are the power spectra of the tensor and scalar curvature perturbations defined at a fiducial wavenumber of @xmath8 . ] the present @xmath9 upper limit of @xmath10 ( seljak 2004 ) gives the constraint @xmath11 , or equivalently @xmath12 . however , a successful model of inflation must also reproduce the observed amplitude of scalar perturbations and produce a sufficiently large number of e - foldings , @xmath13 , between the end of inflation and the time that these perturbations were generated . constraints on more general inflation models can be analysed using a set of ` inflationary flow ' equations ( hoffman and turner 2001 , kinney 2002 , easther and kinney 2003 , p03 , kinney 2004 ) describing the evolution of a hierarchy of ` slow - roll ' parameters . the first two slow - roll parameters are defined in terms of the hubble parameter , @xmath14 , during inflation by @xmath15 where primes denote differentiation of @xmath16 with respect to the inflaton @xmath17 . in terms of @xmath18 , the derivatives of @xmath17 and @xmath14 with respect to the number of e - foldings to the end of inflation are given by @xmath19 @xmath20 and @xmath21 is a necessary criterion for inflation . to second order in slow - roll parameters , the tensor - scalar ratio @xmath0 is given by @xmath22 , \label{i5}\ ] ] where @xmath23 ( see liddle , parsons and barrow 1994 ) . lyth ( 1997 ) noted that for slow - roll inflation , equations ( [ i3 ] ) and ( [ i5 ] ) can be used to relate the change in the inflaton during inflation , @xmath1 , to the tensor - scalar ratio @xmath0 , since if @xmath18 is roughly constant @xmath24 since the universe inflates by @xmath25 during the period that wavelengths corresponding to the cmb multipoles @xmath26 cross the hubble radius , equation ( [ i6 ] ) leads to a bound between @xmath1 and @xmath0 , @xmath27 which we will refer to as the ` lyth bound . ' according to this bound , high values of @xmath0 require changes in @xmath1 of order @xmath28 . in fact , equation ( [ i7 ] ) is a rather crude bound , since at least 50 - 60 e - foldings are required before inflation ends ( see _ e.g. _ liddle and leach 2003 ) . over the full course of inflation @xmath1 could therefore exceed ( [ i7 ] ) by an order of magnitude or more . in this paper , we investigate the lyth bound , making as few assumptions as possible about the nature of inflation . our aim is to establish whether a relation of the form ( [ i7 ] ) applies to general models of inflation that are compatible with observational constraints on the scalar fluctuation spectrum . lyth ( 1997 ) , liddle and lyth ( 2000 ) , kinney ( 2003 ) and others argue that inflation can not be described by a low energy effective field theory if @xmath29 and so high values of @xmath30 are possible only in models for which no rigorous theoretical framework exists . conversely , it has been argued that the lyth bound requires @xmath31 in models of inflation based on well - motivated particle physics . in section 3 , we will comment on these points and discuss the implications of the lyth bound for inflationary model building and for future cmb experiments designed to detect tensor modes . as mentioned in the introduction , inflationary models can be described by an infinite hierarchy of slow roll parameters @xmath32 that satisfy a set of ` inflationary flow ' equations . the inflationary flow approach has been used by a large number of authors and we refer the reader to p03 and kinney ( 2004 ) for a summary of the approach and for the complete set of flow equations . the models generated here followed the prescription given in kinney ( 2004 ) , except that the flow hierarchy was truncated at @xmath33 ( rather than @xmath34 ) and that successful inflationary models were required to expand by at least @xmath35 e - folds , consistent with the analysis of liddle and leach ( 2003 ) . inflation was deemed to end if @xmath36 , in which case the models were evolved backwards by @xmath2 e - foldings to compute various observables , such as the tensor - scalar ratio @xmath0 , scalar spectral index @xmath37 , and the run in the spectral index @xmath38 ( using expressions accurate to second order in slow - roll parameters ) . for models in which inflation occurred at the minimum of a potential ( @xmath39 ) , inflation was simply abruptly terminated and observables calculated using the slow - roll parameters at the end of inflation . physically , such cases can be considered as examples of hybrid inflation ( see liddle and lyth 2000 ) when inflation ends abruptly at some critical field value @xmath40 , or as examples of brane - inflation when inflation ends at a critical inter - brane separation at which open string modes become tachyonic ( see quevedo , 2002 , for a review ) . in practice , any models which achieved @xmath41 e - foldings were grouped into this category . we evolved @xmath42 models using the above prescription . figure 1 shows the parameters @xmath0 , @xmath37 and @xmath43 for a sample of these models plotted against each other . the distributions in these diagrams agree well with the results of previous authors and are straightforward to understand physically . more than 90% of the models are of the ` hybrid - type ' for which @xmath0 is negligible ( equation 4 ) and the scalar spectral index is @xmath44 ( since the inflaton is trapped in the minimum of a potential ) . of the remaining models , those colour coded in blue in figure 1 have @xmath10 and so are consistent with current constraints on the tensor - scalar ratio . the middle panel of figure 1 shows that these models span a wide range of @xmath37 and @xmath45 . 3.5 truein figure 2 shows plots of @xmath46 over the final @xmath2 e - foldings of inflation computed from equation ( 3a ) plotted against the tensor - scalar ratio @xmath0 . the figure to the left shows @xmath47 for all models within the designated ranges of the abscissa and ordinate , colour coded so that hybrid - type models are plotted in red , with the rest of the models plotted in blue . this figure shows that for general inflationary models , there is no well defined lyth relation of the form ( [ i7 ] ) . models can be found , for example , with @xmath48 and @xmath49 in the range @xmath50@xmath51 . likewise , models can be found with low values of @xmath52 and @xmath53 well in excess of unity . however , on closer inspection , one finds that models lying in these regions of the diagram have unusual scalar spectral indices . in the former case , the models have blue spectra with @xmath54 , and in the latter case the models have red spectra with @xmath55 . it is therefore interesting to plot @xmath0 against @xmath1 only for those models that satisfy observation constraints on the shape of the scalar power spectrum . the tightest observational constraints on @xmath37 and @xmath56 at present come from combining observations of the cmb anisotropies , with observations of the matter power spectrum deduced from galaxy redshift surveys and from ly@xmath57 lines in the spectra of high redshift quasars ( see _ e.g. _ viel 2004 , seljak 2004 ) . the recent results of seljak give approximate @xmath58 ranges of @xmath59 ( see figure 3 of seljak 2004 ) . figure 2b shows @xmath0 and @xmath1 for the subset of models that satisfy the constraints given in ( 8) . ( in fact , although we have imposed a constraint on @xmath60 , it is the constraint on @xmath37 that has the most significant effect in defining the distribution of points in figure 2b . ) the models plotted in figure 2b now delineate a much tighter relationship between @xmath0 and @xmath1 , which can be approximated for @xmath3 by @xmath61 this relation is our reformulation of the lyth bound ( 6 ) . 3.5 truein present observations set the constraint @xmath10 ( seljak 2004 ) . to improve on this limit significantly will require more sensitive cmb polarization experiments than those done so far . the _ planck _ satellite should be able to achieve a limit of @xmath62 by measuring the b - mode polarization spectrum , but despite the fact that _ planck _ will survey the entire sky from space it is limited in sensitivity by the small number of polarization sensitive detectors . ground based b - mode optimised experiments , using either bolometer arrays ( clover , taylor 2004 ) , or large arrays of coherent detectors ( quieti.htm ] ) are currently being designed that should be able to set limits of @xmath63 in the presence of realistic foregrounds . for @xmath64 , b - mode anisotropies caused by gravitational lensing of the cmb ( zaldarriaga and seljak 1998 ) will need to be subtracted in order to extract an inflationary tensor component . however , there are limits on how accurately a lensing contribution can be subtracted . for example , even in the absence of instrumental noise and foregrounds , lewis ( 2002 ) find that a survey of the sky of radius @xmath65 leads to a limit of @xmath66 from the sample variance of the lens - induced b - modes . this limit could potentially be improved by a factor of @xmath67 with a noiseless all sky survey ( kesden 2002 , knox and song 2002 ) , and perhaps by another factor of @xmath67 if the lens - induced b - modes can be mapped accurately enough to reconstruct the deflection field ( hirata and seljak 2003 ) . comparing with equation ( 1 ) , under optimistic assumptions the lowest energy scale of inflation that can be probed by cmb b - mode polarization measurements is @xmath68 . more realistically , the next generation of cmb experiments such as clover and quiet may reach @xmath63 , corresponding to an energy scale of @xmath69 . figure 2b shows that to produce a detectable tensor component in any foreseeable cmb experiment , inflation must necessarily involve large field variations , @xmath70 . in fact , the dependence of @xmath1 on @xmath0 in figure 2b is so steep that we would need to achieve @xmath71 to probe inflationary models with low field variations of @xmath72 . thus , for the foreseeable future , we will only be able to test high - field models of inflation . furthermore , it may well prove impossible , given realistic polarized foregrounds , to probe small - field models of inflation using the cmb . it is often argued ( _ e.g. _ lyth 1997 ) that the effective potential can be written as a power series @xmath73 with @xmath74 , and hence that an effective field theory description of inflation becomes invalid for field values @xmath75 . however , as linde ( 2004 ) points out , quantum gravity corrections to @xmath76 should become large only for @xmath77 , not @xmath78 as inferred from equation ( [ c1 ] ) . this is the rationale behind chaotic inflation models such as the simple @xmath79 model ( which is now excluded observationally at about the @xmath80 level ) . however , it has proved difficult to find realisations of chaotic inflation motivated by realistic particle physics . for example , realising the conditions for slow - roll inflation for either large or small field variations has been a long standing problem in @xmath81 supergravity . this may , however , simply reflect our lack of understanding of supergravity . for example , it has been pointed out recently that chaotic inflation can be realised in supergravity models if the potential has a shift symmetry ( kawasaki 2000 , yamaguchi and yokoyama 2001 ) , _ i.e. _ the inflaton potential does not depend on the imaginary part of a complex field @xmath82 ( see linde 2004 for a more detailed discussion ) . it may therefore be possible to construct particle physics motivated models of inflation with @xmath83 . understanding the physics behind such high - field models is important , for as we have shown in this paper they are the only class of inflationary models that can be probed by cmb b - mode polarization experiments in the foreseeable future . _ acknowledgments : _ the research of gpe is supported by the uk particle and astrophysics research council . km would like to thank the institute of astronomy for their hospitality during a summer visit when part of this work was done . bennett c.l . , , 2003 , _ ap j. suppl _ , * 148 * , 1 .	we investigate the lyth relationship between the tensor - scalar ratio , @xmath0 , and the variation of the inflaton field , @xmath1 , over the course of inflation . for inflationary models that produce at least @xmath2 e - folds of inflation , there is a correlation between @xmath0 and @xmath1 as anticipated by lyth , but the scatter around the relationship is huge . however , for inflationary models that satisfy current observational constraints on the scalar spectral index and its first derivative , the lyth relationship is much tighter . in particular , any inflationary model with @xmath3 must have @xmath4 . large field variations are therefore required if a tensor mode signal is to be detected in any foreseeable cosmic microwave background ( cmb ) polarization experiment .
single top quark production at hadron colliders can proceed through three distinct partonic processes : the @xmath0 channel ( @xmath1 and @xmath2 ) which involves the exchange of a spacelike @xmath3 boson , the @xmath4 channel ( @xmath5 ) which proceeds via a timelike @xmath3 boson , and associated @xmath6 production ( @xmath7 ) @xcite . the @xmath0 channel processes are numerically the largest at both the tevatron and the lhc . at the tevatron the @xmath4 channel is second in magnitude and @xmath6 production has the smallest cross section . at the lhc , @xmath6 production has a much bigger cross section than the @xmath4 channel . the cross sections for all these processes receive contributions from soft - gluon emission which can be dominant near threshold @xcite . threshold resummation organizes these contributions and can be used to compute higher - order corrections for many processes @xcite . for the partonic process with momenta @xmath8 we define @xmath9 , @xmath10 , @xmath11 and @xmath12 . near threshold , @xmath13 approaches zero and the soft - gluon corrections take the form @xmath14_+$ ] , where @xmath15 is the top quark mass and @xmath16 for the @xmath17-th order corrections . we calculate these corrections through next - to - next - to - next - to - leading order ( nnnlo ) in the strong coupling @xmath18 at next - to - leading logarithmic ( nll ) accuracy at the tevatron @xcite and the lhc @xcite . this requires one - loop calculations in the eikonal approximation . the nlo soft - gluon corrections can be written in the form @xmath19_+ + c_{2 } \left[\frac{1}{s_4}\right]_+ + c_{1}^{\mu } \ , \delta(s_4)\right\ } \nonumber\end{aligned}\ ] ] where @xmath20 is the born term for each channel and @xmath21 is the renormalization scale . for the @xmath0 and @xmath4 channels the leading logarithm coefficient is @xmath22 while for the @xmath6 channel it is @xmath23 , where @xmath24 and @xmath25 with @xmath26 the number of colors . the nll coefficient is @xmath27 for the @xmath4 channel , where @xmath28 is the factorization scale , and similar expressions can be given for the other channels . the complete virtual corrections ( @xmath29 terms ) can not be derived from threshold resummation but one can derive the factorization and renormalization scale terms denoted by @xmath30 in the above equation @xcite . the nnlo soft - gluon corrections for the @xmath0 and @xmath4 channels are @xmath31_+ \right . \nonumber \\ & & \hspace{20 mm } \left . { } + \left[\frac{3}{2 } c_3 \ , c_2 -\frac{\beta_0}{4 } c_3 + c_f \frac{\beta_0}{8}\right ] \left[\frac{\ln^2(s_4/m_t^2)}{s_4}\right]_+ \right\ } \nonumber\end{aligned}\ ] ] plus subleading terms @xcite , where the appropriate expression for @xmath32 and @xmath33 , @xmath34 for each channel must be used , and where @xmath35 is the lowest - order @xmath36 function , with @xmath37 the number of light quark flavors . a similar expression holds for the @xmath6 channel ( by deleting @xmath38 above ) . since this is a nll calculation , only the leading and nll terms shown above are complete . however we can also calculate exactly terms involving @xmath28 and @xmath21 as well as terms with @xmath39 constants in the subleading logarithms . complete expressions and further details are provided in ref . @xcite . the nnnlo soft - gluon corrections for each channel can be written as @xmath40_+ \right . \nonumber \\ & & \hspace{20 mm } \left . { } + \left[\frac{5}{8 } c_3 ^ 2 c_2 -\frac{5}{48 } \beta_0 c_3 ( 2 c_3-c_f ) \right ] \left[\frac{\ln^4(s_4/m_t^2)}{s_4}\right]_+ \right\ } \nonumber\end{aligned}\ ] ] plus subleading terms @xcite , again with the appropriate expression for @xmath32 and @xmath33 , @xmath34 . -channel single top quark cross section at the tevatron.,title="fig:",width=249 ] -channel single top quark cross section at the tevatron.,title="fig:",width=249 ] [ fig : t - chan.tev ] we now calculate the contribution of these corrections to the single top cross section at the fermilab tevatron . the mrst 2004 nnlo parton densities @xcite are used for the numerical results . we find that the threshold corrections are dominant in all partonic channels . figure 1 shows the results for the cross section in the @xmath0 channel . in the left - hand plot we show the leading - order ( lo ) cross section as well as the cross sections with the nlo , nnlo , and nnnlo soft - gluon corrections included versus the top quark mass @xmath15 , with the factorization and renormalization scales set equal to @xmath15 . on the right - hand plot we show the @xmath41 factors , which are the ratios of the higher - order cross sections to lo . we see that the corrections in this channel are relatively small . our best estimate for the cross section is calculated after matching to the exact nlo cross section @xcite , i.e. by adding the soft - gluon corrections through nnnlo to the exact nlo cross section . below we give results for two choices of the top quark mass , @xmath42 gev and @xmath43 gev . we find @xmath44 pb and @xmath45 pb . the uncertainty indicated includes the scale dependence and the pdf uncertainties . figure 2 shows the results for the cross section and @xmath41 factors in the @xmath4 channel . in this channel the corrections are large , providing up to 65% enhancement of the leading - order cross section . after matching , we find @xmath46 pb and @xmath47 pb . the single top cross section at the tevatron in the @xmath6 channel is rather small , even though the @xmath41 factors are large ( up to 85% enhancement ) . our estimate for the cross section is @xmath48 pb and @xmath49 pb . for all three channels at the tevatron the cross section for single anti - top production is identical to that shown above for single top production . -channel single top quark cross section at the tevatron.,title="fig:",width=249 ] -channel single top quark cross section at the tevatron.,title="fig:",width=249 ] [ fig : s - chan.tev ] finally , we note that there has been recent evidence for single top quark production at the tevatron @xcite with a cross section consistent with the above results . next we calculate the threshold corrections for the single top cross section at the cern lhc . it turns out that in the @xmath0 channel the threshold corrections are not a good approximation of full qcd corrections , hence we only update the exact nlo result @xcite , while in the @xmath4 and @xmath6 channels the threshold approximation holds and we provide results including the nnnlo soft - gluon corrections . also at the lhc the cross section for single top is different from that for single antitop production in the @xmath0 and @xmath4 channels . the exact nlo cross section for single top production in the @xmath0 channel at the lhc is @xmath50 pb and @xmath51 pb . for single antitop production in the @xmath0 channel the exact nlo cross section is @xmath52 pb and @xmath53 pb . figure 3 ( left ) shows results for single top production in the @xmath4 channel at the lhc . the contribution from soft gluons is significant ( up to 55% enhancement ) . after matching to the exact nlo cross section @xcite , we find @xmath54 pb and @xmath55 pb . the corresponding results for single antitop production at the lhc in the @xmath4 channel are @xmath56 pb and @xmath57 pb . here the soft - gluon corrections are somewhat smaller ( less than 20% ) . figure 3 ( right ) shows results for single top production at the lhc in the @xmath6 channel . this channel has a significant cross section at the lhc . also the soft - gluon corrections are quite large , providing around 60% enhancement . after matching to the exact nlo cross section @xcite , we find @xmath58 pb and @xmath59 pb . the cross section for associated antitop production is identical to that for a top quark . -channel single top ( left ) and @xmath6 ( right ) cross sections at the lhc.,title="fig:",width=249 ] -channel single top ( left ) and @xmath6 ( right ) cross sections at the lhc.,title="fig:",width=249 ] [ fig : s - chan.tw.lhc ] this work has been supported by the national science foundation under grant no . phy 0555372 . n. kidonakis and a. belyaev , jhep * 12 * , 004 ( 2003 ) [ hep - ph/0310299 ] ; n. kidonakis and a. sabio vera , jhep * 02 * , 027 ( 2004 ) [ hep - ph/0311266 ] ; r.j . gonsalves , n. kidonakis and a. sabio vera , phys . lett . * 95 * , 222001 ( 2005 ) [ hep - ph/0507317 ] ; n. kidonakis , r.j . gonsalves , and a. sabio vera , in dis 2006 , p. 337 ( 2007 ) [ hep - ph/0606145 ] .	i discuss single top quark production at the tevatron and the lhc . the cross section , including soft - gluon threshold corrections through nnnlo , is presented for each partonic channel . the higher - order corrections provide significant contributions to the single top cross sections at both colliders . * higher - order threshold corrections for single top quark production * nikolaos kidonakis + +
perovskite - type manganites of the chemical formula @xmath6mno@xmath2 ( where @xmath7 is a rare - earth and @xmath8 is an alkaline - earth metal ) have attracted much interest because of their colossal magnetoresistance ( cmr ) @xcite . historically , the magnetic and transport properties of this kind of materials have been understood in terms of double - exchange ( de ) interaction between the localized @xmath9 electrons and the itinerant @xmath4 electrons @xcite . in this model , the kinetic energy gain of the holes doped into the @xmath4 band is maximized when the spins of the @xmath9 electrons are aligned in the same direction , thereby stabilizing the ferromagnetic ground state . however , it has been pointed out by millis _ et al . _ @xcite that the de model is insufficient to explain the huge change in the resistivity under a magnetic field and the high resistivity above the curie temperature ( @xmath5 ) . they have proposed a model including the dynamical jahn - teller ( jt ) distortion of the mno@xmath10 octahedra in addition to the de interaction and reproduced the experimentally observed behavior of the resistivity . it has also been proposed that the extraordinary enhancement of the resistivity in the manganites may result from the emergence of lattice polarons in the paramagnetic insulating ( pi ) phase @xcite . _ @xcite have found that lattice polarons in nd@xmath11sr@xmath12mno@xmath2 are strongly suppressed by applying magnetic field and do not completely disappear at high fields , corresponding to an admixture of the conducting " and insulating " carriers . from x - ray scattering and neutron scattering studies @xcite , indeed , the diffuse scattering due to the lattice polarons has been observed in the high - temperature pi phase of manganites and gradually disappears with decreasing temperature in the ferromagnetic metallic ( fm ) phase . on the other hand , several local structural studies have suggested that a dynamical or local jt distortion persists in the fm phase , too . x - ray absorption fine structure ( exafs ) studies have shown that the lattice distortion of the mno@xmath10 octahedra is found in the fm phase of la@xmath0ca@xmath13mno@xmath2 at low temperatures @xcite . from the pair - density function analysis of pulsed neutron diffraction data , the local jt distortion has been observed in the fm phase of la@xmath0sr@xmath13mno@xmath2 @xcite . if the jt distortion exists in the fm phase , the degeneracy of the @xmath4 band may be lifted already in that phase . ( a ) degenerate two - orbital model ; ( b ) one - orbital model resulting from the jahn - teller splitting . @xmath15 , @xmath5 and @xmath16 denote the chemical potential , the curie temperature and the hole concentration , respectively.,width=264 ] in order to obtain insight into the competition between the de mechanism and the lattice - polaron effect , the measurement of chemical potential shift ( @xmath17 ) as a function of temperature gives much insight . according to the de model , temperature - dependent chemical potential shift occurs due to the change in the @xmath4 band width as schematically shown in fig . 1 @xcite . if the @xmath4 band is split by jt distortion and the one - orbital de model becomes relevant , the chemical potential is shifted upward for hole concentration @[email protected] with decreasing temperature and downward for @[email protected] @xcite . if one takes into account the double degeneracy of the @xmath4 orbitals [ fig . 1(a ) ] , a downward shift with decreasing temperature would be expected in the fm phase for 0@xmath14@xmath16@xmath141 because the the up - spin band of the @xmath4 orbitals is less than half - filled in the @xmath6mno@xmath2 compounds . therefore , the temperature - dependent chemical potential shift is sensitive to the splitting of the @xmath4 band and therefore to the dynamical / local jt effect . @xcite have investigated the change of the work function in la@xmath19sr@xmath20mn@xmath21o@xmath2 as a function of temperature by measurements of photoemission spectra and attributed the change to the temperature - dependent shift of chemical potential . alternatively , the chemical potential shift can be deduced from the shifts of photoemission spectra because the binding energies of the spectra are measured relative to the chemical potential . we employ the latter method in this work . pi : paramagnetic insulating phase ; ci : spin - canted insulating phase ; fm : ferromagnetic metallic phase ; fi : ferromagnetic insulating phase ; co / ooi : charge - orbital ordered insulating phase ; afm : antiferromagnetic metallic phase ; afi : antiferromagnetic insulating phase @xcite.,width=245 ] nd@xmath0sr@xmath1mno@xmath2 ( nsmo ) is a suitable system for clarifying the relationship between the de interactions and the existence of lattice polarons because it shows the fm phase for @xmath22 and the so - called ce - type antiferromagnetic ( af ) charge - ordered ( co ) phase in the doping region close to the half - doping @xmath16 = 0.5 as shown in fig . 2 @xcite . in this work , we study the chemical potential shift in nsmo as a function of temperature by measurements of core - level photoemission spectra . we found that the chemical potential shift was large in the low - temperature part of the fm phase as predicted by the de model and dynamical / local jt effect and was suppressed at high temperatures near @xmath5 . we consider that the different behaviors with temperature to be related to the competition between the de interaction and the lattice - polaron effect . single crystals of nsmo ( @xmath3 and 0.45 ) were prepared by the floating zone method @xcite . x - ray photoemission spectroscopy measurements were performed using the photon energies of @xmath23 1253.6 ev ( mg @xmath24 ) . all the photoemission measurements were performed under the base pressure of @xmath25 torr at 20 - 330 k. the samples were repeatedly scraped _ in situ _ with a diamond file to obtain clean surfaces . the cleanliness of the sample surface was checked by the reduction of the shoulder on the high binding energy side of the o 1@xmath26 core level . photoelectrons were collected using a scienta ses-100 electron - energy analyzer . the energy resolution was about 800 mev . the measured binding energies were stable , because the gold @xmath27 core - level spectrum did not changed in the measurements with the accuracy of @xmath28 mev at each temperature . 3 , we have plotted the spectra of the o @xmath29 , sr @xmath30 , nd @xmath30 and mn @xmath31 core levels in nsmo with @xmath3 . the vertical lines mark the estimated positions of the core levels used in the present study . we employed the midpoint of the low binding - energy slope for the o @xmath29 core level because the line shape on the higher - binding energy side of the o @xmath29 spectra is known to be affected by surface contamination or degradation . we also employed the midpoint for the sr @xmath30 and mn @xmath31 core levels . as for the nd @xmath30 core level , 80 @xmath32 of the peak height of the low binding - energy slope was used because the line shape near the midpoint on the lower - binding energy side slightly changed with temperature . sr@xmath33mno@xmath2 taken with the mg @xmath24 line . ( a ) o @xmath29 ; ( b ) sr @xmath30 ; ( c ) nd @xmath30 ; ( d ) mn @xmath31.,width=351 ] figure 4(a ) shows the binding energy shift of each core level as a function of temperature for nsmo with @xmath3 . one can see that the observed binding energy shifts with temperature were approximately common to the o @xmath29 , sr @xmath30 , nd @xmath30 and mn @xmath31 core levels . from the measurements of the doping - dependent chemical potential shift in pr@xmath0ca@xmath1mno@xmath2 , all the core levels have shown identical shifts with hole concentration except for the mn @xmath31 core level , where the effect of chemical shift is superimposed @xcite . here , all the core levels including the mn 2@xmath34 core level exhibit similar shifts with temperature in contrast to the measurements of doping - dependent core - level shifts @xcite . therefore , we assume that the shifts of the core levels are largely due to the chemical potential shift , and take the average of the shifts of the four core levels as a measure of @xmath17 as a function of temperature in nsmo . sr@xmath33mno@xmath2 . ( a ) binding energy shifts of the o @xmath29 , sr @xmath30 , nd @xmath30 , and mn @xmath31 core levels as functions of temperature relative to 290 k ; ( b ) chemical potential shift @xmath17 as a function of temperature . @xmath5 denotes curie temperature.,width=340 ] figure 4 ( b ) shows the temperature - dependent chemical potential shift in nsmo with @xmath3 . we observed a large upward chemical potential shift with decreasing temperature in the fm phase of nsmo at low temperatures . furukawa @xcite proposed an anomalous temperature - dependent chemical potential shift below @xmath5 due to the de interaction through the change of the @xmath4 band width with temperature . he also predicted that the magnitude of the shift was estimated to be about 0.1 ev when the @xmath4 band width was of order @xmath35 1 ev @xcite . if the @xmath4 band remains degenerate in the fm phase of nsmo , one would expect to see a downward chemical potential shift with decreasing temperature based on the de interaction because the up - spin band of the @xmath4 orbitals is less than half - filled for 0@xmath14@xmath16@xmath141 . on the other hand , if the degeneracy of the @xmath4 band is lifted by the dynamical / local jt distortion , the upward chemical potential shift with decreasing temperature is predicted because the band is more than half filled for @[email protected] ( see fig . 1 ) . therefore , we attribute the large upward shift of the chemical potential with decreasing temperature in the low - temperature fm phase of nsmo to the change in the width of the jt - split @xmath4 band caused by the de mechanism . the magnitude of the observed shift is in quantitative agreement with the results of the one - orbital de model @xcite . we consider that the temperature - dependent splitting of the @xmath4 band may also influence the temperature - dependent chemical potential shift in the fm phase of nsmo because the intensity of the diffuse scattering due to jt effect gradually increases with increasing temperature in the fm phase , which indicates further increase of the energy splitting of the @xmath4 level @xcite . if this is the case , the temperature - dependent shift would be even stronger at higher temperatures for the one - orbital case and the suppression of chemical potential shift in nsmo within the fm phase near @xmath5 can not be explained . at high temperatures near @xmath5 in the pi phase , the diffuse scattering due to the formation of lattice polarons has been observed by means of x - ray scattering and neutron scattering studies @xcite . we consider that the suppression of the shift in the fm phase at high temperatures is connected with the influence of the lattice polarons and the de model is no more effective at those high temperatures . in nd@xmath0sr@xmath1mno@xmath2 ( @xmath3 ) with the shifts of the o 1@xmath26 core level ( in the process of increasing temperature ) and the valence band of la@xmath0sr@xmath1mno@xmath2 ( @xmath36 , 0.3 and 0.4 with @xmath37 , 370 , and 370 k , respectively ) @xcite , @xmath17 for the bilayered system la@xmath38sr@xmath39mn@xmath21o@xmath40 ( @xmath3 with @xmath41 k ) @xcite and correlated hopping model of mn@xmath42/mn@xmath43 mixed - valence state . the spectral shifts have been translated to the chemical potential shift , assuming that the spectral shifts are primarily close to the chemical potential shift.,width=340 ] in fig . 5 , we compare @xmath17 for nsmo ( @xmath3 ) with the shifts of the o 1@xmath26 core level ( in the process of increasing temperature ) and the valence band of la@xmath0sr@xmath1mno@xmath2 ( @xmath36 , 0.3 and 0.4 ) , @xmath17 for bilayered system la@xmath38sr@xmath39mn@xmath21o@xmath40 ( @xmath3 ) and correlated hopping model of mn@xmath42/mn@xmath43 mixed - valence state as a function of temperature @xcite . we consider that the temperature - dependent shifts of the o 1@xmath26 core level and the valence band for la@xmath0sr@xmath1mno@xmath2 are ascribed to the temperature - dependent change in @xmath17 . for nsmo and la@xmath0sr@xmath1mno@xmath2 , the @xmath17 curves show similar temperature dependences in the sense that the chemical potential is shifted upward with decreasing temperature . we consider that the shifts are understood in terms of the de interaction ( and possibly the temperature dependence of the splitting of the @xmath4 band caused by the dynamical / local jt effect ) @xcite . an upward chemical potential shift by applying magnetic field has also been observed in la@xmath0sr@xmath1mno@xmath2 , which may be ascribed to the de interaction in the jt split @xmath4 band @xcite . however , the @xmath17 for the bilayered compound la@xmath38sr@xmath39mn@xmath21o@xmath40 reported by schulte _ @xcite is opposite to the prediction of the de model . they speculated that the two dimensionality might affect the temperature - dependent chemical potential shift , but its origin remains as an open question . also , we have compared the experimentally observed shifts with the results of the correlated hopping model in the splitting of the @xmath4 band due to the dynamical / local jt distortion @xcite . the correlated hopping model is given by @xmath44 , where @xmath45 and @xmath46 are the spin - orbital degeneracies of mn@xmath42 and mn@xmath43 , respectively and @xmath16 is the fraction of mn@xmath43 ions @xcite . the parameters were fixed at @xmath47 , @xmath48 and @xmath3 , corresponding to the mn@xmath42/mn@xmath43 mixed - valence state in the split of the @xmath4 band . we have plotted the calculated @xmath17 as shown by a dashed line in fig . the result is in qualitative agreement with the observed chemical potential shift in high - temperature region of nsmo with @xmath3 . @xmath5 denotes curie temperature.,width=340 ] for other hole concentrations of nsmo , too , we confirmed that the observed temperature - dependent shifts were common to the o @xmath29 , sr @xmath30 , nd @xmath30 and mn @xmath31 core levels and deduced the temperature - dependent chemical potential shifts by taking the average of the shifts of the four core levels . figure 6 shows the @xmath17 for nsmo with @xmath50 as a function of temperature . the shift @xmath17 was suppressed just below the @xmath5 as in the case of @xmath3 , in accordance with the enhancement of the diffuse scattering due to the lattice polarons @xcite . we attribute the shift in the low - temperature region of the fm phase for @xmath50 to the de interaction ( and the temperature dependence of the splitting of the @xmath4 band due to the dynamical / local jt effect ) as in the case of @xmath3 . @xcite . the magnitude of the observed shift for @xmath50 is smaller than that for @xmath3 , consistent with the one - orbital de model resulting from the jt splitting [ see fig . we have measured the chemical potential shift as a function of temperature in nsmo by means of core - level photoemission spectroscopy . we have found an anomalous upward chemical potential shift with decreasing temperature in the low - temperature region of the fm phase and its suppression in the high - temperature region of the fm phase near @xmath5 . we attribute the large shift in the low - temperature region to the change of band width due to the de interaction ( and possibly the temperature dependence of the splitting of the @xmath4 level caused by the dynamical / local jt effect . ) also , the suppression of the shift at higher temperatures is ascribed to the influence of lattice - polaron formation . informative discussions with n. furukawa and h. wadati are gratefully acknowledged . this work was supported by a grant - in - aid for scientific research in priority area invention of anomalous quantum materials " from the ministry of education , culture , sports , science and technology , japan .	we have studied the chemical potential shift as a function of temperature in nd@xmath0sr@xmath1mno@xmath2 ( nsmo ) by measurements of core - level photoemission spectra . for ferromagnetic samples ( @xmath3 and 0.45 ) , we observed an unusually large upward chemical potential shift with decreasing temperature in the low - temperature region of the ferromagnetic metallic ( fm ) phase . this can be explained by the double - exchange ( de ) mechanism if the @xmath4 band is split by dynamical / local jahn - teller effect . the shift was suppressed near the curie temperature ( @xmath5 ) , which we attribute to the crossover from the de to lattice - polaron regimes .
the ultradiscrete kdv equation : @xmath0 \label{udkdv}\ ] ] was first introduced as a dynamical equation for the so - called box - ball system ( bbs)@xcite . since is closed on the binary values @xmath1 or @xmath2 , if we regard @xmath3 as the number of balls in the @xmath4th box at time step @xmath5 , and assume the boundary condition @xmath6 , it describes the time evolution of balls in a one dimensional array of boxes . the updating rule of the bbs according to can be expressed as follows . ( see figure [ fig01 ] . ) * find all the pairs of a ball and an adjacent vacant box to the right of it , and draw arclines from the ball to the vacant box for each of these pairs . * neglecting the pairs connected by arclines , find all the pairs of a ball and an adjacent vacant box to the right of it and draw arclines in a similar way . * repeat the above procedure until all the balls are connected to vacant boxes , then move all the balls into the vacant boxes connected by arclines . an example of the time evolution pattern is shown in figure [ fig02 ] . the consecutive balls behave like solitons in the kdv equation . furthermore , when we denote by @xmath7 the number of arclines drawn in the first step of the updating rule , by @xmath8 one in the second step and so on , we obtain a nonincreasing integer sequence @xmath9 , which is a conserved quantity of the bbs in time@xcite . . ) ] the reason why the bbs has these solitonic properties is understood by the fact that is obtained from the kdv equation through a limiting procedure , ultradiscretization@xcite . the process involved in obtaining from the kdv equation can be explained as follows . the integrable discrete kdv equation is given in the bilinear form as @xmath10 where @xmath11 and @xmath12 is a discretization parameter@xcite . by putting @xmath13 gives the simultaneous equations @xmath14 when we impose the boundary condition @xmath15 , then @xmath16 from and we have @xmath17 we introduce a parameter @xmath18 and put @xmath19 . if there exists a one parameter family of solutions to @xmath20 , and if the limit @xmath21 exists , then , using the identities for arbitrary real numbers @xmath22 and @xmath23 @xmath24,\\ & \lim_{\epsilon \to + 0 } \epsilon \log \left ( e^{a/\epsilon } \cdot e^{b/\epsilon}\right ) = a+b,\\ & -\max[-a ,- b]=\min[a , b],\end{aligned}\ ] ] we find that @xmath3 satisfies . recently , has been studied without the constraint of binary values@xcite . one interesting feature in this extension is that there exist background solutions composed of so - called negative solitons travelling at speed one@xcite . a typical example of the background solutions is shown in figure [ fig03 ] . furthermore , such negative solitons interact nontrivially with ordinary solitons as shown in figure [ fig04 ] . a method to solve the initial value problem for this generalized situation is given in ref.@xcite . from these properties , the dynamical systems described by in the general setting are also regarded as integrable cellular automata , just as the bbs . however , several important features of integrable cellular automata have not been clarified yet , such as the conserved quantities , the relation to integrable lattice models , extension to the case of a cyclic boundary condition and so on , all of which are well understood in the case of bbs . in this article , we report that a simple transformation reveals the underlying integrable structure of for the generalized states . by this transformation , the relation to the bbs and hence to the integrable lattice models , and generalization to a different boundary condition become apparent . we explicitly show how to obtain the conserved quantities and give the expression of general solutions to where negative solitons and ordinary solitons coexist . we start from the ultradiscretization of the coupled equations and . by putting @xmath25 and assuming that the limit @xmath26 exist , we obtain the equations @xmath27 , \label{combinatorialr1 } \\ v_{n+1}^t&=u_n^t+v_n^t - u_n^{t+1}. \label{combinatorialr2}\end{aligned}\ ] ] if we impose the boundary condition @xmath28 , eqs . and are equivalent to . it is easily seen that and are generalized by introducing a positive integer constants @xmath29 and @xmath30 as @xmath31-\min\left[u_n^t , c - v_n^t \right ] , \label{carrierr1 } \\ v_{n+1}^t&=u_n^t+v_n^t - u_n^{t+1}. \label{carrierr2}\end{aligned}\ ] ] equations and are known as the ultradiscrete limit of the modified kdv equation and are interpreted as a bbs of box capacity @xmath29 with carrier of capacity @xmath30 @xcite . at the same time , they represent the action of the combinatorial r - matrix of an @xmath32 crystal on a certain symmetric tensor product representation as long as @xmath33 and @xmath34@xcite . note that eqs . and are closed under the values @xmath33 and @xmath34 . hence the time evolution patterns of the bbs are regarded as the ground states of a 2 dimensional integrable lattice model at temperature 0 and its generalization has been investigated in detail@xcite . when we allow @xmath3 or @xmath35 in and to take negative integer values and/or the values greater than @xmath29 or @xmath30 respectively , they are no longer interpreted as a bbs nor as a realization of the combinatorial r - matrix . for an arbitrary initial state including negative integer values , let @xmath36 $ ] . when we change variables @xmath37 , then we find that eqs . and are unchanged under this gauge transformation@xcite , that is , @xmath38-\min\left[\tilde{u}_n^t,\tilde{c}-\tilde{v}_n^t\right],\\ \tilde{v}_{n+1}^t&=\tilde{u}_n^t+\tilde{v}_n^t-\tilde{u}_n^{t+1}.\end{aligned}\ ] ] since these equations are closed under the values @xmath39 and @xmath40 , and initial values @xmath41 belong to the same sets , we can regard them as equations for a bbs with box capacity @xmath42 and carrier capacity @xmath43 , and regard them as a realization of @xmath32 crystal lattice composed of a larger tensor product representation . thus we can conclude that eqs . and as well as eqs . and do not change their mathematical structure even in the case where the dependent variable take negative integer values . from the above argument , we find that we can identify the ultradiscrete kdv equation as a bbs with larger box capacity when its dependent variables take negative integer values . the conserved quantities of the bbs have been widely investigated in general cases and we can apply the results to this system@xcite . in @xcite , it is shown that the energy functions which are combinatorially determined , are conserved in time , and in @xcite the path description for the conserved quantities is given . here we give another expression of the conserved quantities for . let @xmath44 $ ] . then eqs . and are transformed to the bbs with box capacity @xmath45 ( with infinite carrier capacity ) . let the number of balls in the @xmath4th box be @xmath46 . we consider the time evolution starting from the time step @xmath47 , and choose the two integers @xmath48 and @xmath49 which satisfy the following conditions . 1 . @xmath50 for @xmath51 and @xmath52 ( @xmath53 @xmath54 ) . 2 . @xmath55 ( mod @xmath56 ) . no vacancy in the @xmath49th box is connected by an arcline in the first step of the construction of the conserved quantities which we explain below . the conditions 2 , 3 are used to avoid the redundancy of the definition of conserved quantities . for the time step @xmath5 , from the boundary condition , there exists two integers @xmath57 such that @xmath58 for @xmath59 and @xmath60 ( @xmath61 @xmath62 ) . furthermore we impose the following conditions on them . @xmath63 at each times step @xmath5 , we consider the boxes with indices from @xmath64 to @xmath65 . in other words , we neglect all the balls in the boxes @xmath66 and @xmath67 . then , as in the case of the bbs with box capacity one , we draw arclines from balls to vacancies to determine the values of the conserved quantities . ( see figure [ fig05 ] . ) * draw an arcline from one of the leftmost balls to one of its nearest right vacancies . * draw an arcline from one of the leftmost balls which are located on the right of the vacancy we have drawn the arcline to one of its nearest right vacancies . * repeat the above procedure until one has drawn arclines between all the pairs of balls and vacancies . note that if we drew an arcline to a vacancy at @xmath4th box , then the ball from which we draw an arcline in the next step is located in the box whose index is greater than @xmath4 , and that there must be no arcline starting from a ball in the last box due to the condition 3 . let @xmath68 be the number of arclines drawn in this step . * then neglecting all the boxes and vacancies which are connected by arclines , repeat the above procedure and let @xmath69 be the number of arclines drawn in the second step . * repeat the above procedure until all the balls but those in the @xmath65th box are connected and let @xmath70 ( @xmath71 ) be the number of arclines drawn in the @xmath72th steps . * let @xmath73 ( @xmath74 and @xmath75 @xmath76 . note that we obtain the updated state of the bbs with carrier capacity @xmath30 if we exchange the balls with the vacancies connected to them up to the @xmath30th steps , i.e. , if we exchange the first @xmath77 pairs of balls and vacancies . then we find the following proposition . [ prop2 - 1 ] the sequence ( @xmath78 ) does not depend on @xmath5 . and @xmath79 . from this figure , we have @xmath80 and @xmath81 . since @xmath82 , @xmath83 and @xmath84 , we obtain @xmath85 in both time steps . ] the statement of the proposition is the direct consequence of the fact that @xmath86 is equal to the number of balls moved by the carrier with capacity @xmath87 at one time step which is the value of @xmath87th energy function@xcite . here we map the system to the bbs with box capacity @xmath88 in which the number of balls is finite , that is , the @xmath4th box is vacant for @xmath59 and @xmath89 . of course we can make another choice we could impose the condition @xmath90 ( mod @xmath56 ) or the condition that a vacancy in the @xmath49th box is connected by an arcline in the first step . this redundancy is caused by the fact that there are essentially four types of finite systems which are equivalent to this system . once we determine the type however , the sequence @xmath91 does not depend on @xmath65 and @xmath64 as long as they are sufficiently large and , when there is no negative soliton @xmath92 @xmath93 , where @xmath94 is the @xmath72th conserved quantity defined in the original bbs . for the original bbs where every box capacity is one and only a finite number of balls exist , all the solutions are proved to be constructed by ultradiscretization from the soliton solutions of the discrete kdv equation @xcite . although the ultradiscrete kdv equation over arbitrary integers is shown to be equivalent to the bbs with larger box capacity , the solutions to them can not be the ultradiscrete limit of soliton solutions to , because the boundary condition of the bbs changes such that the number of balls becomes @xmath95 for @xmath96 . nevertheless we shall show that solutions can be constructed from the soliton solutions to through ultradiscretization with an appropriate scaling limit . here we treate only the case @xmath82 . our method for constructing the solutions is equally applicable to the case @xmath97 , although the analysis of that case is more elaborate . the coupled equations for the bbs are explicitly written as @xmath98 , \label{oureq1 } \\ v_{n+1}^t&=u_n^t+v_n^t - u_n^{t+1 } , \label{oureq2}\end{aligned}\ ] ] with the boundary condition @xmath99 from and , if we put @xmath100 ( @xmath101 ) and assume that @xmath102 exists , we find @xmath103 . \label{udkdvbilinear}\ ] ] from , if we put @xmath104 they satisfy and for @xmath105 . hence we will construct solutions @xmath106 of which satisfy the boundary conditions . first let us consider a state where only negative solitons exist . the corresponding solution is called a background solution @xcite . it is a stationary solution of the form @xmath107 where @xmath108 is a @xmath109 sequence with only finite number of @xmath1s . let the number of @xmath1s be @xmath110 ( @xmath111 ) , and let @xmath112 , @xmath113 ( @xmath114 ) and @xmath115 @xmath116 . we define a @xmath109 sequence @xmath117 as @xmath118 apparently any background solution @xmath119 can be expressed as @xmath120 for some integers @xmath121 and @xmath122 . the @xmath123 soliton solution to is given as @xmath124}\atop{j \ne \phi}}\left ( \prod_{i \in j } ( -\gamma_i)\left ( \frac{1-p_i}{p_i}\right)^t \left ( \frac{p_i+\delta}{1+\delta - p_i}\right)^n\right)\frac{\prod_{i > j , i , j\in j}(p_i - p_j)^2}{\prod_{i , j \in j}(1-p_i - p_j)},\ ] ] where @xmath125 and @xmath126 @xmath127 are arbitrary but @xmath128 for @xmath129@xcite . we often denote @xmath130:=\{1,2,3,\ldots , k\}$ ] for a positive integer @xmath87 . to take the ultradiscrete limit , let @xmath100 , @xmath131 , @xmath132 @xmath133 where @xmath134 and @xmath135 for @xmath129 . then taking the limit @xmath136 , we have @xmath137}\atop{j \ne \phi } } \left[\sum_{i \in j}(c_i+tp_i - n\min[p_i , l ] ) -\sum_{{i , j\in j}\atop{i \ne j } } \min[p_i , p_j ] \right ] \right\}. \label{rhomax}\ ] ] for @xmath138 , we find that @xmath3 and @xmath35 satisfy and , and hence . in particular , if @xmath139 , gives an @xmath123 soliton solution of the original bbs . we are , however , concerned with eqs . and with the boundary condition . hereafter we put @xmath105 and prepare several symbols and notations : @xmath140 we also assume that @xmath141 for @xmath142 ( @xmath143 ) , and @xmath144 . let us consider an @xmath145 soliton solution @xmath146 of the form . we assume that @xmath147 ( @xmath148 ) and @xmath149 @xmath150 . the phases are chosen as @xmath151 then we find @xmath152\\ & = \max_{{-n_0\le k_1 \le n_{\infty}-l'-n_0}\atop{0\le k_2 \le l ' } } \left [ -(2k_1+k_2)(n - t+2n_0 ) + { \sum_{i=1}^{k_1}}'\bar{c}_{i}\right.\\ & \qquad \left.+\sum_{i=1}^{k_2}c_i'-2k_1(k_1 - 1)-k_2(k_2 - 1)-2k_1k_2 \right]\\ & \qquad \qquad \qquad \qquad \qquad -2n_0(n - t+2n_0)+2n_0 ^ 2 + 3n_0,\end{aligned}\ ] ] where @xmath153 it is obvious that if @xmath106 is a solution to , @xmath154 is also a solution for an arbitrary function @xmath155 and arbitrary values @xmath156 and @xmath157 . hence we find that @xmath158 is a solution to , where @xmath159.\end{aligned}\ ] ] since @xmath160 are large but arbitrary positive integers , the limit @xmath161 \label{rhoforiginal}\end{aligned}\ ] ] is also a solution to . when we rewrite in as @xmath162\right]\ ] ] and using the relations such as @xmath163 = \max[-2x+a_2,a_0]\ ] ] eq . can be simplified as shown in the proposition [ prop1 ] below . first let us define some symbols and notations . s_1,\quad i : = [ 2k - l]\setminus \bar{i } \\ \bar{i}&:=\left\{m \in [ 2k - l]\big\| m=2b_{i-1}+\delta_{i-1}+2k_i+1,\ k_i=0,1,\ldots , b_{i-1}-1 , \ i \in s_0 \right\}\end{aligned}\ ] ] the quantity @xmath165 is defined for @xmath166 as * if @xmath167 and @xmath168 , then putting @xmath169 @xmath170 , @xmath171 + ( 2i-1)k \nonumber \\ & \quad + ( b_{i-1}+k+\delta_{i-1})(b_{i-1}+k+\delta_{i-1}-1)+(b_{i-1}+k)(b_{i-1}+k-1 ) . \label{theta1}\end{aligned}\ ] ] * if @xmath172 , @xmath173 and @xmath169 @xmath170 , then @xmath165 is given by . * if @xmath172 , @xmath173 and @xmath174 @xmath170 , then @xmath175 [ prop1 ] @xmath176 , \label{eqrho}\ ] ] where @xmath177 , \label{rho_minus}\\ \rho_f^{(+)}(x)&:=\left\ { \begin{array}{l } \dis\max_{{q \ge 1}\atop{q\in\mathbb{z}}}\left [ -(2k - l+2q)x - q(2q+4k-2l+2s-1)-q_0 \right ] \quad ( \delta_{s+1}=0 ) , \\ \dis \max_{{q \ge 1}\atop{q\in\mathbb{z}}}\left [ -(2k - l+q)x-(q-1)\left(\frac{q}{2}+2k - l+s\right)-s - q_0 \right]\quad ( \delta_{s+1}=1 ) , \end{array } \right . \label{rho_pluse}\\ \rho_f^{(0)}(x)&:=\max_{{1 \le r \le 2k - l}\atop{r\in\z}}\left [ \rho^{(r)}(x ) \right ] , \label{rho_zero}\end{aligned}\ ] ] and the function @xmath178 is given as @xmath179 from , if we define @xmath180 @xmath181 gives a solution to and . by using the fact that @xmath182 is a continuous and convex piecewise linear function , and by calculating the intersection points of each linear functions appeared as linear pieces in the function @xmath182 , we obtain the explicit form of @xmath183 . let us define @xmath184 @xmath185 as @xmath186 by straightforward but a little tedious calculation , we find @xmath187 where @xmath188 an example of the function @xmath183 is shown in figure [ fig06 ] . comparing these expressions with @xmath189 , we find that @xmath190 for @xmath191 . thus we have proved the following proposition . it holds that @xmath192 for @xmath191 . [ prop2 ] and @xmath183 . the parameters are @xmath193 . ] furthermore , comparing @xmath189 with the conserved quantities @xmath194 defined in the previous section , we have the following proposition . [ conservation ] the number of negative solitons is equal to @xmath195 , i.e. , @xmath196 , and @xmath197 is the number of solitons with amplitude @xmath2 which is used to construct the background solution , i.e. , @xmath198 . next we consider the solutions in which @xmath199 ordinary solitons exist in a background state . since @xmath123 soliton solutions are given by and the parameters for the background solution are already known , we have only to add new parameters @xmath200 @xmath201 and @xmath202 . when we denote the set of indices for a background solution @xmath203 , then the solution is given by @xmath204 , j_2 \subseteq \mathcal{b}}\left[\sum_{i \in j_1}(-\min[3,\hat{p}_i]n+\hat{p}_it+\hat{c}_i ) + \sum_{i \in j_2}(-p_i(n - t)+c_i ) \right.\\ & \qquad -\sum_{i\ne j , i , j \in j_1}\min[\hat{p}_i,\hat{p}_j ] -\sum_{i\ne j , i , j \in j_2}\min[p_i , p_j ] -2 \sum_{i \in j_1 , j\in j_2 } \min[\hat{p}_i , p_j]\biggr]\\ & = \max_{j_1 \subseteq [ m ] , j_2 \subseteq \mathcal{b}}\left[\sum_{i \in j_1}(-3n+\hat{p}_it+\hat{c}_i ) + \sum_{i \in j_2}(-p_i(n - t)+c_i ) \right.\\ & \qquad -\sum_{i\ne j , i , j \in j_1}\min[\hat{p}_i,\hat{p}_j ] -\sum_{i\ne j , i , j \in j_2}\min[p_i , p_j ] -2 \sum _ { j\in j_2 } \|j_1\|p_j\biggr]\\ & = \max_{j_1 \subseteq [ m]}\left [ \sum_{i \in j_1}(-3n+\hat{p}_it+\hat{c}_i ) -\sum_{i\ne j , i , j \in j_1}\min[\hat{p}_i,\hat{p}_j ] \right.\\ & \qquad+\max_{j_2 \subseteq \mathcal{b}}\left.\left [ \sum_{i \in j_2}(-p_i(n - t+2\|j_1\|)+c_i ) -\sum_{i\ne j , i , j \in j_2}\min[p_i , p_j ] \right ] \right ] . \end{aligned}\ ] ] here we used the abbreviation @xmath205 \equiv \max\left\ { 0 , \max_{j \subseteq \mathcal{a } , j \ne \phi}\left[\cdots \right]\right\}.\ ] ] therefore we obtain the following theorem . [ th1 ] the solution to representing the interaction of @xmath199 soliton solutions with the background given by , has the expression : @xmath206}\left [ \sum_{i \in j}(-3n+\hat{p}_it+\hat{c}_i ) -\sum_{{i\ne j}\atop{i , j \in j}}\min[\hat{p}_i,\hat{p}_j ] + \rho_f(n - t+2\|j\| ) \right ] . \label{lastsolution}\ ] ] the multi - soliton solutions with a background is also given by y. nakata in an algebraic form ( nested maximization ) with the ultaradiscrete vertex operators@xcite . so far we do not find the exact correspondence between our solution and his formula . from , we find that the phase shift of negative solitons after collision of an ordinary soliton is @xmath207 . an example of a time evolution pattern is shown in figure [ fig07 ] . since the solution shows that the ordinary solitons move freely for @xmath208 as those in the bbs with box capacity one , comparing with the construction of conserved quantities , we find the following proposition@xcite . for a solution to determined by , let @xmath209 be the young diagram corresponding to the partition @xmath210 constituted by the conserved quantities , that is , the length of its @xmath211th column of @xmath209 is @xmath212 . then , the length of @xmath72th row in @xmath209 is equal to @xmath213 . given by . here the symbol ` . ' denotes ` 1 ' . the parameters are @xmath215 . when we make gage transformation as @xmath216 , this time evolution pattern coincides with that of figure [ fig04 ] . ] we constructed the conserved quantities of the ultradiscrete kdv equation with negative solitons by using the transformation to the bbs with larger box capacities . the solutions to in which negative solitons and ordinary solitons coexist are also obtained as the solutions to this extended bbs . in the bbs with box capacity one , every state is obtained by ultradiscretization of a soliton solution of the discrete kdv equation , and using this fact , initial value problem is solved with elementary combinatorial methods . we conjecture that this fact also holds for the extended bbs and can be used to solve the initial value problems . investigating this conjecture is a problem we wish to address in the future . the authors wish to thank professors atsushi nagai and ralph willox for useful discussions and comments . 99 d. takahashi and j. satsuma 1990 `` a soliton cellular automaton '' , _ j. phys . jpn . _ * 59 * , 35143519 d. takahashi 1993 `` on some soliton systems defined by using boxes and balls '' , _ proc . int . symp . on nonlinear theory and its applications _ ( nolta93 ) , 555558 m. torii , d. takahashi and j. satsuma 1990 `` combinatorial representation of invariants of a soliton cellular automaton '' , _ physica _ * d92 * , 209220 t. tokihiro , d. takahashi , j. matsukidaira and j. satsuma 1996 `` from soliton equations to integrable cellular automata through a limiting procedure '' , _ phys . * 76 * , 32473250 r. hirota , s. tsujimoto and t. imai 1992 `` difference scheme of soliton equations '' , _ future directions of nonlinear dynamics in physical and biological systems nato adv . b : phys . _ * 312 * , _ plenum _ , 715 r. hirota 2009 `` new solutions to the ultradiscrete soliton equations '' , _ stud . appl . math . _ * 122 * , 361376 r. willox , y. nakata , j. satsuma , a. ramani and b. grammaticos 2010 `` solving the ultradiscrete kdv equation '' , _ j. phys . a : math . theor . _ * 43 * , 482003 , 7pp d. takahashi and j. matsukidaira 1997 `` box and ball system with a carrier and ultradiscrete modified kdv equation '' , _ j. phys . a : math . _ * 30 * , l733739 m. murata , s. isojima , a. nobe and j. satsuma 2006 `` exact solutions for discrete and ultradiscrete modified kdv equations and their relation to box - ball systems '' , _ j. phys . a : math . gen . _ * 39 * , l2734 a. nakayashiki and y. yamada 1997 `` kostka polynomials and energy functions in solvable lattice models '' , _ selecta mathematica , new ser . _ * 3 * , 547599 k. fukuda , m. okado and y. yamada 2000 `` energy functions in box ball systems '' , _ int . * 15 * , 13791392 g. hatayama , k. hikami , r. inoue , a. kuniba , t. takagi and t. tokihiro 2001 `` the @xmath217 automata related to crystals of symmetric tensors '' , _ j. math . phys . _ * 42 * , 274308 this transformation was independently found out by c. gilson , a. nagai and j. nimmo . ( private communication ) . j. mada , m. idzumi and t. tokihiro 2005 `` path description of conserved quantities of generalized periodic box - ball systems '' , _ j. math . phys . _ * 46 * , 022701 , 19pp k. fukuda 2004 `` box - ball systems and robinson - schensted - knuth correspondence '' , _ j. algebraic combin . _ * 19 * , 6789 j. mada , m. idzumi and t. tokihiro 2008 `` the box ball system and the @xmath123-soliton solution of the ultradiscrete kdv equation '' , _ j. phys . a : math . theor . _ * 41 * , 175207 , 23pp d. yoshihara , f. yura and t. tokihiro 2003 `` fundamental cycle of a periodic box - ball system '' , _ j. phys . a : math . gen . _ * 36 * , 99121 y. nakata 2010 `` vertex operators and background solutions for ultradiscrete soliton equations '' , _ doctor thesis , university of tokyo _	we construct generalized solutions to the ultradiscrete kdv equation , including the so - called negative solition solutions . the method is based on the ultradiscretization of soliton solutions to the discrete kdv equation with gauge transformation . the conserved quantities of the ultradiscrete kdv equation are shown to be constructed in a similar way to those for the box - ball system .
the lemniscate is the curve defined by the polar equation @xmath0 : @xmath1{fig1}\ ] ] in the first quadrant , the arc length @xmath2 is related to the radial distance @xmath3 by the elliptic integral @xmath4 the arc length of the first - quadrant portion is @xmath5 , which we denote by @xmath6 in analogy with @xmath7 . thus @xmath8 is the length of the entire lemniscate . following abel , the inverse function of is denoted @xmath9 . the goal of this paper is to compute the galois group @xmath10 where @xmath11 and @xmath12 is a positive odd integer . geometrically , @xmath13 tells us how to find the first @xmath12-division point of the lemniscate , and the size of the galois group determines whether or not we can divide the lemniscate into @xmath12 pieces by ruler and compass . this will lead to a quick proof of abel s theorem on the lemniscate , which characterizes those @xmath12 s for which this can be done ( see section [ abelthmsec ] for a precise statement ) . our main result is an isomorphism @xmath14/n\mathbb{z}[i]\big)^\times\ ] ] when @xmath15 is odd . we will give two proofs : * the first proof uses class field theory and complex multiplication . the key step is to show that @xmath16 is the ray class field of @xmath17 for the modulus @xmath18 . * the second proof is more elementary and uses _ lemnatomic polynomials _ , which are analogs of cyclotomic polynomials . the key step is to prove that lemnatomic polynomials are irreducible . here is a brief summary of the paper . section [ lemfnsec ] explains how @xmath19 extends to an elliptic function @xmath20 and describes the period lattice @xmath21 and associated elliptic curve @xmath22 . we also recall how complex multiplication by @xmath23 $ ] gives explicit formulas for @xmath24 when @xmath25 $ ] . section [ prelimsec ] introduces the field @xmath26 for @xmath27}}$ ] relatively prime to @xmath28 ( we say that @xmath29 is _ odd _ ) and constructs an injection @xmath30/\beta{{\mathbb{z}}}[i])^\times.\ ] ] we also give some alternate descriptions of @xmath31 that clarify its relation to the elliptic curve @xmath22 . in section [ cftsec ] we identify @xmath31 as the ray class field of @xmath32 for the modulus @xmath33 when @xmath34 $ ] is odd and use class field theory and complex multiplication to prove that is an isomorphism . this gives our first proof of . section [ lempolysec ] defines the lemnatomic polynomial @xmath35 and proves its irreducibility over @xmath32 by an elementary argument , leading to our second proof of . we also determine the degree and constant term of @xmath35 . section [ abelthmsec ] shows how abel s theorem on the lemniscate follows from the irreducibility of lemnatomic polynomials , and the final section [ chebysec ] explores a surprisingly strong analogy between lemnatomic polynomials and irreducible factors of chebyshev polynomials . as explained in ( * ? ? ? * ch . 15 ) , the function @xmath19 from the introduction extends to a function of period @xmath8 on @xmath36 and satisfies the addition law @xmath37 and the differential equation @xmath38 to extend @xmath39 to @xmath40 , note that the integral suggests defining @xmath41 . this and enable us to define @xmath42 by ( * ? ? ? 15.3 ) , @xmath43 is an elliptic function for the lattice @xmath44 whose zeros ( all simple ) are at points @xmath45 and whose poles ( also simple ) are at points @xmath46 . furthermore , @xmath43 is an odd function and @xmath47 for the rest of the paper , we will use the notation @xmath48}}.\ ] ] earlier we defined @xmath49 to be _ odd _ if it is relatively prime to @xmath28 . note that @xmath50 we say that @xmath49 is _ even _ if it is not odd , i.e. , if @xmath28 divides @xmath29 . if @xmath49 is nonzero and @xmath21 is the period lattice , then note that @xmath51 as @xmath52-modules . we say that @xmath53 _ is a @xmath29-torsion generator _ if @xmath54 \in \frac1\beta l /l$ ] generates @xmath55 as an @xmath52-module . any two @xmath29-torsion generators @xmath56 satisfy @xmath57 for some @xmath58 \in ( { { \mathcal{o}}}/\beta{{\mathcal{o}}})^\times$ ] . we will use the @xmath29-torsion generator @xmath59 frequently . it is clear that @xmath43 has complex multiplication by @xmath52 . here is a precise description of what this means . [ phipropsthm ] let @xmath60 be odd . then there exist relatively primepolynomials @xmath61 $ ] and @xmath62 such that 1 . for all @xmath63 , @xmath64 , where @xmath65 2 . 3 . @xmath67 and @xmath68 have degree @xmath69 , where @xmath70 is the norm of @xmath29 . the * -division polynomial * @xmath71 has @xmath72 distinct roots given by @xmath73 for @xmath74 \in { { \mathcal{o}}}/\beta { { \mathcal{o}}}$ ] and @xmath75 a fixed @xmath29-torsion generator . 5 . @xmath67 is monic and @xmath76 . suppose @xmath77 is an odd prime in @xmath52 and let @xmath78 . then @xmath79 such that each @xmath80 is divisible by @xmath77 and @xmath81 . parts ( 1 ) , ( 2 ) , ( 3 ) and ( 5 ) are proved in ( * ? ? ? 15.4.4 ) , and part ( 6 ) ( due to eisenstein ) is proved in ( * ? ? ? 15.4.8 ) . as for part ( 4 ) , let @xmath82 . for any @xmath83 , @xmath84 is a root of @xmath71 since @xmath43 vanishes at @xmath85 . now suppose that @xmath86 satisfy @xmath87 . by , we have @xmath88 this implies @xmath89 , so that @xmath90 is even . it follows easily that @xmath91 . thus @xmath71 has at least latexmath:[$\mathrm{n}(\beta ) = @xmath84 for @xmath58 \in { { \mathcal{o}}}/\beta{{\mathcal{o}}}$ ] . since this polynomial has degree @xmath72 , these are all of its roots . then the same holds for any other @xmath29-torsion generator @xmath93 since @xmath94 for some @xmath58 \in ( { { \mathcal{o}}}/\beta{{\mathcal{o}}})^\times$ ] . there are formulas for @xmath95 when @xmath29 is even , but they involve both @xmath96 and @xmath97 . an example is the duplication law @xmath98 we also note that differentiating the formula for @xmath95 from theorem [ phipropsthm ] leads to a formula for @xmath99 as a rational function of @xmath97 . we conclude this section with a few words about the elliptic curve @xmath100 associated to the period lattice @xmath101 from . [ eweq ] the weierstrass equation of @xmath22 is @xmath102 . the weierstrass equation of @xmath103 is @xmath104 , and @xmath105 since @xmath22 has complex multiplication by @xmath106}}$ ] . as for @xmath107 , we know from @xcite that the lattice @xmath108 has @xmath109 . since @xmath110 , we obtain @xmath111 hence @xmath112 , which gives the desired weierstrass equation . let @xmath113 be the weierstrass @xmath114-function for @xmath21 . the elliptic functions @xmath43 and @xmath115 have period lattice @xmath21 and hence are rational functions of @xmath114 and @xmath116 . by analyzing the behavior at zeros and poles , one can show that @xmath117 to see what this means geometrically , note that by , @xmath43 and @xmath115 parametrize the curve @xmath118 while @xmath114 and @xmath116 parametrize the curve of lemma [ eweq ] . then tells us that the curves of lemma [ eweq ] and are related by the birational transformation @xmath119 with inverse @xmath120 , @xmath121 . we begin with the field @xmath122 defined in the introduction . [ fieldkbeta ] if @xmath49 is odd and @xmath93 is any @xmath29-torsion generator , then @xmath123),\ ] ] where @xmath124)$ ] is the field obtained from @xmath32 by adjoining the coordinates of the @xmath29-torsion points of @xmath22 . we first show that @xmath125 is a @xmath29-torsion generator . since @xmath29 is odd , we have @xmath126 for some @xmath127 . multiplying this by @xmath59 gives @xmath128 , and our claim follows . to prove the first equality of the proposition , let @xmath56 be @xmath29-torsion generators . as noted above , @xmath57 for some @xmath58 \in ( { { \mathcal{o}}}/\beta{{\mathcal{o}}})^\times$ ] , where we may assume that @xmath129 is odd since @xmath29 is ( if @xmath129 is even , replace it with @xmath130 . ) then @xmath131 and the latter is in @xmath132 by part ( 1 ) of theorem [ phipropsthm ] . this implies @xmath133 , and then the first equality follows using @xmath134 . for the second equality , we use to obtain @xmath135 this implies @xmath136 , which shows that the second equality holds for @xmath137 . then the addition laws for @xmath43 and @xmath115 show that it also holds for any @xmath29-torsion generator @xmath93 . the third equality follows since is birational , and the final equality follows since @xmath138 $ ] generates @xmath139 $ ] as an @xmath52-module . the final equality of proposition [ fieldkbeta ] shows that @xmath31 is a galois extension of @xmath32 ( this also follows from theorem [ phipropsthm ] ) . here is a preliminary result about the galois group of @xmath140 . [ preliminary ] let @xmath49 be odd . then for any @xmath141 , there is a unique @xmath58 \in ( { { \mathcal{o}}}/\beta{{\mathcal{o}}})^\times$ ] such that @xmath142 for any @xmath29-torsion generator @xmath93 . furthermore , the map @xmath143 $ ] defines an injective homomorphism @xmath144 a similar result is proved in ( * ? ? ? the proposition here follows by essentially the same argument . we omit the details . our eventual goal is to prove that the map of proposition [ preliminary ] is an isomorphism . using class field theory , we provide the first proof of our main result . [ cftthm ] let @xmath60 be odd . then @xmath31 is the ray class field of @xmath32 for the modulus @xmath33 . furthermore , the map of proposition [ preliminary ] is an isomorphism @xmath145 let @xmath146 be odd . the weierstrass equation of @xmath22 has @xmath147 and @xmath148 , so that the weber function of @xmath22 is given by @xmath149 . since @xmath150 is a @xmath33-torsion generator , the theory of complex multiplication ( see ( * ? ? ? 5.6 ) , for example ) tells us that @xmath151 is the ray class field of @xmath32 for the modulus @xmath33 . furthermore , since @xmath52 is a pid , the artin map induces an isomorphism @xmath152 however , since @xmath29 is odd , we have @xmath153 where the last isomorphism follows because @xmath154 maps isomorphically onto @xmath155 . it follows that @xmath156 using the relation between @xmath115 and @xmath157 , we can write the ray class field @xmath21 as @xmath158 we prove that @xmath159 as follows . first observe that @xmath160 . now consider the identity @xmath161 since @xmath162 and @xmath163 , follows from the addition law for @xmath164 by differentiation . since @xmath29 is odd , there exists a @xmath165 such that @xmath166 for @xmath167 . then @xmath168 . using @xmath169 and , we obtain @xmath170 it follows that @xmath171 . since these are galois extensions , we have @xmath172 combining this with the injection of proposition [ preliminary ] , we conclude that @xmath173 and that the injection of proposition [ preliminary ] is an isomorphism . setting @xmath174 in theorem [ cftthm ] gives the following corollary . if @xmath12 is a positive odd integer , then @xmath175 is a galois extension of @xmath32 with galois group @xmath176 furthermore , @xmath177 is the ray class field of @xmath32 for the modulus @xmath18 . this is the class field theory proof of the isomorphism from the introduction . in 1980 , rosen @xcite applied class field theory to the lemniscate . he used the lattice @xmath178 , which corresponds to the elliptic curve @xmath179 defined by @xmath180 . in order to prove abel s theorem on the lemniscate ( see theorem [ abel ] below ) , he used the isomorphism @xmath181 since @xmath182 is the ray class field of @xmath32 for the modulus @xmath12 . in rosen s approach , @xmath12 can be any positive integer . the first person to apply class field theory to the lemniscate was takagi in his 1903 thesis , where he showed that all abelian extensions of @xmath11 are generated by division values of the lemniscatic elliptic function ( see schappacher @xcite for precise references ) . theorem [ cftthm ] for the case when @xmath29 is an odd prime power was first proved by lemmermeyer in ( * ? ? ? * thm . 8.19 ) . in this section we present the second proof of our main result . pursuing the analogy between the roots of unity and the division points of the lemniscate leads to an algebraic theory for the lemniscate akin to the circle s cyclotomy . we begin by developing the foundations of the theory , introducing the _ lemnatomic polynomials _ , and finally proving their irreducibility over @xmath32 . our main result follows as a corollary . here is the key definition of this section . [ lempolydef ] let @xmath60 be odd and set @xmath59 . we call @xmath183\in ( { { \mathcal{o}}}/\beta { { \mathcal{o}}})^\times } \!\!\!\ ! { ( x-{\varphi}(\alpha{\delta_\beta}))}\ ] ] the _ * lemnatomic polynomial_. in this definition , we can replace @xmath137 with any @xmath29-torsion generator @xmath93 . furthermore , the roots of @xmath184 are @xmath185 as @xmath93 ranges over all lattice - inequivalent @xmath29-torsion generators . note also that @xmath186 when @xmath29 and @xmath187 are associates in @xmath52 since @xmath184 depends only on the ideal generated by @xmath29 . it will often be convenient to specify an associate . recall that an odd @xmath49 satisfies @xmath188 for @xmath189 . an odd element @xmath146 is _ normalized * _ if @xmath190 . [ features ] let @xmath184 be as above . then @xmath191 $ ] is monic of degree @xmath192 . all that requires demonstration is @xmath191 $ ] . by proposition [ preliminary ] , the roots of @xmath184 are permuted under the action of @xmath193 and hence the coefficients lie in the fixed field @xmath32 of the galois group . the roots of @xmath184 are algebraic integers since they are also roots of the monic polynomial @xmath194 $ ] . it follows that @xmath195 $ ] . as cyclotomic polynomials provide a factorization of @xmath196 over @xmath197 , the lemnatomic polynomials provide a factorization of @xmath71 over @xmath32 . [ decomp ] let @xmath60 be odd and let @xmath71 be the @xmath29-division polynomial from theorem [ phipropsthm ] . then @xmath198 where the product is over all normalized divisors @xmath199 of @xmath29 . from theorem [ phipropsthm ] , the roots of @xmath71 are @xmath200 for @xmath58\in{{\mathcal{o}}}/\beta{{\mathcal{o}}}$ ] . the proof consists of using gcd s to reorganize the roots as in the analogous cyclotomic result . we refer the reader to ( * ? ? ? . we can also determine the constant term of @xmath184 . [ constantterm ] let @xmath146 be odd and a nonunit . if @xmath201 where @xmath202 , @xmath77 is a normalized prime , and @xmath203 , then @xmath204 . in all other cases , @xmath205 . we may assume that @xmath29 is odd and normalized ( this implies @xmath206 in part ( 1 ) of theorem [ phipropsthm ] . ) using @xmath207 and lhspital s rule , we have @xmath208 since @xmath209 by part ( 5 ) of theorem [ phipropsthm ] . now let @xmath77 be a normalized odd prime . if @xmath210 , then proposition [ decomp ] tells us that @xmath211 . therefore , @xmath212 by . next , if @xmath213 , then we proceed by strong induction . suppose that for all @xmath214 , we have @xmath215 . from proposition [ decomp ] and our induction hypothesis we obtain @xmath216 we conclude that @xmath217 , completing our induction . finally , suppose @xmath29 is not a prime power . again , proposition [ decomp ] tells us that @xmath218 each prime power @xmath219 dividing @xmath29 , where @xmath77 is a normalized prime , contributes @xmath77 to the product ( [ eqn6 ] ) . thus the normalized prime power divisors of @xmath29 contribute a factor of @xmath29 to ( [ eqn6 ] ) . therefore @xmath220 using this and strong induction on the number of prime factors of @xmath29 ( counted with multiplicity ) , it is now easy to prove that @xmath205 when @xmath29 is not a prime power . we leave the details to the reader . the inspiration for proposition [ constantterm ] derives from a similar result for irreducible factors of chebyshev polynomials proved in ( * ? ? ? * prop . 2 ) . we will explore the analogy between lemnatomic polynomials and chebyshev polynomials in section [ chebysec ] . we now come to the main result of this section . [ main ] if @xmath60 is odd , then the lemnatomic polynomial @xmath184 is irreducible over @xmath32 . we will follow the strategy used in ( * ? ? ? 9.1.9 ) to prove that the cyclotomic polynomial @xmath221 is irreducible over @xmath197 . since @xmath222 $ ] is monic , gauss s lemma implies that an irreducible factor @xmath223 of @xmath184 in @xmath224 $ ] lies in @xmath225 $ ] , and we can assume that @xmath223 is also monic . the roots of @xmath223 have the form @xmath226 for @xmath227 with @xmath228 and @xmath229 for some @xmath3 . for @xmath230 an odd prime , consider the polynomial @xmath231 recall that @xmath71 is separable by theorem [ phipropsthm ] . it follows that @xmath184 is separable over @xmath32 and hence is separable modulo any prime of @xmath52 not dividing @xmath232 . for our purposes , we assume that the prime @xmath233 satisfies 1 . @xmath234 . 2 . @xmath184 is separable modulo @xmath77 . 3 . @xmath235 . first , we show that @xmath236 $ ] . let @xmath237 and take a root @xmath238 of @xmath239 . since @xmath240 and @xmath241 are @xmath29-torsion generators , proposition [ preliminary ] implies that there is @xmath242 relatively prime to @xmath29 such that @xmath243 since @xmath226 is a root of @xmath244 $ ] , the first line of implies that @xmath245 is a root as well . then @xmath246 is a root of @xmath239 by definition . but this root is @xmath247 by the second line of . it follows that @xmath248 permutes the roots of @xmath239 , and then @xmath249 $ ] follows easily . we next prove the following claim : @xmath250 suppose is false . since @xmath223 and @xmath239 are monic of the same degree and @xmath223 is irreducible , @xmath251 implies that @xmath223 is coprime to @xmath239 . the roots of @xmath239 are also roots of @xmath184 , so there exists a monic @xmath252 $ ] such that @xmath253 . let us analyze the roots of @xmath254 . from parts ( 5 ) and ( 6 ) of theorem [ phipropsthm ] , we have @xmath255 where @xmath77 divides all the coefficients @xmath80 . let @xmath256 be the ring of integers of @xmath31 and let @xmath257 be a prime ideal of @xmath256 dividing @xmath258 and suppose that @xmath226 is a root of @xmath223 . therefore @xmath259 in @xmath31 . by ( [ candoit ] ) , we may reduce this equality modulo @xmath257 to get @xmath260 let @xmath261 then an argument similar to that given for @xmath239 shows that @xmath262 $ ] . by , @xmath263 and @xmath264 have the same roots in @xmath265 and hence @xmath266.\ ] ] however , @xmath267 is obtained from @xmath223 by raising its roots to the @xmath268 power , and a standard argument implies that @xmath269.\ ] ] ( see ( * ? ? ? 9.1.8 ) for the case of a prime @xmath270 . the proof for a prime @xmath233 is identical . ) combining these congruences shows that @xmath271 $ ] . therefore , @xmath272},\ ] ] which is to say that @xmath184 is not separable modulo @xmath77 . this contradicts our choice of @xmath77 and allows us to conclude . now consider a root @xmath226 of @xmath223 . let @xmath273 be the product of all odd primes dividing @xmath274 but not dividing @xmath29 . if @xmath275 is any root of @xmath184 , then the chinese remainder theorem implies that there is a gaussian integer @xmath276 such that @xmath277 therefore @xmath276 is odd and we may factor @xmath276 as @xmath278 for odd , normalized primes @xmath279 coprime to @xmath280 . iterating ( [ claim ] ) we have @xmath281 hence @xmath282 is a root of @xmath283 . thus @xmath184 and @xmath223 have the same roots ; since both are monic and separable they must be equal . we conclude that @xmath184 is irreducible over @xmath32 as desired . [ maingalois ] let @xmath60 be odd . then @xmath284 in particular , if @xmath12 is a positive , odd integer , then @xmath285 from proposition [ preliminary ] , we have an injection @xmath286 since @xmath287 , theorem [ main ] and lemma [ features ] tell us that @xmath288 = [ k({\varphi}({\delta_\beta})):k ] = \deg({{\lambda}}_\beta(x ) ) = \|({{\mathcal{o}}}/\beta { { \mathcal{o}}})^\times\|.\ ] ] therefore , our injection must be an isomorphism . the final assertion follows since @xmath175 by definition . the final assertion of corollary [ maingalois ] completes the elementary proof of the isomorphism from the introduction . let @xmath233 be an odd prime not dividing @xmath29 , so that @xmath77 is unramified in @xmath31 . under the isomorphism of corollary [ maingalois ] , some @xmath289 maps to @xmath290 \in ( { { \mathcal{o}}}/\beta{{\mathcal{o}}})^\times$ ] , where @xmath248 and @xmath77 are linked via the equation @xmath291 if we assume in addition that @xmath292 is relatively prime to the index @xmath293 $ ] , then easily implies that @xmath294 for any @xmath295 and any prime @xmath257 of @xmath256 dividing @xmath258 . since @xmath140 is abelian , the artin symbol @xmath296 is the unique element of @xmath193 with this property . hence we have proved that @xmath297 which shows that the isomorphism @xmath298 from corollary [ maingalois ] can be identified with the artin map . observe the key role played by here and in the proof of theorem [ main ] . if you look back at the proof , you will see that follows from . proving is actually the hardest part of the proof of theorem [ main ] . the proof of given in ( * ? ? ? 15.4.8 ) follows eisenstein s original argument , which is both intricate and brilliant . let us work out the case @xmath299 . the @xmath300-division polynomial is @xmath301 we can factor @xmath300 into normalized primes as @xmath302 . then @xmath303 note that the constant terms of the factors are as predicted by proposition [ constantterm ] . our work in the previous section allows us to give a new , concise proof of abel s wonderful result about ruler and compass constructions on the lemniscate . [ abel ] for a positive integer @xmath12 , the @xmath12-division points of the lemniscate may be constructed using ruler and compass if and only if @xmath304 where the @xmath305 are distinct fermat primes and @xmath306 is a nonnegative integer . we begin by reducing the equivalence . according to ( 15.1.1 ) , a point on the lemniscate is constructible if and only if its radial component is constructible . therefore , constructing the @xmath12-division points is equivalent to constructing @xmath307 for @xmath308 . we need not worry about all these points , as ( * ? ? ? 15.2.7 ) states that if @xmath309 is constructible , then @xmath307 is constructible for all @xmath310 . finally , ( * ? ? ? * prop . 15.2.3 ) allows us to conclude that @xmath96 is constructible if and only if @xmath311 is constructible . hence we may assume @xmath12 is odd . next observe that @xmath312 $ ] . this follows since lemnatomic polynomials satisfy @xmath313 with respect to complex conjugation . then @xmath314 is irreducible over @xmath197 since it is irreducible over @xmath32 by theorem [ main ] . furthermore , @xmath309 is a root of @xmath314 since @xmath315 is an @xmath12-torsion generator ( see the proof of proposition [ fieldkbeta ] ) . it follows that @xmath316\ ! = \!\deg({{\lambda}}_n)\ ! = \!\|({{\mathcal{o}}}/n{{\mathcal{o}}})^\times\|\ ! = \ ! n^2\prod_{p\mid n } \left(\!1-\frac1p\right ) \left(\!1-\left(\frac{-1}{p}\right)\frac1p\right),\ ] ] where the last equality follows from ( * ? ? ? * ex . 7.29 ) . now assume that @xmath309 is constructible . then @xmath317 $ ] is a power of @xmath318 by ( * ? ? ? 10.1.8 ) , so that by , we have @xmath319 since @xmath12 is odd , this implies that @xmath12 is a product of distinct primes @xmath320 . if @xmath321 , then @xmath322 is a power of @xmath318 , which forces @xmath323 , and if @xmath324 , then @xmath325 is a power of @xmath318 , which easily implies that @xmath320 is a fermat prime . conversely , assume that @xmath12 is a product of distinct fermat primes . observe that @xmath326 , which is galois over @xmath197 since @xmath327 is galois and @xmath328 is real . then shows that @xmath329 = 2[{{\mathbb{q}}}({\varphi}({\textstyle\frac{2{\varpi}}{n}})):{{\mathbb{q}}}]\ ] ] is a power of @xmath318 . hence @xmath309 is constructible by ( * ? ? ? . other proofs of this theorem are due to abel , eisenstein and rosen , though rosen was the first to prove the `` only if '' part of the theorem . rosen s proof @xcite uses class field theory , while the above proof uses only the irreducibility of the lemnatomic polynomial @xmath314 , which was proved without class field theory in theorem [ main ] . the proofs of abel and eisenstein are discussed in ( * ? ? ? 15.5 ) , with references to the original papers . see also the book @xcite by prasolov and solovyev . the analogy between abel s function @xmath96 and the sine function @xmath330 has been recognized since the time of gauss and abel . for example , in his unpublished work on elliptic functions , gauss wrote @xmath96 as @xmath331 ( see @xcite ) . thus the analogy between lemnatomic polynomials and cyclotomic polynomials made earlier in this paper needs to be reconsidered from the point of view of the sine function . as we now show , the analog of the division polynomial @xmath71 is the well - known _ chebyshev polynomial _ @xmath332 $ ] , which is defined by the identity @xmath333 for a positive integer @xmath12 . [ sintn ] if @xmath12 is odd , then @xmath334 furthermore , there is @xmath335 $ ] such that @xmath336 and the roots of @xmath337 are @xmath338 for @xmath339 \in { { \mathbb{z}}}/n{{\mathbb{z}}}$ ] . since @xmath12 is odd , the addition laws for @xmath340 and @xmath341 imply that @xmath342 the formula for @xmath343 implies that @xmath344 is a root of @xmath345 . since @xmath346 is an even function , it follows that @xmath347 is also even and hence is a polynomial in @xmath348 . the assertion concerning the roots of @xmath345 is equally easy and is omitted . notice how @xmath349 , @xmath12 odd , is analogous to the formula for @xmath95 , @xmath29 odd , from part ( 1 ) of theorem [ phipropsthm ] , where @xmath350 corresponds to @xmath351 . also , the crucial identity used in the proof of theorem [ cftthm ] is the lemniscatic analog of the identity @xmath352 used in the proof of lemma [ sintn ] . the analogy between @xmath20 and @xmath330 actually begins with the equations @xmath353 and @xmath354 that define the lemniscate and circle of radius @xmath355 centered at @xmath356 . table 1 shows how these curves lead naturally to the functions @xmath20 and @xmath330 and also records some of the similarities between their properties . ( see also @xcite for another view of the analogy between @xmath20 and @xmath330 . ) ' '' '' & circle + polar equation ' '' '' & @xmath353 & @xmath354 + & @xmath357{fig2}$ ] & @xmath357{fig3}$ ] + arc length ' '' '' & @xmath358 & @xmath359 + inverse function ' '' '' & @xmath360 & @xmath361 + & & + & & + ' '' '' & @xmath362 & @xmath363 + & @xmath364 & @xmath365 + ' '' '' & @xmath366 & @xmath367 + ' '' '' & @xmath368 & @xmath369 + this analogy suggests in particular that the lemnatomic polynomials @xmath184 should correspond to the irreducible factors of the chebyshev polynomial @xmath345 when @xmath12 is odd . the factors of @xmath345 were studied by hsiao @xcite in 1984 . since @xmath345 is not monic , he used the monic polynomial @xmath370.\ ] ] hsiao determines the irreducible factorization of @xmath371 over @xmath197 ( * ? ? ? * prop . 1 ) . when @xmath12 is odd , his result may be restated as a factorization @xmath372 where @xmath373 has degree @xmath374 ( the euler @xmath375-function ) and is given by @xmath376 \in ( { { \mathbb{z}}}/n{{\mathbb{z}}})^\times } \!\!\big(x - 2\sin(a{\textstyle\frac{2\pi}{n}})\big).\ ] ] note the analogy with definition [ lempolydef ] . thus , when @xmath12 is odd , is the analog of proposition [ decomp ] , where the irreducibility of @xmath373 corresponds to the irreducibility of @xmath184 proved in theorem [ main ] . although hsiao s proof of irreducibility ultimately rests on the the irreducibility of cyclotomic polynomials , we note that the polynomials @xmath373 may be shown irreducible over @xmath197 directly by adapting the proof of theorem [ main ] . we also note that in one of his unpublished papers , schur @xcite mentions very briefly ( without proof ) the irreducible decomposition of chebyshev polynomials hsiao also studied the constant term of the factors of @xmath371 . when @xmath12 is odd , his result can be stated as follows ( * ? ? ? * prop . 2 ) . [ hsiaoconstant ] let @xmath377 be odd and positive . if @xmath378 for a prime @xmath320 , then @xmath379 . otherwise , @xmath380 . by tweaking hsiao s proof , one can show that when @xmath320 is an odd prime and @xmath203 , we have @xmath381 . since @xmath382 , we see that @xmath383 is the analog of a normalized prime @xmath233 , so that proposition [ hsiaoconstant ] is the analog of proposition [ constantterm ] . in fact , proposition [ constantterm ] was inspired by hsiao s result . for further results on the factorization of chebyshev polynomials , see @xcite . there is also another interesting analogy to consider , this one involving the _ carlitz polynomials _ @xmath384(x)$ ] for @xmath385[x]$ ] . the irreducible factors of carlitz polynomials have a lot in common with lemnatomic and cyclotomic polynomials . see @xcite for more details . for more on the history and number theory associated to the lemniscate , the reader should consult schappacher s article _ some milestones of lemniscatomy _ @xcite . the results of section [ lempolysec ] are based on the senior honors thesis of the second author , written under the direction of the first author . we are grateful to amherst college for the post - baccalaureate summer research fellowship that supported the writing of this paper . we would like to thank rosario clement and keith conrad for useful conversations . the comments by keith conrad , franz lemmermeyer and michael rosen on earlier versions of the paper are greatly appreciated . n. schappacher , _ on the history of hilbert s twelfth problem : a comedy of errors _ , in _ matriaux pour lhistorie des mathmatiques au xx@xmath387 sicle ( nice , 1996 ) _ , smin . congr . * 3 * , soc . france , paris , 1998 , 243273 .	this article studies the galois groups that arise from division points of the lemniscate . we compute these galois groups two ways : first , by class field theory , and second , by proving the irreducibility of _ lemnatomic polynomials _ , which are analogs of cyclotomic polynomials . we also discuss abel s theorem on the lemniscate and explain how lemnatomic polynomials relate to chebyshev polynomials .
in the iron - based pnictide and chalcogenide superconductors , chemical doping and physical pressure are universal variables by which to tune the superconducting properties.@xmath8 for example , the critical temperature , @xmath3 , of 8 k in fese under ambient conditions@xmath9 can be raised to 38 k by externally applied pressure@xmath10 or 44 k by intercalation of cationic species.@xmath11 the tetragonal ( @xmath12 ) structure of fese ( figure 1 ) consists of sheets of edge - sharing fese@xmath13 tetrahedra held together by van der waals interactions , which makes it an ideal host for intercalation chemistry . negative pressure , or strain , has also been implicated as a parameter in the high @xmath3 of 65 k - 100 k reported for single layered fese.@xmath14 given the propensity of the fese layered system for chemical and physical manipulation , fese is an ideal platform for under - standing the superconductivity of the iron - based systems and for the preparation of new layered functional materials . the recently discovered ( li@xmath1fe@xmath2oh)fese system,@xmath15 which contains pbo - type layers of lioh alternating with the anti - pbo type layers of fese ( figure 1 ) , offers such an opportunity . iron may occupy the lithium site and therefore effectively charge dope the fese layer since the ( li@xmath1fe@xmath2oh ) layer would be positively charged . sun _ _ et al__. have also reported that increased lithiation of the ( li@xmath1fe@xmath2oh ) layer would force iron to occupy any vacancies in the fese layer , which can be detrimental to the superconducting properties.@xmath16 fe@xmath17se@xmath18 ( left ) fese ( centre ) and ( li@xmath1fe@xmath2oh)fese ( right ) ] three outstanding issues in the ( li@xmath1fe@xmath2oh)fese system and related phases are whether 1 ) the parent phase is antiferromagnetic , 2 ) superconductivity coexists with ferromagnetism , and 3 ) any isotope effects on @xmath3 exist . critical to answering all three questions is the preparation of the deuteroxide version of ( li@xmath1fe@xmath2oh)fese and comparing their phase diagrams . furthermore , hydrothermal synthesis under either h@xmath7o or d@xmath7o presents interesting differences in the purity of the resulting superconducting phases due to differences in the reaction kinetics . thus , this study will help workers in the field understand the thermodynamic and kinetic factors in the preparation of phase pure and superconducting fese - based materials . due to the large incoherent scattering of hydrogen and high neutron absorption cross section of @xmath19li , our compounds were prepared doubly isotopically pure , ( @xmath0li@xmath1fe@xmath2od)fese , which allowed for an opportunity to complete a phase diagram for the deuterated series . although some studies have found no evidence of ferromagnetism below @xmath3 in their samples,.@xmath20 under the right synthesis conditions , we have observed a ferromagnetic signal in the superconducting regime as first reported by pachmayr _ _ et al__.@xmath21 herein , we report the phase diagram for ( li@xmath1fe@xmath2od)fese , compare it to that of ( li@xmath1fe@xmath2oh)fese , and investigate the magnetic properties of the non - superconducting and superconducting samples . the preparation of powder samples were modified from a hydrothermal route reported in the literature.@xmath22 for the synthesis of deuterated samples we first prepared the doubly isotopically pure @xmath23liod as a precursor . @xmath23liod was prepared by mixing a stoichiometric amount of @xmath23lico@xmath24 ( sigma aldrich , 99% for @xmath23li ) and cao ( calcined from caco@xmath24 , sigma aldrich , 99% ) in d@xmath7o . the caco@xmath24 precipitate was filtered , and @xmath23liod crystallized by evaporation of the solvent . for a typical preparation of ( li@xmath1fe@xmath2od)fese , 5 mmol of fe powder ( alfa aesar , 99.9% ) , 6 mmol of selenourea ( sigma aldrich , 98% ) and 50 mmol of liod powder were suspended in 5 ml of distilled d@xmath7o ( oxford isotope , 99.9% ) . the mixture was placed in a teflon - lined stainless steel autoclave at 120 - 200 @xmath25c for 3 - 5 days . afterwards , the autoclave was opened in an argon - filled glove bag , and the shiny black precipitate was washed with d@xmath7o . the product was washed and centrifuged several times until the supernatant was clear . the remaining product was collected , vacuumed dried , and stored in a nitrogen - filled glove box . the yield of the product was usually between 50% and 70% . single crystal ( li@xmath1fe@xmath2od)fese and ( li@xmath1fe@xmath2oh)fese samples were prepared by replacing potassium cations with liod or lioh from k@xmath26fe@xmath27se@xmath18 single crystals under hydrothermal conditions similar to those reported by dong _ _ et al__.@xmath28 for the growth of the k@xmath26fe@xmath27se@xmath18 single crystals , 1.8 g ( 13 mmol ) of fese powder was mixed with 0.21 g ( 5.4 mmol ) of potassium metal ( alfa aesar , 99% ) to match the nominal composition of k@xmath29fe@xmath27se@xmath18.@xmath30 the fese precursor was prepared by heating fe ( alfa aesar , 99.9% ) and se ( alfa aesar , 99% ) powders to 700 @xmath25c for 5 h followed by furnace cooling ; the resulting phase does not need to be of the tetragonal @xmath31-fese form for the crystal growth . the fese / k mixtures were loaded in a quartz ampoule inside a nitrogen - filled glovebox , and the ampoules flame sealed under vacuum . in order to avoid oxidation of the samples from breaking of the ampoule due to potassium - induced corrosion of the quartz walls , the sample container was sealed in a larger ampoule . for crystal growth of k@xmath26fe@xmath27se@xmath18 , the mixture was heated to 1030 @xmath25c over 10 h and held at 1030 @xmath25c for 3 hours to form a homogeneous melt . subsequently , the melt was slowly cooled at a rate of 6 @xmath25c / hour to 650 @xmath25c to allow crystal growth . after cooling to room temperature , k@xmath26fe@xmath27se@xmath18 single crystal approximately 8 mm in diameter was recovered . in order to compare the effect of d@xmath7o to the reaction kinetics , single crystals of both ( li@xmath1fe@xmath2oh)fese and ( li@xmath1fe@xmath2od)fese were prepared under identical hydrothermal conditions . for the preparation of ( li@xmath1fe@xmath2oh)fese single crystals , the k@xmath26fe@xmath27se@xmath18 precursor ( 0.2 g - 0.4 g ) , 0.14 g ( 2.5 mmol ) fe powder , and 2 g ( 47 mmol ) lioh monohydrate were added to 5 ml water . for ( li@xmath1fe@xmath2od)fese single crystals , to match the concentration of lioh in water , 1.2 g ( 47 mmol ) liod and 6 ml d@xmath7o were used for reactions . the mixture was placed in a teflon - lined stainless steel autoclave at 120 - 200 @xmath25c for 4 - 5 days . silver colored single crystals were recovered by washing away excess powder with water and drying under vacuum overnight . the as - recovered single crystals retained similar shape to the starting k@xmath26fe@xmath27se@xmath18 single crystals . samples prepared in the absence of excess iron powder were not superconducting , which could be due to either oxidation of the iron or vacancy formation in the fese layer . to study the role of metal powders during the cation exchange reactions , experiments using sn metal ( alfa aesar , 99.9% ) instead of fe powder for the preparation of ( ( li@xmath1fe@xmath2oh)fese and ( ( li@xmath1fe@xmath2od)fese single crystals at 120 @xmath25c were carried out . sn can react with hot concentrated bases to form soluble [ sn(oh)@xmath32@xmath33 species while evolving h@xmath7 gas,@xmath34 thus providing a stronger reducing environment than the hydrothermal reactions without the presence of metal powders . li@xmath1fe@xmath2od)fese with x = 0.166 . no magnetic phase could be indexed in the npd , indicating lack of antiferromagnetic ordering . weight percent fractions from structural refinements are as follows : 98% ( @xmath0li@xmath1fe@xmath2od)fese ( gold ticks ) and 2% li@xmath7co@xmath5 ( green ticks ) . a few broad peaks corresponding to fese were fit by a pawley routine ( asterisk ) . ] room temperature powder x - ray diffraction ( pxrd ) data were collected on a bruker d8 x - ray diffractometer ( cu k radiation , @xmath35 = 1.5418 ) . data were collected with a step size of 0.02@xmath25 between 7@xmath25 and 70@xmath25 for pawley fits to extract lattice constants and 7@xmath25 and 120@xmath25 for rietveld fits to obtain better structural parameters . in order to find any possible crystallographic phase transitions that are coupled to either the superconducting or magnetization order parameters , temperature dependent ( 5 - 300 k ) high - resolution synchrotron x - ray diffraction was carried out for powders of ground single crystals at beamline 11-bm at the advanced photon source ( aps ) . in addition to finding subtle changes in symmetry due to peak splitting , the synchrotron measurements provide high-@xmath36 data and therefore more accurate structural parameters . analysis of the high-@xmath36 reflections help determine any small changes in the iron occupancies in both the fese and lioh layers , which could affect @xmath3 s of the sample as suggested by sun _ _ et al__.@xmath16 an oxford helium cryostat ( closed flow system ) was used to reach a temperature that is close to liquid helium ( @xmath37 4 k ) . ground powders of single crystals were packed in 0.4 mm kapton capillaries tubes and sealed with epoxy . diffraction data were collected between 0.5@xmath25 and 46@xmath25 with a step size of 0.0001@xmath25 using a constant wavelength = 0.413964 ( 30 kev ) . magnetic susceptibility measurements were carried out using a magnetic property measurement system ( quantum design mpms ) . both field - cooled ( fc ) and zero - field - cooled ( zfc ) magnetic susceptibility measurements were taken from 2 k - 300 k in direct current mode with an applied magnetic field of 1 or 3 mt . all the neutron work was carried out with doubly isotopically pure samples ( @xmath0li@xmath1fe@xmath38od)fese at the nist center for neutron research ( ncnr ) . the samples were loaded into he - filled vanadium cans and subsequently into a closed cycle refrigerator for low temperature measurements . low temperature ( 4 k ) diffraction data were collected on the bt-1 high - resolution npd with the cu(311 ) monochromator ( @xmath35 = 1.540 ) . in addition to base temperature measurements , we performed npd measurements at various temperatures 25 k , 75 k , 150 k , and room temperature to search for any crystallographic phase transitions . high - intensity and coarse - resolution diffraction measurements were carried out on the bt-7 spectrometer ( @xmath35 = 2.359 ) using the position sensitive detector ( psd ) to search for magnetic bragg peaks from base temperature up to 150 k.@xmath39 rietveld refinements with both xrd and npd data were carried out with the topas 4.2 software.@xmath40 representative fits to one of the deuteroxide samples are presented in figure 2 for both laboratory x - rays and neutrons . although the samples are mostly phase pure , some starting reagent li@xmath41co@xmath24 was found as an impurity in the neutron data , which is more of a bulk technique than x - ray diffraction . furthermore , two very broad peaks could be indexed as close to the lattice parameters of the parent phase @xmath31-fese . indeed , these peaks have also been observed in previous work.@xmath42 the much broader peak width for the fese impurity is indicative of very small crystallite size and quantitatively fitting this phase is not possible given its nearly amorphous nature . [ crystal_data ] fe@xmath38od)fese prepared at 120 @xmath25c for 5 days ( @xmath3 = 37 k ) . rietveld refinement of data collected at both temperatures did not reveal any lowering of symmetry from tetragonal @xmath12 . tick marks representing the tetragonal phase are shown below the calculated , observed , and differences curves . the insets shown are a zoom in of the high - angle synchrotron data . ] the temperature - dependent synchrotron diffraction data did not reveal any major crystallographic changes in the structure ( figure 3 ) . therefore , unlike the parent fese phase , which undergoes a tetragonal to orthorhombic phase transition near 75 k,@xmath43 ( li@xmath1fe@xmath38oh)fese and ( li@xmath1fe@xmath38od)fese remain tetragonal down to base temperature ( 10 k ) . rietveld refinements of one of the deuteroxide patterns at 7 k and 150 k are presented in figure 3 , and relevant structural parameters are in table 1 for both superconducting and non - superconducting deuteroxide phases . relevant bond distances and bond angles are also shown in table 1 . only results from the synchrotron x - ray dataset are presented in table 1 , and structural parameters from the rietveld refinements , including the neutron data , for the rest of the samples used to construct the full phase diagrams can be found in the esi ( tables s1-s5 ) . the squid magnetic susceptibility measurements for the series of hydroxide samples prepared through the powder routes are presented in figure 4a . the deuteroxide samples , which were all derived from the single crystal route , are shown in figure 4b . only one sample within the hydroxide series expressed a ferromagnetic signal within the superconducting regime . a similar plot ( figure s1 ) for the hydroxide system prepared via the single crystal route can be found in the esi . fe@xmath38od)fese . for @xmath3 = 32 k , a ferromagnetic transition can be noted at @xmath44 = 10 k. zfc data of powder samples and single crystal samples collected with applied fields of 1 mt and 3 mt , respectively , are shown . ] we have constructed superconducting phase diagrams in figure 5 that relate the critical temperatures @xmath3 to the lattice constants compiled from the squid data and diffraction results from all the samples . the lattice parameters of the tetragonal unit cell found at room temperature were used as the x - axis versus @xmath3 in the phase diagrams . more specifically , we found the best correlation to @xmath3 is that of the tetragonality parameter , which is the simple @xmath45 ratio . the corresponding superconducting volume fractions ( 4@xmath46@xmath47 ) were also established by squid magnetometry ( figure 4 ) . we found that @xmath3 and its volume fraction increased with the lattice constant @xmath48 and decreased with lattice constant @xmath49 . therefore , those with the highest tetragonality gave the maximum @xmath3 and superconducting volume fractions . for samples to exhibit significant superconducting volume fractions ( 4@xmath46@xmath47 @xmath50 10 % ) , the lattice constant @xmath48 must be larger than about 9.20 and @xmath49 smaller than about 3.80 . these trends in the lattice parameters are consistent with the findings of sun @xmath51 @xmath52 . on their hydroxide analogues,@xmath16 where they attribute a large a lattice constant to iron vacancies in the fese layers and therefore iron slightly oxidized above 2 + . fe@xmath38oh)fese as comparison for that of ( b ) single crystal ( @xmath0li@xmath1fe@xmath38od)fese and ( c ) powder samples and ( li@xmath1fe@xmath38od)fese . the critical temperatures tc are related to the tetragonality parameter , which is the simple ratio of the lattice parameters @xmath45 . ] our combined diffraction experiments did indeed find variations on the iron occupancies , whether superconducting or non - superconducting . in general , the smaller the tetragonality parameter , the lower the @xmath3 . as table 1 shows , when the occupancy of the fe2 site falls from near full to 91.9(2 ) % , superconductivity is lost . the differences overall between the hydroxide samples prepared by powder routes and the single crystal ones could be due the accommodation of iron vacancies during the syntheses . as sun @xmath53 t @xmath52 . found in their samples,@xmath40 when fese in its tetragonal @xmath31-phase is used as the host for intercalation via hydrothermal synthesis , iron powder is necessary in order to fill in the resulting vacancies . when we start with fe powder and selenourea as the se source , this leads to more variability in the amount of iron vacancies and therefore a larger spread in the tetragonality parameter that can express superconductivity ( figure 5 ) . our powder method therefore would lead to the in - situ growth of alternating fese and li@xmath1fe@xmath2od layers rather than post - synthetic modification ( also known as soft chemistry ) of fese layers as done in our single crystal method . to verify whether any of the samples exhibit antiferromagnetism , we searched for any superlattice peaks in the npd patterns that could arise below 120 k , the afm transition in the parent phase of ( li@xmath1fe@xmath2od)fese reported by dong @xmath51 @xmath52.@xmath4 no superlattice reflections were observed in the bt-1 high - resolution npd patterns , and our deuterated samples allowed for a low background in case of a small fe signal . indeed , in the arsenide systems the iron moment can be small in the parent phases such as 0.36(5 ) @xmath54@xmath55/fe in laofeas30 and 0.25(7 ) @xmath54@xmath55/fe in ndofeas,@xmath56 any hydrogen incoherent background would easily overwhelm such a small signal from long - range magnetic ordering in the npd . the samples were measured up to 50 k on bt-1 , and no long range magnetic ordering was observed . npd patterns measured with a psd on bt-7 , which has a much higher flux than bt-1 at low angles , also revealed no antiferromagnetic peaks in the non - superconducting samples ( figure 6 ) . difference patterns between 150 k and 4 k are shown in figure 6 , revealing no residual intensity and only differences arising from thermal expansion and thus peak positions . in preparing our deuteroxide samples , we found the reaction temperature to influence the lattice constants . mild hydrothermal reaction temperatures ( 120 @xmath25c ) led to samples expressing a higher @xmath3 , while the reaction temperature above 180 @xmath25c led exclusively to either non - superconducting samples or ones with very low volume fractions ( 4@xmath46@xmath47@xmath571% ) . reaction times also affected the lattice constants . longer reaction times ( @xmath503 days ) yielded samples with slightly larger @xmath49 and smaller @xmath48 ( i.e. smaller tetragonality parameters ) . while all the deuterated samples followed the trend shown in the phase diagram ( figure 5 ) , similarly prepared hydrated samples deviated in their behaviour . indeed , some hydroxide samples with lattice parameters matching those in the phase diagram from earlier literature1 did not exhibit superconductivity ( figure 4a ) . interestingly , samples prepared at lower temperatures with the described mixing ratio and longer reaction times expressed coexistence between superconductivity and ferromagnetism ( figure 3 ) . thus , while longer reaction times above 180 @xmath25c led to lower @xmath3 s or non - superconducting samples , longer reaction times at lower temperatures ( 120 @xmath25c ) produced the ferromagnetic signal in superconducting ( li@xmath1fe@xmath2od)fese . as described in the next section , this might be a kinetic effect from the increasing amount of oxidized iron in water from longer reaction times . in several of our non - superconducting samples , we have observed an antiferromagnetic transition close to 120 k. dong et al . have claimed that the hydroxide samples in the non - superconducting dome were antiferromagnetic parent phases with a @xmath58 close to 120 k , and therefore that the selenides and arsenides have the same underlying physics with respect to the superconducting mechanism . this is a very important claim that could have large implications in the field of iron - based superconductors . none of our non - superconducting deuteroxide samples , however , exhibited this antiferromagnetic signal in the squid measurements , which led us to believe that the antiferromagnetism may not be intrinsic to this system . li@xmath1fexod)fese at 150 k and ( b ) 4 k. the difference between the two patterns in ( c ) reveals no antiferromagnetic peaks . ] our findings in the preparation of hydroxide and deuteroxide samples revealed the strong possibility that the 120 k transition observed in the squid magnetization measurements arise from iron oxide impurities . the so - called verwey transition , which corresponds to a fe@xmath33/fe@xmath59 charge ordering transition in fe@xmath5o@xmath6 also occurs near 120 k.@xmath60 furthermore , structurally related li@xmath2feo@xmath7 phases can express @xmath58 from 100 k to 300 k according to amount of intercalated lithium cations.@xmath61 in order to study the formation of iron oxide impurities , a sample was prepared under similar hydrothermal conditions but without the addition of selenourea . as pointed out by sun @xmath51 @xmath52 . in their extensive study of the formation of ( li@xmath1fe@xmath2oh)fese , the strongly basic conditions ( ph @xmath50 14 ) of the synthesis strongly favors the formation of fe@xmath59 species according the electrochemical - ph phase equilibrium diagram ( i.e. pourbaix ) of iron.@xmath62 therefore , without the selenourea reagant to stabilize divalent iron , a large amount of mixed valent iron oxides are produced from hydrothermal synthesis containing large amounts of lioh ( or liod ) . o@xmath6 prepared under similar hydrothermal conditions to that of ( li@xmath1fe@xmath38oh)fese in the absence of selenourea . ( b ) the corresponding magnetization data of the fe@xmath5o@xmath6 sample indicating the verwey transition near 125 k. ( c ) the powder xrd of ( li@xmath1fe@xmath38oh)fese with the fe@xmath5o@xmath6 impurity marked along its strongest reflection . ( d ) the corresponding magnetization data of this non - superconducting ( li@xmath1fe@xmath38oh)fese sample , exhibiting the 125 k transition similar to that of fe@xmath33/fe@xmath59 charge ordering seen in fe@xmath5o@xmath6 ( inset ) . ] the xrd pattern of the as - recovered sample from hydrothermal synthesis without selenourea was fitted to the fe@xmath5o@xmath6 structure , and its magnetic susceptibility measurement was in very good agreement with typical verwey transition at 120 k ( figure 7 ) . as shown in figure 7b , a hydroxide sample with lattice constants in the supposed superconducting region showed no superconductivity , but a transition similar to charge ordering in fe@xmath5o@xmath6 . peaks in the xrd pattern ( indicated by * in figure 7b ) that can not be matched with the ( li@xmath1fe@xmath2oh)fese phase was indexed well with the strongest peaks of fe@xmath5o@xmath6 . in addition , our synchrotron xrd data for a non - superconducting ( li@xmath1fe@xmath2oh)fese single crystal sample revealed small amounts of fe@xmath5o@xmath6 impurity ( figure s2 ) , which was not observed by laboratory x - ray measurements . therefore , it is likely that the 120 k transition in non - superconducting ( li@xmath1fe@xmath2oh)fese samples is extrinsic and due to magnetic impurities fe@xmath5o@xmath6 or structurally related li@xmath2feo@xmath7 , which would modulate the ordering temperature . hydrothermal synthesis of samples with d@xmath7o under similar conditions as those with h@xmath7o did not lead to appreciable oxide impurity . we therefore conclude that the observed differences in the acid - base chemistry of h@xmath7o and d@xmath7o lead to different products for similar reaction conditions . indeed , a hydrothermal treatment of iron powder with d@xmath7o and with liod did not lead to complete conversion to fe@xmath5o@xmath6 but left unreacted iron powder ( approximately 50% ) . since under highly basic conditions , fe@xmath63 should be favoured thermodynamically , we believe the kinetics for the oxidation of iron with d@xmath7o is slower than in h@xmath7o . the autoionization constant of d@xmath7o is smaller than that of h@xmath7o due to the stronger d o bond than the h o bond . all deuteroxide single crystal samples prepared in d@xmath7o showed noticeable higher @xmath3 s than their hydroxide counterparts prepared at identical conditions ( table s2 ) . the @xmath3 of hydroxide samples can be improved by reducing the reaction time ( i.e. 37 k vs. 32 k for 2 d and 4 d , respectively at 120 @xmath25c ) . it is likely that a shorter reaction time reduced the extent of fe oxidation . interestingly , both ( li@xmath1fe@xmath2oh)fese and ( li@xmath1fe@xmath2od)fese single crystals prepared using sn metal instead of fe powder at 120 @xmath25c showed the same @xmath3 at 42 k , higher than other samples without using sn . the advantage of sn metal was to create a reducing environment without introducing iron oxide impurities , due to lack of fe powder . although we have established here that the ( li@xmath1fe@xmath2oh)fese system likely does not have a parent antiferromagnetic phase , we do not suggest that the chalcogenide - based superconductors are not linked to the arsenide - based superconductors from the present results . the fe@xmath64te@xmath65ch@xmath66 for @xmath67 = se and s phases in particular exhibit a rich magnetic phase diagram@xmath68 before superconductivity sets in with chemical substitution.@xmath69 the ordered vacancy phase of k@xmath29fe@xmath17se@xmath18.@xmath30 has also shown an antiferromagnetic transition at large temperatures ( about 559 k)@xmath70 while the disordered vacancy phase exhibits a @xmath3 close to 30 k.@xmath71 what distinguishes those two systems , however , from the present compound and fese , is the lack of a large magnetic moment on iron in the latter compounds.@xmath72 in fe@xmath64te it can be as large as 2 @xmath54@xmath55@xmath73 and in k@xmath29fe@xmath17se@xmath18 as large as 3.3(1 ) @xmath54@xmath55.@xmath70 not surprisingly , when in the superconducting regime , both compounds exhibit a spin resonance energy in the inelastic neutron spectra , which corresponds to spin fluctuations . a large magnetic moment is clearly not the case in the present system . as to the ferromagnetic transition observed at about 10 k in the sample with a @xmath3 of 34 k , several authors have also observed it in in the hydroxide analogues . pachmeyer @xmath51 @xmath52 . attribute the long range magnetic order at 18 k to the iron cations partially substituted on the li site,@xmath21 while lu @xmath51 @xmath52 . assign this transition as being antiferromagnetic ( about 12 k ) according to their nmr studies.@xmath74 our recent small angle neutron scattering study illustrates the formation of long - range magnetic order below 12.5 k , but with a moment too small to see with diffraction.@xmath75 no doubt this observation arises from the crystallographic site where the moment is located is too dilute with iron occupancy . in conclusion , we have successfully mapped out a phase diagram for ( li@xmath1fe@xmath2od)fese and have found that the highest @xmath3 for deuterated samples is 42 k , and the @xmath3 for both deuterated and hydroxide samples correlate with lattice constants . since the highest @xmath3 observed for the hydroxide sample was also approximately 42 k , we conclude that there is no isotope effect on the superconducting properties in substituting h by d. mild hydrothermal preparation for long reaction times can lead to the coexistence of ferromagnetism and superconductivity . finally , any claims of anti - ferromagnetism in the parent phase of this system should be re - evaluated in light of the easy preparation of oxide impurities with transition temperatures near the vicinity of 120 k in h@xmath7o . research at the university of maryland was supported by the nsf career dmr-1455118 and the afosr grant no . fa9550 - 14 - 10332 . we acknowledge the support of the national institute of standards and technology , u. s. department of commerce , in providing the neutron research facilities used in this work . use of the advanced photon source at argonne national laboratory was supported by the u. s. department of energy , office of science , office of basic energy sciences , under contract no . de - ac02 - 06ch11357 . we thank s. lapidus for his assistance with the 11-bm measurements . h. sun , d. n. woodruff , s. j. cassidy , g. m. allcroft , s. j. sedlmaier , a. l. thompson , p. a. bingham , s. d. forder , s. cartenet , n. mary , s. ramos , f. r. foronda , b. h. williams , x. li , s. j. blundell and s. j. clarke , _ inorg . _ , 2015 , * 54 * , 1958 - 1964 .	we report the phase diagram for the superconducting system ( @xmath0li@xmath1fe@xmath2od)fese and contrast it with that of ( li@xmath1fe@xmath2oh)fese both in single crystal and powder forms . samples were prepared via hydrothermal methods and characterized with laboratory and synchrotron x - ray diffraction , high - resolution neutron powder diffraction ( npd ) , and high intensity npd . we find a correlation between the tetragonality of the unit cell parameters and the critical temperature , @xmath3 , which is indicative of the effects of charge doping on the lattice and formation of iron vacancies in the fese layer . we observe no appreciable isotope effect on the maximum @xmath3 in substituting h by by d. the npd measurements definitively rule out an antiferromagnetic ordering in the non - superconducting ( li@xmath1fe@xmath2od)fese samples below 120 k , which has been reported in non - superconducting ( li@xmath1fe@xmath2oh)fese.@xmath4 a likely explanation for the observed antiferromagnetic transition in ( li@xmath1fe@xmath2oh)fese samples is the formation of impurities during their preparation such as fe@xmath5o@xmath6 and lixfeo2 , which express a charge ordering transition known as the verwey transition near 120 k. the concentration of these oxide impurities is found to be dependent on the concentration of the lithium hydroxide reagent and the use of h@xmath7o vs. d@xmath7o as the solvent during synthesis . we also describe the reaction conditions that lead to some of our superconducting samples to exhibit ferromagnetism below @xmath3 .
multiple - input multiple - output ( mimo ) techniques can provide the benefits of spatial diversity and multiplexing gain @xcite-@xcite . spatial multiplexing ( v - blast ) using @xmath5 transmit antennas achieves the full - rate of @xmath5 symbols per channel use . however , full transmit diversity of order @xmath5 is not achieved in v - blast . orthogonal space - time block codes ( e.g. , @xmath1 alamouti code ) achieve full transmit diversity , but suffer from rate loss for increased number of antennas @xcite . achieving both full - rate ( fr ) and full - diversity ( fd ) simultaneously is a desired goal in mimo communications . we refer to mimo systems that simultaneously achieve both fr and fd as frfd mimo systems . one way to achieve frfd in mimo systems is to employ non - orthogonal stbcs @xcite-@xcite , which offer the full - rate of @xmath5 symbols per channel use by having @xmath6 symbols in one @xmath7 stbc matrix , and full - diversity under ml decoding . the @xmath1 golden code with 4 symbols in one stbc matrix is a well known non - orthogonal stbc for 2 transmit antennas @xcite . a drawback with frfd - achieving non - orthogonal stbcs is their high decoding complexities , because ml decoding of these stbcs involves joint decoding of @xmath6 symbols . ml decoding in v - blast , on the other hand , involves joint decoding of only @xmath5 symbols . the inability of v - blast to achieve transmit diversity can be overcome through the use of precoding at the transmitter @xcite-@xcite . precoding based on knowledge of full channel state information at the transmitter ( csit ) and first-/second - order statistics of the channel have been studied widely @xcite-@xcite . with full csit precoding , frfd can be achieved in v - blast @xcite . however , in limited feedback precoding schemes in v - blast , fd is achieved , but with some loss in rate @xcite-@xcite . for e.g. , the precoding scheme in @xcite is based on grassmannian subspace packing , which does not allow simultaneous transmission of more than @xmath8 streams ( i.e. , achievable rate is @xmath9 symbols per channel use ) . in a @xmath1 mimo system , this means a rate loss of 50% . the same is true with any other grassmannian subspace packing based scheme or transmit antenna selection based scheme @xcite . our contribution in this paper is two fold : * first , we present a _ limited feedback _ ( lfb ) based precoding scheme for v - blast _ which achieves frfd _ in small systems like @xmath1 , @xmath2 , and @xmath3 mimo . since the proposed scheme is not based on subspace packing or antenna selection , there is no loss in rate . the proposed scheme involves the design of a codebook having a finite number ( @xmath10 ) of unitary precoding matrices , which are generated from a unitary matrix ( @xmath11 ) parametrized by a single angular parameter , @xmath12 the receiver chooses the precoding matrix which maximizes the minimum distance with ml decoding , and sends the corresponding index to the transmitter . we refer to the proposed scheme as frfd - vblast lfb scheme . * second , we present a ber performance comparison between the two frfd - achieving schemes , namely , the proposed frfd - vblast - lfb scheme and non - orthogonal stbc mimo using golden / perfect codes under ml decoding . our simulation results show that in a @xmath1 mimo system with 4-qam/16-qam , the proposed frfd - vblast - lfb scheme outperforms the golden code by about 0.6 db . in @xmath2 and @xmath3 mimo , the performance of frfd - vblast - lfb scheme is comparable to the performance of perfect codes . the proposed frfd - vblast - lfb scheme is attractive because 1 ) ml decoding becomes less complex compared to that of non - orthogonal stbcs , 2 ) the number of feedback bits required to achieve the above performance is small , 3 ) in slow - fading channels it is adequate to send feedback bits only occasionally , and 4 ) in most practical wireless systems feedback channel is often available ( e.g. , for adaptive modulation , rate / power control ) . the rest of this paper is organized as follows . in section [ sec2 ] , we present the system model . the proposed limited feedback precoding scheme is presented in section [ sec3 ] . ber performance of the proposed scheme along with a performance comparison with golden / perfect codes are presented in section [ sec4 ] . conclusions are presented in section [ sec5 ] . consider a precoded v - blast system with @xmath5 antennas at the transmitter and @xmath13 antennas at the receiver . let @xmath14 denote the channel gain matrix , whose entries are i.i.d and @xmath15 . perfect knowledge of * h * is assumed at the receiver but not at the transmitter . let @xmath16 denote the precoder codebook of size @xmath10 , where the @xmath17 s , @xmath18 , are @xmath7 unitary precoding matrices . this codebook is known to both transmitter and receiver . for a given channel @xmath19 , the receiver chooses the precoding matrix from @xmath20 that maximizes the minimum distance with ml decoding , and sends the corresponding index to the transmitter . let @xmath21 denote the number of feedback bits needed to represent this index . given this index , @xmath22 , the transmitter uses the corresponding precoding matrix , denoted by @xmath23 . let @xmath24 denote the complex data symbol vector at the transmitter , where @xmath25 is the modulation alphabet . the transmitted signal vector , @xmath26 is given by @xmath27 . the received signal vector , @xmath28 , at the receiver is given by @xmath29 where @xmath30 is the noise vector whose entries are i.i.d @xmath31 , where @xmath32 is the average energy of the transmitted symbols , and @xmath33 is the average received snr per receive antenna . for a non - precoded system ( i.e. , for @xmath34 ) , the ml decision is given by @xmath35 the probability of error in the decision depends on the minimum distance , @xmath36 , which is given by @xmath37 it is known that precoding at the transmitter improves @xmath36 @xcite . we illustrate this point using the following example and fig . assume a @xmath1 system with * h * @xmath38 $ ] and pam modulation . as can be seen from fig . [ fig1 ] , @xmath39 . now , consider unitary precoding with @xmath40}$ ] . the new effective channel matrix is given by @xmath41 . from fig . [ fig1 ] , it can be seen that @xmath42 . = 7.00 cm = 6.00 cm [ fig1 ] let @xmath43 unitary matrices parametrized by a single angular variable , @xmath44 . for e.g. , for @xmath45 , @xmath46 is the set of all possible matrices of the form @xmath47 , where @xmath48 is a real or complex scalar function of only @xmath49 , and + @xmath0 @xmath50 , + @xmath4 @xmath51 + one example of a matrix from @xmath46 is given by @xmath52 . \end{aligned}\ ] ] with each matrix @xmath53 , we associate an infinite size codebook , @xmath54}. to define a finite set precoder , we select a finite subset of size @xmath55 from @xmath56 , where @xmath57 is the number of feedback bits . specifically , we define the finite precoding codebook as @xmath58 . for e.g. , for @xmath45 and @xmath59 , @xmath60 with @xmath61 in ( [ eg1 ] ) , the finite precoding codebook is given by @xmath62 , \left[\begin{array}{cc}\frac{1-i}{2 } & \frac{-i}{\sqrt{2}}\\ \frac{-1+i}{2 } & \frac{-i}{\sqrt{2 } } \end{array}\right ] , \left[\begin{array}{cc } \frac{-i}{\sqrt{2 } } & \frac{-1}{\sqrt{2 } } \\ \frac{-i}{\sqrt{2 } } & \frac{1}{\sqrt{2 } } \\ \end{array}\right ] , \\ \left[\begin{array}{cc } \frac{-1-i}{2 } & \frac{i}{\sqrt{2}}\\ \frac{1+i}{2 } & \frac{i}{\sqrt{2 } } \end{array}\right]\right\}$ ] . + the performance of the codebook is dependent on the choice of @xmath63 and @xmath10 . we , therefore , come up with the following performance indicator for a given codebook : in words , @xmath64 is the expected ratio of the maximum squared @xmath36 with precoding to that without precoding . then , the optimal finite precoding codebook with @xmath65 feedback bits is @xmath66 ) , where @xmath67 is given by @xmath68 obtaining an exact solution for @xmath69 analytically is difficult . in the absence of a solution to the above problem , we tried out several @xmath63 matrices for small values of @xmath5 ( e.g. , @xmath70 , which are of interest in practical mimo systems ) , and found that the following designs for @xmath70 work very well achieve very good ber performance ( as we will see in sec . [ sec4 ] ) . ] in the proposed scheme : @xmath71 , \label{2x2case}\end{aligned}\ ] ] @xmath72 , \label{3x3case}\end{aligned}\ ] ] @xmath73 at the receiver , given the knowledge of * h * , we define @xmath74 , @xmath75 , as the receiver sends to the transmitter the index @xmath76 , given by @xmath77 using @xmath57 bits of feedback . hence , the optimum precoding matrix chosen is given by @xmath78 signal detection is performed using the sphere decoding algorithm @xcite . in the following , we present the computation of @xmath74 in ( [ dmin1 ] ) for @xmath79 . we can rewrite the system model equation ( [ eqn1 ] ) for the precoded system , when precoded with the @xmath80th precoding matrix as @xmath81 where @xmath82 . using @xmath83 and @xmath84 to denote the real and imaginary parts of a complex argument , the above equation can be transformed into an equivalent real - valued model as @xmath85 where @xmath86}^t,\ \tilde{\bf x}\ = \ { [ \mathcal{r}\{{\bf x}^t\}\ \ \mathcal{i}\{{\bf x}^t\}]}^t$ ] , @xmath87}^t$ ] , and @xmath88.\ ] ] here , @xmath89 . also , @xmath90 is the real pam constellation corresponding to @xmath25 . henceforth , we shall work with the real - valued system in ( [ realeq ] ) . with the new system model in ( [ realeq ] ) , we can re - write ( [ dmin1 ] ) as where @xmath91 is the difference constellation of @xmath90 . for example , for square @xmath92-qam modulation , @xmath90 is given by @xmath93 and @xmath94 is @xmath95 . since the factor of 2 can be neglected , @xmath94 can therefore be taken to be @xmath94 = @xmath96 . geometrically , @xmath97 defines a @xmath98 dimensional lattice in @xmath99 , denoted as @xmath100 and a finite subset of @xmath101 is @xmath102 = @xmath103 . for the receiver to find the optimal precoding matrix , it needs to evaluate @xmath104 using ( [ dmin_2 ] ) for all @xmath105 . given the geometrical interpretation it is easy to see that calculation of ( [ dmin_2 ] ) is equivalent to the problem of finding the shortest vector in a finite subset of the lattice @xmath101 . this problem is a constrained version of the well known shortest vector problem ( svp ) for any arbitrary lattice . unconstrained svp can be solved by appropriately modifying the closest lattice point search algorithm , as discussed in @xcite . we solve the constrained svp problem by restricting the search to be within the finite subset @xmath102 . we evaluated the ber performance of the proposed limited feedback precoding scheme as a function of average received snr per receive antenna , @xmath33 , through simulations for @xmath106 . for comparison purposes , we also evaluate the performance of golden code / perfect codes in mimo systems with @xmath106 . figure [ fig2 ] shows the simulation results for @xmath1 and @xmath3 v - blast without and with the proposed precoding , for 4-qam and 16-qam using sphere decoding . ber performance plots for different levels of quantization requiring different number of feedback bits ( @xmath107 bits ) are shown . from fig . [ fig2 ] , it is observed that the proposed precoding scheme achieves significantly better diversity than the ` no precoding ' scheme . in fact , a comparison of this precoded performance for @xmath1 v - blast with the performance of @xmath1 golden code in fig . [ fig3 ] shows that both these curves run parallel illustrating that , just like golden code , the proposed scheme also achieves full diversity . another interesting observation in fig . [ fig2 ] is the effect of feedback bits on the ber performance . it can be seen that the ber with @xmath108 and @xmath109 are almost the same , showing that the performance in the proposed scheme remains robust even with a nominal quantization of using 4 bits . = 8.750 cm = 7.25 cm [ fig2 ] = 8.750 cm = 7.25 cm [ fig3 ] next , in figs . [ fig3 ] to [ fig5 ] , we compare the ber performances of the proposed precoding scheme and the golden code / perfect codes in @xmath1 ( fig . [ fig3 ] ) , @xmath2 ( fig . [ fig4 ] ) , and @xmath3 ( fig . [ fig5 ] ) systems , using sphere decoding . the channel is assumed to remain constant for @xmath5 consecutive channel uses in v - blast in order to facilitate the comparison between v - blast and golden code / perfect codes ( which are assumed to have a quasi - static interval of @xmath5 channel uses ) under similar quasi - static channel conditions . from fig . [ fig3 ] , it can be seen that , for both 4-qam and 16-qam , the performance curves of the proposed scheme runs parallel to those of the golden code , showing that the proposed scheme achieves the full diversity of 4 . it is interesting to further observe that the proposed scheme exhibits some coding gain advantage compared to the golden code . particularly , the coding gain attained by the proposed scheme over golden code is about 0.7 db at a ber of @xmath110 . this is significant since precoding achieves this better performance with a lower decoding complexity ( joint detection of 2 symbols in one channel use ) than the golden code ( joint detection of 4 symbols in one stbc matrix ) . next , the ber comparison in fig . [ fig4 ] for @xmath2 system shows that the proposed scheme achieves the same diversity as that of @xmath2 perfect code , but is inferior to perfect code in terms of coding gain . this performance gap , however , is small ( about 0.3 db ) . in @xmath3 system in fig . [ fig5 ] , the performance gap in terms of coding gain is about 1 db . this suggests that better precoding strategies for larger @xmath5 can be investigated to achieve close to perfect code performance . a likely approach can be to consider multiple parameter based precoder designs . a key advantage of the proposed precoder approach compared to the non - orthogonal stbc approach is its lesser decoding complexity . this advantage is captured in the complexity comparison plots in fig . [ fig6 ] , where the complexities in both the approaches using sphere decoding , in terms of number of real operations per decoded symbol , are plotted as a function of @xmath106 at snrs corresponding to a target ber of @xmath111 . it can be seen that the complexity in the proposed approach is much less because it needs to jointly detect only @xmath5 symbols , whereas in the non - orthogonal stbc approach joint detection is over @xmath6 symbols . we note that the complexity comparison in fig . [ fig6 ] does not include the complexity involved in the optimization to choose the optimum precoding matrix at the receiver . it is pointed out that the data decoding complexity dominates over the precoding selection complexity , as decoding is done on a per channel use basis whereas , in slow fading , the precoding selection computation need not be carried out so frequently . = 8.750 cm = 7.25 cm [ fig4 ] = 8.75 cm = 7.25 cm [ fig5 ] = 8.750 cm = 7.25 cm [ fig6 ] we presented a simple , single angular parameter based codebook design for limited feedback precoding in v - blast . the proposed precoding scheme achieves full - rate for any @xmath5 by design , whereas the achievability of full - diversity was established through ber simulations for @xmath106 under ml decoding for 4-qam/16-qam . our simulation results showed that in a @xmath1 mimo system , the proposed scheme outperformed the golden code by about 0.6 db . it performed comparable to perfect codes in @xmath2 and @xmath3 mimo systems as well . the decoding complexity in the proposed scheme was shown to be much less compared to that of golden / perfect codes . it is noted that the feedback channel is an additional resource required in precoding schemes . however , given that a feedback channel is often available in most practical wireless systems ( e.g. , for adaptive modulation , rate / power control , etc . ) , and that the feedback bandwidth required will be very less in slow fading , the proposed scheme can be quite attractive for its full - rate , full - diversity , and low - complexity attributes . investigations related to applicability of the proposed approach to large antenna systems using multiple - parameter based precoder designs can be carried out further .	full - rate ( fr ) and full - diversity ( fd ) are attractive features in mimo systems . we refer to systems which achieve both fr and fd simultaneously as frfd systems . non - orthogonal stbcs can achieve frfd without feedback , but their ml decoding complexities are high . v - blast without precoding achieves fr but not fd . frfd can be achieved in v - blast through precoding given full channel state information at the transmitter ( csit ) . however , with limited feedback precoding , v - blast achieves fd , but with some rate loss . our contribution in this paper is two - fold : @xmath0 we propose a limited feedback ( lfb ) precoding scheme which achieves frfd in @xmath1 , @xmath2 and @xmath3 v - blast systems ( we refer to this scheme as frfd - vblast - lfb scheme ) , and @xmath4 comparing the performances of the frfd - vblast - lfb scheme and non - orthogonal stbcs without feedback ( e.g. , golden code , perfect codes ) under ml decoding , we show that in @xmath1 mimo system with 4-qam/16-qam , frfd - vblast - lfb scheme outperforms the golden code by about 0.6 db ; in @xmath2 and @xmath3 mimo systems , the performance of frfd - vblast - lfb scheme is comparable to the performance of perfect codes . the frfd - vblast - lfb scheme is attractive because 1 ) ml decoding becomes less complex compared to that of non - orthogonal stbcs , 2 ) the number of feedback bits required to achieve the above performance is small , 3 ) in slow - fading , it is adequate to send feedback bits only occasionally , and 4 ) in most practical wireless systems feedback channel is often available ( e.g. , for adaptive modulation , rate / power control ) .
consider a pure @xmath1-dimensional polyhedral complex @xmath2 . the _ graph @xmath9 _ , or _ @xmath10-skeleton _ , of @xmath2 is the undirected simple graph that has the vertices of @xmath2 as nodes and the one - dimensional faces of @xmath2 as edges . the study of graph - theoretic connectivity of skeleta of polyhedral complexes has its beginning with steinitz theorem from 1922 , which states that a graph is the @xmath10-skeleton of the boundary complex of some @xmath11-dimensional convex polytope if and only if it is @xmath11-connected and planar . later , balinski @xcite generalized part of this to higher dimensions by showing that the graph of every @xmath7-dimensional convex polytope is @xmath7-connected . this was generalized further by barnette @xcite who showed that the graph of every @xmath1-dimensional polyhedral pseudomanifold is @xmath7-connected . in this paper we consider the simplicial case . a simplicial pseudomanifold is said to be _ flag _ if its faces coincide with the cliques of its @xmath10-skeleton , and it is said to be _ normal _ if all links of faces are connected . it was shown by athanasiadis @xcite that the graph of a flag simplicial @xmath1-pseudomanifold is @xmath8-connected . we introduce an invariant @xmath0 of a pure @xmath1-dimensional simplicial complex @xmath2 taking values in the range @xmath3 . it is shown to affect connectivity in the following way . [ main ] let @xmath12 be a @xmath1-dimensional normal simplicial pseudomanifold . then the graph @xmath9 is @xmath5-connected . the definition of @xmath0 involves the concept of _ banner _ triangulations , a generalization of flag triangulations . we have that @xmath6 if and only if the complex @xmath12 is banner , and in particular if @xmath2 is flag . thus , for the case of normal simplicial pseudomanifolds our result interpolates between the theorems of barnette and athanasiadis . in this section we collect some definitions and review some auxiliary results needed later in the paper . we assume basic knowledge about simplicial complexes . throughout the paper , @xmath2 will denote a pure finite @xmath1-dimensional simplicial complex on vertex set @xmath13 . let @xmath14 be a face of @xmath2 . the _ link _ of @xmath14 in @xmath2 , denoted @xmath15 , is the subcomplex that contains a face @xmath16 if @xmath17 and @xmath18 . let @xmath19 be a vertex of @xmath2 . the _ closed star _ @xmath20 is the subcomplex of @xmath2 which is the cone over @xmath21 with apex @xmath22 . the _ antistar _ @xmath23 is the subcomplex of @xmath2 induced on the set of vertices @xmath24 . note that @xmath25 . a pure @xmath1-dimensional complex @xmath2 is a _ pseudomanifold _ if * every @xmath26-dimensional face is contained in exactly two facets ( maximal faces ) , * @xmath2 is strongly connected , meaning that the _ facet graph _ ( whose vertices are the facets and edges the pairs of adjacent facets ) of @xmath2 is connected . we use the following property of pseudomanifolds at a crucial point in the paper . [ lem : astar_sc ] the antistar of any vertex in a pseudomanifold is strongly connected . a pure simplicial complex @xmath2 is a _ homology manifold _ if @xmath2 is connected and @xmath15 has the homology of a sphere of the appropriate dimension for every nonempty face @xmath27 . the complex @xmath2 is a _ homology sphere _ if it is a homology manifold and itself has the homology of a sphere of the same dimension . a pseudomanifold is called _ normal _ if all links @xmath15 of dimension at least one are connected . the condition of being normal is quite natural and holds e.g. for all homology manifolds . most importantly , it is not hard to check that the class of normal pseudomanifolds is closed under taking links . this is not the case for pseudomanifolds in general . of the following four properties for a pure simplicial complex @xmath2 , each implies its successor : * @xmath2 is a triangulation of a compact topological manifold , * @xmath2 is a homology manifold , * @xmath2 is a normal pseudomanifold , * @xmath2 is a pseudomanifold . we assume basic knowledge about graphs and refer to @xcite for details . a graph @xmath28 is said to be _ @xmath29-connected _ if @xmath28 has more than @xmath29 vertices and @xmath30 is connected for every set of vertices @xmath31 with @xmath32 , where @xmath30 denotes the graph that one obtains by deleting the vertices in @xmath31 and all incident edges . the following well - known theorem relates the connectivity of a graph to families of independent paths . here a family of paths between two vertices @xmath22 and @xmath33 is called _ independent _ if the only vertices contained in more than one path of the family are @xmath22 and @xmath33 . a graph @xmath28 with at least @xmath34 vertices is @xmath29-connected if and only if any two vertices can be joined by @xmath29 independent paths . it turns out that it suffices to check this condition for vertices at distance two in the graph . this fact plays a central role for proving the connectivity results in this paper . [ liu ] let @xmath28 be a connected graph with at least @xmath34 vertices . if for any two vertices @xmath35 and @xmath36 of @xmath28 with distance @xmath37 there are @xmath29 independent @xmath35 @xmath36 paths in @xmath28 , then @xmath28 is @xmath29-connected . assume that @xmath28 is not @xmath29-connected . choose a set @xmath31 with less than @xmath29 vertices such that @xmath30 is disconnected and such that every proper subset of @xmath31 does not disconnect @xmath28 . then there are two vertices @xmath38 in different components of @xmath30 and a path from @xmath22 to @xmath33 in @xmath28 that contains exactly one element @xmath39 of @xmath31 . consider the vertices @xmath35 and @xmath36 immediately before and after @xmath39 on this path . they are at distance two in @xmath28 and can not be connected by a path in @xmath30 , because @xmath22 and @xmath33 can not . thus , every @xmath35 @xmath36 path in @xmath28 must pass through @xmath31 , so there are at most @xmath40 independent @xmath35 @xmath36 paths in @xmath28 . here we present the concept of banner triangulations , a generalization of flag triangulations . 1 . a _ clique _ is a subset @xmath41 such that @xmath42 for all @xmath43 . it is a @xmath44-clique if @xmath45 . a clique @xmath46 is _ spanning _ if @xmath47 . the complex @xmath2 is said to be a _ flag complex _ if every clique is spanning . flag complexes have been shown to have very strong properties in many situations . this is also the case for the connectivity of their edge graphs @xmath9 as the following result shows . let @xmath2 be a @xmath1-dimensional simplicial pseudomanifold . if @xmath2 is flag , then the graph @xmath9 is @xmath8-connected . the aim of this paper is to interpolate between the connectivity result of barnette for general pseudomanifolds and the one of athanasiadis for flag pseudomanifolds . for that purpose , we introduce the concept of banner complexes . 1 . a clique @xmath46 is _ critical _ if @xmath48 , for some @xmath49 . a pure @xmath1-dimensional complex @xmath2 is said to be a _ banner complex _ if * every critical @xmath7-clique is spanning and * @xmath2 does not contain the boundary complex of a @xmath7-simplex as a subcomplex . -clique ] every clique in a flag complex is spanning . let us call a pure @xmath1-dimensional complex @xmath2 , not containing the boundary of a @xmath7-simplex as a subcomplex , strongly banner if every @xmath7-clique is spanning . then @xmath50 all triangle - free graphs with no isolated vertices are strongly banner . for two - dimensional complexes the concepts of banner and flag coincide . this is so because every @xmath11-clique is critical and a non - spanning @xmath11-clique is the same thing as an empty triangle . furthermore , if every @xmath11-clique is spanning but there is a non - spanning @xmath51-clique then the complex contains the boundary of a @xmath11-simplex as subcomplex . the three concepts become distinct starting in dimension three , as shown by the following examples . [ rem0 ] for any graph @xmath52 we construct a pure @xmath11-dimensional simplicial complex @xmath53 as follows . just expand each edge @xmath54 of @xmath28 to a tetrahedron @xmath55 , where @xmath56 and @xmath57 are the two endpoints of @xmath54 , and @xmath58 and @xmath59 are new vertices specific to @xmath54 . then @xmath53 is defined as the complex with facets @xmath60 , @xmath61 . thus , @xmath53 consists of tetrahedra , one for each edge @xmath61 , that pairwise meet in a vertex exactly when the corresponding edges do . let @xmath62 , the complete graph on @xmath63 vertices . one sees that @xmath64 is strongly banner but not flag , while @xmath65 is banner but not strongly banner . [ rem1 ] we now present a detailed construction of a shellable @xmath11-ball @xmath66 on @xmath67 vertices that is strongly banner but not flag . the vertices are ( @xmath68 ) : @xmath69 and @xmath33 , and the 26 facets are ( @xmath70 ) : @xmath71 figure 2 shows part of the structure of @xmath66 , namely the subcomplex generated by the @xmath72 facets @xmath73 this subcomplex , a `` ring of three bananas , '' consists of @xmath11 octahedra on vertices @xmath74 , @xmath75 , glued together at the common vertices @xmath76 . the rest of @xmath66 is used to `` fill the hole '' in this octahedral ring . ] the complex @xmath66 is shellable . shellings are obtained by starting with the @xmath77 facets in @xmath78 , at the center of @xmath66 , and then moving out to the facets of the three octahedra . thus , being a shellable pseudomanifold with boundary , @xmath66 is a @xmath11-dimensional ball . its @xmath79-vector is @xmath80 . its boundary is a @xmath81-sphere with @xmath79-vector @xmath82 . the complex @xmath66 has a unique empty triangle , namely @xmath83 . hence , it is not flag . however , it is strongly banner . one can reason as follows to see that every @xmath51-clique is spanning . suppose that @xmath84 is a @xmath51-clique in @xmath85 . dividing the vertices into @xmath11 groups @xmath86 we observe that none of the three sets contains a @xmath51-clique and that either @xmath87 or @xmath88 , since there are no edges @xmath89 . so there are two cases to consider : @xmath90 and @xmath91 . we leave to the reader the few easy steps left to verify that in both cases @xmath84 must be one of the @xmath92 facets . from this ball a triangulated @xmath11-sphere that is strongly banner but not flag can be derived , see example [ rem2 ] below . [ rem9 ] a triangulation of a @xmath11-ball that is banner but not strongly banner can be constructed along the same lines as our 3-ball in example 3.5 , starting this time from a @xmath51-clique , embedding its @xmath93 edges into octahedra to form a `` tetrahedron of six bananas '' , and then filling in the rest so that it is banner . the properties of being banner or strongly banner are inherited by some related complexes , such as links , cones and suspensions . here are a few useful such constructions . [ lem : banner_link ] let @xmath2 be a pure @xmath1-dimensional simplicial complex . if @xmath2 is banner and @xmath22 is a vertex of @xmath2 , then @xmath21 is banner . the same is true for being strongly banner . because @xmath21 is @xmath26-dimensional , we need to show that every critical @xmath1-clique is spanning and that @xmath21 does not contain the boundary of the @xmath1-simplex . let @xmath46 be a critical @xmath1-clique in @xmath21 . then @xmath94 is a face of @xmath21 for some vertex @xmath95 . if we set @xmath96 , then @xmath97 is a clique and @xmath98 is a face of @xmath2 . so , @xmath97 is a critical @xmath7-clique and thus spanning because @xmath2 is banner . equivalently , @xmath46 is a face of @xmath21 and thus spanning . assume that @xmath21 contains the boundary of the @xmath1-simplex as a subcomplex . then the set @xmath46 of vertices of that simplex in @xmath21 is a critical @xmath7-clique in @xmath2 . because @xmath2 is banner , @xmath46 is spanning and the subcomplex of @xmath2 on vertices @xmath99 is the boundary of a @xmath7-simplex . if a triangulated @xmath1-ball @xmath2 is extended by raising a cone with apex @xmath22 over its boundary complex @xmath100 , we obtain a triangulated @xmath1-sphere @xmath101 , see figure [ fig : banner3 ] . , scaledwidth=40.0% ] the @xmath44-cliques in @xmath102 are of three kinds : \(1 ) the @xmath44-cliques of @xmath9 , \(2 ) the @xmath103-cliques of @xmath104 augmented by @xmath22 , and \(3 ) the @xmath103-cliques of @xmath9 with no interior vertex and not of type ( 2 ) , augmented by @xmath22 . for example , in figure [ fig : banner3 ] the triangle @xmath105 is a @xmath11-clique of type @xmath106 . [ thm : conesusp ] for a pure simplicial complex @xmath2 , let property x denote either `` banner '' , `` strongly banner '' , or `` flag '' . then , 1 . @xmath107 + 2 . @xmath108 + 3 . if @xmath2 is a @xmath1-ball lacking @xmath103-cliques of type ( 3 ) , then + @xmath109 the proofs are in all cases straightforward verifications . [ rem2 ] parts ( i ) and ( ii ) of the proposition can be used to give examples in all dimensions of complexes that are exactly one of banner , strongly banner and flag . part ( iii ) applied to the @xmath11-ball @xmath66 of example [ rem1 ] shows how to construct a shellable @xmath11-sphere that is strongly banner but not flag . applying the construction instead to the @xmath11-ball of example [ rem9 ] we obtain a banner shellable @xmath11-sphere that is not strongly banner . the three classes of complexes that we have considered have the property that the graph @xmath9 determines the whole complex , or almost . any complex is of course determined by its facets , and in a strongly banner complex @xmath2 the facets are determined by the @xmath7-cliques of @xmath9 . the situation for banner complexes is slightly weaker . suppose that @xmath2 is banner , strongly connected , and every codimension one face is contained in exactly @xmath29 facets ( e.g. , if @xmath2 is a banner pseudomanifold , the @xmath110 case ) . then it can be shown that if we know one particular @xmath7-clique to be a facet , then we can identify the rest of the facets among the other @xmath7-cliques . the aim of this section is to prove the following : [ thm : bannergraph ] let @xmath2 be a normal @xmath1-dimensional pseudomanifold . if @xmath2 is banner , then its graph @xmath9 is @xmath8-connected . to prove theorem [ thm : bannergraph ] , we will use a few lemmas . for @xmath111 let @xmath112 be the set of vertices of @xmath113 , that is , @xmath22 together with its neighboring vertices @xmath114 such that @xmath115 . [ lem : banner_1 ] let @xmath2 be a banner pseudomanifold . if @xmath116 then @xmath117 . let @xmath84 be any maximal face in @xmath21 that contains @xmath33 . then , since @xmath2 is a pseudomanifold , @xmath118 consists of two isolated vertices , one of which is @xmath22 . let @xmath119 be the other such vertex , so that @xmath120 and @xmath121 . assume that @xmath119 is adjacent to @xmath22 , and as usual let @xmath122 . for any vertex @xmath123 , the set @xmath124 is a critical @xmath7-clique in @xmath2 , sincce @xmath125 is a face of @xmath2 . because @xmath2 is assumed to be banner , @xmath124 is a face of @xmath2 . since @xmath123 is arbitrary , this implies that @xmath2 contains the boundary of the @xmath7-simplex on vertex set @xmath126 , in contradiction to our assumption that it is banner . therefore , @xmath119 is not adjacent to @xmath22 and @xmath117 . [ lem : banner_gnotcomplete ] let @xmath2 be a pseudomanifold . if @xmath2 is banner , then @xmath9 is not a complete graph . follows from the preceding lemma and is also easy to see directly from the definition . we remark that banner complexes that are not pseudomanifolds can have a complete graph . however , this happens only for complexes that are the @xmath1-skeleton of a @xmath29-simplex with @xmath127 . let @xmath2 be a banner pseudomanifold , @xmath128 an arbitrary vertex of @xmath2 and let @xmath66 denote the subcomplex of @xmath2 induced on the set @xmath129 . we need for our proof that @xmath66 is nonempty and connected . it is clear from lemma [ lem : banner_1 ] that @xmath130 . that @xmath66 is connected seems very natural but is not to be taken automatically for granted . for instance , for the complex shown in figure [ fig : banner3 ] ( which is not banner ) the subcomplex @xmath66 consists of two isolated vertices @xmath35 and @xmath119 . we offer two proofs . the first one is entirely elementary , relying on barnette s lemma [ lem : astar_sc ] . the second uses lefschetz duality , and is therefore valid only for homology manifolds . [ prop : banner_outsidepath ] let @xmath22 be a vertex of a banner pseudomanifold @xmath2 . then the subcomplex @xmath66 induced on the set of vertices not adjacent to @xmath22 is connected . let @xmath35 and @xmath119 be vertices not adjacent to @xmath22 , and let @xmath131 and @xmath14 be facets of @xmath132 such that @xmath133 and @xmath134 . from lemma [ lem : astar_sc ] we know that @xmath23 is strongly connected , so we may choose a path @xmath135 in the facet graph of @xmath23 . this given , we want to choose vertices @xmath136 such that for all @xmath137 : 1 . @xmath138 , 2 . @xmath139 if this is possible we are done , because then @xmath140 is a path in @xmath141 , that is , in @xmath66 . suppose that for some @xmath137 , such a choice of @xmath142 is impossible . then @xmath143 the ridge @xmath144 has the following properties : \1 . @xmath145 is a @xmath7-clique , \2 . @xmath145 is critical , since @xmath146 , \3 . @xmath145 is not spanning , since @xmath2 is a pseudomanifold and @xmath147 is already contained in the facets @xmath148 and @xmath149 . this contradicts the assumption that @xmath2 is banner . here we assume that @xmath2 is a homology manifold . all homology groups are taken over @xmath150 . we have that @xmath151 , as is true for all pseudo - manifolds ( * ? ? ? * exercise 43.5d ) . let @xmath152 be the subcomplex of @xmath2 induced on the set @xmath112 and let @xmath66 be the complex induced on the complementary set of vertices , as before . consider the long exact sequence for relative homology @xmath153 we have that @xmath154 , and our assumption that @xmath2 is banner implies that @xmath113 and @xmath152 have the same faces of dimensions @xmath1 and @xmath155 . because @xmath113 is contractible , it follows that @xmath156 , and thus that @xmath157 . by lefschetz duality ( * theorem 70.2 ) , we get that @xmath158 which means that the space @xmath159 is connected . finally , @xmath66 is a deformation retract of @xmath159 ( * ? ? ? * lemma 70.1 ) , and is therefore also connected . this is equivalent to @xmath85 being connected . we have now assembled all pieces needed to prove theorem [ thm : bannergraph ] . we use induction on @xmath1 and liu s lemma [ liu ] . for @xmath160 , @xmath2 is a cycle graph and thus @xmath81-connected . assume that @xmath161 , and let @xmath33 and @xmath162 be a pair of vertices at distance two . from lemma [ lem : banner_gnotcomplete ] we know that @xmath9 is not a complete graph , so such pairs exist . let @xmath22 be a vertex of @xmath2 such that @xmath33 and @xmath162 are contained in @xmath21 . because @xmath2 is normal , @xmath21 is a @xmath26-dimensional normal pseudomanifold . using proposition [ lem : banner_link ] we see that @xmath21 is banner . so , by induction , the graph of @xmath21 is @xmath163-connected . this means that @xmath21 contains more than @xmath163 vertices , and we can find @xmath163 independent paths from @xmath33 to @xmath162 in @xmath21 . by lemma [ lem : banner_1 ] we may choose vertices @xmath164 and @xmath165 . thus , @xmath166 . we obtain a path from @xmath33 to @xmath162 by first going to @xmath35 . then , by lemma [ prop : banner_outsidepath ] , we can continue along a path from @xmath35 to @xmath119 with all vertices in @xmath66 . finally we go from @xmath119 to @xmath162 . note that this path has no interior vertices in @xmath21 . thus , together with the path @xmath167 we have found @xmath8 paths from @xmath33 to @xmath162 which by construction are independent . also , @xmath33 and @xmath162 are not adjacent , so the path outside @xmath20 has at least one inner vertex . this vertex and @xmath22 are both outside @xmath21 , so @xmath2 has more than @xmath8 vertices . by definition , a one - dimensional pseudomanifold is banner if and only if it is not the cycle graph @xmath168 , the boundary of a triangle . however , every one - dimensional pseudomanifold including the boundary of a triangle is @xmath81-connected . this observation motivates including @xmath168 in the following definition . it is a consequence of proposition [ lem : banner_link ] that the link of any face @xmath14 with @xmath174 is banner or @xmath168 . let @xmath175 be any face with @xmath176 . then @xmath177 is banner , because @xmath178 . assume that @xmath180 and let @xmath181 be any vertex . let @xmath182 . by induction , @xmath85 is @xmath183-connected . using lemma [ lem : tentlinks ] , we see that @xmath184 . thus , @xmath85 is @xmath185-connected and has at least @xmath186 vertices . together with @xmath22 , then @xmath9 has at least @xmath187 vertices . if @xmath9 is complete , then we are done . if not , let @xmath188 be two vertices at distance @xmath81 in @xmath9 . then we find a vertex @xmath181 such that @xmath189 and again , let @xmath182 . by induction and lemma [ lem : tentlinks ] , there are @xmath190 independent paths from @xmath33 to @xmath162 in @xmath85 . together with the path @xmath191 , this gives us @xmath186 independent paths from @xmath33 to @xmath162 in @xmath9 . by liu s lemma [ liu ] , this proves the result . * acknowledgments . * we want to thank c. athanasiadis for bringing barnette s lemma [ lem : astar_sc ] to our attention , a. goodarzi for suggesting example [ rem0 ] , and s. klee and i. novik for spotting an error in an earlier version . we also thank the anonymous referee who made made several suggestions that led to improvements of the exposition .	this paper is concerned with lower bounds for the connectivity of graphs ( one - dimensional skeleta ) of triangulations of compact manifolds . we introduce a structural invariant @xmath0 of a simplicial @xmath1-manifold @xmath2 taking values in the range @xmath3 . the main result is that @xmath4 influences connectivity in the following way : _ the graph of a @xmath1-dimensional simplicial compact manifold @xmath2 is @xmath5-connected . _ the parameter @xmath0 has the property that @xmath6 if the complex @xmath2 is flag . hence , our result interpolates between barnette s theorem ( 1982 ) that all @xmath1-manifold graphs are @xmath7-connected and athanasiadis theorem ( 2011 ) that flag @xmath1-manifold graphs are @xmath8-connected . the definition of @xmath0 involves the concept of _ banner _ triangulations of manifolds , a generalization of flag triangulations .
the ultraluminous infrared galaxy mkn231 ( ugc8058 , viizw490 ) has the highest luminosity ( @xmath3@xmath4 for a distance of 170mpc ) of the original ulirg sample of @xcite from the bright galaxy survey . that this galaxy hosts a powerful agn is beyond doubt , as evidenced by the wealth of observations : for example , a compact nuclear radio continuum source with jets @xcite , broad absorption lines @xcite , and a compact source of variable hard x - ray emission @xcite . because mkn231 has an absolute blue magnitude @xmath5 close to the cutoff of @xmath6 for qsos @xcite , it is often classified as one of these . however , it is also sometimes classed as a seyfert1 , particularly because it has become clear that the agn is responsible for only a part of the luminosity , and that a significant fraction ( 1/3 to 2/3 ) infact originates in star formation @xcite . it is exactly this quality the co - existence in the nucleus of both the star forming and agn phenomena with massive luminosities combined with its relative proximity that make mkn231 a key object for investigations of whether , as a class , ulirgs do ( e.g. * ? ? ? * ; * ? ? ? * ) or do not ( e.g. * ? ? ? * ) evolve into qsos . the asymmetrical optical morphology and large scale tidal tales @xcite of mkn231 are typical of the remnants predicted by models of merger events between equal mass gas rich spiral galaxies ( e.g. * ? ? ? * ; * ? ? ? such models indicate that mkn231 is most likely to be such a late stage merger . the structural and kinematic properties of a sample of similar late stage ulirg mergers have been studied by @xcite to investigate whether they might evolve later into ( intermediate mass , @xmath7 ) ellipticals ; and by @xcite to test whether , once rid of their gas and dust shells , they might be the progenitors of qsos . the total sample consists of 18 ulirgs , of which 8 contain qso nuclei . mkn231 presents a puzzle since its stellar velocity dispersion of 120 is the smallest of the sample ( measured in 0.6 slits ) . a remnant of a merger between massive gas rich spirals is expected to have a higher velocity dispersion , and indeed the ulirg sample as a whole had a ( logarithmic ) average of 185 . on the other hand , mkn231 appears to be similar to the rest of the sample in having a relaxed stellar population , with @xmath8 and an @xmath9 luminosity profile . however , again the effective radius of the extended h and k band light in mkn231 of 300pc is the smallest of the sample , which had a ( logarithmic ) average of 2kpc . under the assumption that the stellar distribution is spheroidal , @xcite calculated the mass within the effective radius ( 300pc ) to be @xmath10 . from the relation between velocity dispersion and black hole mass , they also estimated @xmath11 , which would imply an eddington efficiency of 6 . the rather uncomfortable result of a super - eddington black hole luminosity led them to consider whether the stars might actually lie in a nearly face - on regularly rotating inner disk , as exhibited by the cold molecular gas @xcite , since this could also lead to a low value for @xmath12 . to address this question in detail and shed light on the nucleus of mkn231 , we have undertaken near infrared spectroscopy using adaptive optics . the aim is to investigate its stellar luminosity profile , mass , and geometry , and to probe its nuclear dynamics at high angular resolution in the central arcsecond . the paper is arranged as follows . after summarising the observations in [ sec : obs ] , we consider the shape and extent of the continuum emission in [ sec : im ] . the radial profile , luminosity , and mass of the nuclear star forming region is discussed in detail in [ sec : starform ] . in [ sec : kin ] we turn to the kinematics of the gas and stars , and construct a simple planar disk model which can reproduce their characteristics . bringing these results together suggests that the starburst in the nucleus of mkn231 is very young , and so in [ sec : young ] we appraise other evidence from the literature that this is the case . finally , we conclude in [ sec : conc ] . the data were obtained on the night of 28 may 2002 using nirc-2 , in a partially commissioned shared risk basis , behind the adaptive optics ( ao ) system @xcite on the keck ii telescope . nirc-2 is designed to take full advantage of the adaptive optics system and has a @xmath13 aladdin-3 insb array , with a number of grisms and slits for the spectroscopic mode . it is unusual in that , in order to select the required wavelength range , the slit and science target are shifted in the ao system s focal plane rather than tilting the grism . spectra were obtained in the h- and k - bands with the camera in widefield mode ( 40mas pixels ) through a 80mas slit with the medium resolution grism , resulting in nominal spectral resolutions of @xmath14 and 2090 respectively . the wavelength coverage is sufficient to include each band entirely in a single setting . two position angles ( pa ) were used : @xmath15 ( close to the kinematic major axis ) and @xmath16 ( reasonably close to the minor axis , but bisecting the extended 1 - 0s(1 ) emission seen by @xcite ) . the two positions of the slit are depicted in fig . [ fig : slits ] . individual exposures of 600sec in the h - band and 300sec in the k - band were used , nodding the slit after each exposure . total integration times were 40mins and 80mins . standard calibrations were performed , including atmospheric standard stars ( using the type g2v star hd117845 in the h - band , and the f2 star hd110105 in the k - band ) , arcs , flatfields , and dark frames . the data were reduced using pc - iraf2.11.3 using standard techniques . the spatial resolution depended on the adaptive optics performance , the reference for which was mkn231 itself . because the seeing was reasonably good and stable , the wavefront sensor was able to run at a frame rate of 55hz with up to 60counts per pixel . for the k - band this resulted in a spatial resolution limited only by the coarse pixel sampling . the fwhm measured on the nucleus of mkn231 was 85mas ( close to the 80mas expected for nyquist sampling ) . for the h - band , both the spatial and temporal scales of the atmospheric turbulence are smaller , and hence the ao performance is not so good . the resulting fwhms measured on the galaxy nucleus were 0.192and 0.160 for the two position angles ( @xmath15 and @xmath16 respectively ) . in section [ sec : starform ] we show that the resolved stellar continuum contributes relatively little to this , and hence we can take it as a reasonable estimate of the spatial resolution . the resulting fwhm spectral resolution measured from the arc lamp lines is 120 at 2.2 , and 164 at 1.6 . the k - band resolution is rather better than the nominal value because of the ao correction which resulted in a psf which had an intrinsic fwhm rather less than 2 pixels across . hence the spectral resolution is limited by the psf rather than simply the slit width . flux calibration for the h - band data could not be performed using a standard star . instead we use the h - band point source magnitude for mkn231 of h=10.09 from 2mass . extrapolating the continuum profile , assuming azimuthal symmetry , over a filled circular aperture with a diameter of 2 ( which contains effectively all the flux we have detected ; see fig . [ fig : profiles ] ) indicates that the flux falling into the slit is a factor of 5.1 less than would be in such an aperture . therefore we assign a magnitude of 11.86 to the flux detected in the central 2 along the slit . similarly , the magnitude for a 1 circular aperture is h=10.19 . a short sequence of images , totalling 20sec , were taken in the k - band to assess the size of the nucleus of mkn231 during adaptive optics wavefront correction . this was measured to be 85mas fwhm , including loss of resolution from undersampling and sub - pixel shifting to align the frames . a similar sequence of images taken on the standard star , while the ao system was running with the same control parameters ( although with 200counts rather than 60 , even after dimming the star with a neutral density filter ) , led to the same measurement of the fwhm and an estimate of the strehl ratio to be @xmath17% in the k - band . thus , although the final images are not diffraction limited , this is purely a result of the data acquisition procedure , and the nucleus of mkn231 is not resolved in the k - band on this spatial scale . a second source about 0.19 to the north has been reported by @xcite who claimed to have seen the same source in deconvolved j- , h- , and k - band images . in the k - band it had a relative intensity of a few percent . the adaptive optics psf for the keck telescope is very complex , and because of pupil rotation it was not possible to perform a satisfactory deconvolution of mkn231 for our data . the feature is not included in the spectroscopic data because the slit orientations were such that neither overlapped with the position where it should be . we are therefore unable to comment on this feature . the h - band continuum profiles along the slit are shown in the top panels of fig . [ fig : profiles ] . at both position angles these have been fitted with a combination of a gaussian and a moffat function . this provides an excellent match to the spatial profile because the gaussian represents both the core of the adaptive optics psf and the bright unresolved point source of the continuum related to hot dust around the agn ; while the moffat represents jointly the halo of the psf and the faint extended continuum component , which has wide shallow wings at a radial offset of 1 , the continuum is @xmath180.5% of its peak intensity . at pa @xmath15 , the fit is remarkably good and has an overall fwhm of 0.192 . the moffat profile has a fwhm 0.41 consistent with both the expected size of the halo for the ambient seeing and the effective radius of the de vaucouleurs fit to imaging data @xcite . it is similar to the broad component at the other position angle and also those measured in the continuum of the k - band spectra . the gaussian component , representing the core of the psf , has a fwhm 0.14 . at a pa @xmath16 the gaussian component has a slightly smaller width giving an overall fwhm of 0.160 , consistent with a better ao correction while these data were taken . however , there is a residual at an offset of @xmath19 which has a peak intensity 18% of the nucleus . we return to this in the discussion about the co absorption in the next section . it is perhaps surprising that such a feature was not seen in the _ pueo _ images of @xcite . however , in the k - band where much of their spatial analysis was performed , the relative strength of this feature would certainly be much less . assuming the agn component of the continuum emission is hot dust at @xmath181000k ( consistent with the slopes of the h- and k - band continua ) and that the feature has a spectrum typical of late type stars , its strength in the k - band relative to the peak would be reduced to only 34% . in fig . [ fig : contk ] we do in fact see a residual at on offset of about @xmath20 in spatial profile of the k - band spectrum at pa @xmath16 with about this relative strength ; but without the much clearer h - band data it would not be considered significant . such a weak feature is not expected to be detected so close to the nucleus in k - band imaging data without prior knowledge or very careful deconvolution techniques . under the assumption that the mean stellar type dominating the h - band stellar continuum is the same across the entire region where stellar absorption features can be seen , the absorption flux ( in contrast to the equivalent width ) of the absorption features traces the spatial extent of the stellar cluster . we have measured this using the 1.62 co6 - 3 bandhead absorption ( by comparing the absorption in the range 1.6171.623 with the continuum in the range 1.6231.629 ) . if there were no hot dust emission associated with the agn , one would expect it to be similar to the continuum profile . however , fig [ fig : profiles ] shows that the profile is indeed substantially different to that of the continuum primarily in not having a strong narrow peak in the core . this results in a sharp reduction in the equivalent width of the co bandhead to @xmath21 at radii smaller than 0.1 ; at radii in the range 0.10.5 ( beyond which the signal to noise becomes too weak to continue ) , @xmath22 is approximately constant . in late type giant and supergiant template stars @xcite , we find that typically @xmath23 , as shown in table [ tab : stars ] . once such stars appear in a cluster , they will dominate the near infrared stellar light : in the h - band , a k2i star is about 10 times brighter than a o9i star . hence , our value for @xmath24 implies that dilution due to hot dust is a factor of at least 13 in the core ( on scales of 0.1 ) and still a factor 56 around it ( out to @xmath25 ) . since a partially corrected adaptive optics psf has strong extended emission on scales similar to the seeing , much of the dilution on these scales may be due to the halo of the psf and hence attributable to the agn rather than being associated with the underlying and surrounding stellar population . if this is the case , then in a 2 length of the slit ( which includes essentially all the near - infrared flux , and for which @xmath26 ) , only 1/8 of the h - band light originates in star formation . extrapolating to a filled circular 2 aperture , we estimate that approximately 2025% of the h - band light is stellar . although the total h - band magnitude in this aperture is 10.09 , the stellar contribution is therefore 11.72mag . a distance modulus of 36.15 gives an absolute magnitude for the stellar light of m@xmath27 . we have applied no extinction correction to this value , because all estimates of the extinction suggest that it is low : optical and near infrared hydrogen recombination line ratios suggest it is in the range @xmath286mag @xcite , consistent with conclusions that the inner disk of mkn231 is close to face - on and hence relatively unobscured @xcite . instead we note that in the following discussion , applying any extinction correction would increase the h - band luminosity , with the result that both the mass and the bolometric luminosity of the star forming region would increase and hence the upper limit of the permissible age range would decrease . whether the stellar light is dominated by active star formation or an old population of stars can be determined from its mass to light ratio . using population synthesis models for m@xmath29 and v - h from starburst99 @xcite , one can calculate the mass of stars for a given age for both instantaneous and continuous star forming scenarios ( we assume solar metallicity and a salpeter imf ) . given that the stellar ` cluster ' ( note : for convenience we use the word ` cluster ' for the nuclear star forming region even though it is not known whether the stars are distributed continuously , or if they actually exist in many compact clusters ) is extended over about 0.40 ( see below ) , equivalent to 330pc , it seems unlikely that all the stars were formed in a single instantaneous burst . we therefore consider only continuous star forming scenarios . the minimum possible mass of the stellar cluster occurs when late type supergiants dominate the h - band continuum , at an age of about 10myr . [ fig : sb99 ] shows that to reach m@xmath27 at this age requires a star formation rate of @xmath30yr@xmath31 , yielding a total mass is @xmath1 and a light / mass ratio of 830@xmath4/. the absolute upper limit to the age is set by the fact that the stellar mass can not exceed the dynamical mass in the same region . as shown in table [ tab : masses ] , the stellar mass is unlikely to exceed @xmath32 . hence , from fig . [ fig : sb99 ] , the stellar cluster must be less than 120myr old , over which timescale the average star formation rate would have to be @xmath33yr@xmath31 , and have a lower light / mass ratio 125@xmath4/. @xcite detected 1.4ghz radio continuum from a @xmath34mas disk with a flux density of 130mjy ( see fig . [ fig : slits ] ) . further analysis of this and other 522ghz radio continuum data at various resolutions were presented by @xcite . following @xcite , these two sets of authors estimate the associated star formation rate for stars with masses @xmath35 to be @xmath36115yr@xmath31 ( adjusted to our adopted distance for mkn231 ) . extending the initial mass function down to 1 will increase the star formation rate by about a factor of two . alternatively , one can use starburst 99 models to derive the star formation rate from the supernova rate @xmath37 , which the radio continuum yields directly since the flux at 1.4ghz is likely to be dominated by non - thermal emission ( indeed the thermal continuum related to ionization by young stars would be only about 10% of that measured for the star formation rate above ) . using the galactic relation between the non - thermal radio luminosity and @xmath37 @xcite yields @xmath38yr@xmath31 and hence a star formation rate of order 200yr@xmath31 . these estimates agree well with those above based on the near infrared co absorption and h - band luminosity . one important question for mkn231 , which has repeatedly been addressed through the literature , concerns the respective fractions of the bolometric luminosity due to nuclear star formation and the agn . previously indirect estimates suggest that about 1/3 of the far infrared luminosity originates in nuclear star formation : @xcite used the 1 - 0s(1 ) and an analogy to the circumnuclear ring in ngc7469 for which the bolometric luminosity due to stars is known @xcite ; @xcite assumed that the stars are in a disk and used the stellar velocity dispersion and point source subtracted k - band image to derive a mass to light ratio and hence a bolometric luminosity . an alternative , albeit empirical , method is to use the far infrared radio correlation with the radio continuum measurement from @xcite . these authors used the correlation to determine whether star formation is a reasonable explanation for the radio continuum , and derived the ratio between the far infrared and radio fluxes to be @xmath39 . the median value is @xmath40 at 1.4ghz @xcite , which would imply @xmath41 . this is about 70% of the observed far infrared ( 40120 ) luminosity of mkn231 , and about 25% of the 81000luminosity . we are able to use a more direct method , based on the co6 - 3 absorption which arises in the stars themselves . the limits on the star formation rate and age we have discussed above constrain the bolometric luminosity of the cluster to be @xmath4212.11 ( the lower luminosity corresponding to the greater age ) . the bolometric luminosity of mkn231 ( taken as the luminosity in the range 8 - 1000 ) is @xmath43 @xcite ) . hence the star formation within 300pc of the agn is responsible for 2540% of the galaxy s luminosity . the range would reduce considerably if the mass of the stars could be constrained more tightly . for this calculation we have assumed that the star formation is continuous . once it ceases , the bolometric luminosity will fall very fast , and the cluster will fade by an order of magnitude over a timescale of up to a few 100myr . repeating the same calculation for smaller apertures shows that within 0.5 ( 165pc , similar to the fwhm of the stellar cluster ) of the agn , stars are still responsible for 1530% of @xmath44 ; while within a radius of 0.25 the fraction drops to 1015% because many of the stars are then excluded . we have fitted three analytical functions to the co absorption profiles out to radii of 1.5 in order to assess how the luminosity ( and hence density ) of the stellar cluster changes with radius . at pa @xmath16 , a small section around the residual peak seen in the continuum was omitted in order to avoid biassing the fit . the simplest profile is a gaussian ( blue curves in lower panels of fig . [ fig : profiles ] ) . however , the best fits with fwhm 0.50 and 0.60 , at pas @xmath15 and @xmath16 respectively , underestimate both the core and the wings of the measured profiles and do not yield good matches . it shows definitively that the spatial profile of the stellar cluster has a sharper cusp and broad shallow wings . we therefore consider two further profiles which have a physical meaning when applied to stellar distributions . the de vaucouleurs @xmath9 profile provides an empirical match to the luminosity profile in elliptical galaxies and other spheroids . the functional form @xmath45}\}$ ] has been convolved ( in 2 dimensions ) with a gaussian which represents our best estimate of the beam smearing ( fwhm 0.192 and 0.160 as described in section [ sec : im ] ) . these provide acceptable fits to the inner part of the profiles , but tend to overestimate the wings at radii beyond 0.5 ( red curves in fig . [ fig : profiles ] ) . the derived effective radii are also surprisingly large , @xmath46 and 4.32 ( 1.4 and 3.5kpc ) . the final function is an exponential profile typical of galaxy disks , which has the form @xmath47 ( note that the effective radius , inside of which half the total luminosity is emitted , is a factor 1.68 times the disk scale length @xmath48 ) . this was also convolved with a guassian representing the beam smearing , and provides acceptable fits ( green lines in the same figure ) over the whole co profile at both position angles . the resulting disk scale lengths are 0.18 and 0.24 ( 150 and 200pc ) , consistent with the size scales over which we have data . which of these two profiles best represents the data has important implications on our understanding of the geometry of the nuclear star forming region : in essence , whether it is spheroidal or a disk . however , it is not immediately clear which is correct . previous measurements of the radial profile of the h- and k - band continua ( after subtracting a nuclear point source ) yield very clear signatures of an @xmath9 law , but with small effective radius @xmath49 and @xmath50 ( 400 and 200pc ) respectively @xcite . in order to reconcile this result with our fits above , we consider also the h - band continuum profile from our spectroscopic data , which is shown in fig . [ fig : dvex ] . to avoid contamination by the agn component we have fit the profile at radii 0.252.5 , in which region it clearly follows the @xmath9 de vaucouleurs function ( dark blue line ) . the effective radius we derive is @xmath51 ( 270pc ) , consistent with the 0.5 ( 400pc ) found by @xcite . however , if we extrapolate this profile back in towards the nucleus ( convolved with our estimate of the psf as earlier ) , we find that it accounts for all of the continuum ( cyan line ) although with some small differences to the observed profile . yet since we know that most of the nuclear continuum is associated with the agn rather than stars ( e.g. from @xmath24 ) , this indicates that an @xmath9 profile with this @xmath52 can not extend all the way in to the nucleus . the same figure also shows the profile of the co absorption flux to radii of 1 together with the @xmath9 fit ( red line ) and the exponential fit ( green line ) from above . it can be seen that the large @xmath46for the former is needed to match the broad shape of the profile . but such a large @xmath52 is inconsistent with that derived for the continuum ( which at large radii @xmath53 is certainly dominated by stellar light ) , since it would imply that @xmath24 increases rapidly and continuously with radius and hence would require a strong gradient in the stellar population . additionally , at radii greater than about 0.5 , the fit appears to begin diverging from the data , although this is close to the noise limit . accordingly , we are led to exclude the @xmath9 fit to the co data . on the other hand , the exponential fit matches the co data well , and does not contradict the continuum profile . our conclusion here is therefore that at radii greater than @xmath181 the continuum traces the luminosity profile of the stellar population , which follows a @xmath9 profile . [ fig : dvex ] hints that there may be a small break in the slope of the profile , and that at radii larger than @xmath180.8(600 - 700pc ) @xmath52 is slightly larger than the value we found above . the result of this would be that at smaller radii , the @xmath9 profile no longer dominates the continuum . instead , as shown by the co data , the stellar light here is dominated by a bright nuclear population no more than 10100myr old , which is present in a disk and hence has an exponential profile . at very small radii , the continuum traces the unresolved hot dust emission associated with the agn . the fact that the continuum appears to follow a single @xmath9 law at all radii less than 2.5 is an unfortunate coincidence , since it hides the true stellar luminosity profile at small radii . a similar configuration is seen in ngc1068 ( @xcite , davies et al . in prep . ) , but in that case the break between the @xmath9 profile and the nuclear exponential star forming region occurs at a radius of @xmath1870pc , and is much clearer . at pa @xmath16 , a strong residual is apparent at an offset of @xmath54 with a peak intensity of 32% of the central point . the position is similar to the residual seen in the continuum . however , this can not be a peculiarity of the adaptive optics correction for two reasons : ( 1 ) the equivalent width of the co is more than a factor 2 higher than that at the nucleus ; ( 2 ) the residual is much narrower than the main profile ; if a scaled replica of the main profile were to be added instead , the residual would be far less distinct and apparent only as an asymmetry . we conclude that this is indeed a real feature , and propose that it may be an unresolved circumnuclear star cluster residing 120140pc from the agn . in fact , fig . [ fig : slits ] shows that the 1.4ghz radio continuum @xcite also has a subsidiary peak at about this position , providing additional evidence that it is a real feature . we estimate that about 1% of the co absorption in the nuclear star forming region originates in this feature . if the star formation here has the same history as discussed above , the mass would be of order @xmath55 . while this is a large mass for a single star cluster , it is only a factor of a few greater than the most massive clusters seen in the antennae @xcite and henize2 - 10 @xcite , and is entirely consistent with the larger cluster masses found in ngc6745 @xcite , arp220 @xcite , and ngc7252 @xcite . mkn231 appears to be a later stage merger between gas rich spiral galaxies . in a study of 18 late stage ulirg mergers , @xcite and @xcite showed characteristics of relaxed stellar populations with @xmath9 luminosity profiles and stellar dynamics dominated by dispersion rather than rotation . @xcite showed that although mkn231 itself is unusual in having a lower than average velocity dispersion ( only 115 ) , it still has a @xmath9 luminosity profile and a low value of @xmath56 . the spheroidal mass enclosed within the effective radius of 300pc would then be @xmath10 . and correlations of @xmath57 with the black hole mass would yield @xmath11 , which leads to a super - eddington agn luminosity . this result led these authors to consider whether the stars are confined to the same disk that the molecular gas occupies . in section [ sec : starform ] we have shown that the nuclear stellar population is very young and has an exponential profile . both of these results tend to support the conclusion that the stars are in a disk . however , before we continue in this vein , we need to consider whether this is consistent with the stellar dynamics . the disk properties determined by @xcite are an inclination @xmath58 , a rotation velocity of @xmath59 at a radius of 0.6 ( 500pc ) , a dynamical mass of @xmath60 within the same radius , and a scale height ( fwhm ) of 23pc . the actual rotation velocity at 500pc is then 345 . using the usual equation for calculating the velocity dispersion perpendicular to the disk plane @xmath61 where @xmath62 is the mass surface density of the disk and @xmath63 is the exponential scale height ( for which we approximate @xmath64fwhm ) , we find a mean value of @xmath65 . hence , @xmath66 , typical of rotationally supported systems . that @xmath57 is rather larger than one might expect for typical galactic disks , for which @xmath57 is a few times 10 to about 100 @xcite , is simply due to the large surface density of the disk , on average @xmath67pc@xmath68 . it seems therefore that it may be possible to reproduce a high dispersion and low rotation velocity from a disk . to test this more thoroughly we have made a model of an exponential disk having @xmath69 and other characteristics consistent with those above . we have convolved it with seeing of 0.6 and extracted line profiles from a 0.6 slit oriented at @xmath15 . integrating from @xmath70 to @xmath71 yields a line width of @xmath72 ; and the range 0.30.7 yields @xmath73 . thus we find an apparent @xmath74 , consistent with the measurements of @xcite and showing that a nearly face - on stellar disk can masquerade as a spheroid if it is sufficiently dense and the details are blurred by seeing . the caveat is that in this case , the correlation between @xmath57 and @xmath75 , which is derived for dynamically hot systems , does not apply since most of the motion is in rotation . the gas dynamics were determined using the 2.12 h@xmath2 1 - 0s(1 ) and the 1.64 [ feii ] emission lines . the long exposures needed to detect the more extended ( i.e. out to radii of 0.5 or more ) line emission resulted in the central few rows of the k - band spectrum being saturated ; due to the sub - pixel shifting associated with straightening the spectral traces , it is not possible to use the central 6 rows . in the h - band , the errors in the line flux and velocity across the central few rows of the spectrum are too large for these data to be useful . we are therefore unable to measure the shape of the rotation curve closer than 0.12(100pc ) to the nucleus . we have also measured the stellar dynamics using the h - band absorption features . however , while these are consistent with the gas dynamics , the error bars are too large to draw any independent conclusions . the h@xmath2 and [ feii ] emission line velocity profiles are shown in figs . [ fig : h2vel ] and [ fig : fevel ] respectively ( centre panels ) , together with the flux distribution ( top ) and the velocity dispersion ( bottom ) . these parameters were all measured by fitting a gaussian to the emission line profile in each row of the spectrum . the velocity zero points were calculated as the weighted mean velocity , giving a value close to 12680 in each line and position angle . uncertainties were calculated simultaneously for all 3 parameters by constructing an 11 pixel segment , which had the same statistical noise as the original spectrum ( after subtracting the fitted gaussian ) , and back to which the fitted gaussian profile was added . a new gaussian profile was then fitted to the segment , producing slightly different parameters . the procedure was repeated 1000 times , yielding the errorbars shown in the figure . the stellar velocities were measured by cross - correlating spectra in the range 1.571.64 ( rest frame 1.5061.574 ) , extracted from each row of the dataset , with a k1.5i template from @xcite ; repeating the procedure with templates of different late stellar types gave the same results . a gaussian was fitted to the cross - correlation peak . as for the emission lines , the errorbars were estimated by measuring the residual noise in the neighbouring 51 pixels , creating a segment with the same statistics to which the original gaussian was added back , and refitting the result . while this method does not permit the velocity dispersion to be measured , it does avoid the necessity of decovolution which can be uncertain and adds significant noise . the velocities and their uncertainties are plotted in fig . [ fig : covel ] . that the cold molecular gas lies in a thin rotating disk has been shown previously @xcite . where we have been able to measure both the h@xmath2 and [ feii ] emission lines , the velocities and dispersions are remarkably consistent and show a clear rotational signature which extends inwards to @xmath76 , as close to the nucleus as we can measure . thus it appears that these lines also originate in the disk structure . the stellar velocities are consistent with the same rotation , which is to be expected if , as we have argued , the stars lie in the same disk as the gas a scenario that would be feasible if the stars were recently formed in the gas disk . as a consistency check on our measurements of the emission lines , we have calculated an ` effective ' velocity dispersion from our data in a similar aperture to that used by @xcite who , for mkn231 , summed data at radii 0.30.7 either side of the nucleus . we have first calculated the mean velocity difference from one side of the nucleus to the other , and then combined this in quadrature with the mean velocity dispersion in the same region . at pa @xmath15 ( close to the kinematic major axis , see section [ sec : sdm ] ) we find @xmath77 and @xmath78}$ ] are both approximately 120 , consistent with the stellar velocity dispersion reported by @xcite . at pa @xmath16 ( where the rotational signature is weaker ) we find @xmath79 and @xmath78 } = 85 $ ] , both less than the value given by these authors . although these results suggest that both the h@xmath2 and the [ feii ] lines trace the kinematic characteristics of a rotationally supported disk in the nucleus of mkn231 , we consider also whether they might be affected by shocks or outflows . the nuclear radio continuum of mkn231 has been resolved into a triple structure oriented north - south and extended over a distance of 0.05 ( 40pc ) @xcite . an additional component 0.1 ( 90pc ) south of the core was reported by @xcite . because the ends of these structure are elongated perpendicular to the direction to the core , they have been interpreted as the working surfaces of jets from the agn . the core itself has also been resolved on parsec scales , showing two or three components aligned with a more east - west orientation @xcite . by analogy with ngc1068 , one might expect these radio jets to be associated with strong [ feii ] emission excited at the interface between the jet and the interstellar medium @xcite . this may be the case , but it can not related to the [ feii ] emission we have observed , which lies at much greater distances ( more than 100pc ) from the agn . it therefore seems unlikely that the extended [ feii ] has an origin in outflows . on the other hand , it is co - spatial with the nuclear star forming region , and therefore may instead originate in supernova remnants . a star formation origin for the extremely high optical feii luminosity of mkn231 has already been proposed by @xcite . our confidence in this conclusion is strengthened by the similarity of the kinematic signatures of the h@xmath2 and [ feii ] lines which , again by analogy with ngc1068 , one would not necessarily expect to be the case if they were produced in outflows rather than stellar processes . there are in fact only two places where the emission line kinematics differ significantly . one of these is the difference in velocities for small negative offsets at pa @xmath15 . the excess in both line fluxes which is also seen at an offset of @xmath80 along this position angle may be connected to this disparate kinematic behaviour . the second occurrence is the difference in dispersions at positive offsets along the same position angle . here the errors in the dispersion for the [ feii ] line are relatively large , and the difference from the h@xmath2 dispersions is therefore not highly significant . there is only one remaining unexpected aspect . at pa @xmath16 the velocities show a symmetrical decay at increasing radii from @xmath81 to zero within 0.30.4 . this is reminiscent of keplerian rotation about an unresolved mass . however , the decrease in velocity is steeper than @xmath82 , and we discuss it further in section [ sec : warp ] . with only two slit positions , the inclination angle and kinematic major axis are unconstrained by our data . instead , we fix them as input parameters to our model using estimates based on interferometric measurements from the literature of the millimetre co emission line and radio continuum . for the kinematic major axis , we adopt the value from @xcite , who used a [email protected] synthesised beam to measure the properties of the co2 - 1 line ( the highest resolution data available for co emission ) . they found the kinematic line of nodes to be oriented at @xmath84 ( i.e. east - west ) , similar to that determined by @xcite in slightly lower resolution data . for the inclination angle , various estimates exist . based on the morphology of the nuclear radio continuum , @xcite estimated @xmath85 . however , we have argued in section [ sec : starform ] that the elongation in their map is due to additional massive star clusters which are not intrinsically part of the nuclear disk . excluding these makes the remaining emission much more symmetrical , suggesting a low inclination angle . such a conclusion is supported by the k - band adaptive optics image in @xcite which has a circular appearance . @xcite reported for their co2 - 1 data that in order to reproduce the spectral profile and the symmetrical morphology , to avoid the gas mass exceeding the dynamical mass , and to be consistent with the lack co absorption towards the nucleus , the inclination had to be less than 20@xmath86 from face - on . additionally , to avoid an excessively high rotation speed ( i.e. rather higher than all other galaxies in their sample ) , the inclination could not be much more face - on than 10@xmath86 . because of the dependence of mass on @xmath87 , this uncertainty equates to a factor of 4 in the dynamical mass . we therefore repeat our analysis using the two limiting values of 10@xmath86 and 20@xmath86 . to create a model of the mass distribution and kinematics , we have used a method outlined in @xcite that has also successfully been applied to ngc7469 @xcite . we assume that the mass is distributed in a thin disk ( see @xcite and @xcite for h@xmath2 , and @xcite for hi ; section [ sec : starprof ] for the stars ) ; that the galaxy is inclined by either 10@xmath86 or 20@xmath86 at a position angle of 90@xmath86 ; and that the disk comprises 3 components : a nuclear stellar cluster with exponential luminosity and mass profile as derived in section [ sec : starform ] , a uniform gas disk as described by @xcite ( note that these authors also describe a 3 outer disk , but since we are only interested in radii @xmath88 , we do not include this explicitly ) , and a black hole . a typical rotation curve that results from optimising the masses is shown in figs . [ fig : h2vel ] and [ fig : fevel ] . given that the gas dynamics could be affected by non - gravitational motions , the model yields a reasonably good representation of the velocities and dispersions for both lines at both position angles . for an inclination of @xmath89 our dynamical mass is only @xmath1 within 0.6 . this inclination is clearly too high since it results in a mass which is barely the minimum we have estimated for the young stars alone based on their luminosity , and leaves no room for the gas mass . including the gas mass of @xmath90 of @xcite , would suggest a maximum inclination of @xmath91 from face - on . at @xmath58 , the inclination preferred by @xcite , we find the enclosed mass within a radius of 0.6 is @xmath92 . that this is rather less than the mass determined by them is due to the different rotation velocities measured at this radius ( but note that their beam size was @xmath93 ) . our and their measurements differ by 20 , which , because mkn231 is close to face - on , corresponds to a factor of 1.5 in the rotation curve and hence a factor 2 in the derived dynamical mass . the measured velocity dispersions are reproduced by imposing a disk thickness of 15pc ( gaussian fwhm ) , similar to that estimated by @xcite . the increase in velocity dispersion towards smaller radii is due to the rising mass surface density of the disk , which is on average 8000pc@xmath68 . if we repeat our earlier experiment of making the same measurements as @xcite ( i.e. 0.6 seeing , and extracting the velocity and dispersion over lengths of a 0.6 slit ) , we find @xmath94 and @xmath95 . while the former is in good agreement with their rotation velocity , the dispersion is somewhat less , indicating there there remains some uncertainty about the disk thickness and perhaps also the mass . the black hole mass determined by the model for @xmath58 is @xmath96 , similar to the mass of stars and gas within a radius of 0.1 , indicating that this is approximately the ` sphere of influence ' of the black hole ( within which it dominates the dynamics ) . in our data this region is not well resolved , but if one could reach the 0.05 diffraction limit of a large telescope by better sampling in the k - band or achieving better ao correction ( e.g. with a laser guide star ) in the h - band , one could impose a much stronger constraint on the black hole mass . the velocity profile at pa @xmath16 shows a symmetrical decay at increasing radii from @xmath81 to zero within 0.30.4 . this is reminiscent of keplerian rotation about a compact mass . however , as the right panel in fig . [ fig : velcurve ] indicates , the decay is in fact faster than r@xmath68 and so can not be explained by any mass distribution confined to a planar disk . such a curve can in principle be produced by a warp in the inner disk of mkn231 , which would have the effect of twisting the inner part of the disk to a different inclination and position angle . in section [ sec : starprof ] we have already encountered a hint that this may be the case : the effective radius we derive for the exponential profile is significantly larger at pa @xmath16 than at pa @xmath15 , opposite to that expected if the kinematic major axis lies at about 90@xmath86 . although the angular scales are small , this can be easily seen in fig . [ fig : profiles ] . it is not an effect of possible differences in the adaptive optics correction because the scale sizes and their difference are much larger than those of the psfs ( in fact , both results are independent of which psf estimate we use ) . this suggests that , at least for the stellar disk and unless it is strongly asymmetrical , the major axis lies closer to @xmath16 than @xmath15 . while it may be difficult to warp a stellar disk once it has formed , we are proposing that the stars have formed very recently in the gas disk . as a result , it is certainly plausible that the warp occured in the gas disk before the stars formed . even combining the kinematic and spatial data , the geometry of a putative warp can not be constrained . here , therefore , we consider only an illustrative case , which demonstrates the feasibility of this hypothesis in explaining the observations . as the basic starting point , we have taken the same mass distribution as derived in the previous section , and set the inclination to be @xmath58 with the major axis at @xmath97 ( slightly different to the @xmath84 previously adopted ) . we define the position angle of the warp in the disk plane as measured from an axis in that plane orthogonal to the major axis ( i.e. if the disk major axis runs east - west then , if the disk is deprojected , the position angle of the warp is measured east of north ) . at a position angle of 30@xmath86 and for radii @xmath98 , we tilt the disk out of its plane by 10@xmath86 . beyond this , the tilt declines uniformly with increasing radius , until it reaches zero for radii @xmath99 . the observable effect this slight warping of a nearly face - on disk is in fact to shift the major axis of the inner part of the disk more towards @xmath16 without greatly altering the effective inclination . for this example , the inner part of the disk would have an effective inclination of @xmath100 but at a position angle of @xmath101 . hence , as shown in the right hand panel of fig . [ fig : velcurve ] , at pa @xmath16 , the signature of rotation is seen at small radii but not at large radii . comparison to fig . [ fig : covel ] suggests that the stellar velocity curve might also match this model better , again a result which one might expect if the stars have formed recently in a disk which was already warped . finally , our warp model is also able to reproduce approximately the velocity contours presented by @xcite with a 0.6 beam . we have provided an example of a weak warp which can explain the kinematics we have observed . however , measurements of both hydroxyl and the radio continuum provide evidence that there may be much stronger warping occuring in the nucleus of mkn231 . the velocity field of the 1.6ghz oh line emission , observed at a resolution of 3040mas over an extent of 100150mas by @xcite , has been modelled in terms of a thick disk or torus inclined by 56@xmath86 at a position angle of 35@xmath86 . this orientation , on scales of 100pc , is rather different from that on scales of @xmath102pc traced by the molecular gas , and is indicative of a warp completely consistent with but more pronounced than that described above . as we noted earlier , radio continuum data suggest that there is a very pronounced shift in the direction of the radio jet at radial scales between 1 and 20pc @xcite . unless the jets are viewed nearly end - on , this represents a substantial mislignment of the jet axis . whether the misalignment is due to deflection of the jets as it strikes dense interstellar medium , or from a twisting of the axis of the central agn , it could plausibly be related ( whether causally or not ) to the warping of the nuclear disk . two such possibilities are that ( 1 ) the jet impinges on the disk and has caused it to warp , at the same time changing the direction of the jet ; ( 2 ) the spin axis in the core has changed , perhaps during the later stages of the merger as clumps of gas thrown off in the early stages are accreted , and this is reflected in both the altered jet direction and the kinematics of the disk . the consequence of a warp is to make modelling of the small scale kinematics rather more complex . however , with the right data it may also make it easier to measure the dynamical effect of the black hole because for such a low inclination of the disk , small uncertainties in either the measured kinematics or the inclination itself propagate into very large uncertainties in the true rotation curve and hence enclosed mass estimate . a warp will almost inevitably lead to the inner region having a larger inclination , and hence reduce the sensitivity of the derived rotation curve to uncertainties in these parameters . with the current data we are unable to constrain the geometry of the warp and hence are unable to make detailed statements about the black hole mass . it is expected that with either integral field spectroscopy at adaptive optics scales , or higher resolution interferometric data at mm wavelengths will help to resolve the situation . in the previous sections we have presented evidence that there is a massive ( log @xmath103 ) young ( @xmath104myr ) starburst occuring in the nucleus of mkn231 . the stars involved in this burst are distributed in the same disk plane occupied by the gas , which is nearly face - on . this evidence suggests that the h and k band stellar light is not tracing stars from the progenitor galaxies which have relaxed into a spheroid during the merger , but instead is dominated by stars which have formed _ in situ _ in the gas disk which itself has formed as a result of the merger and still remain there . it is important , therefore , to consider whether other observations are consistent with such a scenario . as we mentioned in section [ sec : lummass ] , the radio continuum supports the existence of a young active starburst . here we also look at the important characteristics observed with the _ infrared space observatory _ , in particular the mid infrared emission lines reported in @xcite and the polycyclic aromatic hydrocarbon ( pah ) features in @xcite . @xcite presented measurements of the 7.7 pah feature for 62 ulirgs , comparing them templates from 23 agn and 15 starburst and normal galaxies . they showed that agn and starbursts could be distinguished by the ratio of the pah strength at 7.7 to the continuum at 5.9 ( @xmath105 ) : agn have @xmath106 , while starbursts have @xmath107 . under this classification , mkn231 with @xmath108 , is clearly an agn . however , this does not mean that there is no starburst . the spectrum in fig . 2 of @xcite clearly shows a strong pah feature the low @xmath105 ratio arises simply because the continuum is also strong . indeed , comparison with the other ` starburst - like ' ulirgs in this study shows that the absolute stength of the pah feature is consistent with what one would expect from a massive starburst . @xcite did not detect 12.8 [ neii ] , a line that arises in hii regions and commonly seen in the mid infrared spectra of starburst and ulirgs . ne@xmath109 has an ionization potential of 21.6ev , and so is a tracer of the hottest stars . the limit of @xmath110wm@xmath68 is a little lower than the flux that might be expected , but on the other hand , one can not rule out that the true line flux lies within the range of typical values for massive young starbursts . a line that was detected is [ siii ] at 34 . the ionization potential for si@xmath109 is only 8.2ev and hence this line arises in hii regions , at the interface between the ionized and molecular gas . the flux measured was @xmath111wm@xmath68 , again similar to the other ulirgs . all the ulirgs in this sample had similar fluxes for the two lines ; and the detection of [ siii ] but non - detection of [ neii ] at 1/3 the flux level is not an implausible result , particularly given the difficulty of detecting the lines as a result of the low line to continuum ratio . a re - analysis of the data by a. verma ( private communication ) suggests that there is a possible @xmath112 detection of [ neii ] . one final line of evidence about this is the near to far infrared spectral energy distribution ( sed ) . this has been modelled in detail by @xcite , who found that a combination of nearly face - on torus model together with a zero age starburst was needed to match the 1 - 1000 sed . although the exact parameters are not uniquely constrained , this general conclusion was inescapable . we conclude that the radio continuum , mid infrared diagnostics , and sed all support the view that there is vigorous young star formation occuring in the nucleus of mkn231 . the superposition of this starburst and the agn have made mkn231 look similar to a quasar , but at the same time means that it does not fit any of the standard correlations . once the star formation ceases , and the starburst fades on timescales of one to a few 100myr the galaxy will probably look more like a typical seyfert 1 . evidence for the existence of young nuclear stellar disks in a number of spiral galaxies has been presented in the literature in terms of a drop in the stellar velocity dispersion on small spatial scales . @xcite measured the @xmath188600 caii triplet in 5 isolated seyfert spiral galaxies and found a dip in the velocity dispersion across the central 13 for 4 of them ( and a hint of such in the fifth ) . the drop was accompained by a possible local increase in the equivalent width of the caii triplet , indicating the presence of a younger stellar population . a similar effect has been previously reported for 3 barred galaxies by @xcite using the 2.29 co bandhead . this phenomenon has been modelled by @xcite using simulations which include stars , gas , and star formation . their conclusion was that the drop in the line - of - sight stellar velocity dispersion is the kinematical signature of stars formed in a dynamically cold gas disk , which forms as gas is driven to the nucleus by the bar . because these stars are young , they out - shine the old ( dynamically hotter ) population in the nucleus . and the effect is enhanced because the high gas density in the very centre leads to more efficient cooling , reducing further the velocity dispersion of the gas itself , and hence also the stars formed from this gas . thus , although the frequency of such nuclear disks is unkown , it may be a relatively common phenomenon in galaxies where gas is driven to the nucleus whether this be by a bar in a spiral galaxy as for the examples above , or due to a merger event as we have seen for mkn231 . we have presented adaptive optics spectroscopic data of the nearby ultraluminous infrared galaxy and qso / seyfert1 mkn231 in the h- and k - bands at resolutions as small as 0.085 . 1 . we have resolved the region of active star formation in the nucleus of mkn231 through the 1.62 co absorption in late type stars . the region has a fwhm of 0.350.40 ( 300pc ) , and has a spatial profile significantly more cusped and with broader wings than a gaussian . it is fit well by an exponential profile with disk scale length @xmath1130.24 . at larger radii the continuum luminosity profile is well represented by an @xmath9 law , as one might expect , since during a merger the stars relax into a spheroid ; but that this also appears to continue inwards is an unfortunate coincidence . the nuclear star forming region would have a minimum mass of @xmath1 if it is only @xmath114myr old ( i.e. it has just reached the age when late type supergiants dominate the near infrared stellar continuum ) , implying a star formation rate of @xmath30yr@xmath31 . constraints from the dynamical mass indicate that its age is unlikely to exceed 120myr . it is responsible for 2540% of the galaxy s entire bolometric luminosity , the higher fraction corresponding to the younger age limit . these results all point to the fact that the stars lie in a disky rather than spheroidal distribution , and that they are very young . hence it is likely that they have formed recently in the nearly face - on molecular gas disk , which is itself a product of the merger that resulted in mkn231 . 2 . there is an unresolved extranuclear star cluster @xmath115(120140pc ) from the nucleus , with a mass of @xmath116 . this is more massive than ` typical ' young star clusters in star forming galaxies , but is comparable with the most massive clusters seen in a few cases . the dynamics at radii 0.10.7 have been traced using the 2.12 1 - 0s(1 ) h@xmath2 and 1.64 [ feii ] emission lines , as well as the h - band stellar absorption features . the emission lines are extended over the same spatial scales as the stellar co absorption , which is much greater than the scale of the radio jets , and are therefore likely to originate in stellar processes . their kinematics are similar , and in general terms consistent with rotation of a nearly face - on disk . however , a simple dynamical model shows that there are distinct differences . the rotation curve at a position angle of @xmath16 shows a rapid decrease in velocity with radius , faster than @xmath82 . this can be explained if there is a warp in the inner 0.20.3(200pc ) of mkn231 , a hypothesis which appears to be supported by the orientation of the major axis of the stellar disk , by the pronounced shift on small scales of the radio jet axis , and by the velocity field of the @xmath117mas 1.6ghz oh line emission . integral field velocity maps at 0.1 resolution will be needed to constrain the geometry and properties of such a warp . the authors are grateful to the support of the staff of the keck observatory . the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community . we are most fortunate to have the opportunity to conduct observations from this mountain . we are grateful to a. verma for looking again at the iso data of mkn231 , and for making her work on its sed available to us . hr 7479 & g4 i & @xmath118 & @xmath119 & 12 & 1.3 + hr 8465 & k1.5 i & @xmath120 & @xmath121 & 13 & 4.5 + hr 1155 & m2 i & @xmath122 & @xmath123 & 19 & 5.8 + hr 7328 & g9 iii & @xmath124 & @xmath125 & 1.1 & 2.1 + hr 8317 & k1 iii & @xmath126 & @xmath127 & 1.1 & 2.7 + hr 165 & k3 iii & @xmath128 & @xmath129 & 1.2 & 3.3 + hr 4517 & m1 iii & @xmath130 & @xmath131 & 1.2 & 4.2 + young stars & @xmath132 & minimum mass possible for @xmath133 + molecular gas & @xmath134 & from co model of @xcite + dynamical , @xmath89 & @xmath135 & inconsistent with stellar and gas mass + dynamical , @xmath58 & @xmath136 & implies total ( old+young ) stellar mass @xmath137 +	we report adaptive optics h- and k - band spectroscopy of the inner few arcseconds of the luminous merger / ulirg / qso mkn231 , at spatial resolutions as small as 0.085 . for the first time we have been able to resolve the active star forming region close to the agn using stellar absorption features , finding that its luminosity profile is well represented by an exponential function with a disk scale length 0.180.24 ( 150200pc ) , and implying that the stars exist in a disk rather than a spheroid . the stars in this region are also young ( 10100myr ) , and it therefore seems likely that they have formed _ in situ _ in the gas disk , which itself resulted from the merger . the value of the stellar velocity dispersion ( @xmath0 rather than the usual few times 10 in large scale disks ) is a result of the large mass surface density of the disk . the stars in this region have a combined mass of at least @xmath1 , and account for 2540% of the bolometric luminosity of the entire galaxy . at our spatial resolution the stellar light in the core is diluted by more than a factor of 10 even in the h - band by continuum emission from hot dust around the agn . we have detected the 2.12 1 - 0s(1 ) h@xmath2 and 1.64lines out to radii exceeding 0.5 . the kinematics for the two lines are very similar to each other as well as to the stellar kinematics , and broadly consistent with the nearly face - on rotating disk reported in the literature and based on interferometric co1 - 0 and co2 - 1 measurements of the cold gas . however , they suggest a more complex situation in which the inner 0.20.3 ( 200pc ) is warped out of its original disk plane . such a scenario is supported by the projected shape of the nuclear stellar disk , the major axis of which is significantly offset from the nominal direction ; and by the pronounced shift on very small scales in the direction of the radio jet axis which has been reported in the literature .
the study of the properties of hadrons in hot and dense matter is receiving a lot of attention in recent years to understand fundamental aspects of the strong interaction , such as the partial restoration of chiral symmetry @xcite , as well as a variety of astrophysical phenomena , such as the dynamical evolution of supernovas and the composition of neutron stars @xcite . a special effort has been invested to understand the properties of antikaons in the medium , especially after the speculation of the possible existence of an antikaon condensed phase was put forward in @xcite which would soften the equation - of - state producing , among other phenomena , lower neutron star masses @xcite . while it is well established that the antikaons should feel an attractive interaction when they are embedded in a nuclear environment , the size of this attraction has been the subject of intense debate . theoretical models including medium effects on the antikaon - nucleon scattering amplitude , the behaviour of which is governed by the isospin zero @xmath4 resonance , naturally explain the evolution from repulsion in free - space to attraction in the nuclear medium @xcite . however , the presence of the resonance makes the size of the antikaon - nucleus potential be very sensitive to the many - body treatment of the medium effects . the phenomenological attempts of extracting information about the antikaon - nucleus potential from kaonic - atom data favoured very strongly attractive well depths @xcite , but recent self - consistent in - medium calculations @xcite based on chiral lagrangian s @xcite or meson - exchange potentials @xcite predict a moderately attractive kaon - nucleus interaction . in fact , recent analysis of kaonic atoms @xcite are able to find a reasonable reproduction of the data with relative shallow antikaon - nucleus potentials , therefore indicating that kaonic atom data can not really pin down the strength of the antikaon - nucleus potential at nuclear matter saturation density . on the other hand , heavy - ion collisions at energies around @xmath5 agev offer the possibility of studying experimentally the properties of a dense and hot nuclear system @xcite . in particular , a considerable amount of information about strange particles like antikaons is available . since antikaons are produced at finite density and finite momentum , the chiral models have recently incorporated the higher partial waves of the antikaon - nucleon scattering amplitude both in free space@xcite and in the nuclear medium @xcite . the complete scenario taking into account finite density , finite momentum and finite temperature has recently been addressed in refs . @xcite . event generators trying to analyse heavy - ion collision data @xcite need to implement the modified properties of the hadrons in the medium where they are produced . transport models have shown , for instance , that the multiplicity distributions of kaons and antikaons are much better reproduced if in - medium masses rather than bare masses are used @xcite . production and propagation of kaons and antikaons have been investigated with the kaon spectrometer ( kaos ) of the sis heavy - ion synchrotron at gsi ( darmstadt ) . the experiments have been performed with au@xmath6au , ni@xmath6ni , c@xmath6c at energies between @xmath7 and @xmath8 agev @xcite . one surprising observation in c@xmath6c and ni@xmath6ni collisions @xcite is that , as a function of the energy difference @xmath9 , where @xmath10 is the needed energy to produce the particle ( @xmath11 gev for @xmath1 via @xmath12 and @xmath13 gev for @xmath0 via @xmath14 ) , the number of @xmath0 balanced the number of @xmath1 in spite of the fact that in @xmath15 collisions the production cross - sections close to threshold are 2 - 3 orders of magnitude different . this has been interpreted to be a manifestation of the enhancement of the @xmath1 mass and the reduction of the @xmath0 one in the nuclear medium , which in turn influence the corresponding production thresholds @xcite , although a complementary explanation in terms of in - medium enhanced @xmath16 production has also been given @xcite . another interesting observation is that at incident energies of @xmath17 and @xmath18 agev , the @xmath0 and @xmath1 multiplicities have the same impact parameter dependence @xcite . equal centrality dependence for @xmath1 and @xmath0 and , hence , independence of centrality for the @xmath19 ratio has also been observed in au@xmath6au and pb@xmath6pb reactions between @xmath20 agev and rhic energies ( @xmath21 agev ) @xcite . this independence of centrality is astonishing , since at low energies one expects that as centrality increases and with it the participating system size and the density probed the @xmath19 ratio should also increase due to the increased reduction of the @xmath0 mass together with the enhancement of the @xmath1 mass . in fact , the independence of the @xmath19 ratio on centrality has often been advocated as signalling the lack of in - medium effects . a recent interesting interpretation of this phenomenon is given in ref . @xcite , where it is shown that the @xmath0 are predominantly produced via @xmath22 collisions ( @xmath23 ) and , hence , the @xmath0 multiplicity is strongly correlated with the @xmath1 one , since kaons and hyperons are mainly produced together via the reaction @xmath24 . although the transport model calculations show that strangeness equilibration requires times of the order of @xmath25 fm / c @xcite , statistical models , which assume chemical and thermal equilibrium and common freeze - out parameters for all particles , are quite successful in describing particle yields including strange particles @xcite . the kaon and antikaon yields in the statistical models are based on free masses and no medium effects are needed to describe the enhanced in - medium @xmath19 ratio or its independence with centrality . the increased value of the @xmath19 ratio is simply obtained by choosing a particular set of parameters at freeze - out , the baryonic chemical potential @xmath26 mev and the temperature @xmath27 mev , which also reproduce a variety of particle multiplicity ratios @xcite . on the other hand , centrality independence of the @xmath19 ratio is automatically obtained in statistical models within the canonical or grand - canonical schemes because the terms depending on the system size drop out @xcite . however , as shown by brown et al . in ref . @xcite , using the reduced in - medium @xmath0 mass in the statistical model would force , in order to reproduce the experimental value of the @xmath19 ratio , a larger value of the chemical potential and hence a larger and more plausible baryonic density for strangeness production . in addition , brown et al . introduce the concept of broad - band equilibration " according to which the @xmath0 mesons and the hyperons are produced in an essentially constant ratio independent of density , hence explaining also the centrality independence of the @xmath19 ratio but including medium effects . in essence , the establishment of a broad - band relies on the fact that the baryonic chemical potential @xmath28 increases with density an amount which coincides roughly with the reduction of the @xmath0 mass . however , as mentioned above , the antikaon properties are very sensitive to the type of model for the @xmath29 interaction used and of the in - medium effects included . the purpose of the present work is precisely to investigate to which extent the broad - band equilibration " concept holds when more sophisticated models for the in - medium antikaon properties are used . we explore within the context of a statistical model what is the behaviour of the @xmath19 ratio as a function of the nuclear density when the hadrons are dressed , paying special attention to different ways of dressing the antikaons : either treating them as non - interacting , or dressing them with an on - shell self - energy , or , finally , considering the complete antikaon spectral density . we use the antikaon self - energy which has been derived within the framework of a coupled - channel self - consistent calculation in symmetric nuclear matter at finite temperature @xcite , taking as bare meson - baryon interaction the meson exchange potential of the jlich group @xcite . we will show that the determination of temperature and chemical potential at freeze - out conditions compatible with the experimental value of the @xmath0/@xmath1 ratio is very delicate , and depends very strongly on the approximation adopted for the antikaon self - energy . the paper is organised as follows . in sec . ii the formalism of the thermal model is described and the different ingredients used in the determination of the off - shell properties of the @xmath0 are given . the results are presented and discussed in sec . finally , our concluding remarks are given in sec . in this section we present a brief description of the thermal models to account for strangeness production in heavy - ion collisions . the basic hypothesis is to assume that the relative abundance of kaons and antikaons in the final state of relativistic nucleus - nucleus collisions is determined by imposing thermal and chemical equilibrium @xcite . the fact that the number of strange particles in the final state is small requires a strict treatment of the conservation of strangeness and , for this quantum number , one has to work in the canonical scheme . other conservation laws must also be imposed , like baryon number and electric charge conservation . since the number of baryons and charged particles is large , they can be treated in the grand - canonical ensemble . in this way , the conservation laws associated to these other quantum numbers are satisfied on average , allowing for fluctuations around the corresponding mean values . to restrict the ensemble according to the exact strangeness conservation law , as done in refs . @xcite , one has to project the grand - canonical partition function , @xmath30 , onto a fixed value of strangeness @xmath31 , @xmath32 where @xmath33 are the baryon , strangeness and charge fugacities , respectively , and where @xmath34 stands explicitly for @xmath35 . only particles with @xmath36 are included in the grand - canonical partition function because , in the range of energies achieved at gsi , they are produced with a higher probability than particles with @xmath37 . the grand - canonical partition function is calculated assuming an independent particle behaviour and the boltzmann approximation for the one - particle partition function of the different particle species . in principle , one deals with a dilute system , so the independent particle model seems justified . however , medium effects on the particle properties can be relevant . as mentioned in the introduction , the aim of this paper is to study how the dressing of the hadrons present in the gas , especially the antikaons , affects the observables such as the ratio of kaon and antikaon particle multiplicities , in particular for the conditions of the heavy - ion collisions at sis / gsi energies . within the approximations mentioned above , the grand - canonical partition function reads as follows , @xmath38 where @xmath39 is the sum over one - particle partition functions of all particles and resonances with strangeness @xmath36 , @xmath40 the expressions @xmath41 and @xmath42 indicate the one - particle partition function for baryons and mesons respectively , while @xmath43 is the one - particle partition function associated to a baryonic or mesonic resonance . in the latter case , however , the factor @xmath44 would not be present in eq . ( [ reso ] ) . notice that the resonance is described by means of a breit - wigner parameterisation . the quantity v is the interacting volume of the system , @xmath45 , @xmath46 and @xmath47 are spin - isospin degeneracy factors and @xmath28 and @xmath48 are the baryonic and charge chemical potentials of the system . for ni+ni system at sis energies , @xmath48 can be omitted because it is associated to the isospin - asymmetry of the system and , in this case , the deviation from the isospin - symmetric case is only @xmath49 ( see ref . on the other hand , @xmath28 will be fixed to the nucleonic chemical potential , @xmath50 , because the abundance of nucleons is larger than the one for the other baryons produced . the energies @xmath51 , @xmath52 refer to the in - medium single - particle energies of the hadrons present in the system at a given temperature . following ref . @xcite , the canonical partition function for total strangeness @xmath53 is @xmath54 in this work , as well as in ref . @xcite , the small and large volume limits of the particle abundances were studied . these limits were performed to show that the canonical treatment of strangeness in obtaining the particle abundances gives completely different results in comparison to the grand - canonical scheme , demonstrating at the same time that , for the volume considered , the canonical scheme is the appropriate one . the aim of going through these limits again in the following is to remind the reader that , for the specific case of the ratio @xmath19 , the result is independent of the size of the system and is the same for both the canonical and grand - canonical treatments . according to statistical mechanics , to compute the number of kaons and antikaons @xcite one has to differentiate the partition function with respect to the particle fugacity @xmath55 expanding @xmath56 in the small volume limit , @xmath57 and @xmath58 are given by @xmath59 where antibaryons have not been considered because the ratio @xmath60 is negligibly small at gsi / sis colliding energies , where @xmath61 and @xmath62 represent the number of baryons and antibaryons , respectively . the expression for @xmath58 ( @xmath57 ) indicates that the number of @xmath1(@xmath0 ) has to be balanced with all particles and resonances with @xmath63(@xmath64 ) . it can be observed from eq . ( [ eq : smallvolume ] ) that the ratio @xmath0/@xmath65 in the canonical ensemble does not depend on the volume because it cancels out exactly . at the other extreme , i.e. in the thermodynamic limit ( large volumes ) , since it is known that the canonical treatment is equivalent to the grand canonical one , one can compute the ratio explicitly from the grand canonical partition function @xmath66 , @xmath67 where @xmath68(@xmath69 ) is the sum of one - particle partition functions for baryons ( mesons ) with @xmath70 . then , by imposing strangeness conservation on average , @xmath71 one can easily obtain @xmath34 , @xmath72 therefore , from @xmath73 one obtains the ratio @xmath74 the condition @xmath75 and dealing with strange particles of @xmath76 makes the ratio independent of the volume . although the expression obtained is the same as that from the canonical ensemble in the small volume limit , the proof that the ratio @xmath19 is independent of the volume has to be obtained from a general intermediate size situation . this was shown to be the case in ref . @xcite , where expanding @xmath77 of eq . ( [ canon3 ] ) in power series , it was expressed as @xmath78 , where @xmath79 is the partition function that includes all particles and resonances with @xmath53 , @xmath80 is the modified bessel function , and @xmath81 . the computed @xmath19 ratio gave precisely the same expression as those given here for small [ eq.([eq : smallvolume ] ) ] and large [ eq.([eq : largevolume ] ) ] volumes . therefore , as noted in ref . @xcite , the @xmath0/@xmath1 ratio is independent of the volume and , consequently , independent on whether it is calculated in the canonical or grand - canonical schemes . in this subsection we study how the in - medium modifications of the properties of the hadrons at finite temperature affect the value of the @xmath0/@xmath1 ratio , focusing our attention on the properties of the antikaons in hot and dense matter . for consistency with previous papers , we prefer to compute the inverse ratio @xmath1/@xmath0 . as it was mentioned before , the number of @xmath0 ( @xmath1 ) has to be balanced by particles and resonances with @xmath82(@xmath63 ) in order to conserve strangeness exactly . for balancing the number of @xmath1 , the main contribution in the @xmath63 sector comes from the @xmath83 and @xmath84 hyperons and , in a smaller proportion , from the @xmath0 mesons . in addition , the effect of the @xmath85 resonance is also considered because it is comparable to that of the @xmath0 mesons . on the other hand , the number of @xmath0 is balanced only by the presence of @xmath1 . then , we can write the @xmath1/@xmath0 ratio as , @xmath86 where the @xmath87 s indicate the different one - particle partition functions for @xmath0 , @xmath1 , @xmath83 , @xmath84 and @xmath88 , and , for baryons , they now contain the corresponding fugacity . it is clear from eq . ( [ eq : lamb - sig - k ] ) that the relative abundance of @xmath83 , @xmath84 and @xmath88 baryons with respect to that of @xmath0 mesons determines the value of the ratio . in order to introduce the in - medium and temperature effects , the particles involved in the calculation of the ratio are dressed according to their properties in the hot and dense medium in which they are embedded . for the @xmath83 and @xmath84 hyperons , the partition function @xmath89 is constructed using a mean - field dispersion relation for the single - particle energies @xmath90 for @xmath91 , we take the parameterisation of ref . @xcite , @xmath92 . for @xmath93 , we take a repulsive potential , @xmath94 , extracted from analysis of @xmath84-atoms and @xmath84-nucleus scattering @xcite , where @xmath95 @xmath96 is the saturation density of symmetric nuclear matter . a repulsive @xmath84 potential is compatible with the absence of any bound state or narrow peaks in the continuum in a recent @xmath84-hypernuclear search done at bnl @xcite . the @xmath85 resonance is described by a breit - wigner shape , @xmath97 with @xmath98 and @xmath99 . in the case of @xmath1 we take @xmath100 where @xmath101 is obtained from a @xmath102 approximation , as discussed in refs.@xcite . a particular effort has been invested in studying the antikaon properties in the medium since the @xmath29 has a particularly rich structure due to the presence of the @xmath4 resonance @xcite . the antikaon optical potential in hot and dense nuclear matter has recently been obtained @xcite within the framework of a coupled - channel self - consistent calculation taking , as bare meson - baryon interaction , the meson - exchange potential of the jlich group @xcite . in order to understand the influence of the in - medium antikaon properties on the @xmath1/@xmath0 ratio , two different prescriptions for the single - particle energy of the antikaons have been considered . first , the so - called on - shell approximation to the antikaon single - particle energy has been adopted . the antikaon partition function in this approach reads @xmath103 where @xmath104 is the @xmath2 single - particle potential in the brueckner - hartree - fock approach given by @xmath105 which is built from a self - consistent effective @xmath3 interaction in nuclear symmetric matter , averaging over the occupied nucleonic states according to the fermi distribution at a given temperature , @xmath106 . the second approach incorporates the complete energy - and - momentum dependent @xmath2 self - energy @xmath107 via the corresponding @xmath2 spectral density @xmath108 where @xmath109 stands for the @xmath2 propagator . in this case , the @xmath2 partition function reads @xmath110 where @xmath111 . we note , however , that only the s - wave contribution of the jlich @xmath112 interaction has been kept . the reason is that this potential presents some short - comings in the @xmath113 partial wave , which manifest themselves especially in the low energy region of the @xmath2 self - energy . specifically , the @xmath83 and @xmath84 poles of the @xmath29 @xmath114-matrix come out lower about @xmath115 than the physical values and , consequently , the corresponding strength in the antikaon spectral function due to hyperon - hole excitations appears at too low energies , a region very important for the calculation we are conducting here . in addition , the role of the @xmath85 pole , which lies below the @xmath3 threshold , is not included in the jlich @xmath29 interaction . in order to overcome these problems , we have added to our @xmath116 self - energy the p - wave contribution as calculated in ref . @xcite . in this model , the p - wave self - energy comes from the coupling of the @xmath2 meson to hyperon - hole ( @xmath117 ) excitations , where @xmath118 stands for @xmath83 , @xmath84 and @xmath88 . in symmetric nuclear matter at @xmath119 , this self - energy reads @xmath120 the quantities @xmath121 , @xmath122 and @xmath123 are the @xmath124 , @xmath125 and @xmath126 coupling constants , while @xmath127 , @xmath128 , @xmath129 are the @xmath83 , @xmath84 , @xmath88 relativistic recoil vertex corrections and @xmath130 , @xmath131 , @xmath132 the lindhard functions at @xmath119 . for more details , see ref.@xcite . the relevance of this low energy region makes advisable to extend this p - wave contribution to finite temperature . the hyperon - hole lindhard functions at finite temperature are easily obtained from the @xmath133-hole one given in eq . ( a.16 ) of ref . @xcite , by ignoring , due to strangeness conservation , the crossed - term contribution . the spin - isospin degeneracy factors and coupling constants need to be accommodated to the notation used in eq . ( [ eq : self ] ) and this amounts to replace @xmath134 by @xmath135 . finally , the width @xmath136 in eq . ( a.16 ) of ref . @xcite is taken to zero , not only for the stable @xmath83 and @xmath84 hyperons but also , for simplicity , for the @xmath88 hyperon , which allows one to obtain the imaginary part of the lindhard function analytically . in this section we discuss the effects of dressing the @xmath137 mesons in hot and dense nuclear matter on the @xmath137/@xmath1 ratio around the value found in ni+ni collisions at an energy of 1.93 @xmath138 . a preliminary study was already reported in ref . as previously mentioned , we prefer to discuss the results for the inverted ratio @xmath139/@xmath137 . the @xmath139/@xmath0 ratio is shown in fig . [ fig : ratio1 ] as a function of density at two given temperatures , t= 50 and 80 mev , calculated for the three ways of dressing of the @xmath0 : free ( dot - dashed lines ) , the on - shell or mean - field approximation of eq . ( [ eq : onshell ] ) ( dotted lines ) , and using the @xmath140 spectral density including s - wave ( long - dashed lines ) or both s - wave and p - wave contributions ( solid lines ) . the two chosen temperatures roughly delimit the range of temperatures which have been claimed to reproduce , in the framework of the thermal model , not only the @xmath139/@xmath0 ratio but also all the other particle ratios involved in the ni+ni collisions at sis energies@xcite . since the baryonic chemical potential @xmath28 grows with density , the factor @xmath141 in the partition functions of eqs . ( [ eq : lam - sig ] ) , ( [ eq : resonance ] ) allows one to understand why the ratio increases so strongly with density in the free gas approximation ( dot - dashed lines ) . the same is true when the particles are dressed . in this case , however , the @xmath0 feels an increasing attraction with density which tends to compensate the variation of @xmath28 and the curves bend down after the initial increase . this effect is particularly notorious when the full @xmath140 spectral density is used . the results are in qualitative agreement with the broad - band equilibration " notion introduced by brown et al . @xcite . however , in this present model , the gain in binding energy in the on - shell approximation for @xmath0 ( thick dotted line in fig . [ fig : ratio1 ] ) when the density grows does not completely compensate the increase of @xmath28 , as was the case in ref . @xcite . to illustrate this fact we note that the variation of the @xmath0 single particle energy at zero momentum at @xmath142 mev changes in our model @xcite from 434 mev to 375 mev when the density grows from 1.2@xmath143 to 2.1@xmath143 , while @xmath28 changes from 873 mev to 962 mev . therefore the relevant quantity to understand the behaviour of the @xmath1/@xmath0 ratio with density in the on - shell approximation , i.e. the sum of @xmath28 and @xmath144 [ see eq . ( 6 ) of ref . @xcite ] , suffers in our model a variation of about 30 mev . on the other hand , the model of ref.@xcite assumed a slower variation of @xmath28 , from 860 mev to 905 mev , which was almost cancelled by the change of @xmath144 from 380 mev to 332 mev , giving therefore a practically density independent ratio . we note that our chemical potential is derived from a nucleonic energy spectrum obtained in a walecka @xmath145 model , using density dependent scalar and vector coupling constants fitted to reproduce dirac - brueckner - hartree - fock calculation [ see table 10.9 in ref.@xcite ] . the results obtained in the present microscopic calculation show that the broad - band equilibration " only shows up clearly when the full spectral function is used ( solid line in fig . [ fig : ratio1 ] ) . after an increase at low densities , the @xmath146 ratio remains constant at intermediate and high densities . the use of the spectral density implicitly amounts to an additional gain in binding energy for the antikaons and , as density increases , it compensates rather well the variation of @xmath28 . to understand the origin of this additional effective attraction when the full spectral density is used , we show in fig . [ fig : ratio2 ] the two functions that contribute to the integral over the energy in the definition of the @xmath0 partition function [ eq.([eq : offshell ] ) ] , namely the boltzmann factor @xmath147 and the @xmath0 spectral functions including l=0 ( long - dashed line ) and l=0 + 1 ( solid line ) components of the @xmath148 interaction , for a momentum @xmath149 mev at @xmath150 @xmath96 and t@xmath151 mev . as it is clearly seen in the figure , the overlap of the boltzmann factor with the quasi - particle peak of the @xmath0 spectral function is small for this momentum . it is precisely the overlap with the strength in the low energy region that acts as a source of attraction in the contribution to the @xmath0 partition function . this effect is particularly pronounced when the p - waves are included , due to the additional low - energy components in the spectral function coming from the coupling of the @xmath0 meson to hyperon - hole ( @xmath117 ) excitations , where @xmath118 stands for @xmath83 , @xmath84 and @xmath88 . assigning these low energy components to real antikaons in the medium is not clear , since one should interpret them as representing the production of hyperons through @xmath152 conversion . while this is certainly true , it may also happen that , once these additional hyperons are present in the system , they can subsequently interact with fast non - strange particles ( pions , nucleons ) to create new antikaons . a clear interpretation on what fraction of the low energy strength will emerge as antikaons at freeze - out is certainly an interesting question and its investigation will be left here for forthcoming work . once the integral over the energy is performed , the determination of the @xmath0 partition function still requires an integral over the momentum . the integrand as a function of momentum is plotted in fig . [ fig : ratio3 ] for the same density and temperature than in the previous figure . as expected , the integrand is larger when the full spectral density is considered . in this case , the @xmath0 partition function is enhanced and therefore the @xmath139/@xmath0 ratio is smaller than in the on - shell approximation and also in the case where only the l=0 contributions to the spectral density are used . notice the behaviour at large @xmath153 , which decays very quickly in the on - shell approximation but has a long tail for the @xmath154 spectral density originated from the coupling of the @xmath0 meson to @xmath155 configurations . another aspect that we want to consider is how the dressing of the @xmath84 hyperon affects the value of the ratio . in fig . [ fig : ratio4 ] the value of the @xmath139/@xmath0 ratio at @xmath156 mev is shown as a function of density for different situations . in all calculations displayed in the figure , the partition function associated to @xmath0 has been obtained using the full @xmath0 spectral density . the dotted line corresponds to the case where only the @xmath83 hyperons , dressed with the attractive mean - field potential given in the previous section , are included to balance strangeness . when the @xmath84 hyperon is incorporated with a moderately attractive potential of the type @xmath157 mev , the @xmath1/@xmath0 ratio is enhanced substantially ( dashed line ) . this enhancement is more moderate when one uses the repulsive potential @xmath158 instead ( dot - dashed line ) . finally , the additional contribution of the @xmath88 resonance produces only a small increase of the ratio ( solid line ) due to its higher mass . we have checked that heavier strange baryonic or mesonic resonances do not produce visible changes in our results . notice also that , although the ratios obtained with both prescriptions for the mean - field potential of the @xmath84 meson differ appreciably , the present uncertainties in the ratio would not permit to discriminate between them . in the framework of the statistical model , one obtains a relation between the temperature and the chemical potential of the hadronic matter produced in the heavy - ion collisions by fixing the value of the @xmath137/@xmath1 ratio which was measured for ni+ni collisions at gsi to be on the average @xmath137/@xmath159 @xcite . we compare our results with a corresponding inverse ratio of @xmath139/@xmath160 in the following . the temperatures and chemical potentials compatible with that ratio are shown in fig . [ fig : ratio6 ] for different approaches . the dot - dashed line stands for a free gas of hadrons , similar to the calculations reported in refs . the dotted line shows the @xmath161 curve obtained with the on - shell or mean - field approximation [ see eqs.([eq : onshell ] ) and ( [ eq : self0 ] ) ] , while the dashed and solid lines correspond to the inclusion of the off - shell properties of the @xmath137 self - energy by using its spectral density [ eqs . ( [ eq : pik]),([eq : spec ] ) and ( [ eq : offshell ] ) ] , including l=0 or l=0 + 1 components , respectively . in the free gas limit , the temperatures compatible with a ratio @xmath1/@xmath162 imply a narrow range of values for the baryonic chemical potentials , namely @xmath163 $ ] @xmath164 for temperatures in the range of 20 to 100 mev . these values translate into density ranges of @xmath165 $ ] . as it can be seen from the dotted line in fig . [ fig : ratio6 ] , the attractive mean - field potential of the antikaons compensates the effect of increasing baryochemical potential @xmath28 . as a consequence , the density at which the freeze - out temperature compatible with the measured ratio takes place also grows . but this attraction is not enough to get the same @xmath1/@xmath0 ratio for a substantially broader range of density compared to the free case . so we do not see a clear indication of `` broad - band equilibration '' in our self - consistent mean - field calculation in contrast to the results of brown , rho , and song @xcite . the influence of the antikaon dressing on the ratio is much more evident when the spectral density is employed ( long - dashed and solid lines ) . from the preceding discussions , it is easy to understand that the low energy behaviour of the spectral density enhances the @xmath0 contribution to the ratio , having a similar role as an attractive potential and , hence , the value of @xmath28 compatible with a ratio at a given temperature increases . moreover , due to the bending of the @xmath1/@xmath0 ratio with density and its evolution with temperature observed in fig . [ fig : ratio1 ] , it is clear that there will be a maximum value of t compatible with a given value of the ratio . below this maximum temperature , there will be two densities or , equivalently , two chemical potentials compatible with the ratio . for example , the ratio @xmath1/@xmath0=30 will in fact not be realized with the temperatures of @xmath156 mev and @xmath166 mev displayed in fig . [ fig : ratio1 ] , if the antikaon is dressed with the spectral density containing @xmath167 and @xmath113 components . as shown in fig . [ fig : ratio6 ] , only temperatures lower than @xmath168 mev are compatible with ratio values of 30 ! we note that the flat regions depicted by the solid lines in fig . [ fig : ratio6 ] could be considered to be in correspondence with the notion of broad - band equilibration " of brown et al . @xcite , in the sense that a narrow range of temperatures and a wide range of densities are compatible with a particular value of the @xmath1/@xmath0 ratio . nevertheless , the temperature range is too low to be compatible with the measured slope parameter of the pion spectra . explicitly , for @xmath1/@xmath0=30 , we observe a nearly constant ratio observed in the range of @xmath169 mev covering a range of chemical potentials in between @xmath170 mev which translates into a density range @xmath171 $ ] . note that in this case , we can hardly speak of a broad - band equilibration in the sense of that introduced by brown , rho and song in ref . @xcite , where a ratio @xmath1/@xmath0=30 holds over a large range of densities in between @xmath172 and @xmath173 for @xmath142 mev . however , as we indicated at the beginning of this section , this result was obtained in the framework of a mean - field model and our equivalent on - shell results ( dashed lines in figs . [ fig : ratio1 ] , [ fig : ratio6 ] ) , based on a stronger variation of the @xmath28 with density and on a less attractive @xmath174 , seem to be very far from producing the broad - band equilibration behaviour . as pointed out before , our nucleon chemical potential is obtained in the framework of a relativistic model and varies with density more strongly than that used in ref . @xcite , which shows values close to those for a free fermi gas . if we now calculate the @xmath1/@xmath0 ratio using a @xmath175 for a free ( non - interacting ) system in the on - shell approximation , we obtain the thin dotted line in fig . [ fig : ratio1 ] . at @xmath166 mev , we now observe a tendency of a broad - band equilibration but for ratios higher than @xmath176 , of around @xmath177 . this is connected to the particular on - shell potential of the antikaon which , in our self - consistent procedure , turns to be moderately attractive . only if the attraction was larger would the broad band be realized in this on - shell picture for smaller values of the ratio as found by brown et al . @xcite . [ fig : ratio7 ] shows the @xmath1/@xmath0 ratio for the full model calculation as a contour plot for different temperatures and baryochemical potentials . we note , that the ratio is substantially lower at the temperature and density range of interest , being more likely around 15 or so for a moderately large region of baryochemical potential . note that this reduced ratio translates into an overall enhanced production of @xmath0 by a factor two compared to the experimentally measured value . at pion freeze - out , the medium can hold twice as many @xmath0 as needed to explain the measured enhanced production of @xmath0 if one considers the full spectral features of the @xmath0 in the medium . we stress again that this enhancement is not due to an increased attraction in the sense of a mean - field calculation . it is a consequence of the additional strength of the spectral function at low energies which emerges when taking into account p - wave hyperon - hole excitations . the boltzmann - factor amplifies the contribution of the low - energy region of the spectral function so that these excitations are becoming the main reason for the overall enhanced production of the @xmath0 in the medium . we have studied , within the framework of a statistical model , the influence of considering the modified properties in hot and dense matter of the hadrons involved in the determination of the @xmath19 ratio produced in heavy - ion collisions at sis / gsi . we have focused our attention on incorporating the effects of the antikaon self - energy , which was derived within the framework of a self - consistent coupled - channel calculation taking , as bare interaction , the meson - exchange potential of the jlich group for the s - wave and adding the p - wave components of the @xmath178 excitations , with @xmath179 . it is found that the determination of the temperature and chemical potential at freeze - out conditions compatible with the ratio @xmath19 is very delicate and depends very strongly on the approximation adopted for the antikaon self - energy . the effect of dressing the @xmath0 with an spectral function including both s- and p - wave components of the @xmath29 interaction lowers considerably the effective temperature while increasing the chemical potential @xmath28 up to @xmath180 mev , compared to calculations for a noninteracting hadron gas . when the free or on - shell properties of the antikaon are considered , the ratio at a given temperature shows a strong dependence with the density . this is in contrast with the `` broad - band equilibration '' advocated by brown et al . @xcite , which was established , in the context of a mean - field picture , through a compensation between the increased attraction of the mean - field @xmath2 potential as density grows with the increase in the baryon chemical potential . our mean - field properties do not achieve such a compensation due to a stronger increase of the nucleon chemical potential @xmath28 with density . when taking into account the full features of the spectral function of the @xmath0 via hyperon - hole excitations , we find that the @xmath19 ratio exhibits broad - band equilibration . nevertheless , the ratio is even in excess of the measured ratio at temperatures which are deduced from the slope parameter for pions . one can argue in principle , that dynamical non - equilibrium effects can reduce the number of @xmath0 by virtue of annihilation with nucleons to hyperons and pions at freeze - out . the number of the latter produced hadrons will not be altered substantially by this annihilation process as the @xmath0 is much less abundant than hyperons and especially pions . in our approach , we calculate the number density or particle ratios which are in the medium . what needs to be clarified is how the particles get on - shell at freeze - out , a question previously posed e.g. for antiproton and antihyperon production at the gsi . clearly , this has to be addressed in dynamical models and we leave that investigation for forthcoming work . we are very grateful to dr . volker koch and dr . luis alvarez - ruso for useful discussions . this work is partially supported by dgicyt project bfm2001 - 01868 and by the generalitat de catalunya project 2001sgr00064 . l.t . also wishes to acknowledge support from the ministerio de educacin y cultura ( spain ) . a. ramos , j. schaffner - bielich , and j. wambach in _ physics of neutron star interiors _ , edited by d. blaschke , n. k. glendenning , and a. sedrakian , lecture notes in physics vol.578 , ( springer , new york , 2001 ) , p.175 and references therein .	the @xmath0/@xmath1 ratio produced in heavy - ion collisions at gsi energies is studied . the in - medium properties at finite temperature of the hadrons involved are included , paying a special attention to the in - medium properties of antikaons . using a statistical approach , it is found that the determination of the temperature and chemical potential at freeze - out conditions compatible with the ratio @xmath0/@xmath1 is very delicate , and depends very strongly on the approximation adopted for the antikaon self - energy . the use of an energy dependent @xmath2 spectral density , including both s and p - wave components of the @xmath3 interaction , lowers substantially the freeze - out temperature compared to the standard simplified mean - field treatment and gives rise to an overabundance of @xmath0 production in the dense and hot medium . even a moderately attractive antikaon - nucleus potential obtained from our self - consistent many - body calculation does reproduce the `` broad - band equilibration '' advocated by brown , rho and song due to the additional strength of the spectral function of the @xmath0 at low energies . _ pacs : _ 13.75.jz , 14.40.ev , 25.75-q , 25.75.dw , 24.10.pa , 24.60.-k _ keywords : _ @xmath3 interaction , kaon - nucleus potential , @xmath4 , finite temperature , heavy - ion collisions
long gamma - ray bursts ( grbs ) are linked with the death of massive stars ( for a review , see * ? ? ? their association with supernovae ( e.g. * ? ? ? * ; * ? ? ? * ) and the fact that these events are at cosmological distance @xcite make them interesting for studies of cosmology in the redshift range 1 - 15 . however , while the grb features a prompt emission , usually seen in gamma - ray only , and an afterglow seen at all wavelengths , only the former emission has been considered for cosmological studies yet . one of the first attempt to do so , based on the @xmath2 correlation found by @xcite , was done by @xcite , who constrained the @xmath3 parameters . while not very constraining , their findings were compatible with previous tests made with supernova samples . the use of grb for cosmological studies , specially at high redshift where their detection at optical / ir wavelength is difficult , needs to build a robust indicator of their distance , whenever possible based on their intrinsic properties . the afterglow emission is less studied from a cosmological point of view because of its diversity . however , there were hints of standardization of the x - ray afterglows . the first attempt was done by @xcite , who found evidences for clustering in the x - ray light curves of bepposax afterglows . this clustering was confirmed later by @xcite who extended the sample to the xmm - newton and chandra data . in the following we will refer to these articles respectively as paper i and ii . in addition , @xcite showed that supernova and grb light curves had similar behaviors and were converging with time toward a similar luminosity . in paper i we tried also to check if optical light curves were also clustered : this attempt failed due to the poor knowledge of the intrinsic absorption in the burst host galaxy at that time . this study was completed by @xcite and @xcite who found independently that optical afterglows were also clustered in luminosity . the high detection rate and throughput of swift , which provide a large sample of x - ray observations and rapid , accurate localization , enabling a redshift estimation by optical / ir telescopes , allowed us to increase dramatically our sample . with this larger sample , we have tried to derive a method for estimating the burst redshift from the x - ray light curve @xcite . however , before to apply this method it is necessary to understand the nature of these two groups , and to determine a consistent way to derive to which group belongs each burst . this is the purpose of this article . in section [ sec_ana ] we present our sample and the data analysis we performed . we discuss the x - ray clustering in section [ sec_result ] . we compare the x - ray and optical clusterings in section [ sec_discu ] . in section [ sec_redshift ] we present our method of grb source redshift estimation from the x - ray afterglow light curve . we finally test this method on several grbs detected recently by swift , and we propose an estimation of the redshift for two previously detected grbs of unknown distance . our sample of grbs with known redshift and x - ray afterglow observations is listed in table [ table1 ] . it includes all afterglows observed by bepposax , xmm - newton and chandra , which data were retrieved from @xcite and @xcite . as for swift observations , we browsed the swift archive web page and selected all bursts detected prior to the 1st of august 2006 with a measured redshift and a @xmath4 larger than 2.0 seconds ( in order to exclude short bursts ) . from this sample of swift bursts , we further excluded due to data processing errors . we removed the flaring parts of the light curve when applicable ( this totally removed from the sample the data of , which is thus not listed in table [ table1 ] ) . we corrected the flux light curves for distance effects as in paper i and ii : we apply a k - correction using the measured spectral index of each afterglow , assuming a flat universe with @xmath5 , and correcting for the time dilation effect by computing the luminosity at a time @xmath6 using the observed flux at time @xmath7 . this is very important for the correction accuracy , as discussed in sec . [ sec_discu ] . we restricted the light curves to the [email protected] kev x - ray band , where the absorption is negligible . this allowed us to neglect any other corrections for absorption by the ism . we do not take into account any beaming effect due to a possible jet structure . as we done previously , instead of using a luminosity light curve , we express all light curves in flux units at a given distance ( like the optical absolute magnitude ) . this allows to reduce the uncertainties on the correction . for consistency with paper i and ii , we fix the redshift to @xmath9 . cccccccccc & bepposax & 0.695 & @xmath10 & @xmath11 & ii & i & @xmath12 & @xmath13 & + & bepposax & 0.835 & @xmath14 & @xmath15 & ii & ii ? & @xmath16 & @xmath17 & + & bepposax & 3.42 & @xmath11 & @xmath18 & i & i & @xmath19 & @xmath20 & + & bepposax & 0.0085 & @xmath21 & ( 1.1 ) & iii & & @xmath22 & @xmath23 & + & bepposax & 1.096 & @xmath24 & ( 1.1 ) & ii & ii & @xmath25 & @xmath26 & + & bepposax & 0.966 & @xmath27 & @xmath28 & ii & i & @xmath29 & @xmath30 & + & bepposax & 1.60 & @xmath31 & @xmath32 & i & i & @xmath33 & @xmath34 & + & bepposax & 1.619 & @xmath35 & @xmath36 & i & i & @xmath37 & @xmath38 & + & chandra & 1.02 & @xmath39 & @xmath40 & i & i & @xmath41 & @xmath42 & + & bepposax & 0.846 & @xmath43 & @xmath44 & ii & & @xmath45 & @xmath46 & + & bepposax & 0.42 & @xmath47 & @xmath11 & ii & & @xmath48 & @xmath49 & + & bepposax & 2.066 & @xmath50 & @xmath40 & i & i & @xmath51 & @xmath52 & + & bepposax & 1.477 & @xmath53 & @xmath54 & i & i & @xmath55 & @xmath56 & + & bepposax & 0.36 & @xmath57 & ( 1.1 ) & ii & ii & @xmath58 & @xmath59 & + & xmm & 2.14 & @xmath60 & @xmath61 & ii & i & @xmath62 & @xmath63 & + & chandra & 0.69 & @xmath64 & @xmath65 & ii & i & @xmath66 & @xmath67 & + & chandra & 1.25 & @xmath68 & @xmath69 & i & i & @xmath70 & @xmath71 & + & chandra & 2.3 & @xmath72 & @xmath73 & i & i & @xmath74 & @xmath75 & + & chandra & 1.98 & @xmath76 & @xmath77 & ii & i & @xmath78 & @xmath79 & + & chandra & 1.52 & @xmath80 & @xmath81 & ii & i & @xmath82 & @xmath83 & + & xmm & 0.168 & @xmath84 & @xmath65 & ii & i & @xmath28 & @xmath85 & + & xmm & 0.105 & @xmath86 & @xmath14 & iii & & @xmath87 & @xmath88 & + & swift & 1.29 & @xmath89 & @xmath90 & ii & & & & @xmath91 + & swift & 0.5915 & & @xmath92 & iii & & & & + & swift & 1.949 & @xmath93 & @xmath94 & i & i & @xmath95 & @xmath96 & @xmath97 + & swift & 3.24 & @xmath98 & @xmath99 & i & i & & & @xmath100 + & swift & 2.90 & @xmath101 & @xmath11 & i & i & @xmath102 & @xmath103 & @xmath104 + & swift & 2.44 & @xmath105 & @xmath106 & ii & & & & @xmath107 + & swift & 0.6535 & @xmath108 & @xmath109 & ii & & @xmath110 & @xmath111 & @xmath112 + & swift & 4.27 & @xmath113 & @xmath114 & i & i & & & @xmath115 + & swift & 0.606 & @xmath116 & @xmath18 & ii & i & @xmath117 & @xmath118 & @xmath119 + & swift & 2.821 & @xmath120 & @xmath121 & i & i & @xmath122 & @xmath123 & @xmath124 + & swift & 3.968 & @xmath125 & @xmath126 & i & i & & & @xmath127 + & swift & 1.71 & @xmath128 & @xmath129 & ii & & & & @xmath130 + & swift & 5.3 & @xmath131 & @xmath121 & i & & & & @xmath132 + & swift & 0.83 & @xmath133 & @xmath14 & ii & ii & @xmath134 & @xmath135 & @xmath136 + & swift & 0.297 & @xmath137 & @xmath18 & iii & & & & + & swift & 3.344 & @xmath138 & & ii & & & & @xmath139 + & swift & 2.198 & @xmath140 & @xmath10 & ii & i & @xmath141 & @xmath142 & @xmath143 + & swift & 0.9364 & @xmath144 & @xmath44 & ii & & & & @xmath145 + & swift & 2.346 & @xmath146 & @xmath65 & i & & & & @xmath147 + & swift & 0.080 & @xmath148 & @xmath47 & iii & & & & @xmath149 + & swift & 1.549 & @xmath150 & @xmath18 & ii & i & & & + & swift & 2.03 & @xmath151 & @xmath11 & ii & & & & @xmath152 + & swift & 3.53 & @xmath153 & @xmath61 & i & & & & @xmath154 + & swift & 4.048 & @xmath155&@xmath14 & i & & & & @xmath156 + & swift & 3.91 & @xmath157 & @xmath158 & i & & & & @xmath159 + & swift & 0.033 & @xmath160 & @xmath161 & iii & & & & @xmath162 + & swift & 4.41 & @xmath10 & @xmath77 & ii & & & & @xmath163 + & swift & 1.489 & @xmath164 & @xmath165 & ii & & & & @xmath166 + & swift & 4.9 & @xmath167 & @xmath168 & i & & & & @xmath169 + & swift & 0.4428 & @xmath72 & @xmath44 & iii & & & & @xmath170 + & swift & 5.11 & @xmath171 & @xmath81 & i & & & & @xmath172 + & swift & 3.21 & @xmath173 & @xmath174 & i & & & & @xmath175 + & swift & 2.68 & @xmath72 & @xmath10 & ? & & & & @xmath176 + & swift & 3.78 & @xmath177 & @xmath114 & i ? & & & & @xmath178 + & swift & 3.082 & @xmath179 & @xmath180 & ii & & & & @xmath181 + & swift & 0.125 & @xmath182 & @xmath14 & iii & & & & @xmath183 + & swift & 3.43 & @xmath184 & @xmath185 & i & & & & @xmath186 + & swift & 2.71 & @xmath187&@xmath15 & i & & & & @xmath188 + & swift & 0.54 & @xmath189 & @xmath190 & ii & & & & @xmath191 + the light curves of our sample , corrected for distance effects . we indicate swift , xmm - newton , chandra and bepposax bursts with blue , green , black and red diamonds respectively . the error bars are not plotted for clarity . see electronic version for colors . ] figure [ fig1 ] presents the light curves corrected for distance effects . as one can clearly see , no clustering is observed in the early part of the light curves . the addition of swift bursts seems to have blurred out the clustering properties observed in paper i and ii . if we assume that swift is observing grbs similar to those observed previously , the only difference that could explain this result is the addition of the early time x - ray light curves , since previous observations from bepposax , xmm , chandra where performed usually at least 6h after the grb alert . swift has shown that the x - ray light curve is typically composed by a first steep power - law segment , associated with the tail of prompt emission , followed by a flat plateau , a second steepening ( which is the segment usually seen before swift ) , and a possible late steepening related to the jet aperture @xcite . @xcite have interpreted this behavior with a two - components model . according to them , the plateau phase marks the transition between a first component , ascribed to the prompt emission , and a second one which nature is less clear and develops in the afterglow . we have retrieved for swift bursts the values of the end - time @xmath192 of the plateau phase , and excluded all swift data taken before this time . the results are displayed in fig . [ fig2 ] . excluding these data , the two groups reported in papers i and ii become apparent . we note however still some dispersion during the first part of the light curves . we interpret this as a consequence on the error ( sometime large ) on the @xmath192 measurement . indeed , restricting the sample to data collected after @xmath193 ( i.e. being very conservative on the @xmath192 value ) strongly reduce this dispersion , at the price of a drastic reduction of the sample size . in the following , we will use all data taken after @xmath192 ( i.e. a less conservative hypothesis ) and do not discuss the effects seen in the early part of the light curves . we refer to the bright group as _ group i _ and the dimmer one as _ group ii _ , as in paper ii . they are clustering with a mean flux of @xmath194 erg s@xmath195 @xmath196 and @xmath197 erg s@xmath195 @xmath196 for groups _ i _ and _ ii _ respectively , assuming a common redshift of unity ( see above ) and at a time of 1 day . some bursts do not follow this relation : , , , , , , , and . they are all nearby events , the most distant , , being at @xmath198 @xcite . note however that is also at low redshift ( @xmath199 ) but belongs to _ group ii_. we consider that these low luminosity events form a specific group , referred as _ group iii _ in the following . we finally observe two peculiar events . belongs to _ group i _ but , due to its steep decay , changes to _ group ii _ later on . we note that , contrary to the other bursts , this event is not compatible with an ism or wind medium , but is compatible with a jet effect . this would imply that the bor & gendre relation is valid _ before _ jet effects become apparent . is located between _ group i _ and _ group ii_. on 61 events , this is the only one lying within the gap . we note that the light curve of this event presents a strong `` noise '' ( i.e. small scale temporal variations ) during the whole observation . a possible explanation to these variations would be that this event presents several x - ray flares ( like e.g. ) , implying that we do not observe the continuum . under this assumption , this event would be a _ group ii _ event ( and should not be considered for further discussion ) . however , we can not rule out a possible outliers hypothesis and maintain this event within the sample . the probability that a power law luminosity distribution ( letting the index be a free parameter ) represents the observed distribution is , at the maximum , @xmath200 ; thus the observed clustering in two groups is very significant . to compute this probability we impose a lower luminosity limit such that the _ group iii _ bursts are excluded , because of the selection effects ( see sec . [ sec_discu ] ) . we checked if this clustering can be related to the isotropic burst energy @xmath201 or the peak energy @xmath202 values . figure [ ep ] presents the @xmath202 distribution of bursts versus the x - ray afterglow flux at one day . one can clearly see that the two groups share a similar @xmath202 distribution . this also holds for the @xmath201 distribution ( fig . [ eiso ] ; see also sec . [ sec_discu ] for a discussion on the selection effects ) . recently , @xcite reported three classes of grbs based on their fluence and duration . however , their classification and the results we report here are not correlated . moreover the early swift x - ray light curves associated with the tail of the prompt emission do not cluster in luminosity . this leads us to consider that the afterglow emission is not correlated with the prompt emission . ( rest frame ) values . blue circles and red squares represent _ group ii _ and _ group i _ bursts respectively . the flux at one day as been computed by interpolating the flux at that date when the burst was observed one day after the burst ( see text for discussion about extrapolation effects ) , thus lowering the number of events available for comparison . , width=302 ] ) . symbols are the same as in fig . , width=302 ] . , width=302 ] in paper ii , we pointed out a possible segregation of decay indexes : _ group i _ bursts seemed to decay faster than _ group ii _ ones . thanks to the fast monitoring capabilities of swift we can clearly rule out this hypothesis . figure [ decay ] shows that there is no difference between the decay index distributions for the two groups . moreover , swift has added some events to the _ group i _ with a low decay ( e.g. ) . we find the same results when we look at the spectral indexes ( fig . [ alpha ] ) . clearly , the two distributions are similar . in fact , except for two of them , all bursts are compatible with having a common spectral index of @xmath203 ( with the flux @xmath204 ) . . , width=302 ] we present all results using a common distance ; however , as we are correcting for distance effects , one may wonder if this could bias our findings . . , width=302 ] figure [ redshift ] presents the redshift distribution of our sample . there is a clear trend : low luminosity bursts are nearby while bright ones are more distant . this is even more obvious if we include _ group iii _ bursts , which have all a very low redshift . this effect can be partly explained by a selection effect , as low luminosity events can not be detected at high distance . the lack of dim distant events ( located in the upper left corner of fig . [ redshift ] ) is due to that effect . however , we should also have detected bright nearby events . the nearest _ group i _ burst , ( ) , is located at @xmath205 . while this is already lower than the mean swift grb redshift ( 2.7 , * ? ? ? * ) , there is a clear deficit of bright sources closer @xmath206 : whether this is related to the nature , or the intrinsic frequency , of _ group i _ bursts , is still an open question . another important effect to take into account is the time dilation and the estimation of the distribution parameters . x - ray light curves are not taken continuously . before swift , x - ray observations were usually performed as snapshots of 30 - 60 ksec long @xcite . because of the time dilation , the start and end of each observation were located at different times , and the estimation of a parameter at a given time needs extrapolation or interpolation from the value observed at another time . however , the decay index can largely vary during the observations as shown by @xcite ; this uncertainty can affect the accuracy of the distribution parameter estimations . this effect can be illustrated by looking at fig . 6 of @xcite , which is equivalent to fig . [ fig2 ] of the present work : the results are clearly different ; we observe a strong clustering , while @xcite see only a trend . however @xcite extrapolate the burst afterglow light curves , when no data is available at times earlier than 12h in the source rest frame : as an example is observed from 1.00 day to 1.34 days ( source rest frame ) , is observed from 0.69 to 4.36 days ( idem ) . the net effect of doing extrapolations instead of interpolations is to broaden the distribution , blurring in turn the clustering reported here . to avoid this effect , we decided , as in paper i and ii , to avoid extrapolations and to perform only flux interpolations , when relevant , using the nearest measurements . this is of course at the expense of the sample size , as one can note in figs . [ ep ] , [ eiso ] , [ decay ] , [ alpha ] , [ redshift ] where only half of the total sample is plotted . @xcite , @xcite and @xcite have shown that even optical afterglow light curves display a clustering effect . these authors do not include in their works any of our _ group iii _ bursts : for consistency we will restrict the following discussion to groups _ i _ and _ ii_. they found two groups , and we will refer to them as _ optical group i _ ( or _ oi _ ) and _ optical group ii _ ( or _ oii _ ) for the bright and dim group respectively . we indicate in table [ table1 ] the optical class of each bursts . several events of our sample are not included in the work of @xcite or @xcite , and no optical classification is reported for them . according to @xcite , one day after the burst ( rest frame ) , the difference in luminosity between groups _ oi _ and _ oii _ is @xmath207 . at the same time , the difference in luminosity between the x - ray groups is @xmath208 . from table [ table1 ] , we can see that while an _ oi _ burst can belong to _ x - ray group i _ or _ x - ray group ii _ , _ oii _ bursts are dim both at optical and x - ray wavelengths . we present in fig . [ decay - spectral ] the distribution of the decay index versus the spectral index of these three burst classes . again , no clear separation can be observed . in this section we investigate further on the nature of the observed clusterings in the context of the external shock model . @xcite has indicated that around 1 day for almost all bursts , we have @xmath209 ; thus in the following we will assume this repartition of the characteristic emission frequencies . according to @xcite , during the slow cooling regime the predicted flux for a fireball expanding in a uniform interstellar medium ( ism ) is : @xmath210 @xmath211 while for a fireball expanding in a wind - like medium we have : @xmath212 @xmath213 where @xmath214 , @xmath215 , and @xmath6 are the distance , observation frequency and observation time respectively , and @xmath216 is the electron distribution power law index ; we assume @xmath217 for all bursts . we rewrite these equations in terms of @xmath218 and @xmath219 , the energy carried by the magnetic field and relativistic electrons respectively ( @xmath220 is the fireball total energy ) : @xmath221 @xmath222 @xmath223 @xmath224 we thus find the optical - to - x - ray flux ratio to be : @xmath225 for a fireball expanding in a ism and @xmath226 for a fireball expanding in a wind - like medium . we extracted from the work of @xcite the mean optical fluxes at @xmath227 hz . they are 0.178 mjy and 10.0@xmath228jy for groups _ oi _ and _ oii _ respectively . the mean x - ray fluxes ( at @xmath229 hz ) are 2.10 @xmath228jy and 0.09 @xmath228jy for groups _ i _ and _ ii _ respectively . combining equation [ eq_ism_opt2 ] with equation [ eq_ratio_ism ] , and equation [ eq_wind_opt2 ] with equation [ eq_ratio_wind ] , we obtain the results listed in table [ table_equation ] . one can note that the constraints on @xmath230 depend weakly on the medium density ( because of the very low exponent of the density parameter ) . because @xmath231 ( or @xmath232 ) , we can find a constraint on the total fireball energy @xmath220 ( a similar condition is true for @xmath233 , but this is not constraining for @xmath220 ) . cccccc x - ray & optical & medium & @xmath230 & @xmath233 & @xmath220 + group & group & class & constraint & constraint & constraint + & _ oi _ & ism & @xmath234 & @xmath235 & @xmath236 + _ i _ & _ oi _ & wind & @xmath237 & @xmath238 & @xmath239 + & _ oi _ & ism & @xmath240 & @xmath241 & @xmath242 + _ ii _ & _ oi _ & wind & @xmath243 & @xmath244 & @xmath245 + & _ oii _ & ism & @xmath240 & @xmath246 & @xmath242 + _ ii _ & _ oii _ & wind & @xmath247 & @xmath248 & @xmath249 + within each group the flux variations around the mean value are small . this may indicate that either @xmath233 , @xmath230 , @xmath220 , and @xmath250 ( or @xmath251 , depending on the surrounding medium ) are correlated quantities , or are all fixed to a common value . in the former case , because the surrounding medium density is fixed _ before _ the burst occurs , it is this quantity that should be considered as the true variable parameter ( and not the energy injected within the fireball ) . on the other hand , a stellar progenitor produces a wind which affects its surrounding medium ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? thus , the surrounding density is ( partly ) fixed by the progenitor properties . as these properties should also affect the energy quantity emitted by the fireball ( and thus @xmath233 , @xmath230 , and @xmath220 ) , we find no surprising that @xmath233 , @xmath230 , @xmath220 , and @xmath251 or @xmath250 are correlated together . if @xmath233 , @xmath230 , @xmath220 , and @xmath250 ( or @xmath251 ) have fixed values , the constraints on the model are strong since this restrict to only three types the possible environments in which a grb can occur ( according to the three observed behaviors ) . if so , this is a possible explanation of why we do not observe a relativistic outflow associated with each type ib / c supernovae @xcite : in fact , normal type ib / c supernovae ca nt develop as a grb . the flux ratio between the two optical groups and the two x - ray groups are similar . @xcite have investigated the surrounding medium around 10 grbs ( , , , , , , , , , and ) . they found that most of them can be fit with an ism ( all belong to our optical and x - ray bright burst group ) , while only one burst , , requires a wind environment ( it is an optical and x - ray dim burst ) . this suggests that _ group ii - oii _ bursts are surrounded by a stellar wind while _ group i - oi _ are located within an ism . assuming that the only difference between these two groups is due to the medium density profile ( and thus that all other parameters have a similar distribution within the two groups ) , we can explain the difference between the two groups if we set @xmath252 ( assuming @xmath253 ) . however in such a case we then can not explain the behavior of _ group ii - oi _ bursts . we also note that @xcite have found that a wind medium can also accurately describe the surrounding environment of and ( two _ group i - oi _ bursts ) . thus , the surrounding medium type can not explain the origin of the observed clustering . in x - ray , the flux is mostly dependent on @xmath230 and @xmath220 ( see eqn . [ eq_ism_x2 ] and [ eq_wind_x2 ] ) . if one assumes that @xmath201 is a good estimator of @xmath220 , then since the observed distributions of @xmath201 are similar for the two groups ( see fig . [ eiso ] ) , different values of @xmath220 can not explain the differences . in turn this implies that @xmath230 varies within the two groups ( see table [ table_equation ] ) . this however again can not explain the origin of _ group ii - oi _ : if the clustering is due only to a difference in the @xmath230 values , then we can not expect from eqn . [ eq_ism_opt2 ] and [ eq_ism_x2 ] , and [ eq_wind_opt2 ] and [ eq_wind_x2 ] , that a dim x - ray afterglow can be bright in optical . the constraints listed in table [ table_equation ] indicate that the difference between groups _ i - oi _ and _ ii - oii _ on one hand , and group _ ii - oi _ on the other hand can be also due to a different value of @xmath233 . we thus propose that the presence of different groups is ascribed to different families of @xmath230 and @xmath233 values : * a family of magnetized fireball that produces the group _ ii - oi_. in such a case the fireball transfers only a low fraction of its energy into relativistic electrons . * a family of less magnetized fireball that produces the groups _ i - oi _ and _ ii - oii_. the fraction of total energy going into magnetic fields is roughly one order of magnitude lower than group _ ii - oi_. these two groups can be related to an high and low fraction of energy going in relativistic electrons , respectively . interestingly , the group _ ii - oii _ is not very numerous ( compared to the other two groups ) , so that most of grbs of our sample are then either magnetized but not electron - energized or not magnetized but electron - energized. the clustering observed in x - ray can be used for distance estimation . to date , most of the redshift measurements made on grb afterglows were done by optical spectroscopic or photometric observations . however , because of the lymann alpha cut - off , this method can not be applied to high redshift events ( @xmath254 ) , unless doing ir spectroscopy . moreover , not all grbs can be followed in optical ( e.g. the so - called dark - bursts , * ? ? ? * ) , either because they are too faint at the time of the observation ( or even dark ) , or because there is no large enough telescope available , or because of the position of the event respective to the sun . hence many grbs have no known distance , stressing the need for intrinsic distance indicators . after the launch of the swift satellite , thanks to its fast re - pointing capabilities , nearly all grbs have an homogeneous x - ray follow - up . hence , a redshift measurement method based solely on x - ray observations could be very interesting if one wants to use a large sample of grb sources for cosmological studies . we propose here to use the reported clustering of grbs x - ray light curves in three classes to build a redshift estimator . the method is based on the redshift needed for a burst to belong to one ( or both ) of the two groups . the steps are easy to follow : we compute from the x - ray observations ( i.e. obtained from swift ) the flux at 1 day ( observer frame ) in the 2 - 10 kev band , either using a mean spectral index of @xmath255 , or the exact one obtained from x - ray spectral fitting . we note that even if the spectral index is not known , this has little influence on the actual flux at one day ( within reasonable limits ) . table [ table_estimate ] , which provides the redshift needed to comply with the relation for both groups for a given observed flux , is directly usable to estimate the redshift . comparison of the estimated redshift versus the measured one for the bursts defining our sample . the solid line corresponds to equality .,width=302 ] we calibrated this method by deriving the estimated redshift for the bursts of our sample ( for which the group is known ) , and comparing this value with the measured redshift . the results are displayed in fig . [ test_z ] . the estimated redshift agrees with the measured one for most of the bursts . the only discrepancies arises at low redshift . in sect . [ sec_discu ] we already noted that _ group iii _ bursts are located at small distances ( [email protected] ) and do not follow the same relation as groups i and ii : as a conservative approach we prefer to restrict the validity of our method to source located at redshifts larger than 0.5 . we list in table [ table_estimate ] the observed flux one day after the burst ( observer frame ) and the associated redshift needed for a burst to comply with the observed clustering for group i and ii sources . because of the limitation to redshifts larger than 0.5 , all bursts brighter than @xmath257 erg s@xmath195 @xmath196 belong to _ group i_. however , the belonging group is not correlated with easily observable quantities ( absorption , peak energy , isotropic energy , spectral or decay indexes ) , so it can be fixed only through broad - band modeling . in fact this leads to two redshift estimates . if the bursts has an optical afterglow there is a possible way to decide which of the estimate is valid : a burst with flux of @xmath257 erg s@xmath195 @xmath196 which belongs to _ group i _ has a redshift of @xmath258 . at that distance , the lymann @xmath259 cut - off appears to be at 360 nm , in the uvot observation bands . hence , if a grb afterglow is observed by uvot ( or in the b band for ground based telescopes ) then it can not belong to group i , and the ambiguity on the redshift determination is cleared . c c c flux & group i redshift & group ii redshift + ( erg s@xmath195 @xmath196 ) & & + @xmath260 & & 4.43 + @xmath261 & & 3.28 + @xmath262 & & 2.72 + @xmath263 & & 2.35 + @xmath264 & & 2.12 + @xmath265 & 7.80 & 1.96 + @xmath266 & 7.09 & 1.83 + @xmath267 & 6.38 & 1.75 + @xmath268 & 6.05 & 1.68 + @xmath269 & 5.78 & 1.62 + @xmath270 & 4.10 & 1.26 + @xmath271 & 3.48 & 1.09 + @xmath272 & 3.05 & 0.96 + @xmath273 & 2.82 & 0.89 + @xmath274 & 2.59 & 0.82 + @xmath275 & 2.42 & 0.77 + @xmath276 & 2.29 & 0.73 + @xmath277 & 2.10 & 0.69 + @xmath278 & 1.99 & 0.66 + @xmath279 & 1.50 & 0.50 + @xmath280 & 1.29 & + @xmath281 & 1.16 & + @xmath282 & 1.06 & + @xmath283 & 0.99 & + @xmath284 & 0.94 & + @xmath285 & 0.89 & + @xmath286 & 0.83 & + @xmath287 & 0.79 & + @xmath288 & 0.56 & + @xmath289 & 0.50 & + ccc grb name & redshift & estimate + & group i & group ii + & 4.2 @xmath0 1.2 & 1.2 @xmath0 0.2 + & 3.8 @xmath0 0.7 & 1.4 @xmath0 0.2 + & 3.5 @xmath0 0.9 & 1.3 @xmath0 0.3 + & 4.7@xmath290 & 1.6 @xmath0 0.3 + & 2.3 @xmath0 0.7 & 0.8 @xmath0 0.2 + & 5.8 @xmath0 1.8 & 2.2 @xmath0 0.4 + & 5.0 @xmath0 1.5 & 1.5 @xmath0 0.3 + & 0.5 @xmath0 0.4 & @xmath291 + & 3.4 @xmath0 0.5 & 1.0 @xmath0 0.2 + & 5.5@xmath292 & 1.7 @xmath0 0.2 + & 8.0 @xmath0 2.0 & 1.9 @xmath0 0.3 + we retrieved from @xcite and @xcite the light curves of several bursts with good temporal sampling and spectral informations ( table [ table2 ] ) , and estimated their redshift assuming they belong either to _ group i _ or to _ group ii_. for only one burst we compute an associated redshift lower than 0.5 ( ) , and thus we do not consider it in the following discussion . the estimated redshift of is either @xmath293 or 1.4 @xmath0 0.2 . this burst was observed in u band @xcite . as the lymann @xmath259 cut - off cross the u band at @xmath294 , this burst should not have been observed in the high distance hypothesis . thus , u observations can indeed solve the problem of group classification , and we propose a redshift measurement for of 1.4 @xmath0 0.2 based on x - ray observations . for the same reason we argue that is at @xmath295 and not @xmath296 , since an optical afterglow has been observed in optical @xcite . was observed by xmm - newton but no optical afterglow was detected . this event was thus classified as a dark burst @xcite . we note that this one has a large redshift value , most of all if it belong to _ group i_. in such a case , with a calculated redshift of @xmath297 , we can explain the classification as dark burst by the lyman alpha cut - off at @xmath298 . however , since we can not draw conclusions from the absence of detection at optical wavelengths , and as dark bursts can also be intrinsically fainter @xcite , we can not exclude the hypothesis that this event is located at @xmath299 , hence belonging to _ group ii_. finally , the mean redshift for swift bursts is 2.3 , while the pre - swift mean redshift was 1.2 @xcite . lets assume that _ all _ bursts listed in table [ table2 ] are _ group ii _ bursts , a very conservative hypothesis , the mean redshift , in this case , is 1.5 : this is already larger than the pre - swift mean redshift since it is reasonable to consider that at least some of these events belong to _ group i _ , this should be considered as a lower limit . before swift , the x - ray / optical follow - up was late , and only the brightest events were observed . as more distant events will appear fainter , this introduced a bias against distant events . indeed , the fact that these bursts without known redshift are slightly more distant indicates that the pre - swift redshift distribution was biased toward low redshift due to selection effects , and that swift cleared this bias thanks to its fast follow - up . cccc grb name & group i & group ii & measured + & estimate & estimate & redshift + & 7.8 & 1.96 & 2.44 + & 6.05 & 1.68 & 2.04 + & & 3.28 & 3.62 + we have investigated the clustering of afterglow light curves observed previously ( papers i and ii ) in x - ray and in optical from bepposax , xmm - newton and chandra data , using the newly available , well - monitored swift light curves . adding swift bursts to the previous sample reported in paper ii , we still confirm our previous findings . on a sample of 61 events the x - ray light curves cluster in two groups , with a significance larger than 6 @xmath300 . @xcite describe the x - ray light curve as a sum of two components , the first one being related to the prompt emission : we found that it is the second component that clusters in luminosity . this finding is also supported by the fact that the clustering is not related to the properties of the prompt emission . a similar clustering was observed also at optical wavelengths . we compared the classification within each group in x - ray and in optical , and found three classes : _ bright _ optical and x - ray afterglows , _ dim _ ones , and optically _ bright_-x - ray _ dim _ ones . we propose that this clustering is related to the fireball properties and the surrounding medium density , which can be either constant or correlated . we used the mean x - ray and optical fluxes of each group to put constraints on the values of @xmath301 and @xmath302 , and expressed them in terms of the total fireball energy and the surrounding medium density . we stress that the extension of this work at low frequency ( radio , sub - millimeter and infrared ) may help solving the exact origin of the clustering by strongly constraining the medium density and the position of the synchrotron self absorption and injection frequencies . using the observed x - ray afterglow properties of grbs , we propose a new , simple , method for the determination of the source redshift based on x - ray data ; optical photometry in u and v bands may help to clear the degeneracy between the two estimates found . this method has been established on a sample of bursts with known redshift , and we find an excellent correlation . we apply this method to a sample of grb source of unknown redshift . we propose an estimation of the redshift for ( 1.4 @xmath0 0.2 ) and for ( @xmath1 ) . while we were building this distance indicator and writing this paper , several grb were detected by swift and their distance derived using optical spectroscopy . we thus have a sample of bursts which allows to test this correlation independently . to simulate actual conditions for an unknown event we used the swift count rate light curve available from their web site @xcite , and we apply a mean spectral index of 1.2 for the count - to - flux conversion . the results are indicated in table [ table_sup ] , indicating an agreement between 10.3 and 24.5 % ( in accordance with the error estimated in [ toto ] , we estimate far superior than other methods e.g. * ? ? ? bg acknowledge support from cofin grant 2005025417 , and from the centre nationnal detudes spatiales . we are pleased to thank alessandra corsi , giulia stratta and an anonymous referee for very useful and constructive discussions . this work made use of data supplied by the uk swift science data centre at the university of leicester . amati , l. , frontera , f. , tavani , m. , et al . , 2002 , , 390 , 81 amati , l. , 2006 , , 372 , 233 atteia , j.l . , 2003 , , 407 , l1 berger , e. , kulkarni , s.r . , fox , d.b . , et al . , 2005 , , 634 , 501 bloom , j. s. , foley , r. j. , koceveki , d. , & perley , d. , 2006 , gcn # 5217 bor , m. & gendre , b. 2000 , a&a , 361 , l21 bor , m. , atteia , j.l . , damerdji , y. , et al . , 2006 , , 638 , l71 chattopadhyay , t. , misra , r. , chattopadhyay , a.k . , & naskar , m. , 2007 , , 667 , 1017 chevalier , r.a . , li , z.y . , & fransson , c. , 2004 , , 606 , 369 eldridge , j. j. , genet , f. , daigne , f. , & mochkovitch , r. , 2006 , mnras , 367 , 186 evans , p. a. , beardmore , a. p. , page , k. l. , et al . , 2007 , , 469 , 379 gendre , b. , bor , m. , 2005 , a&a , 430 , 465 gendre , b. , corsi , a. , & piro , l. , 2005 , , 455 , 803 gendre , b. , & bor , m. , 2006 aipc , 836 , 558 gendre , b. , galli , a. , corsi , a. , et al . , 2007 , , 462 , 565 ghirlanda , g. , ghisellini , g. , & lazzati , d. , 2004 , , 616 , 331 grupe , d. , nousek , j. , van den berk , d.e . , et al . , 2007 , , 133 , 2216 hjorth , j. , sollerman , j. , mller , p. , et al . 2003 , nature , 423 , 847 jakobsson , p. , levan , a. , fynbo , j. p. u. , et al . , 2006 , , 447 , 897 jaunsen , a.o . , hjorth , j. , bjrnsson , g. , et al . , 2001 , , 546 , 127 kann , d.a , klose , s. , & zeh , a. , 2006 , , 641 , 993 kouveliotou , c. , woosley , s.e . , patel , a. , et al . , 2004 , , 608 , 872 lamb & reichart , 2000 , , 536 , l1 liang , e. & zhang , b. , 2006 , , 638 , l67 malesani , d. , davanzo , p. , melandri , a. , et al . , 2004 , gcn # 2685 metzger , m.r . , djorgovski , s.g . , kulkarni , s.r . , et al . , 1997 , nature , 387 , 879 meszaros , p , 2006 , rep . phys . , 69 , 2259 nardini , m. , ghisellini , g. , ghirlanda , g. , et al . , 2006 , , 451 , 821 nousek , j. a. , kouveliotou , c. , grupe , d. , et al . , 2006 , , 642 , 389 obrien , p.t . , willingale , r. , osborne , j. , et al . , 2006 , , 647 , 1213 panaitescu , a. , & kumar , p. , 2000 , , 543 , 66 panaitescu , a. , & kumar , p. , , 571 , 779 , 2002 de pasquale , m. , piro , l. , perna , r. , et al . , 2003 , , 592 , 1018 de pasquale , m. , piro , l. , gendre , b. , et al . , 2005 , , 455 , 813 ramirez - ruiz , e. , dray , l.m . , madau , p. , tout , c.a . , 2001 , , 327 , 829 pedersen , k. , hurley , k. , hjorth , j. , et al . , 2006 , , 636 , 381 soderberg , a.m. , frail , d.a . , & wieringa , m.h . , 2004 , , 607 , l13 stanek , k. z. , matheson , t. , garnavich , p. m. , et al . , 2003 , , 591 , l17 watson , d. , reeves , j.n hjorth , j. , et al . , 2006 , , 637 , l69 willingale , r. , obrien , p. t. , osborne , j. p. , et al . , 2007 , , 662 , 1093 zhang , b. , fan , y.z . , dyks , j. , et al . , 2006 , , 642 , 354	we investigate the clustering of afterglow light curves observed at x - ray and optical wavelengths . we have constructed a sample of 61 bursts with known distance and x - ray afterglow . this sample includes bursts observed by bepposax , xmm - newton , chandra , and swift . we correct the light curves for cosmological effects and compare the observed x - ray fluxes one day after the burst . we check for correlations between the observed flux and the burst spectral and temporal properties . we confirm the previous result of bor and gendre ( 2000 ) that x - ray afterglow light curves cluster in luminosity , even when we consider the last swift data . we observe this clustering only for the afterglow light curves ; the inclusion of prompt - related data broaden the distribution . a similar clustering is observed for the optical light curves ; grb sources can be divided in three classes , namely optical and x - ray _ bright _ afterglows , optical and x - ray _ dim _ ones , and optically _ bright _ -x - ray _ dim _ ones . we argue that this clustering is related to the fireball total energy , the external medium density , the fraction of fireball energy going in relativistic electrons and magnetic fields . these parameters can be either fixed to a standard value , or correlated . we finally propose a method for the estimation of the grb source redshift based on the observed x - ray flux one day after the burst and optical properties . using this method , we compute a redshift of 1.4 @xmath0 0.2 for and of @xmath1 for . we tested this method on three recently detected swift grbs with known redshift , and found it in good agreement with the reported distance from optical spectroscopy .
in 1998 , anderson and putnam showed that a substitution tiling space can be represented as inverse limit of an inflation and substitution map on the cellular anderson - putnam complex @xcite . the cohomology of the tiling space can then be computed as the direct limit of the homomorphism induced by inflation and substitution on the cohomology of the complex . ghler and maloney @xcite investigated the cohomology of one - dimensional tiling spaces arising from several substitutions which are all acting on the same set of tiles . since each substitution is applied globally , the entropy remains zero and the cohomology remains finitely generated . the next natural goal is to go one step further by studying mixed substitutions that are randomly applied locally , which makes the entropy positive @xcite . in this study , we use the concept of rauzy graphs which , as shown in @xcite , is a very convenient tool to relate complexity and cohomology . for simplicity , we illustrate our method using the random fibonacci substitution even though it also works in more general cases , such as the random noble means substitutions , which were introduced and studied in @xcite . they are given by @xmath6 where @xmath7 is a fixed probability vector , that is @xmath8 and @xmath9 . the random fibonacci substitution is the case where @xmath10 . it was introduced by godrche and luck @xcite in 1989 , and , together with its generalised form , has been analyzed intensively by nilsson @xcite . a _ tile _ is a subset of @xmath11 that is homeomorphic to a closed unit ball . a _ partial tiling _ @xmath12 is a set of tiles , any two of which intersect only on their boundaries , and the _ support of _ @xmath12 is the union of its tiles . we define a _ tiling _ as a partial tiling whose support is @xmath11 . any set of tilings of @xmath11 can be equipped with a metric , in which two tilings are close if , up to a small translation , they agree on a large ball around the origin . we use the metric defined in @xcite : for any two tilings @xmath12 and @xmath13 of @xmath11 , @xmath14 where @xmath15 is the usual norm on @xmath11 and @xmath16 is the open ball of radius @xmath17 centered at @xmath18 in @xmath11 . the topology arising from this metric may be generated by the cylinder sets @xmath19 this means that if @xmath13 is in the cylinder set @xmath20 , then @xmath12 and @xmath13 agree on the ball @xmath21 up to translation by @xmath22 . in this study , we look at a set of tilings in @xmath23 . let @xmath24 and @xmath25 be closed intervals of lengths @xmath26 and 1 , respectively . let @xmath27 be the set of prototiles and let @xmath28 denote the set of all partial tilings that contain only translates of these prototiles . let @xmath5 . define the random fibonacci inflation rule as @xmath29 we say that it is a _ local random inflation _ because we apply @xmath30 locally , that is , for each occurrence of the tile @xmath24 we decide independently which among the @xmath31 possible realizations to choose . in particular , @xmath30 is an inflation map from @xmath27 to @xmath28 with inflation constant @xmath32 where the supports of @xmath33 and @xmath34 are @xmath35 and @xmath36 , respectively . then @xmath30 can be extended to a map @xmath37 by @xmath38 where @xmath39 and @xmath40 . the _ random fibonacci tiling space _ @xmath41 is the set of all tilings @xmath42 such that for any partial tiling @xmath43 with bounded support , we have @xmath44 for some @xmath40 . the following result is of interest because in the case of the global random substitution @xcite , for two distinct substitution sequences @xmath45 and @xmath46 the corresponding hulls may be disjoint . @xmath41 is connected . suppose that @xmath41 is not connected . then there exist open sets @xmath47 with @xmath48 and @xmath49 . let @xmath50 and @xmath51 . since @xmath50 and @xmath52 is open , there is a cylinder set @xmath53 contained in @xmath52 , for some @xmath54 ; that is , @xmath55 . choose some number @xmath56 , and let @xmath57 $ ] be a partial tiling in @xmath58 consisting of all tiles that intersect the ball @xmath59 . it is easy to see that @xmath60 is also contained somewhere in @xmath61 . then some translates of @xmath61 agree with @xmath58 on the ball @xmath59 . consider one such translate of @xmath61 , say @xmath62 for some @xmath63 . since @xmath56 , @xmath62 and @xmath58 also agree on the ball @xmath64 , which means that @xmath65 . now @xmath66 , the translation orbit of @xmath61 , which lies in a path component of @xmath41 . each path component of @xmath41 lies in a component of @xmath41 . also since a component of @xmath41 is a connected subspace of @xmath41 , it lies entirely within either @xmath52 or @xmath67 . since @xmath51 , the path component containing @xmath68 must lie in @xmath67 . hence @xmath69 . we now have @xmath70 and @xmath71 , which contradicts the assumption that @xmath48 . thus @xmath41 is connected . another way to look at one - dimensional tiling spaces is in terms of symbolic dynamics . let our alphabet be @xmath72 . a _ word _ @xmath73 of length @xmath74 over @xmath75 is a sequence @xmath76 of @xmath74 letters from @xmath75 . if @xmath77 and @xmath78 are two words , then we let @xmath79 denote the _ concatenation _ of @xmath80 and @xmath81 ; that is , @xmath82 . similarly , for two sets of word @xmath52 and @xmath67 , we let their product be the set @xmath83 which contains all possible concatenations . we define the set of _ bi - infinite words _ over @xmath75 to be @xmath84 , and let @xmath85 be the _ shift operator _ , @xmath86 . consider the substitution @xmath30 , where @xmath24 and @xmath25 are letters instead of intervals . as before , @xmath30 is randomly applied locally . define the _ symbolic random fibonacci tiling space _ @xmath87 as the set of all elements of @xmath88 such that for any @xmath89 and for any subword @xmath80 of @xmath73 , there exists a power @xmath90 where @xmath80 is a subword of some realisation of @xmath91 . clearly , @xmath87 is totally disconnected , compact , and shift - invariant . now , define the _ suspension _ @xmath92 of @xmath93 by @xmath94 where @xmath95 is the equivalence relation generated by the relations @xmath96 for all @xmath97 and all @xmath98 . the map @xmath99 is an action of @xmath23 defined by @xmath100):=[(w , t+s)].$ ] we have that @xmath92 is an @xmath23-dynamical system . it is easy to see that @xmath41 is homeomorphic to the random fibonacci tiling space defined previously , justifying the use of the same notation . moreover , the cohomology groups do not depend on the length of the tiles , which allows us to work on the symbolic level . before we can define a rauzy graph , we need a finite method for finding the set of subwords of the words in the symbolic random fibonacci tiling space @xmath87 . if we let @xmath30 act on letter @xmath25 repeatedly , it yields an infinite sequence of words @xmath101 . we call @xmath102 an _ inflated word in generation _ @xmath74 , and let @xmath103 be the set of all inflated words in generation @xmath74 . it can be defined recursively as follows : let @xmath104 , @xmath105 and @xmath106 , and for @xmath107 , @xmath108 . let the _ subword set _ @xmath109 be the set of all subwords of length @xmath74 of the words in a set @xmath110 . in @xcite , it is shown that , for @xmath111 , @xmath112 where the numbers @xmath113 are fibonacci numbers . let us now define the rauzy graphs associated with @xmath87 . the following are based on the definitions and results in @xcite . let @xmath114 , the set of all subwords of length @xmath74 of the words in @xmath87 . the @xmath74th rauzy graph @xmath115 associated with @xmath87 is an oriented graph with set of vertices @xmath116 and set of oriented edges @xmath117 , where the oriented edges are defined as follows : there is an edge from @xmath118 to @xmath119 if @xmath120 for all @xmath121 and @xmath122 . for any two vertices @xmath80 and @xmath81 , there is a path from @xmath80 to @xmath81 and a path from @xmath81 to @xmath80 as well , and so the rauzy graphs @xmath115 associated with @xmath87 are connected by oriented paths , that is , they are _ strongly connected_. moreover , the rauzy graphs are not just combinatorial objects but are also metric spaces . we will consider @xmath115 as the disjoint union of copies of the interval @xmath123 $ ] indexed over the edges modulo indentification of endpoints : @xmath124_{(xfy)}/\sim\ ] ] where @xmath125 and @xmath123_{(xfy)}$ ] is an edge with tail @xmath126 and head @xmath127 . two endpoints are identified if they stand for the same subword . we now show that the tiling space @xmath41 is an inverse limit of rauzy graphs . clearly , a word of length @xmath128 can be shortened to a word of length @xmath74 . this fact allows us to define projection maps @xmath129 , which we need to build inverse limits . define @xmath130 in the following way . if @xmath74 is even , @xmath131 is defined by 1 . @xmath132 if @xmath133 ; 2 . if an edge @xmath134 goes from @xmath80 to @xmath81 , then there exists in @xmath115 an edge from @xmath135 to @xmath136 ; the application @xmath131 then maps @xmath137 to that edge , and its restriction on @xmath137 is the identity map on the segment @xmath123 $ ] . if @xmath74 is odd , @xmath131 is similarly defined by 1 . @xmath138 if @xmath133 ; 2 . the definition of @xmath131 on the edges is exactly the same as above . c c [ cols="^ " , ] + the proof of the following proposition is straightforward . [ qou ] the maps @xmath131 are surjective , continuous , open maps . let us now consider the inverse limits of the rauzy graphs with respect to the maps @xmath131 . the inverse limit of the complex @xmath139 is the following subset of @xmath140 : @xmath141 which is denoted by @xmath142 . the inverse limit is a topological space with the relative topology induced by the product topology of @xmath140 . it is a compact set , being a closed subset of a compact set . the following theorem has been proved in @xcite for the one - dimensional standard substitution tiling spaces . that argument can be used with very minimal modification to prove the case for the random fibonacci tiling space @xmath41 . [ invlim ] let @xmath41 denote the suspension of @xmath93 . then we have the following homeomorphism : @xmath143 as before , we denote by @xmath144 the set of all subwords of length @xmath74 of the words in @xmath87 , and denote by @xmath145 the union of @xmath146 for all @xmath74 . note that @xmath145 contains all the subwords of any length of the words in @xmath87 , and for every word @xmath147 there is a pair of letters @xmath148 such that @xmath149 . we define the subword complexity function @xmath150 as the cardinality of the set @xmath146 , and let the first difference of the complexity be denoted by @xmath151 . the number of vertices of the @xmath74th rauzy graph @xmath115 is equal to the complexity @xmath150 , and its number of edges is @xmath152 . the following proposition from @xcite makes an explicit link between @xmath153 and the cohomology of the @xmath74th rauzy graph @xmath115 . [ rk ] the rank of the cohomology @xmath154 of the @xmath74th rauzy graph equals @xmath155 . we now introduce some useful concepts defined in @xcite . every element of length @xmath74 in @xmath145 can be extended in at least one way to an element of length @xmath128 by appending a letter to the right , we call those elements for which this can be done in two different ways as _ right specials_. similarly , the words @xmath147 such that @xmath156 and @xmath157 are called _ left specials_. it is easy to see that @xmath153 may be interpreted as the number of right special elements of length @xmath74 in @xmath145 , or as the number of left special elements of length @xmath74 in @xmath145 . if an element @xmath81 of @xmath145 is both right special and left special , we say that @xmath81 is _ bispecial_. there are three kinds of bispecial elements , depending on the cardinality of the set @xmath158 . when @xmath159 has four elements , @xmath81 is a _ strong bispecial _ , and when it has only two elements , @xmath81 is a _ weak bispecial_. if we let @xmath160 denote the number of strong bispecials of length @xmath74 and @xmath161 the number of weak bispecials of length @xmath74 , then it can be shown that @xmath162 . the following lemma implies that , for the random fibonacci tiling space @xmath41 , there are no weak bispecials in @xmath145 . thus , we have @xmath163 , which tells us that @xmath153 is nondecreasing . [ wb ] suppose @xmath164 such that both @xmath165 and @xmath166 ( @xmath167 and @xmath168 ) are in @xmath145 . then , @xmath165 and @xmath166 ( @xmath167 and @xmath168 ) can be legally extended to the left ( right ) both with an @xmath24 , or both with a @xmath25 . that is , either @xmath169 or @xmath170 ( @xmath171 or @xmath172 ) . we prove the case where @xmath81 is first extended to the right . the other case is analogous . as a legal word , @xmath81 occurs somewhere in an infinite word @xmath73 coming from a substitution . thus , the word @xmath73 has a partitioning into first order supertiles @xmath173 $ ] , @xmath174 $ ] , and @xmath175 $ ] . different occurrencies of @xmath176 may have different partionings into supertiles , but if @xmath81 is extensible to the right in two ways , there must be an occurrence of @xmath176 which ends on a supertile boundary . otherwise , only one continuation would be possible . as @xmath81 can be extended to the right in two ways , it must be possible to extend it with a supertile @xmath173 $ ] , and thus also with @xmath174 $ ] . if only @xmath175 $ ] was possible , there would only be one way to extend . now take an infinite word @xmath73 with a fixed occurrence of @xmath177 $ ] . as @xmath73 is random substitution word , we can swap the pair @xmath173 $ ] to the right of @xmath81 , leaving the rest of @xmath73 unchanged , and obtain another legal word @xmath178 , containing @xmath179 $ ] instead . as to the left of @xmath81 both @xmath73 and @xmath178 agree , the claim follows . in theorem [ invlim ] , we have seen that the suspension space @xmath41 is homeomorphic to the inverse limit of the rauzy graphs . this implies that the @xmath180ech cohomology of @xmath41 is the direct limit of the cohomology of the rauzy graphs . in particular , the @xmath180ech cohomology group @xmath181 is isomorphic to the direct limit of the system of abelian groups @xmath182 we want to show is that for the random fibonacci tiling space @xmath41 , the maps @xmath183 are injective . the proof of this uses the theory of quotient cohomology , which was introduced in @xcite . we now introduce some notation relating to @xmath180ech cohomology . if @xmath184 is a topological space with the structure of a cw - complex , let @xmath185 denote a cell in @xmath184 , and let @xmath186 denote the corresponding cochain . let @xmath187 denote the group of @xmath188-cochains of @xmath184 , and @xmath189 denote the coboundary map . the theory of quotient cohomology applies to topological spaces @xmath184 and @xmath190 for which there is a quotient map @xmath191 such that the pullback @xmath192 is injective on cochains . define the cochain group @xmath193 to be the quotient @xmath194 , and the let the quotient coboundary operator to be @xmath195 . then the quotient cohomology between @xmath184 and @xmath190 is defined to be @xmath196 . then , by the snake lemma , the short exact sequence of cochain complexes @xmath197 induces a long exact sequence @xmath198 [ inj ] if @xmath115 are the rauzy graphs of the random fibonacci tiling space @xmath41 , the induced maps @xmath199 are injective . the following proof is adapted from @xcite . let @xmath200 and @xmath201 . by proposition [ qou ] , the map @xmath202 is a quotient map , and the pullback of this map is injective on cochains . then we have the following exact sequence @xmath203 @xmath204 to prove that @xmath205 is injective , we show that @xmath206 . to this end , we show that if @xmath207 such that @xmath208 , then @xmath209 . suppose @xmath74 is odd . then @xmath138 . in the following , we use these notations : for some @xmath210 , @xmath211 and @xmath212 . the image of a cochain @xmath213 under @xmath205 is @xmath214 . furthermore , supposing @xmath207 and @xmath208 , we have @xmath215 and @xmath216 for some @xmath217 . on the other hand , by definition of @xmath218 , we have @xmath219 where summations are taken over @xmath220 . taking modulo @xmath221 , we get @xmath222 by lemma [ wb ] , for some @xmath223 , if @xmath224 and @xmath225 are elements of @xmath145 , then @xmath226 or @xmath227 . the first case implies that @xmath228 and @xmath229 , while from the second case we have @xmath230 and @xmath231 . thus , in particular , the coefficients @xmath232 depend on @xmath233 alone , and so we have @xmath234 hence , @xmath18 lies in @xmath235 . similar arguments can be used for the case where @xmath74 is even . in @xcite , julien showed that if the complexity function @xmath150 is bounded by @xmath236 for some constant @xmath237 , then the cohomology is rationally finitely generated . this is the case for the global random substitution @xcite where @xmath150 is bounded , and hence the cohomology is also rationally finitely generated . however , in our case the system has positive entropy which means that @xmath150 grows exponentially , and @xmath153 is nondecreasing . by proposition [ rk ] and proposition [ inj ] , the rank of the cohomology of @xmath115 grows without bound , and the kernels of the bonding maps between them vanish . as a consequence , the direct limit can not be finitely generated and we can deduce the following . the ech cohomology group @xmath181 is not finitely generated . the authors would like to thank paolo bugarin , greg maloney , and johan nilsson for fruitful discussions . e. provido would also like to thank the alexander von humboldt foundation for the generous support . this work is also supported by dfg via the collaborative research centre 701 and the research center for mathematical modelling ( rcm^2^ ) at bielefeld university . m. baake and m. moll , in : _ aperiodic crystals _ , doi : 10.1007/978 - 94 - 007 - 6431 - 6_3 . s. schmid , r.l . withers , r. lifshitz , springer , dordrecht 2013 , p. 19 . m. moll , ph.d . thesis , universitt bielefeld , bielefeld 2013 . j. cassaigne , in _ developments in language theory ii , at the crossroads of mathematics , computer science and biology _ , isbn 981 - 02 - 2682 - 9 . j. dassow , g. rozenberg , a. salomaa , world scientific , singapore 1996 , p 25 .	we look at the topology of the tiling space of locally random fibonacci substitution , which is defined as @xmath0 with probability @xmath1 , @xmath2 with probability @xmath3 and @xmath4 for @xmath5 . we show that its ech cohomology group is not finitely generated , in contrast to the case where random substitutions are applied globally . pacs : 02.40.re , 45.30.+s , 61.44.br 2
.sensitivity to quartic anomalous couplings for all possible scattering processes [ cols="<,<,^,^,^,^,^",options="header " , ] [ krk - t4 ] table [ trtmrt ] and table [ krk - t4 ] contain results for weak boson scattering assuming integrated luminosity of 1000fb@xmath3 in @xmath4 mode in su(2)@xmath5 conserving case and broken su(2)@xmath5 . we consider now the reactions @xmath6 and @xmath7 . on tree level the elementary process producing the @xmath8 final state is driven by 15 feynman diagrams . only one of the diagrams contains the quartic coupling and has to be extracted from the other interfering terms . only the part containing a longitudinal gauge boson is expected to give a sizable signal related to electroweak symmetry breaking . for @xmath8 this part is substantially enhanced using polarized beams . we investigate several cases : i ) unpolarized , ii ) @xmath9 right polarized electrons , and iii ) @xmath9 right polarized electrons along with @xmath10 left polarized positrons . for @xmath11 polarization is not substantial , since the standard model background is much smaller ( two diagrams ) . calculations are done using the whizard event generator @xcite . since the gauge bosons are short lived states they decay . presently we consider on - shell gauge bosons only ( narrow width approximation ) and hadronize the final state using pythia @xcite.the three - boson state is characterized by three four - momenta and the bosonic spins . in general three momenta lead to 12 kinematical variables that are reduced by four through energy momentum conservation , by three because of the on - shell condition mentioned before , and by two due to rotational invariance . hence in total three independent kinematical variables are left . we choose two invariant masses of the dalitz plot , @xmath12 , @xmath13 and the angle @xmath14 between the beam axis and the direction of the @xmath15-boson . spin of the bosons leads to additional degrees of freedom , and we may distinguish longitudinal ( @xmath16 ) from transverse ( @xmath17 ) polarization . presently , we do not yet analyze the bosonic spins . the three independent kinematical variables lead to a three dimensional histogram . if the angle @xmath14 is not measured the resulting two dimensional histogram leads a dalitz plot . we investigate the differences on the histograms as a function of @xmath18 and @xmath19 . the observable are discretize into bins and @xmath20 is given by @xmath21 where @xmath22 denotes the error , and @xmath23 the sums over bins of @xmath12 , @xmath13 , and @xmath14 . we use the whizard generator to produce standard model events corresponding to a luminosity of 1000 fb@xmath24 . the detector efficiency is simulated using the fast simulation simdet @xcite . to reconstruct @xmath8 , we use the decay @xmath25 jets . the dominant background is due to @xmath26 jets . in the absence of a full simulation at 1000 gev we estimate from our previous studies at 500 gev @xcite a combined effect of efficiency and purity to be 42% . a full simulation at 1000 gev is presently under development . the probabilities of standard model events are reweighted when introducing anomalous couplings @xmath18 , @xmath19 . since the effective lagrangian is linear in @xmath18 , @xmath19 any observable is of second order in the parameters and can be expressed by a polynomial with five parameters . the parameters are determined by evaluation of @xmath27 for each event and for five pairs of fixed values @xmath28 . by inversion @xmath29 is know for arbitrary values of @xmath28 , viz . @xmath30 for each bin @xmath23 . finally we calculate @xmath20 and determine @xmath31 and @xmath32 for the specific values @xmath33 ( 68.3% confidence ) and @xmath34 ( 90% confidence ) . results are shown in fig . [ fig : result ] . we find that the sensitivity drastically increases with polarization . sensitivity can be improved even further by utilizing meaningful cuts , which has not been done in the present stage of analysis and by using the information of angular distribution of the jets that depends on the polarization stage of the final bosons . expected limits on the measurement of anomalous quartic couplings from all possible weak boson scattering processes were presented . in fig . [ fig : upsala ] we make a comparison of estimated sensitivites from weak boson scattering processes and ongoing triple boson production analysis . with the same integrated luminosity and @xmath9 left @xmath35 and @xmath36 right @xmath37 polarization for scattering and @xmath9 right @xmath37 and @xmath10 left @xmath35 polarization for triple producstion we obtain comparable results . this shows that luminosity sharing of opposite polarization can probably lead to the same overall accuracy for the measurement of quartic boson couplings . the same luminosity based conclusion was made after comparison of @xmath38 and @xmath39 running modes leaving the experimental physicist several ways to achieve the desired precision .	abstract . if no light higgs boson exist , the interaction among the gauge bosons becomes strong at high energies ( @xmath0 1tev ) . the effects of strong electroweak symmetry breaking ( sewsb ) could manifest themselves as anomalous couplings before they give rise to new physical states , thus measurement of all couplings and their possible deviation from standard model ( sm ) values could give valuable information for understanding the true nature of symmetry breaking sector . here we present a detail study of the measurement of quartic gauge couplings in weak boson scattering processes and a possibility for same measurement in triple weak boson production . expected limits on the parameters @xmath1 and @xmath2 in electroweak chiral lagrangian are given .
one important property of field theories with spontaneous symmetry breaking is that the life - time of energetically meta - stable vacuum states can be very long . this can lead to the partitioning of the universe into regions of different meta - stable vacuum states . one of these structures is called cosmic strings @xcite . such kind of structures represent thin tubes of false vacuum and are expected to be of large linear mass density @xmath5 and can also carry intrinsic angular momentum @xmath6 . the cosmic strings affect the spacetime mainly topologically , giving a conical structure to the space region around the cone of the string . the conical topology may be responsible for several gravitational effects , gravitational lensing @xcite and particle production due to the changing gravitational field during the formation of such object @xcite being some examples . there is a lot of papers studying quantum processes in a cosmic string spacetime . of special interest to us are the following : @xcite where pair production in a straight cosmic string spacetime is discussed , @xcite where the response function of detectors in the presence of the cosmic strings is calculated , and @xcite where the spinning cosmic string spacetime is investigated . finally spinning cosmic strings were studied also by galtsov _ et al . _ and also letelier . these authors investigated chiral strings generating a chiral conical spacetime @xcite . in this paper we are interested in two calculations . first , to analyse the energy momentum distribution of a time dependente spinning cosmic string spacetime . second , to estimate the particle production due to the changing gravitational field in the situation of gradual loss of angular momentum . the einsteins gravity equation of general relativity is given by : @xmath7 where the constant @xmath8 is defined to give the correct newtonian limit , resulting in @xmath9 @xmath10 and @xmath11 appearing in the latter equation represent the newtonian gravitational constant and light velocity , respectively . in order to obtain the geometry generated by a rotating cosmic string we proceed in the following way . first , we choose a cylindrical coordinate system ( @xmath12 ) in which an infinitely long and thin cosmic string lies along the @xmath13-axis . we consider a stationary string carrying linear densities of mass @xmath5 and angular momentum @xmath14 . the mass density is proportional to a two - dimensional delta function ( @xmath15-function ) while the angular momentum density is proportional to the derivatives of @xmath15-functions . the @xmath16 and @xmath17 components of the associated energy momentum tensor will be proportional to @xmath15-function and its derivatives , respectively . thus , einstein equations leads to the geometry @xmath18 = \left ( \begin{array}{cccc } c^2 & 0 & 4g{\cal j}_{\scriptscriptstyle 0}/c^2 & 0 \\ 0 & -1 & 0 & 0 \\ 4g{\cal j}_{\scriptscriptstyle 0}/c^2 & 0 & -\left(b^2r^2 - 16g^2{\cal j}_{\scriptscriptstyle 0}^2/c^6\right ) & 0 \\ 0 & 0 & 0 & -1 \end{array } \right ) \label{3}\ ] ] with @xmath19 from these metric components we can construct the associated line element @xmath20 therefore , as was noticed by deser and jackiw @xcite , the two - parameter @xmath21 metric tensor showed in equation ( [ 3 ] ) represents the general time - independent solution to gravitational einstein equations outside any matter distribution lying in a bounded region on the plane and having cylindrical symmetry . in this work we will consider only the exterior region of the cosmic string . thus the @xmath15-functions will be suppressed in our analysis of the energy momentum tensor generating a spacetime configuration due to a cosmic string loosing angular momentum . in the same way the exact spacetime metric representing the vacuum solution of a static cylindrically symmetric cosmic string is found @xcite to be @xmath22 obviously this metric is a particular case of the former , equation ( [ 5 ] ) , with null density of angular momentum . thus it arises the questions : how can we go from a spinning cosmic string spacetime to a static one and what is the rate of particle production between these two asymptotic limits ? first of all , let us assume that the cosmic string is generated in such way that there is spontaneous loss of energy associated with change in its rotation velocity , i.e. , loss of density of angular momentum . quantitatively such process would correspond to a non - stationary spacetime geometry that could be described by the metric ( [ 5 ] ) with @xmath14 replaced by a general function of time @xmath23 : @xmath24^{2 } - dr^{2 } - b^{2}r^{2}d\varphi^{2}-dz^{2 } \label{7}\ ] ] and provided with the asymptotic conditions @xmath25 a specific choice for the density of angular momentum that solves the above relations is given by @xmath26 . \label{10}\ ] ] in the following section we will look for the kind of matter content that generates such spacetime geometry . as we can notice , the region between the two asymptotic limits represents a curved spacetime . from einstein equations we show that the fluid configuration characterizing such geometry is represented by the following energy momentum tensor components @xmath27 = -\frac{2g{\cal j}_{\scriptscriptstyle 0}}{\kappa r \cosh t}\left ( \begin{array}{cccc } 0 & 0 & 0 & 0 \\ 0 & \frac{2gj_{\scriptscriptstyle 0}}{b^2 r}f(t ) & 1 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 0 & \frac{2gj_{\scriptscriptstyle 0}}{b^2 r}f(t ) \end{array } \right ) \label{energia}\ ] ] with @xmath28 defined by : @xmath29 let us now analyze the properties of the above characterized fluid . first of all we define a 4-velocity vector field @xmath30 @xmath31 and the projector tensor @xmath32 @xmath33 in the standard way , we express the energy momentum tensor in terms of its irreducible parts as @xmath34 where we introduced the quantities related with the fluid , i.e. , energy density , isotropic pressure , heat flux and anisotropic pressure , defined respectively as @xmath35 for the case we are analyzing here only isotropic and anisotropic pressure survive , and they result in : @xmath36 and @xmath37 = -\frac{p(t , r)}{2 } \left ( \begin{array}{cccc } 0 & 0 & 0 & 0 \\ 0 & 1 & \tilde{\pi } & 0 \\ 0 & \tilde{\pi } & -2 & 0 \\ 0 & 0 & 0 & 1 \end{array } \right ) \label{energia}\ ] ] where @xmath38 from what we have seen , outside the core of the string there is not energy density of matter but only flux of energy , that appears as pressure . from the einstein equations we obtain the relation between the trace of the energy momentum tensor and the curvature of the spacetime ( @xmath39 ) , that results in : @xmath40 = 80ex = 80ex the geometry of the spacetime is flat in the asymptotic regions corresponding to infinite past and infinite future . therefore it is worth to analyze the behavior of the curvature in the intermediate region . in the figure [ figure ] we plot the curvature @xmath41 as function of time @xmath42 and radial coordinate @xmath43 ( distance from the @xmath13 axes of the string ) . for @xmath44 the spacetime is approximately flat for all values of @xmath43 , and for large values of @xmath43 it will still be flat for all values of @xmath42 , as we would expect . as time @xmath42 goes from negative to positive values the curvature decreases until it reaches a minimum value at @xmath45 . after this point it begins to increase , becomes positive and reaches a maximum value at @xmath46 . finally it asymptotically goes to zero . the two asymptotic spacetimes are flat . the only difference between them is that for @xmath0 the string is spinning with constant density of angular momentum @xmath47 . as a general result of this analysis we obtained that the loss of angular momentum by a cosmic string causes a change in the metric properties of the spacetime . but changes in the gravitational field generate particles @xcite . thus we have to investigate the creation of particles and radiation by the changing gravitational field during the loss of angular momentum of a initially spinning cosmic string . we will treat this problem by performing the calculation of the bogoliubov transformation entailed by this process . from the general geometry ( [ 7 ] ) and the conditions ( [ 8 ] ) and ( [ 9 ] ) , in the two asymptotic regions the infinite past and infinite future the spacetime metric structures reduce to : @xmath48 the above metrics represent the structure of the spacetime in the exterior region of a rotating and a non - rotating cosmic string , respectively . in this section we will analyze the rate of particle production by the changing gravitational field between the two asymptotic spacetimes during the evolution of a cosmic string that looses angular momentum . a similar idea was used by bernard and duncan @xcite to study a two dimensional robertson - walker model where the conformal scale factor has the same functional form as equation ( [ 10 ] ) . in the two asymptotic limits , the spacetime becomes minkowskian , and they obtained the mode solutions of the klein - gordon equation in these two limits . a straightforward calculation of the bogoliubov coefficients between the _ in _ and _ out _ modes gives the rate of particle production during the expansion of the universe . it is worth to notice that such method was first introduced by l. parker @xcite in order to calculate particle creation by the expansion of the universe . let us consider a massive minimally coupled hermitian scalar field @xmath49 defined at all points of the spacetime ( [ 7 ] ) . the klein - gordon equation is given by : @xmath50\phi(t , r,\varphi , z ) = 0 \label{v8}\ ] ] where the symbol @xmath51 represents the covariant derivative with respect to the metric @xmath52 , and @xmath53 is the mass of the scalar field . to maintain the energy produced in a limited region of the space we impose dirichlet boundary conditions at @xmath54 , @xmath55 and periodic boundary conditions in @xmath13 with period @xmath56 . in order to circumvent the problem of the generation of closed time - like curves ( ctc ) , we impose an additional vanishing boundary condition at @xmath57 . thus , the radial coordinate has the domain @xmath58 . for a careful study on how to construct quantum field theory in a spacetime with ctc , see for instance @xcite . the same problem appears in @xmath59 dimensional gravity since the spinning cosmic string spacetime is exactly the solution of @xmath59 einstein equations of a spinning point source @xcite . the klein - gordon equation for the initial spacetime ( [ v6 ] ) reduces to the form : @xmath60\phi(t , r,\varphi , z)=0 . \label{v9}\ ] ] the mode solutions @xmath61 are found to be . ] @xmath62 \label{v16}\ ] ] with @xmath63 and @xmath64 in the above , @xmath65 . choosing the constant @xmath66 to make the set orthonormal results : @xmath67 ^ 2 - v_{0}\left[\frac{j^{'}_{\mu}(qr_{0})}{j_{\mu}(qr ) } - \frac{y^{'}_{\mu}(qr_{0})}{y_{\mu}(qr ) } \right]^2 \right\}^{-\frac{1}{2 } } \label{v33}\ ] ] where we defined the 3-volumes @xmath68 and @xmath69 . the values of the parameter @xmath70 are determined by a transcendental equation which comes from the vanishing boundary conditions , that is , @xmath71 and the infinite number of its roots are labeled by @xmath72 . the modes @xmath73 , form a basis in the space of the solutions of the klein - gordon equation and can be used to expand the field operator as : @xmath74 . \label{v35}\ ] ] the creation and annihilation operators @xmath75 and @xmath76 satisfy the commutation relation : @xmath77 = \delta_{j , j ' } \label{aa}\ ] ] and we define the in - vacuum state by @xmath78 in the same way we can perform the canonical quantization of the field in the infinite future . the klein - gordon equation in the non - rotating cosmic string spacetime ( [ v7 ] ) reads @xmath79\phi(t , r,\varphi , z)=0 . \label{v13}\ ] ] the mode solutions @xmath80 are found to be : @xmath81 \label{v37}\ ] ] with @xmath82 and @xmath83 choosing the constant @xmath84 in order to make the set of modes @xmath85 orthonormal , results : @xmath86 - v_{0}\left[\frac{j^{'}_{\nu}(\bar{q}r_{0})}{j_{\nu}(\bar{q}r ) } - \frac{y^{'}_{\nu}(\bar{q}r_{0})}{y_{\nu}(\bar{q}r)}\right ] \right\}^{-\frac{1}{2 } } \label{v33}\ ] ] where the values of @xmath87 are determined by a transcendental equation of the same type as before . the out modes , solutions of the klein - gordon equation , also form a complete set and can be used to expand the field operator as : @xmath88 . \label{v36}\ ] ] the creation and annihilation operators @xmath89 and @xmath90 satisfy the usual commutation relation : @xmath91 = \delta_{j , j ' } \label{bb}\ ] ] and the out - vacuum state is defined by @xmath92 following parker we will calculate the rate of particle production between two asymptotic spacetimes discussed above : the spinning cosmic string spacetime in the infinite past and a non - spinning cosmic string spacetime in the infinite future . an important point is that in our model we have not to deal with the problems of junction conditions since there is no sudden approximation here . the metric evolves continuously between both asymptotic states . the angular momentum of the spinning cosmic string is lost by particle emission processes . the fundamental quantity we have to calculate is the bogoliubov coefficients between the modes in the non - spinning and spinning cosmic string spacetime . the average number of out - particles in the modes @xmath93 produced by this process is : @xmath94 using the definition of the bogoliubov coefficients @xmath95 given by @xmath96 we have @xmath97 where @xmath98\left[\frac{j_{\nu}(\bar{q}r)}{j_{\nu}(\bar{q}r ) } - \frac{y_{\nu}(\bar{q}r)}{y_{\nu}(\bar{q}r)}\right]\ ] ] by substituting ( [ beta ] ) in equation ( [ number ] ) and using the definitions of the normalization constants @xmath99 and @xmath100 , the average number of particles produced in the mode @xmath101 is : @xmath102 ^ 2 \nonumber\\ & \times&\hspace{-3 mm } \left\ { v \left[\frac{j^{'}_{\mu}(qr)}{j_{\mu}(qr ) } - \frac{y^{'}_{\mu}(qr)}{y_{\mu}(qr ) } \right]^2 - v_{0}\left[\frac{j^{'}_{\mu}(qr_{0})}{j_{\mu}(qr ) } - \frac{y^{'}_{\mu}(qr_{0})}{y_{\mu}(qr ) } \right]^2 \right\}^{-1}.\end{aligned}\ ] ] the expression of the average number of particles in the mode @xmath103 is very complicated , and some simplifications used by mendell and hiscock can not be made here . the key point is that in parker s work use was made of the sudden approximation , i.e. , for @xmath104 the spacetime is minkowskian and for @xmath105 the geometry is conic . mendell and hiscock extended parker s work considering a number of different models still using the sudden approximation . consequently if @xmath106 is the actual time of formation of the string , the production of particles in modes with frequencies that are large compared with @xmath107 are suppressed . in our model , the process of particle production by loss of angular momentum takes an infinite time , since we have two asymptotically flat spacetimes with curved geometry between both states . nevertheless some conclusions can be obtained . note that the part of the above expression that comes between braces does not depend on the energy of the produced particle @xmath108 . from this it is clear that the number of produced particles diverges for low and high energy modes . this behavior is expected for the low energy modes , but is not for the high ones . a way to improve our model is to assume a finite time @xmath109 for the loss of the angular momentum to occur , thereby introducing a natural cutoff in the energy of the produced particles , because now the production of modes for which the energy is large compared with @xmath110 will be also suppressed , as in the case of the sudden approximation . in this paper we discussed the properties of a fluid generating a general spinning cosmic string spacetime with flat limiting cases corresponding to a constant angular momentum in the infinite past and static configuration in the infinite future . we also analyze the particle production by loss of angular momentum in a spinning cosmic string spacetime . to circumvent the problem of ctcs we assumed a cosmic string with a radius fixed . moreover , to keep energy and number of particles produced by the process in a finite region of space , following parker s arguments , we impose vanishing boundary conditions on the wall of a cylinder with finite radius @xmath41 . a possible continuation of this paper is to formulate the energy conservation law , that is , to show if there is a balance between the total energy of the particles created and the energy associated with loss of angular momentum . this can be done comparing the vacuum stress - tensor of the massive field in the spinning and non - spinning cosmic string spacetimes . the calculation for a massless conformally coupled scalar field in the non - spinning cosmic string spacetime has been done by many authors @xcite . the same calculation in the spinning cosmic string spacetime has been done by @xcite . as far as we know the renormalized stress tensor of a massive minimally coupled scalar field has not been investigated in the literature . this paper was partially supported by _ conselho nacional de desenvolvimento cientfico e tecnolgico _ ( cnpq ) , of brazil . p. c. w. davies and v. sahni , class . * 5 * , 1 ( 1988 ) ; b. f. svaiter and n. f. svaiter , class . * 11 * , 347 ( 1994 ) ; l. iliadakis , u. jasper and j. audretsch , phys . rev . * d 51 * , 2591 ( 1995 ) . p. o. mazur , phys . lett . * 57 * , 929 ( 1986 ) ; d. harari and a. p. polychronakos , phys . rev . * d 38 * , 3320 ( 1988 ) ; b. jensen and h. h. soleng , phys . rev . * d 45 * , 3528 ( 1992 ) ; c. doran , a. lasenby and s. gull , phys . rev . * d 54 * , 6021 ( 1996 ) . s. deser , r. jackiw and g. thooft , ann . phys . * 152 * , 220 ( 1984 ) ; ibid * 153 * , 405 ( 1984 ) ; j. r. gott iii and m. alpert , gen . rel . grav . * 16 * , 243 ( 1984 ) ; s. giddings , j. abhott and k. kuchar , gen . 16 * , 751 ( 1984 ) . t. m. helliwell and d. a. konkowski , phys . rev . * d 34 * , 1918 ( 1986 ) ; b. linet , phys . rev . * d 35 * , 536 ( 1987 ) ; w. a. hiscock , phys . lett . * b 188 * , 317 ( 1987 ) ; j. s. dowker , phys . rev . * d 36 * , 3095 ( 1987 ) ; p. c. w. davies and v. sahni , class . * 5 * , 1 ( 1988 ) .	we analyze the properties of a fluid generating a spinning cosmic string spacetime with flat limiting cases corresponding to a constant angular momentum in the infinite past and static configuration in the infinite future . the spontaneous loss of angular momentum of a spinning cosmic string due to particle emission is discussed . the rate of particle production between the spinning cosmic string spacetime ( @xmath0 ) and a non - spinning cosmic string spacetime ( @xmath1 ) is calculated . * from spinning to non - spinning cosmic string spacetime * + v. a. de lorenci@xmath2 r. d. m. de paola@xmath3 and n. f. svaiter @xmath4 + + pacs numbers : 03.60.bz , 04.20.cv , 11.10.qr
comparison between experimental observation and theoretical analysis is a crucial step in condensed matter physics researches . temperature dependence of the specific heat and the magnetic susceptibility , for example , has been studied to extract nature of low energy excitations and magnetic interactions among electrons , respectively , through comparison with the landau s fermi liquid theory @xcite and curie - weiss law @xcite . for flexible and quantitative comparison with experimental data , an exact diagonalization approach @xcite , which can simulate quantum lattice models without any approximations is one of the most reliable numerical tools . most numerical methods for general quantum lattice problems , which is not accompanied by the negative sign difficulty , are not exact and reliable error estimate for them is difficult . this drawback is serious especially when one wants to compare the results with experiments therefore , exact methods are valuable though they are usually available only for small size problems . they are important in two ways ; as a tool for obtaining reliable answers to the original problem ( when the original problem is a small size problem ) and as a generator of reference data with which one can compare the results of other numerical methods ( when the original problem is beyond the applicable range of the exact methods ) . for the last few decades , the numerical diagonalization package for quantum spin hamiltonians , titpack @xcite has been widely used in the condensed matter physics community . based on titpack , other numerical diagonalization packages such as kobepack @xcite and spinpack @xcite have also been developed . there is another program performing such a calculation in the alps project ( algorithms and libraries for physics simulations ) @xcite . however , the size of the target system is limited for titpack , kobepack , and alps because they do not support the distributed memory parallelization . although spinpack supports such parallelization and an efficient algorithm with the help of symmetries , this program package can not handle the general interactions such as the dzyaloshinskii - moriya and the kitaev interactions that recently receive extensive attention . in addition , advances in quantum statistical mechanics @xcite open a new avenue for exact methods to finite - temperature calculations without the ensemble average . this development enables us to calculate finite - temperature properties of quantum many - body systems with computational costs similar to calculations of ground state properties . now , it is possible to quantitatively compare theoretical predictions for temperature dependence of , for example , the specific heat and the magnetic susceptibility with experimental results @xcite . however , the above program packages have not support this calculation yet . to perform the large - scale calculations directly relevant to experiments by utilizing the parallel computing architectures with small bandwidth and distributed memory , a user - friendly , multifunctional , and parallelized diagonalization package is highly desirable . @xmath0 [ _ aitch - phi _ ] , a flexible diagonalization package for solving a wide range of quantum lattice models , has been developed to overcome the problems in the previous program packages . in @xmath0 , we implement the lanczos method for calculating the ground state and properties of a few excited states , thermal pure quantum ( tpq ) states @xcite for finite - temperature calculations , and full diagonalization method for checking results of lanczos and tpq methods at small clusters , with an easy - to - use and flexible user interface . by using @xmath0 , one can analyze a wide range of quantum lattice models including conventional hubbard and heisenberg models , multi - band extensions of the hubbard model , exchange couplings that break su(2 ) symmetry of quantum spins such as the dzyaloshinskii - moriya and the kitaev interactions , and the kondo lattice models describing itinerant electrons coupled to quantum spins . it is easy to calculate a variety of physical quantities such as the internal energy , specific heat , magnetization , charge and spin structure factors , and arbitrary one - body and two - body static green s functions at zero temperature or finite temperatures . as we mention above , one of the aims of developing @xmath0 is to make a flexible diagonalization program package that enables us to directly compare the theoretical calculations and experimental data . in addition to the simple model hamiltonians introduced in textbooks of quantum statistical physics , more complicated hamiltonians inevitably appear to quantitatively describe electronic properties of real compounds . in the lanczos and the tpq simulation of the quantum lattice model in the condensed matter physics , the most time - consuming part is the multiplication of the hamiltonian to a wavefunction . excepting the case for the very small system , it is unrealistic to store all non - zero elements of the hamiltonian in memory . therefore we perform the multiplication by constructing the matrix elements of the hamiltonian on the fly . to carry out the lanczos and tpq calculations efficiently , we implement two algorithms for massively parallel computation ; one is the conventional parallelization based on the butterfly - structured communication pattern @xcite ( the butterfly method ) and the other is newly developed method [ the continuous - memory - access ( cma ) method ] . because the numerical cost of the butterfly method is proportional to the number of terms in the second quantized hamiltonian , it requires very long time to simulate the complicated hamiltonian relevant to the real compounds . in contrast to the conventional algorithm , the newly developed method can realize the continuous memory access and numerical cost which does not depend on the number of terms during multiplication between the hamiltonians and a wavefunction . in this paper , we also explain these two algorithms and compare their computational speed . this paper is organized as follows : in sec . 2 , we introduce the basic usage of @xmath0 . how to download and build @xmath0 are explained in sec . 2.1 , and how to use @xmath0 is explained in sec . we also explain what types of models can be treated by using @xmath0 in sec . 2.3 . in sec . 3 , we detail algorithms implemented in @xmath0 . the representation of the hilbert space adopted in @xmath0 is explained in sec . 3.1 . how to implement multiplication of the hamiltonian to a wavefunction is explained in sec . implementation of full diagonalization method , the lanczos method , and the tpq method is explained in sec . 3.3 , sec . 3.4 . 3.5 , respectively . in sec . 4 , we explain the parallelization of multiplication of the hamiltonian to a wavefunction . how to distribute the wavefunction for each process is explained in sec . 4.1 . in sec . 4.2 , we explain the conventional butterfly algorithm for parallelizing the hamiltonian - wavefunction multiplication . we also explain another parallelization method ( the cma method ) , which is suitable for treating complicated quantum lattices models . in sec . 5 , we show the benchmark results of @xmath0 . in sec . 5.1 , we examine the validity of the tpq method by using @xmath0 . in sec . 5.2 , we show the benchmark results of parallelization efficiency for 18-site hubbard model and 36-site heisenberg model on supercomputers . we also show the benchmark results of the cma method in sec . finally , sec . 6 is devoted to the summary . the gzipped tar file , which contains the source codes , samples , and manuals , can be downloaded from the @xmath0 download site @xcite . for building @xmath0 , a c compiler and the blas / lapack library@xcite are prerequisite . to enable the parallel computations , the message passing interface ( mpi ) library @xcite is also required . for building @xmath0 , the cmake utility @xcite can be used as follows : .... $ cd $ home / build / hphi $ cmake $ pathtohphi $ make .... here , one builds @xmath0 in ` $ home / build / hphi ` and it is assumed that the environment variable , ` $ pathtohphi ` , is set to the path to the source tree path of @xmath0 ( the top directory of @xmath0 ) . if the cmake utility can not find the mpi library on the system , the @xmath0 executable is automatically compiled without the mpi library . in this example , cmake will choose a c compiler automatically . instead , one can specify the compiler explicitly as follows : .... $ cmake -dconfig=$config $ pathtohphi $ make .... where ` $ config ` is chosen from the following configurations : * ` gcc ` : gcc * ` intel ` : intel compiler + mkl library * ` sekirei ` : intel compiler + mkl library on issp system - b ( sekirei ) * ` fujitsu ` : fujitsu compiler + ssl2 library on issp system - c ( maki ) for a system which does not have the cmake utility , we provide another way to generate makefile s using ` hphiconfig.sh ` shell script . one can run ` hphiconfig.sh ` in the @xmath0 top directory as follows : .... $ bash hphiconfig.sh gcc $ make hphi .... once the compilation finishes successfully , one can find the executable , ` hphi ` , in ` src/ ` subdirectory . here , we briefly explain how to use @xmath0 . @xmath0 has two modes ; standard mode and expert mode . the difference between them is the format of input files . for expert mode , one has to prepare the files in the four categories below . + ( 1 ) * parameter files for specifying the model : * in these files , one specifies the transfer integrals , interactions , and types of electronic state ( local spins or itinerant electrons ) . + ( 2 ) * parameter files for specifying the calculation condition : * in these files , one specifies the calculation methods ( full diagonalization , lanczos method , and tpq method ) , the target of models ( spin , kondo , and hubbard models ) , number of sites and electrons , and the convergence criteria . + ( 3 ) * parameter files for correlation functions : * by using these input files , one specifies one - body and two - body equal - time green s functions to be calculated . + ( 4 ) * file for list of above input files : * in this file , one should list the names of all the files that are necessary in the calculations . for typical models in the condensed matter physics , such as the heisenberg model and the hubbard model , one can use @xmath0 in standard mode . in standard mode , one can specify all the input parameters by a single file with a few lines , from which the files described above are generated automatically before starting the calculation . it is thus much more easier to simulate and analyze these models in standard mode as long as the model and the lattice is supported in standard mode . the models supported in standard mode will be explained in the following sections . a typical flow of calculations in @xmath0 is shown as follows : 1 . make input files + an example of input file is shown for standard mode below : + .... l = 4 w = 4 model = " spin " method = " lanczos " lattice = " square lattice " j = 1.0 2sz = 0 .... + here ` l ` and ` w ` are a linear extents of square lattice , in @xmath2 and @xmath3 directions , respectively . ` spin ` means the heisenberg model , and ` j ` is the exchange coupling . by using this input file , one can obtain the ground state of the heisenberg model for @xmath4 sites by performing the lanczos method . details of keywords in the input files can be found in the manuals @xcite . in expert mode , it is necessary to prepare all the files that specify the method , model parameters , and other input parameters . all the names of input files are listed in ` namelist.def ` . details of the input files are also found in the manuals @xcite . 2 . run + after preparing the input files , run a executable ` hphi ` in terminal by setting option ` -s ` " ( or `` ` --standard ` '' ) for standard mode and ` -e ` " for expert mode as follows : * standard mode + ` $ ./hphi -s stdface.def ` * expert mode + ` $ ./hphi -e namelist.def ` 3 . log and results + log files and calculation results are output in the ` output ` directory which is automatically generated in the working directory . for the lanczos and the full diagonalization methods , @xmath0 calculates and outputs the energy , the one - body green s functions , and the two - body green s functions from obtained eigenvectors . for the tpq method , the inverse temperature , the energy and its variance are also obtained at each tpq step . the specified one - body and two - body green s functions are output at specified interval . here , we explain what types of models can be treated in @xmath0 . in expert mode , the target models can be generally defined by the following hamiltonian , @xmath5 where @xmath6 is a generalized transfer integral between site @xmath7 with spin @xmath8 and site @xmath9 with spin @xmath10 and @xmath11 is the generalized two - body interaction which annihilates a spin @xmath10 particle at site @xmath9 and a spin @xmath12 particle at site @xmath13 , and creates a spin @xmath8 particle at site @xmath7 and a spin @xmath14 particle at site @xmath15 . here , @xmath16 ( @xmath17 ) is the creation ( annihilation ) operator of an electron on site @xmath7 with spin @xmath18 or @xmath19 . for the hubbard and kondo models , the 1/2-spin fermions are only allowed . note here that any system of localized spins can be regarded as a special case of the above hamiltonian . therefore , one can use @xmath0 for solving such quantum spin models by straight - forward interpretation of the spin interaction in terms of @xmath20 and @xmath21 in the above expressions . @xmath0 also has a capability of handling spins higher than @xmath22 . in standard mode , @xmath0 can treat the following models . ( the titles of the following items are the corresponding keys for the ` model ` parameter in a standard input file . here , the keys , ` " fermion hubbardgc " ` , ` " spingc " ` and ` " kondogc " ` are the grand - canonical version of ` " fermion hubbard " ` , ` " spin " ` and ` " kondo " ` , respectively . ) \i ) ` " fermion hubbard"/ " fermion hubbardgc " ` + the hamiltonian is given by @xmath23 where @xmath24 is the chemical potential , @xmath25 is the transfer integral between site @xmath7 and site @xmath9 , @xmath26 is the on - site coulomb interaction , @xmath27 is the long - range coulomb interaction between @xmath7 and @xmath9 sites . @xmath28 is the number operator at site @xmath7 with spin @xmath29 and @xmath30 . + + ii ) ` " spin"/ " spingc " ` + the hamiltonian is given by @xmath31 where @xmath32 is the longitudinal magnetic field , @xmath33 is the transverse magnetic field , and @xmath34 is the single - site anisotropy parameter . @xmath35 is the coupling constant between the @xmath36 component of spins at site @xmath7 and site @xmath9 , and @xmath37 is the coupling constant between the @xmath36 component of the spin at site @xmath7 and the @xmath38 component of the spin at site @xmath9 , where @xmath39 . @xmath40 is the @xmath36-axis spin operator at site @xmath7 . + + iii ) ` " kondo"/ " kondogc " ` + the hamiltonian is given by @xmath41 where @xmath24 , @xmath25 , @xmath26 , and @xmath27 are the same as those in the `` ` fermion hubbard ` '' . @xmath42 is the coupling constant between spins of an itinerant electron and a localized one . here , we explain available lattice structures in @xmath0 . in expert mode , arbitrary lattice structures can be treated by specifying connections of sites . in contrast to expert mode , one can easily specify the conventional lattice structures by using standard mode . standard mode supports the chain lattice , the ladder lattice , the square lattice ( examples are shown in insets of figs . [ fig_fdvstpq ] and [ fig_hubbard_speed ] ) , the triangular lattice ( fig . [ fig_triangular ] ) , the honeycomb lattice ( fig . [ fig_latticegp ] ) , and the kagome lattice ( an example of kagome lattice is shown inset of fig . [ fig_kagome ] ) . in standard mode , one can construct the above two - dimensional lattices by specifying four integer parameters , ` a0w ` , ` a0l ` , ` a1w ` , and ` a1l ` and two unit lattice vectors @xmath43 and @xmath44 . by using these parameters and unit vectors , a superlattice spanned by @xmath45 and @xmath46 is specified . we show an example of triangular lattice in fig . [ fig_triangular ] . ( purple dashed arrow ) and @xmath47 ( green dashed arrow ) becomes the supercell to be calculated ( 10 sites ) . red solid arrows ( @xmath43 and @xmath44 ) are unit lattice vectors . , width=151 ] there are some sample input files and reference results in ` samples/ ` directory . for example , ` samples / standard / hubbard / square/ ` is the directory containing files for the computation of the hubbard model on the @xmath48-site square lattice . we can perform the calculation of the ground state by running @xmath0 as standard mode as follows : , .... $ ./hphi -s stdface.def .... while running , @xmath0 dumps information to the standard output . we can check the geometry of the calculated system by using ` lattice.gp ` which is generated by @xmath0 ; it is read by ` gnuplot ` @xcite as follows : .... $ gnuplot lattice.gp .... then , ` gnuplot ` displays the shape of the system and indices of sites ( see fig . [ fig_latticegp ] as an example . this figure is obtained by using ` samples / standard / spin / kitaev / stdface.def ` ) . the calculated results are written in ` output/ ` . in ` output / zvo_energy.dat ` the total energy , the number of doublons ( sites occupied by two electrons with opposite spins ) , and the magnetization along the @xmath49 axis are output . we can compare it with the reference data in ` output_lanczos / zvo_energy.dat ` below ` samples / standard / spin / kitaev/ ` . correlation functions are obtained in ` output / zvo_cisajs.dat ` and ` output / zvo_cisajscktalt.dat ` . when executed in standard mode , @xmath0 creates various input files , such as calcmod.def and modpara.def , that can be used in expert mode . editing those files may be the easiest way of using @xmath0 for models not supported in standard mode . to do so , after making all the necessary modifications , run @xmath0 by the following command : .... $ ./hphi -e namelist.def .... by using expert mode , one can perform more flexible calculations . in @xmath0 , we implement the following three methods ; full diagonalization , exact diagonalization by the lanczos method for ground state calculation , the tpq method for calculation of physical properties at finite temperatures . in this section , we first explain the internal representation of the hilbert space in @xmath0 and how to implement the multiplication of the hamiltonian to a wavefunction in @xmath0 . next , we explain above three methods . to specify a state of a site in @xmath0 , we use an @xmath50-ary , which takes one of @xmath50 different values . for example , in the case of an itinerant electronic system , the local site may take four states , @xmath51 , and @xmath52 , and therefore it is natural to represent it by a 4-ary . in this representation , a basis @xmath53 is labeled by @xmath54_{4}{=[0123]_4}= 0\cdot4 ^ 3 + 1\cdot4 ^ 2 + 2\cdot4 ^ 1 + 3\cdot4 ^ 0 $ ] as @xmath55_4 \rangle$ ] . here , @xmath56_{n}=[a_{m-1}\cdots a_1 a_0]_n$ ] @xmath57 represents the @xmath58-digit @xmath50-ary number , i.e. , @xmath59_{n}=[a_{m-1}\cdots a_1 a_0]_n= \sum_{j=0}^{m-1 } a_{j}\cdot n^{j}.\end{aligned}\ ] ] to use the standard bitwise operations ( _ e.g. _ logical disjunction and exclusive or ) , we decompose the 4-ary representations as the following example : a basis @xmath53 of the 4-site system is indexed by a quartet of @xmath60-digit binary numbers as @xmath61_2,[01]_2,[10]_2,[11]_2\right]_{4 } } \notag \\ & = { [ \overbrace{0,0}^{0},\overbrace{0,1}^{\uparrow},\overbrace{1,0}^{\downarrow},\overbrace{1,1}^{\uparrow\downarrow}]_{2 } } \notag \\ & = 0\cdot2^{7}+0\cdot2 ^ 6 + 0\cdot2 ^ 5 + 1\cdot2 ^ 4 \notag \\ & + 1\cdot2 ^ 3 + 0\cdot2 ^ 2 + 1\cdot2 ^ 1 + 1\cdot2 ^ 0 . \notag\end{aligned}\ ] ] for the kondo system , the local sites are classified into itinerant electron part and localized spin part . however , for simplicity , we use 4-ary representations likewise for the itinerant electronic system , i.e. , we represent a localized spin by @xmath62_2]_{4}\rangle$ ] or @xmath63_2]_{4}\rangle$ ] . for localized spin-@xmath64 systems ( @xmath65 ) , to represent hilbert spaces we use @xmath66-ary representation . for example , in a spin-1 system the wavefunction is represented as @xmath67_{3}\rangle$ ] . to treat the spin-@xmath64 system , we used bogoliubov representation shown in [ bogo ] . in @xmath0 , we can treat the more complicated spin systems such as the spin angular momentums have different values at each site ( mixed spin systems ) . in general , at @xmath7-th site with the spin angular momentum @xmath68 , the hilbert space at @xmath7-th site is represented by @xmath69-ary number and the hilbert space of the system is represented by @xmath70 where @xmath71 and @xmath72 is the whole number of sites . by using the multiplications of the @xmath50-ary number , we represent the mixed spin systems as follows : @xmath73 we note that , when the canonical system is selected by the input file , @xmath0 automatically constructs the restricted hilbert space @xmath74 that has the specified particle number or total @xmath75 . to perform the efficient reverse lookup in the restricted hilbert space , we use the two - dimensional search method @xcite . we also use the algorithm quoted in @xcite as `` finding the next higher number after a given number that has the same number of 1-bits '' to perform an efficient two - dimensional search method . we generate the matrix of @xmath76 by using above - mentioned basis set @xmath77 ( @xmath78 , @xmath79 is the dimension of the hilbert space ) : @xmath80 . by diagonalizing this matrix , we can obtain all the eigenvalues @xmath81 and normalized eigenvectors @xmath82 ( @xmath83 ) . in the diagonalization , we use lapack routine such as ` dsyev ` or ` zheev ` . we also calculate and output the expectation values @xmath84 . these values are used for the finite - temperature calculations . from @xmath84 , we calculate finite - temperature properties by using the ensemble average @xmath85 in the actual calculation , the ensemble averages are performed as a postprocess . here , we explain the main operation in @xmath0 , multiplying the hamiltonian to the wavefunctions . because the dimension of the wavefunction becomes exponentially large by increasing the number of sites , it is impossible to store the whole matrix in the actual calculations . thus , in the lanczos ( sec . 3.4 ) and the tpq ( sec . 3.5 ) methods , we only store the wavefunctions and calculate the matrix elements at each interaction on the fly . we explain the procedure of multiplying the hamiltonian to the wavefunctions by taking the 4-site heisenberg model with @xmath22 on the chain with periodic boundary condition as an example . the hamiltonian is block - diagonalized with blocks being characterized by the total magnetization , @xmath86 . for the block satisfying @xmath87 , the wavefunction can be represented by using the simultaneous eigenvectors of all @xmath88 s as @xmath89 here , the quartet of up or down arrows , which we call a `` configuration '' in what follows , specifies a basis vector . in the actual calculations , we store the list of the configurations . the hamiltonian is defined as @xmath90 where @xmath91 and @xmath92 are defined as follows , @xmath93 to reduce the numerical cost , we store diagonal elements such as @xmath91 for each configuration . in contrast to the diagonal terms , @xmath94 changes the configuration , e.g. , @xmath95 in @xmath0 , if the exchange operation is possible , the operation is performed by bit operations as follows : @xmath96_{2 } \text{\^~ } [ 1010]_{2 } = [ 1001]_{2},\end{aligned}\ ] ] where ^ represents the bitwise exclusive or " . before performing the exchange operator , we count bits in the configuration and judge whether the exchange operation is possible or not . if the exchange operation is not possible , we do not perform the operation and regard the contribution from the configuration as zero , i.e. , @xmath97 we note that the sign arises from the exchange relation between fermion operators in the itinerant electrons systems such as hubbard models , while the sign does not appear in the localized spin systems . to obtain the sign , it is necessary to count the parity of the given bit sequences . we use the algorithm for parity counting that is explained in the literature @xcite . for the sake of completeness , we briefly review the principle of the lanczos method . for more details see , for example , the titpack manual [ 2 ] and the textbook by m. sugihara and k. murota @xcite . the lanczos method is based on the power method . in the power method , by successive operations of the hamiltonian @xmath98 to the arbitrary vector @xmath99 , we generate bases @xmath100 . the obtained linear space @xmath101 is called the krylov subspace . the initial vector is represented by the superposition of the eigenvectors @xmath102 ( corresponding eigenvalues are @xmath81 ) of @xmath98 as @xmath103 we note that all the eigenvalues are real number because hamiltonian is a hermitian operator . by operating @xmath104 to the initial vector , we obtain the relation as @xmath105,\end{aligned}\ ] ] where we assume that @xmath106 has the maximum absolute value of the eigenvalues . this relation shows that the eigenvector of @xmath106 becomes dominant for sufficiently large @xmath50 . we note that the krylov subspace does not change under the constant shift of the hamiltonian @xcite , i.e. , @xmath107 where @xmath21 is the identity matrix and @xmath108 is a constant . by taking constant @xmath108 as the large positive value , the leading part of @xmath109 becomes the maximum eigenvectors and vice versa . thus , by constructing the krylov subspace , we can obtain eigenvalues around both the maximum and the minimum eigenvalues . in the lanczos method , we successively generate the normalized orthogonal basis @xmath110 from the krylov subspace @xmath111 . we define the initial vector and associated components as @xmath112 , @xmath113 . from this initial condition , we can obtain the normalized orthogonal basis as follows : @xmath114 from these definitions , it is obvious that @xmath115 and @xmath116 are real number . in the subspace spanned by these normalized orthogonal basis , the hamiltonian is transformed as @xmath117 here , @xmath118 is the matrix whose column vectors are @xmath119 . we note that @xmath120 ( @xmath121 is an @xmath122 identity matrix ) . @xmath123 is a tridiagonal matrix and its diagonal elements are @xmath124 and subdiagonal elements are @xmath125 . it is known that the eigenvalues of @xmath98 are well approximated by the eigenvalues of @xmath123 for sufficiently large @xmath50 . the original eigenvectors of @xmath98 is obtained by @xmath126 , where @xmath127 are the eigenvectors of @xmath123 . from @xmath118 , we can obtain the eigenvectors of @xmath98 . however , in the actual calculations , it is difficult to keep @xmath118 because its dimension is too large to store [ dimension of @xmath118 = ( dimension of the total hilbert space ) @xmath128 ( number of lanczos iterations ) ] . thus , to obtain the eigenvectors , in the first lanczos calculation , we keep @xmath127 because its dimension is small ( upper bound of the dimensions of @xmath127 is the number of lanczos iterations ) . then , we again perform the same lanczos calculations after we obtain the eigenvalues from the lanczos methods . from this procedure , we obtain the eigenvectors from @xmath118 . in the lanczos method , by successive operations of the hamiltonian on the initial vector , we obtain accurate eigenvalues around the maximum and minimum eigenvalues and associated eigenvectors by using only two vectors whose dimensions are the dimension of the total hilbert space . in @xmath0 , to reduce the numerical cost , we store two additional vectors ; a vector for accumulating the diagonal elements of the hamiltonian , and another vector for the list of the restricted hilbert space explained in sec . [ sec : multiply ] . the dimension of these vectors is that of the hilbert space . as detailed below , to obtain the eigenvector by the lanczos method , one additional vector is necessary . within some number of iterations , we obtain accurate eigenvalues near the maximum and minimum eigenvalues . the necessary number of iterations is small enough compared to the dimensions of the total hilbert space . we note that it is shown that the errors of the maximum and minimum eigenvalues become exponentially small as a function of lanczos iterations ( for details , see ref . details of generating the initial vectors and convergence conditions are shown in [ ape : inilz ] and [ ape : convergencelz ] , respectively . in some cases , the accuracy of obtained eigenvectors by the lanczos method is not enough for calculating the correlation functions . to improve the accuracy of the eigenvectors , we implement the inverse iteration method in @xmath0 , which is also implemented in titpack @xcite . by using the approximate value of the eigenvalues @xmath129 , by successively operating @xmath130 to the initial vector @xmath131 , we can obtain the accurate eigenvector for @xmath129 . the linear simultaneous equation for such procedure is given by @xmath132 by solving this equation by using the conjugate gradient ( cg ) method , we can obtain the eigenvector . we note that we take @xmath131 as the eigenvector obtained by the lanczos method and add small positive number to @xmath129 for stabilizing the cg method . although we can obtain more accurate eigenvector by using this method , additional four vectors are necessary to perform the cg method . the method of the tpq states is based on the fact @xcite that it is possible to calculate the finite - temperature properties from a few wavefunctions ( in the thermodynamic limit , only one wavefunction is necessary ) without the ensemble average . in what follows we describe the prescription proposed in @xcite , which we adopt in @xmath0 . such wavefunctions that replace the ensemble average are called tpq states . because a tpq state can be generated by operating the hamiltonian to the random initial wavefunction , we directly use the routine of the lanczos method to the tpq calculations . here , we explain how to construct the tpq state , which offers a simple way for finite - temperature calculations . let @xmath133 be a random initial vector . how to generate @xmath133 is described in [ ape : initpq ] . by operating @xmath134 ( @xmath13 is a constant , @xmath135 represents the number of sites ) to @xmath133 , we obtain the @xmath15-th tpq states as @xmath136 , we estimate the corresponding inverse temperature @xmath116 as @xmath137 where @xmath138 is the internal energy . arbitrary local physical properties at @xmath116 are also estimated as @xmath139 in a finite - size system , a finite statistical fluctuation is caused by the choice of the initial random vector . to estimate the average value and the error of the physical properties , we perform some independent calculations by changing @xmath133 . usually , we regard the standard deviations of the physical properties as the error bars . in the calculation of the exact diagonalization based on the lanczos method or the finite - temperature calculations based on the tpq state , we have to store the many - body wavefunction on the random access memory ( ram ) . this wavefunction becomes exponentially large with increasing the number of sites . for example , when we compute the 40-site @xmath140 spin system in the @xmath75 unconserved condition , its dimension is @xmath141 and its size in ram becomes about 17.6 tb ( double precision complex number ) . such a large wavefunction can be treated by distributing it to many processes by utilizing distributed memory parallelization . in @xmath0 , the ranks of the processes are labeled by using the leftmost bits in the bit representation of the wavefunction basis . when the leftmost bits corresponding to a @xmath142-site subsystem are chosen to label the ranks of the processes , the number of processes have to be @xmath143 ( for spin systems ) or @xmath144 ( for itinerant electron systems ) . then , each process keeps partial wavefunction of the complementary @xmath145-site subsystem . distribution of wavefunctions of an @xmath146 spin-1/2 system is illustrated in fig . [ fig_parallel ] . we note that the configuration of the @xmath147 sites is fixed in the same process . -site @xmath140 spin system in the @xmath75 unconserved condition . each process has the wavefunction having @xmath148 component ; we can obtain the wavefunction of the entire system by connecting local wavefunction in all processes . , width=340 ] for example , we show the distribution of the wavefunction of the @xmath149-site @xmath140 spin system in the @xmath75 unconserved condition in fig . [ fig_parallel ] ( the number of processes is @xmath150 ) . each process has the wavefunction having @xmath148 components and we can obtain the wavefunction of the entire system by connecting local wavefunctions of all processes . the most time - consuming procedure in a calculation of the lanczos method or the tpq state is the multiplication of the hamiltonian with a vector ( @xmath151 ) , where @xmath152 and @xmath153 are wavefunctions distributed to all the processes . we parallelize this procedure with the aid of message passing interface ( mpi ) @xcite . here , we explain the parallelization procedure by taking a spin exchange term as an example . because the spin exchange occurs between two sites , we use one of the following three procedures according to the type of the two sites ( see fig . [ fig_commun ] ) . when the numbers specifying the both sites are smaller than @xmath154 , we can perform this exchange independently in each process ; there is no inter - process communication [ fig . [ fig_commun](a ) ] . if one of two sites has an index equal to or larger than @xmath154 , first the local wavefunction is communicated between two processes connected with the exchange operator and stored in a buffer . then we compute the remaining exchange operator [ fig . [ fig_commun](b ) ] . we employ this procedure also when both sites are equal to or larger than @xmath154 [ fig . [ fig_commun](c ) ] . in this case , some processes do not participate the calculation ; it causes a load imbalance . however , it is not so serious problem ; the computational time of this case is shorter than those time of the previous two cases because the arrays are accessed sequentially in this case . numerical costs become smaller if all operations can be done similar to fig . [ fig_commun](a ) in which the memory access is limited within each process . in addition , if we can apply simultaneously the all terms in the second quantized hamiltonian acting to the part of wavefunctions continuously stored in the cpu - cache memory , the numerical cost becomes small and independent from the number of the terms . indeed , we can partially convert the conventional algorithm into an efficient algorithm by employing a permutation of the bits . by the permutation , we can also achieve continuous memory access , which is usually faster than random access happening inevitably in the conventional algorithm ( sec . [ parahvprod ] ) . in this section , we explain the newly developed algorithm to realize the continuous memory access and to enhance the computing efficiency . any partial hamiltonians acting on the rightmost @xmath155 bits are represented by a @xmath156 by @xmath156 matrix that acts on @xmath156 successive components , namely , @xmath157 th to @xmath158-th components , of the wavefunction . then , if the @xmath159 ( @xmath160 ) rightmost bits are permuted in the bit representation of the basis as , @xmath161_2\rangle \nonumber\\ & \rightarrow & \|[\sigma_{w-1}\cdots\sigma_{1}\sigma_{0}\sigma_{n_{\rm s}-1}\cdots\sigma_{w+1}\sigma_{w}]_2\rangle,\end{aligned}\ ] ] where @xmath162 ( @xmath163 ) , the partial hamiltonians acting on the @xmath159 th bit , @xmath164 th bit , and so on , are again represented by a matrix that acts on the successive components of the wavefunction . if we find an integer @xmath159 that satisfies @xmath165 and decompose the total hamiltonian into the partial hamiltonians that act on sets of @xmath155 successive bits that overlap each other , the multiplication of the hamiltonian is implemented with continuous memory access by repeating the @xmath159-bits permutation . when the wavefunctions are distributed in many processes , the @xmath159-bits permutation is achieved by the following steps . first , we permute the components in the wavefunction within the rank @xmath166 process , @xmath167 ( @xmath168 ) , with a buffer @xmath169 as @xmath170_2}^ { ( [ { \sigma}_{n_{\rm s}-1 } \cdots { \sigma}_{n_{\rm s}-n_{\rm p } } ] _ 2 ) } \\ & = v _ { [ { \sigma}_{n_{\rm s}-n_{\rm p}-1 } \cdots { \sigma}_{w } { \sigma}_{w-1 } \cdots { \sigma}_{w - n_{\rm p } } { \sigma}_{w - n_{\rm p}-1 } \cdots { \sigma}_{0 } ] _ 2}^ { ( [ { \sigma}_{n_{\rm s}-1 } \cdots { \sigma}_{n_{\rm s}-n_{\rm p } } ] _ 2)}. \end{array}\end{aligned}\ ] ] then , we call an mpi all - to - all routine to transfer the components in the buffer @xmath171 to another buffer @xmath172 in other processes as @xmath173_2}^ { ( [ { \sigma}_{w-1 } \cdots { \sigma}_{w - n_{\rm p}}]_2 ) } \\ & = u _ { [ { \sigma}_{w-1 } \cdots { \sigma}_{w - n_{\rm p } } { \sigma}_{w - n_{\rm p}-1 } \cdots { \sigma}_{0 } { \sigma}_{n_{\rm s}-n_{\rm p}-1 } \cdots { \sigma}_{w}]_2}^ { ( [ { \sigma}_{n_{\rm s}-1 } \cdots { \sigma}_{n_{\rm s}-n_{\rm p } } ] _ 2)}. \end{array}\end{aligned}\ ] ] finally , the following permutation of the components of the wavefunctions completes the @xmath159-bit permutation of the distributed wavefunction : @xmath174_2}^ { ( [ { \sigma}_{w-1 } \cdots { \sigma}_{w - n_{\rm p}}]_2 ) } \\ & = { u ' } _ { [ { \sigma}_{n_{\rm s}-1 } \cdots { \sigma}_{n_{\rm s}-n_{\rm p } } { \sigma}_{w - n_{\rm p}-1 } \cdots { \sigma}_{0 } { \sigma}_{n_{\rm s}-n_{\rm p}-1 } \cdots { \sigma}_{w}]_2}^ { ( [ { \sigma}_{w-1 } \cdots { \sigma}_{w - n_{\rm p}}]_2)}. \end{array}\end{aligned}\ ] ] details of the continuous access will be reported elsewhere @xcite . the cma method is implemented in standard mode for @xmath22 spins without total @xmath75 conservation in the present version of @xmath0 . ) , the algorithm has not been implemented for other systems yet . ] in this section , we show benchmark results of @xmath0 . first , we examine the accuracy of the finite - temperature calculations based on the tpq algorithm @xcite by comparing the results with the exact ensemble averages calculated by the full diagonalization . for small system sizes , we can easily perform the same calculation on our own pcs . second , we show the parallelization efficiency of @xmath0 for large system sizes on supercomputers . we have carried out tpq simulations of two typical models , namely , an 18-site hubbard model on a square lattice and a 36-site heisenberg model on a kagome lattice , with changing numbers of threads and cpu cores , where the number of cpu cores equals the number of threads times the number of mpi processes . we also show performance of the cma method . to examine the validity of the tpq calculation , we compute the temperature dependence of the doublon density of the hubbard model ( @xmath175 ) for an 8-site cluster with the periodic boundary condition . the shape of the cluster is illustrated in the inset of fig . [ fig_fdvstpq ] . the hubbard model is defined by setting @xmath176 , @xmath177 , @xmath178 for the nearest - neighbor pairs of sites , @xmath179 , and @xmath180 for further neighbor pairs of sites . an example of the input file for the 8-site hubbard model used in this calculation is shown as follows : .... a0w = 2 a0l = 2 a1w = -2 a1l = 2 model = " fermion hubbard " method = " tpq " lattice = " square lattice " t = 1.0 u = 8.0 .... in fig . [ fig_fdvstpq ] , we show the temperature dependence of the doublon density , @xmath181 , calculated by the tpq state and by the canonical ensemble . we confirm that the tpq calculations well reproduce the results of the canonical ensemble average . in addition to the finite - temperature calculations , we perform the ground state calculations by the lanczos method and the calculated doublon density at zero temperature is plotted by the dashed line . we also confirm that doublon density at low temperature well converges to the value of the ground state . all the results show the validity of the tpq method . here , we carry out tpq simulations for large system sizes with changing numbers of threads and processes to examine parallelization efficiency of @xmath0 . we choose two typical models ; a half - filled 18-site hubbard model on a square lattice and a 36-site heisenberg model on a kagome lattice . the speedup of the simulation for the 18-site hubbard model with up to 3,072 cores is measured on sgi ice xa ( sekirei ) at issp ( table [ tbl_spec ] ) . the speedup of the large - scale simulation for the 36-site heisenberg model with up to 32,768 cores is examined by using the k computer at riken aics ( table [ tbl_spec ] ) . [ cols="^,^,^",options="header " , ] we perform the tpq simulations for the half - filled 18-site hubbard model on the square lattice illustrated in the inset of fig . [ fig_hubbard_speed ] . here , we employ the subspace of the hilbert space that satisfies @xmath182 . then , the dimension of the subspace is @xmath183 , where @xmath184 represents the binomial coefficient . the input file used in this calculations is shown below : .... a0w = 3 a0l = 3 a1w = -3 a1l = 3 model = " fermion hubbard " method = " tpq " lattice = " square lattice " t = 1.0 u = 8.0 nelec = 18 2sz = 0 .... in fig . [ fig_hubbard_speed ] , we can see significant acceleration caused by the increase of cpu cores . this acceleration is almost linear up to 192 cores , and is weakened as we further increase the number of cores . this weakening of the acceleration comes from the load imbalance in the @xmath1 conserved simulation . for example , when we use 3 openmp threads and 1024 mpi processes , the number of sites associated with the ranks of processes ( @xmath147 in sec . [ sec_dist_wave ] ) becomes 5 , and the number of sites in the subsystem in each process ( @xmath154 in sec . [ sec_dist_wave ] ) becomes 13 . each process has different numbers of up- and down- spin electrons . in the process which has the largest hilbert space , there are six up - spin electrons and six down - spin electrons , and the hilbert space becomes @xmath185 . on the other hand , the process which has the smallest hilbert space has nine up - spin electrons , nine down - spin electrons and the @xmath186 dimensional hilbert space . this large difference of the hilbert - space dimension between each process causes a load imbalance . this effect becomes large as we increase the number of processes . when we fix the total number of cores ( for examples 1536 cores ) , we can find that the process - major computation ( 256 processes with 6 threads ) is faster than the thread - major computation ( 64 processes with 24 threads ) . since the numerical performance of both of them are equivalent , this behavior seems strange at first sight . the origin of difference might be explained as follows : when we double the number of processes , the amount of data communicated by each process at once becomes a half of the original size , and the communication time decreases . on the other hand the communicated data - amount and the communication time is unchanged even if we double the number of threads . in other words , the algorithm described in sec . [ parahvprod ] parallelizes calculation and also communication . ( @xmath187 ) and total @xmath75 conservation ( @xmath182 ) as functions of the total number of cores . we note that , due to changes in size of data transfer between mpi process , the thread number affects the tpq steps per hour even when the total number of cores is fixed . the squares , circles , upward triangles , and downward triangles represent the results with 3 threads , 6 threads , 12 threads , and 24 threads , respectively . the inset shows the shape of the 18-site cluster used in this benchmark calculations.,width=264 ] we perform the tpq calculations for the 36-site heisenberg model on the kagome lattice with hybrid parallelization by using up to 32,768 cores . here , we employ the @xmath188-dimensional ( @xmath189 ) hilbert space without the @xmath1 conservation . although the canonical ( @xmath1 conserved ) calculation is faster than the grand - canonical ( @xmath1 unconserved ) calculation , we sometimes employ the latter for the tpq calculation . the finite - size effect is much smaller in the grand - canonical tpq state than in the canonical tpq state @xcite . figure [ fig_kagome ] shows the speedup of this calculation on the k computer . in contrast to the canonical calculation in the previous section , we obtain the almost linear scaling even we use the very large number of cores ( more than ten thousands cores ) . in the grand - canonical calculation , each process has a hilbert space of the same size . therefore , there is no load imbalance that appears prominently in the canonical calculation . due to the high throughput of the inter - node data transfer on the k computer , the parallelization efficiency from 4,096 cores to 32,768 cores reaches 82% and change in number of mpi processes does not largely affect the speed . except for some cases ( open symbols in fig . [ fig_kagome ] ) , in the group of the same number of cores ( for example , 16384 cores ) , we can not find any significant difference between the performance of the thread - major computation ( 2048 processes with 8 threads ) and that of the process - major computation ( 16384 processes with 1 thread ) . on the other hand when we calculate the same system by using 3,072 cores on sekirei , the speed of the computation by using 128 processes with 24 threads and that speed by using 1024 processes with 3 threads become 29 steps / hour and 71 steps / hour , respectively ; we can find the advantage of the process - major computation on sekirei as in the previous section . it is thought that because the communication time on sekirei is longer than that on the k computer , the parallelization of the communication by increasing the number of processes is more effective on sekirei than on the k computer . for some cases the speed becomes exceptionally slow ( corresponding conditions are shown as open symbols ) . this slowdown occurs when the number of processes is 8,192 , and it remains even if we change the lattice to the one - dimensional chain . we consider the reason of the slowdown as follows . there are three effects on the computational speed by increasing processes : first , the numerical cost per single process and the data size per single communication decrease in inversely proportional to the number of processes ( positive effect ) , second , the frequency of communication increases ( negative effect ) , and , finally , the load balancing becomes worse ( negative effect ) . therefore we consider the slowdown comes from the competition of these effects . conservation as functions of the total number of cores . the squares , circles , upward triangles , and downward triangles represent the results with 1 thread , 2 threads , 4 threads , and 8 threads , respectively . the inset shows the shape of the 36-site cluster used in this benchmark calculations.,width=264 ] in this section , we show benchmark results of the continuous - memory - access parallelization algorithm , namely , the cma method in @xmath0 , which is briefly explained in sec . [ ca ] . the cma method is particularly advantageous when the hamiltonian does not possess the u(1 ) symmetry or the su(2 ) symmetry that make the total charge and the total magnetization good quantum numbers . a typical example is the kitaev model discussed below . for such hamiltonians , we can not use the canonical mode . especially , the cma method is much faster than the conventional parallelization scheme explained in sec.[parahvprod ] when inter - site interactions are dense and complicated . this is because the cost for the computation and memory - transfer for the bit permutation does not depend on the number of terms or the complexity of the hamiltonian as long as the range of interaction is not too long , while , in the conventional parallelization scheme , the cost for the hamiltonian operations increases as it has more terms . to compare performance of the cma method and the conventional parallelization scheme , here , we carried out tpq simulations of a spin hamiltonian with complicated spin - spin interactions . we show the benchmark results of the tpq simulation for a spin hamiltonians of iridium oxide , na@xmath190iro@xmath191 , which was derived by utilizing _ ab initio _ electronic band structures and many - body perturbation theory @xcite . the hamiltonian is defined on a honeycomb structure and has complicated inter - site spin - spin interactions that can be classified into five parts as follows : @xmath192 where the first term @xmath193 represents the kitaev hamiltonian @xcite , the second term @xmath194 describes diagonal exchange couplings between nearest - neighbor ( n.n . ) spins , and the third term @xmath195 describes complicated off - diagonal spin - spin interactions . the fourth term @xmath196 and the fifth term @xmath197 describe the second neighbor ( 2nd n. ) interactions and the third neighbor ( 3rd n. ) interactions , respectively . details of the hamiltonians are given in the previous study @xcite . in the lower panel of fig . [ fig_na2iro3 ] , elapsed time per tpq step for the conventional and the cma method is shown for a kitaev hamiltonian , @xmath193 , a kitaev - heisenberg hamiltonian , @xmath198 , a n.n . hamiltonian , @xmath199 , a hamiltonian including 2nd n. interactions , @xmath200 , and the _ ab initio _ hamiltonian @xmath201 . here , we use 16 nodes ( 24 cores per node ) on sekirei at issp the tpq simulation with 3 threads and 128 processes . even when the number of the coupling term increases , from the sparse kitaev hamiltonian @xmath193 to the dense and complicated hamiltonian of na@xmath190iro@xmath191 , @xmath202 , the elapsed time per tpq step of the cma method does not show significant increase while the elapsed time of the conventional scheme shows significant increase . although , for the sparse kitaev hamiltonian @xmath193 , the conventional scheme shows slightly better performance than that of the cma method , the cma method is three times faster than the conventional one even for the n.n . hamiltonian @xmath203 . furthermore , even though , from @xmath203 to @xmath200 , the number of the bonds clearly increases , the elapsed time per tpq step of the cma method only shows slight increase . ( upper panel ) and elapsed time per tpq step for the conventional butterfly and the cma method ( lower panel ) . the three different nearest - neighbor ( n.n . ) bonds , @xmath204 , @xmath205 , and @xmath206 , the second neighbor ( 2nd n. ) bonds @xmath207 , and the third neighbor ( 3rd n. ) bonds @xmath208 are illustrated in the upper panel . in the lower panel , the ( cyan ) squares denote the elapsed time per tpq step of the conventional scheme and the ( red ) circles denote the elapsed time per tpq step of the cma method.,width=264 ] to summarize , we explain the basic usage of @xmath0 in sec . 2 . by preparing one file within ten lines , for typical models in the condensed matter physics , one can easily perform the exact diagonalization based on the lanczos method or finite - temperature calculations based on the tpq method . by using the same file , one can also perform the large - scale calculations on modern supercomputers . by properly editing the intermediate input files generated by standard mode execution , it is also easy to perform the similar calculations for complicated hamiltonians that describe the low - energy physics of real materials . this flexibility and user - friendly interface are the main advantage of @xmath0 and we expect that @xmath0 will be a useful tool for broad spectrum of scientists including experimentalists and computer scientists . in sec . 3 , we explain the basic algorithms implemented in @xmath0 . although the lanczos method and the tpq method are detailed in the literature , to make our paper self - contained , we explain the essence of these methods . furthermore , in sec . 4 , we detail the parallelization methods of the multiplication of hamiltonian to the wavefunctions ; one is the conventional butterfly method and another one is the newly developed method . we also show the benchmark results of @xmath0 on supercomputers in sec . the results indicate that parallelization of @xmath0 works well on these supercomputers and the computational time is drastically reduced . we also introduce the new algorithm for parallelization , i.e. , the ( cma ) method . the main advantage of this method is that the speed does not depend on the number of terms in the hamiltonians , while the computational cost of the conventional butterfly method is proportional to the number of terms in the hamiltonians . thus , the cma method is suitable for treating the complicated hamiltonians for real materials . actually , for the low - energy model of na@xmath209iro@xmath210 , we show that the speed of the cma method is about six times faster than that of the conventional method . so far , the cma method is applicable to the limited class of hamiltonians and lattice geometries . in addition , its efficiency largely depends on the models . it is a remaining challenge to remove the limitation from the cma method . recent development of the theoretical tools for obtaining the low - energy models of real materials enable us directly compare the theoretical results and experimental results . although the limitation of the system size is severe , the exact analyses based on the lanczos method or the tpq method offer a firm basis for examining the validity of the theoretical results . the present version of @xmath0 can calculate only the static physical quantities , which is difficult to be directly measured in experiments . extending @xmath0 to calculate dynamical properties such as the dynamical spin / charge structure factors and the optical conductivity is a promising way to make @xmath0 more useful . such an extension will be reported in the near future . we would like to express our sincere gratitude to prof . hidetoshi nishimori and mr . daisuke tahara . implementation of the lanczos algorithm in @xmath0 written in c is based on the pioneering diagonalization package titpack ver . 2 written in fortran by prof . nishimori . for developing the user interface of @xmath0 , we follow the design concept of the user interface in the program for variational monte carlo method developed by mr . tahara . a part of the user interface in @xmath0 is based on his original codes . we would also like to thank the support from _ project for advancement of software usability in materials science _ " by the institute for solid state physics , the university of tokyo , for development of @xmath0 ver.1.0 . we thank the computational resources of the k computer provided by the riken advanced institute for computational science through the general trial use project ( hp160242 ) . this work was also supported by grant - in - aid for scientific research ( 15k17702 , 16h06345 , and 16k17746 ) and building of consortia for the development of human resources in science and technology from the mext of japan and supported by presto , jst . we also thank numerical resources from the supercomputer center of the institute for solid state physics , the university of tokyo . in the spin system , the spin indices in input files of ` transfer ` , ` interall ` , and correlation functions are specified by using the bogoliubov representation . spin operators are written by using creation / annihilation operators as follows : @xmath211 in the lanczos method , an initial vector is specified with ` initial_iv`(@xmath212 ) defined in an input file for standard mode or a modpara file for expert mode . a type of an initial vector can be selected from a real number or complex number by using ` initialvectype ` in a modpara file . * for canonical ensemble and ` initial_iv ` @xmath213 + a non - zero component of a target of hilbert space is given by @xmath214 where @xmath215 is a total number of the hilbert space and @xmath216 is added to avoid selecting the special configuration for a default value ` initial_iv ` @xmath217 . when a type of an initial vector is selected as a real number , a coefficient value is given by @xmath218 , while as a complex number , the value is given by @xmath219 . * for grand canonical ensemble or ` initial_iv ` @xmath220 + an initial vector is given by using a random generator , i.e. coefficients of all components for the initial vector is given by random numbers . the seed is calculated as @xmath221 where @xmath222 is a number given by an input file ( parameter ` initial_iv ` ) , @xmath223 is the thread i d , @xmath224 is the number of threads , and @xmath225 is the process i d . therefore the initial vector depends both on ` initial_iv ` and the number of processes . random numbers are generated by using simd - oriented fast mersenne twister ( dsfmt ) @xcite . in @xmath0 , we use ` dsyev ` ( routine of lapack ) for diagonalization of @xmath123 . we use the energy of the first excited state of @xmath123 as a criteria of convergence . in the standard setting , after five lanczos step , we diagonalize @xmath123 every two lanczos step . if the energy of the first excited states agrees with the previous energy within the required accuracy , the lanczos iteration finishes . the accuracy of the convergence can be specified by ` lanczoseps ` ( modpara file in expert mode ) . after obtaining the eigenvalues , we again perform the lanczos iteration to obtain the eigenvector . from the eigenvectors @xmath226 , we calculate energy @xmath227 and variance @xmath228 . if @xmath129 agrees with the eigenvalues obtained by the lanczos iteration and @xmath229 is smaller than the specified value , we finish the diagonalization . if the accuracy of lanczos method is not enough , we perform the cg method to obtain the eigenvector . as an initial vector of the cg method , we use the eigenvectors obtained by the lanczos method in the standard setting . this often accelerates the convergence . for tpq method , an initial vector is given by using a random number generator , i.e. coefficients of all components for the initial vector are given by random numbers . the seed is calculated as @xmath230 where @xmath222 , @xmath223 , @xmath224 , and @xmath225 are the same as those for the lanczos method [ eqn . ( [ fml_initialguess_lanczos ] ) ] . @xmath231 indicates the counter of trials , which is equal to or less than the total number of trials ` numave ` in an input file for standard mode or a modpara file for expert mode . we can select a type of initial vector from a real number or complex number by using ` initialvectype ` in a modpara file . @xmath232 indicate the thread i d , the number of threads , the process i d , respectively ; the initial vector depends both on ` initial_iv ` and the number of processes . b. bauer , l. d. carr , h. g. evertz , a. feiguin , j. freire , s. fuchs , l. gamper , j. gukelberger , e. gull , s. guertler , a. hehn , r. igarashi , s. v. isakov , d. koop , p. n. ma , p. mates , h. matsuo , o. parcollet , g. pawwski , j. d. picon , l. pollet , e. santos , v. w. scarola , u. schollwck , c. silva , b. surer , s. todo , s. trebst , m. troyer , m. l. wall , p. werner , s. wessel , journal of statistical mechanics : theory and experiment 2011 ( 05 ) ( 2011 ) p05001 . http://stacks.iop.org/1742-5468/2011/i=05/a=p05001[[link ] ] . j. jakli , p. prelov ek , phys . b 49 ( 1994 ) 50655068 . http://dx.doi.org/10.1103/physrevb.49.5065 [ ] , http://link.aps.org/doi/10.1103/physrevb.49.5065[[link ] ] . http://link.aps.org/doi/10.1103/physrevb.49.5065 a. hams , h. de raedt , phys . e 62 ( 2000 ) 43654377 . http://dx.doi.org/10.1103/physreve.62.4365 [ ] , http://link.aps.org/doi/10.1103/physreve.62.4365[[link ] ] . http://link.aps.org/doi/10.1103/physreve.62.4365 e. anderson , z. bai , c. bischof , l. blackford , j. demmel , j. dongarra , j. du croz , a. greenbaum , s. hammarling , a. mckenney , d. sorensen , lapack users guide , 3rd edition , society for industrial and applied mathematics , 1999 .	@xmath0 [ _ aitch - phi _ ] is a program package based on the lanczos - type eigenvalue solution applicable to a broad range of quantum lattice models , i.e. , arbitrary quantum lattice models with two - body interactions , including the heisenberg model , the kitaev model , the hubbard model and the kondo - lattice model . while it works well on pcs and pc - clusters , @xmath0 also runs efficiently on massively parallel computers , which considerably extends the tractable range of the system size . in addition , unlike most existing packages , @xmath0 supports finite - temperature calculations through the method of thermal pure quantum ( tpq ) states . in this paper , we explain theoretical background and user - interface of @xmath0 . we also show the benchmark results of @xmath0 on supercomputers such as the k computer at riken advanced institute for computational science ( aics ) and sgi ice xa ( sekirei ) at the institute for the solid state physics ( issp ) . 02.60.dc numerical linear algebra,71.10.fd lattice fermion models,75.10.kt quantum spin liquids * program summary * _ manuscript title : _ quantum lattice model solver @xmath0 + _ authors : _ mitsuaki kawamura , kazuyoshi yoshimi , takahiro misawa , youhei yamaji , synge todo , naoki kawashima + _ program title : _ @xmath0 + _ journal reference : _ + _ catalogue identifier : _ + _ program summary url : _ + http://ma.cms-initiative.jp/en/application-list/hphi + _ licensing provisions : _ gnu general public license , version 3 or later + _ programming language : _ c + _ computer : _ any architecture with suitable compilers including pcs and clusters . + _ operating system : _ unix , linux , os x. + _ ram : _ problem dependent . for example , less than one gb for a few - site system , and 3.8 tb for the 36-site heisenberg model computed in this study . + _ number of processors used : _ problem dependent . we use up to 4,096 processors ( 32,768 cores ) in this study . + _ keywords : _ 02.60.dc numerical linear algebra , 71.10.fd lattice fermion models , 75.10.kt quantum spin liquids + _ classification : _ 4.8 linear equations and matrices , 7.3 electronic structure + _ external routines / libraries : _ mpi , blas , lapack + _ nature of problem : _ + physical properties ( such as the magnetic moment , the specific heat , the spin susceptibility ) of strongly correlated electrons at zero- and finite temperature . + _ solution method : _ + application software based on the full diagonalization method , the exact diagonalization using the lanczos method , and the microcanonical thermal pure quantum state for quantum lattice model such as the hubbard model , the heisenberg model and the kondo model . + _ restrictions : _ + system size less than about 20 sites for a itinerant electronic system and 40 site for a local spin system . + _ unusual features : _ + finite - temperature calculation of the strongly correlated electronic system based on the iterative scheme to construct the thermal pure quantum state . this method is efficient for highly frustrated system which is difficult to treat with other methods such as the unbiased quantum monte carlo . + _ running time : _ + problem dependent . for example , when we compute the heisenberg model on the kagome lattice without @xmath1 conservation , we can perform 400 iterations per hour .
one of the main topic in cosmology is the complicated fact that our universe have an accelerated expansion @xcite . in order to interpret this expansion , some various candidates have been proposed by many authors , in particular , cosmological constant idea , dark energy models and modified gravities . also , transition from linear einstein - hilbert action to nonlinear modification of it , is expected when string / m - theory corrections are considered @xcite . amongst the nonlinear modifications of einstein gravity , the so - called @xmath0 gravity @xcite , whose action is a nonlinear function of the curvature scalar @xmath4 , is completely special . it seems to provide an interpretation to dark energy , the hierarchy problem cognola2006 , the four cosmological phases @xcite , the power law early - time inflation @xcite , late - time cosmic accelerated expansion @xcite , singularity problem arising in the strong gravity regime abdalla , briscese , noj046006,bamba , kobayashi2008 , rotation curves of spiral galaxies @xcite and detection of gravitational waves gwave . in additions , we can give other motivations to consider @xmath0 gravity . first of all , in spite of @xmath0 theory is the simplest modification of the gravitational interaction to higher order known so far , @xmath0 action is sufficiently general to encapsulate some of the basic characteristics of higher order gravity . second , it is believed that there are sufficient theoretical predictions to claim that @xmath0 gravity can be compatible with newtonian and post - newtonian prescriptions capozziello2005a , capozziello2005b . third , there are serious reasons to believe that @xmath0 theories are unique among higher order gravity theories , in the sense that they seem to be the only ones which can avoid the long known and fatal ostrogradski instability @xcite . fourth , @xmath0 theories have no ghosts ( @xmath5 ) , and the stability condition @xmath6 of essentially amounts to guarantee that the scalaron is not a tachyon @xcite . all these properties can be easily obtained either directly , or using conformal equivalence of field equations of @xmath0 theories to those of the einstein gravity interacting with a minimally coupled scalar field with some potential @xmath7 which form is uniquely determined by @xmath0 in all points where @xmath8 @xcite . some of these reasons are sufficient to state why there is much activity in the study of different versions of modified @xmath0 theories with applications to gravity , cosmology and astrophysics . unlike general relativity , which involves metric derivatives no higher than second order , @xmath9 gravity involves also third and fourth order derivatives , which caused complications in the calculation . some of the exact solutions in @xmath0 gravity have been studied in ref . most of this solutions are schwarzschild with or without of cosmological constant . in refs . exactsolrn , dombriz , moon , it has shown that one may obtain reissner - nordstrm solutions in @xmath0-maxwell gravity . when we follow the einstein s thought and desires , we may find that one of the weighty dream of einstein was creating a geometrical unified theory of physics . however the einstein dream is still very much alive , but till now , all attempts of geometrizing physics are unfeasible . in what follows , we consider one of the modified theory of gravity ( @xmath0 gravity ) instead of einstein general relativity , and show that some solutions of @xmath0 gravity ( pure geometry ) are equivalent to the einstein - nonlinear maxwell gravity in the present of cosmological constant . it is notable that ( a)ds solutions of @xmath10 gravity has been investigated before @xcite , but there is not any charged ( a)ds solution of pure @xmath10 gravity , yet . finally , we should notice as an example of charged solutions in pure @xmath0 gravity , we refer to the particular form @xmath11 @xcite ( it is not in the form of @xmath10 ) , which attains an electromagnetic - like curvature source , so that @xmath12 can be interpreted as an electric charge `` without charge '' . in other word , @xmath13 behaves geometrically as if we have @xmath14 ( electrostatic field ) . remarkably , @xmath15 plays the role of `` charge '' so that the geometry @xmath13 becomes equivalent to the reissner - nordstrom geometry in a spherically symmetry spacetime . beside the case of @xmath14 ( electrostatic field ) , and more aptly , cases such as @xmath14 ( non - minimal scalar field ) cases also have been investigated @xcite . the action of @xmath16-dimensional @xmath0 gravity , in the presence of matter field has the form of @xmath17 , \label{action}\]]where @xmath0 is an arbitrary function of ricci scalar @xmath4 , and @xmath18 is the lagrangian of matter fields . variation with respect to metric @xmath19 , leads to the following field equations @xmath20where @xmath21 is the ricci tensor , @xmath22 and @xmath23 is the standard matter stress - energy tensor in which derived from the matter lagrangian @xmath24 in the action ( [ action ] ) . the trace of eq . ( [ fe ] ) reduces to @xmath25 f_{r}-\frac{d}{2}f(r)=\mathrm{t } , \label{trace}\]]where @xmath26 is the trace of matter stress - energy tensor . it is notable that eq . ( [ trace ] ) leads to @xmath27 , for sourceless einstein gravity ( @xmath28 , @xmath29 and @xmath30 ) . here , we want to obtain the static solutions of eq . ( [ fe ] ) with positive , negative and zero curvature horizons . for this purpose , we assume that the metric has the following form @xmath31where @xmath32considering the field equation ( [ fe ] ) with the metric ( [ met1 ] ) , one can obtain the following independent sourceless ( @xmath33 ) field equations ; @xmath34 f_{r}^{\prime } -\left [ rg^{\prime \prime } ( r)+(d-2)g^{\prime } ( r)\right ] f_{r}=rf(r ) , \label{fieldtt } \\ & & \left [ rg^{\prime } ( r)+2(d-2)g(r)\right ] f_{r}^{\prime } -\left [ rg^{\prime \prime } ( r)+(d-2)g^{\prime } ( r)\right ] f_{r}=rf(r ) , \label{fieldrr } \\ & & 2rg(r)f_{r}^{\prime \prime } + 2\left [ rg^{\prime } ( r)+(d-3)g(r)\right ] f_{r}^{\prime } -\frac{2(d-3)}{r}\left ( \frac{rg^{\prime } ( r)}{d-3}% + g(r)-k\right ) f_{r}=rf(r ) , \label{fieldthth}\end{aligned}\]]corresponding to @xmath35 , @xmath36 and @xmath37 ( @xmath38 ) components of eq . ( [ fe ] ) , respectively . it is notable that the prime and double prime are the first and second derivatives with respect to @xmath39 , respectively . in this paper , we study black hole solutions with constant ricci scalar ( so @xmath40 ) , and therefore it is easy to show that the field equations ( [ fieldtt])-([fieldthth ] ) reduce to @xmath41equating the left hand sides of eqs . ( [ eqaa ] ) and ( [ eqbb ] ) , we obtain @xmath42with the trivial topological schwarzschild solution @xmath43 , where @xmath44 . it is important to note that we are looking for a solution which satisfy both eqs . ( eqaa ) and ( [ eqbb ] ) simultaneously and the mentioned trivial topological schwarzschild solution is not the solution of them for arbitrary @xmath0 . inserting the mentioned topological schwarzschild solution ( with nonzero @xmath4 ) in eqs . ( [ eqaa ] ) and ( [ eqbb ] ) , and solve new equations , directly , one can find that the topological schwarzschild solution in eqs . ( [ eqaa ] ) and ( [ eqbb ] ) leads to exponential form of @xmath0 gravity ( @xmath45 ) , which is not in the form of @xmath46 . it is notable that in special case , @xmath29 ( with constant @xmath4 ) , we encounter with sourceless einstein gravity in the absence of cosmological constant which leads to @xmath27 ( see eq . ( [ trace ] ) for more detail ) . therefore it is straightforward to show that for @xmath29 and so @xmath47 , eqs . ( [ eqaa ] ) and ( [ eqbb ] ) reduce to@xmath48 which their general solution is @xmath49 . in this paper , we consider some various class of @xmath10 gravity and surprisingly , we find that for special cases of @xmath0 gravity we can obtain charged black holes in addition to schwarzschild solutions . it is notable that , in general , charged solution could not satisfy eqs . ( [ eqaa ] ) and ( [ eqbb ] ) and we should set some free parameters in @xmath0 models to satisfaction . here , we want to obtain the static solutions of eqs . ( [ fe ] ) without any matter field ( @xmath50 ) . we assume that the metric has the same form of eq . ( [ met1 ] ) . here , to find the function @xmath51 , one may use any components of eq . ( [ fe ] ) with defined function @xmath0 . in what follows , we consider the some special cases of @xmath0 gravity and investigate their properties . one of the initiative @xmath0 models supposed to explaining the positive acceleration of expanding universe has @xmath0 action as @xmath53 @xcite . in this model , for large values of ricci scalar , @xmath0 function tends to @xmath29 , so we expect for these values of @xmath54 the modification become negligible but for small values of ricci scalar , @xmath9 action is not the linear one thus for this values of ricci scalar , gravity is modified . looking at equations ( [ met1 ] ) and ( [ fe ] ) , one could obviously obtain the metric functions as follows @xmath55 where we should set @xmath56 . it is easy to show that in this model the ricci scalar and the kretschmann scalar are @xmath57 and @xmath58 , respectively , and so we encounter with a curvature singularity at @xmath2 . in other word , this solution corresponds to the topological schwarzschild - ads(ds ) black hole solution , when we set @xmath59 , @xmath60 ( @xmath61 , @xmath62 ) . in summary , we start with pure @xmath0 gravity in the absence of cosmological constant , but this solution ( with @xmath63 ) is matched to the einstein gravity in the presence of cosmological constant , exactly . so , one may think that the cosmological constant could emerge from @xmath0 gravity . it is notable that in this model we can not obtain charged solution . because for large values of @xmath4 the modification is negligible , this model can not explain the inflation but there is several viable models unifying inflation and late time acceleration . also , in this model and for positive cosmological constant ( ds solutions ) , @xmath64 , and so one can deduce the presented ds model suffer the dolgov - kawasaki instability @xcite . it is notable that one can remove instability of ds solutions by adding @xmath65 term in the mentioned model @xcite . in addition , this model is not consistent with solar system tests , exactly @xcite . investigation of @xmath0 gravity models show that there are some known examples of viable @xmath0 gravity that exhibit no problems in the weak gravity regime @xcite . but for the strong gravity regime , these modes have a serious drawback such as singularity problem . in order to resolve the singularity problem arising in the strong gravity regime , kobayashi and maeda @xcite have considered a higher curvature correction proportional to @xmath67 where @xmath68 . the field equations , given in eq . ( [ fe ] ) with the metric ( [ met1 ] ) , in this case simply provide @xmath69 here , we should mention that in this model we can not obtain charged solution . straightforward calculations show that our information about the ricci and the kretschmann scalars are @xmath70 these information help us to conclude that there is a curvature singularity at @xmath2 , and this solution is correspond with the topological schwarzschild black hole in the presence of cosmological constant . in other word , one can extract the cosmological constant from @xmath71 gravity by suitable coefficient @xmath72 . for @xmath73 , the solution reduces to topological schwarzschild black hole for vanishing @xmath74 . it is simple to calculate @xmath75 @xmath76 and therefore one can find that the ads ( ds ) solutions follow dolgov - kawasaki stability for @xmath77 ( @xmath78 ) . it is notable that viable modifications in gravity should pass all sorts of tests from the large scale structure of the universe to galaxy and cluster dynamics to the solar system tests . one of the outstanding question in @xmath0 gravity is whether this theory is consistent with solar system tests or not . when the correction term of einstein gravity is exponential form ( see for e.g. @xcite ) , one can prove that in this model , there is no conflict with solar system tests and also it satisfy high curvature condition @xcite . in addition , the solutions of this model is virtually indistinguishable from that in general relativity , up to a change in the newton s constant @xcite . considering the presented @xmath0 model , with eqs . ( [ met1 ] ) and ( [ fe ] ) , we clearly deduce at least two type solutions : as one can check easily , the first solution is corresponding to the schwarzschild ( a)ds in the following form @xmath80when we set @xmath81 with ricci scalar @xmath4 which is calculated in eq . ( [ rii ] ) and free parameter @xmath82 . second solution is the same as topological charged solution @xmath83 where we should set some parameters . considering charged solution , eq . ( giiib ) with @xmath84 , one can show that eqs . ( fieldtt)-([fieldthth ] ) reduce to @xmath85 r^{d}=0 . \label{factorsiii}\ ] ] in order to vanish eq . ( [ factorsiii ] ) we should consider the following equations @xmath86 with the solutions @xmath87 and @xmath88 . in other words , setting @xmath87 and @xmath88 , one can show that the charged solution ( [ giiib ] ) satisfies field equations of the mentioned @xmath0 gravity . we should note that the presented uncharged black holes ( [ giiia ] ) is the same as solutions which obtained in ref . @xcite for @xmath89 . although for @xmath90 in ref . @xcite , the metric function has a charged term @xmath91 in analogy with our charged solution ( @xmath92 ) , but these solutions are completely different . they have different geometry and constant curvature , @xmath4 , with various asymptotic behavior . in addition , it is notable that the metric function @xmath51 , presented here , differ from the standard higher - dimensional reissner nordstrm solutions since the electric charge term in the metric function ( [ giiib ] ) is proportional to @xmath93 but in the standard higher - dimensional charged black hole solutions is proportional to @xmath94 . in order to interpret the charge term , one may think about the scalar - tensor representation of @xmath0 gravity . in order to find a scalar - tensor representation of @xmath0 theories , one can use conformal transformations @xcite . considering the conformal transformations on the metric function , we find that there no new interpretation about new metric function unless its scale . on other hand , comparing the uncharged solution of @xmath10 gravity ( eq . ( [ giiia ] ) ) with the solution of einstein gravity in the presence of cosmological constant shows that @xmath95 has the role of cosmological constant . as we will see , we compare the charged solution of @xmath10 gravity ( eq . ( [ giiib ] ) ) with the solution of einstein gravity coupled to the conformally invariant maxwell field and state that @xmath95 gravity has the role of electromagnetic field in addition to the cosmological constant . by calculating the ricci and the kretschmann scalars , we find @xmath96 in which we can deduce that there is a essential singularity at @xmath2 . it is notable that the second derivative of the @xmath0 function for this specific model is @xmath97 and so we find that ads solutions are stable in arbitrary dimensions . however , for @xmath84 , the solutions are unstable , one can add another small correction to make it stable in the solar system . we choose @xmath65 correction , which was firstly proposed in ref . expform , starobinsky to explain inflation . thus the total model is @xmath99 , where @xmath72 is constant . by adjust the parameters , this new form of @xmath0 can satisfy the high curvature condition , early universe inflation , stability and the late time acceleration @xcite . in the model , we can obtain at least two solutions . first solution is uncharged and is like eq . ( [ giiia ] ) with the same condition on @xmath100 . second solution is corresponding with the topological einstein - nonlinear maxwell solution in the presence of cosmological constant @xmath101 which we should fix some parameters like @xmath100 and @xmath72 . inserting this solution , ( [ givb ] ) , in the field equations ( [ fieldtt])-(fieldthth ) with the mentioned model , one can conclude that the field equations reduce to @xmath102 d(d-2)q^{2}- \nonumber \\ & & \left [ \left ( 2r\xi + d\right ) \lambda e^{-\xi r}+(d-4)\kappa r^{2}-(d-2)r% \right ] r^{d}=0 \label{factorsiv}\end{aligned}\ ] ] which we should adjust @xmath103 , @xmath104 for satisfaction , and @xmath105 is free parameter . these solutions are very close to previous solutions and there is a curvature singularity at @xmath2 . it is matter of calculation to show that @xmath106 and therefore one can set free parameter @xmath105 to obtain positive value for @xmath75 for arbitrary value of cosmological constant . in this subsection , we generalize previous types of @xmath0 gravity and consider a combination of the mentioned corrections in previous subsections . let us take @xmath0 in the following @xmath108-parametric form @xmath109 the solutions of eqs . ( [ met1 ] ) and ( [ fe ] ) , with this choice of @xmath0 , can be written as two kind of charged and uncharged solutions like previous section , but with different conditions on parameters . if we consider topological schwarzschild @xmath110 in the field equations ( [ fieldtt])-([fieldthth ] ) with the mentioned model , we obtain @xmath111 which it help us to find that in the presence of cosmological constant we should set @xmath112 and @xmath113 are free parameters . in order to study dolgov - kawasaki stability , we should calculate @xmath75@xmath114 } { d-2}}% } { d+2\xi r}\left [ ( d-2n)r^{n}\kappa -(d-2)r-(d\ln r-2)\eta \right ] + \nonumber \\ & & r^{-2}\left [ n(n-1)r^{n}\kappa -\eta \right ] . \label{frrva}\end{aligned}\ ] ] it is not simple to present a condition for stability , but as one can see in fig . [ fig1 ] , we can set free parameters to obtain positive @xmath75 . when we consider topological charged solution with nonzero @xmath1 , @xmath115 the field equations ( [ fieldtt])-([fieldthth ] ) reduce to @xmath116 d(d-2)q^{2}- \nonumber \\ & & \left [ \left ( 2\xi r+d\right ) \lambda e^{-\xi r}-\eta d\ln r-(d-2)r+\left ( 2n - d\right ) \kappa r^{n}+2\eta \right ] rr^{d}=0 \label{factorsvb}\end{aligned}\ ] ] in order to satisfy eq . ( [ factorsvb ] ) with nonzero @xmath117 and @xmath74 , the coefficients of @xmath118 and @xmath119 should set to zero , separately , to obtain @xmath120 considering both of eqs . ( [ gva ] ) and ( [ gvb ] ) , one can show that the asymptotic behavior of these solutions near the origin ( @xmath2 ) and for large value of @xmath39 ( @xmath121 ) is the same as presented expressions in eqs . ( [ riii])-([rriiib ] ) . here we want to discuss about dolgov - kawasaki stability . using eqs . ( [ eqeq1 ] ) and ( [ eqeq2 ] ) in second derivative of eq . ( [ fv ] ) , yealds @xmath122 } { % 2d\lambda } + 1\right ) \xi , \nonumber \\ \psi _ { 3 } & = & \frac{(d-2)}{2d\lambda } \left ( \frac{(d-2)r^{n}\kappa n^{2}}{% 2d\lambda } + 1\right ) . \nonumber\end{aligned}\ ] ] we plot eq . ( [ frrvb ] ) versus @xmath105 in fig . [ fig2 ] , for some fixed free parameters and find that @xmath75 is positive for some values of @xmath82 . one of the @xmath0 models that pass all the observational and theoretical constraints is the starobinsky model @xcite . this model is proposed which produce viable cosmology different from the @xmath123 one at recent times and satisfy cosmological , solar system and laboratory tests . in this model the @xmath0 function define as@xmath124 where @xmath125 , @xmath126 and @xmath127 are constants . looking at equations ( [ met1 ] ) and ( [ fe ] ) , one could obviously obtain the metric functions as follows @xmath128 with a constrain on the free parameters as @xmath129 -\left ( \lambda r_{0}-2\lambda \right ) vr_{0}^{2}=0 , \nonumber \\ & & v=\frac{r_{0}^{2}+r^{2}}{r_{0}^{2}}. \label{factorsvia}\end{aligned}\ ] ] considering the eq . ( [ factorsvia ] ) , it is easy to show that all field equations are satisfied when @xmath130 v^{-n}-vr_{0}^{2}\right ) ^{-1}$ ] . here we conclude that this the solution of this model is corresponding to the schwarzchild ( a)ds solution in einstein gravity . straightforward calculations show that in this case we can not obtain charged solution . it is notable that for uncharged solution we obtain @xmath131 } { % vr_{0}^{2}d(1-v^{n})+4nr^{2 } } , \label{frrvia}\ ] ] where , as one can follow in fig . [ fig3 ] , we obtain stable ( a)ds solutions for suitable @xmath126 . in order to obtain the topological charged solution , we generalize this model with adding an extra term to starobinsky model . we define the generalized starobinsky model in the following form@xmath132 the basic motivation for this extra term is that for @xmath133 , the starobinsky model reduce to einstein gravity with @xmath65 correction and in the generalized starobinsky model , we generalize the power of correction term to arbitrary factor @xmath127 . the solution of eqs . ( [ met1 ] ) and ( [ fe ] ) , in the generalized starobinsky model is as follows @xmath134 with two constrain on the free parameters which appear in the following equation @xmath135 + vr_{0}^{2}\left ( d\lambda r_{0}-2d\lambda \right ) \right ) rr^{d}+ \nonumber \\ & & \left [ -vr_{0}^{2}\left ( r+n\kappa \left ( r\right ) ^{n}\right ) + 2r^{2}n\lambda r_{0}v^{-n}\right ] d(d-2)q^{2}=0 . \label{factorsvib}\end{aligned}\ ] ] which for nonzero @xmath117 and @xmath74 , we obtain @xmath136 v^{-(2n+1)/2}-r_{0}\right ) } , \label{eqeq11 } \\ \kappa & = & \frac{r_{0}-\left [ r^{2}\left ( 2n+1\right ) + r_{0}^{2}\right ] v^{-(2n+1)/2}}{nr^{n-1}\left [ \left ( r_{0}^{2}+3r^{2}\right ) v^{-(2n+1)/2}-r_{0}\right ] } . \label{eqeq22}\end{aligned}\ ] ] considering both of eqs . ( [ gvia ] ) and ( [ gvib ] ) , one can show that the asymptotic behavior of these solutions near the origin ( @xmath2 ) and for large value of @xmath39 ( @xmath121 ) is the same as presented expressions in eqs . ( [ riii])-([rriiib ] ) . finally we calculate second derivative of @xmath0 gravity in generalized starobinsky model to obtain @xmath137 } { % nr_{0}^{3}v^{n+2}\left [ \left ( r_{0}^{2}+3r^{2}\right ) v^{-(2n+1)/2}-r_{0}% \right ] } + \nonumber \\ & & \frac{r_{0}-\left [ r^{2}\left ( 2n+1\right ) + r_{0}^{2}\right ] v^{-(2n+1)/2}% } { \frac{r}{n-1}\left [ \left ( r_{0}^{2}+3r^{2}\right ) v^{-(2n+1)/2}-r_{0}% \right ] } , \label{frrvib}\end{aligned}\ ] ] where we plot it in fig . this figure shows that we can obtain stable solution for special values of @xmath126 for positive and negative cosmological constant . in this section , we consider the @xmath16-dimensional einstein-@xmath74 gravity with conformally invariant maxwell field ( e@xmath74cim ) . in other word , the electromagnetic field equation be invariant under conformal transformation ( @xmath138 and @xmath139 ) . for more investigations and motivations one can see ref . the action of e@xmath74cim is @xmath140 , \label{ig}\ ] ] where @xmath141 is a coupling constant . in order to ensure a physical interpretation of our future solutions , we fix @xmath142 and so the energy density , i.e. the @xmath143 component of the energy - momentum tensor in the orthonormal frame , is positive . varying the action ( [ ig ] ) with respect to the metric tensor @xmath19 and the electromagnetic field @xmath144 , the equations of gravitational and electromagnetic fields may be obtained as @xmath145 , \label{graveq}\ ] ] @xmath146 = 0 . \label{maxeq}\ ] ] the maxwell equation ( [ maxeq ] ) with metric ( [ met1 ] ) can be integrated immediately to give @xmath147 where @xmath148 , an integration constant and it is proportional to the electric charge of the spacetime . here , we calculate the electric charge per unit volume of the black hole by finding the flux of the electromagnetic field at infinity , obtaining @xmath149 which confirm that @xmath148 is related to the electrical charge of the spacetime @xmath150 . inserting the maxwell fields ( [ ftr ] ) and the metric ( [ met1 ] ) in the field equation ( [ graveq ] ) , one can show that these equations have the following solutions @xmath151 where @xmath152 is the integration constant which is related to mass parameter and when we set @xmath92 , reissner - nordstrm solution is recovered . comparing eq . ( [ g(r)cim ] ) with eq . ( [ giiib ] ) ( and let @xmath153 ) , show that there is a correspondence between the e@xmath154cim solutions and the solutions of @xmath0 gravity without matter field . considering eq . ( [ charge ] ) , one can find @xmath155 where in four dimension it reduce to total electric charge of reissner - nordstrm spacetime ( @xmath156 ) . in summary we investigated some different well - known types of @xmath0 theory as a correction to einstein gravity with constant ricci scalar metric in higher dimensions . the basic motivation arise from einstein dream which is creating a geometrical unified theory of physics . we found two kind of charged and uncharged black hole solutions with constant ricci scalar and different topology in horizon and investigated some geometrical properties such as singularity and asymptotic behavior . in this paper , we considered some various class of @xmath10 gravity and surprisingly , we found that for special cases of @xmath0 gravity we can obtain charged black holes in addition to schwarzschild solutions . it is notable that we started with pure @xmath0 gravity without matter field and cosmological constant , but we found that , by fixing some of the free parameters , presented charged and uncharged solutions are corresponding with the topological nonlinear charged and the schwarzschild black holes in the presence of cosmological constant . in other word , it is not necessary to insert ( by hand ) the cosmological constant in the field equations , and by setting some of free parameters in @xmath0 gravity , one may obtain ( a)ds solution from pure geometry . in addition , one may think about charged solution in @xmath0 gravity , without considering the ( nonlinear- ) maxwell stress - energy tensor . also , we calculated kreschmann scalar to show that there is a curvature singularity at @xmath2 . it means these solutions are interpreted as topological ( un)charged black holes . it is notable that the kreschmann scalar of charged solutions diverges as @xmath3 , but with a rate slower than that of uncharged solutions . in addition , for any kind of presented @xmath0 models , we analyze dolgov - kawasaki stability and show that one can find stable solutions provided the parameters of the @xmath0 gravity are chosen suitably . finally , it is also be desirable to study the cosmological view , the causal structure and dynamical stability as well as thermodynamical properties of the black hole solutions derived here . generalization of these solutions to non - constant ricci scalar with nontrivial topology of horizons as well as obtaining a special transformation to find a corresponding between @xmath0 gravity and einstein - nonlinear maxwell gravity , remain to be carried out in the future .	in this paper we obtain topological static solutions of some kind of pure @xmath0 gravity . the present solutions are two kind : first type is uncharged solution which corresponds with the topological ( a)ds schwarzschild solution and second type has electric charge and is equivalent to the einstein-@xmath1-conformally invariant maxwell solution . in other word , starting from pure gravity leads to ( charged ) einstein-@xmath1 solutions which we interpreted them as ( charged ) ( a)ds black hole solutions of pure @xmath0 gravity . calculating the ricci and kreschmann scalars show that there is a curvature singularity at @xmath2 . we should note that the kreschmann scalar of charged solutions goes to infinity as @xmath3 , but with a rate slower than that of uncharged solutions .
let @xmath0 be a connected reductive algebraic group over an algebraic closure of a finite field , endowed with an endomorphism @xmath1 , a power of which is a frobenius endomorphism . let @xmath2 be a prime number distinct from the defining characteristic of @xmath0 and @xmath3 a finite extension of @xmath4 , large enough for the finite groups considered . let @xmath5 be the ring of integers of @xmath3 over @xmath6 and @xmath7 the residue field . we will denote by @xmath8 a ring that is either @xmath3 , @xmath5 or @xmath7 . the main tool for the study of representations of @xmath9 over @xmath10 is the deligne - lusztig induction . let @xmath11 be an @xmath1-stable levi subgroup of @xmath0 contained in a parabolic subgroup @xmath12 with unipotent radical @xmath13 so that @xmath14 . consider the deligne - lusztig variety @xmath15 it has a left action of @xmath9 and a right action of @xmath16 by multiplication . the corresponding complex of @xmath2-adic cohomology induces a triangulated functor @xmath17 and a morphism @xmath18 : g_0(\l\lb^f)\to g_0(\l\gb^f).\ ] ] this is the usual harish - chandra construction when @xmath12 is @xmath1-stable . let @xmath19 be a group langlands dual to @xmath0 , with frobenius @xmath20 . consider the set @xmath21 of characters of irreducible representations of @xmath9 over @xmath3 . deligne and lusztig gave a decomposition of @xmath21 into rational series @xmath22 where @xmath23 runs over the set of @xmath24-conjugacy classes of semi - simple elements of @xmath24 . the _ unipotent characters _ of @xmath9 are those in @xmath25 . let @xmath11 be an @xmath1-stable levi subgroup of @xmath0 with dual @xmath26 containing @xmath27 . lusztig constructed a bijection @xmath28 if @xmath29 , then there is a bijection @xmath30 where @xmath31 is the one - dimensional character of @xmath16 corresponding to @xmath32 , and we obtain a bijection @xmath33 this provides a description of irreducible characters of @xmath9 in the rational series @xmath23 in terms of unipotent characters of an other group , when @xmath27 is a levi subgroup of @xmath19 . let us now consider the modular version of the theory described above . let @xmath32 be a semi - simple element of @xmath24 of order prime to @xmath2 . consider @xmath34 , where @xmath35 runs over conjugacy classes of semi - simple elements of @xmath24 whose @xmath36-part is @xmath23 . brou and michel @xcite have shown this is a union of blocks of @xmath37 . the sum of the corresponding block idempotents is an idempotent @xmath38 , and we obtain a decomposition @xmath39 where @xmath23 runs over @xmath24-conjugacy classes of semi - simple @xmath36-elements of @xmath24 . let @xmath11 be an @xmath1-stable levi subgroup of @xmath0 with dual @xmath40 containing @xmath27 . let @xmath12 be a parabolic subgroup of @xmath0 with unipotent radical @xmath13 and levi complement @xmath11 . brou @xcite conjectured that the @xmath41-bimodule @xmath42 induces a morita equivalence between @xmath43 and @xmath44 . this was proven by brou @xcite when @xmath11 is a torus and in @xcite in general . brou also conjectured that the truncated complex of cohomology @xmath45 ( rickard s refinement of @xmath46 , well defined in the homotopy category @xcite ) induces a splendid rickard equivalence between @xmath43 and @xmath44 : it induces not only an equivalence of derived categories , but even an equivalence of homotopy categories , and it induces a similar equivalence for centralizers of @xmath2-subgroups . one of our main results here is a proof of that conjecture . in order to show that there is a homotopy equivalence , for connected groups , we show that the global functor induces local derived equivalences for centralizers of @xmath2-subgroups . since such centralizers need not be connected , we need to extend the results of @xcite to disconnected groups . so , part of this work involves working with disconnected groups . we also extend the `` jordan decomposition equivalences '' ( morita and splendid rickard ) to the `` quasi - isolated case '' : assume now only @xmath47 , and that @xmath40 is minimal with respect to this property . we show that the right action of @xmath16 on @xmath42 extends to an action of @xmath48 commuting with the action of @xmath9 , and the resulting bimodule induces a morita equivalence between @xmath43 and @xmath49 . similarly , the complex @xmath50 induces a splendid rickard equivalence between @xmath43 and @xmath49 . as a consequence , we deduce that the bijection between blocks of @xmath43 and @xmath49 preserves the local structure , and in particular , preserves defect groups . cabanes and enguehard have proven this under some assumptions on @xmath2 ( * ? ? ? * proposition 5.1 ) , and kessar and malle in the setting of @xcite , when one of the blocks under consideration has abelian defect groups ( modulo a central @xmath2-subgroup ) ( * ? ? ? * theorem 1.3 ) , an important step in their proof of half of brauer s height zero conjecture for all finite groups ( * ? ? * theorem 1.1 ) and the second half for quasi - simple groups ( * ? ? ? * main theorem ) . let us summarize this . [ th : introequiv ] assume @xmath47 and that @xmath40 is minimal with respect to this property . the right action of @xmath16 on @xmath51 extends to an action of @xmath52 and the resulting complex @xmath53 induces a splendid rickard equivalence between @xmath43 and @xmath49 . the bimodule @xmath54 induces a morita equivalence between @xmath43 and @xmath49 . the bijections between blocks of @xmath43 and @xmath49 induced by those equivalences preserve the local structure . significant progress has been made recently on counting conjectures for finite groups , using the classification of finite simple groups , and @xcite has proved very useful . we hope this theorem will lead to simplifications and new results . the character - theoretic consequence of this theorem is that , for groups with disconnected center , the jordan decomposition shares many of the properties of that for the connected case ( commutation with deligne - lusztig induction for example ) . in type @xmath55 , the jordan decomposition of characters links all series to unipotent series of smaller groups : even in that case , the good behaviour of those correspondences was known only when @xmath56 is large ( bonnaf @xcite for @xmath57 and cabanes @xcite for @xmath58 ) . one of the two key steps in @xcite was the proof that the category of perfect complexes for @xmath37 is generated by the complexes @xmath59 , where @xmath60 runs over borel subgroups of @xmath0 with an @xmath1-stable maximal torus . we show here a more precise result of generation of the derived category of @xmath37 . let @xmath61 be the set @xmath62 , where @xmath63 runs over @xmath1-stable maximal tori of @xmath0 , @xmath60 over borel subgroups of @xmath0 containing @xmath63 , and @xmath64 over isomorphism classes of @xmath65-modules . the set @xmath61 generates @xmath66 ( as a thick subcategory ) if and only if all elementary abelian @xmath2-subgroups of @xmath9 are contained in tori . this , in turn , requires an extension of the results of brou - michel @xcite on the compatibility between deligne - lusztig series of characters and the brauer morphism , to disconnected groups . we are able to achieve this by refining our result on the generation of the category of perfect complexes to a generation of the category of @xmath2-permutation modules whose vertices are contained in tori ( the crucial case is that of connected groups ) . such a result allows us to obtain a generating result for the full derived category , under the assumption that all elementary abelian @xmath2-subgroups are contained in tori . note that the condition is automatically satisfied for @xmath67 and @xmath68 ( see examples [ ex : elementaryabelian ] ) and when @xmath2 is _ very good _ for @xmath0 . it is known in most cases , and conjectured in general , that the map @xmath69 on grothendieck groups is actually independent of @xmath12 ( @xcite when @xmath11 is a torus and @xcite when @xmath70 and @xmath1 is a frobenius endomorphism over @xmath71 ) . on the other hand , the functor @xmath72 does depend on @xmath12 . our main new geometrical result proves the independence after truncating by a suitable series . let @xmath73 and @xmath74 be two parabolic subgroups admitting a common levi complement @xmath11 . denote by @xmath75 the unipotent radical of the parabolic subgroup of @xmath19 corresponding to @xmath76 . let @xmath32 be a semi - simple element of @xmath77 of order prime to @xmath2 . if @xmath78 then there is an isomorphism of functors between @xmath79 @xmath80 : \drm^b(\l\lb^f e_s^{\lb^f } ) \longto \drm^b(\l\gb^fe_s^{\gb^f } ) , \leqno{\text{\it and}}\ ] ] where @xmath81 . for instance , if @xmath82 , then the assumption of theorem [ thd ] is satisfied . this is the key result to prove theorem [ th : introequiv ] . this result shows that when @xmath47 , the @xmath41-bimodule @xmath42 is independent of @xmath12 , a question left open in @xcite . we deduce that the bimodule is stable under the action of @xmath83 . using an embedding in a group with connected center , we show that the obstruction for extending the action of @xmath16 to @xmath52 does vanish . * remark. * theorem 1.3 is used in @xcite to construct equivalences of categories between tamely ramified blocks of @xmath84-adic general linear groups . roughly speaking , the main idea of @xcite is to `` glue '' the bimodules giving the morita equivalences of @xcite along a suitable building . the gluing process crucially uses the independence of the bimodules on the choice of parabolic subgroups . we begin in [ se : generation ] with the study of generation of the category of perfect complexes , then we move to complexes of @xmath2-permutation modules and finally we derive our result on the derived category . a key tool , due to rickard , is that the brauer functor applied to the complex of cohomology of a variety is the complex of cohomology of the fixed point variety . section [ se : rational ] is devoted to the study of rational series and their compatibility with local block theory . brou and michel proved a commutation formula between generalized decomposition maps and deligne - lusztig induction . we need to extend the compatibility between brauer and deligne - lusztig theory to disconnected groups , and check that the local blocks obtained from a series satisfying @xmath85 also satisfy a similar assumption @xmath86 . from [ se : comparing ] onwards , the group @xmath0 is assumed to be connected . sections [ se : comparing ] and [ sec : proof ] are devoted to the study of the dependence of the deligne - lusztig induction with respect to the parabolic subgroup . the first section is devoted to the particular case of varieties associated with borel subgroups ( and generalizations involving sequences of elements ) . it is convenient there to work with a reference maximal torus . this is the crucial case , from which the general one is deduced in the latter section , where we go back to levi subgroups that do not necessarily contain that fixed maximal torus . the final section [ se : jordan ] is devoted to the jordan decomposition . we start by providing an extension of the action of @xmath52 on the cohomology bimodule by proving that the cocycle obstruction would survive in a similar setting for a group with connected center , where the action does exist . the rickard equivalence is obtained inductively , and that induction requires working with disconnected groups . in an appendix , we provide some results on the homotopy category of complexes of @xmath2-permutation modules for a general finite group . we would like to thank the referee for an extraordinarily thorough list of suggestions , which greatly improved our paper . let @xmath2 be a prime number , @xmath3 a finite extension of @xmath4 large enough for the finite groups considered , @xmath5 its ring of integers over @xmath6 and @xmath7 its residue field . we will denote by @xmath8 a ring that is either @xmath3 , @xmath5 or @xmath7 . given @xmath87 an additive category , we denote by @xmath88 the category of bounded complexes of objects of @xmath87 and by @xmath89 its homotopy category . let @xmath55 be a @xmath8-algebra , finitely generated and projective as a @xmath8-module . we denote by @xmath90 the algebra opposite to @xmath55 . we denote by @xmath91 the category of finitely generated @xmath55-modules and by @xmath92 its full subcategory of projective modules . we denote by @xmath93 the grothendieck group of @xmath91 . we put @xmath94 , @xmath95 and @xmath96 . we denote by @xmath97 the thick full subcategory of perfect complexes ( complexes quasi - isomorphic to objects of @xmath98 ) . let @xmath99 . there is a unique ( up to a non - unique isomorphism ) complex @xmath100 which is isomorphic to @xmath53 in the homotopy category @xmath101 and which has no non - zero direct summand that is homotopy equivalent to @xmath102 . note that @xmath103 for some @xmath104 homotopy equivalent to zero . we denote by @xmath105 the total @xmath106-complex , with degree @xmath107 term @xmath108 . let @xmath109 be @xmath8-algebra , finitely generated and projective as a @xmath8-module . let @xmath53 be a bounded complex of @xmath110-modules , finitely generated and projective as left @xmath55-modules and as right @xmath109-modules . we say that @xmath53 induces a _ rickard equivalence _ between @xmath55 and @xmath109 if the canonical map @xmath111 is an isomorphism in @xmath112 and the canonical map @xmath113 is an isomorphism in @xmath114 . let @xmath115 be a finite group . we denote by @xmath116 the opposite group to @xmath115 . we put @xmath117 . given @xmath118 , we denote by @xmath119 the order of @xmath120 . let @xmath121 be a subgroup of @xmath115 and @xmath122 . we denote by @xmath123 the equivalence of categories @xmath124 where @xmath125 as a @xmath8-module and the action of @xmath126 on @xmath127 is given by the action of @xmath128 on @xmath64 . we also denote by @xmath123 the corresponding isomorphism of grothendieck groups @xmath129 we assume @xmath130 or @xmath131 in the remainder of [ se : finitegroups ] . an _ @xmath2-permutation @xmath132-module _ is defined to be a direct summand of a finitely generated permutation module . we denote by @xmath133 the full subcategory of @xmath134 with objects the @xmath2-permutation @xmath132-modules . let @xmath135 be an @xmath2-subgroup @xmath135 of @xmath115 . we consider the _ _ @xmath136\perm$ ] . given @xmath137 , we define @xmath138 as the image of @xmath139 in @xmath140 , where @xmath140 is the largest quotient of @xmath141 on which @xmath135 acts trivially . we denote by @xmath142 the algebra morphism given by @xmath143 where @xmath144 for @xmath118 . given @xmath145 and @xmath146 an idempotent , we have @xmath147 . let @xmath121 be a subgroup of @xmath115 , let @xmath148 be an idempotent of @xmath149 and @xmath150 an idempotent of @xmath151 . let @xmath152 . we say that @xmath53 is _ splendid _ if the @xmath153 s are @xmath2-permutation modules whose indecomposable direct summands have a vertex contained in @xmath154 . let @xmath84 be a prime number different from @xmath2 and @xmath155 an algebraic closure of @xmath156 . by variety , we mean a quasi - projective algebraic variety over @xmath155 . let @xmath157 be a variety acted on by a finite group @xmath115 . there is an object @xmath158 of @xmath159 , well defined up to a unique isomorphism . it is a representative in the homotopy category of @xmath132-modules of the isomorphism class of the complex of tale @xmath8-cohomology with compact support of @xmath157 constructed as @xmath160 of the godement resolution ( cf @xcite , @xcite , and @xcite ) . we denote by @xmath161 the image of @xmath158 in @xmath162 . assume @xmath130 or @xmath7 and let @xmath135 be an @xmath2-subgroup of @xmath115 . the inclusion @xmath163 induces an isomorphism ( * ? ? ? * theorem 4.2 ) @xmath164 let @xmath0 be a ( possibly disconnected ) reductive algebraic group endowed with an endomorphism @xmath1 , a power @xmath165 of which is a frobenius endomorphism defining a rational structure over a finite field @xmath166 of characteristic @xmath84 . we refer to @xcite for basic results on disconnected groups . recall that a torus of @xmath0 is torus of @xmath167 . following the classical terminology ( cf for example @xcite ) , we define a _ borel subgroup _ of @xmath0 to be a maximal connected solvable subgroup of @xmath0 . we define a _ parabolic subgroup _ of @xmath0 to be a subgroup @xmath12 of @xmath0 such that @xmath168 is complete . we define the _ unipotent radical _ @xmath13 of a parabolic subgroup @xmath12 to be its unique maximal connected unipotent normal subgroup . levi complement _ to @xmath13 in @xmath12 is a subgroup @xmath11 of @xmath12 such that @xmath14 . note that a closed subgroup @xmath12 of @xmath0 is a parabolic subgroup of @xmath0 if and only if @xmath169 is a parabolic subgroup of @xmath167 . let @xmath12 be a parabolic subgroup of @xmath0 . we have @xmath170 . the unipotent radical @xmath13 of @xmath12 coincides with that of @xmath169 . a levi complement to @xmath13 in @xmath12 is a subgroup of the form @xmath171 , where @xmath172 is a levi complement of @xmath13 in @xmath173 ( then @xmath174 and @xmath14 ) . note that our definition of parabolic subgroup is more general than that of `` parabolic '' subgroup of @xcite , which requires @xmath175 . we denote by @xmath176 the set of pairs @xmath177 where @xmath63 is an @xmath1-stable maximal torus of @xmath0 and @xmath178 is an irreducible character of @xmath179 . note that here @xmath63 is a torus of @xmath167 . given an integer @xmath180 , we denote by @xmath181 the set of pairs @xmath182 such that the order of @xmath178 is prime to @xmath180 . we put @xmath183 if @xmath184 and @xmath185 if @xmath130 or @xmath7 ( recall that @xmath7 is a field of characteristic @xmath2 ) . given @xmath12 a parabolic subgroup of @xmath0 with unipotent radical @xmath13 and @xmath1-stable levi complement @xmath11 , we define the deligne - lusztig variety @xmath186 this is a smooth variety , as in the case of connected reductive groups . it has a left action by multiplication of @xmath9 and a right action by multiplication of @xmath16 ( note that the left and right actions of @xmath187 coincide ) . this provides a triangulated functor [ eq : rlg - def ] and a morphism @xmath188 : g_0(\l\lb^f)\to g_0(\l\gb^f).\ ] ] we put @xmath189 . [ re : actionn ] since @xmath190 depends only on @xmath13 , it is endowed with an action of @xmath191 , which is the group of rational points of the maximal levi subgroup with connected component @xmath192 . the aim of this section is to extend ( * ? ? ? * theorem a ) to the case of disconnected groups , and to deduce a generation theorem for the derived category . in this section [ se : generation ] , @xmath0 is a ( possibly disconnected ) reductive algebraic group . [ prop : centralisateur ] let @xmath135 be a finite solvable @xmath195-group of automorphisms of @xmath0 that commute with @xmath1 and normalize @xmath196 . the group @xmath197 is reductive . @xmath198 is a parabolic subgroup of @xmath197 whose unipotent radical is @xmath199 and admitting @xmath200 as an @xmath1-stable levi complement . in particular , @xmath199 is connected . the natural map @xmath201 is an isomorphism of @xmath202-varieties . @xmath203 . the natural map @xmath204 is an isomorphism of @xmath205-varieties . if @xmath135 is an @xmath2-group , it gives rise to an isomorphism @xmath206 in @xmath207 . assume first @xmath135 is cyclic , generated by an element @xmath208 . \(a ) and ( b ) follow from ( * ? ? ? * proposition 1.3 , theorem 1.8 , proposition 1.11 ) . \(c ) note that that both varieties are smooth ( for @xmath209 , this follows from the fact that @xmath135 is a @xmath195-group and @xmath210 is smooth ) . the injectivity of the map is clear . let us prove the surjectivity . let @xmath211 . then , @xmath212 . denote by @xmath213 the automorphism @xmath214 of @xmath0 . since @xmath215 is semisimple , it stabilizes a maximal torus of @xmath169 ( see ( * ? ? ? * theorem 7.5 ) ) , hence it stabilizes the unique levi complement @xmath216 of @xmath169 containing this maximal torus . since all levi complements are conjugate under the action of @xmath13 , there exists @xmath217 such that @xmath218 . it follows that @xmath219 and @xmath220 normalizes @xmath192 , hence @xmath221 , so @xmath222 , as desired . the tangent space at @xmath13 of @xmath209 is the @xmath135-invariant part of the tangent space of @xmath210 at @xmath13 . that last tangent space is a quotient of the tangent space of @xmath0 at the origin . it follows that the canonical map @xmath223 induces a surjective map between tangent spaces at the origin . consequently , the canonical map @xmath201 induces a surjective map between tangent spaces at the origin . we deduce that the map is an isomorphism . \(d ) the number of @xmath1-stable maximal tori of @xmath11 is a power of @xmath84 ( see ( * ? ? ? * corollary 14.16 ) ) . since @xmath135 is a @xmath195-group , it normalizes some @xmath1-stable maximal torus . using now the root system with respect to this maximal torus , we deduce that there exists a @xmath135-stable subgroup @xmath224 of @xmath13 such that @xmath225 and @xmath226 . therefore , @xmath227 and the result follows . \(e ) follows immediately from ( c ) and ( d ) . we prove now the proposition by induction on @xmath228 . let @xmath229 be a normal subgroup of @xmath135 of index a prime number and let @xmath230 , @xmath231 . let @xmath232 be the subgroup of @xmath135 generated by @xmath208 . by induction , the proposition holds for @xmath135 replaced by @xmath229 : we have a reductive group @xmath233 and a parabolic subgroup @xmath234 with unipotent radical @xmath235 and an @xmath1-stable levi complement @xmath236 . these are all stable under @xmath232 . the cyclic case of the proposition applied to the action of @xmath232 on @xmath237 establishes the proposition for the action of @xmath135 on @xmath238 . if @xmath135 is a finite solvable @xmath195-subgroup of @xmath9 , then @xmath239 is reductive . if in addition @xmath135 normalizes @xmath196 , then @xmath240 is a parabolic subgroup of @xmath239 with unipotent radical @xmath199 and levi complement @xmath241 . the maps defined in ( e ) of proposition [ prop : centralisateur ] are equivariant for the diagonal action of @xmath242 . we will need a converse to proposition [ prop : centralisateur ] , in the case of tori . [ lem : bijtori ] let @xmath135 be a finite solvable @xmath195-group of automorphisms of @xmath0 that commute with @xmath1 . we assume @xmath135 stabilizes a maximal torus of @xmath0 and a borel subgroup containing that maximal torus . let @xmath243 be an @xmath1-stable maximal torus of @xmath197 contained in a borel subgroup @xmath244 of @xmath197 . then , @xmath245 is an @xmath1-stable maximal torus of @xmath0 that is contained in a @xmath135-stable borel subgroup @xmath60 of @xmath0 such that @xmath246 . note that the lemma holds for @xmath135 cyclic by ( * ? ? ? * theorem 1.8 ) . we proceed by induction on @xmath228 as in the proof of proposition [ prop : centralisateur ] , and we keep the notations of that proof . we know ( lemma for @xmath232 ) that @xmath247 is an @xmath1-stable maximal torus of @xmath248 . by induction , @xmath249 is an @xmath1-stable maximal torus of @xmath0 . the existence of the borel subgroup can be obtained as in @xcite . let @xmath250 be a maximal torus of @xmath0 stable under @xmath135 and @xmath251 be a borel subgroup of @xmath0 containing @xmath250 and stable under @xmath135 . by proposition [ prop : centralisateur ] , @xmath252 is a maximal torus of @xmath197 and @xmath253 is a borel subgroup containing it . so , there is @xmath254 such that @xmath255 and @xmath256 . let @xmath257 . this is a @xmath135-stable borel subgroup of @xmath0 containing @xmath245 . by proposition [ prop : centralisateur ] , @xmath258 is a borel subgroup of @xmath197 , hence @xmath246 . to complete proposition [ prop : centralisateur ] , note the following result . [ lem : delta ] let @xmath259 be an @xmath2-subgroup of @xmath260 such that @xmath261 . then @xmath259 is @xmath262-conjugate to a subgroup of @xmath263 . replacing @xmath11 by @xmath264 , we can assume that @xmath265 . let @xmath266 ( respectively @xmath267 ) denote the image of @xmath259 through the second ( respectively first ) projection and let @xmath268 . if @xmath269 , then there exists @xmath270 such that @xmath271 . therefore , @xmath272 , hence @xmath273 . this implies that @xmath274 . we denote by @xmath275 the composition @xmath276 . since @xmath277 ( respectively @xmath135 ) acts freely on @xmath210 as they are @xmath2-groups , the previous computation shows that @xmath31 is an isomorphism , and that @xmath278 now , there exists a positive integer @xmath279 such that @xmath280 and @xmath281 . so @xmath282 acts by left translation on @xmath283 . since @xmath282 is a finite @xmath2-group and @xmath284 is a power of @xmath84 , it follows that @xmath282 has a fixed point in @xmath283 . consequently , there exists @xmath217 such that @xmath285 for all @xmath286 . in other words , @xmath287 . this means that , by replacing @xmath288 by @xmath289 if necessary , we may assume that @xmath290 . therefore , @xmath291 and @xmath292 . now , @xmath293 but , since @xmath294 for all @xmath270 , we deduce that @xmath295 . so @xmath296 ( see proposition [ prop : centralisateur](b ) and ( d ) ) . so , by lang s theorem , there exists @xmath297 such that @xmath298 . this implies that @xmath299 , and @xmath300 as expected . [ cor : deltavertex ] the indecomposable summands of the @xmath301-module @xmath302 have a vertex contained in @xmath263 . let @xmath135 be an @xmath2-subgroup of @xmath303 that is not @xmath262-conjugate to a subgroup of @xmath263 . we have @xmath304 in @xmath305 by lemma [ lem : delta ] . the result follows now from lemma [ lem : bounded ] . given @xmath64 a simple @xmath306-module , we denote by @xmath307 the set of pairs @xmath308 such that @xmath63 is an @xmath1-stable maximal torus of @xmath0 and @xmath60 is a ( connected ) borel subgroup of @xmath0 containing @xmath63 such that @xmath309 . we then set @xmath310 . the following two theorems are proved in ( * ? ? ? * theorem a ) whenever @xmath0 is connected . [ theo : a ] let @xmath64 be a simple @xmath306-module . then @xmath311 . moreover , given @xmath312 such that @xmath313 , we have @xmath314 ) = 0\ ] ] for all @xmath315 . by ( [ eq : ind - rlg ] ) , we have @xmath316 ) = \operatorname{{\mathrm{hom}}}_{\drm^b(\l\gb^{\circ f})}(r\g_c(\yb_{\bb}^{\gb^\circ},\lambda ) , \operatorname{{\mathrm{res}}}_{\gb^{\circ f}}^{\gb^f } m[-i]).\ ] ] since @xmath64 is simple and @xmath317 , it follows that @xmath318 is semisimple . since the theorem holds in @xmath319 ( see ( * ? ? ? * proof of theorem a ) ) , we know that @xmath307 is not empty . the second statement follows from the fact that , if two simple @xmath320-modules @xmath321 and @xmath322 occur in the semisimple module @xmath318 , then they are conjugate under @xmath9 , and so @xmath323 . [ theo : a - engendrement ] the triangulated category @xmath324 is generated by the complexes @xmath325 , where @xmath63 runs over the set of @xmath1-stable maximal tori of @xmath0 and @xmath60 runs over the set of borel subgroups of @xmath0 containing @xmath63 . in this subsection [ se : genderived ] , we assume @xmath130 or @xmath7 . we refer to appendix [ app : l - perm ] for the needed facts about @xmath2-permutation modules . let @xmath135 be an @xmath2-subgroup of @xmath9 and let @xmath64 be an indecomposable @xmath2-permutation @xmath326$]-module with vertex @xmath327 . we denote by @xmath328 $ ] the set of pairs @xmath308 satisfying the following conditions : * @xmath63 is an @xmath1-stable maximal torus of @xmath0 contained in a borel subgroup @xmath60 of @xmath0 such that @xmath135 normalizes @xmath308 * @xmath64 is a direct summand of a term of the complex @xmath329 . we set @xmath330=\min_{(\tb,\bb ) \in \yc[m ] } \dim(\yb_{c^\circ_{\bb}(q)}^{c^\circ_{\gb}(q)})$ ] . [ lem : l - perm ] if @xmath135 normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup of @xmath0 , then @xmath328 \neq \vide$ ] . moreover , given @xmath332 $ ] such that @xmath330=\dim(\yb_{c^\circ_{\bb}(q)})$ ] , the degree @xmath333 term of the complex @xmath334 has no direct summand isomorphic to @xmath64 if @xmath335 $ ] . note that @xmath336 , and we identify @xmath337 with @xmath338 via the first projection . let @xmath339 , an indecomposable projective @xmath340-module . let @xmath341 be the simple quotient of @xmath342 . now , let @xmath244 be a borel subgroup of @xmath343 admitting an @xmath1-stable maximal torus @xmath243 . by lemma [ lem : bijtori ] , @xmath344 is an @xmath1-stable maximal torus of @xmath0 and it is contained in a borel subgroup @xmath60 of @xmath0 such that @xmath345 . we set @xmath346 . by proposition [ prop : centralisateur](e ) , we have @xmath347 in @xmath348 . it follows from lemma [ lem : bounded ] that @xmath64 is a direct summand of the @xmath333-th term of @xmath349 if and only if @xmath342 is a direct summand of @xmath350 . so the result follows from theorem [ theo : a ] . note that @xmath330=d[v]=d(l)=\dim \yb_{\bb_q}^{c_\gb(q)}$ ] . recall that given a borel subgroup @xmath60 of @xmath0 with an @xmath1-stable maximal torus @xmath63 , the variety @xmath351 has a right action of @xmath352 ( cf remark [ re : actionn ] ) . let @xmath353 be the thick subcategory of @xmath354 generated by the complexes of the form @xmath355 where * @xmath63 runs over @xmath1-stable maximal tori of @xmath0 * @xmath60 runs over borel subgroups of @xmath0 containing @xmath63 * @xmath135 is an @xmath2-subgroup of @xmath352 * and @xmath341 is a @xmath356-module , free of rank @xmath357 over @xmath10 . let @xmath358 be the full subcategory of @xmath359 consisting of modules whose indecomposable direct summands have a one - dimensional source and a vertex @xmath135 which normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup of @xmath0 . [ theo : engendrement ] we have @xmath360 . given @xmath52 an indecomposable @xmath306-module with a one - dimensional source @xmath341 and a vertex @xmath135 which normalizes a pair @xmath331 , where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup , we set @xmath361 $ ] to be the minimum of the numbers @xmath330 $ ] , where @xmath64 runs over the set of indecomposable @xmath2-permutation @xmath362-modules with vertex @xmath327 and such that @xmath52 is a direct summand of @xmath363 . note that if @xmath64 is an indecomposable @xmath2-permutation @xmath362-module with vertex properly contained in @xmath327 , then the indecomposable direct summands of @xmath363 have vertices of size @xmath364 and a one - dimensional source . since the @xmath362-module @xmath365 is a direct sum of indecomposable modules with vertices contained in @xmath327 , we deduce that there is an indecomposable @xmath2-permutation @xmath362-module @xmath64 with vertex @xmath327 and such that @xmath52 is a direct summand of @xmath363 . we now proceed by induction on the pair @xmath366)$ ] ( ordered lexicographically ) to show that @xmath367 . fix @xmath64 an indecomposable @xmath2-permutation @xmath362-module @xmath64 with vertex @xmath327 and such that @xmath52 is a direct summand of @xmath363 , with @xmath361=d[m]$ ] . let @xmath332 $ ] be such that @xmath368 $ ] and let @xmath369 . if @xmath335 $ ] , then lemma [ lem : l - perm ] and corollary [ cor : deltavertex ] show that the indecomposable direct summands @xmath370 of @xmath350 have vertices of size @xmath364 , or have vertex @xmath327 and satisfy @xmath371 < d[m]$ ] . therefore , the indecomposable direct summands @xmath372 of @xmath373 have vertices of size @xmath364 or have vertex @xmath135 and satisfy @xmath374 < d[n]$ ] . we deduce from the induction hypothesis that @xmath375 for @xmath376 $ ] . since @xmath52 is a direct summand of @xmath377 } \otimes_{\l q } l$ ] and @xmath378 by construction , we deduce that @xmath367 . [ coro : engendrement ] assume that every elementary abelian @xmath2-subgroup of @xmath9 normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup of @xmath0 . then @xmath379 is generated , as a triangulated category closed under direct summands , by the complexes @xmath380 , where @xmath63 runs over the set of @xmath1-stable maximal tori of @xmath0 , @xmath60 runs over the set of borel subgroups of @xmath0 containing @xmath63 , @xmath135 runs over the set of @xmath2-subgroups of @xmath352 and @xmath341 runs over the set of ( isomorphism classes ) of @xmath356-modules which are free of rank @xmath357 over @xmath10 . since the category @xmath381 is generated , as a triangulated category closed under taking direct summands , by indecomposable modules with elementary abelian vertices and one - dimensional source ( * ? ? ? * corollary 2.3 ) , the statement follows from theorem [ theo : engendrement ] . it is easy to show conversely that if @xmath379 is generated by the complexes @xmath380 as in corollary [ coro : engendrement ] , then @xmath381 is generated by indecomposable modules with a one - dimensional source and an elementary abelian vertex that normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup . in particular , the generation assumption for @xmath131 implies that all elementary abelian @xmath2-subgroups of @xmath9 are contained in maximal tori . the particular case @xmath382 ( for arbitrary @xmath107 ) is enough to ensure that @xmath383 is generated by indecomposable modules with elementary abelian vertices and one - dimensional source , for any finite group @xmath121 this fact is a straightforward consequence of serre s product of bockstein s theorem , but we know of no other proof . it would be interesting to find a direct proof of that result for @xmath384 . recall that an element of @xmath385 is _ uniform _ if it is in the image of @xmath386 , where @xmath63 runs over the set of @xmath1-stable maximal tori of @xmath0 . one can actually describe exactly which complexes are `` uniform '' . [ cor : uniform ] let @xmath387 be the full triangulated subcategory of @xmath379 generated by the complexes @xmath388 where @xmath63 runs over the set of @xmath1-stable maximal tori of @xmath0 , @xmath60 runs over the set of borel subgroups of @xmath0 containing @xmath63 and @xmath64 runs over the set of ( isomorphism classes ) of finitely generated @xmath389-modules . assume that every elementary abelian @xmath2-subgroup of @xmath9 normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup of @xmath0 . an object @xmath53 of @xmath390 is in @xmath387 if and only if @xmath391\in g_0(\l\gb^f)$ ] is uniform . the statement follows from corollary [ coro : engendrement ] and from thomason s classification of full triangulated dense subcategories ( * ? ? ? * theorem 2.1 ) . note that corollary [ cor : uniform ] holds also for @xmath184 : in the proof , corollary [ coro : engendrement ] is replaced by theorem [ theo : a - engendrement ] . [ ex : elementaryabelian ] ( 1 ) if @xmath392 or @xmath393 , then all abelian subgroups consisting of semisimple elements are contained in maximal tori . this just amounts to the classical result in linear algebra which says that a family of commuting semisimple elements always admits a basis of common eigenvectors . \(2 ) assume @xmath0 is connected . let @xmath394 denote the set of prime numbers which are bad for @xmath0 or divide @xmath395 . if @xmath396 and if @xmath397 is an @xmath2-element of @xmath9 , then @xmath398 is a levi subgroup of @xmath0 and @xmath399 ( * ? ? ? * proposition 13.12(iii ) ) . an induction argument shows the following fact . [ eq : bons - bons ] so corollary [ coro : engendrement ] can be applied if @xmath396 . this generalizes ( 1 ) . assume here , and only here , that @xmath400 ( so that @xmath401 ) and that @xmath402 . let @xmath397 ( respectively @xmath403 ) denote the class of the matrix @xmath404 ( respectively @xmath405 ) in @xmath0 . then @xmath406 is an elementary abelian @xmath407-subgroup of @xmath0 which is not contained in any maximal torus of @xmath0 ( indeed , since @xmath0 has rank @xmath357 , all finite subgroups of maximal tori of @xmath0 are cyclic ) . we assume in this subsection [ se : rationalseries ] that @xmath0 is connected . [ se : rationalseries ] let @xmath180 be a positive integer divisible by @xmath408 and such that @xmath409 for all @xmath410 and @xmath411 . let @xmath412 be a generator of @xmath413 . recall ( * ? ? ? * proposition 13.7 ) that the map @xmath414 is surjective and it induces an isomorphism @xmath415 . the morphism @xmath416 factors through @xmath417 and induces a morphism @xmath418 . the corresponding morphism @xmath419 is surjective and induces an isomorphism @xmath420 ( * ? ? ? * proposition 13.7 ) . let @xmath421 be a triple dual to @xmath422 ( * ? ? ? * definition 5.21 ) . the isomorphisms @xmath420 and @xmath423 induce an isomorphism @xmath424 . let @xmath425 and let @xmath426 ( respectively @xmath427 ) denote the root ( respectively coroot ) system of @xmath0 relative to @xmath63 . we set @xmath428 and @xmath429 note that @xmath430 is closed and symmetric , hence it defines a root system . we denote by @xmath431 its weyl group . it is a subgroup of the weyl group @xmath432 and it is contained in the stabilizer @xmath433 of @xmath434 . this can be translated as follows in the dual group ( * ? ? ? * proposition 2.3 ) . let @xmath435 be the element corresponding to @xmath436 . identifying the coroot system @xmath427 with the root system of @xmath19 , we obtain that @xmath437 is the root system of @xmath27 . if @xmath438 is a unipotent subgroup of @xmath19 normalized by @xmath439 , then @xmath440 is generated by the one - parameter subgroups of @xmath19 normalized by @xmath439 , contained in @xmath438 , and corresponding to elements of @xmath430 . the group @xmath441 is identified with the weyl group @xmath442 of @xmath443 relative to @xmath439 while @xmath433 is identified with the weyl group @xmath444 of @xmath27 . recall that @xmath445 and @xmath446 are in the same _ geometric series _ if there exists @xmath447 such that @xmath448 and @xmath449 . the pairs are in the same _ rational series _ if in addition the element @xmath450 corresponding to @xmath451 is @xmath24-conjugate to @xmath452 . we have now a direct description of rational series . [ prop : rational - series ] the pairs @xmath445 and @xmath446 are in the same rational series if and only if there exists @xmath447 such that @xmath448 and @xmath453 . note that given @xmath454 such that @xmath455 is @xmath1-stable , then @xmath456 . let @xmath457 be an @xmath20-stable maximal torus of @xmath19 and let @xmath458 be such that the @xmath24-orbit of @xmath459 corresponds to the @xmath9-orbit of @xmath460 . then the statement of the proposition is equivalent to the following : * _ @xmath452 and @xmath461 are @xmath24-conjugate if and only if there exists @xmath462 such that @xmath463 and @xmath464 . _ so let us prove @xmath465 . first , if @xmath452 and @xmath461 are @xmath24-conjugate , then there exists @xmath466 such that @xmath467 . then @xmath468 and @xmath469 are two maximal tori of @xmath470 , so there exists @xmath471 such that @xmath472 . then @xmath473 and @xmath474 as desired . conversely , assume that there exists @xmath462 such that @xmath463 and @xmath475 . by lang s theorem applied to the connected group @xmath470 , there exists @xmath471 such that @xmath476 . then @xmath477 and @xmath478 . the proof of @xmath465 is complete . we can now translate the properties of regularity and super - regularity defined in @xcite . let @xmath12 be a parabolic subgroup of @xmath0 and let @xmath11 be a levi subgroup of @xmath12 . we assume that @xmath11 is @xmath1-stable . let @xmath479 be a rational series . [ prop : regularity ] the rational series @xmath480 is @xmath481-regular ( respectively @xmath481-super - regular ) if and only if @xmath482 ( respectively @xmath483 ) for some ( or any ) pair @xmath484 . this follows immediately from ( * ? ? ? * lemma 11.6 ) . we consider now again a non - necessarily connected reductive group @xmath0 . we fix an element @xmath485 which stabilizes a pair @xmath308 where @xmath60 is a borel subgroup of @xmath0 and @xmath63 is a maximal torus of @xmath60 . such an element is called _ quasi - semisimple _ in @xcite and @xcite . for instance , any semisimple element of @xmath0 is quasi - semisimple . recall from ( * ? ? ? * theorem 1.8 ) that @xmath486 is a reductive group , that @xmath487 is a borel subgroup of @xmath488 and that @xmath489 is a maximal torus of @xmath490 . we shall be interested in determining the coroot system of the fixed points subgroup @xmath486 . let @xmath426 ( respectively @xmath427 ) be the root ( respectively coroot ) system of @xmath167 relative to @xmath63 . let @xmath491 ( respectively @xmath492 ) denote the root ( respectively coroot ) system of @xmath486 relative to @xmath493 . if @xmath494 is a @xmath120-orbit in @xmath426 , we denote by @xmath495 the scalar by which @xmath496 acts on the one - parameter unipotent subgroup associated with @xmath497 ( through any identification of this one - parameter subgroup with the additive group @xmath155 ) . we denote by @xmath498 the set of @xmath120-orbits @xmath494 in @xmath426 such that there exist @xmath497 , @xmath499 such that @xmath500 . we denote by @xmath501 the set of other orbits . we set @xmath502=\{\o \in ( \phi / g)^a~\|~c_\o= 1 \text { and } p{\not=}2\ } \cup \{\o \in ( \phi / g)^b~\|~c_\o = 1\}.\ ] ] finally , if @xmath503 ( respectively @xmath504 ) , let @xmath505 ( respectively @xmath506 ) . note that @xmath507 is @xmath120-invariant , so it belongs to @xmath508 . [ prop : cooot ] @xmath509\}$ ] . the statement depends only on the automorphism induced by @xmath120 on @xmath167 and can be proved with assuming that @xmath167 is semisimple . since this automorphism can then be lifted uniquely to the simply - connected covering of @xmath167 ( see @xcite ) , we may also assume that @xmath167 is simply - connected . therefore , @xmath120 permutes the irreducible components of @xmath167 and an easy reduction argument shows that we may assume that @xmath167 is quasi - simple . let @xmath510 denote the unipotent radical of @xmath60 , @xmath511 the unipotent radical of the opposite borel subgroup and , if @xmath512 , let @xmath513 denote the corresponding one - parameter unipotent subgroup . we also denote by @xmath514 the subgroup generated by @xmath513 and @xmath515 : it is isomorphic to @xmath516 because @xmath167 is simply - connected . if @xmath517 , we denote by @xmath518 the unipotent subgroup generated by @xmath519 . we follow the proof of ( * ? ? ? * theorem 8.2 ) . according to this proof , any one - parameter unipotent subgroup @xmath13 normalized by @xmath493 is contained in some of these @xmath518 s , and one of the following holds : @xmath504 and @xmath520 . @xmath503 and @xmath521 . @xmath503 , @xmath520 and @xmath401 . in all cases , @xmath522 . let @xmath523 . in case ( 1 ) , as @xmath524=1 $ ] if @xmath497 , @xmath499 , the group @xmath525 is a direct product of groups isomorphic to @xmath516 which are permuted by @xmath120 . it then follows that the coroot corresponding to the one - parameter subgroup @xmath526 ( since @xmath520 ) is equal to @xmath527 . in cases ( 2 ) or ( 3 ) , it follows from the classification that @xmath528 ( this case only occurs in type @xmath529 ) . let @xmath530 . then @xmath531 . in case ( 2 ) , the computations done in ( * ? ? ? * proof of theorem 8.2 , item ( @xmath532 ) ) show that @xmath533 . therefore @xmath534 , where @xmath535 is the @xmath120-orbit ( of cardinality @xmath357 ) of @xmath536 and @xmath537 , and we are back to case ( 1 ) . in case ( 3 ) , the computations done in ( * ? ? ? * proof of theorem 8.2 , item ( @xmath532 ) ) show that @xmath538 and that the associated coroot is @xmath539 . [ rem : ordre pair ] if @xmath503 , then it follows from the classification that @xmath540 is even , and so the order of @xmath120 is even . let @xmath541 be a quasi - semisimple element of @xmath0 . let @xmath542 . we then set @xmath543 . it follows from ( * ? ? ? * theorem 1.8(iv ) ) that @xmath544 is a maximal torus of @xmath167 ( containing @xmath545 ) . it is stable under the action of @xmath120 , so we have a map @xmath546 , @xmath547 $ ] ( which is a morphism of groups because @xmath544 is abelian ) . if @xmath548 , then @xmath549 . in particular , if @xmath550 , then @xmath551 . this shows that any element of @xmath552 has order dividing the order of @xmath120 . note further that @xmath553 ( see ( * ? ? ? * theorem 1.8(iii ) ) ) . we have @xmath554 we deduce that [ eq : fini - ell ] now , if @xmath121 is a @xmath120-stable finite subgroup of @xmath544 of order prime to the order of @xmath120 then @xmath555 ( because the order of @xmath556 divides the order of @xmath120 by ( * ? ? ? * proposition 1.28 ) ) and [ eq : produit - h ] h = c_h(g ) _ so , if the linear character @xmath436 of @xmath557 has order prime to the order of @xmath120 , then it can be extended canonically to a linear character @xmath558 of @xmath559 as follows ( we use the discussion above with @xmath121 the subgroup of @xmath559 of elements with order dividing a power of the order of @xmath436 ) : @xmath558 is trivial on @xmath560 , is trivial on elements of @xmath559 of order prime to the order of @xmath436 and coincides with @xmath436 on @xmath557 . the fact that @xmath558 is trivial on @xmath560 is equivalent to [ eq : theta - plus - trivial ] note that , since @xmath561 by ( * ? ? ? * theorem 1.8 ) , we may identify the weyl group of @xmath562 relative to @xmath545 to a subgroup of the weyl group of @xmath167 relative to @xmath544 . through this identification , we get : [ lem : w0 ] if the order of @xmath436 is prime to the order of @xmath120 , then @xmath563 and @xmath564 . let @xmath565 . then @xmath566 stabilizes @xmath543 and its action on @xmath545 commutes with the action of @xmath120 . so it follows from the construction of @xmath558 that @xmath566 stabilizes @xmath558 . let us now prove the second statement . let @xmath567 be a coroot of @xmath562 relative to @xmath545 such that @xmath568 . let @xmath569 denote the corresponding reflection in @xmath570 . it is sufficient to prove that @xmath571 . then it follows from proposition [ prop : cooot ] that there exists a coroot @xmath572 of @xmath167 relative to @xmath544 and @xmath573 such that @xmath574 where @xmath575 is minimal such that @xmath576 . it follows from remark [ rem : ordre pair ] that , if @xmath577 , then @xmath120 has even order . now , @xmath578 because @xmath558 is @xmath120-stable . since @xmath279 and @xmath579 divide the order of @xmath120 , @xmath580 is prime to the order of @xmath558 , so this implies that @xmath581 . in particular , @xmath582 it follows from ( * ? ? ? * proof of theorem 8.2 , statement ( @xmath583 ) ) that then @xmath569 belongs to the subgroup generated by @xmath584 , @xmath585 , , @xmath586 . let @xmath587 . we say that @xmath588 and @xmath589 are _ geometrically conjugate _ ( resp . in the same _ rational series _ ) if there is @xmath590 such that @xmath591 and @xmath589 are geometrically conjugate ( resp . in the same rational series ) for @xmath592 . we denote by @xmath593 the set of rational series . let @xmath135 be the subgroup of @xmath0 generated by @xmath120 and let @xmath594 be a subgroup of @xmath595 containing @xmath343 . [ coro : brauer - rational ] let @xmath596 . if @xmath597 and @xmath598 are geometrically conjugate in @xmath594 , then @xmath599 and @xmath600 are geometrically conjugate in @xmath0 . if @xmath597 and @xmath598 are in the same rational series of @xmath594 , then @xmath599 and @xmath600 are in the same rational series of @xmath0 . so , the injective map @xmath601 induces a map @xmath602 \(a ) if @xmath597 and @xmath598 are geometrically conjugate in @xmath603 then , by definition , there exists @xmath604 such that @xmath605 and @xmath606 ( as linear characters of @xmath607 ) . since @xmath608 commutes with @xmath120 , it sends @xmath609 to @xmath610 , so it is immediately checked that @xmath611 . the case of geometric conjugacy in @xmath594 and @xmath0 follows immediately . \(b ) if @xmath597 and @xmath598 are in the same rational series of @xmath562 , then , by proposition [ prop : rational - series ] , there exists @xmath604 such that @xmath612 , @xmath613 ( as linear characters of @xmath607 ) and @xmath614 . so the result follows from ( a ) and from propositions [ prop : rational - series ] and [ prop : cooot ] . the case of rational series in @xmath594 and @xmath0 follows immediately . let @xmath11 be an @xmath1-stable levi complement of a parabolic subgroup @xmath12 of @xmath0 containing @xmath120 . then @xmath615 is an @xmath1-stable levi complement of @xmath616 ( * ? ? ? * proposition 1.11 ) . [ coro : super - regular ] let @xmath617 be a rational series . if @xmath618 is @xmath619-regular ( respectively @xmath619-super regular ) , then @xmath480 is @xmath620-regular ( respectively @xmath620-super regular ) . this follows from proposition [ prop : regularity ] and lemma [ lem : w0 ] . the results above extend by induction to general nilpotent @xmath195-subgroups . let @xmath135 be a nilpotent subgroup of @xmath9 of order prime to @xmath84 . fix a sequence @xmath621 of normal subgroups of @xmath135 such that @xmath622 is cyclic for @xmath623 . let @xmath624 . the construction above defines a map @xmath625 that preserves rational and geometric series . fix @xmath626 . let @xmath627 . the kernel of the canonical map @xmath628 has order a divisor of a power of the order of @xmath629 , and so does its cokernel , since it is isomorphic to @xmath630 . since @xmath178 is trivial on @xmath631 , it factors through a character of @xmath632 that comes by restriction from a ( unique ) character @xmath633 of @xmath634 . we obtain a pair @xmath635 . this correspondence defines a bijection @xmath636 that preserves rational and geometric series . composing those bijections with the map in ( [ eq : mapquotient ] ) , we obtain a map @xmath637 and composing all those maps , we obtain a map @xmath638 finally , composing with the canonical map @xmath639 , we obtain a map @xmath640 that preserves rational and geometric series . note that this map depends not only on @xmath135 , but on the filtration @xmath641 . summarizing , we have the following proposition . [ pr : seriesq ] let @xmath135 be a nilpotent subgroup of @xmath9 of order prime to @xmath84 . fix a sequence @xmath621 of normal subgroups of @xmath135 such that @xmath622 is cyclic for @xmath623 . the constructions above define a map @xmath642 let @xmath11 be an @xmath1-stable levi complement of a parabolic subgroup @xmath12 of @xmath0 containing @xmath135 . let @xmath643 be a rational series . then * @xmath644 is an @xmath1-stable levi complement of @xmath645 * if @xmath646 is @xmath619-regular then @xmath480 is @xmath647-regular * if @xmath646 is @xmath619-super regular then @xmath480 is @xmath647-super regular . the map @xmath648 is actually independent of the choice of the filtration of @xmath135 , cf remark [ re : indepfiltration ] below . given @xmath649 , we denote by @xmath650 the block idempotent of @xmath651 not vanishing on @xmath178 . we have now a generalization of ( * ? ? ? * thorme a ) . given @xmath652 , let @xmath653 be the thick subcategory of @xmath654 generated by the complexes @xmath655 where @xmath177 runs over @xmath480 and @xmath60 runs over borel subgroups of @xmath167 containing @xmath63 . note that , by definition of rational series for non - connected groups , we obtain the same thick subcategory by taking instead the complexes @xmath656 where @xmath657 . [ th : blocksperf ] let @xmath652 . there is a ( unique ) central idempotent @xmath658 of @xmath659 such that @xmath660 . we have a decomposition in central orthogonal idempotents of @xmath659 @xmath661 note first that the theorem holds for @xmath167 by ( * ? ? ? * thorme a ) . let @xmath662 and let @xmath663 be a borel subgroup of @xmath167 containing @xmath664 for @xmath665 . by ( [ eq : ind - rlg ] ) and ( [ eq : res - rlg ] ) , we have @xmath666 the connected case of the theorem shows this is @xmath102 unless @xmath588 and @xmath667 are in the same rational series of @xmath668 for some @xmath397 . we have shown that the categories @xmath669 and @xmath670 are othogonal for @xmath671 . the theorem follows now from ( * ? ? ? * proposition 9.2 ) and theorem [ theo : a - engendrement ] . let @xmath672 . let @xmath673 be the thick subcategory of @xmath354 generated by the complexes of the form @xmath674 where * @xmath177 runs over @xmath480 * @xmath60 runs over borel subgroups of @xmath167 containing @xmath63 * @xmath135 is an @xmath2-subgroup of @xmath352 * and @xmath341 is a @xmath356-module , free of rank @xmath357 over @xmath10 . let @xmath675 be the full subcategory of @xmath676 consisting of modules whose indecomposable direct summands have a one - dimensional source and a vertex @xmath135 which normalizes a pair @xmath331 where @xmath63 is an @xmath1-stable maximal torus and @xmath60 a borel subgroup of @xmath0 . [ theo : engendrementseries ] let @xmath672 . we have @xmath677 . by theorem [ th : blocksperf ] , we have @xmath678 if @xmath679 . it follows that @xmath680 . since @xmath681 , the theorem follows from theorem [ theo : engendrement ] . the following result generalizes ( * ? ? ? * thorme 3.2 ) to non - cyclic @xmath2-subgroups and to disconnected groups ( needed to handle the non - cyclic case by induction ) . [ theo : brauer ] let @xmath135 be an @xmath2-subgroup of @xmath9 . the map @xmath648 ( cf proposition [ pr : seriesq ] ) is independent of the filtration of @xmath135 and we denote it by @xmath682 . let @xmath683 . we have @xmath684 assume first that @xmath135 is cyclic . let @xmath685 and let @xmath686 . let @xmath244 be a borel subgroup of @xmath343 containing @xmath545 . note that @xmath687 is not acyclic , because its class in @xmath688 is non - zero . we have @xmath689 . let @xmath690 and let @xmath60 be a @xmath135-stable borel subgroup of @xmath0 containing @xmath544 ( cf lemma [ lem : bijtori ] ) . we have @xmath691 similarly , @xmath692 it follows that @xmath693 . since @xmath694 , we deduce that @xmath695 . by transitivity of @xmath696 , we obtain the formula for @xmath696 for a general @xmath135 by induction on @xmath228 , with @xmath697 replaced by @xmath698 . this shows that actually @xmath698 is independent of the chosen filtration of @xmath135 . [ re : indepfiltration ] let @xmath699 be a product of two cyclic groups of coprime orders . fix a filtration @xmath700 and @xmath701 . we have @xmath702 . it is easy to deduce now from theorem [ theo : brauer ] that @xmath698 is independent of @xmath135 for any nilpotent @xmath195-group @xmath135 . brou - michel s proof of theorem [ theo : brauer ] for @xmath0 connected and @xmath135 cyclic relies on the compatibility of deligne - lusztig induction with generalized decomposition maps . this does generalize to disconnected groups , as we explain below . a direct approach along the lines of brou - michel is possible , based on the results of @xcite . while we will not use the results in the remaining part of this section , they might be useful for character theoretic questions . let @xmath703 be a set of prime numbers not containing @xmath84 . an element of finite order of @xmath0 is a @xmath703-element ( resp . a @xmath704-element ) if its order is a product of primes in @xmath703 ( resp . not in @xmath703 ) . let @xmath120 be an automorphism of finite order of an algebrac variety @xmath157 . write @xmath705 where @xmath208 is a @xmath703-element and @xmath608 a @xmath704-element . the following result is an immediate consequence of ( * ? ? ? * theorem 3.2 ) : [ eq : dl ] _ i 0 ( -1)^i ( g , _ c^i ( , ) ) = _ i 0 ( -1)^i ( x,_c^i(^l , ) ) . write @xmath706 , where @xmath32 has order prime to @xmath84 and @xmath707 has order a power of @xmath84 . then @xmath208 , @xmath32 and @xmath707 commute and have coprime orders . by * theorem 3.2 ) , we have @xmath708 @xmath709 so the result follows from the fact that @xmath710 because @xmath711 . given @xmath121 a finite group and @xmath126 a @xmath703-element , we have a generalized decomposition map from the vector space of class functions @xmath712 to the vector space of class functions on @xmath704-elements of @xmath713 given by @xmath714 for @xmath707 a @xmath704-element of @xmath713 . the following result generalizes the character formula for @xmath715 ( * ? ? ? * proposition 2.6 ) , which corresponds to the case where @xmath703 is the set of all primes distinct from @xmath84 , [ prop : dec - rlg ] let @xmath12 be a parabolic subgroup of @xmath0 , let @xmath13 be its unipotent radical , let @xmath11 be a levi complement of @xmath12 and assume that @xmath11 is @xmath1-stable . let @xmath541 be a @xmath703-element . we have @xmath716 given @xmath121 a finite group , we denote by @xmath717 ( respectively @xmath718 ) the set of @xmath703-elements ( resp . @xmath704-elements ) of @xmath121 . the proof follows essentially the same argument as the proof of the character formula ( see for instance ( * ? ? ? * proposition 12.2 ) ) . let @xmath719 be a class function on @xmath16 and let @xmath720 be a @xmath704-element . by definition of the deligne - lusztig induction and by using ( [ eq : dl ] ) , we get _ ^()(gu ) & = & + & = & + but it follows from lemma [ lem : delta ] that @xmath721 if and only if there exists @xmath722 such that @xmath723 . moreover , in this case , then @xmath724 by proposition [ prop : centralisateur ] . therefore , _ ^()(gu ) & = & + & = & now , if @xmath722 is such that @xmath725 , then @xmath726 so the result follows . from now on , and until the end of this article , we assume @xmath0 is connected . deligne - lusztig varieties can be associated to sequences of elements of @xmath728 , and there is a canonical isomorphism @xmath729 when @xmath730 . we will show in this section that while such an isomorphism fails when @xmath731 , its consequence on cohomology remains true for local systems associated to characters of tori satisfying certain regularity conditions with respect to @xmath732 . in this section we will prove the preliminary statements necessary for our proof of theorem [ thd ] . roughly speaking , the main result of this section ( theorem [ theo : theo - d - borel ] ) is almost equivalent to theorem [ thd ] whenever @xmath11 is a maximal torus . as theorem [ thd ] will be proved by reduction to this case , theorem [ theo : theo - d - borel ] may be seen as the crucial step . in this section [ se : comparing ] , we fix an @xmath1-stable maximal torus @xmath63 contained in an @xmath1-stable borel subgroup @xmath60 and we denote by @xmath510 its unipotent radical . we put @xmath733 . we denote by @xmath426 the associated root system , by @xmath734 the set of positive roots and by @xmath735 the basis of @xmath426 . let @xmath736 , we denote by @xmath737 the corresponding reflection and by @xmath738 the corresponding coroot . we put @xmath739 and we denote by @xmath740 the one - parameter subgroup of @xmath0 normalized by @xmath63 and associated with @xmath741 . we define @xmath742 as the subgroup of @xmath0 generated by @xmath740 and @xmath743 . we fix in this section four parabolic subgroups @xmath73 , @xmath74 , @xmath744 and @xmath745 admitting a common levi complement @xmath11 . we denote by @xmath746 , @xmath747 , @xmath748 and @xmath749 the unipotent radicals of @xmath73 , @xmath74 , @xmath744 and @xmath745 respectively . we define the varieties @xmath750 @xmath751 @xmath752 we denote by @xmath753 the closed immersion and we define @xmath754 all these varieties are endowed with a diagonal action of @xmath0 , and the morphisms @xmath755 and @xmath756 are @xmath0-equivariant . [ prop : dim ] we have : @xmath757 . @xmath758 . @xmath759 . \(a ) is well - known . also , ( _ 1,2,3)&=&(/_1 ) + ( _ 1 _ 2/_2 ) + ( _ 2 _ 3 / _ 3 ) + & = & ( /_1 ) + ( _ 1)-(_1 _ 2 ) + ( _ 2)-(_2 _ 3 ) while ( _ 1,3)&=&(/_1 ) + ( _ 1 _ 3/_3 ) + & = & ( /_1 ) + ( _ 1)-(_1 _ 3 ) . so ( b ) follows from the two equalities ( and from ( a ) ) . let us now prove ( c ) . for this , we may assume that @xmath760 . let @xmath761 denote the set of roots @xmath512 such that @xmath762 . then @xmath763 . in particular , @xmath764 therefore @xmath765 on the other hand , by general facts about the cardinality of a union of finite sets , @xmath766 hence ( c ) follows from ( a ) , ( b ) and from these last two equalities . let @xmath767 . by proposition [ prop : dim ] , we have @xmath768 let @xmath769 @xmath770 both maps are @xmath0-equivariant morphisms of varieties . [ prop : iso ] the maps @xmath771 and @xmath772 are isomorphisms of varieties . the fact that @xmath771 is an isomorphism is clear . it is also clear that @xmath772 is a closed immersion . it is so sufficient to prove that @xmath772 is surjective . so , let @xmath773 . using the @xmath0-action and the fact that @xmath771 is an isomorphism , we may assume that @xmath774 . therefore , @xmath775 given @xmath776 , the multiplication map @xmath777 is an isomorphism of varieties , since @xmath778 and @xmath747 have a common levi complement . here , @xmath779 denotes the unipotent radical of the parabolic subgroup opposite to @xmath74 . it follows that @xmath780 so there exists @xmath781 such that @xmath782 . it is then clear that @xmath783 , as desired . [ coro : dim - closed ] the map @xmath756 is a smooth morphism with fibers isomorphic to the affine space of dimension @xmath784 . moreover , @xmath785 using the isomorphisms @xmath771 and @xmath772 of proposition [ prop : iso ] , the map @xmath756 may be identified with the canonical projection @xmath786 . the corollary follows . let us now define _ 1,2,3,4^&=&\{(g_1_1,g_2_2,g_3_3,g_4_4 ) /_1 /_2 /_3 /_4 \| + & & g_1 ^ -1g_2 _ 1 _ 2 , g_2 ^ -1g_3 _ 2 _ 3 , g_3 ^ -1g_4 _ 3 _ 4 + & & g_1 ^ -1g_4 _ 1 _ 4 } , @xmath787 @xmath788 then : [ coro : egalite - y ] assume that at least one of the following holds : @xmath789 . . @xmath791 . @xmath792 . then @xmath793 . using the fact that the map @xmath794 induces an isomorphism @xmath795 ( with obvious notation ) , we see that it is sufficient to prove only one of the statements . so let us assume that @xmath790 . let @xmath796 . then @xmath797 and so @xmath798 . so it remains to prove that @xmath799 . using the action of @xmath0 and the isomorphism @xmath800 of proposition [ prop : iso ] , we may assume that @xmath801 . since @xmath802 , we have @xmath803 , hence @xmath804 , as desired . [ rem : borel - inclusion ] let @xmath805 , @xmath806 and @xmath807 be three elements of @xmath728 and let us assume here that @xmath808 , @xmath809 , @xmath810 and @xmath811 . then the conditions ( 1 ) , ( 2 ) , ( 3 ) and ( 4 ) of corollary [ coro : egalite - y ] are respectively equivalent to : @xmath812 . @xmath813 . @xmath814 . @xmath815 . we fix a positive integer @xmath579 . given a family of objects @xmath816 belonging to a structure acted on by @xmath1 , we put @xmath817 for @xmath818 and @xmath819 . let @xmath820 be a sequence of elements of @xmath821 . we denote by @xmath822 the image of @xmath823 in @xmath728 and we put @xmath824 . we define @xmath825 where @xmath826 means @xmath827 . this variety has a left action by multiplication of @xmath9 and a right action of @xmath828 where @xmath829 acts by right multiplication by @xmath830 . we define the functor @xmath831 and we put @xmath832:g_0(\lambda\tb^{wf})\to g_0(\lambda\gb^f)$ ] as in @xcite . we fix a positive integer @xmath279 such that @xmath833 for all @xmath333 . the action of @xmath834 on @xmath835 restricts to an action on @xmath836 . given @xmath837 a variety of pure dimension @xmath107 , we put @xmath838(n/2)$ ] , where @xmath839 denotes a tate twist . given @xmath840 , we denote by @xmath841 the @xmath834-stable closed subvariety of @xmath836 defined by @xmath842 and we denote by @xmath843 its open complement . they are both stable under the action of @xmath844 . we denote by @xmath845 the morphism of varieties which forgets the @xmath846-th component and we set @xmath847 @xmath848 let @xmath849 denote the closed immersion and @xmath850 denote the restriction of @xmath851 . note that @xmath852 is @xmath853-equivariant and commutes with the action of @xmath834 . [ lem : fibration ] if @xmath840 , then @xmath852 is a smooth morphism with fibers isomorphic to an affine space of dimension @xmath854 . moreover , the codimension of @xmath841 in @xmath836 is also equal to @xmath854 . let @xmath855 , @xmath808 , @xmath856 and @xmath857 . there is a cartesian square @xmath858_{\pi_{\nb , j } } \ar[rrrrrr]^{(g_1\ub,\ldots , g_r\ub)\mapsto ( g_{j-1}\vb_1,g_jn_{j-1}^{-1}\vb_2,g_{j+1}n_j^{-1}n_{j-1}^{-1}\vb_3 ) } & & & & & & \ycb_{\ ! 1,2,3}^\closed\ar[d]^{\pi_{1,3 } } \\ \yb(c_j(\nb ) ) \ar[rrrrrr]_{(h_1\ub,\ldots , h_{r-1}\ub)\mapsto ( g_{j-1}\vb_1,g_{j+1}n_j^{-1}n_{j-1}^{-1}\vb_3 ) } & & & & & & \ycb_{\ ! 1,3 } } \ ] ] the lemma follows from now from corollary [ coro : dim - closed ] by base change . the map @xmath852 induces a quasi - isomorphism of complexes of @xmath859-bimodules [ eq : iso - fibre ] ( _ j^(), ) ( ( c_j()),)[-2 d_j()](-d_j ( ) ) . composing this isomorphism with the morphism @xmath860 , we obtain a morphism of complexes of @xmath859-bimodules @xmath861 which commutes with the action of @xmath834 , and whose cone is quasi - isomorphic to @xmath862 $ ] . we first recall some results from @xcite . we denote by @xmath109 the braid group of @xmath728 , and by @xmath863 the unique map ( not a group morphism ) that is a right inverse to the canonical map @xmath864 and that preserves lengths . we extend it to sequences of elements of @xmath728 by @xmath865 . we denote by @xmath866 the quotient map . we fix @xmath867 a map ( not a group morphism ) such that @xmath868 for all @xmath410 and such that the image of @xmath869 in @xmath870 is equal to @xmath871 . we extend it to sequences of elements of @xmath821 by @xmath872 . the following result is ( * ? ? ? * proposition 5.4 ) . [ lem : tresse - tore ] let @xmath873 be a sequence of elements of @xmath821 . then : if @xmath874 ( they are elements of @xmath875 ) , then the varieties @xmath836 and @xmath876 are canonically isomorphic @xmath9-varieties-@xmath828 . if @xmath877 ( they are elements of @xmath109 ) , then the varieties @xmath836 and @xmath876 are ( non - canonically ) isomorphic @xmath9-varieties-@xmath828 . \(a ) is proved in @xcite , while ( b ) is ( * ? ? ? * proposition 5.4 ) . using lemma [ lem : tresse - tore](a ) , we shall now write @xmath878 when @xmath874 . strictly speaking , lemma [ lem : tresse - tore](a ) says that these two varieties are only isomorphic but , since this isomorphism is canonical , we shall use the symbol @xmath879 to simplify the exposition . we define the _ cyclic shift _ @xmath880 of @xmath881 by @xmath882 the next result is proved in ( * ? ? ? * proposition 3.1.6 ) for the varieties @xmath883 and @xmath884 . the same proof shows the more precise result below . [ lem : facile ] the map @xmath885 induces an equivalence of tale sites . moreover , it is a morphism of @xmath9-varieties-@xmath886 , where @xmath887 acts on @xmath888 by right multiplication by @xmath889 . consequently , the diagram @xmath890 is commutative . assume in the remaining part of [ se : prelim ] that @xmath891 ( in particular , @xmath892 ) . note that @xmath893 . consider the diagram [ eq : transitivite ] ( ( ) , ) ^ _ & & ( ( c_j()), ) ^ + & & + ( ( c_j-1()), ) _ & & ( ( c_j-1(c_j()), ) . it does not seem reasonable to expect that the diagram ( [ eq : transitivite ] ) is commutative in general . however , it is in some cases . let us first define the following two varieties : @xmath894 @xmath895 more concretely , they are the closed subvarieties of @xmath836 defined by @xmath896 @xmath897 [ le : trans ] if @xmath898 , then the diagram ( [ eq : transitivite ] ) is commutative . there is a commutative diagram , in which all the arrows of the form @xmath899 are closed immersions and all the arrows of the form @xmath900 are smooth morphisms with fibers isomorphic to an affine space : @xmath901^{\ds{i_{\nb , j-1 } } } \ar@{->>}[ddd]_{\ds{\pi_{\nb , j-1 } } } & & & & & & \yb_{j}^\closed(\nb ) \ar@{_{(}->}[ulll]_{\ds{i_{\nb , j } } } \ar@{->>}[ddd]^{\ds{\pi_{\nb , j}}}\\ & & \yb_{j-1,j}^\closed(\nb ) \ar@{_{(}->}[ull]_{\ds{i } } \ar@{->>}[ddd]_{\ds{\pi } } \ar@{=}[rr ] & & \yb_{j , j-1}^\closed(\nb ) \ar@{^{(}->}[urr]^{\ds{i ' } } \ar@{->>}[ddd]^{\ds{\pi ' } } & & \\ & \blacksquare & & & & \blacksquare&\\ \yb(c_{j-1}(\nb ) ) & & & & & & \yb(c_j(\nb ) ) \\ & & \yb_{j-1}^\closed(c_{j-1}(\nb ) ) \ar@{_{(}->}[ull]_{\ds{i_{c_{j-1}(\nb),j-1 } } } \ar@{->>}[dr]_{\ds{\pi_{c_{j-1}(\nb),j-1 } } } & & \yb_{j-1}^\closed(c_{j}(\nb ) ) \ar@{^{(}->}[urr]^{\ds{i_{c_{j}(\nb),j-1 } } } \ar@{->>}[dl]^{\ds{\pi_{c_{j}(\nb),j-1 } } } & & \\ & & & \yb(c_{j-1}(c_j(\nb ) ) ) & & & \enddiagram\ ] ] note that the two squares marked with the [email protected] cm are cartesian by definition . by the proper base change theorem , the composition @xmath903 is obtained as the composition of @xmath904 with the inverse of the isomorphism induced by @xmath905 . similarly , the composition @xmath906 is equal to the composition of @xmath907 with the inverse of the isomorphism induced by @xmath908 . the lemma follows . [ lem : transitivite ] assume that one of the following holds : @xmath909 . @xmath910 . @xmath911 . . then the diagram ( [ eq : transitivite ] ) is commutative . it is sufficient , by lemma [ le : trans ] , to prove that , if ( 1 ) , ( 2 ) , ( 3 ) or ( 4 ) holds , then @xmath898 . this follows , after base change , from corollary [ coro : egalite - y ] . we start with the description of varieties of the form @xmath913 in a very special case , which will be the fundamental step in the proof of theorem [ theo : theo - d - borel ] . let @xmath914 . given @xmath915 , we define a subgroup of @xmath916 @xmath917 let @xmath918 . the group morphism @xmath919 restricts to an embedding of @xmath920 in @xmath921 . given @xmath922 and @xmath923 two sequences , we denote the concatenation of the sequences by @xmath924 . [ lem : crucial ] let @xmath925 and let @xmath926 be a representative of @xmath927 in @xmath928 . we assume that @xmath929 . there exists a closed immersion @xmath930 and an action of @xmath931 on @xmath932 such that @xmath933 as @xmath9-varieties-@xmath828 . given @xmath934 , consider a reduced decomposition @xmath935 . we put @xmath936 . note that @xmath921 is isomorphic to the group @xmath937 defined in @xcite : @xmath938 the following computation in @xmath939 @xmath940 shows that the map @xmath941 is an isomorphism of varieties . let @xmath942 . let @xmath943 . we have @xmath944 if and only if @xmath945 furthermore , if @xmath944 , then @xmath946 is determined by @xmath947 and @xmath948 . therefore , one may forget the second coordinate in the definition of the variety @xmath932 and we get [ eq : iso - epaississement ] rcl _ 2^((,^-1 ) ) & & \{(g , g_1, ,g_r ) \| g^-1g_1 _ ^ + & & g_1 g_2 g_r f(g ) } . this description shows that the group @xmath931 acts on @xmath932 ( as the restriction of the action by right multiplication of @xmath916 on @xmath949 ) . also , as @xmath950 is closed in @xmath951 , the natural map @xmath930 is a closed immersion . we have embeddings @xmath952 and @xmath953 and @xmath954 ( see ( * ? ? ? * proposition 4.11 ) ) . the stabilizer of the closed subvariety @xmath955 under this action is @xmath956 , so it is readily checked that the componentwise multiplication induces an isomorphism of @xmath9-varieties-@xmath886 @xmath957 as desired . the next theorem is the main result of this section . it provides a sufficient condition for @xmath958 to induce a quasi - isomorphism @xmath959 . given @xmath960 , we put @xmath961 we define @xmath962 ( cf [ se : rationalseries ] ) . [ theo : theo - d - borel ] let @xmath963 be a character . let @xmath964 and assume that @xmath965 for all @xmath966 . we have @xmath967 and @xmath968 is a quasi - isomorphism of complexes of @xmath859-bimodules commuting with the action of @xmath834 . if @xmath840 , we denote by @xmath969 the following property : @xmath970 we want to prove that @xmath969 implies that @xmath971 . by ( * proposition 5.19 and remark 5.21 ) , it is sufficient to prove it whenever @xmath972 $ ] is simply connected , and we will assume this holds . so assume from now that @xmath969 holds . we will prove by induction on @xmath973 that @xmath971 . note that the induction hypothesis does not depend on @xmath579 . but , first , note that , if @xmath974 , then @xmath969 is equivalent to @xmath975 and that the morphism constructed in lemma [ lem : facile ] sends @xmath843 to @xmath976 . thus @xmath971 is equivalent to @xmath977 . by successive applications of this remark , this shows that we may assume that @xmath978 . _ first case : assume that @xmath979 . _ this means that @xmath980 and it follows from lemma [ lem : fibration ] ( or lemma [ lem : tresse - tore](a ) ) that @xmath981 . so the result follows in this case . _ second case : assume that @xmath982 and @xmath983 . _ let @xmath925 be such that @xmath984 . by lemma [ lem : tresse - tore ] , we may assume that @xmath985 is a representative of @xmath927 lying in @xmath986 . note also that , since @xmath972 $ ] is simply connected , we have @xmath929 . define @xmath987 . lemma [ lem : crucial ] shows that @xmath988 but @xmath989 , so @xmath990 by hypothesis . note also that @xmath991 acts trivially on the cohomology groups of the complex @xmath992 , as its action extends to the connected group @xmath993 . since @xmath994 ( see ( * ? ? ? * proof of proposition 4.11 , equality ( a ) ) ) , this proves that @xmath995 and so @xmath996 , as desired . _ last case : assume that @xmath997 . _ let @xmath925 be such that @xmath998 , with @xmath999 . let @xmath926 be a representative of @xmath927 in @xmath986 and let @xmath1000 . we will write @xmath1001 . then @xmath1002 is a representative of @xmath1003 and , by lemma [ lem : tresse - tore](a ) , we have @xmath1004 ( see also the remark following lemma [ lem : tresse - tore ] ) . it is well - known that @xmath1005 . therefore @xmath1006 hence @xmath1007 let us now consider the diagram ( [ diag ] ) with @xmath881 replaced by @xmath1008 and @xmath846 is replaced by @xmath1009 . since @xmath1010 , it follows from lemma [ lem : transitivite](1 ) that ( [ eq : transitivite ] ) gives a commutative diagram @xmath1011\ar[d ] & \operatorname{{\mathrm{r}\gamma_{\!\ ! c}^{\mathrm{dim}}}}(\yb(c_3(\sdo\bullet\nb')))e_\theta\ar[d ] \\ \operatorname{{\mathrm{r}\gamma_{\!\ ! c}^{\mathrm{dim}}}}(\yb(\nb))e_\theta\ar[r ] & \operatorname{{\mathrm{r}\gamma_{\!\ ! c}^{\mathrm{dim}}}}(\yb(c_2(\nb)))e_\theta } \ ] ] the left vertical map is an isomorphism since @xmath1012 . by @xmath1013 , we have @xmath1014 , hence property @xmath1015 is fulfilled . so , the top horizontal map is an isomorphism by induction . in order to show that the bottom horizontal map is an isomorphism , it remains to show that the right vertical map is an isomorphism . note that @xmath1016 . two cases may occur : @xmath1017 assume first that @xmath1018 . then @xmath1019 , and the result follows . @xmath1017 assume now that @xmath1020 . then , again by lemma [ lem : tresse - tore](a ) , we have @xmath1021 and , through this identification , @xmath1022 is identified with @xmath1023 . so the result now follows from the _ second case _ ( thanks to @xmath1013 ) . [ re : frob ] theorem [ theo : theo - d - borel ] provides a comparison of modules , together with the frobenius action . consider the case @xmath184 . we have an isomorphism of @xmath1024-modules , compatible with the frobenius action @xmath1025 where @xmath1026 . following the same lines as in the proof of theorem [ theo : theo - d - borel ] , we obtain a new proof of the following classical result . [ theoe ] if @xmath10 is a field , then @xmath1027 . by ( * ? ? ? * proposition 5.19 and remark 5.21 ) , it is sufficient to prove the theorem whenever @xmath972 $ ] is simply connected , and we assume this holds . also , by proceeding step - by - step , it is enough to prove that @xmath1028 . for this , let @xmath1029 denote the class of the complex @xmath1030 in @xmath1031 . we only need to prove that @xmath1032 . proceeding by induction on @xmath1033 as in the proof of theorem [ theo : theo - d - borel ] , and following the same strategy and arguments , we see that it is enough to prove theorem [ theoe ] whenever @xmath1034 , @xmath1035 , where @xmath926 is a representative in @xmath986 of @xmath927 ( for some @xmath925 ) . by lemma [ lem : crucial ] , it is sufficient to prove that the class @xmath1036 of the complex @xmath1037 in @xmath1038 is equal to @xmath102 . now , let @xmath1039 denote the subgroup of @xmath1040 consisting of pairs @xmath1041 such that @xmath1042 and let @xmath1043 denote the class of the complex @xmath1044 in @xmath1045 . then @xmath1046 . but the action of @xmath1039 on @xmath931 extends to an action of the connected group @xmath1047 , hence @xmath1039 acts trivially on the cohomology groups of @xmath1047 . since the euler characteristic of a torus is equal to @xmath102 , this gives @xmath1048 , and consequently @xmath1049 , as desired . [ coro : theoe - weyl ] let @xmath1050 be a sequence of elements of @xmath821 , let @xmath1051 and let @xmath1052 denote the image of @xmath1053 in @xmath728 . we assume that @xmath10 is a field and that @xmath1054 . then the diagram @xmath1055is commutative . let @xmath1056 . then , by lemma [ lem : facile ] , @xmath1057 but , by theorem [ theoe ] , @xmath1058 and @xmath1059 . the following result is a reformulation of corollary [ coro : theoe - weyl ] as in @xcite . [ coro : theoe ] let @xmath250 be an @xmath1-stable maximal torus of @xmath0 and let @xmath251 and @xmath1060 be two borel subgroups of @xmath0 containing @xmath250 . then @xmath1061 for all @xmath1062 . corollary [ coro : theoe ] is well - known . in ( * corollary 4.3 ) , this result is first proved `` geometrically '' for @xmath1063 ( * ? ? ? * theorem 1.6 ) by relating the varieties @xmath1064 and @xmath1065 , and extended to the general case using the _ character formula _ * theorem 4.2 ) . note that this result is then used in ( * ? ? ? * theorem 6.8 ) to deduce the _ mackey formula _ for deligne - lusztig induction functors . in @xcite , lusztig proposed another argument : the mackey formula is proved `` geometrically '' and _ a priori _ ( * ? ? ? * theorem 2.3 ) , and corollary [ coro : theoe ] follows ( * ? ? ? * corollary 2.4 ) . our argument relies neither on the mackey formula nor on the character formula : we lift deligne - lusztig s comparison of @xmath1064 and @xmath1065 to a relation between the varieties @xmath1066 and @xmath1067 ( here @xmath1068 and @xmath1069 are the unipotent radicals of @xmath251 and @xmath1060 respectively ) . some of the results in @xcite ( lemma 4.3 , proposition 4.5 and theorem 4.6 ) rely on a disjointness result used in ( * ? ? ? * line 16 of page 30 ) . this disjointness result was `` proved '' using the isomorphism in ( * ? ? ? * line 18 of page 30 ) : it has been pointed out to the attention of the authors by h. wang that this equality is false . however , wang provided a complete proof of this disjointness result ( * ? ? ? * proposition 3.4.3 ) , so ( * ? ? ? * lemma 4.3 , proposition 4.5 and theorem 4.6 ) remain valid . another proof of this disjointness result has been obtained independently by nguyen @xcite ( with slightly different methods ) . using a version of remark [ re : frob ] , wang and nguyen have been able to keep track of the frobenius eigenvalues . we assume in this section [ sec : proof ] that @xmath0 is connected . we fix an @xmath1-stable maximal torus @xmath63 of @xmath0 and we denote by @xmath421 a triple dual to @xmath422 . we fix a family of parabolic subgroups @xmath73 , @xmath74 , , @xmath1070 admitting @xmath11 as a levi complement . the identification of the root system of @xmath0 with the coroot system of @xmath19 allows us to define parabolic subgroups @xmath1071 , @xmath1072 , , @xmath1073 , admitting a common @xmath20-stable levi complement @xmath40 and such that @xmath40 and @xmath1074 and are dual to @xmath11 and @xmath1075 respectively . we denote by @xmath1076 and @xmath1077 the unipotent radicals of @xmath1075 and @xmath1074 respectively . we denote by @xmath1078 the sequence @xmath1079 . finally , we fix a semisimple element @xmath1080 whose order is invertible in @xmath10 . as announced in the introduction , the isomorphism of functors described in theorem [ thd ] is _ canonical_. so , before giving the proof , we will explain how it is realized . for this , let us define @xmath1081 where @xmath1082 and @xmath1083 . given @xmath840 , we set @xmath1084 it is a closed subvariety of @xmath1085 and we denote by @xmath1086 the closed immersion . let @xmath1087 denote its open complement . we define the sequence @xmath1088 as obtained from the sequence @xmath1078 by removing the @xmath846-th component . we then define @xmath1089 as the map which forgets the @xmath846-th component and we set @xmath1090 note that @xmath9 acts diagonally on @xmath1085 by left translation , that @xmath16 acts diagonally by right translation , and that this endows @xmath1085 with a structure of @xmath9-variety-@xmath16 . the varieties @xmath1091 and @xmath1092 are stable under these actions , and the morphisms @xmath1093 and @xmath1094 are equivariant . as for their analogues @xmath1095 and @xmath852 defined in [ sub : notation ] , we have the following properties , which follow from corollary [ coro : dim - closed ] by base change . [ lem : fibration - p ] the map @xmath1094 is smooth with fibers isomorphic to an affine space of dimension @xmath1096 . the codimension of @xmath1091 in @xmath1085 is also equal to @xmath1096 . we deduce that @xmath1094 induces a quasi - isomorphism of complexes of @xmath1097-bimodules @xmath1098(-d_j(\vb_\bullet)).\ ] ] the closed immersion @xmath1086 induces a morphism of complexes of @xmath1097-bimodules @xmath1099 which , composed with the previous isomorphism , induces a morphism @xmath1100 the main result of this section is the following theorem . we put @xmath1101 , where @xmath1102 is the rational series corresponding to the @xmath77-conjugacy class of @xmath32 . [ theo : theod ] let @xmath964 such that @xmath1103 . we have @xmath1104 hence @xmath1105 induces a quasi - isomorphism of complexes of @xmath1097-bimodules @xmath1106 the proof will proceed in two steps . we first prove the theorem when @xmath11 is a maximal torus : in fact , it will be shown that it is a consequence of theorem [ theo : theo - d - borel ] . we then use ( * ? ? ? * theorem a ) to deduce the general case from this particular one . _ first step : assume here that @xmath11 is a maximal torus . _ let @xmath1107 , , @xmath1108 be elements of @xmath0 such that @xmath1109 for all @xmath1110 ) . as usual , we set @xmath1111 . now , let @xmath1112 . it follows from the definition of the @xmath1113 s that @xmath1114 . we set @xmath820 . note that @xmath1115 . we denote by @xmath822 the image of @xmath823 in @xmath728 and we set @xmath1116 . it is then easily checked that the map @xmath1117 induces an isomorphism of varieties @xmath1118 which sends @xmath1119 to @xmath841 . moreover , conjugacy by @xmath1107 induces an isomorphism @xmath1120 and it is easily checked that the above isomorphism is @xmath1121-equivariant through this idenfication . now , to @xmath32 is associated a linear character of @xmath16 which , through the identification @xmath1120 , defines a linear character @xmath963 . by theorem [ theo : theo - d - borel ] , we only need to prove that condition @xmath1103 is equivalent to @xmath969 . so let us prove this last fact . the property @xmath969 can be rewritten as follows : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ property @xmath969 . _ if @xmath1122 is such that @xmath1123 and @xmath1124 , then @xmath1125 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath1126 . note that @xmath969 is equivalent to @xmath1127 by conjugating by @xmath1107 , and since @xmath1128 , we get that @xmath969 is equivalent to @xmath1129 , as desired . _ second step : the general case . _ let us now come back to the general case : we no longer assume that @xmath11 is a maximal torus . since @xmath1130 , and since @xmath1131 lives in the category generated by the complexes @xmath1132 , where @xmath251 runs over the set of borel subgroups of @xmath11 admitting an @xmath1-stable maximal torus @xmath250 whose dual torus contains @xmath32 ( see ( * ? ? ? * theorem a ) ) , it is sufficient to prove that @xmath1133 so let @xmath1134 be a pair as above . let @xmath1068 denote the unipotent radical of @xmath251 , let @xmath1135 be an @xmath20-stable maximal torus of @xmath40 , containing @xmath32 and dual to @xmath250 and let @xmath1136 be a borel subgroup of @xmath40 containing @xmath1135 and dual to @xmath251 . then @xcite @xmath1137 ( as @xmath9-varieties-@xmath1138 ) . here , we have set @xmath1139 . moreover , through this isomorphism , @xmath1140 is sent to @xmath1141 hence , by applying the first step of this proof , we only need to prove that @xmath1142 . since @xmath1143 and @xmath1144 both admit @xmath40 as a levi complement and @xmath1145 , it follows that @xmath1146 . on the other hand , @xmath1147 by assumption and this completes the proof . [ re : froblevi ] theorem [ theo : theod ] provides a comparison of modules , together with the frobenius action . we have an isomorphism of @xmath1097-bimodules compatible with the frobenius action @xmath1148 where @xmath1149 . let @xmath1150 . the map @xmath1151 is @xmath1121-equivariant and induces an equivalence of tale sites . therefore , it induces a quasi - isomorphism of complexes of bimodules @xmath1152 applying twice theorem [ theo : theod ] , we obtain the following result . [ coro : theod - suite ] let @xmath1153 and assume that @xmath1154 the map @xmath1155 is a quasi - isomorphism of complexes of @xmath1097-bimodules @xmath1156 in the case @xmath1157 , corollary [ coro : theod - suite ] becomes the following result . [ coro : theod ] assume @xmath1158 the map @xmath1159 is a quasi - isomorphism of complexes of @xmath1097-bimodules @xmath1160 as a consequence , we obtain a quasi - isomorphism of functors between @xmath1161 : \drm^b(\l\lb^f e_s^{\lb^f } ) \longto \drm^b(\l\gb^fe_s^{\gb^f})\ ] ] @xmath1162 : \drm^b(\l\lb^f e_s^{\lb^f } ) \longto \drm^b(\l\gb^fe_s^{\gb^f } ) . \leqno{\text{\it and}}\ ] ] the isomorphism of functors of corollary [ coro : theod ] comes with a tate twist . keeping track of this twist has important applications @xcite , @xcite . [ rem : condition ] let us make here some comments about the condition @xmath1163 note that if @xmath1164 , then condition @xmath1165 is satisfied . since @xmath1166 is connected , it follows that if @xmath1167 , then condition @xmath1165 is satisfied . [ exemple : trivial ] of course , condition @xmath1168 is fulfilled for all @xmath32 . gluing the quasi - isomorphisms obtained from corollary [ coro : theod ] , we get a quasi - isomorphism of complexes of bimodules @xmath1169 but , since @xmath1170 , it is readily checked that @xmath1171 . [ exemple : frob ] similarly , condition @xmath1172 is fulfilled for all @xmath32 . gluing the quasi - isomorphisms obtained from corollary [ coro : theod ] , we obtain a quasi - isomorphism of complexes of bimodules @xmath1173 but , since @xmath1174 , it is readily checked that @xmath1175 . [ rem : transitivite ] if @xmath1165 holds , we denote by @xmath1176 the quasi - isomorphism defined by @xmath1177 . assume moreover that @xmath1178 and @xmath1179 hold , so that the quasi - isomorphisms of complexes @xmath1180 and @xmath1181 are also well - defined . it is natural to ask the following _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ * question . * _ when does the equality @xmath1182 hold ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for instance , taking example [ exemple : trivial ] into account , when does the equality @xmath1183 hold ? we do not know the answer to this question , but we can just say that the equality does not always hold . indeed , if @xmath279 is minimal such that @xmath1184 , then the isomorphisms @xmath1185 , @xmath1186 , , @xmath1187 are well - defined and all coincide with the frobenius endomorphism @xmath1 ( see example [ exemple : frob ] ) , and so @xmath1188 ( see example [ exemple : trivial ] ) . let @xmath1189 be a parabolic subgroup admitting an @xmath1-stable levi subgroup @xmath1190 containing @xmath11 . we denote by @xmath1191 the unipotent radical of @xmath1189 and @xmath1192 the corresponding levi subgroup of a parabolic subgroup of @xmath19 containing @xmath40 , which is dual to @xmath1190 . we assume in this example that @xmath1193 . then it follows from ( * ? ? ? * theorem 11.7 ) , corollary [ coro : theod ] and remark [ rem : condition ] that we have an isomorphism of @xmath1097-bimodules @xmath1194 where @xmath1195 . let us consider the harish - chandra case : assume that @xmath746 and @xmath747 are @xmath1-stable . the functors @xmath1196 and @xmath1197 are isomorphic without truncating by any series @xcite . such isomorphisms are given by explicit isomorphisms of bimodules , which do not rely on any algebraic geometry . we do not know if after truncation by a series satisfying ( @xmath1198 ) , they coincide with our isomorphisms . we will provide here an analogue to lemma [ lem : transitivite ] in the more general context of this section . assume in this subsection , and only in this subsection , that @xmath891 ( in particular , @xmath892 ) . since @xmath1199 , we can build a diagram [ eq : transitivite - parab ] ( _ _ , ) ^ _ & & ( _ c_j(_), ) _ + & & + ( _ c_j-1(_), ) ^ & & ( _ c_j-1(c_j(_), ) . it does not seem reasonable to expect that the diagram ( [ eq : transitivite - parab ] ) is commutative in general . however , we have the following result , obtained from the results of section [ appendice : dimension ] below by copying the proof of lemma [ lem : transitivite ] . [ lem : transitivite - parab ] assume that one of the following holds : @xmath1200 . @xmath1201 . @xmath1202 . . then the diagram ( [ eq : transitivite - parab ] ) is commutative . in this section , we assume @xmath0 is connected . we fix an @xmath1-stable maximal torus @xmath63 of @xmath0 and we denote by @xmath421 a triple dual to @xmath422 . we start in [ se : quasiisolatedsetting ] with a recollection of some of the results of @xcite on the vanishing of the truncated cohomology of certain deligne - lusztig varieties outside the middle degree . we fix an @xmath1-stable levi subgroup @xmath11 and consider @xmath1204 of order invertible in @xmath8 such that @xmath47 ( and we take @xmath11 minimal with that property ) . we show that the corresponding middle degree @xmath1205-bimodule @xmath1206 does not depend on the choice of the parabolic subgroup @xmath12 , up to isomorphism , thanks to the results of [ sec : proof ] . in particular , it is stable under the action of the stabilizer @xmath52 of @xmath1207 in @xmath1208 . section [ se : clifford ] develops some clifford theory tools in order to extend the action of @xmath16 on @xmath1206 to an action of @xmath52 . we apply this in [ se : morita ] by embedding @xmath0 in a group @xmath1209 with connected center . this provides a morita equivalence , extending the main result of @xcite to the quasi - isolated case . in section [ se : splendid ] , we show that the action of @xmath16 on the complex of cohomology @xmath1210 also extends to @xmath52 , and the resulting complex provides a splendid rickard equivalence . this relies on checking that given @xmath135 an @xmath2-subgroup of @xmath16 , the complex @xmath1211 arises in a jordan decomposition setting for @xmath1212 , and then applying the results of the appendix . the main difficulty is to prove that @xmath1213 is a sum of idempotents associated to a jordan decomposition setting for @xmath1212 . an added difficulty is that the group @xmath1212 need not be connected . we fix a semisimple element @xmath1214 whose order is invertible in @xmath10 . let @xmath1215 , an @xmath20-stable levi complement of some parabolic subgroup @xmath1216 of @xmath19 . note that @xmath40 is a minimal levi subgroup with respect to the property of containing @xmath443 and @xmath1217 is an abelian @xmath36-group ( * ? ? ? * corollary 2.8(b ) ) . in particular , the series corresponding to @xmath32 is @xmath481-regular . we denote by @xmath1218 a pair dual to @xmath1219 . note that @xmath12 is a parabolic subgroup of @xmath0 admitting @xmath11 as an @xmath1-stable levi complement . the unipotent radical of @xmath12 will be denoted by @xmath13 . we put @xmath1220 . the group @xmath27 normalizes @xmath40 and we set @xmath1221 : it is a subgroup of @xmath1222 containing @xmath40 . via the canonical isomorphism between @xmath1223 and @xmath1224 , we define the subgroup @xmath594 of @xmath1225 containing @xmath11 such that @xmath1226 corresponds to @xmath1227 . note that @xmath1228 is @xmath20-stable and so @xmath594 is @xmath1-stable , and that @xmath1229 and @xmath1230 are abelian @xmath36-groups . let us first derive some consequences of these assumptions . note that @xmath1231 , so that @xmath1232 . also , @xmath1233 is the stabilizer , in @xmath1234 , of the @xmath77-conjugacy class of @xmath32 . therefore [ eq : stab ] it follows that @xmath1207 is a central idempotent of @xmath1235 . by ( * ? ? ? * theorem 11.7 ) , we have @xmath1236 our first result on the jordan decomposition is the independence of the choice of parabolic subgroups . [ th : indepjordan ] given @xmath1237 a parabolic subgroup of @xmath0 with levi complement @xmath11 and unipotent radical @xmath224 , then @xmath1238 as @xmath1205-bimodules . the @xmath1205-bimodule @xmath1239 is @xmath1240-stable . the first result follows from remark [ rem : condition ] and corollary [ coro : theod ] . let @xmath1241 . the isomorphism of varieties @xmath1242 induces an isomorphism of varieties @xmath1243 . as a consequence , we have an isomorphism of @xmath1097-bimodules @xmath1244 where @xmath1245 as a left @xmath306-module and the right action of @xmath1246 on @xmath1247 is given by the right action of @xmath1248 on @xmath1249 . since @xmath107 fixes @xmath1207 , we deduce that @xmath1250 on the other hand , the first part of the theorem shows that @xmath1251 it follows that @xmath1252 . recall that , if @xmath1253 ( that is , if @xmath1254 ) , then @xmath1255 induces a morita equivalence between @xmath1256 and @xmath1131 by ( * ? ? ? * theorem b ) . note that the assumption in ( * ? ? ? * theorem b ) is @xmath1257 , but it can easily be seen that the proof requires only the assumption @xmath1254 . theorem [ th : indepjordan ] shows that this morita equivalence does not depend on the choice of a parabolic subgroup . we will generalize the morita equivalence to our situation . the main difficulty is to extend the action of @xmath16 on @xmath1258 to @xmath1240 . let us recall some basic facts of clifford theory . let @xmath1259 be a field . let @xmath1260 be a finite group and @xmath1261 a normal subgroup of @xmath1260 . let @xmath64 be a finitely generated @xmath1262-module that is @xmath1260-stable and let @xmath1263 . given @xmath1264 , let @xmath1265 be the set of @xmath1266 such that @xmath1267 for all @xmath1268 and @xmath1269 . note that @xmath1270 for all @xmath1271 . let @xmath1272 , a subgroup of @xmath1273 containing @xmath1274 as a normal subgroup . the action of @xmath1268 on @xmath64 defines an element of @xmath1275 , and this gives a morphism @xmath1276 . the @xmath1260-stability of @xmath64 gives a surjective morphism of groups @xmath1277 . let @xmath1278 . there is a diagonal embedding of @xmath1261 as a normal subgroup of @xmath1279 . there is a commutative diagram whose horizontal and vertical sequences are exact : @xmath1280 & 1\ar[d ] \\ & & x\ar@{=}[r]\ar[d ] & x\ar[d ] \\ 1\ar[r ] & a^\times\ar[r]\ar@{=}[d ] & \hat{y}\ar[r]\ar[d ] & y\ar[r]\ar[d ] & 1 \\ 1\ar[r ] & a^\times\ar[r ] & \hat{y}/x\ar[r]\ar[d ] & y / x\ar[d]\ar[r ] & 1 \\ & & 1 & 1 } \ ] ] the action of @xmath1261 on @xmath64 extends to an action of @xmath1260 if and only if the canonical morphism of groups @xmath1281 has a splitting that is the identity on @xmath1261 . this is equivalent to the fact that the canonical morphism of groups @xmath1282 is a split surjection . the extension of groups @xmath1283 splits . indeed , since @xmath55 is a finite - dimensional @xmath1259-algebra , there exists a @xmath1259-subalgebra @xmath1284 of @xmath55 such that the composition @xmath1285 is an isomorphism . since @xmath1286 , we have @xmath1287 . if @xmath1288\in\kb^\times$ ] , then every group extension @xmath1289 splits , since @xmath1290 is the finite extension of abelian groups @xmath1291 and those are @xmath1292-modules . consequently , if @xmath1288\in\kb^\times$ ] , then the action of @xmath1261 on @xmath64 extends to an action of @xmath1260 if and only if the extension @xmath1293 splits . consider now @xmath1294 a finite group with @xmath1260 and @xmath1295 two normal subgroups such that @xmath1296 and @xmath1297 . let @xmath1298 , a @xmath1294-stable @xmath1299-module . we define @xmath1300 , @xmath1301 and @xmath1302 as above , replacing @xmath64 by @xmath1303 . given @xmath1264 , we define a map @xmath1304 for @xmath1305 and @xmath1269 . this gives a morphism of groups @xmath1306 extending the canonical morphism @xmath1307 and a morphism of groups @xmath1308 giving a commutative diagram @xmath1309 & a^\times\ar[r]\ar@{^{(}->}[d ] & \hat{y}/x \ar[r]\ar[d ] & y / x\ar[d]^\sim \ar[r ] & 1 \\ 1\ar[r ] & \operatorname{{\mathrm{end}}}_{\kb\tilde{x}}(\tilde{m})^\times\ar[r ] & \hat{\tilde{y}}/\tilde{x}\ar[r ] & \tilde{y}/\tilde{x}\ar[r ] & 1 } \ ] ] it induces a commutative diagram @xmath1309 & a^\times/(1+j(a))\ar[r]\ar@{^{(}->}[d ] & \hat{y}/x(1+j(a))\ar[r]\ar[d ] & y / x\ar[d]^\sim \ar[r ] & 1 \\ 1\ar[r ] & \operatorname{{\mathrm{end}}}_{\kb\tilde{x}}(\tilde{m})^\times/(1+j(\operatorname{{\mathrm{end}}}_{\kb\tilde{x}}(\tilde{m})))\ar[r ] & \hat{\tilde{y}}/\tilde{x}(1+j(\operatorname{{\mathrm{end}}}_{\kb\tilde{x}}(\tilde{m})))\ar[r ] & \tilde{y}/\tilde{x}\ar[r ] & 1 } \ ] ] assume the inclusion @xmath1310 splits ( this happens for example if @xmath1311 for some @xmath107 , for in that case the algebra embedding @xmath1312 has a section ) . if the surjection @xmath1313 splits , then the surjection @xmath1314 splits . as a consequence , we have the following proposition . [ pr : clifford ] let @xmath1294 be a finite group and @xmath1260 , @xmath1295 be two normal subgroups of @xmath1294 . let @xmath1296 . we assume @xmath1297 . let @xmath1259 be a field with @xmath1288\in\kb^\times$ ] . let @xmath64 be a finitely generated @xmath1262-module that is @xmath1260-stable . we assume that @xmath1315 if @xmath1316 extends to @xmath1294 , then @xmath64 extends to @xmath1260 . we fix a connected reductive algebraic group @xmath1209 containing @xmath0 as a closed subgroup , with an extension of @xmath1 to an endomorphism of @xmath1209 such that @xmath165 is a frobenius endomorphism of @xmath1209 defining an @xmath166-structure , and such that @xmath1317 and @xmath1318 is connected ( * ? ? ? * proof of corollary 5.18 ) . the inclusion @xmath1319 is called a _ regular embedding_. let @xmath1320 , an @xmath1-stable maximal torus of @xmath1209 . fix a triple @xmath1321 dual to @xmath1322 . the inclusion @xmath1323 induces a surjection @xmath1324 . let @xmath1325 , so that @xmath1326 . let @xmath1327 . let @xmath1328 be a set of representatives of conjugacy classes of @xmath36-elements @xmath1329 such that @xmath1330 ( recall that @xmath1331 is connected because @xmath1318 is connected ) . note that @xmath1332 . [ le : idemptilde ] we have @xmath1333 and @xmath1334 . the first statement is a classical translation from @xmath19 to @xmath0 , cf for instance ( * ? ? ? * proposition 11.7 ) . let @xmath1335 be a semisimple element of @xmath1336 such that @xmath1337 . if @xmath1338 , we will assume that @xmath1335 has order prime to @xmath2 ( this is always possible as we may replace @xmath1335 by its @xmath36-part if necessary ) . note that @xmath1339 . let @xmath1340 such that @xmath1341 is @xmath1342-conjugate to @xmath1335 . then @xmath1343 . since @xmath1344 , it follows that @xmath1345 . we have @xmath1346 , hence @xmath1347 . it follows that @xmath1229 acts freely on the set of conjugacy classes of @xmath1342 whose image under @xmath1348 is the @xmath77-conjugacy class of @xmath32 . through the identification of @xmath1229 with @xmath1230 , this shows that given @xmath1349 , the stabilizer in @xmath1240 of @xmath1350 is @xmath16 . [ th : moritaquasi ] the action of @xmath1351 on @xmath1352 extends to an action of @xmath1353 . the resulting @xmath1354-bimodule induces a morita equivalence . let @xmath1355 and let @xmath1356 . note that @xmath1357 ( resp . @xmath1358 ) is a levi complement of @xmath1359 ( resp . @xmath1360 ) and it is @xmath1-stable ( resp . @xmath20-stable ) and the pair @xmath1361 is dual to @xmath1362 . we put @xmath1363 @xmath1364 let @xmath1365 . through the embedding @xmath1366 , we identify @xmath1367 with a subvariety of @xmath1368 . the stabilizer in @xmath1295 of the subvariety @xmath1367 of @xmath1368 is @xmath1261 , hence we have an isomorphism of @xmath1295-varieties @xmath1369 . let @xmath1370 , a @xmath1371-module . let @xmath1372 , a @xmath1373-module . we have an isomorphism of @xmath1373-modules @xmath1374 . we put @xmath1375 . we have @xmath1376 and @xmath1377 is a central idempotent of @xmath1378 ( lemma [ le : idemptilde ] ) . the @xmath1379-module @xmath64 is @xmath1240-stable ( theorem [ th : indepjordan ] ) , hence the @xmath1380-module @xmath1303 is @xmath1240-stable as well . it follows that given @xmath1349 and @xmath1381 , we have @xmath1382 as @xmath1380-modules . the classical mackey formula for induction and restriction in finite groups shows now that @xmath1383 hence @xmath1384 lemma [ le : idemptilde ] shows that @xmath1385 induces a morita equivalence between @xmath1386 and @xmath1387 ( cf ( * ? ? ? * theorem b ) ) . in particular , it is a direct sum of indecomposable modules no two of which are isomorphic . since @xmath1388 induces a morita equivalence between @xmath1387 and @xmath1389 , we deduce that the right action of @xmath1390 on @xmath1391 extends to an action of @xmath1392 commuting with the left action of @xmath1393 and the extended bimodule @xmath1394 induces a morita equivalence between @xmath1386 and @xmath1389 . it follows that @xmath1395 given @xmath1396 with @xmath1397 , the central idempotents @xmath1398 and @xmath1399 of @xmath1378 are orthogonal . it follows that @xmath1400},\ ] ] the last isomorphism following from the fact that @xmath1401 induces a morita equivalence between @xmath1402 and @xmath1403 . we deduce that @xmath1404 for some @xmath579 . since @xmath1288=[\nb:\lb]$ ] is prime to @xmath2 , it follows from proposition [ pr : clifford ] that the action of @xmath1261 on @xmath64 extends to an action of @xmath1260 . denote by @xmath370 the extended module . we have @xmath1405 , hence @xmath1406 . it follows that @xmath1407 induces a morita equivalence between @xmath1386 and @xmath1389 . we have @xmath1408 the canonical map @xmath1409 is an isomorphism , hence the canonical map @xmath1410 is an isomorphism as well . also , @xmath64 is a faithful @xmath1411-module , since @xmath1412 is a faithful @xmath1386-module . we deduce that @xmath370 induces a morita equivalence between @xmath1413 and @xmath1414 . recall that @xmath11 is the minimal @xmath1-stable levi subgroup of @xmath0 such that @xmath47 . [ th : rickardk ] the action of @xmath1351 on @xmath1415 extends to an action of @xmath1353 . the resulting complex induces a splendid rickard equivalence between @xmath1413 and @xmath1414 . @xmath1416_step 1 _ : identification of @xmath1417 in @xmath1418 . let @xmath1419 . the vertices of the indecomposable direct summands of components of @xmath53 are contained in @xmath1420 by corollary [ cor : deltavertex ] . let @xmath135 be an @xmath2-subgroup of @xmath16 . we have @xmath1421 in @xmath1422 by proposition [ prop : centralisateur ] . let @xmath480 be the rational series of @xmath1423 corresponding to @xmath32 , so that @xmath1424 . theorem [ theo : brauer ] shows that @xmath1425 let @xmath1426 . proposition [ pr : seriesq ] shows that @xmath1427 is @xmath1428-regular . it follows from ( * ? ? ? * theorem 11.7 ) that @xmath1429 for @xmath1430 , hence @xmath1431 for @xmath1432 . we have shown that the cohomology of @xmath1433 is concentrated in a single degree . note that @xmath1434 is a perfect complex , hence its homology is projective as a @xmath1435-module . we deduce from theorem [ th : splitend ] that @xmath1436 @xmath1416_step 2 _ : study of @xmath1437 . let @xmath1438 . let @xmath259 be a projective resolution of @xmath1439 , i.e. , a complex of @xmath1440 with @xmath1441 for @xmath1442 , together with a quasi - isomorphism @xmath1443 of @xmath1444-modules . as the terms of @xmath104 are projective @xmath1445-modules , we have a commutative diagram @xmath1446 \ar[d]_\sim & \operatorname{{\mathrm{end}}}_{d^b(k(\gb^f\times(\nb^f)^\opp))}(c ' ) \ar[d]^\sim \\ \operatorname{{\mathrm{hom}}}_{\ho^b(k(\nb^f\times ( \nb^f)^\opp))}(k\nb^f,\operatorname{{\mathrm{end}}}_{k\gb^f}^\bullet(c ' ) ) \ar[r ] & \operatorname{{\mathrm{hom}}}_{\ho^b(k(\nb^f\times ( \nb^f)^\opp))}(p,\operatorname{{\mathrm{end}}}_{k\gb^f}^\bullet(c ' ) ) } \ ] ] using the isomorphisms of complexes in @xmath1447 @xmath1448 and @xmath1449 we deduce that @xmath1450 now , the canonical map @xmath1451 is an isomorphism , since @xmath1452 is a complex concentrated in degree @xmath102 . it follows that the top horizontal map in the commutative diagram above is an isomorphism , hence we have canonical isomorphisms @xmath1453 @xmath1416_step 3 _ : construction of a direct summand @xmath1454 of @xmath1455 . we have shown ( theorem [ th : moritaquasi ] ) that there is a direct summand @xmath370 of @xmath1456 whose restriction to @xmath1457 is isomorphic to @xmath1458 . let @xmath333 be the corresponding idempotent of @xmath1459 and @xmath846 its inverse image in @xmath1460 via the isomorphisms above . we have a surjective homomorphism of finite - dimensional @xmath7-algebras @xmath1461 consequently , @xmath846 lifts to an idempotent @xmath1462 of @xmath1463 ( * ? ? ? * theorem 3.2 ) . it corresponds to a direct summand @xmath1454 of @xmath104 quasi - isomorphic to @xmath370 and @xmath1464 is a direct summand of @xmath1465}$ ] . @xmath1416_step 4 _ : @xmath1454 lifts @xmath1466 . let @xmath1467 be a decomposition into a direct sum of indecomposable objects of @xmath1468 . this induces a decomposition @xmath1469 , where @xmath1470 and @xmath1471 and @xmath1472 have no isomorphic indecomposable summands for @xmath1473 ( cf proof of theorem [ th : moritaquasi ] ) . we have @xmath1474 in @xmath1468 for some integers @xmath1475 $ ] and @xmath1476 . it follows that @xmath1477 for all @xmath579 , hence @xmath1478 in @xmath1468 . this shows the first statement . @xmath1416_step 5 _ : rickard equivalence . we have shown above that @xmath1479 in @xmath1480 . on the other hand , @xmath1481 . it follows from corollary [ cor : splendid ] that @xmath1454 induces a splendid rickard equivalence . we now summarize and complete the description of the jordan decomposition of blocks . the complex of @xmath1482-bimodules @xmath1483 extends to a complex @xmath53 of @xmath1484-bimodules . the complex @xmath53 induces a splendid rickard equivalence between @xmath43 and @xmath1485 . there is a ( unique ) bijection @xmath1486 between blocks of @xmath43 and @xmath1485 such that @xmath1487 . given @xmath148 a block of @xmath43 , then : * the bimodule @xmath1488 induces a morita equivalence between @xmath1489 and @xmath1490 * the complex @xmath1491 induces a splendid rickard equivalence between @xmath1489 and @xmath1490 * there is a ( unique ) equivalence @xmath1492 from the category of @xmath1493-subpairs to the category of @xmath148-subpairs such that @xmath1494 . in particular , if @xmath1495 is a defect group of @xmath1493 , then @xmath1495 is a defect group of @xmath148 . theorem [ th : rickardk ] provides a complex @xmath104 of @xmath1496-modules inducing a splendid rickard equivalence . by rickard s lifting theorem ( * ? ? ? * theorem 5.2 ) , there is a splendid complex @xmath53 of @xmath1497-modules , unique up to isomorphism in @xmath1498 , such that @xmath1499 . also , ( * ? ? ? * proof of theorem 5.2 ) shows that @xmath1483 is the unique splendid complex that lifts @xmath1500 . as a consequence , there is an isomorphism of complexes @xmath1501 by ( * ? ? ? * theorem 5.2 ) , the complex @xmath53 induces a splendid rickard equivalence . since @xmath1502 induces a morita equivalence , it follows that @xmath1503 induces a morita equivalence ( cf e.g. ( * ? ? ? * proof of theorem 5.2 ) ) . the existence of the bijection between blocks follows from the isomorphism of algebras @xmath1504 induced by the morita equivalence , and the blockwise statements on morita and rickard equivalence are clear . by ( * ? ? ? * theorem 19.7 ) , it follows that the brauer categories of @xmath1505 and @xmath1506 are equivalent , and in particular , @xmath1505 and @xmath1507 have isomorphic defect groups . if was already known that given @xmath148 a block of @xmath43 , then @xmath148 and @xmath1493 have isomorphic defect groups under one of the following assumptions : * @xmath2 does nt divide @xmath1508 nor @xmath1509 , @xmath1510 and @xmath1511 if @xmath0 has a component of type @xmath1512 ( * ? ? ? * proposition 5.1 ) . * @xmath1513 and either @xmath148 or @xmath1493 has a defect group that is abelian modulo the @xmath2-center of @xmath9 ( * ? ? ? * theorem 1.3 ) . [ exemple : cgs - levi ] assume in this example that @xmath1514 and that @xmath1515 is cyclic . the element @xmath32 defines a linear character @xmath1516 which induces an isomorphism of algebra @xmath1517 . the linear character @xmath1518 is stable under the action of @xmath1240 so , since @xmath1230 is cyclic , it extends to a linear character @xmath1519 . again , @xmath1520 induces an isomorphism of algebra @xmath1521 . combined with this , theorem [ th : moritaquasi ] provides a morita equivalence between @xmath1522 and @xmath43 . [ exemple : type - a ] assume in this example that all the simple components of @xmath0 are of type @xmath55 ( no assumption is made on the action of @xmath1 ) . then @xmath1514 and @xmath1217 is cyclic . therefore , example [ exemple : cgs - levi ] can be applied to provide a morita equivalence between @xmath1522 and @xmath43 . this article was announced at the end of the introduction of @xcite . unfortunately , we have not been able to settle the problem of finiteness of source algebras . on the other hand , in addition to what was announced in @xcite , we have provided an extension of the jordan decomposition to the quasi - isolated case . in this section , we assume @xmath130 or @xmath131 . let us recall here some results of brou and puig , cf @xcite . let @xmath115 be a finite group . note that an @xmath2-permutation @xmath1523-module @xmath64 is indecomposable if and only if @xmath1524 is an indecomposable @xmath1525-module . let @xmath259 be an @xmath2-subgroup of @xmath115 . an indecomposable @xmath2-permutation @xmath132-module @xmath64 has a vertex containing @xmath259 if and only if @xmath1526 . also , given @xmath342 an indecomposable projective @xmath1527$]-module ( it is then an @xmath2-permutation @xmath1525-module ) , there exists a unique indecomposable @xmath2-permutation @xmath132-module @xmath1528 such that @xmath1529 . the @xmath132-module @xmath1528 has vertex @xmath259 . moreover , every indecomposable @xmath2-permutation @xmath132-module with vertex @xmath259 is isomorphic to such an @xmath1528 . [ lem : split ] let @xmath64 and @xmath52 be @xmath2-permutation @xmath132-modules and let @xmath1530 . assume that all indecomposable summands of @xmath52 have a vertex equal to a given subgroup @xmath259 of @xmath115 and that @xmath1531 is a surjection . then @xmath1532 is a split surjection . proceeding by induction on the dimension of @xmath52 , we can assume that @xmath52 is indecomposable . fix a decomposition @xmath1533 where @xmath1534 is indecomposable for all @xmath1535 and let @xmath1536 denote the restriction of @xmath1532 . since @xmath1531 is a surjection and @xmath1537 is an indecomposable projective @xmath1527$]-module , we deduce that @xmath1538 is a split surjection for some @xmath1535 . by ( * ; * ? ? ? * theorem 3.2(4 ) ) , it follows that @xmath52 is isomorphic to a direct summand of @xmath1534 . since @xmath1534 is indecomposable , there is an isomorphism @xmath1539 . the morphism @xmath1540 is an isomorphism , so it is not nilpotent . therefore , @xmath1541 does not belong to the radical of @xmath1542 , hence it is invertible ( because @xmath1542 is a local ring ) . so @xmath1543 is an isomorphism , as desired . [ lem : bounded ] let @xmath53 be a bounded complex of @xmath2-permutation @xmath365-modules and @xmath259 an @xmath2-subgroup of @xmath115 such that @xmath1544 is acyclic for all @xmath2-subgroups of @xmath115 that are not conjugate to a subgroup of @xmath259 . let @xmath1495 be a bounded complex of finitely generated projective @xmath1527$]-modules . we assume that @xmath1545 in @xmath1546)$ ] . then there exists a bounded complex @xmath104 of @xmath2-permutation @xmath365-modules , all of whose indecomposable summands have a vertex contained in @xmath259 , such that @xmath1547 in @xmath1548 and @xmath1549 in @xmath1550)$ ] . up to isomorphism in @xmath1548 , we may assume that @xmath1551 . we write @xmath1552 . we will first show by induction on the length of @xmath53 that @xmath1553 and that the indecomposable summands of @xmath53 have a vertex contained in @xmath259 . let @xmath1557 and let @xmath135 be the vertex of @xmath1534 . assume that the composition @xmath1558 is surjective . it follows from lemma [ lem : split ] that @xmath1559 is a split surjection : this contradicts the fact that @xmath1551 . if @xmath135 is not conjugate to a subgroup of @xmath259 , then @xmath1560 is sujective by assumption , hence a contradiction . we deduce by induction that the indecomposable summands of @xmath53 have a vertex contained in @xmath259 . we deduce also that the complex @xmath1562 has no non - zero direct summand that is homotopy equivalent to @xmath102 . by the induction hypothesis , the complex @xmath1563 has no non - zero direct summand that is homotopy equivalent to @xmath102 . it follows that @xmath1553 . we deduce from this that @xmath1564 , where @xmath1565 is homotopy equivalent to @xmath102 . so @xmath1565 is a sum of complexes of the form @xmath1566 with @xmath342 projective indecomposable ( up to a shift ) , hence there is a bounded complex @xmath104 of @xmath2-permutation @xmath365-modules that is a direct sum of complexes of the form @xmath1567 with @xmath342 projective indecomposable such that @xmath1568 . we have @xmath1569 , as desired . [ le : h0brauer ] let @xmath115 be a finite group and @xmath53 be a bounded complex of @xmath2-permutation @xmath1525-modules . assume @xmath1570 for all @xmath1571 and all @xmath2-subgroups @xmath135 of @xmath115 . let @xmath1442 be maximal such that @xmath1575 . the map @xmath1576 is surjective for all @xmath2-subgroups @xmath135 . it follows from lemma [ lem : split ] that @xmath1577 is a split surjection : this contradicts our assumption on @xmath53 . so @xmath1578 for @xmath1442 . replacing @xmath53 by @xmath1579 , we obtain similarly that @xmath1580 for @xmath1581 . the lemma follows . [ th : splitend ] let @xmath115 be a finite group and @xmath121 a subgroup of @xmath115 . let @xmath53 be a bounded complex of @xmath2-permutation @xmath1582-modules all of whose indecomposable summands have a vertex contained in @xmath154 . let @xmath277 be an @xmath2-subgroup of @xmath1587 . by ( * ? * proof of theorem 4.1 ) , we have @xmath1588 if @xmath277 is not conjugate to a subgroup of @xmath154 , and given @xmath1589 an @xmath2-subgroup , we have @xmath1590 in @xmath1591 . note that the indecomposable summands of @xmath1433 are projective for @xmath1592 since their vertices are contained in @xmath1593 for some @xmath1594 , hence @xmath1595)\ ] ] and this vanishes for @xmath1571 . consequently , @xmath1596 in @xmath1597 . [ cor : splendid ] let @xmath115 be a finite group , @xmath121 a subgroup of @xmath115 , @xmath148 a block idempotent of @xmath1523 , @xmath150 a block idempotent of @xmath1599 . let @xmath53 be a bounded complex of @xmath2-permutation @xmath1600-bimodules all of whose indecomposable summands have a vertex contained in @xmath154 . assume @xmath1583)=0 $ ] for all @xmath1571 and all @xmath2-subgroups @xmath135 of @xmath121 and the canonical map @xmath1601 is an isomorphism . theorem [ th : splitend ] shows that the canonical map @xmath1604 is an isomorphism in @xmath1586 . it follows from ( * ? ? ? * theorem 2.1 ) that @xmath1605 induces a rickard equivalence between @xmath1606 and @xmath1607 . the result follows now from ( * ? ? ? * proof of theorem 5.2 ) . m. cabanes and m. enguehard , `` representation theory of finite reductive groups '' , cambridge university press , 2004 . j .- f . dat , _ equivalences of tame blocks for @xmath84-adic linear groups _ , preprint arxiv:1603.07226 .	this paper is a continuation and a completion of @xcite . we extend the jordan decomposition of blocks : we show that blocks of finite groups of lie type in non - describing characteristic are morita equivalent to blocks of subgroups associated to isolated elements of the dual group this is the modular version of a fundamental result of lusztig , and the best approximation of the character - theoretic jordan decomposition that can be obtained via deligne - lusztig varieties . the key new result is the invariance of the part of the cohomology in a given modular series of deligne - lusztig varieties associated to a given levi subgroup , under certain variations of parabolic subgroups . we also bring in local block theory methods : we show that the equivalence arises from a splendid rickard equivalence . even in the setting of @xcite , the finer homotopy equivalence was unknown . as a consequence , the equivalences preserve defect groups and categories of subpairs . we finally determine when deligne - lusztig induced representations of tori generate the derived category of representations . an additional new feature is an extension of the results to disconnected reductive algebraic groups , which is required to handle local subgroups .
doped manganites l@xmath0a@xmath1mno@xmath2 ( l and a are rare earth and alkaline earth ions ) usually exhibits the ferromagnetic ( f ) ground state in a region @xmath13 0.5 . in nd@xmath0ba@xmath14mno@xmath2 , this state develops below @xmath15 120 k for 0.2 @xmath16 0.35 @xcite . for @xmath17 0.3 , these compounds show insulating ( i ) behavior in the ordered state , whereas for @xmath18 0.3 the f metallic ( m ) state is observed below @xmath8 . a colossal magnetoresistance occurs for all @xmath3 near @xmath8 . the unexpected coexistence of ferromagnetic and insulating behavior can not be explained by the double exchange mechanism and remains to be poor understood . a possible reason of the charge localization can be a long - range orbital ordering due to jahn - teller ( jt ) effect . this source of the localization was suggested to account for the fi ground state in la@xmath0ca@xmath14mno@xmath2 for 0.12 @xmath19 0.2 @xcite . the fi state can also be consequence of a spontaneous orbital ordering , which develops in a cubic manganite , due to a coupling of the spin and orbital degrees of freedom arising from the electron correlations @xcite . la@xmath20sr@xmath21mno@xmath2 was found to exhibit the fi state of such nature @xcite . note that a new explanation of the insulator to metal transition , which occurs in an orbital disordered state , has been suggested recently @xcite . this approach permits the existence an orbital liquid insulating state . as far as we know , however , such a state has not been found so far . an important aspect of this problem is a nature of the paramagnetic ( p ) fi transition which can differ from that of a 3d isotropic ferromagnet . a few works have only concentrated on behavior at this critical point . an unconventional p - fi transition was observed in insulating la@xmath22ca@xmath23mno@xmath2 where a magnetic correlation length increased only from 10 to 35 as @xmath24 @xcite . to investigate this issue in detail , we have used nd@xmath0ba@xmath14mno@xmath2 compounds , which are closed to the cubic perovskites due to the small structural distortions @xcite . also of interest in this system is the effect of a mn - nd interaction on its properties . in nd@xmath25ba@xmath26mno@xmath2 , the ordering of the nd moments , that develops below 20 k @xcite , changed the magnetic and transport properties @xcite . as study of the nd@xmath0sr@xmath14mno@xmath2 ( 0.1 @xmath16 0.5 ) showed , the nd magnetic subsystem came also into play at the low temperatures due to its ordering @xcite . at the same time , no effects of the mn - nd coupling were found near @xmath8 in the investigations of critical static behavior of nd@xmath27pb@xmath28mno@xmath2 @xcite and critical dynamics of nd@xmath0sr@xmath14mno@xmath2 ( @xmath29 0.5 ) @xcite . our studies have included the measurements of the linear and nonlinear susceptibilities as well as esr . the neutron diffraction investigations have been also performed to elucidate the effects of the lattice distortions and orbital ordering . the study of the critical magnetic properties of the @xmath3 = 0.23 ( @xmath4 124 k ) and @xmath3 = 0.25 ( @xmath4 129 k ) manganites revealed a more complicated scenario than that expected for the 3d isotropic ferromagnets @xcite . the conventional critical behavior was observed to proceed for @xmath30 where @xmath9 147 k ( @xmath3 = 0.23 ) and @xmath9 143 k ( @xmath3 = 0.25 ) . in contrast , the data on nonlinear response of the second order clearly indicated the occurrence of an anomalous critical behavior for @xmath31 that suggested the coexistence of the two magnetic phases . although the @xmath3 = 0.23 compound exhibited an orbital ordering due to the jt effect below @xmath32 350k , the neutron diffraction study did not reveal a two phase structure in the anomalous @xmath33range @xcite that would be a natural explanation of the phase separated ( ps ) magnetic state . in addition , according to the esr measurements , the transition did not become hysteretic in temperature and clearly first order . in this paper we present the neutron diffraction , magnetization and esr investigations of @xmath3 = 0.25 manganite . the @xmath34 is reduced on doping , becoming @xmath7 250 k. a character of the jt distortions and orbital ordering differs from that of the @xmath3 = 0.23 compound . @xmath35-evolution of the structural parameters does not show any anomalies that can be definitely attributed to formation of a structural ps state in the anomalous @xmath35-range . the magnetization , @xmath36 , measurements are performed in the field cooled ( fc ) and zero field cooled ( zfc ) regimes at @xmath37 1 koe in a range 4.2 @xmath6 300 k. the fc data agree reasonably with prediction for a 3d isotropic ferromagnet in the paramagnetic critical region from 2@xmath8 down to @xmath8 . at the same time , a difference between the fc and zfc results is unexpectedly found to proceed up to 300 k. esr measurements are carried out above @xmath8 . a relaxation rate of uniform magnetization @xmath38 and @xmath39-factor show nearly the same dependencies above @xmath12 as for the @xmath40 0.23 manganite . the @xmath41 does not exhibit a critical enhancement as it occurs in the conventional cubic ferromagnets cdcr@xmath42s@xmath43 and cdcr@xmath42se@xmath43 @xcite , as well as in the near to cubic manganites la@xmath0ca@xmath14mno@xmath2 ( @xmath44 @xcite . an anisotropic exchange mn - nd coupling is shown can give a substantial contribution in @xmath38 , masking the critical enhancement . in order to examine character of the phase transition , effect of the different temperature treatments of the sample on the esr signals is elucidated . these measurements and @xmath36 data support the existence of the magnetic ps state near @xmath8 . the possible reasons of origination of the heterophase magnetic state are discussed . the polycrystalline samples were prepared by the standard technology @xcite , an oxygen content being controlled by a thermogravimetric analysis . a 48-counter powder neutron diffractometer of pnpi with a wave length 1.83 of an incident neutron beam was used in the structural investigations . the measurements were performed in a cryostat at 4.2 @xmath6 300 k. program fullprof was employed for the structure refinement . the magnetization measurements were carried out using a squid magnetometer in a temperature range 4.2 @xmath6 300 k and a magnetic field 1 koe . a single crystal with a mass $ m$ = 91.8 mg was used that came from the same growth batch as the crystal of work @xcite . in esr study , the small part of single crystal investigated in work @xcite with a mass @xmath45 1.6 mg was used . we employed a special @xmath46-band esr spectrometer ( @xmath47 8.37 ghz ) described earlier @xcite . it registers a component of magnetization of a sample which is proportional to an off - diagonal part of its magnetic susceptibility ( @xmath48 ) when a linear polarized exciting @xmath49field * h@xmath50 ox and a steady field h@xmath50 oz . below @xmath9 143 k the nonlinear response of the compound reveals features of the ps magnetic state that can signal the first order transition @xcite . it is natural , therefore , to expect a temperature hysteresis of the esr signal . besides , a state forming at the ps usually depends on a process that is used to transfer a system in the ps regime . therefore , the several types of the temperature scan without / with an external magnetic field were performed . 1 . the sample was slowly cooled from room temperature down to 128 k below @xmath8 and then slowly heated back to room temperature . in the chosen temperature points ( @xmath33points ) the esr spectra were registered . a temporal stabilization time before recording a signal was about 200 - 300 seconds at each @xmath33points . 2 . the same temperature scan as above was applied but with @xmath51 = 4 koe during cooling between @xmath33points . this procedure also tests a possible ps state because the magnetic field can change a balance between the phase fractions . a fast cooling of the sample was used down to 128 k in zero @xmath51 and then the sample was slow heated up to room temperature . a neutron diffraction pattern at room temperature is plotted in fig.1 . similar to nd@xmath52ba@xmath53mno@xmath2 crystal @xcite , pseudocubic structure of nd@xmath54ba@xmath55mno@xmath2 exhibits distortions to a monoclinic symmetry , as the characteristic peaks at the low angles of scattering evidence ( see the insert in fig.1 ) . they unambiguously originate from a lattice deformation because the x - ray diffraction patterns also show them at room temperature . a width of the peaks is larger than that of the main bragg reflections . it indicates that the regions of the relatively small sizes possess the monoclinic distortions . the peaks exhibit no temperature variations and have a small amplitude . the presence of small amount of such a monoclinic phase is expected to have a minimal effect on the @xmath33dependent properties of this compound . the measured diffraction profiles are found to be well fitted using the orthorhombic @xmath5 space group . this fit gives only the slightly worse description than that of the monoclinic group @xmath56 . therefore , the further analysis will be based on the @xmath5 setting . fig.2 displays the @xmath33variations of structural parameters . the orthorhombic distortions are small in range of the measurements as in the @xmath57 0.23 manganite @xcite . however , there are essential differences in their character . the data on the mn - o bond lengths ( fig.2c ) indicate that the jt transition from the high temperature orbital disordered @xmath58 phase to the @xmath59 phase possessing the cooperative jt distortions occurs at @xmath32 250 k. this transition involves the strong abrupt changes of the mn - o - mn angles between 220 @xmath6 290 k when the curves of monotonic @xmath33variation for the mn - o1-mn and mn - o2-mn angles cross one another ( fig.2d ) . this transformation is related to rotation of the mno@xmath60 octahedra around the pceudocubic [ 111 ] direction @xcite . thus , the @xmath61 transition is driven by both the jt distortions that underlie an orbital ordering and a steric effect ( the oxygen ion displacements connected to the octahedrons tilting ) . the latter leads to peculiarity of the transition when a relationship between the lattice parameters ( @xmath62 , fig.2a ) conserves above and below @xmath34 , the splitting @xmath63 becoming even smaller on cooling below @xmath34 . the @xmath61 transition is not usually accompanied by the octahedral tilting . in this case @xmath64 above @xmath34 and @xmath65 below it , as it is found in lasr @xcite , laca @xcite and @xmath40 0.23 ndba @xcite manganites . the @xmath33variations of the lattice parameters and unit - cell volume ( fig.2a , b ) are observed above @xmath8 . they mainly reflect compression of the lattice on cooling which is related to reduction of the lattice parameter . this process completes at @xmath8 . the mn - o bound lengths ( fig.2c ) coincide above @xmath7 250 k and their splitting remains nearly temperature independent below 230 k , so that the jt phase forms in the narrow temperature interval . the splitting corresponds the orbital ordering closed to 3@xmath66 type which develops in the @xmath63 plane because dmn - o1 @xmath67 dmn - o22 @xmath68 dmn - o21 @xcite . in la@xmath0ca@xmath14mno@xmath2 ( 0.12 @xmath69 0.21 ) with the fi ground state , the jt - distortions monotonically increases on cooling over a large temperature interval @xcite . at @xmath40 0.2 , this behavior is observed from @xmath70 300 k @xmath71 down to 100 k. it was interpreted as the coexistence of the @xmath58 and @xmath59 phases i.e. as the structural ps state . although our measurements of the dmn - o are not a very high accuracy , they contradict the coexistence of the @xmath58 and @xmath59 phase with the comparable volume fractions even at 220 k , and support formation of the nearly structural uniform phase below 230 k. the @xmath33evolution of the structural parameters does not reveal any peculiarities in the anomalous @xmath33range ( @xmath9 143 k @xmath6 129 k @xmath72 ) which can be directly associated with the appearance of a two - phase structural state . note that a character of the jt distortions in the @xmath40 0.23 manganite below 150 k ( dmn - o21 - dmn - o1 @xmath67 dmn - o1 - dmno22 ) , where it depends weakly on temperature @xcite , differs from that of the @xmath40 0.25 manganite . thus , the fi state and the anomalous critical behavior can occur at the different types of these distortions . temperature dependence of the ferromagnetic moment ( fig.2b ) shows the usual behavior for a ferromagnet with @xmath4 129 k. the magnetic moment in the ground state 3.0(1 ) @xmath73 ( @xmath73 is the borh magneton ) is less than that of the mn sublattice 3.75 @xmath73 expected for this composition . this is mainly due to antiparallel alignment of the nd magnetic moment @xmath74 @xcite that gives the additive contribution in measured ferromagnetic moment of the sample . 3 displays the temperature dependences of @xmath75 that are obtained in the fc and zfc measurements at @xmath51 = 1 koe . we consider first the fc data above @xmath8 . to determine a scaling behavior of susceptibility of the material @xmath76 , it is convenient to write inverse susceptibility of the sample @xmath77 as @xmath78 here @xmath76 = @xmath79((g\mu)^2/v_0)\tau^{-\gamma}$ ] ( the conventional units ) , @xmath80 is the boltzmann constant , @xmath81 is the numerical factor , @xmath39 is the @xmath82factor , @xmath83 41.45 @xmath84 is the volume of the unit magnetic cell and @xmath85 is the demagnetization factor . a fit of the data for 1 @xmath86 0.093 ( 258 k @xmath87 141 k ) with @xmath88 129 k and @xmath85 = 1/3 gives @xmath89 1.39(1 ) , @xmath90 2.95(3 ) for @xmath91 = 1.875 , these values being independent on @xmath85 for 0.25 @xmath92 0.4 ( see inset(1 ) in fig.3 ) . the same values of @xmath93 and @xmath81 are obtained by employing an interval 1 @xmath86 0.256 whereas the fit becomes worse below 141 k. this shows that @xmath51 = 1 koe affects the @xmath94 dependence only at @xmath95 0.093 . the @xmath96 linear response @xcite and esr measurements ( see fig.5 ) give a slightly less value of @xmath97 1.32 that is due to the demagnetization effects because a fit of the @xmath98 data yields also a close magnitude of @xmath93 . since the nonlinear response reveals the anomalous critical behavior in the low magnetic fields below @xmath9 143 k , it is important to check whether the critical @xmath99 dependence at @xmath51 = 1 koe is the conventional one or not in the same @xmath35-range . our @xmath99 data in this region are related mainly to an intermediate @xmath51-regime between the weak ( @xmath100 143 k ) and strong ( @xmath101 ) magnetic fields . as can be easily verified , a modified arrot plot scheme frequently employed for analysis of the critical @xmath102 dependences @xcite has the incorrect analytic properties in the weak fields . therefore we use another approximate equation of state @xcite with a correct behavior in the weak and strong fields . it can be presented in a convenient form as @xcite @xmath103 here @xmath104 0 , 1 @xmath105 0 are the coefficients , @xmath106 is the factor determining the amplitude of @xmath76 , @xmath107 + @xmath108 is the magnetic field index , @xmath108 is the index of the spontaneous magnetization , @xmath109 and @xmath110 is the internal field . the function @xmath111 decreases monotonically from 1 down to 0 as @xmath3 increases from 0 up to @xmath112 . the @xmath111 and , consequently , @xmath113 are the even functions of @xmath3 . they can be expended in the power series in @xmath114 for @xmath115 1 because eq.(3 ) can be presented as @xmath116 [ the right part]@xmath117 . strongly speaking , @xmath118 in this approximation . nevertheless , the parameters @xmath119 and @xmath120 ( or some of them ) can be treated as the free ones at an approximate fit ( @xmath93 has been determined earlier ) . insert ( 2 ) in fig.3a displays the best fit of @xmath113 at @xmath121 1.39 , @xmath122 5/3 ( the value for a 3d isotropic ferromagnet ) and @xmath123 1/3 for our sample closed to a cube . it gives @xmath124 0.98 that is not far from the theoretical value @xmath125 0.81 . we tried also to find @xmath120 using @xmath118 . this yields @xmath126 1.87 ( @xmath127 0.74 ) , the fit being worse . these results show that the @xmath128 dependence is comparably well described by the conventional scaling static theory in the anomalous region ( @xmath129 ) at @xmath130 1 koe . thus , the anomalous nonlinear critical behavior , that is detected by the @xmath131 measurements in the low magnetic fields ( @xmath132 100 oe ) , is suppressed by the stronger field . this means also that the anomalous phase occupies a relative small volume ( an upper border does not presumably exceed 1/5 ) . according to the fc and zfc data , @xmath133 exhibits the @xmath33hysteresis ( see fig.3a and inset in fig.3b ) . this phenomenon is usually observed in the manganaties below @xmath8 , whereas its presence above @xmath8 up to 300 k is the unexpected result . the @xmath134 + @xmath135 exhibits the maximum ( @xmath136 290 k , see fig.3b ) at high temperatures between 220 k and 300 k which can be attributed to the structural transition occurring in this @xmath33range . below 220 k , @xmath137 increases monotonically at @xmath138 . this means that difference @xmath139 - @xmath140 is not reduced to a @xmath35-independent distinguish in the parameters accounting for the @xmath139 and @xmath140 dependences . one may expect that observation of @xmath137 here is related to the two - phase coexistence . the @xmath131-measurements reveal the two - phase state below @xmath9 143 k as the anomalous nonlinear response becomes larger than the conventional one . here we use another way for disturbance of the system and analyze another quantity , so that the possible inhomogeneous magnetic state may be observed above @xmath12 . for instance , the increasing of @xmath137 on cooling below 220 k may reflect the differences in the @xmath35-dependences of the volume fractions and/or mutual arrangement of the phases . in addition , @xmath12 seems to be also seen in @xmath141 data as a temperature where @xmath142 has a maximum . at last , @xmath141 exhibits the maximum at @xmath8 . this corresponds to a conclusion based on the @xmath131-data @xcite that the ordering of both phases occurs simultaneously at @xmath8 . these observations agree with an assumption that formation of the two - phase state takes place independently of the structural transition and does not reduce to the coexistence of the @xmath59 and a small amount of the initial @xmath58 phases . note some peculiarities of the @xmath35-hysteresis below @xmath8 . according to fig.3a , difference between the fc and zfc data increases for @xmath143 90 k. this agrees with @xmath35-dependence of the hysteresis loop found in the re@xmath144 measurements . the modification seems to reflect change in a domain formation . below 20 k the @xmath145 exhibits the sharp reduction , indicating onset of ordering of the nd ions whose magnetization directed antiparallel to that of the mn moments . fig.4 shows the typical spectra for some temperatures obtained at different @xmath35-scans . the similar spectra are also observed for other temperatures above @xmath8 . the spectra recorded at the different treatments well coincide for temperatures above approximately @xmath12 , and are fitted by a lorentzian line shape @xmath146 \mbox{cos}\varphi_s- \nonumber\\ \bigl[\frac{\omega ( \omega - g\mu h)}{(\omega - g\mu h)^2+\gamma^2}- \frac{\omega ( \omega+g\mu h)}{(\omega+g\mu h)^2+\gamma^2}\bigr ] \mbox{sin}\varphi_s \biggr\}.\end{aligned}\ ] ] here @xmath147 is the antisymmetric component of the dynamic susceptibility of the sample , @xmath76 is the static susceptibility , @xmath148 is the frequency of @xmath96-field , @xmath38 is the relaxation rate of a uniform magnetization , and @xmath149 is the phase of the signal . since @xmath38 is a rather large , the both circularly polarized components of @xmath49 field are taken into account in eq.4 . on this reason , a control sample ( a polycrystalline stable nitroxyl radical with @xmath39 = 2.055 and @xmath150 60 oe ) is used for tuning a phase of @xmath96 field which is the reference point for @xmath149 . the fitting parameters in eq.4 are @xmath38 , @xmath39 , @xmath149 and the amplitude of the signal @xmath151 . we consider first a range @xmath152 where the critical behavior follows to that of a 3d isotropic ferromagnet @xcite , and the structural refinement does not show any structural transformation . fig.5 presents the @xmath153 and @xmath154 dependencies for the different @xmath35-scans which are obtained between @xmath12 and 2@xmath8 . it is seen that the different treatments of the sample have a little effect on the data . the @xmath39-factor ( = 2.00(1 ) ) is independent of temperature here . @xmath155 is expected to obey a scaling law for a 3d isotropic ferromagnet , @xmath156 4/3 , in a weak field regime at @xmath157 = @xmath158 for resonance field @xmath159 3 koe . the inset in fig.5b shows that @xmath155 curve follows this prediction down to approximately @xmath160 0.11 and deviates from it at @xmath161 0.11 . this behavior is in a reasonable correspondence with that of @xmath94 found in the @xmath36 ( see above ) and low frequency measurements of the linear response on a weak @xmath96 field @xcite . let us go to @xmath41 . in the weakly anisotropic cubic ferromagnets , it is usually given by @xcite @xmath162 here @xmath163 is controlled by the dipolar forces , a single ion anisotropy ( @xmath164 ) and anisotropy of the exchange interaction ( the two latter appear in the @xmath5 symmetry ) @xcite . the term with @xmath165 describes the uncritical contribution including a spin - lattice relaxation rate . the dzyaloshinsky - moria interaction also presents in this space group @xcite . following to an analysis given in ref . @xcite , one can show that this interaction gives contribution in @xmath165 only since it connects fluctuations of the magnetization ( critical ) and staggered magnetization ( uncritical ) whose characteristic energy @xmath166 . the @xmath41 exhibits a critical enhancement at @xmath167 0 below a minimal value at @xmath168 . this behavior is observed in the traditional cubic ferromagnets @xcite and in the near to cubic manganites la@xmath0ca@xmath14mno@xmath169 0.22 ) with @xmath170 0.25 @xcite . in our compound @xmath41 reduces monotonically at @xmath171 . this dependence can not be described by the single term with @xmath165 and suggests the presence of a new relaxation mechanism . since nd@xmath172 ion has a magnetic moment and this magnetic subsystem is not presented in the indicated above compounds , a mn - nd coupling is the most likely to be additional source of the relaxation . the ground state of nd@xmath172 ion ( total moment @xmath173 = 9/2 , @xmath174 = 6 and @xmath91 = 3/2 ) is split by a low symmetric ( lower than cubic ) crystal field into the five doublets described by the effective spins @xmath91 = 1/2 . it is expected that a mn - nd exchange interaction @xmath175 is larger than that of the nd - nd @xcite . we will show that the weak anisotropic coupling @xmath175 can give the needed contribution in @xmath41 . the main effect of @xmath175 can be explained by taking into account the single ground doublet of the nd@xmath172 ion and neglecting the nd - nd interactions . we consider first a result of a perturbation theory for @xmath38 with @xmath176 , where s@xmath177 and s@xmath178 are spins of the mn and nd ( @xmath179 = 1/2 ) ions , respectively . a relaxation rate @xmath180 of total spin of the system s = s@xmath181 + s@xmath42 * is determined by an anisotropy of @xmath182 . to get @xmath180 , a pair spin green function of the operators @xmath183 is analyzed @xcite . this includes a decoupling of a four - spin green function , which appears as a result of a commutating [ @xmath184 into product of the two - spin functions . in our case the latter are the green functions of the mn ( @xmath185 ) and nd ( @xmath186 ) subsystems , @xmath185 having the properties of a 3d isotropic ferromagnet and @xmath186 corresponding to the free nd spins . the @xmath180 is also determined by an uniform static green function of the system @xmath187 ( @xmath188 @xmath189 , @xmath190 at a small @xmath191 because @xmath192 and @xmath193 ) . as a result , we find @xmath194 . this answer differs from the critical term ( @xmath195 , eq.(2 ) ) in a factor @xmath196 since the critical contribution contains product of the two critical functions ( @xmath197 ) whereas @xmath198 that gives @xmath199 . the magnitude of @xmath180 is determined by anisotropic part of the @xmath175(q ) at * q * = 0 , where @xmath200(q ) is the fourier transform of @xmath201 , @xmath202 = @xmath203 8 is the coordination number of the mn - nd sublattice . this part is @xmath204 , where @xmath205 sp@xmath206 is the mean exchange coupling . generally speaking , one can find the different relaxation rates along the axes of the crystal . however , the anisotropy of @xmath38 is not observed . it can be due to a twinning of the crystal in the ( @xmath207 ) plane and an averaging of the relaxation rates by the circular polarized components of @xmath208 . therefore , we introduce an effective exchange anisotropy @xmath209 and write @xmath210 @xmath211 reflects an exchange narrowing that is due to the strong isotropic exchange interaction in the mn subsystem . this law is the consequence of decay of the uniform magnetization into the critical and uncritical modes which is enforced by the anisotropy , a characteristic energy of the free nd - spins being obviously equal to zero . note that the same contribution in @xmath38 appears also as an uncritical ( or rather intermediate ) mode has an uncritical amplitude ( @xmath212 ) and a characteristic energy that is equal that of the critical mode . this occurs in some cubic collinear ferrimagnets @xcite . to determine a region of the validity for this result , the equations of motion for the green functions @xmath213(q,@xmath148 ) and @xmath214(q,@xmath148 ) at * q * = 0 should be considered . the @xmath215 introduces a frequency scale @xmath216 in @xmath217 that gives the main restriction . the @xmath215 is assumed to be rather anisotropic @xmath218 . this is not unusual situation for the exchange coupling between the @xmath219 and @xmath220 ions in the perovskite lattice and an estimation given below supports this assumption . therefore , one can use the isotropic component @xmath221 to get @xmath216 . the traditional mode coupling theory @xcite yields @xmath222 . result ( 6 ) is valid if @xmath223 , i.e. @xmath224 , @xmath225 , where @xmath226 is the characteristic energy of the critical fluctuations for a 3d isotropic ferromagnet . at @xmath227 , a new dynamic regime originates which corresponds to a system with completely destroyed conservation law of the * s. this regime develops when the static behavior weakly changes . employing the usual two sublattice expressions for @xmath228 and @xmath229 for the estimations , we find for a static anisotropic border @xmath230 . note that the small anisotropic spin interactions in the nd subsystem @xmath231 ( for instance , the dipolar forces ) give an additional contribution in @xmath165 . they do not effect @xmath232 if @xmath233 . we have discussed above dynamics of the s * which coincides with that of the total magnetization * m * when the @xmath39-factors of the mn ( @xmath234 ) and nd ( @xmath235 ) ions are isotropic and equal to each other . the anisotropy of @xmath236 can be neglected whereas that of @xmath237 may be substantial . consider first a case of isotropic @xmath238 . when @xmath175 = 0 , the spectrum reveals the two resonance at @xmath239 . with increasing of isotropic part of the mn - nd exchange , the lines begin to overlap . in a regime of a strong exchange narrowing ( @xmath240 ) , the spectrum is dominated by a single resonance determined by an effective @xmath82factor @xmath241 @xcite . contributions of the @xmath234 and @xmath235 in @xmath241 depend on the relationships between @xmath242 and @xmath243 as well as @xmath244 and @xmath179 . when @xmath245 dominates ( @xmath246 if @xmath247 1 and @xmath248 . in virtue of the same reason , the anisotropy of @xmath249 related to that of @xmath250 and @xmath215 is also to be small . the expression for @xmath180 ( 6 ) corresponds to the damping of this resonance . note , that in our case the admixture of @xmath235 to @xmath234 is very small : @xmath251 @xcite , where @xmath252 ^ 2(g_{2}^{(0)}/g_{1}^{(0 ) } ) \approx 4\cdot 10^{-4}$ ] even at @xmath191 = 1 because @xmath253 , @xmath254 2.95 , @xmath255 1.88 and @xmath256 @xmath257 = 0.75 is the concentration of nd atoms and @xmath258 1 for estimated below @xmath259 5 k. the same factor @xmath260 accounts for the @xmath35-dependent corrections to @xmath234 related to the mn - nd coupling . the closeness @xmath234 = 2.00(1 ) to @xmath261 2 for la@xmath0ca@xmath14mno@xmath2 @xcite and its independence from @xmath35 agree well with the smallness of @xmath260 . going to the data , we present @xmath38 by @xmath262 inset in fig.5a shows the fit of the @xmath41 for 1 @xmath263 0.13 . the fitting parameters are @xmath163 = 9(2 ) oe , @xmath211 = 590(20 ) oe and @xmath165 = 505(24 ) oe . the @xmath163 is controlled by the dipolar forces since the corresponding contribution is estimated as @xmath264 16 oe for @xmath265 @xmath84 @xcite . it is remarkable that for @xmath211 = 0 we obtain @xmath266 115 oe at @xmath160 0.25 . this value is very close to @xmath267 130 oe which was found in la@xmath0ca@xmath14mno@xmath2 ( @xmath268 0.22 and @xmath8 = 180 k @xmath6 188 k ) at @xmath170 0.25 @xcite . this observation strongly supports the assumption on the leading role that plays the mn - nd coupling in the suppression of critical enhancement of the @xmath41 . from eq.(6 ) , we find @xmath269 3.2 k. the strength of @xmath270 is not completely known . we will try to estimate @xmath271 exploiting @xmath35-dependence of the small nd magnetic moment ( 0.45 @xmath73 at 1.5 k ) which was measured by neutron diffraction in nd@xmath25ba@xmath26mno@xmath2 for 20 k @xmath272 1.5 k @xcite . this dependence suggests that nd magnetization is induced by the ordered mn subsystem , i.e. it is due to the splitting of the nd ground kramers doublet by the mean field of the mn spins . assuming that @xmath221 dominates and @xmath273 2 , we find @xmath274 5 k of the same order as @xmath275 . the border of the anisotropic dynamical crossover is @xmath276 . the weak anisotropic spin interactions in the nd subsystem do not affect @xmath232 if @xmath277 1.3 k @xmath278 . this condition satisfies for the nd - nd dipolar interaction whose magnitude does not exceed that of the mn - mn ( @xmath279 0.5 k ) . the exchange coupling of the nd spins seems to be an order of 1 k or less @xcite . the @xmath232 is close to the minimal @xmath160 0.13 which is available for the precise analysis of the @xmath41 . since @xmath41 can not exhibit a critical enhancement in the anisotropic regime , this effect is completely destroyed in our compound that is confirmed by estimation of the line width at @xmath95 0.13 . the anisotropic static crossover occurs at a very small @xmath280 ( @xmath281 + 0.5 k ) . it is not surprising , therefore , that @xmath94 dependence of our compound and the static critical properties of nd@xmath27pb@xmath28mno@xmath2 ( @xmath282 156 k ) @xcite correspond to those of a 3d isotropic ferromagnet . we have characterized above the nd@xmath172 ion by its single ground doublet . other states are ineffective if a distance between the ground doublet and a next one is larger than 2@xmath8 . it can be a realistic situation , since a cubic component , which dominates in the crystal field of the weakly distorted cubic crystal , can produce a large splitting between the ground doublet and the two upper quartets . in a comparably strongly distorted ndcuo@xmath43 , for example , this splitting is 15 mev whereas the total splitting is 109 mev @xcite . if the next several states are relevant they can be simply taken into account since a distance between the neighboring doublets is definitely larger than mn - nd exchange . in this case these states can be treated separately by the same manner as the ground state so that they produce the obvious additive effect . we consider now the esr data below @xmath12 where a pronounced deviation of the signal from lorenzian starts to be observed . a reason is that im4@xmath283 1 at @xmath284 as 4@xmath285 1 ( 4@xmath286 0.6 at @xmath12 ) . this leads to a complex nonuniform distribution of the @xmath49field in the sample because a wave vector @xmath80 in a medium @xmath287 for a wave spreading along * h * , where , for simplicity , we neglected the @xmath288 . as a result , phase @xmath149 of the signal begins to exhibit a strong @xmath51-dependence that can not be correctly determined for our sample in the resonator . therefore below @xmath12 , we present directly the signals for the several types of the @xmath35-scans . the pronounced @xmath35-hysteresis observed in all the measurements ( fig.4 ) is the clear indication of the anomalous behavior that develops below @xmath12 . since a precise quantitative analysis of the signal is not feasible here , we can not determine whether this hysteresis is caused by that of @xmath75 or it is related also to hysteresis of the @xmath38 and @xmath82factor . a rough estimation gives approximately @xmath35-independent values for @xmath38 and @xmath82factor below @xmath12 down to @xmath8 . the same reason does not allow us to elucidate a possible two - phase character of the signal . at the same time , above @xmath12 phase @xmath149 reflects mainly phase shift of the @xmath208 in the sample which is related to a comparably large @xmath289 10 and conductivity @xmath290 . this effect , being independent of @xmath51 , can be taken into account correctly by eq.(4 ) . for the largest conductivity ( @xmath291cm)@xmath292 ) at @xmath293 260 k , we get @xmath294 0.25 for @xmath295 1 , were @xmath296 0.1 mm is the thickness of the sample . since @xmath290 decreases sharply on cooling ( @xmath297 ( @xmath298cm)@xmath292 at @xmath12 ) , the @xmath299-distribution in the sample is closed to uniform between @xmath12 and @xmath300 . the @xmath35-hysteresis observed in the magnetization and esr measurements above @xmath8 evidences at least a weak first order transition . according to the @xmath131-data , a reason of changing regime of the transition at @xmath12 is formation of the new phase with the anomalous strong nonlinear magnetic behavior in the weak fields which is stronger than that of a 3d isotropic ferromagnet near @xmath8 . this phase coexists with the normal one at least from @xmath12 down to @xmath8 . this ps magnetic state can not be trivially related to a two - phase structure because , according to the neutron diffraction , the sample remains the structural uniform compound below @xmath12 . note that , first , the anomalous phase coexists with the conventional one possessing the different orbital ordering for the @xmath3 = 0.23 and @xmath3 = 0.25 compounds . second , it arises in the manganites that are closed to border of the i - m transition ( @xmath301 0.3 ) , where , according to ref.@xcite and our preliminarily data , the orbital ordering disappears . these observations suggest that the anomalous behavior can be attributed to an orbital disordered metallic phase . the lack of observability of the anomalous phase in the structural measurements means that this phase is structurally closed to the conventional one and/or its volume is small . fragmentation of the sample in the ps state may be such that the anomalous phase does not form a percolative conductive cluster , and the system remains an insulator . the strong @xmath51-nonlinear behavior of this phase is expected to be a combined effect of its own properties and a magnetic coupling of its fragments through the normal phase with the developed ferromagnetic correlations . it is very important to stress a difference between this transition and a first order one whose character is determined on the base of @xmath36 data . in the latter , type of the transition ( first / second ) is characterized by sign of coefficient @xmath302 in an expansion of @xmath303 in @xmath304 above @xmath8 : @xmath305 ( @xmath306 0 ( first order ) , @xmath307 0 ( second order ) ) . such a criterion suggests that the magnetization is the single critical degree of freedom accounting for the transition . this peculiarity is indeed observed in the @xmath36 dependence of traditional ferromagnet mnas @xcite and a number of the manganites @xcite . besides the conventional @xmath36 measurements , sign of @xmath302 can be found from the @xmath131-data because re@xmath308 for @xmath309 0 and a small @xmath148 when re@xmath131 dominates in the response @xcite . our results on the @xmath3 = 0.23 and @xmath3 = 0.25 compounds obtained by this method indicate clearly that @xmath307 0 for both coexisting phases @xcite . this implies that @xmath75 is not the single degree of freedom involved in the transition , and one should consider the charge , orbital and jt - phonon variables ( or some of them ) as the critical ones as well . it is not the unexpected conclusion for the manganites that are characterized by close interconnection of the magnetization and these degrees of freedom . in our system , the charge and orbital degrees of freedom together with @xmath75 are the most likely to be the critical variables . the jt - subsystem seems to be less important because the jt distortions are small . we thank v.i . voronin for help in magnetization measurements . we are also grateful to v.p . khavronin for helpful discussions . this work was supported from the joint russian - belorussian foundation for basic research , grants no.04 - 02 - 81051 rfbr - bel2004_a , f04r-076 and partly from the state program of neutron investigation ( ni ) as well as russian foundation for basic research , grant no . 02 - 02 - 16981 . fig . 1 . the neutron diffraction pattern , calculated profile and residual curve for nd@xmath54ba@xmath55mno@xmath2 at room temperature . the insert shows the wide and weak monoclinic peaks at the small angles . temperature dependences of the structural parameters @xmath310 ( panel ( a ) ) ; volume of the unit cell @xmath311 and magnetic moment of the sample per mn ion ( panel ( b ) ) ; the mn - o bond lengths ( panel ( c ) ) ; and the mn - o - mn angles ( panel ( d ) ) . fig . temperature dependence of the magnetization for the cooling and heating regimes under @xmath51 = 1 koe . panel ( a ) shows these dependencies in the full temperature range 6 @xmath6 300 k. insert ( 1 ) displays fit of the @xmath312 measured in the zfc regime to the function @xmath313 . the fitting parameters are found to be @xmath314 3.98(3 ) g / emu and @xmath315 = 1.39(1 ) at @xmath85 = 1/3 . insert ( 2 ) represents fit of the zfc @xmath316 to expressions ( 2 ) , ( 3 ) ( see the text ) in @xmath35-range @xmath317 261 k. panel ( b ) shows the relative difference @xmath137 versus @xmath35 in the @xmath35-range 80 @xmath6 300 k. insert in ( b ) displays the @xmath145 curves registered in the zfc and fc regimes in @xmath35-range 140 @xmath6 180 k. fig . 4 . the esr spectra for the different @xmath35-scans at some close temperatures : cooling ( solid line , panels ( a - d ) ) , heating after fast cooling ( dashed line , panels ( a - c ) ) , cooling under @xmath51 = 4 koe ( dotted line , panels ( a ) and ( d ) ) . panels ( a ) and ( c ) represent also fit of the spectra recorded on cooling to a lorenzian ( dash - dotted line ) . temperature dependencies of parameters of the esr spectra for different @xmath35-treatments of the sample : cooling ( full symbols ) and heating ( open symbols ) under @xmath51 = 0 ( squares ) and @xmath51 = 4 koe ( triangles ) ; heating after fast cooling ( stars ) . panel ( a ) displays the spin relaxation rate @xmath38 versus temperature . insert shows @xmath41 dependence on cooling , and its fit described in the text . panel ( b ) represents amplitudes @xmath154 of the spectra versus @xmath35 . insert shows fit of @xmath318 to the scaling law .	results of structural neutron diffraction study , magnetization and electron spin resonance ( esr ) measurements are presented for insulating nd@xmath0ba@xmath1mno@xmath2 ( @xmath3 = 0.25 ) with the curie temperature @xmath4 129 k. its pseudocubic structure reveals the definite distortions to a lower symmetry . detailed analysis of the data is performed by using @xmath5 space group in a temperature range 4.2 @xmath6 300 k. the compound is found to exhibit the jahn - teller ( jt ) transition at @xmath7 250 k. character of the coherent jt distortions and their temperature evolution differ from those of the @xmath3 = 0.23 manganite . the field cooled magnetization data are in a reasonable agreement with the predictions for a 3d isotropic ferromagnet above @xmath8 . these measurements , however , reveal a difference between the field cooled and zero field cooled data in the paramagnetic region . the esr results are also in a correspondence with behavior of a 3d isotropic ferromagnet above @xmath9 143 k ( @xmath10 = @xmath11 ) . it is shown that an anisotropic exchange coupling of the mn and nd magnetic moments can give a substantial contribution in esr linewidth masking its critical enhancement . the different temperature treatments ( slow / fast cooling / heating , with / without external magnetic field ) of the sample reveal a temperature hysteresis of the esr spectra below @xmath12 indicating an anomalous response in the paramagnetic region . the study of the magnetic phase transition in the @xmath3 = 0.23 and 0.25 manganites suggests change in its character from the second to first order at @xmath12 . the conventional free energy including the magnetization and magnetic field is not found to describe this first order transition . this suggests that the charge , orbital and jt phonon degrees of freedom , in addition to magnetization , may be the critical variables , the unusual character of the transition being determined by their coupling . the unconventional critical behavior is attributed to an orbital liquid metallic phase that begins to coexist with the initial orbital ordered phase below @xmath12 .
fragmentation is a natural phenomena that occurs in a wide range of contexts and at all scales . the stochastic models used to describe this type of process have attracted a lot of attention lately and form a fascinating class of mathematical objects in their own right , which are deeply connected to branching processes , continuum random trees and branching random walks . a good introduction to the study of fragmentation and coalescence is @xcite which also contains many further references . in the present work , we intend to explore and make use of the connection between random fragmentation processes and branching random walks . more precisely we define the notion of the fragmentation fisher - kolmogorov - petrovskii - piscounov ( fkpp ) equation and study the solutions of this equation . let us start by explaining the connection between fragmentation and branching random walk in the simple framework of _ finite - activity _ conservative fragmentations . in this context everything can be defined and constructed _ by hand_. more general constructions will follow . let @xmath0 be a finite measure on @xmath1 with total mass @xmath2 the homogeneous mass - fragmentation process with dislocation measure @xmath3 is a @xmath4-valued markov process @xmath5 , where @xmath6 , which evolves as follows : @xmath7 this initial fragment then waits an exponential time @xmath8 with parameter @xmath9 after which it splits according to the distribution @xmath10 each of these pieces then starts to fragment independently of the others with the same law as the original object . that is each fragment @xmath11 waits an independent exponential-@xmath12 time after which it splits into @xmath13 where @xmath14 and so on . when a fragment splits we need to relabel all fragments since their relative ranks have changed . this process can be seen as a continuous - time branching random walk . more precisely , if we let @xmath15 then @xmath16 evolves according to the following dynamic . suppose @xmath17 then each individual in the population behaves independently , waits an exponentially distributed length of time with parameter @xmath9 , and then branches into offspring which are situated at distance @xmath18 relative to their parent s position where , as before , @xmath19 has distribution @xmath20 figure [ f : frag rw ] shows an example where @xmath21 [ f : frag rw ] this is not , however , the whole story . in particular it is possible to define fragmentation processes for which the dislocation measure @xmath3 is infinite ( but sigma - finite ) . in this case the fragmentation happens continuously in the sense that there is no first splitting event , and in fact along a branch of the fragmentation tree the branching points are dense on any intervals . the construction of such processes , known as homogeneous fragmentation processes , requires some care and was essentially carried out by bertoin ( @xcite , see also @xcite ) . we defer a brief overview of the general construction to the next section . it is enough here to note that @xmath22 is the rate " at which a given fragments split into pieces whose relative sizes are given by @xmath23 this measure must verify the integrability condition @xmath24 some information about @xmath3 is captured by the function @xmath25 where @xmath26 from now on , we always assume that @xmath27 as we shall reveal in more detail later , the function @xmath28 turns out to be the laplace exponent of a natural subordinator associated to the fragmentation . hence @xmath29 is strictly increasing , concave , analytic and such that @xmath30 the equation @xmath31 on @xmath32 is known to have a unique solution in @xmath33 which we shall denote by @xmath34 ( cf . moreover @xmath35 when @xmath36 . this implies that the function @xmath37 reaches its unique maximum on @xmath38 at @xmath34 and this maximum is equal to @xmath39 homogeneous fragmentation processes will be denoted by @xmath40 and we will let @xmath41 be an enumeration of the sizes of the fragments at time @xmath42 ( more details are given in section [ s : disc ] ) . we define the filtration @xmath43 where @xmath44 to be the natural filtration of @xmath45 the main aim of this paper is to formulate the analogue of the fisher - kolmogorov - petrovskii - piscounov ( fkpp ) equation for fragmentation processes and thereby to analyse the existence , uniqueness and asymptotics of its travelling waves . consider a homogeneous fragmentation process @xmath46 with dislocation measure @xmath3 . the equation @xmath47 is the _ fragmentation travelling - wave _ equation with wave speed @xmath48 equation ( [ e : integ dif equ - first - view ] ) is the analogue in the fragmentation context of the classical traveling - wave equation associated with the fkpp equation . this is discussed in greater detail in the section [ s : disc ] , as well as existing results of the same flavour concerning branching brownian motion and branching random walks . in this paper we study the existence , uniqueness and asymptotics of the solution of this _ fragmentation travelling - wave _ equation . we first introduce two classes of functions . the class of functions @xmath49 such that @xmath50 and @xmath51 and @xmath52 is monotone increasing is denoted by @xmath53 . for each @xmath54 $ ] we further define @xmath55 as the set of @xmath56 such that @xmath57 is monotone increasing . our main result follows [ t : main thm ] fix @xmath54 $ ] and suppose that @xmath58 $ ] belongs to @xmath59 then @xmath60 is a @xmath61-martingale if and only if @xmath62 solves ( [ e : integ dif equ - first - view ] ) with @xmath63 . furthermore , up to an additive translation , there is only one such function @xmath64 which is given by @xmath65 where @xmath66 a.s . is @xmath67-measurable ( see theorem [ t : mgcgce ] for the definition of @xmath68 ) . note that when @xmath69 is a martingale , then it is necessarily uniformly integrable since it is bounded in @xmath70 $ ] and therefore converges in @xmath71 to its limit which we denote by @xmath72 . we start by recalling the definition and some results concerning homogeneous fragmentations . the construction and manipulation of general homogeneous fragmentations is best carried out in the framework of partition valued fragmentations . more precisely let @xmath73 an element @xmath74 of @xmath75 can be identified with an infinite collection of blocks ( where a block is just a subset of @xmath76 and can be the empty set ) @xmath77 where @xmath78 , @xmath79 when @xmath80 and the labelling corresponds to the order of the least element , i.e. if @xmath81 is the least element of @xmath82 ( with the convention @xmath83 ) then @xmath84 the reason for such a choice is that we can discretize the processes by looking at their restrictions to @xmath85:=\{1,\cdots , n\}$ ] ( if @xmath86 we denote by @xmath87}$ ] the natural partition it induces on @xmath85 $ ] ) . roughly speaking , a homogeneous fragmentation is a @xmath75-valued process @xmath88 such that blocks split independently of each other and with the same intensity . let @xmath89 be a @xmath75-valued markov process with cdlg and @xmath90 , in @xmath91 , their distance is defined to be @xmath92 where @xmath93 is the largest integer such that @xmath87 } = \pi'_{\|[n]}$ ] . ] sample path . @xmath46 is called a homogeneous fragmentation if its semi - group has the following , so - called , _ fragmentation property _ : for every @xmath94 the conditional distribution of @xmath95 given @xmath96 is that of the collection of blocks one obtains by splitting the blocks @xmath97 of @xmath74 by an iid sequence @xmath98 of exchangeable random partition whose distribution only depends on @xmath99 we also impose that @xmath100 is the trivial partition with a single block made up of the whole set @xmath76 ( see @xcite for further discussion ) . given @xmath101 we say that it has asymptotic frequencies if for each @xmath102 @xmath103}{n}\ ] ] exists . we write @xmath104 for the decreasing rearrangement of the frequencies of the blocks . it is known that if @xmath46 is a homogeneous fragmentation then almost surely , for all @xmath105 has asymptotic frequencies . the process @xmath106 is called a mass fragmentation . one can define directly mass fragmentations ( i.e. markov processes with state space @xmath4 such that each fragments split independently with the same rate ) but all such processes can be seen as the frequency process of an underlying integer partition fragmentation given the `` stationary and independent increments '' flavor of the definition of a fragmentation , it is no surprise that there is a lvy - kintchin type description of the law of these processes . bertoin shows that the distribution of a homogeneous fragmentation @xmath46 is completely characterised by a sigma - finite measure @xmath3 on @xmath4 ( _ the disclocation measure _ ) which satisfies @xmath107 and a parameter @xmath108 ( _ the erosion rate _ ) . the meaning of the dislocation measure @xmath3 is best understood through a poissonian construction of the process @xmath46 which is given in @xcite . roughly speaking , for each @xmath109 we have an independent poisson point process on @xmath110 with intensity @xmath111 if @xmath112 is an atom of the point process with label @xmath102 , then at time @xmath113 the block @xmath114 of @xmath115 is split into fragments of relative size given by @xmath116 ( or more precisely it is split according to kingman s paintbox partition directed by @xmath117 ) . to make the construction rigorous , one needs to show that the point processes can be used to construct a compatible family of markov chains , @xmath118 , each of which lives on @xmath119 , the space of partitions of @xmath120 . hence @xmath22 is the rate at which blocks fragment into subfragments of relative size @xmath121 the role of the erosion term @xmath122 is easier to explain in terms of mass fragmentations . indeed the process @xmath123 for @xmath124 is a @xmath125-mass fragmentation . as the erosion is essentially a deterministic phenomenon , unless otherwise specified we will always suppose @xmath126 the dislocation measure @xmath3 thus plays the same role as in our introductory example where @xmath3 was finite . recall the definition @xmath127 and that the function @xmath28 is the laplace exponent of a pure jump subordinator associated to the fragmentation . this subordinator is precisely the process @xmath128 it is also not difficult to show that the associated lvy measure is given by @xmath129 our main theorem is in the spirit of earlier results which concern branching brownian motion or branching random walks . we discuss these here and we explain why equation ( [ e : integ dif equ - first - view ] ) is the analogue of the fkpp traveling wave equation for fragmentations . the classical fkpp equation , in its simplest form , takes the form of the non - linear parabolic differential equation @xmath130 the fkpp equation is one of the simplest partial differential equation which admits a traveling wave solution of the form @xmath131 where @xmath132 $ ] and @xmath133 is the speed of the traveling wave . it is a classical result that a traveling wave solution to ( [ e : classical ] ) exists with speed @xmath133 if and only if @xmath134 , in which case it solves @xmath135 there are many different mathematical contexts in which the fkpp equation appears , least of all within the original setting of gene concentrations ( cf . @xcite ) . however , mckean @xcite , ikeda et al . @xcite , neveu @xcite and chauvin @xcite all show that the fkpp equation has a natural relationship with one - dimensional dyadic branching brownian motion ( bbm ) . the latter process consists of an individual particle positioned at @xmath136 which executes a brownian motion until a unit - mean , independent and exponentially distributed time at which point it divides into two particles . each of these particles behaves independently of one another and their parent and undergo the same life - cycle as the initial ancestor but space - time shifted to their point of creation . the process thus propagates via the obvious iteration of this procedure . let @xmath137 denote the positions of the particles at time @xmath42 in the bbm . neveu @xcite ( see also theorem 3.1 in @xcite ) shows that @xmath132 $ ] is a solution of ( [ e : classicaltw ] ) with speed @xmath138 if and only if @xmath139 is a martingale for all @xmath140 . a natural extension of this model is to replace the bbm by a continuous - time branching random walk ( brw ) . this process can be described as follows . a particle positioned at @xmath141 lives for an unit mean , exponentially distributed length of time and upon dying it scatters a random number of offsprings in space relative to its own position according to the point process whose atoms are @xmath142 . each of this particle then iterates this same procedure , independently from each other and from the past , starting from their new positions . as above , let @xmath137 denote the positions of the particles at time @xmath42 . this was for instance studied by kyprianou in @xcite where it was shown that in this context the analogue of ( [ e : classicaltw ] ) is @xmath143 - \psi(x ) \right ) & = & 0 \\ \nonumber \psi(-\infty)&=&0 , \qquad \psi(\infty)=1 . \label{e : brw tw } \end{aligned}\ ] ] and that @xmath144 is a martingale for all @xmath140 if and only if @xmath52 is a solution to ( [ e : brw tw ] ) with speed @xmath133 . in the case of discrete time brw ( [ e : brw tw ] ) reduces to the so - called smoothing transform and a similar result holds , see for example @xcite . given that homogeneous fragmentations can be seen as generalized branching random walks , it is natural to formulate analogous results in this wider setting . to derive the analogue of ( [ e : classical ] ) , let us consider the following intuitive reasoning where for simplicity we assume that @xmath3 is a finite dislocation measure . the classical technique for solving ( [ e : classical ] ) with initial condition @xmath145 consists of showing that @xmath146 is a solution . we take the same approach for fragmentations , where the role of the position @xmath147 is played by @xmath148 . it is easily seen from the fragmentation property that @xmath149 recall also from the definition of a fragmentation process that in each infinitesimal period of time @xmath150 , each block independently experiences a dislocation , fragmenting smaller blocks of relative size @xmath151 with probability @xmath152 and more than one split dislocation occurs with probability @xmath153 . roughly speaking , it follows that as @xmath154 @xmath155 this suggestively leads us to the integro - differential equation @xmath156 with initial condition @xmath145 , which we now formally identify as the analogue to the fkpp equation , even for the case that @xmath3 is an infinite measure . a travelling wave @xmath157 $ ] to ( [ e : defn - fkpp ] ) with wave speed @xmath158 therefore solves @xmath159 in the setting of both bbm and brw , a key element to the analysis of ( [ e : classicaltw ] ) and ( [ e : brw tw ] ) has been the study of certain ` additive ' martingales ; cf . @xcite . in the current setting , these martingales are defined as follows . for all @xmath160 we define @xmath161 the _ additive martingale _ and @xmath162 the _ derivative martingale_. the fact that the processes in the above definition are martingales is not difficult to establish , we refer the reader to bertoin and rouault @xcite for further details . as we shall shortly see , consistently with the case of branching brownian motion and branching random walks , it is the limits of these two martingales for certain parameter regimes in @xmath163 which play a central role in the analysis of ( [ e : integ dif equ ] ) . note that the additive martingale is positive and therefore converges almost surely . the derivative martingale , so called because it is constructed from the derivative in @xmath163 of the additive martingale , is a signed martingale and it is not a priori clear that it converges almost surely . moreover , when limits for either of thee two martingales exist , it is also not clear if they are non - trivial . the following theorem , lifted from bertoin and rouault @xcite , addresses precisely these questions . [ t : mgcgce]@xmath164 ( i ) : : if @xmath165 then @xmath166 converges in mean ( as well as almost surely ) to its limit , @xmath167 which is almost surely strictly positive and when @xmath168 it converges almost surely to zero . ( ii ) : : if @xmath169 then @xmath170 converges almost surely to a non - trivial positive limit , say @xmath171 , which is almost surely strictly positive and has infinite mean . we define @xmath172 if @xmath173 and @xmath174 if @xmath175 . let us briefly discuss in informal terms the proof of this theorem . for convenience we shall define the integro - differential operator @xmath176 moreover we shall say that @xmath177 is a _ multiplicative martingale function _ if ( [ e : prod mart ] ) is a martingale . the equivalence of the analytical property @xmath178 and the probabilistic property of @xmath62 being a multiplicative martingale function emerges from a classical feynman - kac representation as soon as one can apply the appropriate stochastic calculus ( which in the current setting is necessarily driven by the underlying poisson random measure used to defined the fragmentation process ) in order to give the semi - martingale representation of @xmath69 . to show the unique form of solutions in @xmath179 , we start by studying the asymptotics of multiplicative martingale functions . we shall elaborate slightly here for the case @xmath180 . to this end , consider @xmath181 a multiplicative martingale function which makes ( [ e : prod mart ] ) a martingale . as @xmath182 is a uniformly integrable martingale @xmath183 our objective is therefore to understand in more detail the martingale limit @xmath72 . taking logs of the multiplicative martingale we see that @xmath184 recall that @xmath185 hence , if we replace @xmath186 by @xmath187 ( assuming that the argument of the log is asymptotically close to 1 ) and multiply by @xmath188 we see that @xmath189 as @xmath190 . we know however that the @xmath191 is the additive martingale and converges to @xmath192 on the other hand , for each fixed @xmath193 we know that @xmath194 is bertoin s tagged fragment subordinator minus some drift so it is a lvy process with no positive jumps . accordingly it respects the law of large numbers @xmath195 as @xmath190 for some constant @xmath196 which can be shown to be positive . heuristically speaking it follows that , if we can substitute @xmath197 in place of @xmath198 in ( [ e : make the sub ] ) then we can `` factor out '' the @xmath199 terms to get @xmath200 as @xmath190 and since both martingale limits @xmath72 and @xmath167 are non - trivial , we deduce that @xmath201 for some constant @xmath202 . a direct consequence of this is that , irrespective of the multiplicative martingale function @xmath62 that we started with , the @xmath71 limit @xmath72 is always equal ( up to an additive constant in @xmath141 ) to @xmath203 . the stated uniqueness of @xmath62 follows immediately . clearly this argument can not work in the case @xmath175 on account of the fact that theorem [ t : mgcgce ] tells us that @xmath204 almost surely . none the less , a similar , but more complex argument in which a comparison of the @xmath71 martingale limit @xmath205 is made with the derivative martingale limit @xmath206 is possible . one particular technical difficulty in the above argument is that one needs to uniformly control the terms @xmath207 in order to factor out " the common approximation @xmath208 . one could try to control the position of the left - most particle to do this ( as in @xcite ) , however , this turns out to be inconvenient and another technique -taking the martingales along the so - called stopping - lines- appears to work better . this is a classical idea in the context of bbm ( see for example neveu @xcite and kyprianou @xcite ) and brw ( see for example biggins and kyprianou @xcite ) . thanks to their uniform integrability , the martingales @xmath209 and @xmath210 may be seen as the projection of their limits on to @xmath211 , the information generated by the fragmentation tree when it is cut " at fixed time @xmath42 . however , we could also project these limits back on to filtrations generated by the fragmentation tree up to different increasing " sequences of cuts " or _ stopping lines _ as they turn out to be known as . to give an example , a particular instance of a increasing sequence " of stopping line is studied by bertoin and martinez @xcite who _ freeze _ fragments as soon as their size is smaller than a certain threshold @xmath212 . after some time all fragments are smaller than this threshold and the process stops thereby cutting " through the fragmentation tree . the collection of fragments @xmath213 one gets in the end generates another filtration @xmath214 . projecting the martingale limits @xmath72 and @xmath215 back on to this filtration produces two new uniformly integrable martingales which have the same limits as before and which look the same as @xmath216 and @xmath217 , respectively , except that the role of @xmath218 is now played by @xmath219 . considering the case @xmath220 , so that @xmath221 , one could now rework the heuristic argument given in the previous paragraphs for this sequence of stopping lines and take advantage of the uniform control that one now has over the fragment sizes on @xmath213 . a modification of this line of reasoning with a different choice of stopping line when @xmath222 can and will be used as a key feature in the proof of our main theorem . the rest of the paper is organized as follows . in section [ s : existence ] we show that the function @xmath62 given by ( [ e : unique candidate for psi ] ) @xmath223 indeed makes @xmath224 a martingale and that it belongs to @xmath179 . section [ s : stop line ] introduces the notions and tools related to martingale convergence on stopping lines . more precisely we generalize the notion of _ frozen fragmentation _ defined in @xcite to other stopping lines which allow us to study the whole range of wave speeds . one particular feature which will emerge in this section will be the fact that , for certain parameter ranges , the stopping lines we consider will sweep out a crump - mode - jagers ( cmj ) process embedded within the fragmentation process . a central result of the proof is given in section [ s : lln ] where we provide a law of large numbers for the empirical distribution of the sizes of the blocks on the stopping lines defined in the previous section . a key tool here is the connection between these stopped fragmentations and the aforementioned embedded cmj processes which allows us to use classical results from nerman @xcite . next , in section [ s : exact ] , we show that @xmath225 where @xmath226 if @xmath36 and @xmath227 are the only functions in @xmath179 for @xmath228 $ ] which make @xmath182 a product martingale . we conclude by showing that , on the one hand this function @xmath62 solves ( [ e : integ dif equ ] ) , and on the other hand that if @xmath181 solves ( [ e : integ dif equ ] ) then it makes @xmath182 a martingale . our main objective here is to establish the range of speeds @xmath133 in ( [ e : prod mart ] ) for which multiplicative martingale functions exist . in short , the existence of multiplicative martingale functions follows directly from the existence of a non - trivial limit of the martingales @xmath229 for @xmath36 and @xmath230 . we begin with a classification of possible speeds . we shall call wave speeds @xmath133 _ sub - critical _ when @xmath231 , _ critical _ when @xmath232 and _ super - critical _ when @xmath233 . [ existence]@xmath164[t : existence ] ( i ) : : at least one multiplicative martingale function ( which makes ( [ e : prod mart ] ) a martingale ) exists in @xmath53 ( the set of montone , @xmath234 functions with limit 0 in @xmath235 and 1 in @xmath236 ) for all wave speeds @xmath237 $ ] . ( ii ) : : for all wave speeds @xmath233 there is no multiplicative martingale function in @xmath238 before proceeding with the proof we make the following observation . the constants @xmath239 and @xmath240 also play an important role in @xcite . more precisely , following bertoin ( @xcite ) it was shown in @xcite that one could find fragments decaying like @xmath241 provided that @xmath242 $ ] ( the case @xmath243 being particular ) such that @xmath244 hence , the set of admissible wave speeds @xmath245 \}$ ] is closely related to the set of admissible speeds of fragmentation . however , the only @xmath163 such that @xmath246 is @xmath247 if @xmath3 is conservative , then @xmath248 and @xmath249.$ ] the map @xmath28 can have a vertical asymptote at @xmath250 or a finite value with finite or infinite derivative . first let us assume that @xmath251 . since @xmath252 where @xmath253 are independent copies of @xmath254 , it follows that taking laplace transforms and then limits as @xmath255 that if we take @xmath62 as in ( [ e : unique candidate for psi ] ) @xmath256 then @xmath257 hence , we see that @xmath62 is a multiplicative martingale function with wave speed @xmath258 note the fact that @xmath259 follows by dominated convergence . note also that from the definition of @xmath62 it follows automatically that @xmath260 . moreover , suppose that @xmath261 , that is to say @xmath262 . it would then follow automatically from ( [ prob0 ] ) that @xmath263 or @xmath264 . the latter case is ruled out on account of the fact that @xmath167 is an @xmath71 . it therefore follows that @xmath265 . hence @xmath266 next assume that @xmath267 . the method in the previous part of the proof does not work because , according to the conclusion of theorem [ t : mgcgce ] ( i ) , @xmath268 almost surely . in that case it is necessary to work instead with the derivative martingale . using the conclusion of theorem [ t : mgcgce ] ( ii ) we note that by conditioning on the @xmath211 , with @xmath269 as i.i.d copies of @xmath270 we have appealing to similar analysis to previously @xmath271 where the second equality follows since @xmath272 . the same reasoning as above tells us that @xmath273 . now assume that @xmath233 . the following argument is based on ideas found in @xcite . suppose that @xmath274 is the largest fragment in the process @xmath275 . then we know that @xmath276 almost surely . see ( @xcite ) . suppose that a multiplicative martingale function @xmath277 exists within this regime . set @xmath278 . it follows by virtue of the fact that @xmath52 is bounded in @xmath279 $ ] that @xmath280 for all @xmath124 . from the above remark on the rate of decay of @xmath281 we can easily deduce that @xmath282 almost surely as @xmath190 since @xmath283 . taking limits in ( [ takelimits ] ) as @xmath190 , we deduce by dominated convergence and the fact that @xmath50 that @xmath284 for all @xmath136 . this contradicts the assumption that @xmath52 is in @xmath285 recall that @xmath286 [ p : in t2 ] fix @xmath54 $ ] and define @xmath62 as in ( [ e : unique candidate for psi ] ) recalling that @xmath226 if @xmath287 and otherwise @xmath288 when @xmath289 . then @xmath290 we have already established that @xmath266 consider @xmath291 for some non - negative and non - trivial random variable @xmath68 . write @xmath292 . from chapter xiii.2 of feller @xcite it is known that @xmath293/y$ ] is the laplace transform of a positive measure and hence is decreasing in @xmath294 . in turn this implies that @xmath295 is an increasing function in @xmath141 . the concept of _ stopping line _ was introduced by bertoin @xcite in the context of fragmentation processes , capturing in its definition the essence of earlier ideas on stopping lines for branching processes coming from the work of neveu @xcite , jagers @xcite , chauvin @xcite and biggins and kyprianou @xcite . roughly speaking a stopping line plays the role of a stopping time for branching random walks . the tools and techniques we now introduce also have an intrinsic interest in themselves and cast a new light on some earlier results by bertoin . the following material is taken from @xcite . recall that for each integer @xmath193 and @xmath296 we denote by @xmath297 the block of @xmath298 which contains @xmath102 with the convention that @xmath299 while @xmath300 is the @xmath102th block by order of least element . then we write @xmath301 for the sigma - field generated by the history of that block up to time @xmath302 we call stopping line a family @xmath303 of random variables with values in @xmath304 $ ] such that for each @xmath193 * @xmath305 is a @xmath306-stopping time . * @xmath307 for every @xmath308 for instance , first passage times such as @xmath309 for a fixed level @xmath310 define a stopping line . the key point is that it can be checked that the collection of blocks @xmath311 is a partition of @xmath76 which we denote by @xmath312 where as usual the enumeration is by order of least element . observe that because @xmath313 when @xmath314 the set @xmath315 has repetitions , @xmath316 is simply a way of enumerating each element only once by order of discovery . in the same way the @xmath317 s can be enumerated as @xmath318 such that for each @xmath319 corresponds to the stopping time of @xmath320 if @xmath321 is a stopping line , it is not hard to see that both @xmath322 and @xmath323 are also stopping lines . this allows us to define @xmath324 and the sigma field @xmath325 the following lemma ( lemma 3.13 in @xcite ) can be seen as the analogue of the strong markov property for branching processes ; it is also known as the extended fragmentation property cf . lemma 3.14 in bertoin @xcite . [ l : markov ] let @xmath326 be a stopping line , then the conditional distribution of @xmath327 given @xmath328 is @xmath329 where @xmath330 heuristically , we are going along the rays from the root to the boundary , one at a time ( each integer defines a ray , but observe that there are some rays which do not correspond to an integer ) . on each ray @xmath331 we have a stopping time @xmath332 , i.e. we look only at what happens along that ray @xmath333 , and based on that information an alarm rings at a random time . when later we go along another ray @xmath334 , then if @xmath335 ( i.e. the two rays have not branched yet at @xmath336 ) then @xmath337 following chauvin @xcite and kyprianou @xcite we now introduce the notion of _ almost sure dissecting _ and @xmath71-_dissecting _ stopping lines . let @xmath303 be a stopping line . * we say that @xmath321 is a.s . dissecting if almost surely @xmath338 * let @xmath339 then we say that @xmath321 is @xmath340-dissecting if @xmath341 in this section we will discuss changes of measures and subsequent path decompositions which were instigated by lyons @xcite for the branching random walk and further applied by bertoin and rouault @xcite in the setting of fragmentation processes . the following lemma is a so - called many - to - one principle . it allows us to transform expectations of functionals along a stopping line into expectations of functions of a single particle , namely bertoin s tagged fragment . [ l : many to one ] let @xmath321 be a stopping line then , for any measurable non - negative @xmath342 we have @xmath343 to see this , observe that because @xmath344 is an exchangeable partition , the pair @xmath345 is a size - biased pick from the sequence @xmath346 ( i.e. given @xmath347 then @xmath348 ) . the indicator function in the right - hand side comes from the possibility that there is some dust in @xmath344 observe that if @xmath321 is almost surely dissecting , as @xmath46 is non - dissipative , then @xmath344 has no dust . the converse is not true . the second tool we shall use is a probability tilting that was introduced by lyons pemantle and peres for galton watson processes @xcite , lyons @xcite for branching random walks and by bertoin and rouault for fragmentation @xcite . first note that since @xmath349 is a subordinator with laplace exponent @xmath29 , it follows that @xmath350 is a @xmath351-martingale . if we project this martingale on the filtration @xmath61 we obtain the martingale @xmath352 we can use these martingales to define the tilted probability measures @xmath353 and @xmath354 the effect of the latter change of measure is described in detail in proposition 5 of @xcite . more precisely , under @xmath355 the process @xmath356 is a subordinator with laplace exponent @xmath357 and lvy measure given by @xmath358 where @xmath359 was given in ( [ m(dx ) ] ) . under @xmath355 the blocks with index not equal to unity split with the same dynamic as before . the block containing 1 splits according to the atoms of poisson point process @xmath360 on @xmath361 with intensity @xmath362 observe that because @xmath363 is measurable in @xmath364 it is measurable with respect to the filtration of the aforementioned poisson point process and thus the above description is enough to determine the law of @xmath363 under @xmath365 adopting the notation @xmath366 and writing @xmath367 instead of @xmath368 , we have the following result . [ l : proba tilt ] for all positive measurable @xmath369 , @xmath370 from lemma [ l : many to one ] with @xmath371 we have @xmath372 and the result follows . as a first application of these tools we prove the analogue of theorem 2 in @xcite which gives a necessary and sufficient condition for a stopping line to be @xmath340 dissecting . [ t : l1 dissect ] a stopping - line @xmath321 is @xmath340-dissecting if and only if @xmath373 from the previous lemma , @xmath374 it follows that the limit of the left hand side as @xmath190 is zero if and only if @xmath375 . we define an ordering on stopping lines as follows : given @xmath376 and @xmath377 two stopping lines , we write @xmath378 if almost surely , for all @xmath379 . so given a family @xmath380 of stopping lines we say that @xmath381 is increasing if almost surely , for all @xmath382 given @xmath380 an increasing family of stopping lines we may define two filtrations @xmath383 and @xmath384 defined as follows : @xmath385 and then define @xmath386 finally we say that an increasing family of stopping lines @xmath387 is _ proper _ if @xmath388 which is equivalent to @xmath389 for all @xmath102 almost surely . the next lemma mirrors analogous results that were obtained for branching brownian motion by chauvin @xcite . recall that @xmath390 . [ stopmgs ] fix @xmath391 $ ] and let @xmath62 be a solution to ( [ e : prod mart ] ) , i.e. a function that makes @xmath392 a martingale and is bounded between 0 and 1 . then ( i ) : : let @xmath380 be a proper increasing family of a.s . dissecting lines . then the stochastic process @xmath393 is a uniformly integrable martingale with respect to @xmath394 having limit equal to @xmath395 . ( ii ) : : let @xmath380 be a proper increasing family of @xmath340-dissecting lines , then the stochastic process @xmath396 is a unit mean martingale with respect to @xmath397 . furthermore when @xmath398 @xmath399 in @xmath71 where @xmath167 is martingale limit described in theorem [ t : mgcgce ] ( i ) . a straightforward analogue result can not hold for @xmath400 as it is a signed martingale on the stopping lines we consider . we start with ( i ) . with the help of lemma [ l : markov ] . we have for all @xmath136 @xmath401 where in the second and third equalities , given @xmath384 the quantities @xmath402 are independent copies of @xmath403 . as @xmath387 is almost surely dissecting , we know that as @xmath404 the set @xmath405 becomes empty almost surely . since @xmath62 is positive and bounded by unity , we may apply dominated convergence to deduce as @xmath190 that for all @xmath136 @xmath406 since @xmath407 is arbitrarily valued , the last equality in combination with the extended fragmentation property is sufficient to deduce the required martingale property . indeed , suppose that @xmath408 we have that @xmath409 using ( [ constexp ] ) and lemma [ l : markov ] we see that for all @xmath384-measurable @xmath410 , @xmath411 so the martingale property follows . uniform integrability follows on account of the fact that @xmath412 and lemma [ l : markov ] gives us that @xmath413 we now prove ( ii ) . let @xmath321 be a @xmath340 dissecting stopping line . by the monotone convergence theorem @xmath414 \\ & = \lim_{t \to \infty } { \mathbb{e}}\left [ \mathbf{1}_{\{\ell_1 \le t\ } } \|\pi_1(\ell_1)\|^{p } e^{\phi(p ) \ell_1}\right]\\ & = \lim_{t \to \infty } { \mathbb{e}}^{(p)}\left [ \mathbf{1}_{\{\tau \le t\}}\right ] \\ & = 1\end{aligned}\ ] ] where we have used the many - to - one principle , the probability tilting ( [ e : change of pb ] ) and theorem [ t : l1 dissect ] . to prove the martingale property , fix @xmath415 and observe that @xmath416 we apply the strong markov property lemma [ l : markov ] to obtain that @xmath417 the @xmath71 convergence of @xmath418 when @xmath36 is a consequence of the fact that lemma [ l : markov ] applied to @xmath387 again gives us @xmath419 together with theorem [ t : mgcgce ] ( i ) . in this paragraph , we introduce the families of stopping lines we will be using and we show that they satisfy the desired properties . fix @xmath420,$ ] and for each @xmath421 let @xmath422 be the stopping line defined as follows : for each @xmath193 @xmath423 where @xmath424 in other words , @xmath425 is the first time when @xmath426 crosses the line @xmath427 ( see figure [ stop line ] ) . recall that @xmath428 is increasing on @xmath429 $ ] with @xmath430 [ stop line ] for any fixed @xmath431 $ ] , the family @xmath432 is a.s . and @xmath340 dissecting as well as proper . as @xmath433 when @xmath180 we have from ( [ left - most - speed ] ) that for all @xmath434,$ ] @xmath435 \ ; -\log \|\pi_1(t)\|^\downarrow -c_p t \rightarrow \infty \label{superspeed}\ ] ] as @xmath190 . this entails immediately that the lines are a.s . dissecting . to prove that for any @xmath436 is @xmath71 dissecting we use theorem [ t : l1 dissect ] . recall that under @xmath355 the process @xmath437 is a subordinator with laplace exponent @xmath438 @xmath439 and this is equal to one if and only if @xmath440 which is equivalent to @xmath441 consider the stopping line @xmath442 two fundamental differences in the way this stopping line dissects the fragmentation process occurs in the regimes @xmath443 and @xmath444 . when @xmath445 it is easily seen that for all @xmath193 we have that @xmath446 so that @xmath447 is the trivial partition with all integers in one block . on the other hand when @xmath448 then we claim that almost surely , for all @xmath193 we have @xmath449 by exchangeability it is enough to prove it for @xmath450 observing that @xmath451 is a spectrally positive lvy process with negative drift , standard theory ( cf . chapter vii of @xcite ) tells us it has the property that @xmath452 is irregular for @xmath453 which , in turn , is the claim . hence , when @xmath444 , the partition @xmath447 is a non - trivial collection of non - singleton blocks . in this section we make an interesting observation concerning the process of blocks that are swept out of the underlying fragmentation process by the stopping lines @xmath454 . our observation builds on an idea which goes back to neveu @xcite . a cmj process is a branching process in which a typical individual reproduces at ages according to a random point process on @xmath455 and may or may not live for ever . the coming generation at time @xmath42 of a cmj . process consists of the collection of individuals born after time @xmath42 whose parent was born before time @xmath42 . in this section we show that the coming generation of a fragmentation with respect to a stopping line @xmath422 ( i.e. the collection of blocks @xmath456 ) is also the coming generation at time @xmath457 of certain cmj process which is path wise embedded into the fragmentation process . the case @xmath220 corresponds to a vertical line in figure [ stop line ] and has been studied in detail in bertoin and martinez @xcite and harris _ et al . indeed , the process @xmath458 can be described as the fragmentation processes @xmath459 where one freezes fragments when they are smaller than some @xmath460 for the first time . it is evident that the final state of such a process corresponds exactly to the coming generation associated to the stopping line @xmath461 ( with @xmath220 ) and @xmath462 in the following we let @xmath463 $ ] be fixed , and we consider the collection of distances @xmath464 i.e. the point process of distances to the line @xmath465 of the individuals in the first generation . note specifically that the latter point process can not be defined when @xmath466 on account of the fact that for all @xmath102 we have that @xmath446 . define @xmath467 to be the point process of the @xmath468 s and let @xmath469 be its intensity measure . the following proposition shows that the associated intensity measure has several convenient properties . [ p : cmj - conditions ] fix @xmath463 $ ] . then @xmath470 is a non - lattice measure with the following properties . 1 . its malthusian parameter is equal to @xmath471 , i.e. @xmath472 2 . for all @xmath460 such that @xmath473 , @xmath474 we first introduce some more notation : we define the _ martingale weights _ @xmath475 note that for any borel set @xmath476 we have with the help of the many - to - one principle ( [ e : many - to - one ] ) that @xmath477 where @xmath478 and @xmath479 . it is well known that since @xmath480 is spectrally positive the law of @xmath481 is diffuse and hence non - lattice . note also that @xmath482 which establishes the proof of part ( i ) . for the proof of part ( ii ) , our objective is to compute @xmath483 noting that @xmath452 is irregular for @xmath453 for @xmath480 ( cf . chapter vii of @xcite ) and , moreover , that @xmath484 , it follows that the ascending ladder height of @xmath480 is a compound poisson process . the jump measure of the latter , say @xmath485 , is therefore proportional to @xmath486 and , from corollary 7.9 in @xcite , it can also be written in the form @xmath487 where @xmath488 is the largest root in @xmath455 of the equation @xmath489 . in order to verify ( [ needtocheck ] ) it therefore suffices to prove that @xmath490 whenever @xmath473 . to this end note that , with the help of fubini s theorem and the fact that @xmath491 , we have @xmath492-\frac{\eta}{\epsilon(\eta+\epsilon)}\int_0^\infty ( 1-e^{-\epsilon z})e^{-(p-\epsilon)z}m(dz ) \\ & & \hspace{6cm}+ \frac{1}{\eta+\epsilon}\int_0^\infty ( 1- e^{-\eta z } ) e^{-pz } m(dz)\\ & = & \frac{1}{\eta+\epsilon}[\phi(p ) - \phi(p-\epsilon ) ] + \frac{1}{\eta+\epsilon } [ - \phi(p)]\\ & = & \frac{1}{\eta+\epsilon } [ \phi(p+\eta ) - \phi(p-\epsilon)]\end{aligned}\ ] ] which is indeed finite when @xmath473 . we conclude this section by showing an important relationship between first passage times and the coming generation of an embedded cmj process . to this end , define for each @xmath421 @xmath493 [ embedded - cmj ] for each @xmath494,$ ] the process @xmath495 is the process of the birth times of the individuals in the coming generation at level @xmath457 of a cmj process whose individuals live and reproduce according to the point process @xmath496 note that proposition [ p : cmj - conditions ] ( i ) implies that @xmath497 for all @xmath460 and hence there is an almost sure first atom in the point process @xmath498 . the theorem is trivially true for each @xmath499 the process @xmath456 is constant on @xmath500 and has a jump at time @xmath501 where a single block splits . thanks to the branching property it give birth to a collection of blocks in @xmath502 whose positions to the right of their parent is again an instance of the point process @xmath496 this shows that as the line @xmath422 sweeps to the right , the coming generation process @xmath503 describes a cmj process which is our claim . before proceeding to the proof of asymptotics and uniqueness of multiplicative - martingale functions in the class @xmath504 , we need to establish some further technical results which will play an important role . the following result is of a similar flavour to the types of laws of large numbers found in bertoin and martinez @xcite and harris _ et al . _ @xcite . the next theorem gives us a strong law of large numbers for fragments in @xmath422 as @xmath505 with respect to the weights ( [ mgweights ] ) when @xmath506 $ ] . recall that the lvy measure @xmath507 was defined in ( [ m(dx ) ] ) and the definition of the martingale weights ( [ mgweights ] ) @xmath508 [ t : sandwlln ] fix @xmath509 $ ] let @xmath510 such that @xmath511 . suppose @xmath512 for some @xmath513 and @xmath460 satisfying @xmath514 then @xmath515 in probability where @xmath516 if @xmath342 is uniformly bounded , then , the above convergence may be reinforced to almost sure convergence . for convenience we shall define @xmath517 note that from the many - to - one principle ( [ e : many - to - one ] ) that @xmath518 where @xmath519 and for @xmath124 , @xmath520 . note that @xmath480 is in fact a subordinator on account of the fact that when @xmath506 $ ] , @xmath521 . moreover , it has finite mean with @xmath522 for @xmath124 . note also that the assumption that @xmath511 implies that the expectation on the right hand side of ( [ frommt1 ] ) does not include the possible contribution that comes from the event that @xmath480 creeps over @xmath457 . let us first prove that if @xmath512 for some @xmath513 and @xmath460 satisfying @xmath473 ( and in particular for uniformly bounded @xmath342 ) we have that @xmath523 a classical result from the theory of subordinators ( cf . chapter 3 of @xcite ) tells us that for @xmath524 @xmath525 where @xmath526 is the potential measure associated with @xmath480 under @xmath527 , meaning that @xmath528 , and we recall that @xmath529 is the lvy measure of @xmath331 under @xmath527 . hence it follows that @xmath530 where @xmath531 it can also be shown that @xmath532 is a classical renewal measure of a renewal process with mean inter - arrival time given by @xmath533 ; see lemma 5.2 of @xcite . the latter result also indicates that the associated inter - arrival time of @xmath534 has distribution @xmath535 and hence an easy computation shows that the mean inter - arrival time is equal to @xmath536 . applying the key renewal theorem to @xmath537 we deduce that , whenever @xmath369 is directly reimann integrable , @xmath538 note that @xmath369 has no discontinuities on account of the fact that , for each @xmath539 , @xmath540 which equals zero thanks to the assumption that @xmath511 . moreover , thanks to the assumption that @xmath541 and @xmath473 we have that @xmath542 which shows that @xmath369 is directly reimann integrable . we have thus established ( [ simon s assumption ] ) . next we turn to establishing the limit ( [ intwosenses ] ) in the almost sure sense when @xmath342 is uniformly bounded . @xcite show that , when @xmath220 , the required strong law of large numbers holds for all bounded measurable @xmath342 in the sense of almost sure convergence . although we are interested in conservative fragmentation processes in this paper , the proof of ( [ intwosenses ] ) for the case that @xmath543 is mathematically similar to the dissipative case that was handled when @xmath220 in @xcite . in the notation of @xcite , the role of the quantity @xmath544 is now played by the martingale weights @xmath545 in that case , using ( [ frommt1 ] ) in place of the limit ( 9 ) in @xcite , all of the proofs go through verbatim or with obvious minor modification with the exception of their lemma 5 which incurs a moment condition . in fact , this moment condition turns out to be unnecessary as we shall now demonstrate . the aforementioned lemma requires us to show that , in the notation of the current setting , @xmath546 thanks to jensen s inequality the process @xmath547 is a submartingale . hence , recalling that the almost sure limit of @xmath418 is @xmath167 , as soon as it can be shown that @xmath548 for some @xmath549 , then ( [ no - moment ] ) is satisfied . note however , the latter has been clearly stated and established in the proof of theorem 2 of @xcite despite the fact that the aforementioned theorem itself does not state this fact . this completes the proof of the almost sure convergence ( [ intwosenses ] ) for uniformly bounded @xmath342 . we shall now obtain the required weak law of large numbers for @xmath550 . to this end , let us suppose that @xmath551 is an increasing sequence of bounded positive functions such that , in the pointwise sense , @xmath552 . it follows from the aforementioned strong law of large numbers for each @xmath553 that @xmath554 for all @xmath555 and hence by monotone convergence , @xmath556 almost surely . next note by fatou s lemma , @xmath557 where the final equality follows by ( [ simon s assumption ] ) , and hence we are led to the conclusion that @xmath558 almost surely . next define for @xmath407 @xmath559 note by ( [ simon s assumption ] ) , monotone convergence and ( [ a.s.liminf ] ) that @xmath560 it now follows as a simple consequence from the markov inequality that @xmath561 converges in probability to zero . using the latter together with the almost sure convergence in ( [ a.s.liminf ] ) , it follows that @xmath562 converges in probability , to zero as @xmath505 . in the above proof , when dealing with the almost sure convergence ( [ intwosenses ] ) for uniformly bounded @xmath342 , the replacement argument we offer for lemma 5 of harris _ et al . _ @xcite applies equally well to the case that the fragmentation process is dissipative . this requires however book keeping through the paper of bertoin @xcite , which assumes conservativeness , in order to verify that the @xmath563 estimates are still valid . the consequence of this observation is that the moment condition ( a3 ) in @xcite is unnecessary . this clears up the inconvenience of the aforementioned condition which , in fact , is equivalent to the condition that the dislocation measure has finite total mass . the next result gives us a strong law with respect to the weights ( [ mgweights ] ) for the regime @xmath564 $ ] . for this we recall also that the measure @xmath565 was defined for @xmath566 $ ] in proposition [ p : cmj - conditions ] by @xmath567 and its intensity was denoted by @xmath470 . it is also worth recalling that the lvy measure associated with the tagged fragment @xmath568 is denoted by @xmath359 for @xmath569 . moreover , under the measure @xmath527 where @xmath32 , the aforementioned lvy measure takes the form @xmath570 for @xmath569 . [ t : slln ] fix @xmath571 $ ] . let @xmath510 such that @xmath512 for some @xmath513 and @xmath460 satisfying @xmath473 . then , almost surely , @xmath572 where @xmath573 first recall from theorem [ embedded - cmj ] that when @xmath574 $ ] , the sequence of stopping lines @xmath575 sweeps out the coming generation of an embedded cmj process with malthusian parameter @xmath471 and whose associated birth process is described by the point process @xmath576 . for the aforementioned cmj process we denote by @xmath577 the birth times of individuals , where the enumeration is in order of birth times starting with the initial ancestor , counted as @xmath453 , having birth time @xmath578 . define @xmath579 then , following jagers classical theory of counting with characteristics ( cf . jagers @xcite ) , our cmj processes has count at time @xmath407 given by @xmath580 where , for each @xmath102 , @xmath581 has the same definition as @xmath582 except that the counting measure @xmath565 is replaced by the counting measure of the point process describing the age of the @xmath102-th individual at the moments it reproduces . in this respect the characteristics @xmath583 are i.i.d . writing @xmath584 for the ages at which the @xmath102-th individual reproduces and @xmath585 for the index set of individuals which form the coming generation at time @xmath407 , we have @xmath586 where the final equality follows from theorem [ embedded - cmj ] . in the particular case that @xmath342 is identically equal to unity , we denote @xmath587 by @xmath588 and we see that @xmath589 . recall that @xmath470 was defined in proposition [ p : cmj - conditions ] as the intensity of the counting measure @xmath565 . the strong law of large numbers ( [ e : slln ] ) now follows from the classical strong law of large numbers for cmj processes given in theorem 6.3 of @xcite which says that @xmath590 almost surely provided that the following two conditions hold for some @xmath591 . firstly @xmath592 and secondly @xmath593 the first condition holds thanks to proposition [ p : cmj - conditions ] ( ii ) for all @xmath594 sufficiently close to @xmath471 . for the second condition , we note that when @xmath512 we may estimate for all @xmath594 sufficiently close to @xmath471 , @xmath595 which is finite , again thanks to proposition [ p : cmj - conditions ] , providing @xmath473 . note in particular that these conditions also ensure that the right hand side of ( [ nerman - limit ] ) is positive but neither zero nor infinity in value . the proof of part ( i ) of the theorem is thus complete as soon as we note that ( [ nerman - limit ] ) is the desired limit . in this section we establish the asymptotics of multiplicative martingale functions in the class @xmath504 which will quickly lead to the property of uniqueness within the same class . we start with the following first result . [ t : sv ] suppose that @xmath391 $ ] and that @xmath596 is a product martingale function which makes @xmath597 in [ e : prod mart ] a martingale . then for all @xmath598 @xmath599 that is to say @xmath199 is additively slowly varying . the proof is an adaptation of arguments found in @xcite . first note that the monotonicity of @xmath199 implies that for all @xmath598 @xmath600 in turn , this implies that for each @xmath598 there exists an increasing subsequence @xmath601 tending to infinity along which we have the following limit , @xmath602 suppose now that there exists a @xmath603 and @xmath604 such that @xmath605 the monotonicity of @xmath199 implies that for all @xmath606 @xmath607 the crux of the proof will be to show that this leads to a contradiction . to this end , recall the identity ( [ constexp ] ) in the proof of theorem [ stopmgs ] . starting from this expression and with the help of a telescopic sum we have that for all @xmath598 @xmath608 \prod_{j < i}\psi_p(x -\log\|\pi_j(\ell^{(p , z)})\| -c_p \ell^{(p , z)}_j).\ ] ] recalling the definition of @xmath199 it follows easily that @xmath609 \label{ktoinf}\end{aligned}\ ] ] now pick @xmath610 . next recalling that @xmath260 and that @xmath611 we can take limits in ( [ ktoinf ] ) as @xmath612 , applying fatou s lemma twice , to reach the inequality @xmath613 however , theorem [ stopmgs ] implies that the expectation above is equal to unity and we reach a contradiction . we are forced to conclude that @xmath614 and the required additive slow variation follows . the following lemma is a key ingredient which will help to extract exact asymptotics . [ l : lswitch ] fix @xmath391 $ ] . suppose that @xmath615 is a monotone increasing function and additive slowly varying at @xmath236 , that is to say , it satisfies the property that for all @xmath616 @xmath617 then @xmath618 where the limit is understood almost surely when @xmath566 $ ] and in probability when @xmath506 $ ] . the proof will follow closely ideas in theorem 8.6 of biggins and kyprianou @xcite . first define @xmath619 then using the fact that @xmath369 is increasing and @xmath620 we have @xmath621 as @xmath369 is additively slowly varying , we may appeal to the classical representation of slowly varying functions ( cf . viii . 9 of feller @xcite ) to deduce that for all @xmath622 there exists a @xmath623 such that for all @xmath539 @xmath624 this allows for the upper estimate on the second term on the right hand side of ( [ upperest ] ) for all @xmath457 sufficiently large @xmath625 where @xmath626 now note from theorem [ t : slln ] that the right hand side of ( [ needlln ] ) converges almost surely to @xmath627 when @xmath566 $ ] , where the definition of @xmath627 is given by ( [ bigpqp ] ) . when @xmath506 $ ] the convergence occurs in probability and @xmath627 is defined by ( [ smallpqp ] ) . in either case , thanks to the appropriate integrability of the function @xmath628 for sufficiently small @xmath629 , we have that @xmath630 as @xmath631 . moreover , in both cases , using ( [ needlln ] ) and ( [ upperest ] ) we have @xmath632 when @xmath566 $ ] , thanks to the preceding remarks , as @xmath505 , the almost sure limit of the right hand side above can be made arbitrarily small by choosing @xmath133 sufficiently large . when @xmath506 $ ] , again thanks to the preceding remarks , we see that for each @xmath633 , we may choose @xmath133 sufficiently large such that @xmath634 thus establishing the required convergence in probability . we are now ready to establish the asymptotics of multiplicative martingale functions from which uniqueness will follow . for any product martingale function @xmath62 , with speed @xmath635 , where @xmath636 $ ] , which belongs to the class @xmath504 , recall that we have defined @xmath637 moreover , it has already been shown in theorem [ t : sv ] that @xmath199 is additively slowly varying . [ t : l cvg to cstt ] suppose that @xmath62 is any product martingale function in @xmath504 with speed @xmath635 such that @xmath636 $ ] . then there exist constants @xmath202 such that when @xmath165 we have that @xmath638 and when @xmath169 we have @xmath639 suppose that @xmath636 $ ] . it is not difficult to show that for any given @xmath633 we may take @xmath457 sufficiently close to @xmath264 such that @xmath640 \label{estimate}\ ] ] thanks to theorem [ stopmgs ] , and in particular the fact that @xmath641 is a uniformly integrable martingale with limit @xmath395 @xmath642 since @xmath643 and @xmath260 we may apply the estimate in ( [ estimate ] ) and deduce that @xmath644 next we consider the restriction that @xmath165 . recalling from theorem [ stopmgs ] that @xmath645 converges almost surely in @xmath71 to @xmath167 and from theorem [ t : sv ] that @xmath199 is additive slowly varying , we may apply the conclusion of lemma [ l : lswitch ] to ( [ logmglimit ] ) and deduce that @xmath646 the right hand side above is purely deterministic and the left hand side is bounded and strictly positive which leads us to the conclusion that there exists a constant @xmath202 such that @xmath647 as @xmath648 . now suppose that @xmath649 . from bertoin and rouault @xcite ( see also the method in kyprianou @xcite ) it is known that @xmath650 where @xmath651 is a uniformly integrable positive martingale with mean @xmath141 ; we denote its limit by @xmath652 . moreover thanks to ( [ left - most - speed ] ) it is also true that there exists a part of the probability space , say @xmath653 , satisfying @xmath654 , such that @xmath655 on @xmath653 . again thanks to lemma [ l : markov ] we may project the limit @xmath652 back on to the filtration @xmath656 to obtain @xmath657 is a positive uniformly integrable martingale with almost sure and @xmath71 limit @xmath652 where @xmath658 note also that this implies that for each @xmath569 , on @xmath653 we have @xmath659 almost surely . as we may take @xmath141 arbitrarily large , the above almost sure convergence occurs on the whole of the probability space . next turning to lemma [ l : lswitch ] , we note that both @xmath660 and @xmath661 are suitable functions to use within this context . we therefore have for @xmath569 @xmath662 almost surely . thanks to ( [ logmglimit ] ) and ( [ ell - dw ] ) it follows that @xmath663 almost surely . in particular , as the left hand side above is deterministic , and the right hand side is a random variable in @xmath452 , it follows that the limit must be equal to some constant in @xmath664 which we identify as @xmath665 . this completes the proof . [ t : unique mmf ] for @xmath54,$ ] there is a unique multiplicative martingale function @xmath62 which is solutions to ( [ e : prod mart ] ) with speed @xmath635 in @xmath666 ( up to additive translation in its argument ) . in particular , when @xmath36 the shape of the multiplicative martingale function is given by @xmath667 and the shape of the critical multiplicative martingale function is given by @xmath668 first suppose that @xmath165 and take any travelling wave @xmath62 at wave speed @xmath635 . thanks to the uniform integrability of the associated multiplicative martingale as well as ( [ -logm = w ] ) we have that @xmath669 note that from ( [ e : integ dif equ ] ) , if @xmath670 is a travelling wave then so is @xmath671 for any @xmath672 . we therefore deduce from ( [ psi - is - wlt ] ) that travelling waves at wave speed @xmath635 and @xmath36 are unique up to an additive translation in the argument . moreover , without loss of generality , the shape of the travelling wave may be taken to be of the form given on the right hand side of ( [ psi - is - wlt ] ) but with @xmath673 . exactly the same reasoning applies in the case @xmath674 except that we appeal to the distributional identity ( [ gives uniqueness ] ) instead of ( [ -logm = w ] ) . given the conclusion of theorem [ t : unique mmf ] , it remains only to prove the first part of theorem [ t : main thm ] . to this end , suppose first that @xmath181 and that @xmath675 ( and hence implicitly we understand that @xmath62 is in the domain of @xmath676 ) . define @xmath677 for @xmath136 and @xmath678 where @xmath69 is given by ( [ e : prod mart ] ) . also for convenience write @xmath679 the change of variable formula gives us @xmath680.\ ] ] henceforth we will use the notation @xmath681 recall the poisson point process construction of the fragmentation @xmath682 described in the introduction . write @xmath683 for the poisson random measure on @xmath684 with measure intensity @xmath685 which describes the evolution of the fragmentation process . using classical stochastic analysis of semi - martingales and the poissonian construction of fragmentation processes we deduce that @xmath686 and it is obvious that changing @xmath687 into @xmath688 does not affect the value of the integral in the final step above . the assumption ( [ parabolic ] ) now implies that for all @xmath689 and @xmath678 @xmath690 it is now a simple application of the fragmentation property to deduce that @xmath691 where @xmath692 is the fragmentation process initiated by the @xmath102-fragment at time @xmath42 of the original fragmentation process . hence @xmath693 is a @xmath61-martingale . + for the converse direction , we know from theorem [ t : unique mmf ] that if @xmath181 makes @xmath597 a martingale then , without loss of generality we may take @xmath694 where @xmath695 if @xmath36 and @xmath696 if @xmath674 . for the rest of the proof , @xmath62 will be given by the above expression . note that since @xmath697 is monotone increasing ( see proposition [ p : in t2 ] ) we have that @xmath698 so that @xmath699 this estimate and that it implies the uniform boundedness of @xmath700 will be used at several points in the forthcoming text . we will use the following fact . given @xmath704 and @xmath705 two sequences in @xmath70 $ ] we have that @xmath706 as an immediate application we have that @xmath707 we bound the two terms separately . for the first term , chose @xmath708 small enough so that @xmath709 to obtain @xmath710 where @xmath711 and @xmath712 are finite constants and we have used the mean value theorem and ( [ e : bound psi ] ) in the second inequality and @xmath713 in the final inequality . this shows that @xmath714 for the second term in ( [ e : two terms ] ) , we first observe that thanks to theorem [ t : sv ] , for each @xmath715 we can bound @xmath716 as @xmath32 we can chose @xmath708 small enough so that @xmath717 which then implies that @xmath718 hence , putting the two bounds together we see that for each @xmath689 it holds that @xmath719 and hence @xmath62 belongs to the domain of @xmath720 let us now show that @xmath701 is continuous . as @xmath62 is @xmath721 it is enough to show that @xmath722 is continuous . we start by writing @xmath723 next we decompose the integrand as a sum @xmath724 the proof will be complete once we will have shown that the integral with respect to @xmath22 of each term on the right hand side of ( [ e : fourterms ] ) goes to 0 as @xmath725 * first term :* the first term is @xmath726 the term @xmath727 is uniformly bounded between 0 and 1 . on the other hand @xmath728 where @xmath729 $ ] and @xmath730.$ ] observe that for each @xmath731 because @xmath62 is @xmath732 we have that @xmath733 as @xmath725 ( the choice of @xmath734 and @xmath735 might not be unique , but by adopting the convention that we always chose the lowest possible such value the above argument becomes tight ) . fix @xmath736 and decompose @xmath737 \nu ( ds ) \\ & = \int_{\{1-s_1>\delta\ } } ( -\log s_1 ) \left[\psi_p'(x+\eta_1(\epsilon , s_1 ) ) - \psi_p'(x+ \eta_2(\epsilon , s_1 ) \right ] \nu ( ds ) \\ & \qquad \qquad+ \int_{1- s_1\leq \delta } -\log s_1 \left[\psi_p'(x+\eta_1(\epsilon , s_1 ) ) - \psi_p'(x+ \eta_2(\epsilon , s_1 ) \right ] \nu ( ds ) \\ & \le c(\delta ) + c'(\delta)\int_{\{1- s_1\leq \delta\ } } \left[\psi_p'(x+\eta_1(\epsilon , s_1 ) ) - \psi_p'(x+ \eta_2(\epsilon , s_1 ) \right ] ( 1-s_1)\nu ( ds)\end{aligned}\ ] ] where the first integral is bounded by a constant @xmath738 which is arbitrarily small according to the choice of @xmath739 since @xmath740 is uniformly bounded by ( [ e : bound psi ] ) . as @xmath741 is a finite measure we can use the dominated convergence theorem and we see that @xmath742 ( 1-s_1)\nu ( ds ) = 0\ ] ] which proves that the first term converges to 0 . * second term :* using again ( [ e : useful bound ] ) to bound the difference of the two products in the second term we see , again with the help of ( [ e : bound psi ] ) and the monotonicity of @xmath62 , that @xmath743\}\\ & \qquad \le \epsilon \sum_{i\geq 2 } ( 1 - \psi(x-\log s_i))\end{aligned}\ ] ] and we can use ( [ e : bound for 1 - psi ] ) to see that @xmath744 we conclude that the integral of the second term converges to 0 as @xmath725 * third term :* let us now consider the third term @xmath745 we have already shown ( for the second term of ( [ e : two terms ] ) ) that @xmath746 and @xmath747 as @xmath748 so the integral of this term also converges to 0 . we now return to the proof of theorem [ s : proof of thm ] and show that @xmath178 where @xmath62 is as above . suppose for contradiction that there exists some @xmath141 such that @xmath749 ( a similar argument will work to refute the case @xmath750 ) . we introduce the process @xmath751 ( which is a functional of the fragmentation process @xmath752 ) @xmath753 observe that since @xmath182 is a martingale we have as in ( [ e : u(xt)-u(x,0 ) ] ) that for all @xmath754 we have @xmath755 we claim that almost surely @xmath756 indeed , we start by observing that as @xmath757 is a uniformly bounded martingale , then as @xmath758 @xmath759 so we only need to show that @xmath760 but since @xmath761 because @xmath701 is continuous ( lemma [ l : psi in t2 is in domain of l ] ) and @xmath762 , it is enough to show that @xmath763 ( where we have used that for all @xmath42 and @xmath764 ) . we thus have @xmath770 using that for any @xmath460 we have @xmath771 for some constant @xmath711 ( which , again , may change from line to line ) we see that @xmath772 and for @xmath708 small enough so that @xmath773 , @xmath774 almost surely when @xmath775 on account of the fact that @xmath776 almost surely as @xmath775 . since @xmath160 . the claim that @xmath777 now follows . the almost sure right - continuity at 0 of @xmath778 implies that the stopping time @xmath779 is almost surely strictly positive . because @xmath182 is a uniformly integrable martingale we must have @xmath780 but @xmath781 so we have a contradiction to the assumption that there exists some @xmath141 such that @xmath749 . this completes the proof of theorem [ s : proof of thm ] . kolmogorov , a. , petrovskii , i. and piskounov , n. ( 1937 ) tude de lquation de la diffusion avec croissance de la quantit de la matire at son application a un problm biologique . _ moscow univ . _ * 1 * , 125 . lyons , r. ( 1997 ) a simple path to biggins martingale convergence for branching random walk . in _ classical and modern branching processes ( minneapolis , mn , 1994 ) _ , 217221 , i m a vol . 84 , springer , new york . mckean , h.p . ( 1976 ) a correction to : `` application of brownian motion to the equation of kolmogorov - petrovskii - piskunov '' ( comm . pure appl . math . * 28 * ( 1975 ) , no . 3 , 323331 ) . pure appl . _ * 29 * 553 - 554 . neveu , j. ( 1988 ) multiplicative martingales for spatial branching processes . _ seminar on stochastic processes , 1987 ( princeton , nj , 1987 ) _ , 223242 , progress in probability and statistics , birkhuser boston , boston , ma .	we formulate the notion of the classical fisher - kolmogorov - petrovskii - piscounov ( fkpp ) reaction diffusion equation associated with a homogeneous conservative fragmentation process and study its travelling waves . specifically we establish existence , uniqueness and asymptotics . in the spirit of classical works such as mckean @xcite , neveu @xcite and chauvin @xcite our analysis exposes the relation between travelling waves certain additive and multiplicative martingales via laws of large numbers which have been previously studied in the context of crump - mode - jagers ( cmj ) processes by nerman @xcite and in the context of fragmentation processes by bertoin and martinez @xcite and harris _ et al . _ @xcite . the conclusions and methodology presented here appeal to a number of concepts coming from the theory of branching random walks and branching brownian motion ( cf . @xcite ) showing their mathematical robustness even within the context of fragmentation theory . key words and phrases : fisher - kolmogorov - petrovskii - piscounov equation , travelling waves , homogeneous fragmentation processes , product martingales , additive martingales , spine decomposition , stopping lines . msc 2000 subject classifications : 60j25 , 60g09 . j. berestycki , s. c. harris , a. e. kyprianou
according to dirac @xcite : _ `` classical mechanics must be a limiting case of quantum mechanics . we should thus expect to find that important concepts in classical mechanics correspond to important concepts in quantum mechanics and , from an understanding of the general nature of the analogy between classical and quantum mechanics , we may hope to get laws and theorems in quantum mechanics appearing as simple generalizations of well known results in classical mechanics '' . _ in the dirac approach to quantum mechanics , the carrier space is a hilbert space , i.e. it is a vector space endowed with a hermitian structure which give rise to a riemannian structure ( the real part of the inner product ) , a symplectic structure ( the imaginary part of the inner product ) and a connecting ( 1 - 1)-tensor , the complex structure @xmath0 satisfying @xmath1 . on the other hand in classical mechanics the prevailing structure is provided by the symplectic structure , it is therefore a natural question to ask what happens , in the appropriate limit as suggested by dirac , of both the complex and riemannian structures . of course a similar question can be asked with respect to the linear structure . here , however , the situation is more involved because , due to the probabilistic interpretation of quantum mechanics , one requires that only the probability density is physically meaningful and not the probability amplitude , this means that from the measurement view point , the carrier space of quantum mechanics is the complex projective space associated with @xmath2 rather than @xmath3 itself . on this manifold there is still a khler structure @xcite but the linear structure has disappeared , showing that it is not the linear structure which matters but rather a superposition ( composition rule ) for solutions , allowing to construct new solutions out of two given ones , as it is well known this phenomenon is well understood in the framework of riccati - type equations and we will not insist any further on the subject , we refer to the existing literature @xcite on the argument . as for quantum mechanics , there has been a recent proposal to define a superposition rule on pure states @xcite ( the complex projective space of @xmath3 ) which amounts to the introduction of a connection on the complex projective space . this connection is essentially given by the pancharatnam phase , which is also known in the literature as the berry phase or geometric phase.@xcite apart from the details concerning the linear structure or the superposition principle of quantum mechanics , the khler structure is always present and it is a challenge to understand what happens of the complex structure when the appropriate limit to go from quantum to classical is taken . because we want to consider the fate of the complex structure in the quantum - classical transition , it is more appropriate to deal directly with a real vector space and a complex structure @xmath0 defined on it , rather than consider the hilbert space and hide the complex structure in the defining properties of the space itself . before entering our discussion we should point out that the role played by the complex structure in quantum mechanics was already considered by h. reichenbach in his book _ philosophic foundations of quantum mechanics _ .@xcite it was also extensively considered by stueckelberg and his collaborators.@xcite afterwards we find comments by g.w . mackey @xcite and a more extensive treatment by v. cantoni . a decade ago a new analysis started @xcite also in connection with a proposal by s. weinberg @xcite on a non - linear quantum mechanics . in referring to previous papers we have not distinguished between papers actually focused on the role of the complex structure and those dealing with the role of the riemannian structure , the reason being that for us the symplectic structure is not questioned because ( as a structure ) it survives the appropriate limit to classical mechanics . therefore , with a symplectic structure granted , the complex structure and the riemannian structure are not independent and each one determines the other one . because we are mostly interested in the classical limit , we are obliged to take into account various pictures of quantum mechanics to consider the classical limit separately for each one of them . traditionally , one considers the schroedinger picture and the heisenberg picture , in which the equations of motion are considered on the space of states or on the algebra of observables respectively . here we would like to replace the heisenberg picture with what we may call the ehrenfest picture , that is we deal with quadratic functions , expectations values , rather than with operators . in this section we discuss in a slightly generalized form the quoted argument of stueckelberg which shows the necessity of introducing a complex structure @xmath0 to get indetermination relations and formulate quantum mechanics . therefore we start with a real hilbert space . an indetermination relation _ la _ heisenberg usually is a relation like @xmath4\right\rangle _ { \psi } ^{2}\ ] ] where the error operators of the _ observables _ @xmath5 and @xmath6@xmath7 are related to their commutator through the hermitian operator @xmath8 $ ] . however , on a _ real _ hilbert space , the multiplication by the imaginary unit @xmath9 in @xmath8 $ ] does not make sense . moreover , @xmath10 $ ] is a skew - hermitian operator , so that its mean value with respect to a _ real _ scalar product always vanishes . then , motivated by scaling arguments , in principle there are two possibilities : @xmath11^{2}\right\rangle _ { \psi } \\ \mu ^{2}\left\langle k(a , b)\right\rangle _ { \psi } ^{2 } \end{array } \right.\ ] ] where @xmath12 is a real number and @xmath13 a hermitian bilinear expression to be determined . the first possibility leads to a contradiction for bounded operators unless @xmath14 in fact , choosing @xmath15 in an orthonormal basis \{@xmath16 } in which @xmath5 is diagonal : @xmath17 and @xmath18 is bounded if @xmath6 is bounded , so their product vanishes while on the contrary @xmath19^{2}\right\rangle _ { \psi } = \sum\nolimits_{n}(a_{\psi } -a_{n})^{2}\left\langle \psi \|b\|\phi _ { n}\right\rangle ^{2}$ ] is strictly positive . only the existence of a complex structure @xmath0 leads to an indetermination relation : an indetermination relation ( inequality ) holds on a real hilbert space for hermitian operators @xmath5,@xmath6 ( observables ) commuting with a complex structure @xmath0 . it has the form : @xmath20\right\rangle _ { \psi } ^{2}\ ] ] the following is a detailed proof which shows that a generalization is possible . consider the following real scalar product : @xmath21 where @xmath22 is a real number and @xmath23 is an operator which has to be determined in such a way that an indetermination relation holds . expanding the preceding expression leads to @xmath24 therefore the condition is obtained : @xmath25 the expression is positive for every value of @xmath22 when : @xmath26 where @xmath27 if a complex structure @xmath0 is available , we may choose @xmath28 and look for the maximum of @xmath29 we have @xmath30_{+}\right\rangle _ { \psi } \cos \theta + \left\langle j[a , b]\right\rangle _ { \psi } \sin \theta = : \alpha \cos \theta + \beta \sin \theta\ ] ] then @xmath31 so that eventually @xmath32_{+}\right\rangle _ { \psi } ^{2}+\frac{1}{4}\left\langle j[a , b]\right\rangle _ { \psi } ^{2}\ ] ] this general form of the indetermination relation , first discovered by schroedinger @xcite and robertson @xcite in the 30 s , reduces to the weaker heisenberg relation for uncorrelated states ( see also @xcite ) . * remark * the same expression for the indetermination relation holds also on the _ complexified _ ( _ via _ @xmath0 ) hilbert space . before we have discussed the need of a complex structure @xmath0 in quantum mechanics . in this section we discuss the relevant geometric structures which appear in quantum mechanics and the relations among them . to avoid technicalities , for the time being , while we deal with general aspects , we shall confine ourselves to finite dimensional carrier spaces ; later , to consider the transition to classical mechanics , we will deal with operators acting on infinite dimensional hilbert spaces . we consider a vector space @xmath33 equipped with a complex structure @xmath0 , that is a ( 1 - 1)-tensor with the property @xmath1 . as @xmath33 is a vector space its linear structure determines -and is determined by- a vector field @xmath34 , the infinitesimal generator of dilation ( often it is also called the liouville vector field or the euler operator ) . if we introduce cartesian coordinates for @xmath33 , say @xmath35 , we have @xmath36 while for the complex structure we find @xmath37 with the property @xmath38 the existence of @xmath0 on the vector space @xmath33 implies that @xmath39 is even , say @xmath40 we denote by @xmath41 the metric tensor ( euclidean metric tensor on @xmath33 ) defined by @xmath42 with @xmath43 and @xmath44 a positive - definite quadratic function . the admissibility condition of @xmath41 and @xmath0 is stated @xcite by requiring that @xmath45 @xmath46 i.e. @xmath0 is an orthogonal transformation both at the finite level ( eq.([finj ] ) ) and at the infinitesimal level ( eq.([infinj ] ) ) . out of @xmath41 and @xmath0 we may construct a symplectic structure @xmath47 defined by @xmath48 it follows that @xmath47 is not degenerate and @xmath49 that is @xmath47 is indeed a symplectic structure . having these four structures on @xmath33 , we may consider the group of diffeomorphisms of @xmath33 and identify subgroups by requiring that one or more of the previous structures are preserved . by requiring that @xmath50 we collapse the infinite dimensional diffeomorphism group to the finite dimensional general linear group @xmath51 the requirement @xmath52 provides us with the infinite dimensional subgroup of symplectic ( canonical ) transformations . while @xmath53 gives us the finite dimensional rotation group and @xmath54 identifies the general linear group @xmath55 by taking the intersection of any two linear subgroups associated with any pair of admissible structures we get the subgroup of unitary transformations with infinitesimal generators provided by the antihermitian operators with respect to the hermitian structure @xmath56 we recall that an hermitian structure on @xmath57 as a real vector space , can be defined as a map : @xmath58 such that @xmath59 we can also induce a complex structure on @xmath33 by setting , for @xmath60 and @xmath61@xmath62 to consider the product structure on functions and more specifically on quadratic functions , it is convenient to consider directly the target space @xmath63 in eq.([targhet ] ) , as @xmath64 the complex numbers . in this way , if @xmath33 is considered as a real vector space , @xmath65 is a complex valued quadratic function of real variables . * remarks * i ) notice that if the admissibility condition of eq.([finj ] ) does not hold , we can always build a hermitian structure out of a given @xmath41 by substituting it with the symmetrized metric tensor : @xmath66 which will be positive and nondegenerate if @xmath41 is . quite similarly @xcite , if the relation of eq.([admj ] ) does not define a complex structure @xmath0 , that is if @xmath67 , then riesz s theorem tells us that there exists a nonsingular linear operator @xmath5 such that : @xmath68 and the antisymmetry of @xmath47 implies : @xmath69 i.e. that @xmath5 is skew - hermitian : @xmath70 , and : @xmath71 . let then @xmath72 be a ( symmetric ) nonnegative square root of @xmath73 . @xmath72 will be injective , so @xmath74 will be well defined is bounded and injective , and that @xmath72 is also injective and densely defined , so that @xmath75 is well defined in the infinite - dimensional case as well . ] . we define then : @xmath76 and : @xmath77 therefore : @xmath78 and : @xmath79 @xmath80 . the triple @xmath81 will be then an admissible triple , eq.([infinj ] ) will hold true for @xmath82 and , moreover : @xmath83 and eq.([finj ] ) will be satisfied as well . \ii ) the adjoint @xmath84 of any linear operator @xmath5 with respect to a metric tensor @xmath41 is defined by the standard relation : @xmath85 and we can read eq.([finj ] ) as saying that the complex structure @xmath0 is skew - adjoint with respect to the metric tensor @xmath86 although it may seem elementary , it is worth stressing here that , despite the fact that we are working in a real vector space , the adjoint of @xmath5 does * not * coincide with the transpose @xmath87 for a general @xmath41 . indeed , spelling out explicitly eq.([acroce ] ) in terms of matrices leads to : @xmath88 and therefore , even for real matrices : @xmath89 only if the metric is standard euclidean one and , in general , symmetric matrices need not be self - adjoint . to completely turn entities depending on the linear structure on the space of states into tensorial objects , we notice that with every matrix @xmath90 we can associate both a @xmath91tensor @xmath92 and a linear vector field @xmath93 the two are connected by @xmath94 and are both homogeneous of degree zero , i.e. @xmath95 the correspondence @xmath96 is a full associative algebra and a corresponding lie algebra isomorphism . the correspondence @xmath97 is instead only a lie algebra ( anti)isomorphism , that is @xmath98 while @xmath99=-x_{[a , b]}.\ ] ] moreover , for any @xmath100 : @xmath101}.\ ] ] out of the liouville vector field @xmath34 and the metric tensor @xmath41 , we can construct the quadratic function @xmath102 along with the associated hamiltonian vector field @xmath103 via the @xmath104 it is possible to show that @xmath105 or @xmath106 . the vector field @xmath103 preserves all three structures @xmath107 and @xmath0 . thus the vector field @xmath103 will be a generator of the unitary group and may be associated with a schroedinger - type equation @xmath108 where we have used the more familiar notation @xmath109 instead of @xmath110 . more generally , starting from eqs.([met ] ) , ( [ gamma ] ) and ( [ jdelta ] ) , we can construct quadratic functions associated with any linear vector field @xmath111 by defining @xmath112 then again we have @xmath113 where now @xmath114 whenever the matrix @xmath5 satisfies the condition of the following eq.([omegaa+aomega ] ) . in general , any vector field @xmath111 will define the equations of motion of a quantum systems if @xmath115 and , _ _ a fortiori , _ _ @xmath116 . this implies that the matrix @xmath5 , representative of the vector field , satisfies the relations @xmath117 while the condition @xmath118 implies that @xmath119 where , with abuse of notation , we have used @xmath47 and @xmath41 to represent a matrix associated with the corresponding tensors . by using the vector fields @xmath34 and @xmath120 , which are an involutive distribution , i.e. in particular they commute , we get out of @xmath121 the complex projective space associated with the hilbert space structure on @xmath33 defined by the hermitian form @xmath122 ( see eq.([innerprod ] ) ) , which will be denoted by @xmath123 . by construction , the unitary group acting on @xmath33 in terms of linear transformation will induce an action on @xmath123 which turns out to be a transitive action . the first outcome of our description in terms of tensors is that now instead of linear operators on @xmath123 we can use vector fields on @xmath123 which are autonomously defined without making recourse to the underlined vector space @xmath124 the main point here is that @xmath123 is not a linear space therefore it makes no sense to consider on it linear operators , one has to induce an action from linear operators acting on @xmath33 . the use of vector fields allows to bypass the construction of the action in terms of @xmath125 we have remarked that the linear action of the unitary group on @xmath33 is provided by the intersection of the symplectic group and the rotation group . because the group acts symplectically , there will be an associated momentum map from the vector space to the dual of the lie algebra , say @xmath126 in explicit terms we have @xmath127 where we have used the bra and ket notation of dirac , and identified @xmath128 with @xmath129 through a trace operation . by using our previous structures , it would be @xmath130 which acts on @xmath129 in the following way : @xmath131 for any @xmath132 to have a momentum map equivariant with respect to the action of @xmath34 and @xmath120 we may set @xmath133 or @xmath134 this map represents the momentum map associated with the symplectic action of @xmath135 on the complex projective space @xmath136 it is now clear that all vector fields associated with hermitian operators on @xmath33 will pass to the quotient @xmath123 because they commute with @xmath137 and preserve @xmath0 . with any hermitian operator @xmath5 we associate @xmath111 which projects onto @xmath138 , giving a symplectic vector field with generating function @xmath139 that is @xmath140 we observe that functions here defined may be expressed through the quadratic functions defined by eq.([quad ] ) of the previous section : they are nothing but the quadratic functions projected onto @xmath123@xmath141 it is now possible to formulate the equations of motion in terms of this mean values of the hermitian operators . the ehrenfest picture @xcite of quantum mechanics originates from the ehrenfest theorem . we may formulate it within our approach in the following way . we consider the lie algebra of the unitary group realized in terms of skew - hermitian operators on @xmath33 and we define a map @xmath142 this evaluation map has the property that @xmath143 coincides with the momentum map associated with the symplectic action of the unitary group acting on @xmath124 our aim now is to realize the algebra of linear operators on @xmath33 in the ehrenfest picture , that is by means of quadratic functions . we observe that the definition of eq.([quad ] ) in sec.(3)is not suitable to this end , because the real scalar product @xmath41 annihilates the skew - symmetric part of any operator @xmath144 therefore , we have to use the hermitian form @xmath122 defined by eq.([innerprod ] ) of sec.(3 ) and deal with complex - valued functions of real variables : @xmath145 this definition reduces to the previous one and yields real functions for hermitian operators , while associates imaginary functions to skew - hermitian operators , so that it realizes a one to one correspondence between operators and complex - valued quadratic functions . we notice that neither @xmath41 nor @xmath146 separately can be used to recover the full associative algebra : in fact @xmath47 allows to recover the lie algebra structure of the hermitian operators , but it is not able to recover the symmetrized product on hermitian operators , i.e. the jordan algebra structure @xcite existing on the space of hermitian operators . to recover this product we have to use the metric tensor @xmath41 . it is only the use of the entire admissible triple @xmath147 or the equivalent hermitian structure @xmath122 that allows to give a complete description of the full algebra of operators in the ehrenfest scheme . we now introduce two brackets on quadratic functions corresponding respectively to lie and jordan operator algebras and from them recover the associative operator product.@xcite in fact , it is well known that on the symplectic vector space @xmath148 we can define a poisson tensor @xmath149 which is the inverse of @xmath150 : in coordinates we have @xmath151 with @xmath152 . the associated poisson bracket is given by @xmath153 on quadratic functions we have the following result : @xmath154}(x)=-if_{[a , b]}(x)\ .\ ] ] in fact , supposing @xmath155 hermitian operators , in coordinates @xmath156 then , bearing in mind that @xmath157 we have : @xmath158 @xmath159 @xmath160x , x)=-\frac{i}{2}h([a , b]x , x)=-if_{[a , b]}(x).\ ] ] moreover @xmath161}(x)\ = \frac{1}{2}h([a , b]x , x)=\frac{i}{2}\omega ( [ a , b]x , x)=\frac{i}{2}g(j[a , b]x , x)=if_{j[a , b]}(x)\ ] ] the proof is analogous in the cases when one or both operators @xmath5 and @xmath6 are skew - symmetric , due to the relation @xmath162 . by using in a similar way the inverse of the metric tensor @xmath163 say @xmath164 with @xmath165 , we may define a new bracket ( we shall call it the riemann - jordan bracket ) on functions as : @xmath166 by using quadratic functions we get the relation @xmath167 in fact , assuming again @xmath168 hermitian operators , and bearing in mind that @xmath169 , we have in coordinates @xmath170 @xmath171_{+}x , x)=\ ] ] @xmath172_{+}x , x)=f_{(ab+ba)}(x).\ ] ] finally , we introduce an associative product on quadratic functions as @xmath173 so that @xmath174 and @xmath175 so we obtain the following result : @xmath176 which is the analog of the operator decomposition @xmath177_{+}+% \frac{1}{2}[a , b]$ ] . it is also apparent that the following relation which connects gradient and hamiltonian fields holds ( see eq.([grad - ham ] ) of sec.(3 ) ) : @xmath178 equations of motion in this picture have the classical - like form @xmath179 within the one - to - one correspondence between operators and quadratic functions , we recover the heisenberg equations of motion in hamiltonian form . we conclude this section by observing that the @xmath180 product is reminiscent of the weyl - wigner approach and the moyal bracket . it shows also that a full isomorphism of algebras between quadratic functions and operators requires the introduction of a non - local product . summarizing , we have replaced the initial description provided by dirac in terms of hilbert spaces and operators acting on them with a description on the differentiable manifold of pure states ( the complex projective space ) in terms of vector fields and functions . it is clear that if we do not have a linear structure we have no meaning of quadratic functions , therefore the image of quadratic functions on @xmath33 shall be identified autonomously . this identification is achieved by requiring that hamiltonian vector fields associated with them should be killing vectors with respect to the metric structure on @xmath123 arising from the khler structure . this ends up being the infinitesimal version of wigner s to consider the quantum - classical transition , the best way is to consider a description which uses the same carries space for both classical and quantum descriptions , i.e. smooth functions on the phase space . the classical picture uses the point - wise product , while the quantum picture uses the @xmath181product . in particular we shall restrict to a phase - space which is a symplectic vector space and the @xmath182-product is the moyal product . given a symplectic vector space @xmath183 , a weyl system is defined to be a strongly continuous map from @xmath184 to unitary transformations on some hilbert space @xmath3 : @xmath185 with @xmath186 with @xmath187 the identity operator . consider a lagrangian subspace @xmath188 and an associated isomorphism @xmath189 on @xmath188 we consider square integrable functions with respect to a lebesgue measure on @xmath188 , a measure invariant under translations . the splitting of @xmath184 allows to define @xmath190 and set @xmath191 @xmath192 @xmath193 it is obvious that @xmath194 are unitary operators and moreover they satisfy weyl condition of eq.([weyl ] ) with @xmath47 being the canonical one on @xmath195 . the strong continuity allows to use stone s theorem to get infinitesimal generators @xmath196 such that @xmath197 and @xmath198 for any @xmath199 . when we select a complex structure on @xmath184 @xmath200 we may define creation and annihilation operators by setting @xmath201 @xmath202 by using this complex structure on @xmath184 we may construct an inner product on @xmath184 as @xmath203 therefore creation and annihilation operators are associated with a khler structure on @xmath204 the weyl map can be extended to functions on @xmath205 ; indeed we first define the symplectic fourier transform of @xmath206 @xmath207 @xmath208 and then associate with it the operator @xmath209 defined by @xmath210 _ vice versa _ , with any operator @xmath5 acting on @xmath3 we associate a function @xmath211 on the symplectic space @xmath184 by setting@xmath212 this map is called the wigner map and , _ via _ a symplectic fourier transform , is the inverse of the weyl map . a new product of functions may be introduced on @xmath213 by setting @xmath214 we thus find that a symplectic structures on @xmath184 give rise to an associative product on @xmath215 which is not commutative . the dynamics on @xmath213 can be written in terms of this non - commutative product as @xmath216 having clarified the setting , let us consider a very simple situation to actually carry on the limiting procedure . without any loss of generality we may limit to consider a bidimensional symplectic vector space @xmath217 the map @xmath218 @xmath219 given by @xmath220 induces a moyal product @xmath221 by using the specific form of @xmath222 it is possible to provide an explicit form in terms of bidifferential operators . we have @xmath223 \,f_{b}(x , y)\ ] ] where a standard notation for physicists has been used , i.e. @xmath224 and @xmath225 mean that the operators act on the left or on the right , respectively . alternatively , we may write @xmath226 \,f_{a}(x^{\prime } , y^{\prime } ) f_{b}(x , y)\ ] ] where the bidifferential operator @xmath227 has been introduced@xmath228 _ { x^{\prime } = x , y^{\prime } = y}\ ] ] the action of @xmath227 may be expressed in terms of the poisson bracket of two functions @xmath229 as : @xmath230 @xmath231 and in general ( all @xmath232 s commute among themselves ) @xmath233 so that @xmath234 from which we see that the quantum corrections to the point - wise abelian product are expressed by means of the complex structure times the planck s constant and the poisson bracket of the two functions and of all their derivatives . now , even or odd powers of @xmath227 are respectively symmetric or skewsymmetric with respect to the exchange of @xmath211 and @xmath235 , so that the commutator of @xmath211 and @xmath235 contains only the odd powers of @xmath236 @xmath237\ ] ] and the quantum equation of motion of @xmath235 with respect to the hamiltonian @xmath238 reads @xmath239\ ] ] and finally , dropping terms @xmath240 we recover the classical equations of motion in the hamiltonian form : @xmath241 when the hamiltonian @xmath238 is a quadratic polynomial , the @xmath242 terms vanish , so this result is exact and we recover the ehrenfest s theorem . this appropriate classical limit shows very clearly that while in quantum mechanics the associative and non commutative product among functions determines all the various structures , i.e. the complex structure and the poisson bracket , because the lie product and therefore the jordan product are uniquely determined by the operator - product @xcite , in the classical picture the point - wise product does not fix a unique compatible poisson bracket . indeed all possible poisson tensors on phase space define derivations for the point - wise product on functions on the phase space . to exhibit more clearly this aspect , i.e. the role of the symmetric tensor in the definition of the @xmath180product we use a slightly modified product with respect to the one given by moyal . this generalization arises from the fact that one may give a different , more general , definition of weyl systems where different ordering procedures are considered instead of the fully symmetrized one provided by the weyl prescription . it is the following one @xcite @xmath243 in analogy with this modification we may consider a @xmath180product given by @xmath244 \,f_{a}(x^{\prime } , y^{\prime } ) f_{b}(x , y)\ ] ] where the bidifferential operator @xmath245 has been introduced : @xmath246 _ { x^{\prime } = x , y^{\prime } = y}.\ ] ] and we have used a dimensionless deformation parameter @xmath22 instead of @xmath247 now we get @xmath248 we consider the classical limit of the lie product and the classical limit of the jordan product to find @xmath249 @xmath250 respectively . we see immediately that if the product should reduce to the commutative point - wise product the symmetric bidifferential operator can not be there . let us summarize our line of argumentation . in quantum mechanics the product of operators ( observables ) uniquely defines the lie product compatible with it ( i.e. such that it defines derivations of the given product ) . if we associate operators with functions ( a necessary replacement if we want to deal with physical pure states , points of the complex projective space ) , the product among operators induces a @xmath251product on functions . in the weyl - wigner approach this product can be expressed by means of bidifferential operators ( generated by a poisson tensor in the case of the weyl ordering and by a poisson and a metric tensors in the case of normal or anti - normal ordering ) . in the classical limit , the product gives rise to a point - wise product which does not privilege any specific poisson bracket . however , a poisson bracket may be selected by considering the classical limit of the lie product . a similar classical limit for the jordan product would anyway start with the point - wise term and therefore would not give rise naturally to a symmetric tensor on phase - space . our argument has been presented without privileging any dynamical system . it is however true that if we consider special systems like the harmonic oscillator , by considering the limit of operators realized in terms of creation and annihilation operators , we would get a classical phase - space described in terms of complex coordinates . however , these would not exist for generic hamiltonian systems . here the the classical limit would require first that the operators are realized as differential operators on some `` configuration space '' , acting on square integrable functions defined on the same configuration space . the classical limit would correspond to the replacement of the differential operators with their symbols @xcite and the complex structure would appear only at the level of the so called `` quantum potential '' having @xmath252 as coefficient . moreover , out of the coupled equations for the real and the imaginary part of the wave function , we would obtain only one equation ( for @xmath253 ) represented by the hamilton - jacobi equation . a complete solution for this equation would allow for the construction of a solution for the other one ( the transport equation).@xcite e.c.g . stueckelberg , quantum theory in real hilbert space , _ helv phys acta _ * 33 * ( 1960 ) , 727 - 751 ; + e.c.g . stueckelberg and m. guenin , quantum theory in real hilbert space ii ( addenda and errata ) , _ helv phys acta _ * 34 * ( 1961 ) , 621 . v. cantoni , generalized transition probability , _ commun . * 44 * ( 1975 ) , 125 - 128 ; + v. cantoni , the riemannian structure on the states of quantum - like systems , _ commun . * 56 * ( 1977 ) , 189 - 193 ; + v. cantoni , intrinsic geometry of the quantum - mechanical phase space , hamiltonian systems and correspondence principle , _ lincei - rend . sc . fis . mat . _ * lxii * ( 1977 ) , 628 - 636 . r. cirelli , a. mani and l. pizzocchero , quantum mechanics as an infinite - dimensional hamiltonian systems with uncertainty structure : part i , _ j. math . * 31 * ( 1990 ) , 2891 - 2897 ; r. cirelli , a. mani and l. pizzocchero , quantum mechanics as an infinite - dimensional hamiltonian systems with uncertainty structure : part ii , _ j. math * 31 * ( 1990 ) , 2898 - 2903 . s. weinberg , testing quantum mechanics , _ ann . _ * 194 * ( 1989 ) , 336 - 386 ; + s. weinberg , precision tests of quantum mechanics,_phys . * 62 * ( 1989 ) , 485 - 488 ; + s. weinberg , weinberg replies,_phys . rev . _ * 63 * ( 1989 ) , 1115 . robertson , a general formulation of the uncertainty principle and its classical interpretation , _ phys . rev . _ * 35 * ( 1930 ) , 667 ; + h.p . robertson , an indeterminacy relation for several observables and its classical interpretation , _ phys . * 46 * ( 1934 ) , 794 - 801 . sudarshan , c.b . chiu , and g. bhamati , generalized uncertainty relations and characteristic invariance for the multimode states , _ phys . rev . a _ * 52 * ( 1995 ) , 43 - 54 . g. marmo , g. morandi , a. simoni and f. ventriglia , alternative structures and bi - hamiltonian systems , _ j.phys . a : math.gen._ * 35 * ( 2002 ) , 8393 - 8406 ; + a. canas da silva , _ lectures on symplectic geometry _ ( springer , berlin , 2001 ) . p. aniello , v.i . manko , g. marmo , s . solimeno , f . zaccaria on the coherent states , displacement operators and quasidistributions associated with deformed quantum oscillator , _ jour . of optics b _ * 2*(2000 ) , 718 - 725 .	according to dirac , fundamental laws of classical mechanics should be recovered by means of an `` appropriate limit '' of quantum mechanics . in the same spirit it is reasonable to enquire about the fundamental geometric structures of classical mechanics which will survive the appropriate limit of quantum mechanics . this is the case for the symplectic structure . on the contrary , such geometric structures as the metric tensor and the complex structure , which are necessary for the formulation of the quantum theory , may not survive the classical limit , being not relevant in the classical theory . here we discuss the classical limit of those geometric structures mainly in the ehrenfest and heisenberg pictures , i.e. at the level of observables rather than at the level of states . a brief discussion of the fate of the complex structure in the quantum - classical transition in the schroedinger picture is also mentioned .
coalescing binaries are the most promising sources of gravitational waves @xcite for laser interferometric gravitational wave detectors . the basic reason for the importance of these type of sources is their broadband nature which makes them ideally suited for their detection by the interferometers . the binary systems which are of relevance here are those consisting of compact objects i.e. black holes and neutron stars . it has been estimated that three such coalescences occur per year out to a distance of 200 mpc @xcite . a lot of attention has recently been focussed on the issues of detecting the presence of the signal and the extraction of astrophysical information from the estimated parameters of the signal . there are plans to construct such laser interferometers around the globe and by the end of this century the american ligo @xcite and french / italian virgo @xcite will be in operation . the emphasis in their construction is on the reduction of noise which may be thermal , seismic , quantum , or photon shot noise . in laser interferometric detectors the lower cuttoff is decided by the seismic noise which is very dominant at low fequencies . it is expected that the ligo will be able to go down to 40 hz in its initial stage and to 10 hz in its final stage . this means that that we can start observing the binary when its orbital frequency is 20 hz in the case of the initial detectors and 5 hz in the case of the advanced ones . this leads to sufficiently large integration times which enhances the signal to noise ratio . it was suggested by thorne that matched filtering would be an ideal filtering technique for this purpose . matched filtering is a standard technique used in signal analysis when the waveform is known . it determines for us an optimal linear filter which can decide on the presence or absence of the signal waveform in a given data train @xcite . this requires accurate modelling of the waveforms , which is possible for the coalescing binary systems . they are clean systems and their inspiral waveform depends on a few parameters such as the individual masses and spins . tidal interactions do not matter until the very end of the inspiral @xcite . a lot of research activity has gone in the direction of obtaining accurate templates under the various approximation schemes such as the quadrupole and the post - newtonian @xcite . recently it has been shown that post - newtonian ( pn ) corrections , spin - orbit ( s.o . ) and spin - spin ( s.s . ) couplings , produce in the waveform an accumulating phase error as compared to the newtonian expression @xcite . therefore , a template constructed from the newtonian waveform would go out of phase with the signal and the so called `` matched filtering '' technique for detection would woefully fail . in this paper we show that as long as we are only _ searching _ for signals a newtonian filter would perform remarkably well even though the signal contains pn corrections . the key idea here is that we allow the parameters of the newtonian filter to vary and adjust so as to produce the maximum possible correlation with the signal . we have found that this flexibility allows for fairly high values of the correlation . in many cases of interest the correlation obtained is 70% of its maximum possible value which would have been obtained had the template been perfectly matched to the signal . on the other hand , a template with the same parameters as those of the signal produces correlations of about 10 to 20% . we have carried out the analysis for the two noise curves assuming a ligo type detector . the two noise curves are the power spectral densities of the noise for the ligo in its initial and advanced stages , as given in @xcite . in the case of the initial ligo detector the analysis is also carried out for the case of white noise for the sake of comparision . also a correspondence between the parameters of the filter and the signal could be set up , it might be possible to estimate the parameters of the signal from those of the filter . in other words the filter parameters may be `` renormalized '' . the paper is divided as follows . in section [ match ] we elaborate on the chirp waveform , and the conventional detection strategy . we discuss the technique of matched filtering and define a quantity which shall be a measure of how well a newtonian waveform can match with a post - newtonian signal.we also make some comments about the signal power spectrum . in section [ numeric ] we discuss the numerical results of the simulations carried out . and finally in section [ end ] we summarise our results and indicate future directions . the waveform of the signal from the coalescing binary system henceforth called the ` chirp ' has been modelled under various approximations . in the quadrupole approximation the chirp has three parameters other than the amplitude . these are the initial phase @xmath0 , the time of arrival @xmath1 ( i.e the time at which the instantaneous frequency of the equals the lower cuttoff of the detector ) and the coalescence time @xmath2 which form a convenient set of parameters for our purpose @xcite . the newtonian waveform @xmath3 is given by , @xmath4 + \phi_0],\ ] ] where @xmath5 and @xmath6 here the lower cuttoff frequency is denoted by @xmath7 , and @xmath8 is called the chirp mass where @xmath9 is the total mass and @xmath10 the reduced mass of the binary system . @xmath11 is the solar mass and it is convenient unit for our purpose . given the form of the signal and the statistical description of the noise one has to design an adequate set of filters to detect the signal . the noise is assumed to be stationary and is further specified by its power spectral density @xmath12 which is defined by the relation , @xmath13 where @xmath14 is the fourier transform of a particular realization of noise and the overbar indicates an ensemble average . the @xmath12 defined above is the two sided power spectral density . we are primarily in search of a filter with an impulse response @xmath15 which correlates best with the signal i.e. when the correlation as defined below takes its maximum value for a particular value of time shift @xmath16 . @xmath17 this implies that the fourier transform of the matched filter @xmath15 to detect the signal @xmath18 is given by the relation , @xmath19 for the numerical computations that follow we use the fast fourier transform algorithm as given in _ numerical recipes _ @xcite . the definition of the fourier transform is the same as given there i.e. @xmath20 the impulse response of the filter @xmath15 depends on the parameters @xmath2 , @xmath1 , @xmath0 . it also depends the time shift @xmath16 . it now becomes important to judiciously space out the filters in the parameter space keeping in mind the constraints of computing power . such an analysis has been carried out in great detail for both white and coloured noise by sathyaprakash and dhurandhar ( see @xcite ) . we discuss briefly their major results : 1 . it was found that @xmath2 the coalescence time is a convenient parameter to use since the filters are equally spaced in this parameter , where the spacing is decided by a fixed drop in the correlation . 2 . for the phase @xmath0 we require just two filters for every value of the @xmath2 parameter , one with @xmath21 and the other with @xmath22 . due to their orthogonality the correlation is maximised over the phase by simply taking the square root of the sum of squares of the individual correlations , i.e. @xmath23 where @xmath24 and @xmath25 are the correlations corresponding to filters with phases @xmath21 and @xmath22 respectively . we assume @xmath26 in the design of the filter and therefore the value of @xmath16 for which the maximum of the correlation occurs is equal to the time of arrival of the signal . such a procedure of maximising the correlation over the phase and the time is carried out for each value of @xmath2 . the final maximization of the correlation is then carried over the @xmath2 parameter . the set of parameters for which the correlation is maximum are then presumed to be the most likely values of the parameters of the gravitational wave signal . the post - newtonian corrections lead to corrections to the phase and the amplitude of the newtonian signal and also lead to additive terms which are qualitatively different from the quadrupole term . in the case of a general binary system it is tedious and difficult to get the various corrections to the evolution of the orbits of the binary . if one of the bodies is large compared to the other as in the case of a black hole neutron star binary system one can apply the regge wheeler perturbation formalism @xcite to get the evolution of the orbit . this provides us with the evolution of the orbital frequency as a function of time . this has been worked out @xcite and is given by @xmath27 where , @xmath28 here @xmath29 represents the first time derivative of frequency and @xmath30 the pn expansion parameter . the newtonian waveform is obtained from the above equation by setting @xmath31 . the phase is obtained by integrating equation ( [ orbev ] ) . for the amplitude we use the newtonian dependence on the frequency _ @xmath32 const@xmath33 . this waveform shall be called ` restricted post - newtonian ' henceforth . although this is not exact , we do not expect the errors in the amplitude to affect the correlation significantly . we assume the initial phase and the arrival time of the signal to be 0 . as the matched filtering process can also be viewed as a correlation between the incoming signal and the filter it is evident that any secular growth of phase difference will reduce the correlation drastically . thus to have a matched filter one must add one or more parameters . this would increase the number of filters enormously with corresponding increase in computational time . it is worthwhile to explore whether we can substantially increase the correlation by allowing for a shift in the parameters of the newtonian filter i.e. whether the signal is able to achieve better correlation with a newtonian filter whose parameters are different from those of the signal . obtaining large correlations depends on the function space spanned by the signal and filter waveforms and to what extent they overlap . we obtain reasonably large correlations . here effects due to s.o . and s.s . coupling are not taken into account . the addition of such terms will not alter the thrust of the argument in that , some other newtonian filter would perform best . a detailed account of the formalism and notation used here and the theory of hypothesis testing using maximum likelihood methods as applied to detection of gravitational waves from coalescing binaries is given in @xcite . we define a scalar product and its corresponding norm in the function space between two functions @xmath34 and @xmath15 for future use ; @xmath35 and @xmath36 if an exact matched filter were present then the signal to noise ratio @xmath37 would be simply equal to @xmath38 where @xmath39 is the signal . note that our definition of snr is different by a factor of two from the one given in @xcite as they work with the one sided power spectral density . the quantity we are interested in computing is @xmath40 where @xmath18 is the chirp corresponding to the filter . we shall term @xmath41 as the normalized correlation . henceforth when we use the word correlation we shall mean the quantity @xmath41 unless specified otherwise . as mentioned above the initial phase and the time of arrival of the signal is taken to be 0 . the aim is to maximise @xmath41 over the range of parameters of the filter . the quantity @xmath41 takes the value between @xmath42 and @xmath43 and tells us how well a newtonian filter can substitute for a post - newtonian one . geometrically one can visualize @xmath41 as the cosine of the angle between the signal vector and the chirp vector . in figure [ fig1 ] we show the impulse response of a filter . as the noise is very high at lower frequencies the amplitude of the impulse response is very small at earlier times and becomes appreciable only at the end . due to the same reason the increase of amplitude with time is different from that of the newtonian chirp . figure [ fig2 ] justifies the high correlations obtained . it shows how well the filter matches with the restricted post - newtonian signal . it is to be noted that the filter matches with the signal very well during the late stages where the amplitude is largest . figure [ fig3 ] shows the power spectrum of the signal which is the square of the magnitude of the fourier transform of the signal divided by the power spectral density of noise as a function of frequency , @xmath44 this quantity peaks near 200 hz and it is in this frequency range therefore that the filter must match the signal very well i.e. it should try to keep in phase with the signal to yield a high correlation . this is borne out in figure [ fig2 ] . the stationary phase approximation used in the fourier transform of the chirp waveform @xcite predicts the power spectrum to be a smooth power law . this however is not true , and numerical results and further investigations into the stationary phase approximation bears this out . the stationary phase analysis leads to a fresnel integral and the smooth power law fall off ( @xmath45 ) in the power spectrum is obtained if the limits of integration extend from @xmath46 to @xmath47 . however since the chirp waveform is taken to be of finite duration we actually get an incomplete fresnel integral . this leads to oscillations in the power spectrum of the signal . the oscillations are pronounced at the two ends of the bandwidth of the power spectrum of the signal as the limits of the integration are curtailed from the ideal @xmath46 to @xmath47 . as the noise is very high at low frequencies the amplitude of the oscillations in @xmath48 is very less at low frequencies . it is easy to explain these oscillations using the cornu s spiral @xcite . the cornu s spiral does not get wound up before the limits of the integration are reached . the thickness of the line indicates the presence of sub structure in the power spectrum . the signal waveform was obtained by numerically integrating equation ( [ cut ] ) . we get time as a function of frequency which we then invert to get frequency as a function of time . this is now used to generate the phase and the amplitude of the signal . we take the initial phase and the time of arrival of the signal to be zero . we now present the results of the numerical simulations . we have considered black hole masses in the range 5 to 10 @xmath11 . the masses taken for the smaller mass are 0.5 , 1.0 , and 1.4 @xmath11 . the analysis has been carried out for the ligo detector both in the initial and the advanced stages . we retain @xmath2 as a parameter for the restricted post - newtonian waveform also defined by the equation ( [ eqxi ] ) though this quantity does not represent the coalescence time of the signal anymore . in general the amount of time the signal lasts is less for the post - newtonian signal as compared with the newtonian one which follows from the fact that the quantity @xmath49 in equation ( [ cut ] ) is less than one in the frequency range considered . in table [ tab1 ] we list the normalised correlations obtained for the ligo detector in the initial stage . the mass of the larger component of the binary ( @xmath50 ) increases from left to right along each row . the mass of the other component ( @xmath51 ) increases from top to bottom in each column . the values of the masses are listed accordingly in the table . the correlations show a very regular behavior in the table . there are two factors controlling the drop of the correlation : 1 . increase of the magnitude of phase corrections with increase of total mass of the binary system , and 2 . decrease of the integration time due to the increase of total mass of the system . these two factors work against each other in producing the total amount of phase error between the newtonian filter and the restricted post - newtonian signal . thus when we increase @xmath50 , the increase in the magnitude of the phase corrections dominates over the loss in integration time and we get lower correlations when we go from left to right . exactly the opposite happens when we increase @xmath51 i.e. the effect of the decrease of integration time dominates and the correlations increase . in table [ tab3 ] we list the correlations for the advanced ligo . the same pattern is observed in this table too . we observe that the correlations for the larger frequency range are smaller as may be expected since the filter is more likely to go out of phase in a broader bandwidth . however , it should be emphasised that these correlations are _ normalised_. the correlation would be unity if the filter were exactly matched to the signal . in absolute terms , if we consider a signal with given parameters having the same amplitude then the correlation for the advanced ligo will be much larger than the initial ligo since the noise is less ; first , we get a larger integration time and second , the power spectral density is an order of magnitude less in the common bandwidth . we find that for the parameters considered , the absolute values of the correlations are larger by a factor of twenty for the advanced ligo . we next take up the issue of the shift in the parameters of the filters which produce maximum correlations . in table [ tab2 ] we list the shift in the parameters @xmath2 and @xmath1 for the case of the initial ligo detector . the phase parameter @xmath0 is an extremely sensitive parameter and its shifts are not regular . the value of @xmath52 is always negative . this is because as mentioned earlier a post - newtonian signal will last for a smaller length of time as compared to a newtonian signal with the same values of @xmath50 and @xmath51 . also it can be seen from equation ( [ orbev ] ) and the definition of @xmath2 that the first derivative of the frequency @xmath29 is approximately proportional to @xmath53 . therefore in order to obtain a higher value of @xmath29 the value of @xmath2 is reduced . here again the factors , the integration time and magnitude of phase corrections , compete against each other in determining how @xmath52 varies with an increase in either of the masses . the value of @xmath52 decreases with increase in @xmath50 and increases with increase in @xmath51 . also @xmath54 is always negative . this parameter tries to compensate for the reduction in the coalescence time by pushing the filter forward in time . as the table shows there is apparently a very strong covariance between these two parameters . the value of @xmath54 also decreases with increase in @xmath50 and increases with increase in @xmath51 . typically for @xmath55 , and @xmath56 we get shifts of @xmath57 secs and @xmath58 secs . this should be compared with the coalescence time of the waveform which is about 9 secs . for the case of the advanced ligo detector the magnitude of the shifts is much more , but the time for which the signal spends in the frequency range 10 to 400 hz is about a factor of fifty more than that for the initial ligo for similar masses . here the same pattern is observed in the variation of @xmath52 and @xmath54 as in the initial ligo . the typical values of the shifts observed are @xmath59 secs and @xmath60 secs for @xmath55 , and @xmath56 . simulations were also done for the band limited white noise with the power spectral density having a constant value between 40 to 400 hz . the results were compared with those of the initial stage of ligo . the effect of coloured noise of the type considered here is to narrow band the signal . thus the newtonian filter has to match with the signal over a smaller range of frequencies . however if the narrow banding occurs at higher frequencies , for the chirp , the magnitude of the phase corrections is more . thus in addition to changing the first derivative of the frequency through the parameter @xmath2 the values of the higher derivatives of the frequencies would also have to be changed to get a good match . had the narrowbanding been at lower frequencies where the time derivatives of the frequency are relatively less , the correlation would have been much higher as the shifts in the newtonian parmeters would have been sufficient for the purpose . in table [ tab5 ] we show the correlations obtained for band limited white noise and table [ tab6 ] shows the corresponding parameter shifts . we observe that in the case of the initial ligo the correlations obtained are less than those for the white noise case for higher values of the total mass and vice - versa . this can also be seen as an effect of narrow banding . the values of the shifts in the parameters is also much smaller in the case of white noise as a most of the contribution to the correlation comes from the lower frequencies where a small shift in @xmath2 is sufficient for the filter to match well with the signal . till now we have considered our filter bank to have an infinite number of filters i.e. we have allowed for a continuous variation of @xmath2 . however one is limited by the computing power available and one must confine oneself to a finite number of filters . thus in general the signal will be unable to achieve its maximum correlation . our aim is to estimate the drop in the correlation for a given computing speed . the maximum drop in the correlation because of the finiteness of the filter bank will have to be kept small . we consider a discrete set of newtonian filters corresponding to distinct values of the @xmath2 parameter . the filter spacing in the @xmath2 parameter is taken to be constant ( @xmath61 ) across the entire range of values @xmath2 can take ( see @xcite ) . we first consider the initial ligo and assume a 1 gigaflop machine on which we intend to do on - line search . the maximum time the signal lasts is found to be 25 secs for the mass range considered . however the data train needs to be padded with zeroes to four times the original length which is optimal for computational purposes ( see @xcite ) . this will increase the length of the data train to 100 secs . we allow for an overlap of 25 secs between consecutive data trains.thus we have 75 secs in which to calculate @xmath62 correlations where @xmath62 is the number of filters . we sample the waveform at 1000 hz . thus we get approximately @xmath63 data points per data train . we have to perform one fast fourier transform ( fft ) operation per filter . the fourier transforms will have already been calculated once and for all for the filters in the bank and one inverse fourier transform will have to be performed to obtain the correlation as a function of the time lag @xmath64 . the computation time will be mostly taken up by the ffts as each fft involves @xmath65 operations where @xmath66 is the number of points in the data train . in this particular case therefore each fourier transform will require about 6.4 million floating point operations ( mfo ) . this has to be compared with the number of floating point operations which can be carried out over the period of 75 secs which is @xmath67 mfos . thus the number of filters that can be accommodated is about 11700 filters . we require two filters for the phase for each value of @xmath2 . as the maximum value of the coalescence time @xmath2 is 25 secs for the range of masses considered , we get a filter spacing in the @xmath2 parameter to be around 4.3 msecs . in the case of the advanced ligo this number is about 172 msecs . let @xmath68 where @xmath69 is the value of @xmath52 corresponding to the maximum correlation . figure [ fig4 ] shows how the correlation for a given signal normalized to its maximum value varies with @xmath70 along a line of curvature i.e. along the curve parameterised by @xmath70 along which the drop in the correlation is the slowest . in other words the figure shows the correlation maximized over @xmath54 and @xmath71 as a function of @xmath70 . the curve has been plotted for the initial ligo and for @xmath72 , and @xmath73 . however the shape of this curve and the magnitudes in the drop of the correlation is insensitive to the values of @xmath50 and @xmath51 . we observe that even for shifts of 100msecs the correlation does not drop by more than @xmath74 . for the case of the advanced ligo this drop in the correlation is even lower . thus the filter spacing which we have calculated is sufficient for our purpose . we have demonstrated here the possibility of using newtonian filters for detecting the presence of a restricted post - newtonian signal . such a strategy would be very useful in providing a preliminary on line analysis of the data train . the analysis which we have carried out here is valid only for the point mass case where @xmath75 where @xmath10 is the reduced mass and @xmath9 the total mass . for the initial ligo the correlation is 0.65 on an average and for the advanced ligo it is around 0.45 . these are only the normalised correlations as we have already stressed before . the absolute values of the signal to noise will be much higher ( by a factor of about 20 ) for the advanced ligo . it must be noted that the drop of the correlation will translate into a loss in the event rate . the distance upto which we can detect the binary will come down by a factor equal to the normalised correlation . this means that for the initial ligo the distance to which we can detect the binary will be brought down by @xmath76 and for the advanced ligo it will be brought down by @xmath77 from their respective maximum ranges . in absolute terms the advanced ligo will still be able to look further than the initial ligo . the effect of the discreteness of the filter bank in producing a further drop in the correlations was investigated . it was found that for a one gigaflop machine the drop in correlation due to the discreteness was very small . with better and faster machines we can make the bank of filters still more efficient . if we consider higher derivatives of frequency @xmath78 say @xmath79 _ etc . _ as parameters @xcite we should get a better match , but the computation is very likely to increase . it should be possible to construct filters which not only enable us to save on the computation time but also span the set of signal waveforms adequately . a deeper analysis of the signal waveforms is in order so that efficient techniques can be developed . this work is now in progress . .this table displays the correlations for the initial ligo detector for a wide range of masses in units of solar masses . the black hole mass varies from 5 to 10 @xmath11 and the other mass takes the values 0.5 , 1.0 and 1.4 @xmath11 . [ tab1 ] [ cols="^,^,^,^,^,^,^ " , ]	as coalescing binary systems are one of the most promising sources of gravitational waves , it becomes necessary to device efficient detection strategies . the detection strategy should be efficient enough so as not to miss out any detectable signal and at the same time minimize the false alarm probability . the technique of matched filtering used in the detection of gravitational waves from coalescing binaries relies on the construction of accurate templates . until recently filters modelled on the quadrupole or the newtonian approximation were deemed sufficient . such filters or templates have in addition to the amplitude , three parameters which are the chirp mass , the time of arrival and the initial phase . recently it was shown that post - newtonian effects contribute to a secular growth in the phase difference between the actual signal and its corresponding newtonian template . this affects the very foundation of the technique of matched filtering , which relies on the correlation of the signal with the filter and hence is extremely sensitive to errors in phase . in this paper we investigate the possibility of compensating for the phase difference caused by the post - newtonian terms by allowing for a shift in the newtonian filter parameters . the analysis is carried out for cases where one of the components is a black hole and the other a neutron star or a small black hole . the alternative strategy would be to increase the number of parameters of the lattice of filters which might prove to be prohibitive in terms of computing power . we find that newtonian filters perform adequately for the purpose of detecting the presence of the signal for both the initial and the advanced ligo detectors .
to discuss actually the nature of the putative nonet of the light scalar mesons : the red putative @xmath3 ( or @xmath4 ) and @xmath5 mesons and the red well established @xmath2 and @xmath1 mesons , one should explain not only their mass spectrum , particularly the mass degeneracy of the @xmath2 and @xmath1 states , but answer the next real challenges . 1 . the copious @xmath6 decay and especially the copious @xmath7 decay , which looks as the decay plainly forbidden by the okubo - zweig - izuka ( ozi ) rule in the quark - antiquark model @xmath8 . 2 . absence of @xmath9 and @xmath10 with copious @xmath11 , @xmath12 , @xmath13 if @xmath1 and @xmath2 are @xmath14-wave states of @xmath15 like @xmath16 and @xmath17 respectively . absence of @xmath18 with copious @xmath19 and @xmath20 if @xmath2 is @xmath14-wave state of @xmath15 like @xmath17 or @xmath21 . 4 . suppression of @xmath22 and @xmath23 with copious @xmath24 , @xmath25 , @xmath26 if @xmath1 and @xmath2 are @xmath14-wave state of @xmath15 like @xmath16 and @xmath17 respectively . unfortunately , i have opportunity to dwell on the @xmath0 radiative decays only in this talk . study of the nature of light scalar resonances has become a central problem of non - perturbative of qcd . the point is that the elucidation of their nature is important for understanding both the confinement physics and the chiral symmetry realization way in the low energy region , i.e. , the main consequences of qcd in the hadron world . @xmath27 @xmath28 @xmath29 mixes left and right spaces @xmath30 and @xmath31 . but in chiral limit @xmath32 these spaces separate realizing @xmath33 flavour symmetry , @xmath34 , which , however , is broken by the gluonic anomaly up to @xmath35 . 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 ocd at low energy . 1 . non - linear @xmath36-model . @xmath37 2 . linear @xmath36-model . @xmath38 the experimental nonet of the light scalar mesons , the putative @xmath3 ( or @xmath4 ) and @xmath39 mesons and the well established @xmath2 and @xmath1 mesons as if suggests the @xmath33 linear @xmath36-model . hunting the light @xmath36 and @xmath40 mesons had begun in the sixties already and a preliminary information on the light scalar mesons in particle data group ( 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 @xmath36 and @xmath40 mesons was the fact that both @xmath41 and @xmath42 scattering phase shifts do not pass over @xmath43 at putative resonance masses . situation changes when we showed ( 1994 ) that in the linear @xmath36-model @xmath44 + \left(\mu^2/2\right)\left [ ( \sigma)^2+(\vec{\pi})^2\right ] \nonumber\\ & & - \left(\lambda/4\right)\left [ ( \sigma)^2+(\vec{\pi})^2\right ] ^2 \nonumber\end{aligned}\ ] ] there is a negative background phase which hides the @xmath36 meson . 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 @xmath36 and @xmath40 mesons . we considered the simplest dyson equation for the @xmath41 scattering amplitude with real intermediate @xmath41 states only ( 1994 ) . @xmath45,\nonumber\\[24pt ] & & t^0_0=\frac{t_0^{0(tree)}}{1-\imath\rho_{\pi\pi}t_0^{0(tree)}}= { \frac{e^{2\imath\left(\delta_{bg}+\delta_{res}\right ) } -1}{2\imath\rho_{\pi\pi } } = \frac{1}{\rho_{\pi\pi}}\left [ \left ( \frac{e^{2\imath\delta_{bg}}-1}{2\imath}\right ) + e^{2\imath\delta_{bg}}t_{res}\right ] } , \nonumber\\[24pt ] & & t_{res}=\frac{\sqrt{s}\gamma_{res}(s)}{m^2_{res } - s + \re(\tilde \pi_{res}(m^2_{res}))-\tilde \pi_{res}(s)}=\frac{e^{2\imath\delta_{res}}-1}{2\imath\rho_{\pi\pi}}\,.\nonumber\end{aligned}\ ] ] in theory the principal problem is impossibility to use the linear @xmath36-model in the tree level approximation inserting widths into @xmath36 meson propagators because such an approach breaks the both unitarity and adler self - consistency conditions , as we showed ( 1994 ) . strictly speaking , the comparison with the experiment requires the non - perturbative calculation of the process amplitudes . nevertheless , now there are the possibilities to estimate odds of the @xmath33 linear @xmath36-model to underlie physics of light scalar mesons in phenomenology . really , even now there is a huge body of information about the @xmath46- waves of different two - particle pseudoscalar states and what is more the relevant information go to press almost continuously from bes , bnl , cern , cesr , da@xmath47ne , fnal , kek , slac and others . as for theory , we know quite a lot about the scenario under discussion : the nine scalar mesons , the putative chiral shielding of the @xmath4 and @xmath48 mesons , the unitarity and adler self - consistency conditions . in addition , there is the light scalar meson treatment motivated by field theory . the foundations of this approach were formulated in our papers ( 1979 - 1984 ) . the nontrivial nature of the well established light scalar resonances @xmath2 and @xmath1 is no longer denied practically anybody . in particular , there exist numerous evidences in favour of the @xmath49 structure of these states . as for the nonet as a whole , even a dope s look at pdg review gives an idea of the four - quark structure of the light scalar meson nonet , @xmath4 , @xmath39 , @xmath2 , and @xmath1 , inverted in comparison with the classical @xmath14-wave @xmath15 tensor meson nonet , @xmath17 , @xmath16 , @xmath50 , @xmath51 . really , while the scalar nonet can not be treated as the @xmath14-wave @xmath52 in the naive quark model , it can be easy understood as the @xmath49 nonet , where @xmath4 has no strange quarks , @xmath39 has the @xmath53 quark , @xmath2 and @xmath1 have the @xmath54 pair , r.l . jaffe , j. schechter et al .. the scalar mesons @xmath1 and @xmath2 , discovered more than thirty years ago , became the hard problem for the naive @xmath15 model from the outset . really , on the one hand the almost exact degeneration of the masses of the isovector @xmath1 and isoscalar @xmath2 states revealed seemingly the structure similar to the structure of the vector @xmath55 and @xmath56 mesons , and on the other hand the strong coupling of @xmath2 with the @xmath57 channel as if suggested a considerable part of the strange pair @xmath54 in the wave function of @xmath2 . in 1977 jaffe noted that in the mit bag model , which incorporates confinement phenomenologically , there are light four - quark scalar states . he suggested also that @xmath1 and @xmath2 might be these states with symbolic structures @xmath58 from that time @xmath1 and @xmath2 resonances came into beloved children of the light quark spectroscopy . ten years later we showed that the study of the radiative decays @xmath59 and @xmath60 can shed light on the problem of @xmath1 and @xmath2 mesons . over the next ten years before experiments ( 1998 ) the question was considered from different points of view . now these decays have been studied not only theoretically but also experimentally . the first measurements have been reported by the snd and cmd-2 collaborations which obtain the following branching ratios @xmath61 more recently the kloe collaboration has measured @xmath62 in agreement with the novosibirsk data but with a considerably smaller error . note that @xmath1 is produced in the radiative @xmath0 meson decay as intensively as @xmath63 containing @xmath64 of @xmath54 , responsible for @xmath65 . it is a clear qualitative argument for the presence of the @xmath54 pair in the isovector @xmath1 state , i.e. , for its four - quark nature . when basing the experimental investigations , we suggested one - loop model @xmath66 . this model is used in the data treatment and is ratified by experiment . below we argue on gauge invariance grounds that the present data give the conclusive arguments in favor of the @xmath67-loop transition as the principal mechanism of @xmath1 and @xmath2 meson production in the @xmath0 radiative decays . this enables to conclude that production of the lightest scalar mesons @xmath1 and @xmath2 in these decays is caused by the four - quark transitions , resulting in strong restrictions on the large @xmath68 expansions of the decay amplitudes . the analysis shows that these constraints give new evidences in favor of the four - quark nature of @xmath1 and @xmath2 mesons . to describe the experimental spectra @xmath69 the function @xmath70 should be smooth ( almost constant ) in the range @xmath71 gev . but the problem issues from gauge invariance which requires that @xmath72\nonumber\\[3pt ] & & = g_r(m)\left [ p_\mu e_\nu(\phi ) - p_\nu e_\mu(\phi)\right]\left [ k_\mu e_\nu(\gamma ) - k_\nu e_\mu(\gamma)\right].\nonumber\end{aligned}\ ] ] consequently , the function @xmath73 is proportional to the photon energy @xmath74 ( at least ! ) in the soft photon region . stopping the function @xmath75 at @xmath76 mev with the help of the form - factor @xmath77 $ ] requires @xmath78 gev@xmath79 . it seems to be incredible to explain such a huge radius in hadron physics . based on rather great @xmath80 gev@xmath79 , one can obtain an effective maximum of the mass spectrum only near 900 mev . to exemplify this trouble let us consider the contribution of the isolated @xmath81 resonance : @xmath82 . let also the mass and the width of the @xmath81 resonance equal 980 mev and 60 mev , then @xmath83 . so stopping the @xmath84 function is the crucial point in understanding the mechanism of the production of @xmath1 and @xmath2 resonances in the @xmath0 radiative decays . the @xmath67-loop model @xmath85 solves this problem in the elegant way : fine threshold phenomenon is discovered , see fig . = 8.5 cm to demonstrate the threshold character of this effect we present fig . [ gg ] and fig . [ ggg ] in which the function @xmath86 is shown in the case of @xmath87 meson mass is 25 mev and 50 mev less than in reality . as seen from figs . [ gg ] and [ ggg ] , the function @xmath86 is suppressed by the @xmath75 law in the region 950 - 1020 mev and 900 - 1020 mev respectively . in the mass spectrum this suppression is increased by one more power of @xmath88 , so that we can not see the resonance in the region 980 - 995 mev . the spectrum maximum is effectively shifted to the region 935 - 950 mev and 880 - 900 mev respectively . in truth this means that @xmath1 and @xmath2 resonances are seen in the radiative decays of @xmath0 meson owing to the @xmath67 intermediate state , otherwise the maxima in the spectra would be shifted to 900 mev . so the mechanism of production of @xmath1 and @xmath2 mesons in the @xmath0 radiative decays is established . both real and imaginary parts of the @xmath89 amplitude are caused by the @xmath67 intermediate state . the imaginary part is caused by the real @xmath67 intermediate state while the real part is caused by the virtual compact @xmath67 intermediate state , i.e. , we are dealing here with the four - quark transition . needless to say , radiative four - quark transitions can happen between two @xmath15 states as well as between @xmath15 and @xmath49 states but their intensities depend strongly on a type of the transitions . a radiative four - quark transition between two @xmath52 states requires creation and annihilation of an additional @xmath15 pair , i.e. , such a transition is forbidden according to the okuba - zweig - izuka ( ozi ) rule , while a radiative four - quark transition between @xmath15 and @xmath49 states requires only creation of an additional @xmath15 pair , i.e. , such a transition is allowed according to the ozi rule . let us consider this problem from the large @xmath68 expansion standpoint , using the g.t hooft rules : @xmath90 at @xmath91 and a gluon is equivalent to a quark - antiquark pair ( @xmath92 ) . the point is that the large @xmath68 expansion is the most clear heuristic understanding of the ozi rule because the ozi forbidden branching ratio as a rule is suppressed by the factor @xmath93 , but not @xmath94 , in comparison with the ozi allowed decay . in our case the results of the analysis are even more interesting . the analysis shows that the @xmath95 decay intensity in the two - quark model and the @xmath96 decay intensity in the @xmath97 model are suppressed by the factor @xmath98 in comparison with the ones in the four - quark model , but in the @xmath99 model the @xmath100 decay intensity is suppressed only by the factor @xmath101 in comparison with the one in the four - quark model . in this model , in addition to the serious trouble with the @xmath102 mass degeneration , the @xmath103 order transition without creation of an additional @xmath15 pair @xmath104 ( similar to the principal mechanism of the @xmath105 decay ) is bound to have a small weight in the large @xmath68 expansion of the @xmath106 amplitude , because this term does not contain the @xmath67 intermediate state , which emerges only in the next to leading term of the @xmath107 order , i.e. , in the ozi forbidden transition . so , in this model the @xmath106 amplitude has the @xmath107 order like the @xmath108 one . emphasize that the mechanism without creation and annihilation of the additional @xmath109 pair can not explain the @xmath2 spectrum because it does not contain the @xmath67 intermediate state ! so , the fine threshold phenomenon is discovered , which is to say that the @xmath67 loop mechanism of the @xmath1 and @xmath2 scalar meson production in the @xmath0 radiative decays is established at a physical level of proof . the case is rarest in hadron physics . this production mechanism is the four - quark transition what constrains the large @xmath68 expansion of the @xmath7 and @xmath110 amplitudes and gives the new strong ( if not crucial ) evidences in favor of the four - quark nature of @xmath1 and @xmath2 mesons . this work was supported in part by the rfbr grant no . 02 - 02 - 16061 and the presidential grant no . 2339.2003.2 for support of leading scientific schools . + _ nb _ * references * contain only author s articles on basis of which the present talk has been written . detailed references to other authors are in these articles . 99 n.n . achasov , s.a . devyanin , and g.n . shestakov , phys . lett . * 88b * , 367 ( 1979 ) ; + phys . lett . * 96b * , 168 ( 1980 ) ; + yad . fiz . * 32 * , 1098 ( 1980)[sov . . phys . * 32 * , 566 ( 1980 ) ] ; + z. phys . c * 22 * , 53 ( 1984 ) ; + usp . nauk . * 142 * , 361 ( 1984 ) [ sov . * 27 * , 161 ( 1984 ) ] . achasov , s.a . devyanin , and g.n . shestakov , phys.lett . b * 108 * 134 ( 1982 ) ; + z. phys . c * 16 * 55 ( 1982 ) achasov and g.n . shestakov , z. phys . c * 41 * 309 ( 1988 ) 309 ; + usp . fiz . nauk . * 161 * ( 6 ) 53 ( 1991 ) [ sov . * 34 * ( 6 ) 471 ( 1991 ) ] . n.n . achasov and v.n . ivanchenko , nucl . b * 315 * , 465 ( 1989 ) . achasov , nucl . b ( proc . suppl . ) * 21 * 189 ( 1991 ) . n.n . achasov and g.n . shestakov , phys . d * 49 * , 5779(1994 ) . n.n . achasov and v.v . gubin , phys . d * 56 * , 4084 ( 1997 ) ; + phys . d * 63 * , 094007 ( 2001 ) . achasov , usp . . nauk . * 168 * , 1257 ( 1998 ) [ phys . usp . * 41 * , 1149 ( 1998 ) ] ; + nucl . a * 675 * , 279c ( 2000 ) ; + yad.fiz . * 65 * , 573 ( 2002 ) [ phys . at . nucl . * 65 * , 546 ( 2002 ) ] . n.n . achasov and a.v . kiselev , phys . d * 68 * , 014006 ( 2003 ) . achasov , nucl . a * 728 * , 425 ( 2003 ) . n.n . achasov and a.v . kiselev , hep - ph/0405128 .	the following topics are considered . 1 . confinement , chiral dynamics and light scalar mesons . 2 . @xmath0-meson radiative decays about nature of light scalar resonances . arguments in favor of the four - quark model of the @xmath1 and @xmath2 mesons are given .
the problem of gravitational radiation in particle collisions has a long history and has been studied in a variety of approaches and approximations . the problem of gravitational radiation in _ massless-_particle collisions is worth studying in its own right and has attracted the interest of many authors in the past as well as very recently . apart from its obvious relevance in the context of tev - scale gravity models with large extra dimensions @xcite , it is very important in relation to the structure of string theory and the issue of black - hole formation in ultra - planckian collisions @xcite . nevertheless , to the best of our knowledge , complete understanding of all facets of the problem is still lacking . the emission of radiation in the form of _ soft _ gravitons was computed in@xcite in the context of quantum field theory , but in that computation the contribution of the non - linear graviton self - couplings i.e. the stress part of the energy - momentum tensor , was argued to be negligible . the result of the quantum computation for low - frequency graviton emission was reproduced by a purely classical computation in@xcite . more recently , a new approach was put forward for the computation of the characteristics of the emitted radiation @xcite , based on the fraunhofer approximation of radiation theory . the purpose here is to extend the method used in@xcite to the study of gravitational radiation in collisions of massless particles with center - of - mass energy @xmath3 and impact parameter @xmath4 . the formal limit @xmath5 ( or equivalently @xmath6 for the lorentz factor ) of the massive case leads to nonsensical answers for the radiation efficiency , i.e. the ratio @xmath7 of the radiated to the available energy , the characteristic radiation frequency @xmath8 , or the characteristic emission angle @xmath9 . we study classically the gravitational radiation in the collision of massless particles using the same perturbative approach as in @xcite . this presentation is mainly based on results obtained in@xcite . the action describing the two massless particles and their gravitational interaction reads @xmath10 where @xmath11 is the einbein of the trajectory @xmath12 in terms of the corresponding affine parameter @xmath13 , @xmath14 and the summation is over the two particles . we will be using unprimed and primed symbols to denote quantities related to the two particles . for identical colliding particles in the center - of - mass frame @xmath15 with @xmath16 the energy of each colliding particle . thus , the particles move on null geodesics , while variation of @xmath17 leads to the particle equation of motion : @xmath18 and similarly for @xmath19 . the particle energy - momentum is defined by @xmath20 . at zeroth order in the gravitational interaction , the space - time is flat and the particles move on straight lines with constant velocities , i.e. @xmath21 given that @xmath22 is traceless , the perturbation @xmath23 satisfies for each particle separately the equation @xmath24 whose solution in coordinate representation is the aichelburg sexl metric @xmath25 where @xmath26 is the @xmath27dimensional fourier transform of @xmath28 : @xmath29 with @xmath30 an arbitrary constant length parameter . write for the metric @xmath31 and substitute in ( [ peom ] ) to obtain for the first correction of the trajectory of the unprimed particle the equation @xmath32 the interaction with the self - field of the particle has been omitted and @xmath33 due to the primed particle is evaluated at the location of the unprimed particle on its unperturbed trajectory . integrating ( [ peom1 ] ) one obtains the trajectory in components : @xmath34 which vanish for all . indeed , the massless particle trajectories should remain undisturbed before the collision . thus , we reproduce the leading order expressions of the two well - known facts @xcite about the geodesics in an aichelburg - sexl metric , namely : ( i ) the _ time delay _ at the moment of shock equal to @xmath35 ( ii ) the _ refraction _ caused by the gravitational interaction by an angle @xmath36 in the direction of the center of gravity . the gravitational wave source has two parts . one is the particle energy - momentum contribution , localized on the accelerated particle trajectories . the direct particle contribution to the source of radiation is called local " , because it is localized on the particle trajectories . the first order term in the expansion of @xmath37 is @xmath38 \delta^4 \!\!{\kern 1pt}\left(x-\ ! \ ! { \ , ^0\!}z(\sigma ) \right)\ , , \end{aligned}\ ] ] where @xmath39 is evaluated at @xmath40 and @xmath41 is evaluated at @xmath42 . in fourier space one obtains effectively @xmath43\,.\end{aligned}\ ] ] the contribution to the source at second - order coming from the expansion of the einstein tensor , reads @xcite @xmath44 upon substitution of @xmath23 and @xmath45 of the previous section in the above expression we obtain for the fourier transform of @xmath46 @xmath47 \nonumber\end{aligned}\ ] ] in terms of the integrals ( @xmath48 ) @xmath49 * total radiation amplitudes . * defining in the center - of - mass frame the radiation wave - vector by @xmath50 and contracting with the two polarizations ( defined as usual ) , we obtain the final ( finite for ) expressions for the source of the gravitational radiation separately for the two polarizations ( here @xmath51 , @xmath52 ) : @xmath53,\\ & \tau_{\times}(k ) = -\frac{16{\kern 1pt } g e^2}{\sqrt{2}}\ , \sin \varphi\ , \!\!\int_0 ^ 1 dx \,{{\rm e}}^{-i ( k b ) x } \biggl [ 2i \frac{\hat{k}_{2 } ( \zeta ) -\hat{k}_{1 } ( \zeta ) } { \omega b \sin \vartheta } + ( 2x-1 ) \cos \varphi \,\hat{k}_{1 } ( \zeta ) \biggr]\ , . { \nonumber}\end{aligned}\ ] ] the emitted radiation frequency spectrum and of the total emitted energy are obtained from ( sum over the two polarizations ) : @xmath54 it will be convenient in the sequel to treat separately the six angular and frequency regimes shown in fig.[table ] . ) .,width=415 ] * zero - frequency limit . * in the low - frequency regime ( @xmath55 ) the amplitude @xmath56 dominates and has the form @xmath57 while @xmath58 is finite ( as @xmath59 ) and gives subleading contribution to . the angular integration leads to @xmath60 , which diverges and implies that our formulae are not valid for @xmath61 close to zero . we can not trust our formulae in regime i and should repair them . a quick way to do it , is to impose a small - angle cut - off @xmath62 on the @xmath63integration , so as to obtain for @xmath64 the value computed quantum mechanically in@xcite : @xmath65 which agrees with our expression for @xmath66 thus , in regime ii we use our formula , which is identical to weinberg s . now for any frequency we have to compare our matter part , which makes the @xmath63integration divergent , with the weinberg s : @xmath67 to leading order in our approximation the scattering process , we are dealing with , is elastic with @xmath68 . write for the incoming particles @xmath69 and for the outgoing ones @xmath70 , substitute into @xmath71 , and expand in powers of @xmath2 to have @xmath72 , to obtain the tensor which is _ identical _ to the local part @xmath37 we use , at _ any _ frequency . therefore , it is natural to impose the angular cutoff @xmath73 for all frequencies of the emitted graviton . * regime vi . * for @xmath74 the two radiation amplitudes read : @xmath75 from which one can obtain an estimate for the frequency distribution of the emitted radiation in regime vi by integrating over @xmath76 and over @xmath77 , namely @xmath78 as well as an estimate for the emitted energy and the corresponding efficiency in regime vi , by integrating also over @xmath79 , @xmath80 * regime iv . * inside the regime iva the amplitude is damped as in regime vi . however , near the left border of regime ivb ( with @xmath81 ) one may expand the amplitudes in powers of @xmath82 and obtain ( [ scalampl2dd ] ) . upon integration over regime ivb , i.e. for @xmath83 , one obtains @xmath84 thus , for @xmath81 one may approximate @xmath85 by @xmath86 it should be pointed out here that the integral of @xmath87 over @xmath88 receives most of its contribution from frequencies in the neighborhood of @xmath89 in both regimes iv and vi . thus , one can say that the characteristic frequency of the emitted radiation is around @xmath90 . * total emitted radiation . * upon integration ( [ scalampl2dd ] ) leads to the angular distribution @xmath91 integration over @xmath92 gives @xmath93 using the same approach as in @xcite , based on standard gr , with the leading non - linear gravity effects taken into account , we studied collisions of massless particles and computed the gravitational energy of arbitrary frequency , which is emitted outside the cone of angle @xmath94 in the forward and backward directions . the value @xmath95 was obtained for the radiation efficiency , with characteristic frequency @xmath96 . in fact , this value represents a lower bound of the efficiency , since it does not include the energy emitted inside that cone . the frequency distribution of radiation in the characteristic angle - frequency regimes is shown in fig.[table ] . unfortunately , we can not yet confirm the presence of any other characteristic frequency@xcite , such as e.g. @xmath97 , or characteristic emission angle smaller than @xmath2 . we hope to return to these issues with a better understanding of regime v in the near future . we would like to acknowledge enlightening discussions with d.galtsov , g.veneziano and d.colferai . this work was supported in part by the eu program thales " ( mis 375734 ) as well as by the rfbr grant 14 - 02 - 01092 . ps acknowledges the non - commercial `` dynasty '' foundation ( russian federation ) for financial support . the work of tnt is implemented under the aristeia ii " action of the operational program education and lifelong learning " and is co - funded by the european social fund ( esf ) and greek national resources and was also partially supported by the european union seventh framework program ( fp7-regpot-2012 - 2013 - 1 ) under grant agreement no .	the angular and frequency characteristics of the gravitational radiation emitted in collisions of massless particles is studied perturbatively in the context of classical general relativity for small values of the ratio @xmath0 of the schwarzschild radius over the impact parameter . the particles are described with their trajectories , while the contribution of the leading nonlinear terms of the gravitational action is also taken into account . the old quantum results are reproduced in the zero frequency limit . the radiation efficiency @xmath1 _ outside _ a narrow cone of angle @xmath2 in the forward and backward directions with respect to the initial particle trajectories is given by and is dominated by radiation with characteristic frequency . the comparison with previous works and the known literature is presented .
the distribution of matter in the cosmos depends on the gravitational interaction and the expansion history of the universe . assuming that galaxies trace the mass distribution , a measurement of galaxy clustering can be used to derive fundamental properties of the universe . on large scales the movement of galaxies is dominated by the hubble recession , while on small scales the gravitational field introduces so - called peculiar velocities . individual galaxy redshifts combine both hubble recession and peculiar velocities indistinguishably . however , these effects can be statistically distinguished in a large sample of galaxy redshifts . this is the purpose of this paper . the difference between the redshift - inferred distance and the true distance is known as redshift - space distortion . redshift - space distortions effectively couple the density and velocity fields , complicating the models needed to accurately describe observed galaxy samples . on the other hand they permit measurements of the properties of the galaxy velocity field , which are difficult to access otherwise . in standard gravity we can use the amplitude of peculiar velocities to measure parameters that describe the matter content of the universe such as @xmath18 and @xmath8 . in this paper we report measurements of the parameter combination @xmath19 , where @xmath20 is the growth rate of cosmic structure ( in terms of the linear growth factor @xmath21 and cosmic scale factor @xmath22 ) and @xmath8 is the r.m.s . of the matter fluctuations in spheres of @xmath9mpc . the measurement of @xmath19 can be used to test theories of dark energy and modified gravity , since a stronger gravitational interaction causes a larger growth rate @xmath23 . it is interesting to note in this context the fact that a different form of gravitational interaction on large scales could be responsible for the accelerating expansion of the universe ( e.g. @xcite ) . probes such as type ia supernovae , the cosmic microwave background ( cmb ) or baryon acoustic oscillations , which have proven the existence of the current acceleration of the expansion of the universe ( e.g. @xcite ) , can not distinguish between acceleration due to a dark energy component with negative pressure or due to a modification of general relativity . however , measurements of the growth of structure are able to distinguish between these models . in order to measure @xmath19 we have to model the effect of redshift - space distortions on the correlation function . while on large scales linear theory can be used to model these effects , on smaller scales non - linear contributions complicate the process . several new approaches have been suggested in recent years to extend linear theory . we will discuss some of these models and apply them to our dataset . redshift - space distortions have previously been analysed using both the correlation function and power spectrum using data from the 2df galaxy redshift survey ( 2dfgrs ; @xcite ) and the sloan digital sky survey ( sdss ; @xcite ) . more recently it has become possible to do similar studies at higher redshift using the vvds @xcite , wigglez @xcite and vlt vimos surveys @xcite . redshift - space distortion measurements are also sensitive to the overall amplitude of the clustering pattern of matter , commonly parameterised by @xmath8 @xcite . this parameter is used to normalise the amplitude of clustering statistics such as the correlation function , @xmath24 . from the cmb we have a very accurate measurement of the matter fluctuations in the early universe ( the scalar amplitude @xmath25 ) at the time of decoupling , @xmath26 . in order to derive @xmath27 we have to extrapolate this measurement to redshift zero , involving assumptions about the expansion history of the universe . the cmb constraint on @xmath8 heavily depends on these assumptions . hence there is a clear advantage in obtaining low - redshift measurements of this parameter . the 6df galaxy survey , which we analyse in this study , is one of the largest galaxy redshift surveys available . its very small effective redshift and wide areal coverage ( @xmath28 of the sky ) make it a powerful sample for the study of the local galaxy distribution . galaxy surveys usually have to consider degeneracies between redshift - space distortions and the alcock - paczynski effect , which arises from the need to assume a cosmological model to transform redshifts into distances . at low redshift this effect is very small , meaning that our measurement is fairly independent of the choice of the fiducial cosmological model . while at high redshift ( @xmath29 ) the matter density dominates both the expansion of the universe and the growth of perturbations , at low redshift these two are partially decoupled , with dark energy mostly dominating the background expansion and the matter density dominating the growth of perturbations . as a result in @xmath30cdm and most proposed modified gravity models , low redshift measurements of the growth rate have a better constraining power than high redshift measurements . the measurement of the growth rate in 6dfgs therefore not only provides a new independent data point at very low redshift , but also promises to make a valuable contribution to tests of general relativity on cosmic scales . the outline of this paper is as follows : in section [ sec : survey ] we introduce the 6df galaxy survey . in section [ sec : data ] we describe the details of the correlation function estimate and introduce the 2d correlation function of 6dfgs . in section [ sec : error ] we derive the covariance matrix for the 2d correlation function . in section [ sec : theory ] we summarise the theory of redshift - space distortions , including extensions to the standard linear approach . we also discuss wide - angle effects and other systematics , such as the alcock - paczynski effect . in section [ sec : fit ] we fit the 2d correlation function to derive @xmath31 and @xmath8 . cosmological implications are investigated in section [ sec : impl ] . in section [ sec : future ] we make fisher matrix predictions for two future low redshift galaxy surveys , wallaby and the proposed taipan survey . we conclude in section [ sec : conc ] . throughout the paper we use @xmath32 to denote real space separations and @xmath33 to denote separations in redshift - space . our fiducial model assumes a flat universe with @xmath34 , @xmath35 and @xmath36 . the hubble constant is set to @xmath37 km s@xmath38mpc@xmath38 . the 6df galaxy survey ( 6dfgs ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) is a near - infrared selected ( @xmath39 ) redshift survey of @xmath40 galaxies across four - fifths of the southern sky , with secondary samples selected in @xmath41 and @xmath42 . the @xmath43 region around the galactic plane is avoided by the @xmath39 surveys to minimise galactic extinction and foreground source confusion in the plane ( as is @xmath44 for @xmath41 and @xmath42 ) . the near - infrared photometric selection was based on total magnitudes from the two - micron all - sky survey extended source catalog ( 2mass xsc ; * ? ? ? the spectroscopic redshifts of 6dfgs were obtained with the six - degree field ( 6df ) multi - object spectrograph of the uk schmidt telescope ( ukst ) between 2001 and 2006 . the effective volume of 6dfgs is about the same as the 2df galaxy redshift survey ( 2dfgrs ; * ? ? ? * ) and is a little under a third that of the sloan digital sky survey main spectroscopic sample at its seventh data release ( sdss dr7 ; * ? ? ? a subset of early - type 6dfgs galaxies ( approximately @xmath45 ) have measured line - widths that will be used to derive fundamental plane distances and peculiar motions . the 6dfgs @xmath0-selected sample used in this paper contains 81971 galaxies selected to a faint limit of @xmath46 . the 2mass magnitudes are on the vega system . the mean completeness of the 6dfgs is 92 percent and median redshift is @xmath47 . completeness corrections are derived by normalising completeness - apparent magnitude functions so that , when integrated over all magnitudes , they equal the measured total completeness on a particular patch of sky . this procedure is outlined in the luminosity function evaluation of @xcite and also in jones et al . , ( in prep ) . the original survey papers @xcite describe in full detail the implementation of the survey and its associated online database . the clustering in a galaxy survey is estimated relative to a random ( unclustered ) distribution which follows the same angular and redshift selection function as the galaxy sample itself . we base our random mock catalogue generation on the 6dfgs luminosity function , where we use random numbers to pick volume - weighted redshifts and luminosity function - weighted absolute magnitudes . we then test whether the redshift - magnitude combination falls within the 6dfgs @xmath0-band faint and bright apparent magnitude limits ( @xmath48 ) . figure [ fig : wred ] shows the redshift distribution of the 6dfgs @xmath0-selected sample ( black solid line ) compared to a mock catalogue with the same number of galaxies ( black dashed line ) . we calculate the co - moving distances for each galaxy using the measured redshift @xmath49 with @xmath50^{1/2},\ ] ] where we assume a flat universe with @xmath36 and @xmath51 and describe dark energy as a cosmological constant ( @xmath35 ) . given the low redshift of our dataset , these assumptions have a very small impact on our final results ( see section [ sec : ap ] ) . we define the positions of two galaxies as @xmath52 and @xmath53 . the redshift - space separation is then given by @xmath54 , while @xmath55 is the mean distance to the pair of galaxies . now we can calculate the separation along the line - of - sight @xmath56 and the separation across the line - of - sight @xmath57 @xmath58 the absolute separation is then given by @xmath59 . we measure the separation between all galaxy pairs in our survey and count the number of such pairs in each separation bin . we do this for the 6dfgs data catalogue , a random catalogue with the same selection function , and a combination of data - random pairs . we call the pair - separation distributions obtained from this analysis step @xmath60 , @xmath61 and @xmath62 , respectively . in the analysis we used @xmath63 random catalogues with the same size as the real data catalogue and average @xmath62 and @xmath61 . the redshift - space correlation function itself is then given by @xcite : @xmath64 where the ratio @xmath65 is given by @xmath66 and the sums go over all random ( @xmath67 ) and data ( @xmath68 ) galaxies . the galaxies are weighted by the inverse completeness @xmath69 of their area of the sky @xmath70 we will discuss further weighting techniques in the next section . there is a possible bias in the estimation of the correlation function due to the fact that we estimate both the mean density and the pair counts from the same survey . this leads to a non - zero difference between the true correlation function estimate of an ensemble of surveys and the ensemble average of @xmath71 from each survey . this is commonly known as the integral constraint ( e.g. @xcite ) , which can be calculated as ( see e.g. @xcite ) @xmath72 and enters our correlation function estimate as @xmath73 where @xmath74 is the redshift - space correlation function from eq . [ eq : ls2 ] and @xmath75 is the model for the correlation function . in 6dfgs @xmath76 is typically around @xmath77 and so has no significant impact on the final result . in figure [ fig : kaiser ] we show the 2d correlation function calculated from the 6dfgs dataset . in this figure we use bins of @xmath78mpc , while for the analysis later on we use larger bins of @xmath79mpc ( see figure [ fig : kaiser_models ] ) . the figure shows clearly the two effects of redshift - space distortions which we will discuss later in section [ sec : theory ] , the `` finger - of - god '' effect at small @xmath57 , and the linear infall effect at larger @xmath57 which gives the correlation function a non - circular shape . in fourier space the error in measuring the amplitude of a mode of the linear power spectrum is given by @xmath80 where @xmath10 is the linear bias , @xmath23 is the growth rate , @xmath81 is the cosine of the angle to the line of sight and @xmath82 is the matter power spectrum . the first term on the right hand side of this equation represents the sample - variance error , while the second term ( @xmath83 ) represents the poisson error . if the sample - variance error is dominant we can reduce the power spectrum error by employing a weighting scheme which depends upon the galaxy density @xmath84 , such as the one suggested by @xcite @xmath85 where @xmath86 describes the amplitude of the weighting . a stronger weighting ( larger value of @xmath86 ) yields a smaller sample - variance error since it increases the survey volume by up - weighting sparsely sampled regions . however , such a weighting scheme also increases the poisson error because it shifts the effective redshift to larger values with a smaller galaxy number density . this is illustrated in figure [ fig : kaiser_error ] and [ fig : kaiser_error2 ] . such a weighting scheme is standard for large scale structure analyses . in a magnitude - limited sample such as 6dfgs , up - weighting higher redshift galaxies also has the effect of shifting the galaxy bias to larger values . the sample - variance error is proportional to the clustering amplitude , and so a larger bias results in a larger error . however , the weighting will still ensure that the relative error of the power spectrum , @xmath87 , is minimised . the redshift - space distortion signal is inversely proportional to the galaxy bias , @xmath88 . if weighting increases the bias @xmath10 , it also reduces the signal we are trying to measure . we therefore must investigate whether the advantage of the weighting ( the reduced relative error ) outweighs the disadvantage ( increasing galaxy bias ) . the situation is very different for measuring a signal that is proportional to the clustering amplitude , such as the baryon acoustic peak . in this case the error and the signal are proportional to the bias , and so weighting will always be beneficial . we stress that an increasing bias with redshift is expected in almost all galaxy redshift surveys . therefore redshift - space distortion studies should first test whether galaxy weighting improves the measurement . the 6df galaxy survey is quite sensitive to the weighting scheme employed because it has a high galaxy density , making the sample - variance error by far the dominant source of error . finally , we have to consider the correlation between the bins in the measured power spectrum or correlation function . if the error is sample - variance dominated , the bins will show large correlation ( especially in the correlation function ) , while in the case of poisson - noise dominated errors , the correlation is much smaller . weighting will always increase the poisson noise and hence reduce the correlation between bins . all of the density weighting effects discussed above need to be considered , before deciding which weighting is best suited to the specific analysis . we summarise as follows : * if a galaxy sample is sample - variance limited , density weighting will reduce the ( relative ) error of the clustering measurement . * in most galaxy redshift surveys , the density weighting increases the galaxy bias , which reduces the redshift - space distortion signal , by flattening the clustering anisotropy . * galaxy weighting also reduces the correlation between bins in the power spectrum and correlation function . whether the signal - to - noise is actually improved by density weighting depends on the specific sample . for the 6dfgs redshift - space distortion analysis we found @xmath89mpc@xmath90 leads to the most accurate constraint on the growth rate . this discussion also indicates that low - biased galaxy samples have an advantage over a highly biased galaxy sample in measuring redshift - space distortions . we will discuss this point further in section [ sec : future ] . in this section we will derive a covariance matrix for the 2d correlation function using jack - knife re - sampling . we also use log - normal realisations to test the jack - knife covariance matrix . [ fig : kaiser_errors ] we divide the dataset into @xmath91 subsets , selected in r.a . and each re - sampling step excludes one subset before calculating the correlation function . the @xmath92 remaining subsets have a volume which is @xmath93 times the volume of the original data . the covariance matrix is then given by @xmath94\left[\xi^k(s_j ) - \overline{\xi}(s_j)\right],\ ] ] where @xmath95 is the correlation function estimate at separation @xmath96 with the exclusion of subset @xmath97 . the mean value is defined as @xmath98 the case @xmath99 gives the error ignoring correlations between bins @xmath100 . we can create a log - normal realisation @xcite of a galaxy survey by deriving a density field from a model power spectrum , @xmath82 , assuming gaussian fluctuations . this density field is then poisson sampled , taking into account the window function and the total number of galaxies . the assumption that the input power spectrum has gaussian fluctuations can only be used if the fluctuations are much smaller than the mean density , otherwise the gaussian model assigns a non - zero probability to regions of negative density . a log - normal random field , @xmath101 , can avoid this unphysical behaviour . it is obtained from a gaussian field @xmath102 by @xmath103 - 1,\ ] ] which is positive - definite but approaches @xmath102 whenever the perturbations are small . as an input power spectrum for the log - normal field we use the linear model @xcite , given by @xmath104 with @xmath105 and @xmath106 , which is consistent with the parameters we measure for 6dfgs ( see section [ sec : fit ] ) . @xmath107 is a linear density power spectrum in real space obtained from camb @xcite and @xmath108 is the galaxy power spectrum in redshift - space . our method is explained in more detail in @xcite , appendix a. we produce @xmath109 such realisations and calculate the 2d correlation function for each of them , deriving a covariance matrix @xmath110\cr & \times\left[\xi^j_n(r_p,\pi ) - \overline{\xi^j}(r_p,\pi)\right ] , \end{split}\ ] ] where the mean value @xmath111 is defined as @xmath112 and @xmath113 is the 2d correlation function estimate of realisation @xmath114 at a specific separation @xmath115 . figure [ fig : cov2d ] shows the correlation matrix @xmath116 derived from the two different covariance matrices , @xmath117 . we plot the jack - knife result in the upper - left corner and the log - normal result in the lower - right corner . both the correlation and covariance matrix are symmetric . while the log - normal correlation matrix indicates somewhat more correlation between bins , overall the correlation matrices are in rough agreement . the diagonal errors for both the log - normal and jack - knife covariance matrices are plotted in figures [ fig : kaiser_error ] and [ fig : kaiser_error2 ] and show very good agreement at large scales . every row / column in figure [ fig : cov2d ] shows the correlation of one bin with all other bins in @xmath118 . figure [ fig : cov2d_2 ] shows an example of such a row / column obtained from jack - knife re - sampling as a @xmath119 matrix , in this case for bin @xmath120 ( @xmath121mpc , @xmath122mpc ) . log - normal realisations do not account for non - linear mode coupling and are very model - dependent in the quasi - linear and non - linear regime . since our analysis relies on fits to fairly small scales , we decided to use the jack - knife covariance matrix in our analysis . however , we find that none of the results reported in this paper depend significantly on which of the two covariance matrices is used . in this section we discuss the theory of redshift - space distortions , starting with the standard linear perturbation theory . we then describe different approaches for extending the linear model to include non - linear structure formation . we also discuss deviations from the plane - parallel approximation . the position of a galaxy in real space @xmath123 is mapped to the position in redshift - space , @xmath124 , via @xmath125 where the unit vector @xmath126 indicates the line - of - sight direction , and the quantity @xmath127 is the line - of - sight component of the velocity field , namely @xmath128 . the scale factor @xmath22 is defined as @xmath129 and @xmath130 is the hubble constant at redshift @xmath131 . the second term in the equation above represents peculiar velocities caused by gravitational interaction . on small scales this elongates structures along the line - of - sight and leads to the so - called `` finger - of - god '' effect @xcite . on large scales matter falls in towards over - dense regions and systematically influences our distance measurement , making the over - densities appear more over - dense . the latter effect can be described by linear theory , while the `` finger - of - god '' effect is a non - linear phenomenon . the model of linear redshift - space distortions has been developed by @xcite and @xcite assuming a plane - parallel approximation . in fourier space the linear model can be written as @xmath132 where @xmath133 is the matter density power spectrum and @xmath108 is the galaxy density power spectrum . in this model , linear redshift - space distortions are quantified by the parameter @xmath134 , which is defined as @xmath135 where @xmath10 is the linear galaxy bias factor , @xmath136 is the growth factor and @xmath14 is the gravitational growth index , which takes the value @xmath137 in @xmath30cdm @xcite . @xmath138 is the matter density at redshift @xmath131 , and is defined as @xmath139 with @xmath140^{-1},\ ] ] where we again follow our fiducial model of @xmath141 and @xmath35 . modifications of the gravitational force mainly affect @xmath14 , while changes in the expansion history of the universe affect @xmath138 . from now on we will formulate our equations in terms of @xmath142 and @xmath143 . we choose the parameter set [ @xmath144 , @xmath145 instead of [ @xmath134 , @xmath8 , @xmath10 ] , because @xmath8 and the linear bias , @xmath10 , are degenerate and difficult to disentangle @xcite . within this parameterisation the power spectrum is expressed as @xmath147 where @xmath148 is the unnormalised matter density power spectrum ( see eq . [ eq : d3 ] ) . in terms of our new parameters , we can write @xmath149 for the remainder of this section we will discuss possible extensions of the linear model to include non - linear structure formation and wide - angle effects . the simplest model of non - linearities is the so - called streaming model @xcite . here , the redshift - space correlation function is just a convolution of the linear correlation function in redshift - space with a pairwise velocity probability density function , @xmath150 , @xmath151 where @xmath152 is the scale factor and @xmath153 is the hubble constant at the effective redshift . we chose @xmath150 to be an exponential distribution @xcite @xmath154 , \label{eq : fv}\ ] ] which has been shown to successfully describe observations @xcite . although the pairwise velocity dispersion within a halo is expected to follow a gaussian distribution instead of an exponential , galaxies populate halos of a wide range of masses and velocity dispersions , which combine to approximately form an exponential function @xcite . the parameter @xmath155 depends on galaxy type ( e.g. @xcite ) and hence its use for cosmological constraints is limited . in recent years many improvements to the model discussed above have been suggested . we will initially discuss these models in fourier space because the theoretical motivation is clearer and many expressions needed for these models simplify considerably . however , since our data is in real space we will also give real - space expressions later on . we start with the streaming model of eq . [ eq : stream ] , which in fourier space is given by @xmath156 hence in fourier space , the convolution with an exponential function becomes a multiplication by a lorentzian distribution . the model above assumes that there is a perfect correlation between the velocity field and the density field , which is given by @xmath157 where @xmath133 is the matter density power spectrum as before . @xmath158 is the velocity divergence power spectrum where @xmath159 is the velocity divergence , and @xmath160 is the cross power spectrum . non - linear effects will violate these assumptions , since the density power spectrum is expected to increase in amplitude at small scales because of non - linear effects , while the velocity field becomes randomised at small scales ( e.g. within virialized galaxy clusters ) and hence @xmath161 will decrease in amplitude ( e.g. @xcite ) . @xcite suggested expressing the 2d power spectrum without the assumption of linear relations between the density field and velocity field , by @xmath162\cr & = f_q(k,\mu,\sigma_v)\cr & \times\left[g_b^2q_{\delta\delta}(k ) + 2\mu^2g_bg_{\theta}q_{\delta\theta}(k ) + \mu^4g_{\theta}^2q_{\theta\theta}(k)\right ] , \end{split } \label{eq : nl4}\ ] ] where the different @xmath163 are defined as @xmath164 and the damping function @xmath165 is usually chosen to be a gaussian of the form @xmath166 the parameter @xmath167 quantifies the non - linear dispersion in the bulk motion of halos . it is different to the @xmath155 parameter we introduced earlier that describes small - scale randomised motion ( e.g. of galaxies within a halo ) . we can derive @xmath167 from the velocity power spectrum as @xmath168 @xcite provide fitting formulae for @xmath160 and @xmath161 derived from n - body simulations . they find the following relation between the different power spectra @xmath169 where @xmath133 can be obtained from camb by including halofit @xcite . for the cross power spectrum @xmath170 we use the ( updated ) parameters ( @xmath171 , @xmath172 , @xmath173 , @xmath174 ) = ( @xmath175 , @xmath176 , @xmath177 , @xmath178 ) and for @xmath179 we use ( @xmath180 , @xmath181 , @xmath182 , @xmath183 ) . the fitting formula reproduces @xmath184 to better than @xmath185 for @xmath186mpc@xmath38 , to @xmath187 for @xmath188mpc@xmath38 , and to @xmath189 for @xmath190mpc@xmath38 . it also reproduces @xmath191 to less than @xmath192 over the whole range @xmath193mpc@xmath38 ( jennings , private communication ) . we cut off the integral when the jennings formula predicts negative values , although because of the high precision of the fitting formula up to large @xmath97 , such a cut - off will not affect our measurement . we can express eq . [ eq : nl4 ] in real - space as @xmath194\mathcal{p}_0(\mu)\cr & + \left[\frac{4}{3}g_bg_{\theta}\xi_{2,\delta\theta}(r ) + \frac{4}{7}g^2_{\theta}\xi_{2,\theta\theta}(r)\right]\mathcal{p}_2(\mu)\cr & + \frac{8}{35}g^2_{\theta}\xi_{4,\theta\theta}(r)\mathcal{p}_4(\mu ) , \end{split } \label{eq : song}\ ] ] where @xmath195 are the legendre polynomials and the spherical harmonic moments @xmath196 are given by @xmath197 appendix [ sec : ana ] shows a partial analytic solution for the double integral above . we therefore have two different models which we will apply to our data : the simple streaming model @xmath198 and the scoccimarro models @xmath199 . note that @xmath199 does not include the parameter @xmath155 and hence has one less free parameter than the streaming model . all equations above are based on the plane - parallel approximation . in the case of 6dfgs we also need to account for wide - angle effects . this means we will replace the equations above with more general descriptions , which we discuss in the next section . so far we have assumed that the separation between galaxy pairs is much smaller than the distance of the galaxies from the observer . the 6df galaxy survey has a maximum opening angle of @xmath200 and the ( effective ) redshift is fairly low at @xmath4 ( see section [ sec : data ] ) . we therefore include wide - angle correction terms . the wide - angle description of redshift - space distortions has been laid out in several papers @xcite . here we will expand on this work by formulating the equations in terms of @xmath144 and @xmath201 and by distinguishing between the density - density , velocity - velocity and density - velocity contributions ( @xmath202 and @xmath203 ) . the model will then correspond to the scoccimarro model ( eq . [ eq : song ] ) in the last section . the general redshift - space correlation function ( ignoring the plane - parallel approximation ) depends on @xmath204 , @xmath205 and @xmath33 . here , @xmath33 is the separation between the galaxy pair , @xmath205 is the half opening angle , and @xmath204 is the angle of @xmath33 to the line - of - sight ( see figure @xmath206 in @xcite ) . the angles @xmath204 and @xmath205 are not independent , but the relation between them is usually expressed through the two angles @xmath207 and @xmath208 given by @xmath209 and @xmath210 . the total correlation function model , including @xmath211 correction terms , is then given by @xcite @xmath212\theta^2 + o(\theta^4 ) . \end{split } \label{eq : wide1}\ ] ] this equation reduces to the plane - parallel approximation if @xmath213 . the factors @xmath214 and @xmath215 in this equation are given by @xmath216\frac{1}{r } - \frac{g_{\theta}^2}{5r}\xi^1_{3,\theta\theta}(r)\\ a_{11 } & = \frac{4g_{\theta}^2}{3r^2}\left[\xi^0_{0,\theta\theta}(r ) - 2\xi^0_{2,\theta\theta}(r)\right]\\ a_{12 } & = \frac{g_{\theta}^2}{5r}\left[2\xi^1_{1,\theta\theta}(r ) -3\xi^1_{3,\theta\theta}(r)\right]\\ b_{11 } & = \frac{4g_{\theta}^2}{3r^2}\left[\xi^0_{0,\theta\theta}(r ) + \xi^0_{2,\theta\theta}(r)\right]\\ b_{12 } & = \frac{2g_{\theta}^2}{5r}\left[\xi^1_{1,\theta\theta}(r ) + \xi^1_{3,\theta\theta}(r)\right ] , \end{split } \label{eq : co}\ ] ] where the spherical harmonic moments , @xmath217 , are @xmath218 with @xmath148 , @xmath219 and @xmath220 as defined in equation [ eq : d3 ] ( see appendix [ sec : ana ] for an analytic solution ) . in order to obtain a model for the 2d correlation function ( including wide - angle effects ) , we can simply integrate eq . [ eq : wide1 ] over @xmath205 . in case of the plane - parallel approximation this would correspond to eq . [ eq : song ] . to reduce the equation to the simple streaming model we have to set @xmath221 , @xmath222 and convolve with @xmath150 . @xcite studied wide angle effects in the sdss - lrg sample and found that such effects are very small . the lower redshift and larger sky coverage of the 6df galaxy survey mean that wide - angle effects are certainly larger in our data set than sdss - lrg . however , we found no significant impact of such effects on our results . this is mainly because our analysis includes only rather small scales ( @xmath223mpc , @xmath224mpc , see section [ sec : fit ] ) . this agrees with the findings of @xcite . we stress that the correction terms discussed above only capture first - order effects and so we additionally restrict our analysis to @xmath225 . many authors have found , using n - body simulations , that both the simple streaming and scoccimarro models do not capture all non - linear and quasi - linear effects present in redshift surveys ( e.g. @xcite and references therein ) . for example the linear bias model seems to be too simplistic at small scales . in this study we restrict ourselves to the models discussed above but would like to point out that many other models have been suggested @xcite which could be compared to our data in future work . a dataset such as 6dfgs , containing galaxies with a high linear bias factor , may be prone to scale - dependent galaxy bias on small scales . we used the gigglez simulations ( poole et al . , in preparation ) to derive the form of the scale - dependent bias . we find a @xmath226 correction to a linear bias at scales of @xmath227mpc , and note that this correction has a small but non - negligible impact on our results , depending on the smallest scales included in the fit . a further systematic uncertainty comes from the so - called alcock - paczynski ( ap ) effect @xcite . by assuming a cosmological model for the conversion of redshifts and angles to distances ( see section [ sec : data ] ) we could introduce an additional anisotropic signal in the correlation function which depends on the angle to the line - of - sight , if the true cosmology differs from our fiducial cosmological model . these distortions are degenerate with the linear redshift - space distortion signal and hence both effects need to be modelled to avoid systematic bias @xcite . using the two scaling factors @xmath228 we can account for the ap effect by rescaling the corresponding axis @xmath229 while this is a matter of concern for high - redshift surveys , the 6df galaxy survey is very nearly independent of the alcock - paczynski distortions because the distances @xmath56 and @xmath57 are almost independent of the fiducial cosmological model when expressed in units of @xmath230mpc . for example the value of @xmath231 at @xmath232 changes by only ( @xmath233 , @xmath234 ) for @xmath187 changes in ( @xmath18 , @xmath235 ) . the corresponding values for @xmath236 are ( @xmath237 , @xmath238 ) . the relative tangential/ radial distortion depends on the value of @xmath239 relative to the fiducial model : @xmath240 this parameter combination changes by only @xmath241 ( @xmath242)% for @xmath187 changes in @xmath18 ( @xmath235 ) . finally , it has been shown that @xmath14 has a degeneracy with the equation of state parameter for dark energy , @xmath235 @xcite . high redshift measurements usually assume @xmath243 , which could introduce a bias in the measured value of @xmath14 if dark energy is not exactly a cosmological constant . again 6dfgs is robust against such effects . lll + @xmath244 & @xmath245 & + @xmath246 & @xmath247 & + @xmath248 & @xmath249 & + @xmath8 & @xmath250 & @xmath251 $ ] + @xmath18 & @xmath252 & @xmath251 $ ] + b & @xmath253 & @xmath251 $ ] + @xmath254 & @xmath255 & @xmath256 $ ] + @xmath14 & @xmath257 & @xmath258 $ ] + @xmath18 & @xmath259 & @xmath258 $ ] + [ tab : para ] we now fit the two models for the 2d correlation function we developed earlier ( @xmath198 and @xmath199 ) to our data . for the final constraints on @xmath16 we use the @xmath199 model since it gives similar results to the streaming model with one less free parameter . other studies ( e.g. @xcite ) prefer to analyse the correlation function moments @xmath260 ( and if possible @xmath261 ) , which carry the same information as the 2d correlation function . the correlation function moments have the advantage that the number of bins grows linearly with the highest scales analysed , while for @xmath5 the number of bins grows quadratically . this makes it easier to get reliable covariance matrices . @xcite also show that the measurement errors of @xmath262 are more gaussian . however , the correlation function moments are integrals over @xmath81 and hence carry information from all directions , including @xmath263 . finger - of - god distortions can influence the correlation function moments up to large scales ( @xmath264mpc ) , while in the 2d correlation function they can be excluded via a cut in @xmath57 . in 6dfgs we have found that these non - linear effects have a strong impact on the correlation function moments up to @xmath265mpc . we have therefore decided to focus on the 2d correlation function instead of the correlation function moments . in figure [ fig : kaiser_models ] we show the 6dfgs 2d correlation function . for our analysis we bin the data in @xmath267mpc bins from @xmath268 to @xmath265mpc in @xmath57 and @xmath56 . including larger scales does not add further information . at small @xmath57 , the finger - of - god effect becomes dominant and we expect any linear model to fail . since our description of non - linearities , in both of our models , is limited in its capability to capture all non - linear effects , it is necessary to include a cut - off scale @xmath269 marking a lower limit of the fitting range in @xmath57 . figure [ fig : can ] shows the measured value of @xmath144 as a function of the cut - off scale @xmath270 for our two different models . above @xmath271mpc the streaming model , @xmath198 , approaches a constant value of @xmath144 . our second model , @xmath199 , contains a systematic error up to much larger scales , before it comes into agreement with the streaming model at about @xmath272mpc . this is expected since this model does not include a description of effects in the non - linear regime . for the final constraints we choose @xmath273mpc for the streaming model and @xmath272mpc for the scoccimarro model . we also note that since the scoccimarro model is based on only two free parameters ( @xmath144 and @xmath201 ) , the error is generally smaller compared to the streaming model , which has three free parameters ( @xmath144 , @xmath274 and @xmath155 ) . other studies fit for the parameter @xmath167 ( e.g. @xcite ) in the scoccimarro model , but we derive it using eq . [ eq : sigv ] . for @xmath199 we use the fitting range @xmath275mpc and @xmath276mpc , which results in a total of @xmath277 bins . the best - fitting results are @xmath278 and @xmath279 , where the errors for each parameter are derived by marginalising over all other parameters . the @xmath280 of this fit is @xmath281 with @xmath282 degrees of freedom ( d.o.f . ) , indicating a good fit to the data . for @xmath198 , we have the fitting range @xmath275mpc and @xmath283mpc , which results in a total of @xmath284 bins . the best fitting parameters are @xmath285 , @xmath286 and @xmath287km / s . the reduced @xmath280 of this fit is given by @xmath288 . we compare the constraints on @xmath144 and @xmath274 from both models in figure [ fig : chi2_1 ] . in the scoccimarro model we could use the parameter @xmath289 instead of @xmath144 to test cosmology , as suggested by @xcite . our best fit gave @xmath290mpc . however , this parameter depends on an additional integral over the velocity power spectrum , which adds a theoretical uncertainty . we therefore prefer to use @xmath144 in the following discussions . we can also express our results in terms of @xmath134 which is given by @xmath291 . we summarise all measured and derived parameters in table [ tab : para ] . in this section we use redshift - space distortions to directly measure @xmath8 . the angular dependence of the redshift - space distortion signal in the 2d correlation function allows us to measure @xmath134 , which quantifies the amplitude of redshift - space distortions . together with @xmath138 and @xmath292 , this constrains the linear bias @xmath10 through the equation @xmath293 knowing @xmath10 we can use the absolute amplitude of the correlation function , @xmath294 ^ 2 $ ] , to constrain @xmath295 \times \sigma_8(z_{\rm eff})$ ] . for computational reasons we use our first model , @xmath198 , in this sub - section and fit the five parameters @xmath8 , @xmath18 , @xmath10 , @xmath296 and @xmath155 using an mcmc approach . since the shape of the correlation function is only sensitive to @xmath297 , we can not constrain @xmath18 and @xmath296 at the same time . for the final results we include a prior on the hubble constant ( @xmath298kms@xmath38mpc@xmath38 , @xcite , from now on referred to as hst prior ) and marginalise over it . we use the same binning and fitting ranges as in the previous section . the best - fitting model results in @xmath299 . we find @xmath300 , @xmath301 , @xmath302 and @xmath303km / s . the remaining degeneracy between the bias @xmath10 and @xmath8 is illustrated in figure [ fig : chi2_sigma8_b ] . we include all these results in table [ tab : para ] . figure [ fig : chi2_om_sigma8 ] compares the 6dfgs @xmath304 probability distribution to measurements from several other datasets : the cfht wide synoptic legacy survey ( cfhtls ) @xcite , the sfi++ peculiar velocity survey @xcite , cluster abundance from x - ray surveys @xcite and wmap7 @xcite . many of the experiments shown in this figure have systematic modelling uncertainties when extrapolating to @xmath305 arising from assumptions about the expansion history of the universe . only 6dfgs and sfi++ are at sufficiently low redshift to be independent of such effects . to illustrate the impact of these effects we plot in figure [ fig : wmap_om_sigma8 ] the probability distribution @xmath304 from wmap7 for different cosmological models . the cmb measures the scalar amplitude @xmath25 , which needs to be extrapolated from redshift @xmath306 to redshift zero to obtain @xmath8 . every parameter that influences the expansion history of the universe in this period affects the value of @xmath8 derived from the cmb alone . the relation between @xmath25 and @xmath8 is given by ( e.g. @xcite ) @xmath307 ^ 2 , \label{eq : sig8z}\end{aligned}\ ] ] where @xmath308mpc , @xmath309mpc@xmath38 and @xmath310 @xcite . @xmath311 is the growth factor at redshift @xmath131 and @xmath312 is the transfer function . if higher - order cosmological parameters such as the dark energy equation of state parameter @xmath235 are marginalised over , then the measurements of @xmath18 and @xmath8 weaken considerably ( see figure [ fig : wmap_om_sigma8 ] ) . we now assess the influence of the @xmath296 prior . we replace the result of @xcite with a measurement derived from the 6dfgs dataset using baryon acoustic oscillations @xcite . the prior from this study is lower than the former value and is given by @xmath313kms@xmath38mpc@xmath38 . using the 6dfgs value of @xmath296 results in @xmath314 , @xmath315 , @xmath316 and @xmath317km / s . the quality of the fit is @xmath318 , very similar to the value obtained with the hst prior . comparing the two results shows that a different prior in @xmath296 shifts the constraint in @xmath8 and @xmath10 along the degeneracy shown in figure [ fig : chi2_sigma8_b ] . however , we note that the 6dfgs measurement of @xmath296 is derived from the same dataset as our present study and hence could be correlated with our measured growth rate . we use these results only for comparison , and include the values obtained using the hst prior in table [ tab : para ] as our final results of this section . alternative methods for deriving @xmath8 or @xmath16 at low redshift are provided by peculiar velocity surveys ( e.g. @xcite ) . 6dfgs will soon provide its own peculiar velocity survey of around @xmath45 galaxies . velocity surveys have the advantage of tracing the matter density field directly , without the complication of a galaxy bias . however , they are much harder to obtain and current velocity surveys are @xmath319 orders of magnitude smaller than galaxy redshift surveys . in this section we test general relativity by measuring the growth index @xmath14 . we would like to stress that the @xmath14-parameterisation of modified gravity has its limitations and other more general parameterisations have been proposed ( see e.g. @xcite ) . however , this is going beyond the scope of this paper . we combine our result for @xmath244 with the latest results from wmap7 @xcite , where we use the wmap7.2 dataset provided on the nasa webpage . while @xmath30cdm predicts @xmath320 , alternative theories of gravity deviate from this value . one example of such an alternative model is the dgp braneworld model of @xcite , in which our observable universe is considered to be a brane embedded in a higher dimensional bulk space - time and the leakage of gravity force propagating into the bulk can lead to the current accelerated expansion of the universe . because of the missing dark energy component , this model predicts a larger growth index of @xmath321 @xcite . in figure [ fig : growth ] we compare measurements of the growth of structure @xmath144 from different galaxy redshift surveys . for the wigglez survey we include data points which assume a correct fiducial cosmology ( solid black triangles ) as well as data points which account for the alcock - paczynski effect ( empty black triangles ) . the degeneracy between the alcock - paczynski effect and the linear redshift - space distortion signal increases the error by about a factor of two . as we showed in section [ sec : ap ] , the alcock - paczynski effect is very small in 6dfgs , which therefore yields a direct measurement of the redshift - space distortion signal . all data points seem to be in good agreement with the @xmath30cdm model ( black solid line ) , while the dgp model generally predicts smaller values of @xmath144 . the value of @xmath8 for the two different models has been derived from the cmb scalar amplitude @xmath25 @xcite , where we use the corresponding friedmann equation to calculate @xmath8 . the analysis method we apply in this section is summarised in the following four points : 1 . we produce a monte carlo markov chain ( mcmc ) with cosmomc @xcite for a @xmath30cdm universe by fitting the wmap7 dataset . the cmb depends on dark energy through the distance of last scattering and the late - time integrated sachs - wolfe ( isw ) effect . we avoid the contributions of the isw effect by limiting the wmap7 dataset to multipole moments @xmath322 . 2 . now we importance - sample the cosmomc chain by randomly choosing a value of @xmath14 in the range @xmath323 for each chain element . since the value of @xmath324 depends on @xmath14 we have to recalculate this value for each chain element . first we derive the growth factor @xmath325,\ ] ] where @xmath326 is the scale factor at the effective redshift @xmath327 . in order to derive @xmath328 we have to extrapolate from the matter dominated region to the effective redshift , @xmath329 where we use @xmath330 from eq . [ eq : sig8z ] and @xmath331 , well in the matter - dominated regime . we now calculate the growth rate using @xmath332 and construct @xmath333 . finally we compare the model with @xmath278 from table [ tab : para ] and combine the likelihood from this comparison with the wmap7 likelihood . the result is shown in figure [ fig : gamma_om ] . marginalising over the remaining parameters we get @xmath334 and @xmath335 , which is in agreement with the prediction of a @xmath30cdm universe ( @xmath320 ) . our analysis depends only on the growth rate measured in 6dfgs and wmap7 . this makes our measurement of @xmath14 independent of systematic effects like the alcock - paczynski distortion which is a matter of concern for galaxy redshift surveys at higher redshift . in this section we make predictions for the accuracy of @xmath16 measurements from future low - redshift galaxy surveys using a fisher matrix analysis based on @xcite the wide - field askap l - band legacy all - sky blind survey ( wallaby ) is an hi survey planned for the australian ska pathfinder telescope ( askap ) , currently under construction at the murchison radio - astronomy observatory ( mro ) in western australia . the survey will cover @xmath336 of the sky and a proposal exists to fill up the remaining @xmath337 using the westerbork radio telescope . in this analysis we follow the survey parameters employed by @xcite ( see also duffy et al . , in preparation ) , where for wallaby we assume a @xmath338 survey containing @xmath339 galaxies at a mean redshift of @xmath340 . the linear bias of a typical wallaby galaxy is @xmath341 @xcite and the volume of the survey is @xmath342gpc@xmath343 . the taipan survey proposed for the uk schmidt telescope at siding spring observatory in new south wales will cover a similar sky area as 6dfgs but will extend to a larger redshift such that @xmath344 . for our taipan forecast we assumed the same sky - coverage as 6dfgs ( @xmath345 ) , a bias of @xmath346 , a total of @xmath347 galaxies and a volume of @xmath348gpc@xmath343 . first we test the fisher matrix prediction for @xmath16 in the case of 6dfgs . we assume a survey volume of @xmath349gpc@xmath343 , with @xmath1 galaxies . using @xmath350mpc@xmath38 we forecast a measurement of @xmath16 of @xmath351 , while using @xmath352mpc@xmath38 produces a @xmath353 error . the actual error in @xmath16 we found in this paper is @xmath354 , somewhere between these two values . for wallaby and taipan we will report constraints for both @xmath350mpc@xmath38 and @xmath352mpc@xmath38 . with the specifications given above and using @xmath355mpc@xmath38 , the fisher matrix forecast for wallaby is a measurement of @xmath16 with @xmath356 error . we included the wallaby forecast with a @xmath192 error - bar in figure [ fig : growth ] . the model taipan survey produces forecast errors of @xmath357 , improving the results from 6dfgs by almost a factor of two . although taipan maps a larger volume of the universe compared to wallaby , it does not produce a better measurement of @xmath16 . wallaby has a smaller galaxy bias , which increases the redshift - space distortion signal by a factor of two compared to taipan . this will also be very useful for breaking the degeneracy between bias and @xmath8 using the technique of section [ sec : sig8 ] . the wallaby survey will target galaxies rich in hi gas . such galaxies will mostly populate under - dense regions of the universe , because in groups and clusters galaxies are stripped of their gas by interactions with other galaxies and the intra - group and intra - cluster medium . this is the reason that hi - selected galaxies possess a low bias ( @xmath358 ) . however , this fact also implies that wallaby galaxies sample the density field in a manner that avoids high - density regions . these high - density regions are an important source of non - linear redshift - space distortions ( `` finger - of - god '' effect ) . we can hence suppose that non - linear effects will be smaller in amplitude in an hi survey compared to highly - biased surveys such as 6dfgs ( see e.g. @xcite or figure 4 in @xcite ) . this should allow the inclusion of much smaller scales in the analysis , producing more accurate measurements . a more detailed analysis using wallaby mock catalogues is in preparation . @xcite have suggested that multiple tracers within the same cosmic volume can be used to reduce the sampling variance and improve cosmological parameter constraints ( see also @xcite ) . using the ratio of the perturbation amplitudes of two surveys with different bias factors gives @xmath359 where @xmath360 . the angular dependence of this expression allows one to extract @xmath23 without any dependence on the density field ( see figure 1 in @xcite ) . the density field is the source of the sampling variance error , since it will change , depending on the patch of the sky which is observed . using the ratio of two tracers , the precision with which the growth rate @xmath23 can be determined is ( in principle ) only limited by the shot noise , and not by the sampling variance . the three surveys discussed above , 6dfgs , wallaby and taipan , have a large overlapping volume which allows the potential application of this method . the technique works best for densely - sampled surveys with very different bias factors , @xmath361 and @xmath362 . while taipan and 6dfgs have very similar bias , the bias of wallaby will be much smaller . we assume an overlap volume of @xmath363gpc@xmath364gpc@xmath343 , where we multiply the sky coverage of 6dfgs and taipan with the effective volume of wallaby . for the different surveys we use the parameters as stated above . we forecast fisher matrix constraints on @xmath16 of @xmath365 using @xmath366mpc@xmath38 . because the overlap volume is only @xmath367 of the wallaby volume , this result does not improve the measurement arising from wallaby alone , especially considering that wallaby itself may be able to include modes up to large @xmath368 in the fitting process . our results show that future surveys such as wallaby and taipan will provide an accurate measurement of @xmath16 at low redshift , and will be able to complement future high - redshift surveys such as boss , which will have a similar accuracy for several data points over the higher redshift range @xmath369 @xcite . in this paper we have measured the 2d correlation function of the 6df galaxy survey . we derived a covariance matrix using jack - knife resampling as well as log - normal realisations and showed that both techniques give comparable results . we have modelled the 2d correlation function with a simple streaming model and a more advanced approach suggested by @xcite combined with the n - body calibrated results from @xcite . we formulated these models in real - space including wide - angle corrections . for the final results on @xmath16 we chose the model by @xcite , although we found that both models gave consistent results at sufficiently large scales . we analysed the measurement in two different ways . first we fitted for the two parameters @xmath370 and @xmath371 , where these constraints depend only on the 6dfgs data . our second analysis method assumes a growth index from standard gravity ( @xmath320 ) and fits for @xmath8 , @xmath10 , @xmath18 , @xmath296 and @xmath155 , where we combine the 6dfgs measurement with a prior in the hubble constant . all parameter measurements are summarised in table [ tab : para ] . we can summarise the results as follows : * our first analysis method found @xmath372 and @xmath373 , at an effective redshift of @xmath4 . the 6dfgs measurement of @xmath144 , unlike high - redshift measurements , does not depend on assumptions about the expansion history of the universe and the alcock - paczynski distortion . * in our second analysis method we used the angle dependence of redshift - space distortions in the 2d correlation function to break the degeneracy between the galaxy bias @xmath10 and the normalisation of the matter clustering statistic @xmath8 , assuming standard gravity . we found @xmath300 , @xmath301 , @xmath302 and @xmath374km / s . this result uses a prior on @xmath296 from @xcite . * combining our measurement of @xmath244 with wmap7 @xcite allows us to measure the growth index @xmath14 , directly testing general relativity . we found @xmath334 and @xmath335 , in agreement with the predictions of general relativity ( @xmath320 ) . the 6dfgs measurement of this parameter is independent of possible degeneracies of @xmath14 with other parameters which affect the correlation function at high redshift , such as the dark energy equation of state parameter @xmath235 . * we used a fisher matrix analysis to forecast the constraints on @xmath16 that would be obtained from two future low - redshift galaxy surveys , wallaby and taipan . we found that wallaby will be able to measure @xmath16 to a forecast accuracy of @xmath375 for @xmath350mpc@xmath38 and @xmath376 for @xmath377mpc@xmath38 . a combination of 6dfgs , taipan and wallaby , using the multiple - tracer method proposed by @xcite , will be able to constrain @xmath16 to @xmath378 . these measurements would complement future large - volume surveys such as boss , which will measure the growth rate at much higher redshift ( @xmath379 ) , and contribute to future precision tests of general relativity on cosmic scales . the authors thank alex merson for providing the random mock generator . we thank alexandra abate , liping fu , patrick henry , adi nusser , adam mantz and alexey vikihlinin for providing their results for comparison . we are also grateful to david parkinson for help with cosmomc and yong - seon song , takahiko matsubara , martin meyer , morag scrimgeour and david rapetti for fruitful discussions . is supported by the australian government through the international postgraduate research scholarship ( iprs ) and by scholarships from icrar and the aao . part of this work used the ivec@xmath380uwa supercomputer facility . gbp acknowledges support from two australian research council ( arc ) discovery projects ( dp0772084 and dp1093738 ) . the 6df galaxy survey was funded in part by an australian research council discovery projects grant ( dp-0208876 ) , administered by the australian national university . 99 abate a. and erdogdu p. , mon . not . * 400 * ( 2009 ) 1541 [ arxiv:0905.2967 [ astro-ph.co ] ] . abazajian k. n. _ et al . _ [ sdss collaboration ] , astrophys . j. suppl . * 182 * ( 2009 ) 543 [ arxiv:0812.0649 [ astro - ph ] ] . alcock c. and paczynski b. , nature * 281 * ( 1979 ) 358 . baker t. , ferreira p. g. , skordis c. and zuntz j. , phys . d * 84 * ( 2011 ) 124018 [ arxiv:1107.0491 [ astro-ph.co ] ] . ballinger w. e. , peacock j. a. and heavens a. f. , mon . not . * 282 * ( 1996 ) 877 [ arxiv : astro - ph/9605017 ] . basilakos s. , plionis m. , kovac k. and voglis n. , mnras * 378 * , 301 ( 2007 ) [ arxiv : astro - ph/0703713 ] . bean r. and tangmatitham m. , phys . d * 81 * ( 2010 ) 083534 [ arxiv:1002.4197 [ astro-ph.co ] ] . bernstein g. m. , cai y. -c . , [ arxiv:1104.3862 [ astro-ph.co ] ] . beutler f. , blake c. , colless m. , jones d. h. , staveley - smith l. , campbell l. , parker q. and saunders w. _ et al . _ , mon . not . * 416 * ( 2011 ) 3017 [ arxiv:1106.3366 [ astro-ph.co ] ] . bielby r. _ et al . _ , arxiv:1005.3028 [ astro-ph.co ] . blake et al . 2011a , mon . not . * 415 * ( 2011 ) 2876 [ arxiv:1104.2948 [ astro-ph.co ] ] . blake et al . 2011b , arxiv:1108.2637 [ astro-ph.co ] . blake et al . 2011c , mon . not . roy . astron . * 418 * ( 2011 ) 1707 [ arxiv:1108.2635 [ astro-ph.co ] ] . cabre a. and gaztanaga e. , mon . not . * 393 * ( 2009 ) 1183 [ arxiv:0807.2460 [ astro - ph ] ] . carlson j. , white m. and padmanabhan n. , phys . d * 80 * ( 2009 ) 043531 [ arxiv:0905.0479 [ astro-ph.co ] ] . clifton t. , ferreira p. g. , padilla a. and skordis c. , arxiv:1106.2476 [ astro-ph.co ] . cole s. _ et al . _ [ the 2dfgrs collaboration ] , mnras * 362 * ( 2005 ) 505 [ arxiv : astro - ph/0501174 ] . coles p. and jones b. , mnras * 248 * ( 1991 ) 1 . colless m. _ et al . _ [ the 2dfgrs collaboration ] , mnras * 328 * ( 2001 ) 1039 [ arxiv : astro - ph/0106498 ] . daniel s. f. and linder e. v. , phys . d * 82 * ( 2010 ) 103523 [ arxiv:1008.0397 [ astro-ph.co ] ] . davis m. and peebles p. j. e. , astrophys . j. * 267 * ( 1982 ) 465 . davis m. , nusser a. , masters k. , springob c. , huchra j. p. and lemson g. , arxiv:1011.3114 [ astro-ph.co ] . diaferio a. and geller m. j. , astrophys . j. * 467 * ( 1996 ) 19 arxiv : astro - ph/9602086 . dvali g. r. , gabadadze g. and porrati m. , phys . b * 485 * ( 2000 ) 208 [ arxiv : hep - th/0005016 ] . feldman h. a. , kaiser n. and peacock j. a. , astrophys . j. * 426 * ( 1994 ) 23 [ arxiv : astro - ph/9304022 ] . fisher k. b. , davis m. , strauss m. a. , yahil a. and huchra j. p. , mon . not . * 267 * ( 1994 ) 927 [ arxiv : astro - ph/9308013 ] . fu l. _ et al . _ , astron . astrophys . * 479 * ( 2008 ) 9 [ arxiv:0712.0884 [ astro - ph ] ] . gordon c. , land k. and slosar a. , phys . * 99 * ( 2007 ) 081301 [ arxiv:0705.1718 [ astro - ph ] ] . guzzo l. _ et al . _ , nature * 451 * ( 2008 ) 541 [ arxiv:0802.1944 [ astro - ph ] ] . hamilton a. j. s. , apj , * 385 * , l5 hatton s. j. and cole s. , mon . not . roy . astron . * 296 * ( 1998 ) 10 [ arxiv : astro - ph/9707186 ] . hawkins e. _ et al . _ , mnras * 346 * , 78 ( 2003 ) [ arxiv : astro - ph/0212375 ] . hojjati a. , pogosian l. and zhao g. -b . , jcap * 1108 * ( 2011 ) 005 [ arxiv:1106.4543 [ astro-ph.co ] ] . hudson m. j. and turnbull s. j. , arxiv:1203.4814 [ astro-ph.co ] . jackson j. c. , mnras * 156 * ( 1972 ) 1 [ arxiv:0810.3908 [ astro - ph ] ] . jarrett t. h. , chester t. , cutri r. , schneider s. , skrutskie m. and huchra j. p. , astron . j. * 119 * ( 2000 ) 2498 [ arxiv : astro - ph/0004318 ] . jennings e. , baugh c. m. and pascoli s. , ( 2011a ) mon . not . * 410 * ( 2011 ) 2081 [ arxiv:1003.4282 [ astro-ph.co ] ] . jennings e. , baugh c. m. and pascoli s. , ( 2011b ) astrophys . j. * 727 * ( 2011 ) l9 [ arxiv:1011.2842 [ astro-ph.co ] ] . jones d. h. _ et al . _ , mnras * 355 * ( 2004 ) 747 [ arxiv : astro - ph/0403501 ] . jones d. h. , peterson b. a. , colless m. and saunders w. , mnras * 369 * ( 2006 ) 25 [ erratum - ibid . * 370 * ( 2006 ) 1583 ] [ arxiv : astro - ph/0603609 ] . jones d. h. _ et al . _ , arxiv:0903.5451 [ astro-ph.co ] . kaiser n. , mnras * 227 * , 1 ( 1987 ) . komatsu e. _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 180 * ( 2009 ) 330 [ arxiv:0803.0547 [ astro - ph ] ] . komatsu e. _ et al . _ [ wmap collaboration ] , astrophys . j. suppl . * 192 * ( 2011 ) 18 [ arxiv:1001.4538 [ astro-ph.co ] ] . kwan j. , lewis g. f. and linder e. v. , arxiv:1105.1194 [ astro-ph.co ] . lahav o. , bridle s. l. , percival w. j. , peacock j. a. , efstathiou g. , baugh c. m. , bland - hawthorn j. and bridges t. _ et al . _ , mon . not . * 333 * ( 2002 ) 961 [ astro - ph/0112162 ] . landy s. d. and szalay a. s. , astrophys . j. * 412 * ( 1993 ) 64 . landy s. d. , astrophys . j. * 567 * , l1 ( 2002 ) [ arxiv : astro - ph/0202130 ] . lewis a. , challinor a. and lasenby a. , astrophys . j. * 538 * , 473 ( 2000 ) [ arxiv : astro - ph/9911177 ] . lewis a. and bridle s. , phys . d * 66 * ( 2002 ) 103511 [ arxiv : astro - ph/0205436 ] . linder e. v. , phys . d * 72 * ( 2005 ) 043529 [ arxiv : astro - ph/0507263 ] . madgwick d. s. _ et al . _ , mon . not . * 344 * ( 2003 ) 847 [ arxiv : astro - ph/0303668 ] . mantz a. , allen s. w. , rapetti d. and ebeling h. , mon . not . soc . * 406 * , 1759 ( 2010 ) [ arxiv:0909.3098 [ astro-ph.co ] ] . marzke r. o. , geller m. j. , da costa l. n. and huchra j. p. , astron . j. * 110 * ( 1995 ) 477 [ arxiv : astro - ph/9504070 ] . matsubara t. , astrophys . j. * 615 * ( 2004 ) 573 [ arxiv : astro - ph/0408349 ] . matsubara 2008a , phys . d * 77 * ( 2008 ) 063530 [ arxiv:0711.2521 [ astro - ph ] ] . matsubara 2008b , phys . d * 78 * ( 2008 ) 083519 [ erratum - ibid . d * 78 * ( 2008 ) 109901 ] [ arxiv:0807.1733 [ astro - ph ] ] . mcdonald p. , seljak u. , jcap * 0910 * , 007 ( 2009 ) . [ arxiv:0810.0323 [ astro - ph ] ] . nusser a. and davis m. , astrophys . j. * 736 * ( 2011 ) 93 [ arxiv:1101.1650 [ astro-ph.co ] ] . papai p. and szapudi i. , arxiv:0802.2940 [ astro - ph ] . peacock j. a. and dodds s. j. , mnras * 280 * ( 1996 ) l19 [ arxiv : astro - ph/9603031 ] . peacock j. a. , cole s. , norberg p. , baugh c. m. , bland - hawthorn j. , bridges t. , cannon r. d. and colless m. _ et al . _ , nature * 410 * ( 2001 ) 169 [ astro - ph/0103143 ] . peebles p. j. e. `` the large - scale structure of the universe '' princeton , n.j . , princeton university press , 1980 . 435 p. percival w. j. _ et al . _ , astrophys . j. * 657 * ( 2007 ) 645 [ arxiv : astro - ph/0608636 ] . raccanelli a. , samushia l. and percival w. j. , arxiv:1006.1652 [ astro-ph.co ] . reid b. a. and white m. , arxiv:1105.4165 [ astro-ph.co ] . roche n. d. , almaini o. , dunlop j. , ivison r. j. and willott c. j. , mon . not . * 337 * ( 2002 ) 1282 [ arxiv : astro - ph/0205259 ] . riess a. g. _ et al . _ , astrophys . j. * 730 * ( 2011 ) 119 [ arxiv:1103.2976 [ astro-ph.co ] ] . samushia l. , percival w. j. and raccanelli a. , arxiv:1102.1014 [ astro-ph.co ] . scoccimarro r. , phys . d * 70 * ( 2004 ) 083007 [ arxiv : astro - ph/0407214 ] . seljak u. , phys . * 102 * ( 2009 ) 021302 . [ arxiv:0807.1770 [ astro - ph ] ] . seljak u. and mcdonald p. , jcap * 1111 * ( 2011 ) 039 [ arxiv:1109.1888 [ astro-ph.co ] ] . seto n. and yokoyama j. , astrophys . j. * 492 * ( 1998 ) 421 [ arxiv : astro - ph/9707312 ] . sheth r. k. , mon . not . * 279 * ( 1996 ) 1310 [ arxiv : astro - ph/9511068 ] . silvestri a. and trodden m. , rept . * 72 * ( 2009 ) 096901 [ arxiv:0904.0024 [ astro-ph.co ] ] . simpson f. and peacock j. a. , phys . d * 81 * ( 2010 ) 043512 [ arxiv:0910.3834 [ astro-ph.co ] ] . simpson f. , james j. b. , heavens a. f. , heymans c. , [ arxiv:1107.5169 [ astro-ph.co ] ] . slosar a. , jcap , 3 , 4 ( 2009 ) smith r. e. _ et al . _ [ the virgo consortium collaboration ] , mnras * 341 * ( 2003 ) 1311 [ arxiv : astro - ph/0207664 ] . song y. s. and percival w. j. , jcap * 0910 * ( 2009 ) 004 [ arxiv:0807.0810 [ astro - ph ] ] . song y. s. , sabiu c. g. , kayo i. and nichol r. c. , arxiv:1006.4630 [ astro-ph.co ] . szalay a. s. , matsubara t. and landy s. d. , arxiv : astro - ph/9712007 . szapudi i. , astrophys . j. * 614 * ( 2004 ) 51 [ arxiv : astro - ph/0404477 ] . takada m. , komatsu e. and futamase t. , phys . d * 73 * ( 2006 ) 083520 [ arxiv : astro - ph/0512374 ] . taruya a. , nishimichi t. and saito s. , phys . d * 82 * ( 2010 ) 063522 [ arxiv:1006.0699 [ astro-ph.co ] ] . tegmark m. _ et al . _ [ sdss collaboration ] , phys . rev . d * 69 * ( 2004 ) 103501 [ arxiv : astro - ph/0310723 ] . tegmark m. _ et al . _ [ sdss collaboration ] , phys . d * 74 * ( 2006 ) 123507 [ arxiv : astro - ph/0608632 ] . de la torre s. , guzzo l. , [ arxiv:1202.5559v1 ] turnbull s. j. , hudson m. j. , feldman h. a. , hicken m. , kirshner r. p. and watkins r. , arxiv:1111.0631 [ astro-ph.co ] . wang y. , jcap * 0805 * ( 2008 ) 021 [ arxiv:0710.3885 [ astro - ph ] ] . white m. , song y. s. and percival w. j. , mon . not . * 397 * ( 2008 ) 1348 [ arxiv:0810.1518 [ astro - ph ] ] . zehavi i. _ et al . _ [ sdss collaboration ] , astrophys . j. * 621 * ( 2005 ) 22 [ arxiv : astro - ph/0411557 ] . the double integrals in eqs . [ eq : xilm1 ] and [ eq : xilm2 ] are difficult to solve numerically . here we show an analytical solution for the integral over @xmath81 which allows for a faster numerical solution of the full integral . first we re - write eq . [ eq : xilm2 ] as @xmath381 where @xmath382 is the complex number . more generally the integral over @xmath81 can be written as @xmath383 the integral on the right can be solved analytically . if we set @xmath384 and @xmath385 we obtain @xmath386 . \end{split}\ ] ] taking the n - th derivatives of the real part of the term above gives @xmath387 . the dawson integral @xmath388 can be calculated using the imaginary error function @xmath389 : @xmath390 we can than construct eq . [ eq : xilm2 ] for the different correlation function moments : @xmath391\cr \xi^m_{\ell=3,xy}(r ) & = \int^{\infty}_{0}\frac{k^mdk}{(2\pi)^2}q_{xy}(k ) \frac{1}{2}\left[5f_3(\mu ) - 3f_1(\mu)\right]\cr \xi^m_{\ell=4,xy}(r ) & = \int^{\infty}_{0}\frac{k^mdk}{(2\pi)^2}q_{xy}(k ) \frac{1}{8}\left[35f_4(\mu ) - 30f_2(\mu ) + 3f_0(\mu)\right]\cr \dots & \end{split}\ ] ] and so on .	we present a detailed analysis of redshift - space distortions in the two - point correlation function of the 6df galaxy survey ( 6dfgs ) . the @xmath0-band selected sub - sample which we employ in this study contains @xmath1 galaxies distributed over @xmath2deg@xmath3 with an effective redshift @xmath4 . by modelling the 2d galaxy correlation function , @xmath5 , we measure the parameter combination @xmath6 , where @xmath7 is the growth rate of cosmic structure and @xmath8 is the r.m.s . of matter fluctuations in @xmath9mpc spheres . + alternatively , by assuming standard gravity we can break the degeneracy between @xmath8 and the galaxy bias parameter , @xmath10 . combining our data with the hubble constant prior from @xcite , we measure @xmath11 and @xmath12 , consistent with constraints from other galaxy surveys and the cosmic microwave background data from wmap7 . + combining our measurement of @xmath13 with wmap7 allows us to test the cosmic growth history and the relationship between matter and gravity on cosmic scales by constraining the growth index of density fluctuations , @xmath14 . using only 6dfgs and wmap7 data we find @xmath15 , consistent with the prediction of general relativity . we note that because of the low effective redshift of 6dfgs our measurement of the growth rate is independent of the fiducial cosmological model ( alcock - paczynski effect ) . we also show that our conclusions are not sensitive to the model adopted for non - linear redshift - space distortions . + using a fisher matrix analysis we report predictions for constraints on @xmath16 for the wallaby survey and the proposed taipan survey . the wallaby survey will be able to measure @xmath16 with a precision of @xmath17 , depending on the modelling of non - linear structure formation . this is comparable to the predicted precision for the best redshift bins of the baryon oscillation spectroscopic survey ( boss ) , demonstrating that low - redshift surveys have a significant role to play in future tests of dark energy and modified gravity . [ firstpage ] cosmology : observations , cosmological parameters , large - scale structure of universe , surveys , galaxies : statistics
let @xmath1 be an @xmath2-valued continuous path over @xmath0 $ ] with bounded variation ( @xmath3 ) . according to @xcite , @xcite , for @xmath4 we can define the sequence of iterated integrals @xmath5 where @xmath6 @xmath7 is regarded as an element in the tensor space @xmath8 and @xmath9 is hence an element in the tensor algebra @xmath10 @xmath9 is multiplicative in the sense that it satisfies the following chen s identity : @xmath11 @xmath9 is uniquely determined by the original path @xmath12 or intuitively speaking , the original path @xmath1 contains all information about its differential @xmath13 . a remarkable consequence is that a theory of integration along @xmath1 can be established in the sense of riemannstieltjes , which leads to a theory of differential equations driven by @xmath14 such a theory for paths with bounded variation is classical and well - studied . if the path @xmath1 is less regular , for example , @xmath1 has finite @xmath15-variation for some @xmath16 it may not be possible to establish an integration theory along @xmath1 by using the information of the original path only . the fundamental reason is that the path @xmath1 itself does not reveal enough information on its differential @xmath13 , which is essential to be fully understood if we want to develop an integration theory along @xmath14 as pointed out by t. lyons in @xcite , for this purpose , together with the path itself , a finite sequence of iterated integrals up to level @xmath17 $ ] satisfying chen s identity should be specified in advance . such a finite sequence of iterated integrals @xmath18})\ ] ] is regarded as a multiplicative functional @xmath19 from the simplex @xmath20 to the truncated tensor algebra @xmath21)}(\mathbb{r}^{d})=\oplus_{n=0}^{[p]}\mathbb{r}^{nd}.\ ] ] @xmath19 is called a rough path with roughness @xmath22 according to @xcite , @xmath19 extends uniquely to a multiplicative functional from @xmath23 to @xmath24 in the founding work of t. lyons in @xcite , a general theory of integration and differential equations for rough paths was established . for a rough path @xmath19 with roughness @xmath25 the signature of @xmath19 is defined as the formal sequence @xmath26},\cdots),\ ] ] where for @xmath27,$ ] @xmath7 is the unique extension of @xmath19 as mentioned before . the signature @xmath28 , proposed by k.t . chen in @xcite and t. lyons in @xcite , can be regarded as the collection of overall information of any arbitrary level @xmath29 about the rough path @xmath30 it is of central interest and conjectured in the theory of rough paths that the signature @xmath28 contains sufficient information to recover the path @xmath19 completely . in the groundbreaking paper @xcite by b. hambly and t. lyons , they proved that for a path @xmath1 with bounded variation , the signature of @xmath1 uniquely determines the path up to a tree - like equivalence . however , for paths with unbounded variation , very few results are available and it remains a lot of work to do . in the work @xcite by y. le jan and z. qian , they considered the case of multidimensional brownian motion and proved that for almost surely , the brownian paths can be recovered by using the so - called stratonovich s signature , which is defined via iterated stratonovich s integrals of arbitrary orders . since the brownian paths are of unbounded variation and can be regarded as rough paths with roughness @xmath31 , we may need to specify the second level in order to make sense in terms of rough paths . however , according to @xcite , @xcite , there is a canonical lifting of the brownian paths to the second level by using dyadic approximations , which is called the l@xmath32vy s stochastic area process and it coincides exactly with the iterated stratonovich s integral defined in the same way as ( [ iterated integrals ] ) . such lifting is determined by the brownian paths itself , and in @xcite when regarding the brownian motion as rough paths such lifting was used by the authors . therefore , the recovery of brownian motion as rough paths is essentially the recovery of the brownian paths in terms of stratonovich s signature . in the present paper , we are going to generalize the result of y. le jan and z. qian in @xcite to the case of multidimensional diffusion processes ( possibly degenerate ) . the main idea of the proof is similar to the case of brownian motion , in which the authors used a specially designed approximation scheme and chose special differential 1-forms to define the so - called extended stratonovich s signatures to recover the brownian paths . however , there are several difficulties in the case of diffusion processes . firstly , we need quantitative estimates for rare events of diffusion processes to prove a convergence result similar to the case of brownian motion . in @xcite , the authors used the symmetry and explicit distribution of brownian motion , which are not available in the case of diffusion processes and hence we need to proceed in a different way . secondly , to construct special differential @xmath33-forms , a quite special case of h@xmath34rmander s theorem was used to ensure the existence of density , in which the so - called h@xmath34rmander s condition was easily verified . in the case of diffusion processes , the construction of differential @xmath33-forms is more complicated to ensure similar kind of hypoellipticity . lastly , in the brownian motion case , the laplace operator is well - posed so that pde methods could be applied to obtain a crucial estimate which enables us to relate the extended stratonovich s signatures to the brownian paths . however , for a general diffusion process , the generator @xmath35 may not be well - posed any more ( we do nt impose uniform ellipticity assumption on @xmath35 ) and pde methods may no longer apply ( in fact , to ensure the application of pde methods , rather technical assumptions should be imposed on the differential operator @xmath35 and the domain if without uniform ellipticity ) . therefore , we need a different approach to recover the diffusion paths by using extended stratonovich s signatures . in this section , we are going to state our main result and illustrate the idea of the proof . let @xmath36 be a complete probability space and let @xmath37 be a @xmath38-dimensional brownian motion on @xmath39 consider an @xmath40-dimensional @xmath41 diffusion process @xmath1 defined by the following sde ( possibly degenerate ) : @xmath42 with @xmath43 we are going to make the following three assumptions on the generating vector fields @xmath44 . \(a ) @xmath45 \(b ) for any @xmath46 h@xmath34rmander s condition ( see @xcite ) holds at @xmath47 in the sense that @xmath48,0\leqslant\alpha,\beta\leqslant d,\ [ v_{\alpha},[v_{\beta},v_{\gamma}]],0\leqslant\alpha,\beta,\gamma\leqslant d,\ \cdots\ ] ] generate the tangent space @xmath49 , where @xmath50 $ ] denotes the lie bracket . \(c ) there exists a positive orthonormal basis @xmath51 of @xmath52 such that for any @xmath53 and @xmath54 @xmath55 is not perpendicular to @xmath56 for some @xmath57 assumptions ( a ) and ( b ) are made to ensure the hypoellipticity of the generator @xmath58 of the diffusion process ( [ sde ] ) . assumption ( c ) is made to ensure the escape condition and the non - tangential condition proposed in @xcite hold on some domain of interest which is relatively small . under these assumptions , we are able to apply results in @xcite to obtain the existence of a continuous density function of the poisson kernel for some domain of interest and a quantitative estimate on the density function , which are both crucial in the proof of our main result . it should be pointed out that if the diffusion process ( [ sde ] ) is nondegenerate , that is , if @xmath59 generate the tangent space @xmath49 at each point @xmath46 then assumptions ( a ) , ( b ) , ( c ) are all verified . for @xmath60 @xmath61 define the iterated stratonovich s integral of order @xmath29 : @xmath62_{s , t}=\int_{s < t_{1}<\cdots < t_{n}<t}\circ dx_{t_{1}}^{j_{1}}\circ dx_{t_{2}}^{j_{2}}\circ\cdots\circ dx_{t_{n}}^{j_{n}},\ \ \ 0\leqslant s < t\leqslant1.\ ] ] alternatively , @xmath63_{s , t}$ ] can be defined inductively by the following relation : @xmath62_{s , t}=\int_{s < u < t}[j_{1},\cdots , j_{n-1}]_{s , u}\circ dx_{u}^{j_{n}},\ \ \ 0\leqslant s < t\leqslant1,\ ] ] where @xmath64_{s , t}$ ] is defined to be @xmath65_{s , t}=\int_{s < u < t}\circ dx_{u}^{j_{1}}=x_{t}^{j_{1}}-x_{s}^{j_{1}},\ \ \ 0\leqslant s < t\leqslant1.\ ] ] for convenience , if @xmath66 we denote @xmath63_{s , t}=1.$ ] the family @xmath67_{0,1}:\ j_{1},\cdots , j_{n}\in\{1,2,\cdots , n\},\ n\geqslant0\}\ ] ] of iterated stratonovich s integrals is called the _ stratonovich signature _ of @xmath1 over @xmath0.$ ] let @xmath68 be the completion of the @xmath69-algebra generated by the diffusion process @xmath1 over @xmath0,$ ] and let @xmath70 be the completion of the @xmath69-algebra generated by the stratonovich s signature of @xmath1 over @xmath0.$ ] more precisely , @xmath71_{0,1}:\ j_{1},\cdots , j_{n}\in\{1,2,\cdots , n\},n\geqslant0\})}.\end{aligned}\ ] ] for the case of brownian motion , it was proved by y. le jan and z. qian in @xcite that @xmath72 such result for diffusion processes in our setting can be proved in the present paper . however , we are going to formulate the problem in a more illustrative way , which to some extend reveals how we can reconstruct the diffusion paths from the stratonovich s signature over @xmath0 $ ] in a conceivable way . first we need the following definition . a _ piecewise linear trajectory _ ( p.l.t . ) @xmath73 in @xmath74 is a finite sequence of points in @xmath74 ( not necessarily all distinct ) . here we always assume that the number of points in @xmath73 is greater than one ( if @xmath73 consists of only one point @xmath75 we will regard @xmath73as the finite sequence @xmath76 ) . for a p.l.t . @xmath73 in @xmath74 , the number of points in @xmath73will be denoted by @xmath77 if the points of @xmath73 belongs to a subset @xmath78 we say that @xmath73 is a p.l.t . in @xmath79 the reason why we use the notion piecewise linear trajectory is that when given @xmath80 we actually think of @xmath73 as a piecewise linear graph by connecting the points in @xmath73 by line segments in order . here we should point out that the order of points in @xmath73 is rather important , and no parametrizations are involved . for @xmath81 a _ parametrization _ @xmath69 of order @xmath29 is a partition of the time interval @xmath0 $ ] into @xmath82 nontrivial subintervals : @xmath83 the space of all parametrizations of order @xmath29 will be denoted by @xmath84 let @xmath73 be a p.l.t . in @xmath74 and let @xmath69 be a parametrization of order @xmath77 the piecewise linear path over @xmath0 $ ] defined by applying linear interpolation of @xmath73 along the parametrization @xmath69 is denoted by @xmath85 our formulation of the problem is related to a kind of convergence which is parametrization free . therefore , we need the following definition of convergence in trajectory . let @xmath86 be a continuous path in @xmath87 a sequence @xmath88 of p.l.t.s is said to be _ converging in trajectory _ to @xmath86 if @xmath89 such kind of convergence modulo parametrization is similar to the notion of fr@xmath32chet distance , which was originally introduced by m. fr@xmath32chet in the study of shapes of geometric spaces . now we are in a position to state our main result . [ thm 2.1]let @xmath90 be the space of p.l.t.s in @xmath91 equipped with the discrete @xmath69-algebra . then there exists a sequence @xmath88 of @xmath90-valued @xmath70-measurable random variables ( random p.l.t.s ) , such that with probability one , @xmath92 converges in trajectory to the diffusion paths @xmath93 it seems that the statement of theorem [ thm 2.1 ] does not contain much information about the approximating sequence @xmath94 however , when from the proof in the next section , we will see that @xmath95 is constructed in a quite explicit way . it should be pointed out that the result of theorem [ thm 2.1 ] was already implicitly proved in @xcite for the case of brownian motion . a direct consequence of theorem [ thm 2.1 ] is the result based on y. le jan and z. qian s formulation . @xmath96 before proving our main result theorem [ thm 2.1 ] in the next section , we first illustrate the idea and main steps of the proof . we adopt the scheme and the key observation that the diffusion paths can be recovered by reading out the maximal sequence of well - chosen compactly supported differential @xmath33-forms such that the iterated stratonovich s integral of those @xmath33-forms ( extended stratonovich s signature ) along the diffusion paths over the duration of visiting their supports is nonzero , which were proposed in @xcite . the idea of the proof of theorem [ thm 2.1 ] is the following . firstly , decompose the euclidean space @xmath74 into disjoint small boxes and narrow tunnels . by recording the successive visit times of those small boxes , we can construct a piecewise linear approximation of the diffusion paths . a convergence theorem can be proved by developing certain types of estimates of rare events for the diffusion process . by enlarging the size of those small boxes a little bit ( by a higher order infinitesimal relative to the size of boxes ) , we can similarly get another piecewise linear approximation also converging to the diffusion paths as the size of boxes goes to zero . secondly , we construct a family of `` special '' differential 1-forms on @xmath74 ( depending on the size of boxes ) in a way that for any larger box , we construct a @xmath33-form supported in it such that it is highly non - degenerate on the inner smaller box . the crucial observation is that the stratonovich s integral of any of those @xmath33-forms along the diffusion paths over the duration of visit of its support is nonzero . it turns out that for a diffusion path , we can read out an associated unique maximal finite sequence of @xmath33-forms ( a p.l.t . ) recording a sequence of boxes in order such that the iterated stratonovich s integral of this sequence of @xmath33-forms ( extended stratonovich s signature ) along the diffusion path over the duration of visiting their supports is nonzero . it provides us with sufficient information to recover the diffusion path by taking limit in a reasonable way . this is due to the fact that based on our construction , we can prove that such a maximal sequence always `` lies '' between the two piecewise linear approximations constructed before , both of which converge to the diffusion path . here we need to develop a kind of squeeze theorem for the type of convergence ( convergence in trajectory in the setting of p.l.t.s defined as before ) in our situation . to carry out the above idea , we are going to establish the following three steps . \(1 ) step one : proving a convergence result for the piecewise linear approximation based on successive visit times of small boxes . the proof consists of two ingredients . the first one is a probabilistic estimate of the number of boxes visited over the time duration @xmath0,$ ] which can be developed by using a random time change technique . it turns out that we can reduce to the brownian motion case . the importance of such an estimate is that we can get an asymptotic rate of the probability that the number of boxes visited over @xmath0 $ ] is quite large . the second one is the probabilistic estimate of the uniform distance between the piecewise linear approximation path and the original diffusion path , provided the number of boxes visited over @xmath0 $ ] is fixed . this can be done by using the strong markov property and a quantitative result in @xcite by g. ben arous , s. kusuoka and d.w . stroock , which gives us control on the density of the poisson kernel of a given bounded domain in @xmath74 and enables us to estimate the probability that the diffusion process travels through narrow tunnels . combining the two ingredients , it is not hard to prove the convergence result by using the borel - cantelli s lemma via a subsequence . \(2 ) step two : constructing special differential @xmath33-forms and using extended stratonovich s signatures . for any larger box , we are going to construct a suitable differential @xmath33-form supported in it and highly nondegenerate on the inner smaller box . the construction of such a differential @xmath33-form can be reduced to the construction of a differential @xmath33-form such that the generator of some associated sde with dimension @xmath97 is hypoelliptic on the support of the differential @xmath33-form . the family of diffenrential @xmath33-forms constructed in such a way will be used to construct extended stratonovich s signatures , which in turn will be used to recover the diffusion paths as stated in the idea of the proof . \(3 ) step three : proving a squeeze theorem for convergence in trajectory to recover the diffusion paths . from the above two steps we constructed two sequences of piecewise linear approximations of the diffusion paths , and between which a sequence of p.l.t.s in terms of extended stratonovich signatures . we will formulate the term `` lying between '' in a rigorous way in the setting of p.l.t.s and prove a squeeze theorem for convergence in trajectory which fits our situation . here the squeeze theorem we are going to prove is not in the most general case ( we need to make use of special parametrizations ) , so we need to modify the piecewise linear approximation associated to larger boxes to fit our case . an advantage of using such a squeeze theorem is that we can get around the estimates based on potential theory and partial differential equations , which was used in @xcite for the laplace operator . in fact , for a general elliptic operator @xmath98 the associated partial differential equation may not be well - posed and the conditions to ensure a ( regular ) probabilistic representation of a solution is quite restrictive and technical . in this section , we will give the detailed proof of our main result theorem [ thm 2.1 ] . recall that @xmath99 is an @xmath40-dimensional diffusion process defined by the following sde : @xmath100 with @xmath101 in which the generating vector fields satisfy assumptions ( a ) , ( b ) , ( c ) . in the following the coordinates of @xmath53 is taken with respect to the orthonormal basis given in assumption ( c ) . similar to the idea of y. le jan and z. qian , we first construct a suitable approximation scheme for the diffusion paths . for convenience , a constant is called universal if it depends only on the generator @xmath35 and the dimensions @xmath102 moreover , sometimes we may use the same notation to denote universal constants coming out from estimates , although they may be different from line to line . let @xmath103 for @xmath104 let @xmath105 be the @xmath40-cube in @xmath74 defined by @xmath106 where @xmath107 is some universal constant to be chosen later on . for technical reasons we assume that the boundary of @xmath105 is smoothed to the order of @xmath108 . such a smoothing procedure can be done in a simple geometric way , or by using standard mollifiers . in the case of @xmath109 we can simply replace each corner of @xmath105 by a quarter of a circle with radius @xmath108 . the space @xmath74 is then divided into disjoint small boxes and narrow tunnels . now we are going to construct an approximation of diffusion paths @xmath1 over the time duration @xmath0.$ ] let @xmath110 and @xmath111 for @xmath112 define @xmath113 if @xmath114 define @xmath115@xmath116 be the integer point in @xmath91 such that @xmath117 if @xmath118 define @xmath119 intuitively , the sequence of hitting times @xmath120 records the successive visit times of the small boxes and the sequence of integer points @xmath121 records the boxes visited by the diffusion paths in order ( revisit of the same box before visiting other boxes does nt count ) . note that it is possible that @xmath122 since with positive probability the process can always stay in narrow tunnels after leaving some box . let @xmath123 be the number of boxes visited by the diffusion paths over the time duration @xmath0.$ ] formally , @xmath124 it follows from uniform continuity of the diffusion paths over @xmath0 $ ] that @xmath125 for almost surely . by a standard random time change argument , we can prove the following . [ lem 3.1 ] let @xmath126 then for any @xmath127 @xmath128 for @xmath112 it is obvious that @xmath129 where the last inequality comes from the fact that the distance between two different boxes is bounded from below by @xmath130 by the strong markov property , it suffices to estimate @xmath131 where @xmath1 is the diffusion process defined by ( [ sde ] ) starting at @xmath132 by rewriting ( [ sde ] ) in the sense of it@xmath133 , we have @xmath134 where @xmath135 . it follows that for @xmath136 , we have @xmath137 by using a standard random time change technique and the inequality @xmath138 it is then easy to obtain that @xmath139 therefore , we have @xmath140 and the proof is complete . now we define polygonal approximations of the diffusion paths through successive visits of those boxes . more precisely , if @xmath141 define @xmath142 on @xmath0;$ ] otherwise for @xmath143 , define @xmath144,\ ] ] and on @xmath145,$ ] define @xmath146 figure 1 illustrates the construction . . ] now we have the following convergence result . the proof is developed for arbitrary dimension @xmath147 but in the case of @xmath148 the idea is easier to visualize . [ prop 3.1]there exists a subsequence @xmath149 such that with probability one , @xmath150 converges uniformly to the diffusion paths @xmath151 on @xmath0 $ ] as @xmath152 we aim at estimating the following probability @xmath153 where @xmath154 is a large universal constant to be chosen later on . for convenience , we will assume that @xmath155 is a positive integer . for this purpose , let @xmath156 be a large integer to be chosen later on ( may depend on @xmath157 ) . it follows that @xmath158 @xmath159+p(m_{h}^{\varepsilon}>k).\end{aligned}\ ] ] we first estimate @xmath160 the idea is the following : the event @xmath161 implies that after time @xmath162 the process must have travelled through many narrow tunnels and spread far away from @xmath163 by many boxes before visiting another box . define @xmath164 to be the first time after @xmath165 that the process arrives at the entrance of some narrow tunnel which is far away from @xmath163 with distance at least @xmath166 without hitting any other boxes . for @xmath115@xmath167 define @xmath168 to be the first time after @xmath169 that the process travels through a narrow tunnel without hitting any boxes other than @xmath163 ( define @xmath170 if there is no such arrival and @xmath171 if there is no such travel through ) . it is easy to see that @xmath172 thus it suffices to estimate @xmath173 this can be done by using the strong markov property and a quantitative estimate for the poisson kernel of some nice domain in @xcite . in fact , by the strong markov property , @xmath174\\ = & e[p^{x_{\sigma_{\lambda/12 - 1}}(\omega)}(\{\omega':\ \omega'\in t(\omega)\}),\ \sigma_{0}(\omega)<\sigma_{1}(\omega)<\cdots<\sigma_{\lambda/12 - 1}(\omega)<\infty],\end{aligned}\ ] ] where @xmath175 denotes the set of sample paths @xmath176 of the diffusion process starting at @xmath177 such that the first time of traveling through a narrow tunnel without hitting any boxes is finite . by the assumptions on the generating vector fields , the generator @xmath35 and those small boxes @xmath105 verify the conditions of lemma 2.6 in @xcite . it follows from the lemma that the poisson kernel @xmath178 of any small box has a continuous density @xmath179 with respect to the normalized surface measure @xmath180 moreover , there are universal constants ( in particular , not depending on @xmath157 ) @xmath181 such that @xmath182 for any @xmath47 in the box and @xmath183 on the boundary . since traveling through narrow tunnels implies escaping through narrow windows of the boundary of some associated domain , it follows that on @xmath184 @xmath185 for some universal constant @xmath186 now it is clear that if we choose @xmath107 to be universal and far greater than @xmath187 then on @xmath184 we have @xmath188 for some universal constant @xmath186 therefore , @xmath189 by induction , it is immediate that @xmath190 therefore , we arrive at @xmath191 the estimate of @xmath192 is exactly the same as above since on @xmath193 there will be no visit of boxes other than @xmath194 during @xmath195.$ ] now consider @xmath196 by lemma [ lem 3.1 ] , if @xmath136 , @xmath197 where @xmath198 choose a universal constant @xmath199 and let @xmath200 $ ] ( when @xmath157 is small , the condition @xmath136 in lemma [ lem 3.1 ] is satisfied ) . it follows that @xmath201 where @xmath202 is a positive universal constant . combining with the estimates before , we arrive at @xmath203 finally , choose a positive universal integer @xmath154 such that @xmath204 and @xmath155 is a positive integer . by taking @xmath205 we have @xmath206 borel - cantelli s lemma then yields the desired result . now the proof is complete . let @xmath207 be another universal constant . define @xmath208 in the same way as @xmath209 only with @xmath107 replaced by @xmath210 , then proposition [ prop 3.1 ] is also true for @xmath211 ( with @xmath212 as in the proof of proposition [ prop 3.1 ] ) . to complete the proof , it remains to trace the diffusion paths via extended stratonovich s signatures `` between '' smaller boxes @xmath105 and larger boxes @xmath213 and prove a squeeze theorem so that we are able to pass to the same limit @xmath14 to trace the diffusion paths by using extended stratonovich s signatures , we first need to construct suitable compactly supported differential @xmath33-forms , such that the stratonovich s integral of any such @xmath33-form @xmath214 along the diffusion paths over the duration of visit of @xmath215 is nonzero with probability one . to this end , it suffices to construct a suitable differential 1-form @xmath214 on @xmath74 with compact support such that the family of vector fields on @xmath216 : @xmath217 satisfies h@xmath34rmander s condition on @xmath218 so that the generator of the diffusion process on @xmath216 defined by @xmath219 is hypoelliptic on @xmath218 , which ensures the existence of smooth probability densities of certain wiener functionals . here and thereafter we use the geometric notation for convenience ( so @xmath220 is a regarded as a column vector and @xmath214 is regarded as a row vector in @xmath74 ) . in fact , if this is possible , then we can proceed in the same way as lemma 2.1 , lemma 2.2 and lemma 2.3 in @xcite to show that the stratonovich s integral of @xmath214 along the diffusion paths over the duration of visit of @xmath215 is nonzero with probability one , since starting from this point the proof relies only on the strong markov property and again the results in @xcite , which hold true from our assumptions on the generating vector fields @xmath221 now to make it more precise , for @xmath222 and @xmath223 we are interested in constructing a differential @xmath33-form @xmath224 such that @xmath225 and @xmath224 has the property mentioned before . the following result gives the desired construction . [ prop 3.2]assume that the family of vector field @xmath44 satisfies h@xmath34rmander s condition at every point @xmath47 in @xmath74 . let @xmath226 be a bounded domain in @xmath74 and @xmath227 be an open subdomain of @xmath226 such that @xmath228 let @xmath229 be a cut - off function of @xmath230 , that is , @xmath231 @xmath232 on @xmath230 and @xmath233 outside a small neighborhood of @xmath230 . then there exists @xmath234 such that for any @xmath235 with @xmath236 , if we define the differential @xmath33-form @xmath214 on @xmath74 by @xmath237 and define the vector field @xmath238 on @xmath216 ( independent of @xmath239 ) by @xmath240 then the family of vector fields @xmath241 satisfies h@xmath34rmander s condition at every point on @xmath242 . in other words , the differential operator @xmath243 on @xmath216 defined by @xmath244 is hypoelliptic on @xmath245 . for a differential @xmath33-form @xmath214 on @xmath74 defined by ( [ 1-form ] ) , define the vector fields @xmath241 on @xmath216 by ( [ vector fields in r^(n+1 ) ] ) . note that @xmath215 is independent of @xmath246 , which will be denoted by @xmath247 let @xmath248 for @xmath249 denote @xmath250 , and we use the notation @xmath251}$ ] ( @xmath252}$ ] , respectively ) to denote @xmath253]$ ] ( @xmath254]$ ] , respectively ) . we first prove that for any @xmath255 @xmath252}$ ] can be written as @xmath256}=\left(\begin{array}{c } v_{[\theta]}\\ g_{[\theta]}+\phi\cdot v_{[\theta ] } \end{array}\right)\ ] ] for some @xmath257}\in c_{b}^{\infty}(\mathbb{r}^{n+1})$ ] independent of @xmath258 in fact , when @xmath259 it is just the definition of @xmath252}$ ] . assume that it is true for any @xmath260 let @xmath261 then there exists some @xmath262 and @xmath263 , such that @xmath264}=[v_{\alpha},v_{[\theta']}],\ \widetilde{v}_{[\theta]}=[\widetilde{v}_{\alpha},\widetilde{v}_{[\theta']}].\ ] ] by the induction hypothesis , we have @xmath265 } & = & \left[\left(\begin{array}{c } v_{\alpha}\\ \phi\cdot v_{\alpha } \end{array}\right),\left(\begin{array}{c } v_{[\theta']}\\ g_{[\theta']}+\phi\cdot v_{[\theta ' ] } \end{array}\right)\right]\\ & = & \left(\begin{array}{c } [ v_{\alpha},v_{[\theta']}]\\ \nabla^{n}(g_{[\theta']}+\phi\cdot v_{[\theta']})\cdot v_{\alpha}-\nabla^{n}(\phi\cdot v_{\alpha})\cdot v_{[\theta ' ] } \end{array}\right)\\ & = & \left(\begin{array}{c } v_{[\theta]}\\ g_{[\theta]}+\phi\cdot v_{[\theta ] } \end{array}\right),\end{aligned}\ ] ] where @xmath266}=\nabla^{n}g_{[\theta']}\cdot v_{\alpha}+v_{[\theta']}^{t}\cdot\nabla^{n}\phi^{t}\cdot v_{\alpha}-v_{\alpha}^{t}\cdot\nabla^{n}\phi^{t}\cdot v_{[\theta']}\in c_{b}^{\infty}(\mathbb{r}^{n+1}),\ ] ] which is independent of @xmath239 . here @xmath267 denotes the gradient operator with respect to @xmath268 and @xmath269 denotes the transpose operator . now we are going to prove the result by a compactness argument . a key observation is that for any fixed @xmath115@xmath255 let @xmath257}\in c_{b}^{\infty}(\mathbb{r}^{n+1})$ ] be such that @xmath256}=\left(\begin{array}{c } v_{[\theta]}\\ g_{[\theta]}+\phi\cdot v_{[\theta ] } \end{array}\right)\ ] ] as in the previous discussion , then @xmath257}$ ] is of the form @xmath266}(x)=p_{[\theta]}(\xi;x)e^{-\frac{1}{2}\|x-\xi\|^{2}},\ ] ] where @xmath270}(\xi;x)$ ] is a polynomial in @xmath271 with @xmath272 coefficients depending only on @xmath273 here the degree of @xmath270}$ ] is at most @xmath274 in other words , @xmath275}(\xi;x)=\sum_{j=0}^{\|\theta\|-1}\sum_{\|\boldsymbol{\alpha}\|=j}c_{\boldsymbol{\alpha}}(x^{(n)})\xi^{\boldsymbol{\alpha}}.\ ] ] fix @xmath276 where @xmath277 by the hypoellipticity of @xmath35 and continuity , there exists a neighborhood @xmath278 of @xmath279 and @xmath280 such that for any @xmath281 , @xmath282}(x^{(n)}),\cdots , v_{[\theta^{(n)}]}(x^{(n)})\}\ ] ] constitutes a basis of @xmath74 . it follows that for any @xmath283 the family of vectors in @xmath216 @xmath284}(x),\cdots,\widetilde{v}_{[\theta^{(n)}]}(x)\}\ ] ] generate an @xmath40-dimensional subspace of @xmath285 let @xmath286 again by the assumptions on @xmath35 and continuity , it is possible to choose @xmath287 with @xmath288 , such that @xmath289})}}>m\label{degree}\ ] ] in some compact neighborhood @xmath290 of @xmath291 in particular , the choice of @xmath292 and @xmath293 is independent of the @xmath246 since the coefficents of @xmath270}$ ] are functions of @xmath47 only . now we are going to show that there exists @xmath234 such that when @xmath235 with @xmath236 , the vector field @xmath252}$ ] can not be generated by @xmath115@xmath294},\cdots,\widetilde{v}_{[\theta^{(n)}]}\}$ ] in @xmath295 so that @xmath296},\cdots,\widetilde{v}_{[\theta^{(n)}]},\widetilde{v}_{[\theta]}}\}}=n+1,\ ] ] which yields the hypoellipticity of @xmath243 defined by [ tilde l ] in @xmath297 to prove this , first notice that there exists @xmath298 such that @xmath264}(x^{(n)})=\sum_{i=1}^{n}\lambda^{i}(x^{(n)})v_{[\theta^{(i)}]}(x^{(n)}),\ \ \ \mbox{for } x^{(n)}\in u_{0}.\ ] ] moreover , from ( [ degree ] ) it is easy to see that there exists @xmath234 such that @xmath275}(\xi;x)\neq\sum_{i=1}^{n}\lambda^{i}(x^{(n)})p_{[\theta^{(i)}]}(\xi;x)\label{degree not equal}\ ] ] for @xmath235 with @xmath236 and @xmath299 if @xmath256}(x_{1})\in\mbox{span}\{\widetilde{v}_{[\theta^{(1)}]}(x_{1}),\cdots,\widetilde{v}_{[\theta^{(n)}]}(x_{1})\}\ ] ] for some @xmath300 then we must have @xmath256}(x_{1})=\sum_{i=1}^{n}\lambda^{i}(x_{1}^{(n)})\widetilde{v}_{[\theta^{(i)}]}(x_{1}).\ ] ] it follows from simple calculation that @xmath266}(x_{1})=\sum_{i=1}^{n}\lambda^{i}(x_{1}^{(n)})g_{[\theta^{(i)}]}(x_{1}).\label{g=00003dlambda g}\ ] ] this is a contradiction to ( [ degree not equal ] ) when @xmath301 therefore , @xmath252}$ ] can not be generated by @xmath294},\cdots,\widetilde{v}_{[\theta^{(n)}]}\}$ ] in @xmath302 if we choose @xmath246 with @xmath236 in the definition of @xmath303 the case when @xmath304 can be proved in the same way by replacing @xmath293 with @xmath305 finally , combining with the above local results and by the compactness of @xmath306 we are able to choose @xmath307 ( depending on @xmath308 ) , such that for any @xmath235 with @xmath309 the differential operator @xmath243 is hypoelliptic on @xmath310 now the proof is complete . for @xmath222 and @xmath223 by taking @xmath311 and @xmath312 we can construct a differential @xmath33-form @xmath224 supported in @xmath226 according to lemma [ prop 3.2 ] ( just take some fixed admissible @xmath235 as in the lemma ) . by proceeding in the same way as in @xcite , we conclude that the stratonovich s integral of @xmath224 along the diffusion paths over the duration of visit of @xmath313 is nonzero with probability one . now we are going to construct extended stratonovich s signatures to trace the original diffusion paths by using these differential @xmath33-forms @xmath224 . we first define extended stratonovich s signatures formally . for smooth differential forms @xmath314 on @xmath74 , the iterated stratonovich s integral @xmath315_{s , t}$ ] ( @xmath316 ) defined inductively by @xmath317_{s , t}=\int_{s < u < t}[\psi^{1},\cdots,\psi^{k-1}]_{s , u}\psi^{k}(\circ dx_{u}),\ ] ] where @xmath318_{s , t}=\sum_{i=1}^{n}\int_{s < u < t}\psi_{i}^{1}(x_{u})\circ dx_{u}^{i},\ ] ] is called an _ extended stratonovich s signature _ of the diffusion process @xmath14 the following lemma allows us to use extended stratonovich s signatures for our study . the case of brownian motion was proved in @xcite , but we can easily adopt the proof to the our case without changing anything ( in fact , the proof does not rely on probabilistic features , but only on paths ) . recall that @xmath70 is the completion of the @xmath69-algebra generated by the stratonovich s signature of @xmath1 over @xmath0.$ ] if @xmath314 are smooth differential @xmath33-forms on @xmath74 with compact supports , then @xmath317_{0,1}\ ] ] is @xmath70-measurable . see @xcite , lemma 1.3 . for @xmath319 let @xmath320 an element @xmath321 is called an admissible word of length @xmath322 . for @xmath223 define the @xmath70-measurable random variable @xmath323 to be the supremum of those @xmath324 such that @xmath325_{0,1}\neq0\ ] ] for some admissible word @xmath326 it follows that @xmath327 for almost surely . for @xmath319 @xmath328 define @xmath329_{0,1}\neq0\},\ ] ] then @xmath115@xmath330 are mutually disjoint @xmath70-measurable sets whose union is the whole space @xmath39 see @xcite for a more detailed discussion . let @xmath331 be the space of admissible words . for @xmath223 define the mapping @xmath332 @xmath333 if @xmath334 is such that @xmath335 it follows that @xmath336 is well - defined and @xmath70-measurable . intuitively , @xmath336 is the maximal admissible word such that the associated extended stratonovich s signature is nonzero . it is natural to believe that @xmath336 records a reasonable amount of information of the diffusion paths and as @xmath337 it might be possible to recover the diffusion paths . in section 2 , we defined piecewise linear trajectories ( p.l.t.s ) , parametrization of a p.l.t . , and introduced the concept of convergence in trajectory . in this section , we are going to show that if @xmath336 is regarded as a p.l.t . in @xmath91 , then by taking @xmath338 with probability one , @xmath339 converges in trajectory to @xmath340 which completes the proof of our main theorem . recall that a p.l.t . @xmath73 is essentially a finite sequence of points in @xmath74 ( not necessarily all distinct ) . for a p.l.t . @xmath80 @xmath341 is denoted as the new p.l.t . by removing the last point of @xmath342 let @xmath343 be two p.l.t.s . @xmath344 is called a _ sub_-p.l.t . of @xmath345 ( denoted by @xmath346 ) if @xmath344 is a subsequence of @xmath347 by the convergence result and the construction of @xmath224 in the last two subsections , if we denote @xmath348 ( respectively , @xmath349 ) as the associated p.l.t . of the piecewise linear path @xmath350 ( respectively , @xmath211 ) , then it is immediate that @xmath351 with probability one , and @xmath352 and @xmath353 both converges in trajectory to @xmath151 . therefore , it is natural to claim a certain kind of squeeze theorem for convergence in trajectory so we may conclude that @xmath354 also converges in trajectory to @xmath151 with probability one . the following result is a squeeze theorem for convergence in trajectory we are looking for , which is sufficient for our use . [ prop 3.3]assume that @xmath355 , @xmath356 are two sequence of p.l.t.s such that : \(1 ) the first points of @xmath357 and @xmath358 are identical ; \(2 ) the last two points of @xmath359 are identical ( @xmath360 ) . let @xmath361 be a parametrization of @xmath359 ( @xmath360 ) such that the partition points in @xmath362 belong to the partition points in @xmath363 and for any @xmath364 in @xmath365 @xmath366 ( this assumption implies that @xmath367 . ) let @xmath88 be a sequence of p.l.t.s such that @xmath368 and @xmath86 be a continuous path in @xmath74 such that @xmath369 then we can choose a parametrization @xmath370 of @xmath371 such that @xmath372 in particular , @xmath95 converges in trajectory to @xmath373 for any @xmath223 there exists @xmath374 such that for any @xmath375 @xmath376 on the other hand , it is obvious that we are able to construct a parametrization @xmath370 of @xmath371 such that : \(1 ) the partition points of @xmath362 belong to the partition points in @xmath377 and for any @xmath364 in @xmath362 , @xmath378 \(2 ) the partition points in @xmath370 belong to the partition points in @xmath379 and for any @xmath364 in @xmath377 @xmath380 let @xmath381 be the largest time spot in @xmath362 such that @xmath382 . let @xmath383 be any consecutive time spots in @xmath377 then on @xmath384 $ ] both @xmath385 and @xmath386 are linear . therefore , by an elementary result on the comparison for linear paths , we have @xmath387 if @xmath388 then @xmath389 it follows from ( [ gamma ] ) and ( [ comparison ] ) that @xmath390 if @xmath391 since the last two points of @xmath357 are identical ( denoted by @xmath392 ) , it follows that @xmath393 obviously @xmath394 but it may not be true for @xmath395 since it is possible that @xmath396 however , since @xmath397 there exists some @xmath398 such that @xmath399 due to the fact that @xmath400 on @xmath401,$ ] we arrive again at @xmath390 consequently , @xmath402 it follows that @xmath403 and in particular , @xmath95 converges in trajectory to @xmath373 now the proof is complete . in order to apply proposition [ prop 3.3 ] , we are going to modify @xmath349 and choose a suitable parametrization based on the one for @xmath349 specified in subsection 3.1 , which is chosen according to the successive visit time of larger boxes for the diffusion paths ( excluding revisit of the same box before visiting other boxes ) , so that the assumptions of proposition [ prop 3.3 ] are all verified . the method is the following . by using the notation in section 3.1 , if @xmath404 is such that @xmath405 then we modify the linear path @xmath211 on @xmath406 $ ] to a new path such that it does not move during @xmath407 $ ] and goes directly from its initial position at @xmath408 to @xmath409 at @xmath410 with constant velocity . if @xmath115 @xmath411 then we modify the linear path @xmath211 on @xmath412 $ ] ( in fact , @xmath211 remains still on @xmath412 $ ] ) to a path such that during @xmath407 $ ] and @xmath195 $ ] it remains still ( equals @xmath413 ) . it seems that such modification is trivial and does not change anything , but it does make a slight difference if we are using the associated p.l.t .. let @xmath414 be the modified piecewise linear path of @xmath211 and let @xmath415 be the associated p.l.t . of @xmath416 if we can prove that @xmath417 converges uniformly to @xmath151 with probability one , then all the assumptions in proposition [ prop 3.3 ] for the triple sequence @xmath418 are verified , and we will complete the proof of theorem [ thm 2.1 ] . in fact , it is just a simple modification of the arguments in subsection 3.1 . with probability one , @xmath419 converges uniformly to the diffusion paths @xmath93 as in the proof of proposition [ prop 3.1 ] , we need to estimate @xmath420 for some universal constant @xmath154 , which reduces to the estimation of @xmath421 @xmath115 @xmath422 and @xmath423 . for the first quantity , from the definition of @xmath414 we have @xmath424)=\widetilde{x}^{\varepsilon}([\zeta_{j-1}^{\varepsilon},\zeta_{j}^{\varepsilon}])\ ] ] on @xmath425 regardless of whether the path has visited the smaller box @xmath426 during @xmath427 $ ] . therefore , the event @xmath428 again implies that during @xmath427 $ ] , the path must have traveled through many narrow tunnels and spread far away from the box @xmath429 before visiting another box . more precisely , again we have @xmath430 the same as in the proof of proposition [ prop 3.1 ] . similar arguments apply to the estimation of the second quantity , and the third quantity has nothing to do with the polygonal approximation . therefore , we can apply exactly the same arguments as in the proof of proposition [ prop 3.1 ] to concluded that @xmath431 where @xmath154 is the universal constant chosen in that proof . now the proof of theorem [ thm 2.1 ] is complete . from the proof of theorem [ thm 2.1 ] , it is not hard to see that the global assumption ( c ) on the generating vector fields can be weakened to a local one to some extend . in fact , the only property of the vector fields we ve used from assumption ( c ) is that at every point on the boundary of @xmath432 , the vector fields @xmath433 do not generate a subspace of the tangent space at that point . therefore , it suffices to assume that for each @xmath434 and @xmath157 , there exists a small rotation @xmath435 ( an orthogonal transformation ) such that after rotating the box @xmath432 by @xmath435 with respect to its center , the vector fields do not generate a subspace of the tangent space at every point on the boundary . the smallness of the rotation @xmath435 can be quantified as follows . if we let @xmath436 be the rotated box , then @xmath435 should satisfy the condition that for any @xmath437 , @xmath438 this is to ensure that the geometric configuration , in particular the tunnel structure , is not damaged , so that the whole proof of theorem [ thm 2.1 ] carries through in the same way . the authors wish to thank the referees for their very careful reading on the manuscript and very useful suggestions . the authors are supported by the oxford - man institute at university of oxford .	in the present paper , we prove that with probability one , the stratonovich sigatures of a multidimensional diffusion process ( possibly degenerate ) over @xmath0,$ ] which is the collection of all iterated stratonovich s integrals of the diffusion process over @xmath0,$ ] determine the diffusion sample paths . _ msc : _ 60j60 ; 60g17 ; 60j45 _ keywords : _ hypoelliptic diffusions ; rough paths ; stratonovich s signatures
the effects of uniform magnetic fields on relativistic quark matter have been studied extensively in recent years ( for an overview , see @xcite ) . the main interest arises from the relevance of these effects on the physics of neutron stars , as well as on the dynamics of the quark - gluon plasma created in relativistic heavy ion collisions . it is known , that magnetars , a certain class of neutron stars , exhibit magnetic fields up to @xmath9 gau on their surface , and @xmath10 gau in their inner core @xcite . in @xcite , the dynamo effect during the first few seconds after the star s gravitational collapse is made responsible for the generation of these unusual magnetic fields . even larger magnetic fields are believed to be created in early stages of non - central heavy ion collisions . depending on the collision energies and impact parameters , the strength of these magnetic fields are estimated to be of the order @xmath11 gev@xmath12 at rhic and @xmath13 gev@xmath12 at lhc @xcite . gev@xmath12 corresponds to @xmath14 gau . ] the mechanism for the creation of strong magnetic fields in heavy ion collisions is originally discussed in @xcite ( see e.g. @xcite for a recent review on this topic ) . a uniform and spatially constant magnetic field breaks the lorentz invariance of the physical system , and consequently induces anisotropies in the parallel and perpendicular directions with respect to the direction of the background magnetic field . the anisotropies include those in the neutrino emission and magnetic winds in the astrophysics of magnetars @xcite , or anisotropies arising in the refraction indices @xcite , and decay constants of mesons in hot and magnetized quark matter @xcite . the latter may be relevant for the physics of heavy ion collisions . recently , the pressure anisotropies for a gas of protons and a gas of neutrons are studied in @xcite . it is , in particular , demonstrated that the inclusion of the anomalous magnetic moment ( amm ) of protons and neutrons increases the level of anisotropies in both cases . pressure anisotropies , arising from uniform and spatially fixed magnetic fields , are supposed to have significant effects on the elliptic flow @xmath15 in heavy ion collisions @xcite . in the present paper , we apply the method used in @xcite , to study the effect of various @xmath16 dependent and independent amm of quarks on the phase diagram of a two - flavor magnetized nambu jona - lasinio ( njl ) model at finite temperature @xmath0 and chemical potential @xmath1 . we will explore the phenomenon of magnetic catalysis ( mc ) , and inverse magnetic catalysis ( imc ) , as well as the reentrance of chiral symmetry broken and restored phases in the phase space of our model . the phenomenon of reentrance is already observed in the condensed matter physics of superfluidity @xcite and liquid crystals @xcite , as well as in the astrophysics of neutron stars @xcite . in our previous works @xcite , we have already studied in detail the complete @xmath0-@xmath1 , @xmath1-@xmath2 , and @xmath0@xmath2 phase portraits of the magnetized two - flavor njl model , without inclusion of the quark amm . we mainly worked in the supercritical regime of the njl coupling , i.e. , the coupling constant of the njl model was chosen in such a way that the system exhibited chiral symmetry breaking even for vanishing magnetic fields . in this case , the magnetic field only enhances the production of bound states ; they consist of mesons in chiral symmetry broken phase ( @xmath17sb ) and diquarks in color symmetry broken phase . in this sense , in @xcite , we focused on the phenomenon of mc of chiral and color symmetry breaking , and reported about various interesting phenomena , such as de haas - van alphen ( dhva ) oscillations @xcite , that lead to reentrant chiral and color symmetry broken phases , mainly in the regime of weak magnetic fields . the phenomenon of mc is discussed intensively in many contexts @xcite . at zero temperature , it arises from a dimensional reduction from @xmath18 to @xmath19 dimensions in the lowest landau level ( lll ) . this is believed to be responsible for the aforementioned enhancement of bound state formation . as a consequence , in a massless theory , whose original lagrangian density satisfies the chiral symmetry , a finite mass is dynamically created , which then breaks the chiral symmetry of the theory . finite temperature and baryonic chemical potential compete with external magnetic fields in this regard , and , except in the regime of very strong magnetic fields , suppress the formation of mesonic bound states ( see @xcite for more details ) . in our model , this regime is characterized by a threshold magnetic field @xmath20 gev@xmath12 , above which the dynamics of the fermions is solely dominated by lll . in this regime , the constituent quark mass monotonically increases with increasing @xmath2 as a consequence of mc . in the regime @xmath21 gev@xmath12 , however , the @xmath2 dependence of the condensate exhibits dhva oscillations , which occur whenever landau levels pass the quark fermi level . thus , because of these oscillations in the regime of @xmath22 , at a fixed temperature and for both vanishing and nonvanishing chemical potential , the chiral condensate exhibits two different behaviors : in some regimes , it increases with increasing @xmath2 , this is related to the phenomenon of mc . but , in some other regimes , it decreases with increasing @xmath2 . this is related to the phenomenon of imc . in @xcite , we have also studied the effect of external magnetic fields on @xmath0@xmath1 , @xmath0@xmath2 and @xmath1@xmath2 phase portraits of magnetized two - flavor njl model . we have shown that , as a consequence of the background magnetic fields , second order chiral phase transitions turn into first order ones , and , in the regime above @xmath20 gev@xmath12 , for fixed @xmath1 and @xmath0 , the critical temperature @xmath23 and chemical potential @xmath24 of @xmath17sb , monotonically increase with increasing @xmath2 . this has been supposed to be an indication of the phenomenon of mc , which is mainly a lll dominated effect . on the other hand , it has been shown , that in the regime of weak magnetic fields , the aforementioned dhva oscillations , lead again to two different behaviors of @xmath23 and @xmath24 as a function of @xmath2 : in some regimes , @xmath23 and @xmath24 increase with increasing @xmath2 , because of mc . in some other regimes , however , as a result of imc , @xmath23 and @xmath24 decrease with increasing @xmath2 . the phenomenon of reentrance of chiral ( color ) symmetry broken phases at certain fixed @xmath16 is also believed to be a consequence of dhva oscillations in the weak field regime @xcite . several other groups also investigate the effects of external magnetic fields on the phase diagram of hot qcd . these includes the groups working on lattice qcd at finite @xmath0 , zero @xmath1 and nonvanishing @xmath2 @xcite , or investigating the qcd phase diagram by making use of functional renormalization group method @xcite . many other groups elaborate different qcd - like models at finite @xmath16 , that exhibit @xmath17sb . the latter consists of , e.g. , njl - model @xcite , polyakov - linear-@xmath25 model @xcite , polyakov - njl model @xcite , polyakov - quark - meson model @xcite , njl model including axial chemical potential @xcite , ginzburg - landau model @xcite , charged scalar model with spontaneous @xmath17sb @xcite , njl model with dynamical amm generation @xcite . let us notice that the interest on this subject grows up after the still not fully comprehended lattice results by bali et al . in @xcite . they reported an unprecedented decrease of @xmath23 as a function of @xmath2 within an ab initio lattice qcd simulation , and declared this as a signature of imc in this framework . the physical explanation of the phenomenon of imc is , in particular , still under investigation : in @xcite , the interaction of the magnetic field with the sea quarks leading to a backreaction of the polyakov loop , and in @xcite the magnetic inhibition because of neutral meson fluctuations are made responsible for imc . in @xcite , it is explained how imc is induced by sphalerons . recently , in @xcite , the running of the corresponding coupling constants to the considered models , and therefore their @xmath2 dependence are taken into account , and it is shown how the lattice results on the @xmath2 dependence of @xmath23 in @xcite can be successfully reproduced . in a parallel development , the competition between mass and @xmath2 contributions to qcd pressure is considered up to two - loop radiative corrections in @xcite . it is argued that the deviation of the land @xmath26-factor from @xmath27 , which is of about @xmath28 , may produce sizable corrections to the qcd pressure . this affects the energy of the lll by turning the mass into @xmath29 . for the relevant @xmath30 gev@xmath12 , the correction will be in the mev range . it may thus compete with @xmath31 , and can not be ignored in quantitative studies @xcite . the effect of dynamically generated amm of fermions on the phase diagram of a magnetized one - flavor njl model is recently investigated in @xcite . here , inspired by the results presented in @xcite , a nonperturbative mechanism for the generation of the quark amm is introduced . to this purpose , a one - flavor njl lagrangian density including an appropriate tensor channel is used . it is in particular shown , that in the subcritical regime of the njl coupling , where the phenomenon of mc is solely responsible for @xmath17sb , two independent condensates emerge in the lll dominant regime . they correspond to the dynamical mass and the amm of quarks . the fact that the dynamical generation of amm is mainly a lll effect suggests that the schwinger linear - in-@xmath3 ansatz @xcite for the quark amm is not appropriate for massless fermions . this is argued to be in sharp contrast to theories with massive fermions , where a the schwinger ansatz is allowed in the limit of weak magnetic fields @xcite . moreover , in @xcite , the critical temperature of the second order phase transition of @xmath17sb is determined in a lll approximation , and it is shown that the generation of the quark amm increases @xmath23 as a function of @xmath2 , and therefore it can not be responsible for the phenomenon of imc . in @xcite , however , following the aforementioned proposal of @xmath2 dependent coupling constant , the effect of running coupling constant of the one - flavor njl model on @xmath23 is studied . it is shown , that in the regime of strong magnetic fields , characterized by @xmath32 , the coupling constant of the model becomes anisotropic , and only the coupling parallel to the direction of the background magnetic field receives contributions from quarks in a lll approximation . interestingly , as a result of a certain antiscreening arising from the these quarks , this coupling decreases with increasing @xmath2 . the dependence of @xmath23 on this specific coupling leads to the desired phenomenon of imc in the strong magnetic field limit . in the present paper , motivated by the above intriguing developments , we will consider the quark amm in the one - loop effective potential of a magnetized two - flavor njl model at finite @xmath0 and @xmath1 . in contrast to @xcite , and similar to the method used in @xcite , we will induce the quark amm by an additional term @xmath33 in the original lagrangian of the model , including massive quarks . the coefficient @xmath34 , proportional to the bohr magneton @xmath35 , will then be identified with the quark amm . here , @xmath36 depends on the constituent quark mass @xmath37 , where @xmath38 is the current quark mass and @xmath39 is the chiral condensate . this is also in contrast to @xcite , where the bohr magneton is inversely proportional to current ( bare ) quark mass @xmath38 . the aforementioned additional term @xmath33 in the lagrangian density of the njl model leads to the energy dispersion @xmath40^{2}},\end{aligned}\ ] ] for up and down quarks in the presence of a constant magnetic field . here , @xmath41 with @xmath42 , @xmath43 labels the landau levels , @xmath44 states for the spin of quarks , @xmath45 labels the up ( @xmath46 ) and down ( @xmath47 ) flavors , and @xmath48 is the charge of up and down quarks moreover , @xmath49 is related to the deviation of the land @xmath26-factor from @xmath27 . the above additional term @xmath50 is equivalent to the linear - in-@xmath3 ansatz by schwinger @xmath51 for @xmath50 , with the nonperturbative ( effective ) bohr magneton @xcite @xmath52 . let us notice , that in the one - loop level , @xmath49 is proportional to the electromagnetic fine structure constant @xmath53 . perturbatively , it receives radiative corrections from the vertex function of quarks and the background photon field . in the framework of constituent quark model @xcite , it appears in the ratio @xmath54 , with @xmath55 the constituent ( effective ) quark mass . ] , @xmath56 the nuclear magneton , @xmath57 the magnetic moment of @xmath58-th quark flavor , and @xmath59 the proton mass ( see appendix [ appa ] for more details ) . as it is argued in @xcite , choosing the experimental values for @xmath60 and @xmath59 , the ratio @xmath61 is fixed to be @xmath62 mev for the up quark and @xmath63 mev for the down quark @xcite . a phenomenological constant value for @xmath64 , compatible with the constituent quark model , can then be determined by choosing an appropriate constant value for the constituent ( effective ) mass @xmath55 , that yields @xmath49 , and consequently @xmath64 through given values of @xmath65 and the definition @xmath66 . in this paper , we will consider three different sets for the dimensionful coupling @xmath67 in @xmath41 ; two of them will be @xmath16 independent , and arise , in the framework of constituent quark model , by separately choosing @xmath68 mev and @xmath69 mev , and fixing @xmath49 through the phenomenologically given ratio @xmath61 , as described above . the third set of @xmath67 will depend on @xmath7 and @xmath2 , and include the @xmath0-dependent one - loop perturbative correction to @xmath49 , generalized from the qed results presented in @xcite . plugging first ( [ z1 ] ) with given values of @xmath64 into the one - loop effective potential of our model , and then minimizing the resulting expression with respect to @xmath31 , we will determine the @xmath7 and @xmath2 dependence of the constituent quark mass for each fixed set of @xmath67 . the complete @xmath0-@xmath1 , @xmath1-@xmath2 and @xmath0-@xmath2 phase portrait of the model for various fixed @xmath70 and @xmath64 will also be explored . we will show that for large enough phenomenological value for @xmath64 , the phenomenon of imc occurs , i.e. in certain regimes of the parameter space , the critical temperature ( chemical potential ) @xmath23 ( @xmath24 ) decreases with increasing @xmath2 . moreover , for @xmath16 dependent @xmath64 , including the one - loop perturbative correction for @xmath71 , the phenomenon of reentrance of @xmath17sb occurs for certain fixed values of @xmath0 and @xmath1 , and in the regime of weak magnetic fields . the organization of this paper is as follows : in sec . [ sec2 ] , we will introduce the quark amm in the magnetized two flavor njl model and will determine the one - loop effective action of this model at finite @xmath8 and @xmath2 . we will also introduce three different sets of the factor @xmath67 , mentioned above . in sec . [ subsec3a ] , the @xmath7 and @xmath2 dependence of the constituent quark mass @xmath31 will be presented for each fixed @xmath64 . in sec . [ subsec3b ] , the complete phase portrait of the model will be explored for various fixed @xmath64 . we will also study , in sec . [ subsec3c ] , the effect of different sets of @xmath64 on the pressure anisotropies in the longitudinal and transverse directions with respect to the direction of the magnetic field . eventually , the @xmath2 dependence of the magnetization @xmath55 will be demonstrated , and the effect of @xmath64 on the product of @xmath72 as a function of @xmath2 will be studied . section [ sec4 ] is devoted to our concluding remarks . in appendix [ appa ] , we will use the constituent quark model , and argue how the constant values of @xmath67 can be determined by the phenomenological data of the magnetic moment of protons and neutrons . we start with the lagrangian density of a two - flavor njl model in the presence of a uniform magnetic field , @xmath73 ^ 2 + [ \bar{\psi}(x)i\gamma_5\boldsymbol{\tau}\psi(x)]^2\bigg\}.\end{aligned}\ ] ] here , the dirac field @xmath74 carries two flavors , @xmath75 , and three colors , @xmath76 . in the presence of a constant magnetic field , the covariant derivative @xmath77 is defined by @xmath78 , where the quark charge matrix is given by @xmath79 , and the gauge field @xmath80 is chosen so that it leads to a uniform magnetic field @xmath81 , aligned in the third direction . the up and down current quark masses are assumed to be equal , and denoted by @xmath82 . this guarantees the isospin symmetry of the theory in the limit of vanishing magnetic field . the latter is introduced by the term proportional to @xmath83 in the fermionic kernel of the lagrangian density @xmath84 . here , @xmath85 $ ] , and the abelian field strength tensor is given by @xmath86}^{\mbox{\tiny{ext.}}}$ ] , with @xmath87 given above . the proportionality factor @xmath34 is related to the quark amm ( for more details , see appendix a ) . in the present two - flavor njl model , @xmath34 is defined by @xmath88 , where @xmath5 and @xmath89 are @xmath90 matrices in the flavor space . at one - loop level , and for a system with isospin symmetry @xmath91 for both @xmath45 flavors . here , @xmath53 is the electromagnetic fine structure constant , and @xmath35 is given in term of the electric charge @xmath92 and the quark constituent mass @xmath37 , where @xmath38 is the current quark mass and @xmath39 the chiral condensate . ( see below ) . ] to introduce the chiral condensate @xmath39 , let us rewrite the njl lagrangian ( [ f1 ] ) in a semi - bosonized form @xmath93 where the mesonic fields @xmath25 and @xmath94 are defined by @xmath95 here , as in ( [ f1 ] ) , @xmath96 is the njl mesonic coupling and @xmath97 are the pauli matrices . to arrive at ( [ f2 ] ) , @xmath98 is used . integrating out the fermionic degrees of freedom , and using @xmath99}=\int { \cal{d}}\bar{\psi}{\cal{d}}\psi\exp\left(i\int d^{4}x{\cal{l}}_{sb}\right),\end{aligned}\ ] ] the effective action of mesons @xmath100 , @xmath101 is derived . it is given in terms of a tree - level action @xmath102=-\int d^{4}x \frac{(\sigma^{2}+\boldsymbol{\pi}^{2})}{4g},\end{aligned}\ ] ] and a one - loop effective action @xmath103,\end{aligned}\ ] ] where the inverse fermion propagator is formally given by @xmath104 here , @xmath105 . expanding the effective action around a constant configuration @xmath106 for the mesonic fields @xmath100 , the constituent mass @xmath107 turns out to be constant , and can be given by @xmath108 . by carrying out the trace operation over color ( @xmath109 ) , flavor ( @xmath58 ) , spinor ( @xmath25 ) degrees of freedom , as well as over the four - dimensional space - time coordinate ( @xmath110 ) , the one - loop effective action of this model reads @xmath111 here , the energy spectrum of up and down quarks in the presence of external magnetic fields and for nonvanishing amm is given by @xmath112+m^{2}\right)^{1/2}-s\kappa_{f}q_{f}eb ] ^{2}},\nonumber\\\end{aligned}\ ] ] where @xmath113 labels the landau levels , and @xmath44 stands for the spin of the quarks . moreover , @xmath114 and @xmath42 , with @xmath45 . the energy @xmath115 arises either by computing the fermion determinant of the present two - flavor njl model including the quark amm , or by solving the modified dirac equation @xmath116 , with @xmath117 , using _ e.g. _ , the ritus eigenfunction method @xcite . following this method , which is descibed also in @xcite for two- and four - dimensional space - time , it turns out that for nonvanishing fermion amm , only the energy eigenvalues of the dirac operator are modified , while the ritus eigenfunctions remain the same as for vanishing amm . let us notice that , the additional @xmath118 term in the dirac operator commutes with @xmath119 . this is why the modified dirac operator @xmath120 has the same eigenfunctions as the ordinary dirac @xmath121 for vanishing amm . let us compare , at this stage , the energy dispersion ( [ f9 ] ) , with the energy dispersion ( [ z1 ] ) appearing also in @xcite . wherever it was necessary . ] according to schwinger @xcite , @xmath50 is linear in @xmath3 , and is given by @xmath122 . plugging @xmath52 into this relation and using @xmath123 , we obtain @xmath124 , which appears also in ( [ f9 ] ) . thus , starting from the above @xmath125 , the two energy dispersion relations ( [ f9 ] ) and ( [ z1 ] ) are equal . the schwinger linear - in-@xmath3 ansatz for amm is recently discussed in the literature @xcite . in particular , it is shown , that this ansatz is only valid in the weak - field limit @xmath126 . in the present paper , however , we will use it for the whole range of @xmath2 . to explain this apparent discrepancy , let us notice that the crucial difference between our approach and the one presented in @xcite , is that in our approach the bohr magneton @xmath36 is defined in term of the constituent quark mass @xmath108 , with @xmath38 the current quark mass and @xmath39 the chiral condensate , while in @xcite , the schwinger ansatz is defined in term of @xmath127 , and is independent of @xmath39 . having this in mind , we expect , that the difficulties related to the schwinger linear ansatz with @xmath128 , described in @xcite , do not occur in our work . as described in sec . [ sec1 ] , in the present paper , the quark amm will be introduced in the framework of a constituent quark model . this is in contrast to @xcite , where a dynamical symmetry breaking is responsible for its generation . in @xcite , for instance , the quark amm is dynamically induced through a nonvanishing spin - one condensate . here , starting from a massless theory , it is shown that since a nonperturbative mechanism of quark pairing , mainly in the lll , is responsible for the dynamical generation of amm , the schwinger linear - in-@xmath3 ansatz , @xmath129 , is not even valid in the aforementioned weak - field approximation .. moreover , the nonperturbative result for amm , arising in the lll approximation @xcite is non - analytic , and can not be taylor expanded in the orders of @xmath3 . ] in the present paper , however , in contrast to @xcite , we do not start with a massless dirac theory . moreover , we will work , in contrast to @xcite , in the supercritical regime of the njl model , i.e. , we will choose the njl coupling @xmath96 in such a way that the model exhibits a dynamical mass , even for zero @xmath2 . we will consider the contributions of all landau level , and will not restrict ourselves to lll , as in @xcite , nor to one - loop approximation in the lll , as in @xcite . hence , starting from the schwinger linear ansatz with @xmath36 instead of @xmath128 seems to be reasonable within the constituent quark model . here , although the quark amm is not dynamically generated , as in @xcite , but it is related to the dynamically generated quark mass @xmath31 through the nonperturbative ( effective ) bohr magneton @xcite @xmath36 in @xmath122 . the constituent quark mass @xmath31 is determined by minimizing the thermodynamic ( one - loop effective ) potential , arising from ( [ f8 ] ) . for a constant configuration @xmath39 , the one - loop effective potential of the theory is given by performing the remaining determinant over the coordinate space in ( [ f8 ] ) . this leads to the one - loop effective potential @xmath130 , defined by @xmath131 , where @xmath132 denotes the four - dimensional space - time volume . in momentum space , the aforementioned determinant is evaluated by the standard replacement @xmath133 where temperature @xmath0 and chemical potential @xmath1 are introduced by replacing @xmath134 with @xmath135 . here , the matsubara frequencies @xmath136 with @xmath137 are labeled by @xmath138 . after performing the sum over @xmath138 , the effective potential of the two - flavor njl model , including the tree - level and the one - loop part is given by @xmath139 in the rest of this paper , we will use the above effective potential to study the phase diagram of the magnetized two - flavor njl model at finite @xmath16 and for a given value of @xmath67 . to do this , we will follow two different approaches : in the first approach , we assume that the value of @xmath123 , appearing in the quark energy dispersion relation @xmath140 from ( [ f9 ] ) , is constant and independent of @xmath16 . minimizing the effective potential @xmath141 from ( [ f11 ] ) with respect to @xmath31 , we will numerically determine the constituent quark mass @xmath31 as a function of @xmath16 and for a fixed value of @xmath67 . in this approach , @xmath142 will depend on @xmath16 only through the constituent mass @xmath31 . to fix @xmath6 , we will follow the method described in appendix a. here , two different sets of constant @xmath143 are introduced , which are compatible with the constituent quark model [ see ( [ appa10 ] ) and ( [ appa12 ] ) ] . to have a sizable quark amm , we will use @xmath144 [ see also ( [ appa9 ] ) ] . the second set yields smaller values of @xmath145 , @xmath146 [ see also ( [ appa11 ] ) ] . in sec . [ sec3 ] , we will in particular show , that once @xmath6 in ( [ f9 ] ) is chosen to be @xmath147 from ( [ f12 ] ) , the phenomenon of imc occurs in the phase diagram of our two - flavor magnetized njl model . in the second approach , we will consider the leading one - loop correction to the quark ( electromagnetic ) amm at nonzero @xmath0 and vanishing @xmath1 and @xmath2 . let us denote it by @xmath148 ( see below ) . to determine the constituent quark mass @xmath31 , we will insert @xmath123 with @xmath149 into ( [ f9 ] ) , and then ( [ f9 ] ) into the one - loop effective potential ( [ f11 ] ) , whose minima will then lead to @xmath16 dependent @xmath31 . to determine the one - loop correction @xmath150 , we use the result presented in @xcite for thermal qed , and generalize it to the case of qcd . in @xcite , the anomalous magnetic moment of electrons at one- and two - loop orders are computed at finite @xmath0 , zero @xmath1 and for vanishing @xmath2 . since in our model the electromagnetic coupling between the quarks and the external photons is to be considered , the qed results presented in @xcite are directly applicable for our qcd - like model , including quarks with different flavors @xmath45 . it is enough to multiply the corresponding qed result with @xmath151 , where @xmath152 is the charge of the @xmath58-th quark flavor . we therefore have @xmath153 with the one - loop contribution @xmath154 for @xmath45 . here , @xmath155 is the constituent quark mass at zero @xmath16 , and @xmath156+{\cal{a}}(z)-{\cal{b}}(z ) \right),\end{aligned}\ ] ] with @xmath157 the electromagnetic fine structure constant , and the functions @xmath158 and @xmath159 given by @xmath160 here , @xmath161 is the well - known exponential integral function , defined by @xmath162 let us notice , that since @xmath163 , the one - loop contribution @xmath150 is indeed the dominant radiative correction . this is why , the higher order loop corrections will be neglected in the present paper . the aim of this paper is to study the effect of the quark amm on a hot and magnetized quark matter . to this purpose , the thermodynamic potential of a two - flavor njl model is determined in the previous section . it mimics the thermodynamic properties of quark matter at high temperature , finite density and in the presence of an external magnetic field . the quantity , @xmath164 , which is introduced in the original lagrangian ( [ f1 ] ) as the coefficient corresponding to the spin - field interaction term @xmath165 , appears in the quark energy dispersion relation @xmath140 from ( [ f9 ] ) in the one - loop effective potential ( [ f11 ] ) , essentially in the combination with the constituent quark mass in @xmath166 . using ( [ f11 ] ) , it is now possible to determine the thermodynamic properties of the present quark model , and , in particular , to study various effects of the quark amm on hot and magnetized quark matter . as we have discussed in sec . [ sec1 ] , the most important effect of the presence of an uniform magnetic field on a system of charged fermions is the phenomenon of mc of chiral symmetry breaking . the first signature of the occurrence of this phenomenon is that the value of the chiral condensate @xmath39 , and consequently the constituent quark mass @xmath108 increase with increasing the strength of the external magnetic field @xmath2 ( see @xcite for detailed discussions ) . to study the impact of the quark amm on this specific effect , we will first determine , in sec . [ subsec3a1 ] , the @xmath2 dependence of @xmath31 for zero chemical potential , at fixed temperature and for three different sets of @xmath6 . they will be denoted by @xmath167 and @xmath168 . here , @xmath169 and @xmath170 correspond to the pair @xmath171 from ( [ f12 ] ) and ( [ f13 ] ) , and , @xmath168 corresponds to the @xmath0-dependent @xmath64 given in ( [ f14 ] ) . the latter includes one - loop perturbative correction to the amm of quarks and is written as a function of the constituent quark mass @xmath31 . we will show that whereas for @xmath170 and @xmath168 , @xmath31 increases with increasing @xmath2 , for @xmath169 , it decreases with increasing @xmath2 . moreover , in the latter case , the @xmath2 dependence of @xmath31 indicates a first order phase transition at certain critical @xmath172 even at @xmath173 mev . this preliminary , unexpected result can be regarded as an indication of the phenomenon of imc . as it is shown in @xcite , the formation of the chiral condensate is suppressed by increasing temperature . in sec . [ subsec3a2 ] , we will study the effect of finite temperature on @xmath31 for zero chemical potential and various values of @xmath2 and @xmath6 . we will then compare the corresponding results to @xmath167 and @xmath168 , and elaborate on the difference between the effect of these three different choices for @xmath6 . in sec . [ subsec3b ] , we will then explore the complete phase portrait of our two - flavor hot and magnetized njl model . the effect of external magnetic field on the nature of phase transition for @xmath174 gev@xmath175 is previously studied in @xcite . it is , in particular , shown that for zero chemical potential , the critical temperature @xmath176 essentially increases with increasing @xmath2 . this can be regarded as the another indication of the phenomenon of mc . in sec . [ subsec3b1 ] , we will first study the @xmath0@xmath2 phase diagram of our two - flavor hot and magnetized njl model for zero and nonzero @xmath1 and for different @xmath6 . we will , in particular , show that the critical temperature of the phase transition decreases with increasing @xmath2 once @xmath6 is chosen to be @xmath169 . this can be regarded as another signature of the aforementioned phenomenon of imc . in secs . [ subsec3b2 ] and [ subsec3b3 ] , we will then study the @xmath1@xmath2 and @xmath0@xmath2 phase diagrams for various fixed temperatures and magnetic fields , respectively . we will , in particular , compare the results for @xmath169 and @xmath170 with @xmath174 gev@xmath175 , previously discussed in @xcite . in sec . [ subsec3c ] , we will finally study the pressure anisotropy between the longitudinal and transverse pressures with respect to the direction of the external magnetic field . moreover , the @xmath2 dependence of the magnetization of the quark matter will be determined for zero and nonzero chemical potential , temperature and for @xmath167 and @xmath168 . we will show that whereas @xmath6 has essentially no effect on the pressure anisotropy , larger values of @xmath177 leading to sizable values of the quark amm , suppress the product @xmath72 , where @xmath55 is the quark matter magnetization . the relation between discontinuities appearing in @xmath55 and the first order phase transitions will be also discussed . to perform the numerical analysis in this section , we will essentially use the same method as described in @xcite and @xcite . in order to determine the constituent quark mass @xmath31 , we will numerically solve the gap equation corresponding to the thermodynamic potential ( [ f11 ] ) @xmath178 here , @xmath108 . our specific choice for free parameters of our model , the ultraviolet ( uv ) momentum cutoff @xmath179 , the njl ( chiral ) coupling constant @xmath96 , and the current quark mass @xmath38 , is as follows : @xmath180 the numerical integration over @xmath181 , appearing in ( [ f11 ] ) will be performed using a smooth cutoff function @xmath182 for vanishing @xmath2 , and @xmath183}- \lambda}{a}\right)},\end{aligned}\ ] ] for nonvanishing @xmath2 . here , @xmath152 is the electric charge of the @xmath58-th quark , @xmath43 labels the landau levels , @xmath44 stands for positive and negative spins of the quarks and @xmath184 . moreover @xmath185 is a free parameter , which is related to the sharpness of the cutoff scheme . as in @xcite , @xmath185 is chosen to be @xmath186 , where the uv cutoff @xmath179 is given in ( [ a2 ] ) . to determine the constituent quark mass @xmath31 , the global minima of @xmath141 from ( [ f11 ] ) are to be determined . in the chiral limit @xmath187 , the @xmath17sb is characterized by nonvanishing chiral condensate @xmath39 , and the chiral symmetry restored ( @xmath17sr ) phase by @xmath188 . as it is shown in @xcite , in the limit of vanishing @xmath38 and @xmath6 , the presence of external magnetic field induces a first order phase transition from the @xmath17sb into the @xmath17sr phase . for @xmath189 and @xmath174 gev@xmath175 , however , we expect a smooth crossover from the chiral @xmath190 , characterized by @xmath191 , into the pseudo - chiral symmetry restored ( p@xmath17sr ) phase , characterized by @xmath188 mev and @xmath192 . in sec . [ subsec3c ] , the order of the phase transition for @xmath189 and @xmath193 will be elaborated . dependence of quark constituent mass @xmath108 is demonstrated for @xmath169 ( red dashed line ) and @xmath170 ( black solid line ) at @xmath173 mev . the behavior of @xmath31 as a function of @xmath2 for @xmath169 suggests a first order phase transition at a critical @xmath194 gev@xmath12 . this can be regarded as possible signature of the phenomenon of imc . ( b ) the @xmath2 dependence of @xmath31 is compared for @xmath170 ( black solid line ) and @xmath168 ( red dashed line ) at @xmath173 mev . the fact that @xmath31 increases with increasing @xmath2 is related to the phenomenon of mc.,title="fig:",width=321,height=226 ] dependence of quark constituent mass @xmath108 is demonstrated for @xmath169 ( red dashed line ) and @xmath170 ( black solid line ) at @xmath173 mev . the behavior of @xmath31 as a function of @xmath2 for @xmath169 suggests a first order phase transition at a critical @xmath194 gev@xmath12 . this can be regarded as possible signature of the phenomenon of imc . ( b ) the @xmath2 dependence of @xmath31 is compared for @xmath170 ( black solid line ) and @xmath168 ( red dashed line ) at @xmath173 mev . the fact that @xmath31 increases with increasing @xmath2 is related to the phenomenon of mc.,title="fig:",width=321,height=226 ] dependence of quark constituent mass @xmath108 is demonstrated for @xmath169 ( red dashed line ) and @xmath170 ( black solid line ) at @xmath195 mev and @xmath196 mev . as in the @xmath197 mev case of fig . [ fig1 ] , the behavior of @xmath31 as a function of @xmath2 for @xmath169 suggests a first order phase transition at a critical @xmath198 gev@xmath12 . this is a possible signature of imc . ( b ) the @xmath2 dependence of @xmath31 is compared for @xmath170 ( black solid line ) and @xmath168 ( red dashed line ) at @xmath195 mev and @xmath196 mev . , title="fig:",width=321,height=226 ] dependence of quark constituent mass @xmath108 is demonstrated for @xmath169 ( red dashed line ) and @xmath170 ( black solid line ) at @xmath195 mev and @xmath196 mev . as in the @xmath197 mev case of fig . [ fig1 ] , the behavior of @xmath31 as a function of @xmath2 for @xmath169 suggests a first order phase transition at a critical @xmath198 gev@xmath12 . this is a possible signature of imc . ( b ) the @xmath2 dependence of @xmath31 is compared for @xmath170 ( black solid line ) and @xmath168 ( red dashed line ) at @xmath195 mev and @xmath196 mev . , title="fig:",width=321,height=226 ] is demonstrated for @xmath196 mev and at different temperatures @xmath199 mev , @xmath200 mev and @xmath201 mev . here , @xmath6 is held fixed to be @xmath169 . the critical magnetic fields corresponding to @xmath202 mev are denoted by @xmath203 , @xmath204 and @xmath205 , respectively . they are given by @xmath206 gev@xmath12 , @xmath207 gev@xmath12 and @xmath208 gev@xmath12 . as it turns out , @xmath172 deceases with increasing @xmath0 . this is an indication of imc , that apparently occurs once the quark amm is large enough.,width=321,height=226 ] in figs . [ fig1 ] and [ fig2 ] , the @xmath2 dependence of the constituent quark mass @xmath108 is demonstrated for @xmath169 , @xmath170 and @xmath168 , @xmath196 mev and at @xmath197 mev ( fig . [ fig1 ] ) as well as @xmath195 mev ( fig . [ fig2 ] ) . in figs . [ fig1](a ) and [ fig2](a ) , the results for @xmath169 ( red dashed lines ) are compared with the corresponding data to @xmath170 ( black solid lines ) . the latter is then compared with the results for @xmath168 ( red dashed lines ) in figs . [ fig1](b ) and [ fig2](b ) . as it turns out that , whereas for @xmath170 and @xmath168 the constituent quark mass @xmath31 increases with increasing @xmath2 , for @xmath169 , that yields a sizable quark amm , @xmath31 abruptly decreases at a certain critical magnetic field @xmath172 to a value @xmath192 . , @xmath31 turns out to be equal to @xmath38 after the transition into the p@xmath17sr phase is occurred . ] this indicates that , once @xmath6 is chosen to be @xmath169 , for @xmath196 mev a first order phase transition occurs at this specific @xmath172 , which is for @xmath197 mev and @xmath195 mev equal to @xmath194 gev@xmath12 and @xmath209 gev@xmath12 , respectively . as we have argued above , the fact that @xmath31 increases with increasing @xmath2 can be regarded as one of the most important signatures of the phenomenon of mc , which is supposed to enhance the formation of the quark condensate in the supercritical regime of njl coupling @xmath96 . on the other hand , by definition , the phenomenon of imc is related to the suppression of bound state formation even for large values of @xmath2 . thus , the fact that @xmath31 decreases with increasing @xmath2 for @xmath210 from ( [ f12 ] ) is related to the phenomenon of imc for @xmath196 mev and @xmath211 mev . to study the effect of temperature on the aforementioned critical magnetic field @xmath172 for @xmath169 , the @xmath2 dependence of @xmath31 is considered in fig . [ fig3 ] for three different temperatures , @xmath199 mev ( red dashed line ) , @xmath212 mev ( black solid line ) and @xmath201 mev ( blue dashed line ) . the critical magnetic fields corresponding to these temperatures are denoted by @xmath213 and @xmath205 . numerically , they are given by @xmath214 gev@xmath12 , @xmath215 gev@xmath12 and @xmath216 gev@xmath12 . it turns out , that @xmath172 decreases with increasing @xmath0 . this can again be considered as a signature of the phenomenon of imc , which seems to occur once the quark amm is large enough . later , we will study the @xmath0@xmath2 phase diagram for @xmath196 mev and @xmath167 and @xmath168 . we will show , that in contrast to the case of @xmath170 and @xmath168 , the critical temperature corresponding to @xmath169 decreases with increasing @xmath2 ( see fig . [ fig8 ] for more details ) . according to the results from @xcite , in the chiral limit @xmath187 , the two - flavor njl model at finite @xmath0 and zero @xmath1 and @xmath2 exhibits a second order phase at @xmath217 mev , and for nonvanishing @xmath2 , @xmath23 increases with increasing @xmath2 up to @xmath218 mev for @xmath219 gev@xmath12 @xcite . hence , the temperature interval @xmath220 $ ] mev can be regarded as the regime of phase transition of this qcd - like model . on the other hand , as it is shown in @xcite , for @xmath189 , the above - mentioned second order phase transition turns into a smooth crossover . in this case the temperature interval @xmath220 $ ] mev indicates the crossover region from the @xmath17sb into the p@xmath17sr phase . in figs . [ fig4 ] and [ fig5 ] , the @xmath2 dependence of the constituent quark mass @xmath31 is demonstrated at @xmath221 mev , for @xmath196 mev and different @xmath177s in the crossover regime of our model . comparing the results for @xmath169 in fig . [ fig3 ] at @xmath222 mev with the results for @xmath169 at @xmath223 mev from fig . [ fig4](a ) , it turns out that the first order phase transition , appearing at @xmath222 mev , turns into a smooth crossover at @xmath223 mev . as concerns the results for @xmath170 and @xmath168 , we observed in figs . [ fig1](b ) and [ fig2](b ) , that at @xmath222 mev , @xmath31 increases with increasing @xmath2 . at @xmath223 mev , however , the situation changes . whereas for @xmath224 mev , @xmath31 increases with increasing @xmath2 for both @xmath170 and @xmath168 [ see the black circles in figs . [ fig4](b ) and [ fig4](c ) ] , at @xmath225 mev and @xmath226 mev , @xmath31 decreases first with @xmath2 and then suddenly increases for certain critical magnetic fields [ see gray squares for @xmath225 mev and red diamonds for @xmath226 mev in figs . [ fig4](b ) and [ fig4](c ) ] . comparing the data for @xmath170 and @xmath168 at @xmath225 mev [ gray squares in figs . [ fig4](b ) and [ fig4](c ) ] , it turns out that for @xmath170 , one and for @xmath168 two critical magnetic fields exist . they are given by @xmath227 gev@xmath12 for @xmath170 and @xmath228 gev@xmath12 and @xmath229 gev@xmath12 for @xmath168 . this means that in the latter case , the system is first in the @xmath17sb phase , enters at @xmath230 the p@xmath17sr phase , and reenters the @xmath17sb phase at @xmath231 ( see fig . [ fig8 ] for more detailed analysis of the phase transitions for different @xmath6 and the discussions related to the phenomenon of reentrance from the @xmath17sb into the p@xmath17sr phase ) . let us now compare the data corresponding to @xmath170 and @xmath168 at @xmath226 mev and @xmath196 mev in figs . [ fig4](b ) and [ fig4](c ) ( red diamonds ) . it turns out , that for @xmath170 , @xmath31 decreases with increasing @xmath2 in the whole range @xmath232 $ ] gev@xmath12 , while , for @xmath168 , there are two critical magnetic fields @xmath233 gev@xmath12 and @xmath234 gev@xmath12 . in other words , for @xmath168 , the quark matter is first in the @xmath17sb phase , enters the p@xmath17sr phase , and then reenters the @xmath17sb phase . our findings in fig . [ fig8 ] confirm this conclusion . in fig . [ fig5 ] , we compare the @xmath2 dependence of @xmath31 for different @xmath6 : @xmath174 gev@xmath175 ( gray circles ) , @xmath235 ( blue squares ) , @xmath236 ( black diamonds ) @xmath168 ( red triangles ) , for @xmath196 mev and at @xmath224 mev [ fig . [ fig5](a ) ] , @xmath225 mev [ fig . [ fig5](b ) ] and @xmath226 mev [ fig . [ fig5](c ) ] . we limited ourselves to the regime @xmath237 $ ] gev@xmath12 in order to magnify the difference in the @xmath2 dependence of @xmath31 , especially for @xmath174 gev@xmath175 , @xmath170 and @xmath168 . according to these results , no difference between the data corresponding to @xmath174 gev@xmath175 and @xmath168 occurs as long as @xmath238 mev . on the other hand , whereas , according to figs . [ fig1](b ) and [ fig2](b ) , there is no difference between the data corresponding to @xmath170 and @xmath168 at @xmath239 mev , at @xmath223 mev the @xmath2 dependence of @xmath31 are different for @xmath170 and @xmath168 . in particular , it turns out that at @xmath226 mev , @xmath196 mev and for @xmath168 , similar to the case of @xmath169 , @xmath31 first decreases with increasing @xmath2 up to a certain critical @xmath240 gev@xmath12 , then suddenly falls down to a value @xmath241 mev . this is again an indication of the phenomenon of imc for @xmath196 mev , at high temperature @xmath242 mev and @xmath243 gev@xmath12 . let us finally notice that the data for @xmath170 in figs . [ fig5](b ) and [ fig5](c ) indicates a smooth crossover from the @xmath17sb into the p@xmath17sr phase in the regime @xmath244 mev and @xmath245 gev@xmath12 and for @xmath196 mev ( see also fig . [ fig8 ] for more details ) . according to the results from figs . [ fig1]-[fig5 ] for fixed @xmath1 , @xmath2 and @xmath6 , finite temperature suppresses the formation of the chiral condensate @xmath39 ( see also @xcite ) . consequently , it is expected that @xmath31 decreases with increasing @xmath0 . in fig . [ fig6 ] , the @xmath0 dependence of the constituent quark mass is demonstrated for @xmath196 mev , @xmath246 gev@xmath12 ( blue circles ) , @xmath247 gev@xmath12 ( gray squares ) , @xmath248 gev@xmath12 ( red diamonds ) and different @xmath169 ( panel a ) , @xmath170 ( panel b ) and @xmath168 ( panel c ) . as it turns out from the results in fig . [ fig6](a ) , whereas for small values of @xmath246 gev@xmath12 , @xmath31 decreases continuously with increasing @xmath0 , for larger values of @xmath249 gev@xmath12 , @xmath31 decreases with increasing temperature up to a certain critical temperature @xmath23 , which is @xmath250 mev for @xmath247 gev@xmath12 and @xmath251 mev for @xmath248 gev@xmath12 . this behavior indicates a first order phase transition from the @xmath17sb into the p@xmath17sr phase for @xmath196 mev and @xmath169 . interestingly , the critical temperature corresponding to this phase transition decreases with increasing @xmath2 , which is , as aforementioned , another indication of the phenomenon of imc once @xmath6 is large enough . for @xmath170 and @xmath168 , the situation is rather different . according to the results from fig . [ fig6](b ) , for @xmath170 , in contrast to @xmath169 , @xmath31 decreases smoothly with increasing @xmath0 for all values of @xmath252 gev . for @xmath0-dependent @xmath168 , however , whereas for @xmath246 gev@xmath12 the constituent quark mass @xmath31 decreases smoothly with increasing @xmath0 up to a certain critical temperature @xmath253 mev , for @xmath247 gev@xmath12 and @xmath248 gev@xmath12 , discontinuities occurs at critical temperatures @xmath254 mev and @xmath255 mev , respectively . these kind of discontinuities in the @xmath0 dependence of @xmath31 are indications of first order phase transitions from the @xmath17sb into the p@xmath17sr phase . to compare the data for different @xmath6 for fixed @xmath1 and @xmath2 , the @xmath0 dependence of @xmath31 is plotted in fig . [ fig7 ] for @xmath196 mev and @xmath246 gev@xmath12 ( panel a ) , @xmath247 gev@xmath12 ( panel b ) and @xmath248 gev@xmath12 ( panel c ) for @xmath0 independent @xmath169 ( blue circles ) , @xmath170 ( gray squares ) and @xmath168 ( red triangles ) . as it turns out , for small value of @xmath246 gev@xmath12 , there is almost no difference between the @xmath0 dependence of @xmath31 for @xmath167 and @xmath168 , and the transition from the @xmath17sb into the p@xmath17sr phase turns out to be a smooth crossover . in contrast , according to the results from figs . [ fig7](b ) and [ fig7](c ) , for @xmath196 mev and @xmath247 gev@xmath12 as well as @xmath248 gev@xmath12 , whereas in the case of @xmath170 , @xmath31 decreases smoothly with increasing @xmath0 , in two other cases of @xmath169 and @xmath168 , there exists a certain critical temperature below which @xmath31 decreases smoothly and above which @xmath31 turns out to be @xmath192 ( for the values of @xmath23 , see above ) . this indicates a first order phase transition for @xmath247 gev@xmath12 and @xmath248 gev@xmath12 . in what follows , we study the full phase portrait of the two - flavor njl model for finite @xmath16 and nonzero @xmath6 . @xmath2 phase diagram of a hot and magnetized two - flavor njl model is presented for @xmath196 mev , @xmath167 and @xmath168 . the black ( for @xmath169 ) and dark blue ( for @xmath170 ) solid lines denote the smooth crossovers , and the gray , green and light blue dashed lines the first order phase transitions for @xmath256 and @xmath168 , respectively . the starting points of the first order transition lines are denoted by @xmath203 ( for @xmath169 ) and @xmath204 ( for @xmath170 ) . the phenomenon of imc occurs for @xmath169 in @xmath257 $ ] gev@xmath12 and @xmath168 in the interval @xmath258 $ ] gev@xmath12 . for @xmath168 , the dhva oscillations in the regime @xmath259 $ ] gev@xmath12 lead to the phenomenon of reentrance from the @xmath17sb into the p@xmath17sr phase.,width=321,height=226 ] in fig . [ fig8 ] , the @xmath0@xmath2 phase diagram of a hot and magnetized two - flavor njl model is presented for @xmath196 mev , constant @xmath167 and @xmath16dependent @xmath168 . the black and dark blue solid lines denote smooth crossover for @xmath169 and @xmath170 , respectively , and the gray , green and light blue dashed lines the first order phase transitions for @xmath167 and @xmath168 , respectively . the starting points of the first order phase transitions are denoted by @xmath203 ( for @xmath169 ) and @xmath204 ( for @xmath170 ) . as aforementioned , for nonzero @xmath38 , the @xmath17sb phase is characterized by @xmath191 , and the p@xmath17sr phase by @xmath188 mev and @xmath192 . hence , by definition , the regimes below the critical lines denote the @xmath17sb phases and the regimes above them , the p@xmath17sr phases . to compare the effect of different choices for @xmath6 , let us first consider the transition curve for @xmath169 . starting from @xmath260 gev@xmath12 , the model exhibits a smooth crossover in the interval @xmath261 $ ] gev@xmath12 ( see the black solid line ) . as it turns out , the crossover temperatures is defined by @xmath262 . ] decreases with increasing @xmath2 from @xmath263 mev for @xmath260 mev to @xmath264 mev for @xmath265 gev@xmath12 . the end point of the crossover transition line is denoted in fig . [ fig8 ] by @xmath203 . for larger values of @xmath266 $ ] gev@xmath12 , a first order phase transition occurs ( gray dashed line ) . the critical temperatures for this first order phase transition decreases with increasing @xmath2 . this result confirms our findings from figs . [ fig1]-[fig7 ] . let us notice , e.g. that for @xmath197 mev , the point @xmath267 gev@xmath12 on the @xmath2 axis , is the same @xmath268 appearing in fig . [ fig3 ] for @xmath169 at @xmath173 mev . the fact that , for @xmath169 , @xmath23 decreases with increasing @xmath2 is related to the phenomenon of imc . this is in contrast to what happens for @xmath170 . here , as it turns out , the critical temperature essentially increases with increasing @xmath2 in the interval @xmath232 $ ] gev@xmath12 . first , in the regime @xmath269 $ ] gev@xmath12 , the system exhibits a smooth crossover from the @xmath17sb into the p@xmath17sr phase ( blue solid line ) . at a certain critical temperature @xmath270 mev and magnetic field @xmath271 gev@xmath12 , denoted by @xmath204 , the crossover transition line turns into a first order transition line in the regime @xmath272 $ ] gev@xmath12 ( green dashed line ) . let us notice at this stage , that , according to our arguments in @xcite , the regime @xmath273 gev@xmath12 is the regime of lll dominance . in this regime , in the most cases which we have considered in this paper , the constituent quark mass @xmath31 as well as the critical temperature @xmath23 monotonically increase with increasing @xmath2 . as concerns the @xmath0@xmath2 phase diagram for @xmath168 ( light blue dashed line ) . as it turns out , @xmath23 decreases with increasing @xmath2 in the regime @xmath274 $ ] gev@xmath12 , exhibits some oscillations in the regime @xmath275 $ ] gev@xmath12 and increases monotonically in the regime @xmath276 $ ] gev@xmath12 . hence , an imc is followed first by some oscillations and then the mc . moreover , for @xmath168 , the transition from the @xmath17sb into the p@xmath17sb phase is of first order in the whole regime @xmath277 $ ] gev@xmath12 ( light blue dashed line ) . for @xmath168 , because of this special shape of the first order transition line , another interesting effect occurs . to describe this effect , let us assume the temperature to be @xmath278 mev . in this case , starting from @xmath260 gev@xmath12 , the system is first in the @xmath17sb phase , then for @xmath279 gev@xmath12 a first order phase transition into the @xmath280sr phase occurs . with increasing the strength of the magnetic field up to @xmath281 gev@xmath12 , the system remains in this phase , and then reenters the @xmath17sb phase for @xmath282 gev@xmath12 phase . the same phenomenon occurs for all @xmath283 mev . let us notice , that the phenomenon of reentrance has been previously observed for nonvanishing @xmath1 and vanishing @xmath6 for the same hot and magnetized two - flavor njl model , as considered in the present paper @xcite ( see also figs . [ fig9](a ) and [ fig9](b ) for @xmath284 mev and @xmath285 mev ) . the results from fig . [ fig8 ] show , that for @xmath168 , the same phenomenon occurs also for @xmath196 mev . we believe that this phenomenon , as well as the observed oscillations in the @xmath2 dependence of @xmath31 appearing in figs . [ fig1 ] , [ fig2 ] and [ fig5 ] , are essentially related to the dhva @xcite in the weak magnetic field regime @xmath286 $ ] gev@xmath12 . these effect had been also studied in @xcite and most recently in @xcite . it occurs whenever landau levels pass the quark fermi level . as it turns out , the dhva oscillations are weakened , once the system enters the lll dominant regime @xmath273 gev@xmath12 . to compare the results for vanishing @xmath1 from fig . [ fig8 ] with the results for nonvanishing @xmath1 , the @xmath0@xmath2 phase diagram of hot and magnetized njl model is demonstrated in fig . [ fig9 ] for different @xmath174 gev@xmath175 ( panel a ) , @xmath169 ( panel b ) and @xmath170 ( panel c ) and for @xmath287 mev , @xmath288 mev and @xmath289 mev . solid lines denote the crossover transition lines , and dashed lines the first order phase transitions from @xmath17sb into p@xmath17sr phase . let us first consider fig . [ fig9](a ) . we observe , that for @xmath174 gev@xmath175 the crossover transition temperature decreases with increasing @xmath1 . the aforementioned dhva oscillations are stronger for larger @xmath1 and disappear in the lll dominant regime @xmath290 gev@xmath12 . in this regime , @xmath23 increases with increasing @xmath2 . this can be regarded as a signature of mc , especially in the lll dominant regime . for @xmath169 , however , @xmath23 decreases with increasing @xmath2 . moreover , whereas for @xmath287 mev and @xmath288 mev the crossover transitions in the weak magnetic field regime turn into first order transition lines , for @xmath289 mev only a crossover transition occurs in the regime @xmath291 $ ] gev@xmath12 . the critical points corresponding to @xmath292 are denoted by @xmath293 . they are given by @xmath294 @xmath295 . as in the case of @xmath174 gev@xmath175 , @xmath23 decreases with increasing @xmath1 for each fixed @xmath2 ( see also the @xmath0@xmath1 phase diagram from fig . [ fig11 ] ) . let us now compare the results for @xmath170 from fig . [ fig9](c ) with those for @xmath174 gev@xmath175 from fig . [ fig9](a ) . we observe that although the @xmath2 dependence of @xmath23 for @xmath170 , is in general similar to the case with @xmath174 gev@xmath175 , but for @xmath170 , the crossover transitions , appearing for @xmath174 gev@xmath175 in the whole range of @xmath296 $ ] gev@xmath12 and for all @xmath297 , turn into a first order phase transitions in the lll dominant regime @xmath290 gev@xmath12 . the critical points corresponding to @xmath297 are given by @xmath298 @xmath299 and @xmath300 . similar to the case of @xmath174 gev@xmath175 , the aforementioned phenomenon of reentrance from @xmath17sb into p@xmath17sr phase occurs also in the case @xmath170 , in particular , for @xmath284 mev . moreover , no dhva oscillations occur in the lll dominant regime . in fig . [ fig10 ] , the @xmath1@xmath2 phase diagram of a hot and magnetized two - flavor njl model is presented for @xmath199 mev and @xmath212 mev , as well as for @xmath174 gev@xmath175 ( panel a ) , @xmath169 ( panel b ) and @xmath170 ( panel c ) . the black and blue solid lines denote the smooth crossovers for @xmath301 and @xmath302 , respectively , and the gray dashed lines the first order phase transitions from the @xmath17sb phase ( the region below the critical lines ) into the p@xmath17sr phase ( the region above the critical lines ) . the starting points of the first order phase transitions are denoted by @xmath293 for @xmath303 . as it turns out , for @xmath174 gev@xmath175 , crossover transitions occur in the whole regime of @xmath304 $ ] gev@xmath12 . in contrast , for @xmath169 and @xmath170 , crossover transitions occur up to certain critical points @xmath293 and then with increasing @xmath2 , they turn into first order phase transitions . these critical points for @xmath169 and @xmath199 mev as well as @xmath212 mev are , @xmath305 and @xmath306 [ see fig . [ fig10](b ) ] . for @xmath170 and @xmath199 mev as well as @xmath212 mev they are given by @xmath307 and @xmath308 , respectively [ see fig . [ fig10](c ) ] . here , similar to the @xmath0@xmath2 phase diagrams , for @xmath169 the critical chemical potential @xmath24 decreases with increasing @xmath2 [ see fig . [ fig10](b ) ] . we conclude therefore that once the quark amm is sizable enough an inverse magnetic catalysis occurs . keeping @xmath2 fixed , @xmath24 decreases with increasing @xmath0 for all values of @xmath193 ( see also the @xmath0@xmath1 phase diagram in fig . [ fig11 ] ) . according to the results from figs . [ fig10](a ) and [ fig10](c ) , for @xmath174 gev@xmath175 and @xmath170 , the aforementioned dhva oscillations lead to the phenomenon of reentrance in the regime @xmath309 $ ] mev and @xmath310 $ ] mev for @xmath199 mev and @xmath174 gev@xmath175 and @xmath170 , respectively . for @xmath212 mev , these regimes are given by @xmath311 $ ] mev for @xmath174 gev@xmath175 and @xmath312 $ ] mev for @xmath170 . and @xmath313 , defined in ( [ a7 ] ) and ( [ a8 ] ) , are plotted as functions of @xmath2 , at @xmath314 mev and for @xmath196 mev as well as @xmath169 ( red squares ) , @xmath170 ( empty diamonds ) and @xmath168 ( gray circles ) . different choices of @xmath6s have no significant effect on the @xmath2 dependence of @xmath315 and @xmath313.,title="fig:",width=336,height=226 ] and @xmath313 , defined in ( [ a7 ] ) and ( [ a8 ] ) , are plotted as functions of @xmath2 , at @xmath314 mev and for @xmath196 mev as well as @xmath169 ( red squares ) , @xmath170 ( empty diamonds ) and @xmath168 ( gray circles ) . different choices of @xmath6s have no significant effect on the @xmath2 dependence of @xmath315 and @xmath313.,title="fig:",width=336,height=226 ] in fig . [ fig11 ] , the @xmath0@xmath1 phase diagram of a hot and magnetized two - flavor njl model is presented for @xmath316 gev@xmath12 and @xmath174 gev@xmath175 ( panel a ) , @xmath169 ( panel b ) and @xmath170 ( panel c ) . the crossover transition lines for @xmath316 gev@xmath12 are denoted by black and blue solid lines , and the blue dashed line denotes the first order phase transition for @xmath248 gev@xmath12 . as expected from the results in figs . [ fig9 ] and [ fig10 ] , for a fixed @xmath2 , the critical temperature of the chiral phase transition decreases with increasing @xmath1 , and different values of @xmath177 do not essentially change this specific feature . however , for @xmath169 , in contrast to the cases of @xmath174 gev@xmath175 and @xmath170 , for fixed @xmath1 , the critical temperature decreases with increasing @xmath2 ( imc ) . comparing the results from fig . [ fig11](a ) and [ fig11](c ) with [ fig11](b ) , it turns out that the crossover transitions for @xmath248 gev@xmath12 , @xmath174 gev@xmath175 and @xmath170 turn into a first order phase transition for @xmath169 . this result confirms our findings in figs . [ fig8 ] , [ fig9](b ) and [ fig10](b ) , and can be regarded as one of the main indications of the effect of large @xmath6 on qcd phase diagram . nonvanishing magnetic fields break the lorentz invariance , and induce certain anisotropies in the pressure of a hot and magnetized quark matter with respect to the direction of the background field . the @xmath2 dependence of pressure of a hot and magnetized quark matter for vanishing and nonvanishing @xmath6 has been already demonstrated in @xcite , where the effect of the quark amm in the strong magnetic field regime @xmath317 is compared with the weak - field amm by schwinger @xcite . the anisotropy in the pressure of hot and magnetized qcd is also investigated recently in @xcite in the framework of lattice gauge theory . let us denote the longitudinal and transverse pressures with respect to the direction of the magnetic field by @xmath318 and @xmath319 . according to @xcite , they are defined by @xmath320 where @xmath321 is the value of @xmath141 from ( [ f11 ] ) , evaluated at the minimum of the effective potential and @xmath322 is the classical magnetic energy . moreover , @xmath3 and @xmath323 are the external and induced magnetic fields , respectively . here , @xmath55 is the magnetization of the quark matter , defined by @xcite @xmath324 in what follows , we are , in particular , interested in the effect of different sets of @xmath6 on the @xmath2 dependence of @xmath325 and on the product @xmath72 . in fig . [ fig12 ] , the longitudinal and transverse pressures , @xmath318 ( black solid curves ) and @xmath319 ( red dashed curves ) are plotted as functions of @xmath2 for @xmath314 mev , @xmath196 mev and @xmath169 ( panel a ) , @xmath170 ( panel b ) as well as @xmath168 ( panel c ) . , the anisotropic pressures from ( [ a5 ] ) are slightly modified by @xmath326 and @xmath327 with @xmath328 and @xmath329 . here , @xmath330 is replaced by @xmath331 , with @xmath332 the electromagnetic fine structure constant . ] similar to the results presented in @xcite , @xmath318 ( @xmath319 ) decreases ( increases ) with increasing @xmath2 . as it turns out , different choices of @xmath6 do not significantly affect the final results for the @xmath2 dependence of anisotropic pressures . let us notice at this stage , that there is indeed an ambiguity in determining the longitudinal and transverse pressures in the literature . in @xcite , for instance , the maxwell term @xmath322 is not considered neither in the effective potential , nor in the definitions of @xmath318 and @xmath319 . in @xcite , however , whereas the maxwell term is not included in the effective potential @xmath141 , it is included in the definitions ( [ a5 ] ) of anisotropic pressures . to partly overcome these ambiguity , and at the same time to define a measure for the splitting of @xmath318 and @xmath319 for nonvanishing @xmath2 , we define the splitting coefficient @xmath315 for a fixed @xmath333 as @xcite @xmath334 where @xmath335 is the pressure at fixed @xmath333 and for vanishing magnetic field . in fig . [ fig13](a ) , @xmath315 is plotted as a function of @xmath2 at fixed temperature @xmath314 mev and for vanishing chemical potential . the results for @xmath167 and @xmath168 are denoted by red squares , empty diamonds and gray circles , respectively . as it turns out , @xmath315 increases with increasing magnetic field @xmath2 , as expected @xcite . this specific feature is , in particular , not affected by different choices for @xmath6 . small deviations , up to maximum @xmath336 occur only in the strong field regime , and increase with increasing @xmath2 . in comparison with @xmath174 gev@xmath175 , the same maximum deviation occurs is in the strong - field regime , and as it turns out nonvanishing @xmath6 has a negative effect on @xmath315 , especially in the strong - field regime . the same observation is also made in @xcite . another useful quantity that quantifies the relation between @xmath318 and @xmath319 is @xmath313 , @xcite @xmath337 where @xmath333 are fixed temperature and chemical potential . as it turns out from fig . [ fig13](b ) , @xmath313 has a minimum for a certain @xmath338 gev@xmath12 , and then increases and has a maximum for another @xmath339 gev@xmath12 . it then decreases to values @xmath340 . almost no differences occurs between different @xmath6s . let us notice , that @xmath341 and @xmath342 are related to the specific magnetic fields , where @xmath319 and @xmath318 almost vanish . as it turns out , @xmath343 . this tiny difference is not visible in fig . [ fig12 ] . the pressure anisotropy has various effects on astrophysics of dense stellar objects @xcite and the experiments of heavy ion collisions @xcite . in @xcite , for instance , the magnetization of quark matter @xmath55 is determined as a function of temperature . in @xcite is shown that in the vicinity of the chiral transition point , the magnetization @xmath55 is positive , and therefore hot and dense qcd at transition point exhibits a paramagnetic response . it is further shown , that the paramagnetic behavior of qcd matter affects the phenomenology of heavy ion collision , and in particular , has `` significant impact on the value of elliptic flow @xmath15 '' . in fig . [ fig14](a ) , the product of the magnetization @xmath55 and the magnetic field strength @xmath3 is plotted for our hot and dense njl model at @xmath173 and for @xmath174 gev@xmath175 ( black circles ) , @xmath169 ( red squares ) , @xmath170 ( gray diamonds ) and @xmath168 ( empty triangles ) . as it turns out , the results for @xmath169 have significant difference with the results corresponding to @xmath174 gev@xmath175 , @xmath170 and @xmath168 . as expected , at @xmath197 mev , there is no difference between the @xmath2 dependence of @xmath72 for @xmath174 gev@xmath175 and @xmath168 . the difference between the data for @xmath174 gev@xmath175 and @xmath170 becomes only significant in the lll dominant regime @xmath344 gev@xmath12 . in this regime , as expected , the dhva oscillations arising in the regime @xmath21 gev@xmath12 weaken , and @xmath72 monotonically increases with increasing @xmath2 . these oscillations are also previously observed in @xcite . the discontinuity arising in the data for @xmath169 at @xmath345 gev@xmath12 ( red squares ) , is related to the first order phase transition from the @xmath17sb into the p@xmath17sr phase , demonstrated in fig . the latter leads also to a discontinuity in the @xmath2 dependence of the constituent quark mass exactly for @xmath345 gev@xmath12 [ see fig . [ fig1](a ) ] . in figs . [ fig14](b ) and ( c ) , the dimensionless quantity @xmath346 is plotted as a function of @xmath347 at @xmath197 mev and @xmath348 mev [ fig . [ fig14](b ) ] and @xmath284 mev [ fig . [ fig14](c ) ] as well as different @xmath6s . black circles , red squares and gray diamonds denote the results for @xmath174 gev@xmath175 , @xmath169 and @xmath170 , respectively . this dependence is also studied in @xcite in the regime @xmath349 $ ] for @xmath350 mev . similar to the results of @xmath196 mev , the difference between @xmath174 gev@xmath175 and @xmath170 increases with increasing @xmath2 . except in the regime of weak magnetic field @xmath351 gev@xmath12 ( @xmath352 for @xmath348 mev ) , and @xmath353 gev@xmath12 ( @xmath354 for @xmath284 mev ) , @xmath346 for @xmath169 is smaller than the data for @xmath174 gev@xmath175 and @xmath170 . according to the @xmath0@xmath2 phase diagram for @xmath284 mev in fig . [ fig9 ] , we expect discontinuities in the @xmath2 dependence of @xmath346 in the regime @xmath355 gev@xmath12 ( @xmath356 ) for @xmath174 gev@xmath175 , @xmath357 gev@xmath12 ( @xmath358 ) for @xmath169 and @xmath359 gev@xmath12 ( @xmath360 ) for @xmath170 . we conclude that discontinuities in the @xmath2 dependence of @xmath72 or @xmath346 as functions of @xmath2 are related , as expected , to first order chiral phase transitions at certain magnetic fields and for fixed values of @xmath0 and @xmath1 . in recent years , there were a number of attempts to explore the effect of the quark amm on the phase diagram of qcd at finite temperature , chemical potential and in the presence of uniform magnetic fields @xcite . following the method used in @xcite , we have studied , in the present paper , the effects of the ( effective ) quark amm on the thermodynamic properties of the constituent quark mass @xmath31 , and on the full phase portrait of a two - flavor magnetized njl model at finite @xmath0 and @xmath1 . the quark amm is introduced via an additional minimal coupling term , @xmath33 , in the lagrangian density of the njl model @xcite . the coefficient @xmath34 , defined by @xmath361 , includes the nonperturbative ( effective ) bohr magneton @xmath52 , and is herewith , as function of the constituent quark mass , and receives @xmath16 corrections . as it turns out , @xmath33 leads to an additional term proportional to @xmath41 in the quark energy dispersion relation ( [ f9 ] ) . in the above expression , @xmath45 stands for up ( @xmath46 ) and down ( @xmath47 ) quark flavors , and the dimensionful coupling @xmath64 is defined by @xmath123 , with @xmath49 related to the deviation of the land @xmath26-factor from @xmath27 . our aim was , in particular , to study the effects of constant as well as @xmath16 dependent effective coupling @xmath64 on the thermodynamic behavior of @xmath31 , as well as on the phase portrait of the hot and magnetized quark matter , described by our model . to this purpose , three different sets for the effective coupling @xmath64 are chosen by making use of the method presented in sec . [ sec2 ] and app . [ appa ] [ see ( [ f12])-([f14 ] ) ] . the dependence of the constituent quark mass @xmath31 on @xmath7 and @xmath2 is then determined for each fixed @xmath64 in sec . [ subsec3a ] ( see figs . [ fig1]-[fig7 ] ) . here , we have explicitly described the signatures related to the phenomena of mc and imc . we have further shown that for large enough @xmath64 , even in the regime of strong magnetic fields , the @xmath2 dependence of @xmath31 is strongly affected by the phenomenon of imc . then , using the one - loop effective potential @xmath141 from ( [ f11 ] ) in term of @xmath64 , we have explored the complete phase portrait of the model in the parameter space @xmath362 , and @xmath64 ( see sec . [ subsec3b ] ) . we have shown , that for large enough @xmath64 and mainly in the regime of weak magnetic fields @xmath21 gev@xmath12 , the critical temperature @xmath23 and critical chemical potential @xmath24 decrease with increasing @xmath2 . this is related to the phenomenon of imc . moreover , it is shown that in certain regimes of the parameter space , the phenomenon of reentrance of chiral symmetry broken phase occurs , mainly as a consequence of dhva oscillations @xcite . also , the order of the phase transition turns out to be affected by @xmath64 ( see figs . [ fig8]-[fig11 ] ) . in sec . [ subsec3c ] , the pressure anisotropy of the quark matter in the longitudinal and transverse directions with respect to the magnetic field is considered . we have shown that different choices of @xmath64 have no specific effect on the pressure anisotropies , demonstrated in fig . [ fig12 ] , and on @xmath315 and @xmath313 , as quantitative measures for these anisotropies ( see fig . [ fig13 ] ) . in fig . [ fig14 ] , the @xmath2 dependence of the product of the field strength and the magnetization is demonstrated . we have shown that for large enough @xmath64 , this product becomes smaller than for the cases of @xmath174 gev@xmath175 and small @xmath64 , but the magnetization is always positive . according to @xcite , this is believed to be an indication of paramagnetic behavior of the hot and magnetized quark matter , especially in the vicinity of the phase transition point . let us notice at this stage , that the method used in the present paper to introduce the quark amm is different from the method presented in @xcite . here , starting from a one - flavor magnetized njl model with an appropriate tensor channel , a mechanism for the dynamical generation of the quark amm in the lll is presented . it is also shown that the scalar and tensor couplings of the njl model become anisotropic , and receive longitudinal and transverse components with respect to the direction of the magnetic field . for constant anisotropic couplings , in the subcritical regime , the phenomenon of imc does not occur . in a subsequent paper @xcite , however , the authors consider the running of these couplings as a function of @xmath2 , and show that because of a certain antiscreening effect , induced by quarks that are confined by the magnetic fields to the lll , the critical temperature of the @xmath17sb decreases with increasing the magnetic field strength . this is believed to be a natural explanation for the phenomenon of imc , that arises originally in a number of model calculations @xcite , in the framework of gauge / gravity duality @xcite , and from an ab initio lattice qcd simulation at finite @xmath0 and @xmath2 @xcite . let us also notice that the linear - in-@xmath3 ansatz , used in the present paper , is different from the one which is used in @xcite . here , the schwinger term is defined to be proportional to the bohr magneton @xmath363 in term of the current ( bare ) quark mass @xmath38 , in contrast to our approach , described above . thus far , the inconsistencies from the schwinger linear - in-@xmath3 ansatz in the weak - magnetic field approximation , described in @xcite , are not expected to occur in our approach . moreover , in contrast to the previous approaches , we have considered the contributions of all landau levels , and not restricted ourselves to lll @xcite , nor to one - loop approximation in the lll , as in @xcite . there are many possibilities to improve the approach presented in this paper . as we have argued in sec . [ sec2 ] , the coefficient @xmath64 and the constituent quark mass @xmath31 are closely entangled . in other words , it is not possible to determine one of them without determining the other one . the reason is indeed formulated in @xcite , where it is stated that since one and the same symmetry is broken by the quark amm and the chiral condensate , nothing can guarantee a vanishing amm , once the chiral symmetry is broken by a nonvanishing chiral condensate . in the present paper , using a method compatible with the constituent quark model , we have fixed @xmath64 and determined @xmath31 . this method can gradually be improved . the main idea is to determine @xmath64 from the relation @xmath364 , where @xmath61 for up and down quarks are given in ( [ appa8 ] ) . instead of fixing @xmath31 with the phenomenologically reliable @xmath68 mev and @xmath69 mev , as it is performed in the present paper , we can replace it , e.g. by @xmath365 , where @xmath366 and @xmath367 are fixed values of chemical potential and magnetic field . we then obtain @xmath368 for up quarks , and @xmath369 for down quarks . plugging these relations into ( [ f9 ] ) , and the latter into @xmath370 , and eventually looking for the global minima of the resulting expression , a new set of constituent quark mass arises , which replaces the data demonstrated in figs . [ fig6 ] , for instance . the same procedure may be repeated for the sets @xmath371 or @xmath372 , where the fixed values of @xmath7 and @xmath2 are denoted by the superscript `` star '' . it would be interesting to look for the phenomenon of mc and imc in this framework . we will report about the results of this new approach in the a future publication . the authors acknowledge f. ardalan for discussions about the effect of the anomalous magnetic moment of quarks on their spectra , and m. mohammadi najafabadi for providing insight into the experimental data for the magnetic moment of protons and neutrons . let us consider a system including up and down quarks in the presence of a uniform magnetic field @xmath373 . the spin magnetic moment @xmath374 of this system is given by @xmath375 here , @xmath376 is the land @xmath26-factor , with @xmath377 denoting its anomalous contribution . in our two - flavor njl model @xmath164 and @xmath378 are @xmath90 diagonal matrices , including the anomalous magnetic moment @xmath145 and electric charge @xmath379 of the up and down quarks . moreover , @xmath380 is the quark spin angular momentum of the quarks , and @xmath381 with @xmath382 , the nonperturbative ( effective ) bohr magneton in the flavor space , which is given in term of the ( bare ) electric charge @xmath92 , as well as the up and down quark constituent ( effective ) masses @xmath383 and @xmath384 , arising in the mass matrix @xmath385 . here , @xmath386 are the three pauli matrices . using ( [ appa1 ] ) , and assuming that the magnetic field @xmath373 is directed in the third direction , the @xmath58-th matrix element of the third component of @xmath374 , @xmath387 , is given by @xmath388 here , @xmath389 is the third pauli matrix . the eigenvalues of @xmath57 in the spinor space are therefore given by @xmath390 with @xmath44 and @xmath45 . using this expression , we can define the following positive ratio @xmath391 which turns out to be a phenomenologically relevant quantity @xcite . here , @xmath392 gev the proton mass , and @xmath56 the nuclear magneton , whose phenomenological values are fixed by experiments . using at this stage , @xmath393 for the magnetic moment of proton ( neutron ) @xmath394 ( @xmath395 ) , and their relationship to the magnetic moment of up and down quarks , @xmath396 and @xmath397 , @xmath398 that yield @xmath399 we obtain @xmath400 using these data , the ratio @xmath401 for @xmath45 is fixed to be @xmath402 same results are also reported in @xcite . the above phenomenological values for @xmath65 can be used to determine phenomenological values for @xmath145 through @xmath403 from ( [ appa3 ] ) . in order to have a sizable quark amm , we choose @xmath404 gev for the quark ( effective ) constituent mass @xmath405 @xcite . we arrive at @xmath406 which satisfy the condition @xmath407 . the latter guarantees the isospin symmetry @xcite . plugging ( [ appa9 ] ) into the relation @xmath408 from ( [ f9 ] ) , and choosing @xmath404 gev , we obtain @xmath409 [ see also ( [ f12 ] ) ] . choosing , on the other hand , @xmath410 gev , and following the same steps as above , we obtain @xmath411 which lead to @xmath412 [ see also ( [ f13 ] ) ] . as described in sec . [ sec2 ] , in the present work , @xmath6 , appearing explicitly in the quark energy dispersion relation ( [ f9 ] ) is fixed by ( [ appa10 ] ) and ( [ appa12 ] ) [ see [ f12 ] ) and ( [ f13 ] ) ] . the @xmath16 dependent constituent quark mass are then determined by plugging this dispersion relation into the thermodynamic potential ( [ f11 ] ) , and minimizing it appropriately . we have shown that for @xmath147 from ( [ appa10 ] ) , leading to large @xmath377 from ( [ appa9 ] ) , the phenomenon of imc occurs . 99 d. e. kharzeev , k. landsteiner , a. schmitt and h. u. yee , _ strongly interacting matter in magnetic fields : an overview _ , lect . notes phys . * 871 * , 1 ( 2013 ) , arxiv:1211.6245 [ hep - ph ] . v. de la incera , _ nonperturbative physics in a magnetic field _ , aip conf . proc . * 1361 * , 74 ( 2011 ) , arxiv:1004.4931 [ hep - ph ] . r. c. duncan and c. thompson , _ formation of very strongly magnetized neutron stars - implications for gamma - ray bursts _ , astrophys . j. * 392 * , l9 ( 1992 ) . v. skokov , a. y. illarionov and v. toneev , _ estimate of the magnetic field strength in heavy - ion collisions _ , j. mod . a * 24 * , 5925 ( 2009 ) , arxiv:0907.1396 [ nucl - th ] . d. e. kharzeev , l. d. mclerran and h. j. warringa , _ the effects of topological charge change in heavy ion collisions : event by event p and cp violation _ , nucl . a * 803 * , 227 ( 2008 ) , arxiv:0711.0950 [ hep - ph ] . k. tuchin , _ particle production in strong electromagnetic fields in relativistic heavy - ion collisions _ , adv . high energy phys . * 2013 * , 490495 ( 2013 ) , arxiv:1301.0099 [ hep - ph ] . s. fayazbakhsh , s. sadeghian and n. sadooghi , _ properties of neutral mesons in a hot and magnetized quark matter _ , phys . d * 86 * , 085042 ( 2012 ) , arxiv:1206.6051 [ hep - ph ] . s. fayazbakhsh and n. sadooghi , _ weak decay constant of neutral pions in a hot and magnetized quark matter _ , phys . d * 88 * , 065030 ( 2013 ) , arxiv:1306.2098 [ hep - ph ] . j. o. andersen , _ thermal pions in a magnetic background _ d * 86 * , 025020 ( 2012 ) , arxiv:1202.2051 [ hep - ph ] . j. o. andersen , _ chiral perturbation theory in a magnetic background - finite - temperature effects _ , jhep * 1210 * , 005 ( 2012 ) , arxiv:1205.6978 [ hep - ph ] . m. strickland , v. dexheimer and d. p. menezes , _ bulk properties of a fermi gas in a magnetic field _ , phys . d * 86 * , 125032 ( 2012 ) , arxiv:1209.3276 [ nucl - th ] . g. s. bali , f. bruckmann , g. endrodi and a. schafer , _ paramagnetic squeezing of qcd matter _ , phys . lett . * 112 * , 042301 ( 2014 ) , arxiv:1311.2559 [ hep - lat ] . k. g. klimenko , _ three - dimensional gross - neveu model at nonzero temperature and in an external magnetic field _ , z. phys . c * 54 * , 323 ( 1992 ) . v. p. gusynin , v. a. miransky and i. a. shovkovy , _ dimensional reduction and catalysis of dynamical symmetry breaking by a magnetic field _ , nucl . b * 462 * , 249 ( 1996 ) , arxiv : hep - ph/9509320 . i. a. shovkovy , _ magnetic catalysis : a review _ , lect . notes phys . * 871 * , 13 ( 2013 ) , arxiv:1207.5081 [ hep - ph ] . f. preis , a. rebhan and a. schmitt , _ inverse magnetic catalysis in dense holographic matter _ , jhep * 1103 * , 033 ( 2011 ) , arxiv:1012.4785 [ hep - th ] . t. inagaki , d. kimura and t. murata , _ four fermion interaction model in a constant magnetic field at finite temperature and chemical potential _ phys . * 111 * , 371 ( 2004 ) , arxiv : hep - ph/0312005 . v. zdravkov , _ et al . _ , _ reentrant superconductivity in nb / cu@xmath413ni@xmath414 bilayers _ , phys . lett . * 97 * , 057004 ( 2006 ) . a. n. berker and j. s. walker , _ frustrated spin - gas model for doubly reentrant liquid crystals _ , phys . 47 * , 1469 ( 1981 ) . j. margueron and e. khan , _ suppression , persistence and reentrance of superfluidity in overflowing nuclear systems _ c * 86 * , 065801 ( 2012 ) , arxiv:1203.2134 [ nucl - th ] . s. fayazbakhsh and n. sadooghi , _ color neutral 2sc phase of cold and dense quark matter in the presence of constant magnetic fields _ , phys . d * 82 * , 045010 ( 2010 ) , arxiv:1005.5022 [ hep - ph ] . s. fayazbakhsh and n. sadooghi , _ phase diagram of hot magnetized two - flavor color superconducting quark matter _ , d * 83 * , 025026 ( 2011 ) , arxiv:1009.6125 [ hep - ph ] . w. j. de haas and p. m. van alphen , _ the dependence of the susceptibility of diamagnetic metals upon the field _ , proc . sci , amsterdam * 33 * , 1106 ( 1930 ) . l. d. landau and e. m. lifshitz , _ statistical physics _ , pergamon press , new york , 1980 . d. ebert and a. s. vshivtsev , _ van alphen - de haas effect for dense cold quark matter in a homogeneous magnetic field _ , arxiv : hep - ph/9806421 . t. inagaki , d. kimura and t. murata , _ njl model at finite chemical potential in a constant magnetic field _ suppl . * 153 * , 321 ( 2004 ) , arxiv : hep - ph/0404219 . j. l. noronha and i. a. shovkovy , _ color - flavor locked superconductor in a magnetic field _ , phys . d * 76 * , 105030 ( 2007 ) , erratum - ibid . d * 86 * , 049901 ( 2012 ) , arxiv:0708.0307 [ hep - ph ] . v. d. orlovsky and y. . a. simonov , _ magnetic susceptibility at zero and nonzero chemical potential in qcd and qed _ , arxiv:1406.1056 [ hep - ph ] . p. v. buividovich , m. n. chernodub , e. v. luschevskaya and m. i. polikarpov , _ numerical study of chiral symmetry breaking in non - abelian gauge theory with background magnetic field _ , phys . b * 682 * , 484 ( 2010 ) , arxiv:0812.1740 [ hep - lat ] . p. v. buividovich , m. i. polikarpov and o. v. teryaev , _ lattice studies of magnetic phenomena in heavy - ion collisions _ notes phys . * 871 * , 377 ( 2013 ) [ arxiv:1211.3014 [ hep - ph ] ] . g. s. bali , f. bruckmann , g. endrodi , z. fodor , s. d. katz , s. krieg , a. schafer and k. k. szabo , _ the qcd phase diagram for external magnetic fields _ , jhep * 1202 * , 044 ( 2012 ) , arxiv:1111.4956 [ hep - lat ] . g. s. bali , f. bruckmann , g. endrodi , z. fodor , s. d. katz and a. schafer , _ qcd quark condensate in external magnetic fields _ , phys . d * 86 * , 071502 ( 2012 ) , arxiv:1206.4205 [ hep - lat ] . ilgenfritz , m. muller - preussker , b. petersson and a. schreiber , _ magnetic catalysis ( and inverse catalysis ) at finite temperature in two - color lattice qcd _ d * 89 * , 054512 ( 2014 ) , arxiv:1310.7876 [ hep - lat ] . v. g. bornyakov , p. v. buividovich , n. cundy , o. a. kochetkov and a. schfer , _ deconfinement transition in two - flavour lattice qcd with dynamical overlap fermions in an external magnetic field _ , arxiv:1312.5628 [ hep - lat ] . c. bonati , m. delia , m. mariti , f. negro and f. sanfilippo , _ magnetic susceptibility and equation of state of @xmath415 qcd with physical quark masses _ , phys . d * 89 * , 054506 ( 2014 ) , arxiv:1310.8656 [ hep - lat ] . f. bruckmann , g. endrodi and t. g. kovacs , _ inverse magnetic catalysis and the polyakov loop _ , jhep * 1304 * , 112 ( 2013 ) , arxiv:1303.3972 [ hep - lat ] . g. s. bali , f. bruckmann , g. endrodi , s. d. katz and a. schafer , _ the qcd equation of state in background magnetic fields _ , arxiv:1406.0269 [ hep - lat ] . k. fukushima and j. m. pawlowski , _ magnetic catalysis in hot and dense quark matter and quantum fluctuations _ , phys . d * 86 * , 076013 ( 2012 ) , arxiv:1203.4330 [ hep - ph ] . j. o. andersen , w. r. naylor and a. tranberg , _ chiral and deconfinement transitions in a magnetic background using the functional renormalization group with the polyakov loop _ , jhep * 1404 * , 187 ( 2014 ) , arxiv:1311.2093 [ hep - ph ] . k. kamikado and t. kanazawa , _ chiral dynamics in a magnetic field from the functional renormalization group _ , jhep * 1403 * , 009 ( 2014 ) , arxiv:1312.3124 [ hep - ph ] . k. fukushima and y. hidaka , _ magnetic catalysis vs magnetic inhibition _ , phys . lett . * 110 * , 031601 ( 2013 ) , arxiv:1209.1319 [ hep - ph ] . r. l. s. farias , k. p. gomes , g. i. krein and m. b. pinto , _ the importance of asymptotic freedom for the pseudocritical temperature in magnetized quark matter _ , arxiv:1404.3931 [ hep - ph ] . a. j. mizher , m. n. chernodub and e. s. fraga , _ phase diagram of hot qcd in an external magnetic field : possible splitting of deconfinement and chiral transitions _ , d * 82 * , 105016 ( 2010 ) , arxiv:1004.2712 [ hep - ph ] . r. gatto and m. ruggieri , _ deconfinement and chiral symmetry restoration in a strong magnetic background _ , d * 83 * , 034016 ( 2011 ) , arxiv:1012.1291 [ hep - ph ] . m. ferreira , p. costa , d. p. menezes , c. providncia and n. scoccola , , phys . d * 89 * , 016002 ( 2014 ) , arxiv:1305.4751 [ hep - ph ] . m. ferreira , p. costa , o. loureno , t. frederico and c. providncia , _ inverse magnetic catalysis in the ( 2 + 1)-flavor nambu jona - lasinio and polyakov nambu jona - lasinio models _ d * 89 * , 116011 ( 2014 ) , arxiv:1404.5577 [ hep - ph ] . e. s. fraga , b. w. mintz and j. schaffner - bielich , _ a search for inverse magnetic catalysis in thermal quark - meson models _ , phys . b * 731 * , 154 ( 2014 ) , arxiv:1311.3964 [ hep - ph ] . j. o. andersen , w. r. naylor and a. tranberg , _ chiral and deconfinement transitions in a magnetic background using the functional renormalization group with the polyakov loop _ , jhep * 1404 * , 187 ( 2014 ) , arxiv:1311.2093 [ hep - ph ] . r. gatto and m. ruggieri , _ hot quark matter with an axial chemical potential _ , d * 85 * , 054013 ( 2012 ) , arxiv:1110.4904 [ hep - ph ] . j. chao , p. chu and m. huang , _ inverse magnetic catalysis induced by sphalerons _ d * 88 * , 054009 ( 2013 ) , arxiv:1305.1100 [ hep - ph ] . m. ruggieri , l. oliva , p. castorina , r. gatto and v. greco , _ critical endpoint and inverse magnetic catalysis for finite temperature and density quark matter in a magnetic background _ , arxiv:1402.0737 [ hep - ph ] . a. ayala , m. loewe , a. j. mizher and r. zamora , _ inverse magnetic catalysis for the chiral transition induced by thermo - magnetic effects on the coupling constant _ , arxiv:1406.3885 [ hep - ph ] . a. ayala , m. loewe and r. zamora , _ anticatalysis in the linear sigma model with quarks _ , arxiv:1406.7408 [ hep - ph ] . e. j. ferrer , v. de la incera and x. j. wen , _ quark antiscreening at strong magnetic field and inverse magnetic catalysis _ , arxiv:1407.3503 [ nucl - th ] . j. p. blaizot , e. s. fraga and l. f. palhares , _ effect of quark masses on the qcd presssure in a strong magnetic background _ , phys . b * 722 * , 167 ( 2013 ) , arxiv:1211.6412 [ hep - ph ] . e. j. ferrer , v. de la incera , i. portillo and m. quiroz , _ new look at the qcd ground state in a magnetic field _ , arxiv:1311.3400 [ nucl - th ] . e. j. ferrer , de la incera , d. m. paret and a. p. martnez , _ anomalous - magnetic - moment effects in a strongly magnetized and dense medium _ , arxiv:1307.5947 [ nucl - th ] . d. m. paret , a. p. martinez , e. j. ferrer and v. de la incera , _ effects of amm on the eos of magnetized dense systems _ , arxiv:1401.5006 [ nucl - th ] . e. j. ferrer and v. de la incera , _ dynamically generated anomalous magnetic moment in massless qed _ b * 824 * , 217 ( 2010 ) , arxiv:0905.1733 [ hep - ph ] . e. j. ferrer and v. de la incera , _ dynamically induced zeeman effect in massless qed _ , phys . * 102 * , 050402 ( 2009 ) , arxiv:0807.4744 [ hep - ph ] . j. s. schwinger , _ on quantum electrodynamics and the magnetic moment of the electron _ , phys . rev . * 73 * , 416 ( 1948 ) . b. l. ioffe and a. v. smilga , _ nucleon magnetic moments and magnetic properties of vacuum in qcd _ , nucl . b * 232 * , 109 ( 1984 ) . j. a. bicudo , j. e. f. t. ribeiro and r. fernandes , _ the anomalous magnetic moment of quarks _ , phys . c * 59 * , 1107 ( 1999 ) , arxiv : hep - ph/9806243 . g. peressutti and b. s. skagerstam , _ finite temperature effects in quantum field theory _ , phys . b * 110 * , 406 ( 1982 ) . j. shen , h. -h . xu and c. -h . tsai , _ quantum electrodynamics at finite temperature : one loop corrections to electron mass and magnetic moment _ , commun . phys . * 15 * , 209 ( 1991 ) . s. s. masood and m. q. haseeb , _ second order corrections to the magnetic moment of electron at finite temperature _ , int . j. mod . a * 27 * , 1250188 ( 2012 ) , arxiv:1203.3628 [ hep - th ] . v. i. ritus , _ radiative corrections in quantum electrodynamics with intense fields and their analytical properties _ , ann . phys . * 69 * , ( 1972 ) 555 . g. murguia , a. raya , a. sanchez , e. reyes , _ the electron propagator in external electromagnetic fields in lower dimensions _ , am . j. phys . * 78 * , 700 - 707 ( 2010 ) , arxiv:0910.1881 [ hep - th ] . n. sadooghi and f. taghinavaz , _ local electric current correlation function in an exponentially decaying magnetic field _ , phys . d * 85 * , 125035 ( 2012 ) , arxiv:1203.5634 [ hep - ph ] . d. m. paret , j. e. horvath and a. p. martinez , _ anisotropic stellar structure equations for magnetized stars _ , arxiv:1407.2280 [ astro-ph.he ] . m. chaichian , s. s. masood , c. montonen , a. perez martinez and h. perez rojas , _ quantum magnetic and gravitational collapse _ lett . * 84 * , 5261 ( 2000 ) , arxiv : hep - ph/9911218 .	the effect of anomalous magnetic moment of quarks on thermodynamic properties of the chiral condensate is studied , using of a two - flavor nambu jona - lasinio model at finite temperature @xmath0 , chemical potential @xmath1 , and in the presence of a uniform magnetic field @xmath2 . to this purpose , the schwinger linear - in-@xmath3 ansatz for the quark anomalous magnetic moment in term of the nonperturbative bohr magneton is considered . in a two - dimensional flavor space , it leads to the correction @xmath4 in the energy dispersion relation of quarks . here , @xmath5 is the quark charge matrix . we consider three different sets for @xmath6 , and numerically determine the dependence of the constituent quark mass on @xmath7 and @xmath2 for fixed @xmath6 . by exploring the complete phase portrait of this model in @xmath0-@xmath1 , @xmath1-@xmath2 , and @xmath0-@xmath2 phase spaces for various fixed @xmath2 , @xmath0 , @xmath1 and @xmath6 , we observe that inverse magnetic catalysis occurs for large enough @xmath6 . moreover , in the regime of weak magnetic fields , the phenomenon of reentrance of chiral symmetry broken and restored phases occurs for @xmath8 and @xmath2 dependent @xmath6 .
galaxies , galaxy groups and clusters are widely held to form under the drive of the gravitational instability that acts on initial perturbations modulating the cosmic density of the dominant cold dark matter ( dm ) component . the instability at first is kept in check by the cosmic expansion , but when the local gravity prevails collapse sets in . the standard sequence runs as follows : a slightly overdense region expands more slowly than its surroundings , progressively detaches from the hubble expansion , halts and turns around ; then it collapses , and eventually virializes to form a dm ` halo ' in equilibrium under self - gravity . the amplitude of more massive perturbations is smaller , so the formation is hierarchical with the massive structures forming typically later ( see peebles 1993 ) . such a formation history has been confirmed and resolved to a considerable detail by many intensive @xmath3body simulations . early on ( white 1986 ) these added an important block of information , i.e. , the growth of a halo actually includes merging events with other clumps of sizes comparable ( ` major ' mergers ) or substantially smaller ( ` minor ' mergers ) , down to nearly smooth accretion ( see springel et al . 2006 ) . more recently , a second round of features has been increasingly recognized in highly resolved simulations of individual halos forming in cosmological volumes ( zhao et al . 2003 ; diemand et al . 2007 ) , to the effect of identifying in the growth two definite stages : an early fast collapse including a few violent major mergers , and a later calmer stage comprising many minor mergers and smooth accretion . during the early collapse a substantial mass is gathered , while the major mergers reshuffle the gravitational potential wells and cause the collisionless dm particles to undergo dynamical relaxation ; therefrom the halo emerges with a definite structure of the inner density and gravitational potential . during the later stage , moderate amounts of mass are slowly accreted mainly onto the outskirts , little affecting the inner structure and potential , but quiescently rescaling upwards the overall size ( see hoffman et al . 2007 ) . in the resulting simulated halos , the coarse - grained proxy @xmath13 to the phase - space density defined in terms of the dm density @xmath2 and of the radial velocity dispersion @xmath1 is empirically found to be stable after the collapse , and distributed in the form of a powerlaw with _ uniform _ slope @xmath14 around @xmath15 throughout the halo s bulk where it is amenable to noise - free measurements ( see taylor & navarro 2001 ; dehnen & mclaughlin 2005 ; ascasibar & gottlber 2008 ) . taking up such a notion , the present paper starts ( 2 ) from investigating the wide set of macroscopic equilibria for the dm under self - gravity , as described by the jeans equation . it proceeds ( 3 ) to the semianalytic description of their time buildup to constrain their governing parameters , and so predict the halo realistic structure . it proposes ( 4 ) physical interpretations for such parameters . the paper is concluded ( 5 ) with a summary of our findings and a discussion of the perspectives they open . throughout we adopt the standard , flat cosmological model with normalized matter density @xmath16 , dark energy density @xmath17 , and hubble constant @xmath18 km s@xmath19 mpc@xmath19 . the static equilibria of the dm halos obey the jeans equation that , on neglecting for the time being anisotropies and the blend into the large - scale environment to be discussed in 5 , reads @xmath20 to elicit the contents of this macroscopic - like equation , an explicit link is required between @xmath21 and @xmath22 . such a link may be provided by @xmath23 , or by the equivalent quantity @xmath24 ; this is often named dm ` entropy ' in view of its formal analogy with the adiabat of a gas in thermal equilibrium , and is increasingly used in the context of dm equilibria , from taylor & navarro ( 2001 ) to faltenbacher et al . we focus on this function of state @xmath25 in the configuration space , and comment on our choice in 5 . in terms of the slope @xmath26 , eq . ( 1 ) is recast into the particularly simple form @xmath27 here @xmath28 is the ( squared ) circular velocity , that provides a natural scaling for the depth of the dm potential well . taking up from the simulation phenomenology only the notion that @xmath5 be _ uniform _ throughout the halo bulk ( see 1 ) , we will discuss the range of values to be expected for @xmath5 as a clue to their origin . in fact , in their pioneering work taylor & navarro ( 2001 ) selected a value of the exponent close to @xmath29 . other , recent studies have provided similar values for @xmath5 : around @xmath30 according to ascasibar et al . ( 2004 ) , and around @xmath31 according to rasia et al . ( 2004 ) . beyond the minor differences , it is the _ narrow _ empirical range that constituted an unsolved challenge . we begin with looking first ( after taylor & navarro 2001 ) at powerlaw solutions of eq . ( 2 ) of the form @xmath32 , with the entropy run set to @xmath33 ; then not only @xmath5 is a given parameter , but also the slope @xmath34 is a _ constant _ to be determined from eq . ( 2 ) in the form @xmath35 here the ` bare ' constant @xmath36 compares the gravitational potential to the random kinetic energy associated with the velocity dispersion at a reference point @xmath37 . at the r.h.s @xmath38 appears as modulated by @xmath39dependencies , produced by the integrated dynamical mass @xmath40 to yield @xmath41 in the powerlaw case , and by the local dependencies from @xmath42 . for a powerlaw solution to actually hold the r.h.s . must be constant , which yields the link @xmath43 as pointed out by taylor & navarro ( 2001 ) , hansen ( 2004 ) , and austin et al . ( 2005 ) . given this , the balance of the three terms in eq . ( 3 ) imposes on the dimensionless gravitational pull the constraint @xmath44 whose nonlinearities reflect the dm self - gravity . thus the allowed powerlaw solutions must lie on a parabola in the plane @xmath45 and on a straight line in the plane @xmath46 , as illustrated in the top panels of fig . 1 ; note that to ensure @xmath47 the overall bounds @xmath48 must hold , implying that @xmath49 applies for the powerlaw solutions . two notable instances correspond to the values @xmath50 and @xmath51 . the former yields the standard , singular isothermal sphere with @xmath52 and @xmath53 ; the latter holds for the value of @xmath5 originally selected by taylor & navarro ( 2001 ) , and implies @xmath54 and @xmath55 . note that they are located on the intersections of the parabola eq . ( 5 ) with the line @xmath56 , see fig . 1 . the powerlaw solutions of the jeans equation with slope @xmath57 will turn out to be tangent to the full solutions at @xmath58 so as to provide a useful reference slope in the middle of the halo . however , they do not qualify as relevant solutions since at both extremes they suffer of unphysical features : since @xmath49 holds as shown above , the force @xmath59 would diverge at the center , while the mass @xmath60 would diverge toward large @xmath61 . to avoid these drawbacks , in all physically relevant solutions the running slope @xmath62 is to bend down from the middle value in going toward large and small @xmath61 . in fact , for @xmath63 one can infer from eq . ( 3 ) that asymptotically @xmath64 is to hold if the gravitational force is to be finite , which implies that both sides of eq . ( 3 ) vanish ( see austin et al . 2005 ) ; on the other hand , for @xmath65 it is seen that either @xmath66 holds or a density cutoff occurs , if the mass @xmath40 is to saturate to a constant , implying @xmath67 . in other words , powerlaws with inner @xmath68 and outer slope @xmath69 will constitute asymptotic or effective approximations to the full solutions . for example , for the value @xmath70 one recovers the slopes @xmath71 and @xmath72 , found by taylor & navarro ( 2001 ) in their archetypical full solution . thus all full solutions may be conveniently visualized as _ bent down _ from the corresponding tangent powerlaw so as to meet the above asymptotic forms at both small and large radii . these solutions will be analytically studied and constrained next . in fact , the solution space of the jeans equation is analytically rich , so we refer the interested reader to the works by austin et al . ( 2005 ) , dehnen & mclaughlin ( 2005 ) , and barnes et al . ( 2006 , 2007 ) for extensive coverage . here we summarize the main results and proceed to focus on the ` critical ' solutions in the terminology of taylor & navarro ( 2001 ) . to highlight the properties of these solutions with running @xmath62 , it is convenient ( see dehnen & mclaughlin 2005 ; austin et al . 2005 ; also barnes et al . 2006 , 2007 ) to replace the original jeans equation @xmath73 with its first and second derivative , and use the differential mass definition @xmath74 , to obtain @xmath75 here @xmath76 and @xmath77 denote the first and second logarithmic derivative of @xmath78 with respect to @xmath61 ; in addition , @xmath37 is the point where the density slope matches @xmath57 ( see eq . [ 4 ] ) , while @xmath79 and @xmath80 are shorthands for the inner and the outer effective slopes anticipated above . in the second equation the normalization constant @xmath81 does not appear explicitly ; however , from the first eq . ( 6 ) evaluated at @xmath37 it is easily recognized that @xmath82 is related to @xmath83 , and in a full solution specifically measures the upward deviation of @xmath38 from the corresponding powerlaw , that is , above the parabola expressed by eq . ( 5 ) , see also fig . 1 . in fact , as the value of @xmath38 increases at given @xmath5 , the solutions will be progressively curved down both inward and outward of @xmath37 , to constitute a bundle _ tangent _ to the related powerlaw with slope given by eq . ( 4 ) . but , as stressed by taylor & navarro ( 2001 ) , only a definite maximal value @xmath12 can be withstood by the halo at equilibrium , lest the solution develops pathologies like a central hole ; this value marks a ` critical ' solution . correspondingly , the values for @xmath8 may be found by a painstaking trial - and - error procedure , as done by taylor & navarro ( 2001 ) for the particular case @xmath84 . more conveniently , they may be computed directly through eqs . ( 6 ) and ( 7 ) ; the former also provides the following exhaustive description of the solutions . in the _ inner _ region only three situations possibly arise ( williams et al . 2004 , dehnen & mclaughlin 2005 ) : ( i ) @xmath78 attains the value @xmath85 asymptotically for @xmath86 , which is unphysical since such a solution would end there with a diverging gravitational force ; ( ii ) the solution attains the value @xmath68 at a finite radius , and then continues to flatten down to the point of developing a central hole , which is obviously to be avoided ; ( iii ) @xmath78 approaches the value @xmath68 asymptotically for @xmath86 , that is the only _ viable _ possibility . thus physical solutions must approach the inner slope @xmath68 , and must steepen monotonically outwards . however , not all of them are tenable in view of their _ outer _ behavior ( austin et al . 2005 , dehnen & mclaughlin 2005 ) . here , again three situations formally arise that depend on the value of @xmath5 : ( i ) @xmath78 approaches the value @xmath85 asymptotically for @xmath65 , which occurs for @xmath87 and is unphysical since with such a solution the mass would diverge ( in practice , it would be ill - defined , see 5 ) ; ( ii ) @xmath78 attains the slope @xmath69 at a finite radius and then goes to an outer cutoff , that occurs for @xmath88 and is physically _ viable _ since the cutoffs actually set in at large radii exterior to the halo s bulk , as discussed below ; ( iii ) @xmath78 approaches the slope @xmath69 asymptotically for @xmath89 as it occurs for @xmath90 , which also provides a physically _ viable _ solution . summing up , the critical solutions satisfy both sets of criteria , but exist only for @xmath91 ( which excludes the case corresponding to the isothermal powerlaw ) ; hereafter these will be named ` @xmath5-_profiles _ ' . we compute them by integrating the second of eqs . ( 6 ) with boundary conditions @xmath92 and @xmath93 for @xmath94 . the results in the plane @xmath95 for various @xmath5 between @xmath96 and @xmath97 are plotted as different lines in fig . 1 ( left bottom panel ) ; there we can read out the values of @xmath83 at @xmath98 ( squares in the figure ) , and find the related @xmath12 after eq . ( 7 ) . the relation @xmath8 is plotted in fig . 1 ( right bottom panel ) ; the resulting correlation is fitted by the expression @xmath99 to better than @xmath100 . our solutions include the prototypical one corresponding to @xmath84 ( highlighted in blue ) , first found by taylor & navarro ( 2001 ) and shown to be well fitted , if only in a region around @xmath37 , by the nfw formula ( navarro et al . 1997 ) ; for such a solution the precise value of @xmath12 reads @xmath101 , see dehnen & mclaughlin ( 2005 ) . our solutions also include the one corresponding to the limiting value @xmath90 ( highlighted in red ) studied by dehnen & mclaughlin ( 2005 ) ; for this @xmath102 is found . in fig . 2 we illustrate the radial dependencies of various interesting quantities for the @xmath5-profiles : density @xmath2 and its logarithmic slope @xmath78 , circular velocity @xmath103 , velocity dispersion @xmath104 , entropy @xmath105 , mass @xmath40 , and gravitational potential @xmath106 ; the profiles are normalized to @xmath107 at the point @xmath37 within the halo s middle . outwards of this , solutions corresponding to larger values of @xmath5 have larger overall masses , flatter density profiles , higher velocity dispersions , and deeper potential wells . as mentioned above , the @xmath5-profiles with @xmath108 show an outer cutoff , to which the following remarks apply . first , the cutoffs generally occur beyond the virial radius ( see 5 ) . second , the velocity dispersion @xmath109 falls rapidly outside of @xmath37 and vanishes at the density cutoff , thus preventing any finite outflow related to residual random motions of dm particles . third , the microscopic distribution function as obtained through the eddington formula ( e.g. , binney & tremaine 1987 ) from pairs of @xmath22 and @xmath110 is realistic being nowhere negative ( see also zhao 1996 ; widrow 2000 ) . in fig . 3 we show the ratio @xmath111 entering eq . ( 2 ) ; we also show the energy distributions associated to the @xmath5-profiles : the specific potential energy @xmath112 @xmath113 , the random kinetic energy @xmath114 , and the local virial ratio @xmath115 . note that the latter tends to @xmath107 at large @xmath61 , as is expected for stable self - gravitating halos ( see okas & mamon 2001 ) ; in fact , it closely approaches unity already for @xmath116 . in fig . 4 we plot as a function of @xmath5 various characteristic radii ( in units of @xmath37 ) of the @xmath5-profiles ; specifically , we show the radius @xmath117 where the circular velocity @xmath118 peaks , the radius @xmath119 where the velocity dispersion @xmath21 peaks , the radius @xmath120 where the slope of the density profile reads @xmath121 , and the radius @xmath122 where the slope of the profile takes on the value @xmath123 . note that @xmath124 is always large , and formally recedes to infinity ( as represented in the plot with an arrow ) for the limiting case @xmath90 that has a powerlaw falloff . for the sake of clarity , in these figures we have confined the values of @xmath5 to between @xmath96 and @xmath97 ; although the upper bound is already required for having no central hole and a finite mass as discussed above , limited information existed so far concerning lower bounds from the isotropic jeans equation . here we show that the cosmogonic _ histories _ of dm halos enforce additional , stringent _ limitations _ to the range of @xmath5 ; on the high side the resulting bound turns out to be pleasingly close to the value @xmath97 derived above from the analysis of the static jeans equation , but on the low side it constitutes a new stringent result . consider a dm halo with current bounding radius @xmath125 , density @xmath2 , mass @xmath126 , velocity scales @xmath127 , and entropy @xmath128 . during the hierarchical buildup the running growth rate scales as @xmath129 with the inflow speed @xmath130 set by the depth of the dm well and hence proportional to @xmath131 ; thus the above quantities may be expressed in terms of @xmath132 and @xmath133 to provide the scaling laws : @xmath134 primarily we look for a link of the entropy slope @xmath5 with the mass growth rate in simple terms ; this analysis will have only heuristic scope , introductory to the full study performed in 3.2 . toward our scope , we make contact with the classic line of developments ( fillmore & goldreich 1984 ; lu et al . 2006 ) that analytically relate the collapse histories to the _ shape _ and growth rate of the primordial dm perturbations . the former may be expressed in terms of @xmath135 ; but since a halo virializes when @xmath136 attains a critical threshold ( e.g. , @xmath137 in the critical universe ) , the shape parameter @xmath138 effectively governs also the ( inverse ) mass growth rate defined in terms of the scaling @xmath139 . then we use the standard relations between redshift @xmath140 and cosmic time @xmath141 , that may be expressed in the form @xmath142 . in the critical cosmology @xmath143 holds ; in the canonical concordance cosmology the value @xmath144 still applies to @xmath145 , but as @xmath140 decreases @xmath146 progressively rises to values @xmath147 for @xmath148 . then the mass growth history writes simply @xmath149 and the growth rate reads @xmath150 ; from eqs . ( 8) , simple algebra yields the scaling laws @xmath151 , @xmath152 , and @xmath153 . on eliminating @xmath141 one obtains @xmath154 and @xmath155 with slopes @xmath156 the latter being related to the former by the simple scaling link @xmath157 , as expected at @xmath116 from eq . ( 4 ) . from eq . ( 9 ) @xmath5 is seen to be slightly larger than unity by a quantity slowly depending on @xmath158 ; this is consistent with @xmath159 , the straightforward message from eqs . ( 8) with @xmath160 . these relationships for @xmath5 and @xmath78 embody in a simple form some relevant cosmogonic and cosmological information . for @xmath161 where @xmath144 applies , one explicitly has @xmath162 and relatedly @xmath163 , the latter being the well - known expression for the similarity solutions to the spherical infall model in a critical universe . for @xmath148 instead when @xmath164 applies , one has @xmath165 and @xmath166 . as said , the remaining parameter @xmath167 governs both the shape of the initial perturbations and the times of their collapse . for instance , if the standard top - hat shape applied with @xmath168 corresponding to a monolithic collapse , from the above relations one would obtain @xmath169 and @xmath170 for halos that virialize at @xmath161 , as pointed out early on by bertschinger ( 1985 ) ; for more massive halos that virialize at @xmath148 , from the scaling laws one anticipates a slightly _ steeper _ @xmath171 and a _ flatter _ @xmath172 . qualitatively , different values of the slope @xmath78 may be related to other values of @xmath167 , such as occurring in an articulated shape of the initial perturbations , that also yields a sequence of collapse times . with a realistic bell - shaped perturbation @xmath168 still applies to its middle ; in the outer regions @xmath173 implies the accretion to become slower , and correspondingly the density slope @xmath78 to steepen there relative to the middle one . the central regions are described by @xmath174 , corresponding to fast collapse ; the scaling law then suggests the density slope to be flatter there . however , this simple approach can not be carried very far . at small radii also angular momentum effects come into play to lower the density profile ; they introduce another radial scaling related to centrifugal barriers , as discussed in detail by lu et al . ( 2006 ) ; these authors show how a density slope around @xmath175 is enforced if efficient isotropization of particles orbits takes place during the fast collapse , while still flatter slopes occur with more even angular momentum distributions . these , however , must be consistent with the overall equilibrium requirements ; these orbital features are actually subsumed into the macroscopic jeans eq . ( 2 ) with @xmath176 and provide the running @xmath177 ( the last term is shown in fig . 3 ) that is to replace the stiff scaling value @xmath157 above , but still recovers it at the single radius @xmath37 ( cf . [ 4 ] and [ 6 ] ) . so we will _ depart _ from the above heuristic approach , and focus on computing the values of @xmath5 to be inserted into jeans , also implementing precise mass and time dependencies of the growth rate @xmath167 . to focus these sensitive issues a refined analysis and a careful discussion are required and performed next . we derive a robust time evolution and mass dependence of @xmath133 from an updated version of the extended press & schechter formalism ( eps , see lacey & cole 1993 ) . we are mainly interested in the average evolutionary behavior of halos , so we compute the mean growth rate @xmath133 of the current mass @xmath132 as @xmath178 in terms of the differential _ merger rate _ @xmath179 given in appendix a. this incorporates the full cosmological framework , and the detailed cold dm perturbation spectrum ; as to the former we implement the concordance cosmology , and as for the latter we adopt the shape given by bardeen et al . ( 1986 ) corrected for baryons ( sugiyama 1995 ) , and normalized as to yield a mass variance @xmath180 on a scale of @xmath181 mpc . we improve over previous studies of this kind ( e.g. , miller et al . 2006 ; li et al . 2007 ; salvador - sol et al . 2007 ) by using the merger rate appropriate for ellipsoidal collapses ( sheth & tormen 2002 ) as provided by zhang et al . ( 2008 ) ; this is shown by the latter authors to yield formation times and growth rates in much closer agreement with numerical simulations than the standard eps theory . further details are presented in appendix a. eq . ( 10 ) provides the relation @xmath182 illustrated in fig . 5 . this may be viewed as a first order differential equation for @xmath183 , that is numerically solved once the mass @xmath184 at the present time @xmath185 has been assigned in the way of a boundary condition . in terms of @xmath183 so computed , the evolution of the halo properties are obtained through eqs . ( 8) ; the results are illustrated in fig . 6 , where we plot the time and redshift evolutions for the inverse growth rate @xmath167 , the overall radius @xmath125 , the mass @xmath132 , the characteristic velocities @xmath186 , the entropy @xmath105 , and finally the corresponding slope @xmath187 . the curves are for different _ current _ values of the halo masses @xmath184 ranging from @xmath188 to @xmath189 . in fact , in these detailed mass accretion histories one can recognize a _ transition _ epoch when the circular velocity @xmath131 of the halo attains its highest value along an accretion history ( see li et al . the transition separates two stages : an early stage of fast _ collapse _ , when @xmath190 corresponds to a growth rate still high on the running hubble timescale , and nearly constant ; and a later phase of slower _ accretion _ when @xmath167 takes off from the value @xmath191 , i.e. , the growth rate is low and decreasing . small halos with @xmath192 have high transition redshifts , hence are now well advanced into their slow accretion stage ; on the other hand , very massive ones with @xmath193 have low transition redshifts , so today they have just completed their fast accretion stage . a similar behavior has been identified in a number of recent numerical simulations ( zhao et al . 2003 ; diemand et al . 2007 ) , and has been recognized also in other semianalytic computations based on the simpler , less precise eps theory ( see li et al . 2007 ; neistein et al . 2006 ) . from fig . 6 we also see that the values of @xmath5 at the transition are narrowly _ constrained _ to within @xmath194 , i.e. , a @xmath195 range . we note that this range explains why narrowly constrained phenomenological values of @xmath5 are found within the halo _ bulk _ from many and different simulations since the year 2000 . in particular , our values turn out to agree with the best fitting exponent of @xmath23 [ or @xmath25 ] with its scatter recently given in detail by ascasibar & gottlber ( 2008 ) within the halo bulk during calm infall ; these authors warn against extrapolating their results to early times and outer radii . correspondingly , in our findings the values of @xmath5 before the transition should not be overinterpreted nor inserted into the jeans equation . in fact , at these early epochs the halos experience rapid changes and are clearly out of equilibrium ; during this stage the @xmath3body experiments ( see zhao et al . 2003 ; peirani et al . 2006 ; hoffman et al . 2007 ) highlight turmoil associated with fast collapse and major mergers , effective dynamical relaxation and mixing of dm particles . taking up from 3.1 , it is during these events that the orbits angular momentum must set to a near isotropic distribution such as to lower the inner densities . but as the turmoil associated with the fast collapse subsides , the orbital structure is bypassed and replaced by the macroscopic entropy distribution @xmath176 with uniform values of @xmath5 , as phenomenologically found to hold throughout the now stable halos bulk . then the values of @xmath5 at the transition epoch and the corresponding edge @xmath197 can be derived from the scaling laws and applied to the whole inner region , to obtain @xmath198 , taking on the value @xmath199 _ only _ at the point @xmath37 according to the results of 2.1 . the value of @xmath5 is expected to persist in the bulk during the subsequent development in the slow accretion stage , when mass is added smoothly and mainly onto the outskirts , with little mixing . such a development is borne out by the widely shared scenario of halo buildup from the inside out , championed by bertschinger ( 1985 ) , taylor & navarro ( 2001 ) , and many others up to salvador - sol et al . specifically , the analysis by the former authors concerning the randomization and the stratification of particles orbits during secondary infall indicates that their internal disposition is little affected by subsequent , smoother mass additions on top of a closely static potential well . thus during the late , slow accretion stage the structure for @xmath200 will be unaffected . meanwhile , the _ scale @xmath201 beyond @xmath37 is increasingly stretched out to substantially larger values ( see fig . 6 , top right panel ) ; correspondingly , we find the outer slope @xmath202 at @xmath203 to decline in time ( see fig . 6 , bottom right panel ) . this translates into a radial decrease @xmath204 on maintaining entropy stratification also into these regions ( see bertschinger 1985 , taylor & navarro 2001 ) . such a decrease is to be expected also from considering that @xmath205 applies , on the basis of the related asymptotic behaviors of @xmath206 and @xmath207 where @xmath208 . we find the values of @xmath5 to decrease by less than @xmath209 out to @xmath210 , and less than @xmath211 out to @xmath212 . concerning the jeans equation , this too ( with its boundary conditions set at @xmath213 , see 2.2 ) technically works from the inside out , and may be extended to the range @xmath214 with decreasing slope @xmath204 . the corresponding density run is somewhat steepened for @xmath215 , as if shifting from the upper to the lower curves in fig . 2 ; we have checked that the outer density profile lowers by less than @xmath216 for @xmath217 . state - of - the - art simulations still yield little information on the outskirts for @xmath218 owing to low densities and noisy data there discussed by ascasibar & gottlber ( 2008 ) . we present elsewhere the complete solutions outside @xmath37 as a testbed for future simulations aimed at resolving the outskirts , and at fitting gravitational lensing observations ( see discussion in 5 ) . in a similar vein , we predict a minor increase of @xmath5 at the transition from @xmath6 to @xmath31 in moving from galaxy- to cluster - sized current masses . a fit accurate to better than @xmath219 to this average relation is provided by @xmath220 ^ 2 $ ] ; in particular for galaxy clusters , this implies values of @xmath221 for abell richness classes @xmath222 and @xmath171 for richness class @xmath223 . the correlation @xmath224 may help to understand the differences in the values of @xmath5 found in numerical simulations ; in fact , the highest value @xmath31 to date has been reported by rasia et al . ( 2004 ) in looking at cluster - sized halos , while the value @xmath6 first found by taylor & navarro ( 2001 ) refers to galaxy - sized halos . observational evidence supporting the above picture is presented in 5 . to sum up , from the cosmological buildup we find for the entropy slope the narrow allowed range @xmath225 this we use in our jeans - like description of dm equilibria within the halos _ bulk_. the bounds on @xmath5 are relevant in two respects : the upper one strengthens the formal limit @xmath226 required for the solutions of the static jeans equation to have viable profiles in the far outskirts ( see 2.2 and discussion in 5 ) ; the lower one complementarily restricts the range of the viable values . here we propose to discuss the physical meaning of the quantity @xmath12 ( defined in 2.1 following taylor & navarro 2001 ) in terms of a direct energetic argument . we first note that given the asymptotics of the @xmath5-profiles , the velocity dispersion @xmath227 must peak at a point @xmath228 ( see fig . 4 ) to the value @xmath229 ; at this easily identified point the constant @xmath38 also writes @xmath230 , and thus is seen to compare the estimate @xmath231 for the random kinetic energy with the estimate @xmath232 for the gravitational potential . we then recall from 2.1 that a halo in the viable equilibrium condition for a given @xmath5 is marked by a maximal value @xmath8 , that is , by a _ minimal _ level of randomized energy @xmath233 . we now show that to a very good approximation the value for @xmath12 may be obtained from considering _ incomplete _ conversion of the infall kinetic energy @xmath234 of dm particles into their dynamically randomized energy @xmath235 during formation ; a process introduced by bertschinger ( 1985 ) , different from the complete violent relaxation proposed by lynden - bell ( 1967 ) and discussed widely , including the recent arad & lynden - bell ( 2005 ) , trenti & bertin ( 2005 ) and trenti et al . ( 2005 ) . we propose to derive the critical value of @xmath236 from the relation @xmath237 this expression for the effective conversion efficiency @xmath238 has been constructed on considering that : ( i ) it has to depend on the ratio of the ( @xmath239d ) infall velocity squared @xmath240 to the depth of the potential well as measured by the ( @xmath241d ) circular velocity squared @xmath242 ; ( ii ) the formal limit @xmath243 corresponding to @xmath53 ( the isothermal powerlaw , see 2.1 ) must be approached for a fast inflow with @xmath244 , which quantifies the incompleteness of the energy conversion during the transit of a bullet with crossing time short compared with the halo s dynamical time . as a simple example guided by the canonical theory of gravitational interactions ( see saslaw 1985 ; cavaliere et al . 1992 ) we may consider that @xmath245 is to hold for having considerable energy transfer ; using this simple approximation in eq . ( 12 ) yields @xmath246 , a value pleasingly close to @xmath247 as obtained numerically in 2.2 for the particular @xmath5-profile with @xmath248 . on the other hand , this simple argument can not pinpoint values of @xmath12 to the precision required to discriminate different values of @xmath5 . so , we may test our proposed @xmath238 as given in eq . ( 12 ) for different values of @xmath5 by using instead energy conservation to express the infall velocity @xmath249 in terms of the adimensional potential drop @xmath250 ( normalized to @xmath242 ) from the turnaround radius to @xmath117 ( see bertschinger 1985 ; lapi et al . 2005 ) . inserting this into eq . ( 12 ) yields the expression @xmath251 this highlights the deviation of @xmath12 from the value @xmath252 that would apply for the isothermal powerlaw ( see 3 ) . using the runs of @xmath253 appropriate for the @xmath5-profile as given in fig . 2 , we compute values for @xmath12 as a function of @xmath5 ; note that these decrease for increasing @xmath5 which implies larger @xmath254 , see fig . 2 . 7 shows how well eq . ( 13 ) recovers the @xmath8 relation obtained through numerical integration of the jeans equation . this excellent fit to @xmath12 means that the expression for @xmath238 in eq . ( 12 ) , more than a successful guess , is describing what actually occurs in the macroscopic equilibria expressed by the jeans equation in keeping with the numerical simulations . at the referee s suggestion we add that what ultimately the first eq . ( 12 ) describes with the isotropic form of its l.h.s . is a partial effective conversion at @xmath255 from radial infall into tangential motions . the general picture of dm halo formation as it emerges from recent @xmath3body simulations envisages a progressive buildup through a two - stage development . at early epochs major mergers associated with fast reshaping of the bulk gravitational potential cause particles orbits to dynamically relax and mix into a definite equilibrium disposition . at later epochs the halo outskirts progressively build up through rare if persisting minor mergers and by smooth mass accretion , settling to a quasi - static equilibrium . the stable equilibria soon after the early stage are effectively described in terms of a phenomenological entropy run @xmath176 with uniform slope @xmath5 somewhat larger than @xmath107 . spurred by this powerlaw behavior , one might look for powerlaw solutions of the jeans equation for the equilibrium density profiles , only to find these just tangent to the actual solutions . from here our analysis goes beyond powerlaw densities , and provides two related blocks of novel results . first , we have computed and illustrated a wide set of novel solutions to the isotropic jeans equation , that we name @xmath5-_profiles _ , including as two particular instances the prototypes investigated by taylor & navarro ( 2001 ) and dehnen & mclaughlin ( 2005 ) . we have stressed that at given @xmath5 the solutions form a tangent bundle with @xmath22 progressively curved down as the gravitational pull measured by @xmath236 increases above the value corresponding to the pure powerlaw profile @xmath157 ; from such bundles the @xmath5-profiles stand out for having a shape maximally curved to the point of complying with sensible outer and inner behaviors , i.e. , finite total mass and no central hole . this occurs for a maximal value @xmath8 of the gravitational pull @xmath242 relative to the random component @xmath231 , i.e. , for a minimal @xmath231 compared to @xmath242 ; this also corresponds to the maximal slope @xmath256 at the point @xmath117 . such solutions still constitute a large set , being parameterized in terms of a wide overall range @xmath257 for the values of @xmath5 . second , we have shown that toward further constraining these values the key step is provided by the halos cosmological buildup ; in fact , the scaling laws of 3 narrowly restrict the viable _ range _ to @xmath258 , corresponding to current masses @xmath259 . we have found these bounds to apply at the epochs of _ transition _ from fast collapse to slow accretion , when the potential wells settle to their maximal depths along an accretion history ( see 3 ) . in fact , our semianalytic analysis not only recovers the two - stage development observed in many recent @xmath3body simulations , but also explains _ why _ only such narrowly constrained values of @xmath5 are actually found in many different numerical works since the year 2000 . the values @xmath12 is correspondingly constrained to the range @xmath260 ; we trace back these values to limited _ randomization _ for the infall kinetic energy of dm particles during accretion , which leads to minimal values of @xmath104 . the @xmath261profiles from the jeans equilibrium with @xmath5 and @xmath12 so constrained imply that galaxies have a generally more _ compact _ central structure than groups and clusters , with appreciable _ deviation _ from self - similarity ; on the other hand , the former have currently developed extensive outskirts , while the very rich clusters do not yet . to embody in a handy expression the basic features of the density runs for the @xmath261profiles , we provide here an analytic fitting formula of the form often advocated on phenomenological grounds ( see zhao 1996 ; kravtsov et al . 1998 ; widrow 2000 ; barnes et al . 2006 ) , that reads @xmath262^{q}},\ ] ] with @xmath68 governing the inner profile , @xmath146 the outer one , @xmath263 and @xmath264 the position and sharpness of the radial transition . here for the first time such a profile is _ substantiated _ with parameters expressed in terms of the derivatives of the jeans equation ( see 2.2 ) and of the values for @xmath5 as detailed in appendix b. we stress that the jeans equlibrium ( see fig . 2 ) requires that the central slopes are always _ flatter _ , and the outer ones always _ steeper _ than in the empirical nfw fit . the goodness of the above fitting formula is discussed in appendix b , and illustrated in fig . we probe the physical relevance of the @xmath261profiles by comparing in fig . 9 the related projected density runs with the results from recent , exquisite observations of strong and weak gravitational lensing in and around massive galaxy clusters , which just require _ flat _ inner , and _ steep _ outer behaviors ( broadhurst et al . 2008 , umetsu & broadhurst 2008 , and references therein ) . it is seen how _ straightforward _ are our fits in terms of high - mass @xmath261profiles that naturally exhibit just such features , compared to those in terms of the nfw empirical formula that not only has problems of ill - defined mass , but also would require particularly high concentrations and old ages . our findings substantiate the @xmath5-profiles , and stress their value as to specific predictions for future simulations aimed at resolving , and for weak gravitational lensing observations aimed at probing the outer cluster densities . a full discussion of these issues deserves a dedicated paper . two comments are in order concerning the validity of the jeans equation in the simple form of eq . first , reasonable anisotropies in the dm are described by the standard parameter @xmath265 ( binney 1978 ) , which numerical simulations suggest to increase with @xmath61 from central values @xmath266 meaning near isotropy to outer values close to @xmath267 , meaning progressive outward prevalence of radial motions ( see hansen & moore 2006 ) . such anisotropies inserted into jeans produce small variations of the @xmath261profiles , to the effect of slightly increasing the limiting value @xmath268 ; thus in the related limiting solutions the inner slope is flattened @xmath269 , the middle slope @xmath270 is steepened , and the outer slope @xmath271 is also steepened with respect to the isotropic case , see dehnen & mclaughlin ( 2005 ) . the oppositely extreme case of total radial anisotropy with constant @xmath272 discussed by austin et al . ( 2005 ) would yield a lower limiting @xmath273 , while also producing a steep inner slope @xmath274 ( cf . 2.1 ) . in brief , the distinctive features of the @xmath261profiles are somewhat enhanced , if anything , with radially increasing anisotropies . second , the static jeans equation becomes increasingly vulnerable in going into the outermost regions , since static or even dynamic equilibrium is not granted beyond the virial radius . specifically , in the concordance cosmology the hubble expansion takes over at radii exceeding a few times the virial radius where @xmath275 holds . this weakens the case for the formal upper limit @xmath276 as derived from jeans analysis but correspondingly highlights the relevance of the ceiling @xmath277 for a few @xmath189 halo we have derived from cosmogonic evolution . on the other hand , any halo termination will be uncertain by full factors of @xmath252 ; with flat outer density slopes , as the case with @xmath87 ( see end of 2.1 ) , the related total masses will be ill - defined by similar factors , corresponding for galaxy clusters to a disturbing uncertainty by @xmath278 in abell richness class ( see yee & lpez - cruz 1999 ) . finally , we comment on our use of the dm ` entropy ' @xmath279 actually the analog of the thermodynamic adiabat , related to entropy by @xmath280 rather than the equivalent functional ` phase - space density ' @xmath281 that underlies the ` information ' @xmath282 with the obvious minus sign . matter - of - factly , the pressure gradient @xmath283 that withstands the self - gravity in eq . ( 1 ) is most directly expressed in terms of the macroscopic quantity @xmath105 , in the form @xmath284 . in the way of interpretations , our lead is provided by the positive - definite gradient @xmath285 , and specifically by its adimensional version @xmath286 . we find @xmath287 to be convenient in understanding the empirical evidence as to positive and uniform of @xmath5 within the halos bulk after the transition ; then and there , in fact , the particle orbits settle to a definite dynamical regime , relaxed toward a periodic , ordered behavior ( bertschinger 1985 ) , corresponding to _ lower _ entropy layers at the center . in the outer regions , on the other hand , the particle orbits stratify but mix poorly during the late , slow buildup ( see bertschinger 1985 ; taylor & navarro 2001 ) with their apocenters distributed over larger and larger sizes ; we understand this dynamical regime in terms of @xmath105 or @xmath288 spread over increasingly stretched radii @xmath289 , resulting in _ flatter _ slopes . it rests with the numerical values of @xmath5 to translate the fine orbital details into the macroscopic average of jeans , as clearly indicated by the consistency of the results from the latter with semianalytic cosmogony and with simulations . we end by stressing that the formulation of dm distribution in terms of ` entropy ' despite the standard objections about defining entropy for a collisionless medium out of thermal equilibrium and dominated by long range weak gravity opens an interesting perspective ; this envisages direct comparisons ( see faltenbacher et al . 2007 ) with the truly thermodynamic entropy of the other major component in galaxy systems , namely , the collisional plasma in local thermal equilibrium constituting the intracluster medium . in fact , in the outer regions these entropies though related as being both originated by conversion of the infall energy under _ weak _ gravity , differ owing to their origin from _ different _ conversion processes ( lapi et al . 2005 ; cavaliere & lapi 2006 ) ; in addition , in the central region of most clusters the hot plasma is heated up by the activity of supermassive black holes originating under _ strong _ gravity conditions . work supported in part by asi and inaf . we thank giuseppe bertin , luigi danese , roberto fusco - femiano , paolo salucci and chiara tonini for helpful discussions . we acknowledge the constructive comments by an anonymous referee , especially concerning the role of angular momentum and tangential motions in 4 and 5 . to reliably compute the growth rate @xmath133 in eq . ( 10 ) , we use the merging rate of dm halos under ellipsoidal collapse as provided by zhang et al . ( 2008 ) ; this reads @xmath290\right\}\ ] ] where @xmath291 is the difference of the mass variances corresponding to the masses @xmath292 and @xmath132 , and @xmath293^{-0.615}\right\}~,\\ & \nonumber & \\ & & a_1\equiv 0.308\,\left[\delta_c(z)^2/ \sigma_m^2\right]^{-0.115}~,\\ & \nonumber & \\ & \nonumber & a_2\equiv 0.0373\,\left[\delta_c(z)^2/ \sigma^2_m\right]^{-0.115}~.\end{aligned}\ ] ] note that the standard eps expression for the spherical collapse model ( lacey & cole 1993 ; kitayama & suto 1996 ) is recovered on setting @xmath294 and @xmath295 . ( a1 ) and ( a2 ) involve the critical threshold @xmath296 for collapse , which is computed according to @xmath297 in terms of the normalization value @xmath298 $ ] and of the growth factor @xmath299^{-1}~;\ ] ] in the above the quantity @xmath300 $ ] represents the evolved matter density parameter of the universe in the concordance cosmology . ( a1 ) and ( a2 ) also involve the mass variance of the primordial perturbation spectrum , that writes @xmath301 where @xmath302 is the wavenumber corresponding to the mass @xmath132 , and @xmath303 mpc@xmath304 is the mean background matter density of the universe . the mass variance is then normalized to the value @xmath180 for the mass corresponding to a scale of @xmath181 mpc . as for the power spectrum , we adopt the shape @xmath305 in terms of the transfer function @xmath306^{-1/4}\ ] ] given by bardeen et al . ( 1986 ) . here @xmath307 includes the shape correction factor @xmath308 due to baryons , after sugiyama ( 1995 ) ; we recall that the values @xmath309 , @xmath310 , and @xmath311 are adopted throughout the paper . here we provide an explicit analytic fit to the @xmath5-profiles , whose numerical computation is sensitive to precise initialization . we use the expression @xmath312 inspired by the exact solution for @xmath90 given by dehnen & mclaughlin ( 2005 ) , and representing an extension of the nfw ( navarro et al . 1997 ) formula ; in the above @xmath313 is the adimensional radius , the inner slope has been fixed to @xmath314 , while @xmath264 , @xmath146 , and @xmath263 are parameters dependent on @xmath5 to be determined . to this purpose , we impose that at @xmath98 the density slope from eq . ( b1 ) and its first and second logarithmic derivatives equal those of the @xmath5-profiles . thus we find @xmath315 then we solve the above expressions for @xmath263 , @xmath264 , and @xmath146 , to obtain @xmath316 the values of @xmath317 and @xmath318 are easily computed as a function of @xmath5 . first , eqs . ( 6 ) evaluated at @xmath98 give @xmath319 then these quantities are easily recast in terms of @xmath5 on recalling from the main text ( 2.2 ) that @xmath314 , @xmath320 , @xmath123 , and @xmath321 . to sum up , ( b3 ) and ( b4 ) provide the parameters @xmath322 , @xmath323 , and @xmath324 as a function of @xmath5 . e.g. , for @xmath90 we obtain @xmath325 , @xmath326 , and @xmath327 ; thus our fitting formula recovers the exact solution by dehnen & mclaughlin ( 2005 ) . at the other extreme , for @xmath328 we obtain @xmath329 , @xmath330 , and @xmath331 ; here our fitting formula approximates well ( see fig . 8) the exact solution , improving upon the fit by taylor & navarro ( 2001 , their eq . [ 10 ] ) emulates the cutoff occurring for @xmath328 ; for @xmath90 the lower @xmath332 emulates the asymptotic falloff of this solution . ] . in fact , we find that for _ any _ @xmath5 between @xmath6 and @xmath97 our fit works to better than a few @xmath333 throughout the significant radial range from @xmath334 to several @xmath37 ; note that both the radial range and the quality of the fit increase for increasing @xmath5 , up to the point of pinning down the exact solution for @xmath90 . williams , l.l.r . , austin , c. , barnes , e. , babul , a. , & dalcanton , j.j . 2004 , in baryons in dark matter halos , eds . r. dettmar , u. klein , and p. salucci ( trieste , italy : sissa ) , see http://pos.sissa.it , p.20.1	in the dark matter ( dm ) halos embedding galaxies and galaxy systems the ` entropy ' @xmath0 ( a quantity that combines the radial velocity dispersion @xmath1 with the density @xmath2 ) is found from intensive @xmath3body simulations to follow a powerlaw run @xmath4 throughout the halos bulk , with @xmath5 around @xmath6 . taking up from phenomenology just that @xmath7 const applies , we cut through the rich analytic contents of the jeans equation describing the self - gravitating equilibria of the dm ; we specifically focus on computing and discussing a set of novel physical solutions that we name @xmath5-_profiles _ , marked by the entropy slope @xmath5 itself , and by the maximal gravitational pull @xmath8 required for a viable equilibrium to hold . we then use an advanced semianalytic description for the cosmological buildup of halos to constrain the values of @xmath5 to within the narrow range @xmath9 from galaxies to galaxy systems ; these correspond to halos current masses in the range @xmath10 . our range of @xmath5 applies since the transition time that @xmath11 both in our semianalytic description and in state - of - the - art numerical simulations @xmath11 separates two development stages : an early violent collapse that comprises a few major mergers and enforces dynamical mixing , followed by smoother mass addition through slow accretion . in our range of @xmath5 we provide a close fit for the relation @xmath8 , and discuss a related physical interpretation in terms of incomplete randomization of the infall kinetic energy through dynamical mixing . we also give an accurate analytic representation of the @xmath5-profiles with parameters derived from the jeans equation ; this provides straightforward _ precision _ fits to recent detailed data from gravitational lensing in and around massive galaxy clusters , and thus replaces the empirical nfw formula relieving the related problems of high concentration and old age . we finally stress how our findings and predictions as to @xmath5 and @xmath12 contribute to understand hitherto unsolved issues concerning the fundamental structure of dm halos .
the field of exact enumeration of tilings ( equivalently , perfect matchings ) dates back to the early 1900 s when macmahon proved his classical theorem on the number of plane partitions that fit in a given box ( see @xcite ) . this theorem is equivalent to the fact that the number of unit rhombus ( or lozenge ) tilings of a hexagon @xmath0 of side - lengths @xmath1 , @xmath2 , @xmath3 , @xmath1 , @xmath2 , @xmath3 ( in cyclic order ) drawn on the triangular lattice is equal to @xmath4 ( we use the operator @xmath5 to denote both the number of tilings of a lattice region and the number of perfect matchings of a graph , as the two objects can be identified by a well known bijection ) . another classical result , from the early 1960 s , is the enumeration of domino tilings of a rectangle on the square lattice , due independently to kasteleyn @xcite and temperley and fisher @xcite . this states that the number of domino tilings of a @xmath6 by @xmath7 rectangle on the square lattice equals @xmath8 in the early 1990 s , elkies , kuperberg , larsen and propp @xcite considered another family of simple regions on the square lattice called aztec diamonds ( see figure [ aztecdiamond ] for an example ) , and proved that the number of domino tilings of the aztec diamond of order @xmath9 is given by the simple formula @xmath10 a large body of related work followed ( see e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and the references in @xcite for a more extensive list ) , centered on families of lattice regions whose tilings are enumerated by simple product formulas . the state of affairs at the end of that decade is captured by propp s paper @xcite published in 1999 , which presented a list of 32 open problems in the field of enumeration of perfect matchings . most of those 32 problems have been solved in the meanwhile , but some are still open . in this paper we solve and generalize one of these open problems ( problem 16 on propp s list @xcite ) . our methods also provide a new proof and a generalization for a related result of douglas @xcite . problem 16 on propp s list @xcite concerns a family of quasi - hexagonal regions on the lattice obtained from the square lattice by drawing in every third southwest - to - northeast diagonal . the case when the side - lengths of the quasi - hexagon are @xmath11 , @xmath12 , @xmath12 , @xmath11 , @xmath12 , @xmath12 ( clockwise from top ) is illustrated in figure [ hexagonp ] . problem 16 of @xcite asks for a formula for the number of tilings of the quasi - hexagon of sides @xmath1 , @xmath2 , @xmath3 , @xmath1 , @xmath2 , @xmath3 ( the sides of length @xmath1 are the ones along _ _ diagonals _ _ of the square lattice ) , where the allowed tiles are unions of two fundamental regions of the resulting dissection of the square lattice sharing an edge . ( 0,0 ) # 1#2#3#4#5 @font ( 4289,2665)(222,-1852 ) ( 237,485)(0,0)[lb ] ( 2009,603)(0,0)[lb ] ( 2836,-224)(0,0)[lb ] ( 2481,-1523)(0,0)[lb ] ( 1182,-1759)(0,0)[lb ] ( 237,-933)(0,0)[lb ] as mentioned in @xcite , the case @xmath13 ( to which also the cases @xmath14 and , by symmetry , @xmath15 turn out to reduce ) has been solved by ben wieland ( in unpublished work ) , who showed that the number of tilings is given in these situations by powers of 2 . in this paper we prove a simple product formula in the case @xmath16 , which was previously open . the result turns out to fully justify propp s initial motivation , stated in @xcite : `` one reason for my special interest in problem 16 is that it seems to be a genuine hybrid of domino tilings of aztec diamonds and lozenge tilings of hexagons . '' indeed , we show that by a sequence of graph transformations which can also be used to prove the aztec diamond theorem ( [ diamond ] ) the dual graph can be transformed into a honeycomb graph whose number of perfect matchings is given by macmahon s formula ( [ mcmahon ] ) . we in fact enumerate the tilings of more general regions , which we describe next . rather than considering quasi - hexagonal regions on the square lattice on which diagonals are drawn in so that the distance . ] between any two successive ones is 3 , we allow the general situation when the distances between successive diagonals are _ arbitrary_. let @xmath17 be a fixed diagonal ( this will contain the middle two vertices of the quasi - hexagon ) , and assume that @xmath18 diagonals have been drawn in above it , with the distances between successive ones , starting from top , being @xmath19 ( @xmath20 is the distance between the bottommost of these @xmath18 diagonals and @xmath17 ) . assume also that @xmath21 diagonals have been drawn in below @xmath17 , with the distances between successive ones , starting from the bottom , being @xmath22 ( @xmath23 is the distance between the topmost of these @xmath21 diagonals and @xmath17 ) . given a positive integer @xmath1 , we define the quasi - hexagonal region @xmath24 as follows ( see figure [ hexagonregion ] for an example ) . its top and bottom boundaries are along the top and bottom diagonals that have been drawn in . its southwestern and northeastern boundaries are defined in the next two paragraphs . color the resulting dissection of the square lattice black and white so that any two fundamental regions that share an edge have opposite color , and assume that the triangles just below the top diagonal are white . let @xmath25 be a lattice point on the top diagonal . start from @xmath25 and take unit steps south or east so that for each step the color of the fundamental region on the left is black . we arrive @xmath17 at a lattice point @xmath26 . continue downward from @xmath26 in similar fashion , with the one difference that now we require the fundamental region on the left to be white for each step . let @xmath27 be the lattice point on the bottom diagonal that is reached by these steps . the described path from @xmath25 to @xmath27 is the southwestern boundary of our region . ( 0,0),title="fig : " ] # 1#2#3#4#5 @font ( 5276,5169)(406,-4616 ) ( 421,-1200)(0,0)[lb ] ( 1350,-2900)(0,0)[lb ] ( 3200,-4141)(0,0)[lb ] ( 4500,-2986)(0,0)[lb ] ( 1500,0)(0,0)[lb ] ( 3900,-1329)(0,0)[lb ] ( 1239,-586)(0,0)[lb ] ( 4300,-646)(0,0)[lb ] ( 3025,131)(0,0)[lb ] ( 3400,-283)(0,0)[lb ] ( 3500,-706)(0,0)[lb ] ( 2500,-4231)(0,0)[lb ] ( 1800,-3700)(0,0)[lb ] the northeastern boundary is defined analogously , starting from the lattice point @xmath28 on the top diagonal that is @xmath1 unit square diagonals to the northeast of @xmath25 ( i.e. @xmath29 ) . take unit steps south or east so that all fundamental regions on the left are white , until a point @xmath30 on @xmath17 is reached , and continue with similar unit steps so that the fundamental regions on the left are black , until a point @xmath31 on the bottom diagonal is reached . the described path from @xmath28 to @xmath31 completes the boundary of @xmath32 . we are only interested in connected regions , so we assume in addition that _ the northeastern and the southwestern boundaries do not intersect each other . _ our generalization of problem 16 of @xcite consists of providing an explicit formula for the number of tilings of such regions . we need some additional definitions and terminology as below . call the fundamental regions inside @xmath32 _ cells _ , and call them black or white according to the coloring described above . note that there are two kinds of cells , square and triangular . the latter in turn come in two orientations : they may point towards @xmath17 or away from @xmath17 . a cell is called _ regular _ if it is a square cell or a triangular cell pointing away from @xmath17 . a _ row of cells _ consists of all the triangular cells of a given color with bases resting on a fixed lattice diagonal , or consists of all the square cells ( of a given color ) passed through by a fixed lattice diagonal . define the _ upper height _ of our region to be the number of rows of regular black cells above @xmath17 . the _ lower height _ is the number of rows of white regular cells below @xmath17 ( for example , the quasi - hexagon in figure [ hexagonregion ] has both the upper and lower heights equal to 6 ) . we are now ready to state our main result . [ main ] let @xmath1 be a positive integer , and let @xmath33 be positive integers . @xmath36 unless the upper and lower heights are equal and the bottom row of cells is black . @xmath34b@xmath35 . suppose that the upper and lower heights of @xmath32 are both equal to @xmath37 , and that the bottom row of cells is black . let @xmath27 be the number of regular black cells above @xmath17 , and @xmath38 the number of regular white cells below @xmath17 . then for any positive integer @xmath1 we have @xmath39 where @xmath40 is the number of rows of black triangular cells above @xmath17 pointing away from @xmath17 , and @xmath9 is the number of rows of black triangular cells above @xmath17 pointing towards @xmath17 . by macmahon s formula ( [ mcmahon ] ) and equality ( [ maineq ] ) , the number of tilings of the quasi - hexagon is equal to @xmath41 when @xmath42 . note that the original problem corresponds to @xmath43 and @xmath44 . the resulting region is then precisely the quasi - hexagon of sides @xmath1 , @xmath18 , @xmath18 , @xmath1 , @xmath18 , @xmath18 . in particular , for @xmath45 and @xmath46 , ( [ maineq ] ) states that the number of tilings of the quasi - hexagon shown in figure [ hexagonp ] is equal to @xmath47 , thus explaining how this factorization ( mentioned explicitly in @xcite ) comes about . in the case of odd @xmath48 s and @xmath49 s , our quasi - hexagons have the following nice facts : the bottom row of cells is always black , all the black cells above @xmath17 and all the white cells below @xmath17 are regular , and the upper and lower heights are equal to @xmath50 and @xmath51 , respectively . therefore , we have a special form of theorem [ maineq ] as follows . [ odd - main ] let @xmath19 and @xmath22 be odd positive integers . assume that @xmath52 let @xmath27 be the number of black cells above @xmath17 , and @xmath38 be the number of white cells below @xmath17 . then for any positive integer @xmath1 we have @xmath53 note that when @xmath43 in the original quasi - hexagons , the balancing condition ( [ balancing ] ) is equivalent to @xmath44 . as mentioned in @xcite , we do not have a simple product formula for the number of tilings of the quasi - hexagon of problem 16 in @xcite when @xmath1 , @xmath2 , @xmath3 are all distinct . however , we show that our graph transformations can still be used to turn the dual graph of the quasi - hexagon into a conceptually simple graph consisting of a honeycomb part and an aztec rectangle part , and therefore the number of perfect matchings of the former is equal to a power of @xmath12 times the number of perfect matchings of the latter ( see theorems [ asymodd ] and [ asymmetric1 ] ) . ( 0,0 ) , @xmath54 and @xmath55.,title="fig : " ] # 1#2#3#4#5 @font ( 6283,2966)(402,-2292 ) ( 2363,-1641)(0,0)[lb ] ( 6260,-1878)(0,0)[lb ] ( 828,-696)(0,0)[lb ] we also extend a theorem of douglas @xcite concerning a particular family of regions on the lattice obtained from @xmath56 by drawing in every _ second _ diagonal ( see figure [ douglas ] ) , to the case of arbitrary distances between the diagonals . let @xmath57 be the portion of @xmath32 that is above the diagonal @xmath17 ( i.e. the top half of the generalized quasi - hexagon ) ; we call it a _ generalized douglas region_. the _ height _ of @xmath57 is defined to be equal to the upper height of @xmath32 . a triangle pointing towards @xmath17 in @xmath57 is called a _ down - pointing triangle _ , and a triangle pointing away from @xmath17 in this region is called a _ up - pointing triangle_. our extension of douglas result is the following . [ douggen ] let @xmath1 be a positive integer , and let @xmath19 be positive integers . let @xmath37 be the height of @xmath57 , @xmath40 ( resp . , @xmath9 ) be the number of rows of up - pointing ( resp . , down - pointing ) triangular black cells in @xmath57 , and let @xmath27 be the number of black cells that are not down - pointing triangles . @xmath34a@xmath35 . if the the bottom row of cells of @xmath57 is black , or if the equality @xmath58 fails , then @xmath59 . @xmath34b@xmath35 . suppose ( [ balancing2 ] ) holds and the bottom row of cells of @xmath57 is white . then we have @xmath60 douglas theorem corresponds to the special case when @xmath18 is odd and @xmath61 , @xmath62 , when the balancing condition ( [ balancing2 ] ) requires @xmath63 ( strictly speaking , in douglas original definition the top and bottom rows of cells consist of squares rather than triangles ; however , the dual graph is isomorphic to the dual graph of our region , so the two versions have the same number of tilings ) . thus , one can view theorem [ main ] as a common generalization of douglas theorem @xcite and propp s open problem on quasi - hexagons . a _ perfect matching _ of a graph @xmath64 is a collection of edges such that each vertex of @xmath64 is incident to precisely one edge in the collection . the number of perfect matchings of @xmath64 is denoted by @xmath65 . more generally , if the edges of @xmath64 have weights on them , @xmath65 denotes the sum of the weights of all perfect matchings of @xmath64 , where the weight of a perfect matching is the product of the weights on its constituent edges . given a lattice in the plane , a ( lattice ) _ region _ is a finite connected union of fundamental regions of that lattice . tile _ is the union of any two fundamental regions sharing an edge . a _ tiling _ of the region @xmath66 is a covering of @xmath66 by tiles with no gaps or overlaps . the tilings of a region @xmath66 can be naturally identified with the perfect matchings of its dual graph ( i.e. , the graph whose vertices are the fundamental regions of @xmath66 , and whose edges connect two fundamental regions precisely when they share an edge ) . in view of this , we denote the number of tilings of @xmath66 by @xmath67 . a _ forced edge _ of a graph is an edge contained in every perfect matching of @xmath64 . assume that @xmath64 is a weighted graph with weight function @xmath68 on its edges , and @xmath69 is obtained from @xmath64 by removing forced edges @xmath70 , and removing the vertices incident to those edges . then one clearly has @xmath71 _ hereafter , whenever we remove some forced edges , we remove also the vertices incident to them . _ the main results of this section are two graph transformation rules that change the number of perfect matchings in a simple predictable way ( see lemmas 3.4 and 3.5 ) . in proving them we will employ three basic preliminary results stated below . ( 0,0 ) # 1#2#3#4#5 @font ( 4188,1361)(593,-556 ) ( 1336,591)(0,0)[lb ] ( 3549,621)(0,0)[lb ] ( 3757 , 89)(0,0)[lb ] ( 3999,621)(0,0)[lb ] ( 820,-541)(0,0)[lb ] ( 1840,-535)(0,0)[lb ] ( 3031,-535)(0,0)[lb ] ( 4426,-484)(0,0)[lb ] [ vertex - splitting lemma][vs ] let @xmath64 be a graph , @xmath72 be a vertex of it , and denote the set of neighbors of @xmath72 by @xmath73 . for any disjoint union @xmath74 , let @xmath69 be the graph obtained from @xmath75 by including three new vertices @xmath76 , @xmath77 and @xmath78 so that @xmath79 , @xmath80 , and @xmath81 ( see figure [ vertexsplitting ] ) . then @xmath82 . [ star ] let @xmath64 be a weighted graph , and let @xmath72 be a vertex of @xmath64 . let @xmath69 be the graph obtained from @xmath64 by multiplying the weights of all edges incident to @xmath72 by @xmath83 . then @xmath84 . part ( a ) of the following result is a generalization ( due to propp ) of the urban renewal " trick first observed by kuperberg . parts ( b ) and ( c ) are due to ciucu ( see lemma 2.6 in [ 5 ] ) . ( 0,0 ) # 1#2#3#4#5 @font ( 4054,1735)(340,-948 ) ( 355,-156)(0,0)[lb ] ( 1064,-933)(0,0)[lb ] ( 1891,-106)(0,0)[lb ] ( 1182,603)(0,0)[lb ] ( 2717,-189)(0,0)[lb ] ( 3426,-933)(0,0)[lb ] ( 4253,-106)(0,0)[lb ] ( 3426,603)(0,0)[lb ] ( 904,-382)(0,0)[lb ] ( 1396,-388)(0,0)[lb ] ( 1418,130)(0,0)[lb ] ( 946,130)(0,0)[lb ] ( 2968,284)(0,0)[lb ] ( 3934,311)(0,0)[lb ] ( 3964,-544)(0,0)[lb ] ( 2965,-526)(0,0)[lb ] ( 2197,-817)(0,0)[lb ] [ spider lemma][spider ] ( a ) let @xmath64 be a weighted graph containing the subgraph @xmath85 shown on the left in figure [ spider1 ] ( the labels indicate weights , unlabeled edges have weight 1 ) . suppose in addition that the four inner black vertices in the subgraph @xmath85 , different from @xmath86 , have no neighbors outside @xmath85 . let @xmath69 be the graph obtained from @xmath64 by replacing @xmath85 by the graph @xmath87 shown on right in figure [ spider1 ] , where the dashed lines indicate new edges , weighted as shown . then @xmath88 . \(b ) consider the above local replacement operation when @xmath85 and @xmath87 are graphs shown in figure [ spider2](a ) with the indicated weights ( in particular , @xmath89 has a new vertex @xmath31 , that is incident only to @xmath25 and @xmath27 ) . then @xmath90 . \(c ) the statement of part ( b ) is also true when @xmath85 and @xmath87 are the graphs indicated in figure [ spider2](b ) ( in this case @xmath69 has two new vertices @xmath27 and @xmath31 , they are adjacent only to one another and to @xmath26 and @xmath25 , respectively ) . , ( b ) @xmath91 , ( c ) @xmath92 , and ( d ) @xmath93,width=340 ] the following four families of graphs will play a special role in our proofs . consider a @xmath94 rectangular chessboard @xmath26 and suppose the corners are black . aztec rectangle _ @xmath95 is the graph whose vertices are the white squares and whose edges connect precisely those pairs of white squares that are diagonally adjacent ( the case @xmath96 , @xmath97 is shown in figure [ arvariants](a ) ) . if one removes the bottom row of the board @xmath26 and then applies the same procedure , the resulting graph is denoted by @xmath98 , and called a _ baseless aztec rectangle _ ( see figure [ arvariants](c ) for an illustration ) the _ combed aztec diamond _ @xmath99 is the graph obtained from @xmath95 by appending a vertical edge to each of its bottom vertices ( an example is shown in figure [ arvariants](b ) ) . the _ combed baseless aztec rectangle _ @xmath100 is the graph obtained from @xmath98 by appending a vertical edge to each of its bottom vertices ( see figure [ arvariants](d ) for an example ) . the _ connected sum _ @xmath101 of two disjoint graphs @xmath64 and @xmath69 along the ordered sets of vertices @xmath102 and @xmath103 is the graph obtained from @xmath64 and @xmath69 by identifying vertices @xmath104 and @xmath105 , for @xmath106 . [ t1 ] let @xmath64 be a graph and let @xmath107 be an ordered subset of its vertices . then @xmath108 where the connected sum acts on @xmath64 along @xmath107 , and on @xmath109 and @xmath110 along their bottom @xmath111 vertices ( ordered from left to right ) . for @xmath112 , the above transformation is illustrated in figures [ lemmat1](a ) and ( e ) . let @xmath113 be the graph obtained from @xmath114 by applying the vertex - splitting lemma at all vertices of @xmath115 that are not @xmath116 and not adjacent to any of them ( figures [ lemmat1](a ) and ( b ) illustrate this for the case + @xmath117 and @xmath118 ) . apply the spider lemma around all @xmath119 shaded diamond in the graph @xmath113 , and remove the @xmath120 edges adjacent to a vertex of degree 1 , which are forced edges ( see figure [ lemmat1](c ) , the forced edges are the circled ones ) . the resulting graph is isomorphic to @xmath121 , where @xmath122 is the graph obtained from @xmath123 by changing all the weights of edges to @xmath124 ( see figure [ lemmat1](d ) ; dotted edges have weight @xmath124 ) . by lemmas [ vs ] and [ spider ] , we have @xmath125 applying the star lemma with factor @xmath126 to all @xmath127 shaded vertices of the graph @xmath122 ( see figure [ lemmat1](d ) ) , the graph @xmath121 is turned into @xmath128 . by ( [ t1eq2 ] ) and lemma [ star ] , we obtain @xmath129 which proves ( [ t1eq1 ] ) . [ t2 ] let @xmath64 be a graph and let @xmath130 be an ordered subset of its vertices . then @xmath131 where the connected sum acts on @xmath64 along @xmath130 , and on @xmath132 and @xmath133 along their bottom @xmath134 vertices ordered from left to right ( see figures [ lemmat2](a ) and ( e ) , for @xmath135 ) . one can prove lemma [ t2 ] similarly to lemma [ t1 ] by using vertex - splitting lemma , spider lemma and star lemma to transform the graph on the left hand side of ( [ t2eq1 ] ) into the graph on the right hand side . this transforming process is illustrated by figure [ lemmat2 ] , for @xmath135 . an _ induced subgraph _ of a graph @xmath64 is a graph whose vertex set is a subset @xmath136 of the vertex set of @xmath64 , and whose edge set consists of all edges of @xmath64 with endpoints in @xmath136 . for a graph @xmath64 , we denote by @xmath137 its vertex set . [ graphsplitting ] let @xmath64 be a bipartite graph , and let @xmath138 and @xmath139 be the two vertex classes . let @xmath140 be an induced subgraph of @xmath64 . \(a ) assume that @xmath140 satisfies the following two conditions . 1 . the _ separating condition _ : there are no edges of @xmath64 connecting a vertex in @xmath141 and a vertex in @xmath142 , 2 . the _ balancing condition _ : @xmath143 . then @xmath144 \(b ) if @xmath140 satisfies the separating condition and but @xmath145 , then @xmath146 . color all vertices of @xmath138 white , and all vertices of @xmath139 black . to prove part ( a ) of the lemma , it suffices to show that that none of the perfect matchings of @xmath64 contain any edge connecting a vertex of @xmath140 to a vertex of @xmath147 . since there are no edges of @xmath64 connecting a white vertex of @xmath140 to a vertex of @xmath147 ( by the separating condition ) , the previous assertion follows by the balancing condition . part ( b ) is perfectly analogous , the only difference is that @xmath140 has no tilings when @xmath148 the proof of theorem 2.3 and as we will see in the next section , also the proof of theorem 2.2 is based on a common generalization of the transformations @xmath149 and @xmath150 of lemmas [ t1 ] and [ t2 ] , presented below in proposition [ composite ] . the diagonals divide @xmath57 into @xmath18 parts called _ layers_. recall that @xmath40 and @xmath9 are the numbers of rows of triangular black cells of @xmath57 that point up and down , respectively . as one moves down from top , each row of black up - pointing triangles causes the length of the rows of black cells in the next layer to go up by one , and each row of black down - pointing triangles causes the length of the rows of black cells in that layer to be one less than in the previous layer . this implies that the bottom row of cells of @xmath57 consists of @xmath151 cells ( note that these are always triangular and up - pointing , but they may be either black or white ) . the _ width of a layer _ is one less than the length of its rows of blacks cells . let @xmath152 be the dual graph of the region @xmath57 . , ( b ) @xmath153 , ( c ) @xmath154 , and ( d ) @xmath155.,width=453 ] [ composite ] let @xmath1 and @xmath19 be positive integers . recall that @xmath27 is the number of black cells of @xmath57 that are either squares or up - pointing triangles , and @xmath37 is the number of rows of black cells consisting of squares or up - pointing triangles . let @xmath64 be a graph , and consider an ordered subset of its vertices @xmath156 . \(a ) if the bottom row of cells in @xmath57 is white , then @xmath157 where the connected sum acts on @xmath64 along @xmath156 , and on the other two summands along their bottom @xmath151 vertices ( ordered from left to right ) . see figures [ mass](c ) and ( d ) for an example , the white circles indicate the vertices @xmath158 @xmath159 . \(b ) if the bottom row of cells in @xmath57 is black , then @xmath160 where the connected sum acts on @xmath64 along @xmath156 , and on the other two summands along their bottom @xmath151 vertices ( ordered from left to right ) . see figures [ mass](a ) and ( b ) for an example . and ( b ) @xmath161 ; and their dual graphs ( c ) @xmath162 and ( d ) @xmath163.,width=453 ] \(a ) we prove ( [ composeq1 ] ) by induction on @xmath18 . for @xmath164 , the dual graph @xmath152 of the region is just the aztec rectangle @xmath165 , and @xmath166 . in particular , the number @xmath27 of black cells is equal to @xmath167 , and ( [ composeq1 ] ) holds . for the induction step , suppose @xmath168 and assume ( [ composeq1 ] ) holds for all regions with strictly less than @xmath18 _ layers _ ( the layers of @xmath152 correspond to the layers of @xmath57 , which are the portions of the latter between consecutive diagonals ) . note that the top layer of @xmath152 is either an aztec rectangle or a baseless aztec rectangle , based on whether @xmath169 is even or odd , respectively . we treat the two cases separately . suppose that the top layer of @xmath152 is the aztec rectangle @xmath170 , i.e. @xmath169 is even and @xmath171 . applying the transformation @xmath149 of lemma [ t1 ] with respect to this layer , we obtain @xmath172 where @xmath85 is the graph obtained from @xmath152 by replacing its top part , isomorphic to the combed aztec rectangle @xmath173 , by the baseless aztec rectangle @xmath174 . note that the graph @xmath85 is also the dual of a generalized douglas region @xmath175 with white bottom row of cells , but the number of layers of @xmath85 is @xmath176 ( see figures [ compare1](c ) and ( d ) for an example ; the subgraph above the dotted line in graph ( c ) is replaced by the subgraph above that line in graph ( d ) ) . in precise , @xmath177 . note that @xmath178 is the width of the second layer in @xmath179 , so @xmath180 . therefore , by the induction hypothesis , we have @xmath181 where the primed symbols refer to the region @xmath175 and denote quantities corresponding to their unprimed counterparts in @xmath57 . it is clear by definition of @xmath175 that @xmath182 . two regions @xmath57 and @xmath175 are different only in their top parts , and their bottom parts are the same ( see figures [ compare1](a ) and ( b ) ; the parts below the dotted line in two regions are the same ) . the number of rows of black up - pointing triangles is the same in @xmath57 and @xmath175 , but the latter has one fewer rows of black down - pointing triangles ( because the former region has such a row just below its top layer , and that is no longer present in @xmath175 ) . thus @xmath183 , and @xmath184 . moreover , the numbers of rows of black squares in @xmath185 and @xmath175 are equal , so we get @xmath186 . finally , one easily sees that @xmath187 . therefore , we obtain @xmath188 then , by ( [ composeq3])([composeq5 ] ) , we have @xmath189 which completes the induction step in this case . and ( b ) @xmath190 ; and their dual graphs ( c ) @xmath163 and ( d ) @xmath191.,width=453 ] suppose now that the top layer of @xmath152 is isomorphic to the baseless aztec rectangle @xmath192 , i.e. @xmath169 is odd and @xmath193 . applying this time the transformation @xmath150 of lemma [ t2 ] with respect to this layer , we obtain that @xmath194 where @xmath85 is now the graph obtained from @xmath152 by replacing its top part , isomorphic to the combed baseless aztec rectangle @xmath195 , by the aztec rectangle @xmath196 ( see figures [ compare2](c ) and ( d ) for an example ) . again , @xmath85 is the dual of a generalized douglas region @xmath197 of the same kind as @xmath57 ( i.e. , with a white bottom row of cells ) , but with fewer layers . therefore , by the induction hypothesis we have again the equality ( [ composeq4 ] ) . again , it is clear that @xmath198 , and two regions @xmath57 and @xmath175 are different only in their top layers ( see figures [ compare2](a ) and ( b ) for an example ) . one can verify that @xmath199 and @xmath200 in this case . moreover , the number of rows of black squares is now one fewer in @xmath185 than in @xmath175 , thus we still have @xmath186 . finally , one readily sees that @xmath201 equals now @xmath202 . this yields @xmath203 which in turn , by ( [ composeq7 ] ) , ( [ composeq4 ] ) and ( [ composeq9 ] ) , implies @xmath204 this completes the induction step in the second case , and thus also the proof of ( [ composeq1 ] ) . part ( b ) is proved by a perfectly analogous argument . the enumeration of tilings of the region @xmath57 follows easily using the above result . apply proposition [ composite ] with @xmath64 chosen to be the graph consisting of @xmath151 vertices and no edges . if the bottom row of cells in @xmath57 is black , then ( [ composeq2 ] ) implies that @xmath205 since a baseless aztec rectangle has no perfect matchings ( because it is impossible to cover all its bottom vertices by disjoint edges ) , it follows that in this case @xmath206 . assume therefore that the bottom row of cells in @xmath57 is white . then with @xmath64 chosen like above , ( [ composeq1 ] ) implies that @xmath207 it follows thus that unless @xmath208 , one again has @xmath209 this proves part ( a ) of the theorem . part ( b ) follows directly from ( [ dougeq2 ] ) and the aztec diamond theorem ( [ diamond ] ) . our proof uses the composite transformations of proposition [ composite ] to transform the dual graph of a generalized quasi - hexagon into a honeycomb graph whose perfect matchings are enumerated by macmahon s formula ( [ mcmahon ] ) . ( 0,0 ) ( left ) and @xmath210 ( right),title="fig : " ] # 1#2#3#4#5 @font ( 5994,3441)(913,-4120 ) ( 928,-1906)(0,0)[lb ] ( 1564,-3448)(0,0)[lb ] ( 3229,-4105)(0,0)[lb ] ( 3889,-3460)(0,0)[lb ] ( 3142,-1831)(0,0)[lb ] ( 1762,-1183)(0,0)[lb ] ( 3913,-1894)(0,0)[lb ] ( 4660,-3088)(0,0)[lb ] ( 6037,-3634)(0,0)[lb ] ( 6742,-2923)(0,0)[lb ] ( 6097,-1642)(0,0)[lb ] ( 4843,-889)(0,0)[lb ] and ( b ) @xmath211.,width=302 ] [ proof of theorem [ main ] . ] first , we consider the case where the bottom row of cells is white . then we have the same situation when showing that a baseless aztec rectangle does not have a perfect matching . it is also impossible to cover the bottom vertices in the dual graph by disjoint edges . therefore the dual graph has no perfect matchings , i.e. the region has no tilings . hereafter , we assume that the bottom row of cells of our region is black . there are two cases to distinguish , depending on whether the row of triangular cells just above @xmath17 and with bases resting on @xmath17 is black or white ( these two cases are illustrated in figure [ conhex ] ; the corresponding dual graphs are shown in figure [ conhex2 ] ) . suppose the row of cells just above @xmath17 consists of black triangles ( note that this is the case for our motivating problem ; see figure [ hexagonp ] ) . denote by @xmath212 the dual graph of the generalized quasi - hexagon @xmath213 apply the composite transformation in proposition [ composite](b ) separately to the portions of @xmath214 corresponding to the parts above and below @xmath17 of the region @xmath140 ( which we will call the _ upper _ and _ lower parts _ of the dual graph ; illustrated in figure [ conhex2 ] as the parts above and below the dotted lines ) . we obtain that @xmath215 where the connected sum acts along the bottom vertices of the two modified aztec rectangles ( ordered from left to right ) , and @xmath216 , @xmath217 , @xmath218 and @xmath219 are quantities having the same significance for the lower part of @xmath214 as their unprimed correspondents have for its upper part ( except that due to the geometry of @xmath32 and the resulting coloring of its fundamental regions they count _ white _ cells ) . the transformation involved in ( [ maineq1 ] ) is illustrated in figures [ symmetricfig](a ) and ( b ) . as we have seen in the second paragraph of section 4 , @xmath151 is the number of triangular cells above @xmath17 with bases resting on @xmath17 . by the same argument , @xmath220 is the number of triangular cells below @xmath17 , with bases resting on @xmath17 . these two numbers are equal by construction , and therefore we have @xmath221 clearly , a necessary condition for the graph on the right hand side of ( [ maineq1 ] ) ( which is a bipartite graph ) to have a perfect matching is that the numbers of its black and white vertices are the same . denote for brevity @xmath222 . then its upper part has @xmath223 white and @xmath224 black vertices , while its lower part has @xmath225 black and @xmath226 white vertices . the balancing condition is then @xmath227 which is equivalent to @xmath186 . this proves in particular the rest of part ( a ) of the theorem . to prove part ( b ) , we consider first the case when @xmath228 . the key idea here is to realize that the composite transformation of proposition [ composite](b ) can also be used to turn a honeycomb graph into the graph on the right hand side of ( [ maineq1 ] ) . more precisely , consider the honeycomb graph of sides @xmath229 ( clockwise from top ) , illustrated in figure [ symmetricfig](c ) . note that this is the dual graph of a generalized quasi - hexagon , in which all the inter - diagonal distances @xmath48 s and @xmath49 s are equal to one ( i.e. the lozenge hexagonal region of the same sides ) ! therefore equality ( [ maineq1 ] ) applies in this case , and the question is what is exactly its form in this particular case ( see figures [ symmetricfig ] ( c ) and ( b ) ) . by symmetry , the primed and unprimed quantities of @xmath27- , @xmath37- , @xmath1- , @xmath40- and @xmath9-type are equal . furthermore , one readily sees that the number of black regular cells above @xmath17 is @xmath230 the upper height is equal to @xmath37 and so is the @xmath40-parameter , while the @xmath9-parameter is 0 . the @xmath1-parameter equals @xmath231 . it follows that the particular form ( [ maineq1 ] ) takes is @xmath232 equations ( [ maineq1 ] ) ( in which @xmath233 by ( [ maineq2 ] ) , and @xmath234 by assumption ) and ( [ maineq3 ] ) imply the statement of part ( b ) of the theorem , provided @xmath235 . ( 0,0 ) # 1#2#3#4#5 @font ( 1917,2321)(3057,-2614 ) ( 3000,-815)(0,0)[lb ] ( 3000,-2114)(0,0)[lb ] note that if @xmath236 , then the lozenge hexagonal region of side - lengths @xmath237 is _ not _ a member of our family of quasi - hexagons anymore ( in the definition of a quasi - hexagon , we always assume that @xmath25 and @xmath28 are two distinct lattice vertices ) ; and if @xmath238 , then the lozenge hexagonal region with side - lengths above does not exist ( two side - lengths are negative ) . this means that the method used to prove ( [ maineq ] ) when @xmath228 does _ not _ work for the case when @xmath239 . to complete the proof of ( [ maineq ] ) , we need to show that the graph @xmath240 has @xmath241 perfect matchings when @xmath236 , and that it has no perfect matchings when @xmath238 . the top @xmath242 rows of its vertices induce a subgraph isomorphic to @xmath243 , and the bottom @xmath242 rows of its vertices induce also a subgraph isomorphic to @xmath243 . denote by @xmath244 and @xmath245 the two aztec rectangles ( illustrated by the subgraph above and the subgraph below two dotted lines in figure [ ben1s ] ) . if @xmath238 , then @xmath244 satisfies the conditions in graph - splitting lemma [ graphsplitting](b ) as an induced subgraph of @xmath246 , and thus @xmath247 . assume that @xmath236 , then the two aztec rectangles @xmath244 and @xmath245 become two aztec diamonds of order @xmath248 . the aztec rectangle @xmath244 satisfies the conditions in graph - splitting lemma [ graphsplitting](a ) as an induced subgraph of @xmath246 , and the rectangle @xmath249 satisfies also the conditions of this lemma as an induced subgraph of @xmath250 . therefore @xmath251 where graph @xmath252 consists of @xmath37 disjoint vertical edges ( see the circled edges in figure [ ben1s ] ) , so it has exactly one perfect matching . this finishes the proof of part ( b ) of the theorem in the case when the triangular cells above @xmath17 with bases resting on it are black . suppose now that the latter are white . apply in this case the composite transformation in proposition [ composite](a ) to the upper and lower parts of the graph @xmath253 . we obtain that @xmath254 where the connected sum acts along the bottom vertices of the two aztec rectangles ( ordered from left to right ) , and @xmath216 , @xmath217 , @xmath218 and @xmath219 are quantities having the same significance for the lower part of @xmath253 as their unprimed correspondents have for its upper part ( except that , as in the first case , they now count white cells ) . the transformation involved in ( [ maineq4 ] ) is illustrated in figures [ symmetricfig2](a ) and ( b ) . apply the vertex - splitting lemma [ vs ] to all bottom vertices of @xmath255 in the graph on the right hand side of ( [ maineq4 ] ) ( see figures [ symmetricfig2](b ) and ( c ) ) . next , apply the transformation @xmath149 of lemma [ t1 ] to the top and bottom of the resulting graph , the latter is transformed into the graph on the right hand side of ( [ maineq1 ] ) ( see figures [ symmetricfig2](c ) and ( d ) ) . then both parts ( a ) and ( b ) of the theorem are reduced to the case treated above . the generalized quasi - hexagons we treated so far had the special property that the point on the southwestern boundary where its defining property `` black on left '' changed to `` white on left '' ( i.e. vertex @xmath26 ) was on the same southwest - to - northeast lattice diagonal as the analogous point on the northeastern boundary ( i.e. vertex @xmath30 ) . therefore , the perpendicular bisector of segment @xmath256 is the symmetric axis of the quasi - hexagons . these include as a special case propp s quasi - hexagons of side - lengths @xmath1 , @xmath2 , @xmath2 , @xmath1 , @xmath2 , @xmath2 ( clockwise from top ) , and will be called from now on symmetric ( generalized ) quasi - hexagons . in order to include the general case of propp s quasi - hexagons when the side - lengths are @xmath1 , @xmath2 , @xmath3 , @xmath1 , @xmath2 , @xmath3 , we extend the definition of our generalized quasi - hexagons as follows . besides @xmath17 , we consider a second distinguished lattice diagonal @xmath257 below it , and draw in also @xmath258 additional diagonals in between them , so that the distances between successive ones ( starting from top ) are @xmath259 . as in the symmetric case , we consider @xmath18 diagonals above @xmath17 at successive distances ( starting from top ) @xmath19 , and @xmath21 diagonals below @xmath257 , at successive distances ( starting from the bottom ) @xmath22 . define our generalized quasi - hexagon as in the symmetric case , with the one change that on the northeastern boundary we make the switch from the rule `` white on left '' to `` black on left '' on the diagonal @xmath257 , rather than on @xmath17 . denote the resulting region by @xmath260 ; an illustrative example is shown in figure [ asymmetrichex ] . as mentioned in section 2 , there is no simple product formula for the number of tilings of asymmetric quasi - hexagons . however , it turns out that we can construct a conceptually simple graph so that the number of tilings of an asymmetric quasi - hexagon is equal to a certain power of @xmath12 times the number of perfect matchings of this graph . we present this construction next . roughly speaking , the graph we need is a connected sum of a baseless aztec rectangle and the bottom half of a honeycomb graph . to be precise , let @xmath1 , @xmath2 , @xmath3 be positive integers with @xmath261 , and let @xmath262 be the portion of the honeycomb graph of sides @xmath1 , @xmath263 , @xmath3 , @xmath1 , @xmath263 , @xmath3 ( clockwise from top ) that is below or on the horizontal through its center . let @xmath264 and @xmath265 be positive integers , and consider the baseless aztec rectangle @xmath266 . the latter has @xmath267 vertices on the bottom ; label them from left to right by @xmath268 . the graph @xmath262 has @xmath269 vertices at the top ; label them from left to right by @xmath270 . ( 6349,3716)(222,-3488 ) ( 1891,-2941)(0,0)[lb ] ( 2836,-2468)(0,0)[lb ] ( 592,-1996)(0,0)[lb ] ( 150,-696)(0,0)[lb ] ( 3948,-2178)(0,0)[lb ] ( 5250,-3005)(0,0)[lb ] ( 6203,-2414)(0,0)[lb ] ( 3450,-814)(0,0)[lb ] ( 1182,-224)(0,0)[lb ] ( 5080 , 12)(0,0)[lb ] on the other hand , if @xmath277 , define @xmath273 by @xmath278 where @xmath279 is the subgraph of @xmath266 obtained by deleting its vertices with labels greater than @xmath269 , and the connected sum identifies the vertices labeled @xmath275 in the two components , for @xmath280 . examples illustrating the two cases are shown in figure [ cfigure ] . for the sake of simplicity , we detail below the case when all distances between successive diagonals are odd ( recall that the original problem , problem 16 in @xcite , corresponds to the special case when all these distances are equal to 3 ) . the general case is addressed in theorem [ asymmetric1 ] . [ asymodd ] let @xmath19 , @xmath22 and @xmath281 be odd integers , and set @xmath282 let @xmath283 be an asymmetric quasi - hexagon , and let @xmath27 be the number of black cells above @xmath17 , and @xmath38 the number of white cells below @xmath257 . in order to prove this result we will need some more graph transformations , which we present in the next two lemmas . let @xmath291 be the graph obtained from the aztec rectangle @xmath95 by deleting its leftmost @xmath40 vertices , and let @xmath292 be the graph obtained from @xmath95 by deleting its rightmost @xmath40 vertices . let @xmath294 be the graph obtained from the graph @xmath291 by appending vertical edges to its top vertices , and @xmath295 the graph obtained from @xmath292 by appending vertical edges to its bottom vertices . then @xmath296 where the connected sum acts along the top and bottom vertices of @xmath294 and @xmath295 ( ordered from left to right and from top to bottom ) , and along the ordered set @xmath136 of vertices of @xmath64 . the above transformation is illustrated by figures [ lemmat3](a ) and ( e ) , for @xmath96 and @xmath297 . one can prove lemma [ t3 ] by following the argument in the proof of lemma [ t1 ] , based on figure [ lemmat3 ] . note that @xmath300 has @xmath305 vertices on the bottom , and @xmath301 has @xmath264 vertices on top . if @xmath306 , define @xmath307 where the connected sum identifies the @xmath275th vertex on the bottom in @xmath300 with the @xmath275th vertex on top in @xmath301 ( both counted from left to right ) , for @xmath308 ( the graph @xmath309 is shown in figure [ lgraph](a ) ) . on the other hand , if @xmath310 , define our graph by @xmath311 where @xmath312 is the graph obtained from @xmath313 by deleting the rightmost @xmath314 of its bottom vertices , and the connected sum identifies the @xmath275th vertex on the bottom in @xmath312 with the @xmath275th vertex on top in @xmath301 , for @xmath315 @xmath316 ( @xmath302 is pictured in figure [ lgraph](b ) ) . our second family of graphs is the result of an analogous construction , in which the role of @xmath291 is replaced by the graph @xmath317 obtained from it by deleting its top @xmath9 vertices ( i.e. , @xmath317 is obtained from @xmath318 by deleting the @xmath40 leftmost and @xmath9 topmost vertices ) . let @xmath298 be positive integers . note that @xmath319 has @xmath264 vertices on top ( just like @xmath301 ) . in analogy to ( [ lgrapheq1 ] ) , if @xmath306 , define @xmath320 where the connected sum identifies the @xmath275th vertex on the bottom in @xmath313 with the @xmath275th vertex on top in @xmath301 but now counted from _ right to left _ for @xmath308 ( the graph @xmath321 is shown in figure [ lgraph](c ) ) . finally , if @xmath310 , define @xmath322 by @xmath323 where @xmath324 is the graph obtained from @xmath313 by deleting the _ leftmost _ @xmath314 of its bottom vertices , and the connected sum identifies the @xmath275th vertex on the bottom in @xmath324 with the @xmath275th vertex on top in @xmath319 ( both counted from right to left ) , for @xmath325 ( @xmath326 is shown in figure [ lgraph](d ) ) . to state our second lemma , it will be useful to introduce the following notation . given an l - shaped graph @xmath64 from one of the two families defined above , define @xmath327 to be the subgraph obtained from @xmath64 by removing all its bottommost vertices . in the spirit of section 3 , if @xmath64 belongs to one of the above two families , or if it is obtained from a member of these families by deleting all its bottom vertices , we denote by @xmath328 ( the `` combed version '' of @xmath64 ) the graph obtained from @xmath64 by appending a vertical edge to each of its bottommost vertices . in each of the equalities ( [ lgrapheq3 ] ) , ( [ lgrapheq4 ] ) and ( [ lgrapheq5 ] ) , the connected sum acts on @xmath64 along @xmath136 , and on the other two summands along their bottommost vertices , ordered from left to right . we call the portions of an asymmetric quasi - hexagon @xmath140 above @xmath17 , below @xmath257 and between @xmath17 and @xmath257 _ the upper _ , _ the lower _ and _ the middle parts _ of the region , respectively . we also use these terms for the corresponding parts of the dual graph @xmath64 of the region . it means the upper , the lower and the middle parts of @xmath64 are the dual graphs of the upper , the lower and the middle parts of the region @xmath140 , respectively . we are now ready to present the proof of our theorem . \(a ) apply the composite transformation of proposition [ composite](b ) to the upper part of @xmath64 , and also to its lower part . this yields @xmath337 where @xmath217 is the length of the southeastern side of @xmath140 . the arguments that led to ( [ maineq2 ] ) still work , because the middle part of @xmath64 is balanced ( i.e. it has the same number of vertices in the two color classes of a proper 2-coloring ) , and they give @xmath338 furthermore , the arguments that gave ( [ maineq2 ] ) also work , because of the same reason ( the middle part of @xmath64 is balanced ) . it follows that @xmath64 is balanced if and only if @xmath186 , which proves part ( a ) . consider first the first equality in ( [ asymoddeq ] ) , provided @xmath341 . in this case @xmath342 , for @xmath343 . to prove this we use the same key idea in the proof of theorem [ main ] as follows . the lozenge hexagon @xmath344 can be viewed as an asymmetric quasi - hexagon , in which all the inter - diagonal distances @xmath48 s , @xmath49 s and @xmath345 s are equal to one . by the same argument in the equality ( [ maineq3 ] ) , we obtain @xmath346 therefore , the first equality in ( [ asymoddeq ] ) follows from ( [ asyproofeq3 ] ) and ( [ asyproofeq3 ] ) . one readily checks that , for @xmath349 , the expression @xmath350 in the exponent of 2 in the second equality of ( [ asymoddeq ] ) becomes precisely @xmath351 , and thus ( [ asyproofeq4 ] ) follows from the case @xmath352 of the second equality in ( [ asymoddeq ] ) . * stage 1 . * _ reduction to the case @xmath354 . _ if @xmath355 ( and @xmath356 ) , apply the transformation @xmath357 of lemma [ t3 ] to the graph @xmath64 , regarded as being obtained as a connected sum of type ( [ t3eq1 ] ) , in which the second summand is the portion of @xmath64 that corresponds to the portion of @xmath140 in between @xmath257 and the first diagonal above it ( see figure [ middle7 ] ; we replace the subgraph between the two dotted lines in a graph by the subgraph between those lines in the next graph ) . one readily sees that the resulting graph is isomorphic to the graph @xmath358 by lemma [ t3 ] , we have @xmath359 . if @xmath360 , then @xmath361 and the latter equality is obviously true . repeating this argument , we get that @xmath362 for @xmath363 , where @xmath364 thus @xmath365 * stage 2 . * _ transforming @xmath64 into the connected sum of an aztec rectangle with missing vertices and a half - honeycomb . _ apply the transformation @xmath150 of lemma [ t2 ] to the part of @xmath366 above @xmath17 , and then apply ( in reverse ) the composite transformation of proposition [ composite](b ) to replace the part of @xmath64 below @xmath257 by a half - honeycomb ( these transformations are illustrated in figure [ stage2 ] ) . denote the resulting graph by @xmath367 . the quoted results imply that @xmath368 note that @xmath367 is not far from being a @xmath369-type graph the only difference is in how its upper part relates to an aztec rectangle . this difference is eliminated in the remaining two stages . * stage 3 . * _ bringing the missing vertices of the aztec rectangle to the top left . _ this stage is illustrated in figures [ middle8](a)(d ) ; we replace the subgraph above the dotted line in a graph by the subgraph above this line in the next graph . note that we have the graph isomorphism @xmath370 ( recall that we are assuming @xmath352 ) . apply lemma [ ltransform](b ) to the graph @xmath367 regarded as a connected sum as shown on the right hand side of ( [ asyproofeq7 ] ) . we obtain that @xmath371 where @xmath372 shown in figures [ middle8](a ) and ( b ) . * stage 4 . * _ moving the missing vertices of the aztec rectangle to the bottom right . _ this stage is illustrated in figures [ midnew3](a)(e ) . apply lemma [ ltransform](c ) to the graph @xmath378 . we obtain that @xmath379 where @xmath380 ( see figures [ midnew3](a ) and ( b ) ) . note that for the last of these graphs we have @xmath387 the second equality in ( [ asymoddeq ] ) follows now by ( [ asyproofeq5 ] ) , ( [ asyproofeq6 ] ) , ( [ asyproofeq10 ] ) , ( [ asyproofeq12 ] ) , ( [ asyproofeq14 ] ) and ( [ asyproofeq16 ] ) . denote by @xmath388 the dual graph of @xmath349 . the top @xmath242 rows of the vertices of @xmath388 induce a subgraph isomorphic to @xmath243 , and the bottom @xmath242 rows of its vertices induce also a subgraph isomorphic to @xmath243 . denote by @xmath244 and @xmath245 the two aztec rectangles ( shown by the subgraph above and the subgraph below two dotted lines in figure [ ben2s ] ) . if @xmath288 , then @xmath244 satisfies the conditions in graph - splitting lemma [ graphsplitting](b ) as an induced subgraph of @xmath388 , so @xmath389 . assume that @xmath289 , then two aztec rectangles @xmath244 and @xmath245 are now two aztec diamonds of order @xmath390 . they satisfy the conditions in the part ( a ) of the graph - splitting lemma [ graphsplitting ] as two induced subgraphs of @xmath388 and @xmath391 , respectively . we can check that the graph @xmath393 has a unique perfect matching ( it consists of all vertical edges in the middle part of @xmath388 when @xmath341 , and its pattern is similar to the set of circled edges in figure [ ben2s ] when @xmath347 ) . therefore , we get @xmath394 from ( [ power2b ] ) . this completes the proof for part ( c ) . the main reason we chose to relate quasi - hexagons to the graphs @xmath273 as in theorem [ asymodd ] is that the number of perfect matchings of the latter can be written as a sum of quantities expressed by simple product formulas . this is presented in corollary [ sumformulas ] below , whose proof employs the following two lemmas . [ aztecdent ] label the bottom vertices of the baseless aztec rectangle @xmath98 from left to right by @xmath395 , and denote by @xmath64 the graph obtained from it by deleting the vertices with labels in the set @xmath396 , where @xmath397 are given integers . then we have @xmath398 [ hexdent ] consider the dual of the bottom half of a lozenge hexagon of side - lengths @xmath2 , @xmath1 , @xmath1 , @xmath2 , @xmath1 , @xmath1 ( clockwise from top ) on the triangular lattice . label its topmost vertices from left to right by @xmath399 , and the number of perfect matchings of the graph obtained from it by removing the vertices with labels in the set @xmath400 is equal to @xmath401 where @xmath402 are given integers . @xmath34b@xmath35 . if @xmath407 , then @xmath408 where the sum is taken over all pairs of disjoint sets @xmath25 and @xmath26 whose union is @xmath409 and whose cardinalities are given by @xmath410 and @xmath411 . @xmath34c@xmath35 . if @xmath412 , then @xmath413 where the sum is taken over all pairs of disjoint sets @xmath25 and @xmath26 whose union is @xmath414 and whose cardinalities are given by @xmath410 and @xmath415 . consider the vertices of @xmath273 that belong both to the baseless aztec rectangle part and to the half - honeycomb part ( these are the vertices on the dotted line in figure [ cfigure ] ) . in any given perfect matching , some of them are matched upward , and the remaining ones are matched downward . let @xmath18 be the number of vertices belonging to both parts , and partition the set of perfect matchings of @xmath273 into @xmath416 classes corresponding to all the possible choices for each of these vertices to be matched upward or downward . each class is then the set of perfect matchings of a disjoint union of two graphs , the top one being of the kind in lemma [ aztecdent ] , and the bottom one ( after adding some vertical forced edges that are shown as circled ones in figure [ cfigure ] ) being of the kind in lemma [ hexdent ] . part ( a ) follows from the requirement that these ( bipartite ) graphs are balanced , while parts ( b ) and ( c ) follow from lemmas [ aztecdent ] and [ hexdent ] . we end this section by presenting the general version of theorem [ asymodd ] , corresponding to the case when the distances between successive diagonals are arbitrary ( and not necessarily odd ) . to state the general result we need some more definitions . we define the upper and the lower heights of the region to be the number of rows of regular black cells above @xmath17 and the number of rows of regular white cells below @xmath257 , respectively . the diagonals divide the middle part of an asymmetric quasi - hexagon into parts , called _ middle layers_. in the case of odd distances ( as in theorem [ asymodd ] ) , the bottom row of cells is black . however , we face the same problem in investigating symmetric quasi - hexagons : the bottom row of cells may be white . by the same argument that we used in the proof of theorem [ main](a ) , when the latter happens the region has no tilings . thus , we assume from now on that the bottom row of cells is black . define the _ slope _ of the asymmetric quasi - hexagon by @xmath420 set @xmath421 since the middle part is balanced , one can verify that the numbers of black and white vertices in the dual graph of @xmath140 are equal if and only if the upper and lower heights are equal . the same arguments that we used to prove theorem [ asymodd ] can be used to obtain the following result . [ asymmetric1 ] let @xmath283 be an asymmetric quasi - hexagon . denote by @xmath27 the number of regular black cells above @xmath17 , and by @xmath38 the number of regular white cells below @xmath257 , and assume the upper and lower heights of @xmath140 are both equal to @xmath37 . let @xmath40 be the number of rows of regular black triangular cells above @xmath17 , and @xmath9 be the number of rows of irregular black triangular cells above @xmath17 . then we have @xmath422 where @xmath423 .	we solve and generalize an open problem posted by james propp ( problem 16 in _ new perspectives in geometric combinatorics _ , cambridge university press , 1999 ) on the number of tilings of quasi - hexagonal regions on the square lattice with every third diagonal drawn in . we also obtain a generalization of douglas theorem on the number of tilings of a family of regions of the square lattice with every second diagonal drawn in .
consider the similarity transformations @xmath5 given by @xmath6 where @xmath7 . because these maps are contractions , there is a nonempty compact attractor @xmath8 associated with the _ iterated function system _ ( or ifs ) given by the pair @xmath9 . the attractor @xmath10 coincides with the set of accumulation points of the forward orbit of any @xmath11 under the semigroup @xmath12 . in this article , we study the topology of certain subsets of the parameter space @xmath13 for @xmath14 . the first set we consider is the _ connectedness locus _ , denoted by @xmath0 ; that is , the set of parameters @xmath1 for which @xmath10 is connected . standard ifs arguments prove that the limit set @xmath10 is either connected , or it is a cantor set ( for details , see lemma [ lemma : disconnected_cantor ] ) . the second subset of the parameter space we examine is related to the geometry of @xmath10 . for all values of the parameter @xmath15 , the map @xmath16 fixes @xmath17 , and the map @xmath18 fixes @xmath19 . as both of these maps are contracting by the same factor ( in fact , by a factor of @xmath1 ) around their respective fixed points , the limit set @xmath10 has a center of symmetry about the point @xmath20 in the dynamical plane . the set @xmath21 is defined to be the set of parameters @xmath1 for which @xmath10 contains the point @xmath20 . the sets @xmath0 and @xmath21 have been studied by various mathematicians over the past 30 years : barnsley - harrington @xcite , bousch @xcite , bandt @xcite , solomyak @xcite , shmerkin - solomyak @xcite , and solomyak - xu @xcite , to name a few . there is a profound and unexpected connection between the sets @xmath0 and @xmath21 and the set of roots of power series with prescribed coefficients ( see section [ section : regular ] ) . in particular , @xmath0 can be identified with the closure of the set of roots of polynomials with coefficients in @xmath22 ( which are in @xmath23 , and @xmath21 can be identified with the closure of the set of roots of polynomials with coefficients in @xmath24 ( which are in @xmath13 ) . via this formulation , the set @xmath21 is related to roots of the minimal polynomials associated to the _ core entropy _ of real quadratic polynomials as defined by thurston @xcite , and established by tiozzo @xcite . we further elaborate on the history of @xmath0 and @xmath21 in section [ subsection : history ] . in @xcite and @xcite , bousch proved that the sets @xmath0 and @xmath21 are connected and locally connected . however , the complement of @xmath0 and the complement of @xmath21 are _ disconnected_. the complement of @xmath0 and the complement of @xmath21 both contain a prominent central component ( see figure [ mandelbrot ] and figure [ set_b ] ) . in 1985 , barnsley and harrington numerically observed other connected components of the complement , or `` holes '' in @xmath0 , and they conjectured that these holes are genuine . in 2002 , bandt rigorously established the existence of one hole in @xmath0 . in theorem [ theorem : hole_limit ] , we prove that there are _ infinitely many _ holes in @xmath0 . these `` exotic holes '' in @xmath0 are quite interesting and somewhat mysterious ; they appear to be very well - organized in parameter space , suggesting that there may be a combinatorial classification of them . we currently have found no such classification . we prove that all of the connected components of @xmath25 are _ schottky _ , in the sense that if @xmath1 in @xmath25 , there is a topological disk @xmath26 containing @xmath10 , so that @xmath27 , and @xmath28 and @xmath29 are contained in the interior of @xmath26 . the semigroup @xmath14 has disconnected @xmath30 if and only if @xmath14 is schottky . to prove that these exotic components in the complement of @xmath0 exist , we introduce the method of _ traps _ ( see section [ subsection : traps_holes ] ) , which allows us to numerically certify that a parameter @xmath31 . this technique is different from bandt s proof of the existence of these exotic holes . in fact , the existence of a trap is an open condition , so if there is a trap for the parameter @xmath15 , then necessarily @xmath32 . traps therefore allow us to access the interior points of @xmath0 . in @xcite , bandt conjectured that the interior of @xmath0 is dense away from @xmath33 ( see conjecture [ conjecture : bandt ] ) . in theorem [ theorem : interior_dense ] , we prove bandt s conjecture using traps . the interior of @xmath0 is dense away from the real axis ; that is , @xmath34 interestingly , the proof of theorem [ theorem : interior_dense ] requires a complete characterization of the set of @xmath35 for which the limit set @xmath10 is convex . this is established in lemma [ lemma : convex_zonohedra ] . in section [ section : renormalization ] , we examine families of exotic holes in @xmath0 which appear to spiral down and limit on a distinguished point @xmath36 ( see figure [ figure : hexahole_spiral ] ) . let @xmath37 be the root of the polynomial @xmath38 with the given approximate value . then 1 . @xmath39 is in @xmath0 , and @xmath21 ; in fact , the intersection of @xmath40 and @xmath41 is exactly the point @xmath20 ; 2 . there are points in the complement of @xmath0 arbitrarily close to @xmath39 ; and 3 . there are infinitely many rings of concentric loops in the interior of @xmath0 which nest down to the point @xmath39 . thus , @xmath0 contains infinitely many holes which accumulate at the point @xmath39 . we continue section [ section : renormalization ] by generalizing the methods of theorem [ theorem : hole_limit ] . we define the notion of _ renormalization _ and _ limiting traps _ to show that at certain renormalization points @xmath31 , the set @xmath0 is asymptotically similar to @xmath42 , where @xmath42 is the limit set of the 3 generator ifs @xmath43 previous results of solomyak established this asymptotic similarity at certain ` landmark points ' in @xmath44 . we reprove his results with a more algorithmic approach using traps , and as a consequence , we obtain `` asymptotic interior . '' suppose that @xmath39 is a renormalization point . there are constants @xmath45 and @xmath46 , depending only on @xmath39 , such that 1 . if @xmath47 , then for all @xmath48 , there is a @xmath49 such that @xmath50 and for all sufficiently large @xmath51 , there is a trap for @xmath52 . if @xmath53 is a single point , then there is @xmath54 such that for all @xmath55 with @xmath56 , the limit set for the parameter @xmath57 is disconnected for all sufficiently large @xmath51 . in section [ section : holes_in_m0 ] , we prove that the complement of @xmath21 is also disconnected by numerically certifying a loop in @xmath21 which bounds a component of the complement . there is a hole in @xmath21 . in section [ semigroups of similarities ] , we establish key definitions and survey some previous results about @xmath0 and @xmath21 . in section [ elementary_estimate ] , we collect a few elementary estimates about the geometry of @xmath10 . in section [ section : regular ] , we explore the connection the sets @xmath0 and @xmath21 have with roots of power series with prescribed coefficients in a more general context involving regular languages . in section [ section : topology_of_lambda ] , we establish some important results about the topology and geometry of the limit set , and we prove theorem [ theorem : disconnected_is_schottky ] . we also present an algorithm ( similar to an algorithm used by bandt in @xcite ) to certify that the limit set @xmath10 is disconnected . in section [ section : differences ] , we examine the set of differences between points in @xmath10 . this set of differences is actually the limit set @xmath42 of the 3 generator ifs @xmath43 in section [ section : interior ] , we introduce the notion of traps , and characterize the set of @xmath35 for which @xmath10 is convex in lemma [ lemma : convex_zonohedra ] . in theorem [ theorem : interior_dense ] , we prove that the interior of @xmath0 is dense away from the real axis , establishing bandt s conjecture [ conjecture : bandt ] . in section [ section : holes ] , we describe our trap - finding algorithm and prove the estimates required to certify that @xmath0 has holes . in section [ section : renormalization ] , we introduce the notions of renormalization and limiting traps , and we prove theorem [ theorem : hole_limit ] and theorem [ theorem : similarity ] . in section [ section : whiskers ] , we discuss the `` real whiskers '' of @xmath0 , and we use a 2-dimensional real ifs for this analysis . and lastly , in section [ section : holes_in_m0 ] , we prove that there is a hole in @xmath21 ; that is , we prove that the complement of @xmath21 is disconnected . we would like to thank christoph bandt , emmanuel breuillard , giulio tiozzo and especially boris solomyak for comments , corrections , pointers to references , and enthusiasm and interest in this project . danny calegari was supported by nsf grant dms 1405466 . sarah koch was supported by nsf grant dms 1300315 and a sloan research fellowship . alden walker was supported by nsf grant dms 1203888 . a _ contracting similarity _ ( or just a _ similarity _ ) with _ center _ @xmath58 and _ dilation _ @xmath59 with @xmath2 is the complex affine map @xmath60 given by @xmath61 the composition of any positive number of similarities is again a similarity . the set of all similarities is topologized as @xmath62 . we are concerned in the sequel with semigroups generated by finitely many similarities . let @xmath63 be a finitely generated semigroup of contracting similarities . limit set _ @xmath64 ( also called the _ attractor _ ) is the closure of the set of fixed points of elements of @xmath63 . the limit set of @xmath63 is the unique compact , nonempty invariant subset of @xmath65 for the action of @xmath63 . in particular the action of @xmath63 on @xmath64 is minimal ( every orbit is dense ) . [ example : middle_third ] the semigroup @xmath66 , @xmath67 has the middle third cantor set as limit set . [ example : sierpinski_carpet ] the semigroup @xmath68 , @xmath69 , @xmath70 for @xmath71 has the sierpinski triangle as limit set . [ definition : schottky ] let @xmath72 be a finite set of contracting similarities , and let @xmath63 be the semigroup they generate . we say that @xmath63 is a _ schottky semigroup _ if there is an embedded loop @xmath73 bounding a closed ( topological ) disk @xmath26 , so that the elements of @xmath72 take @xmath26 to disjoint disks contained in the interior of @xmath26 . a loop @xmath74 with this property , and the disk @xmath26 it bounds is said to be _ good _ for @xmath63 . the schottky semigroup @xmath63 is free ( on @xmath72 ) and discrete as a subset of @xmath62 . actually , _ every _ finitely generated semigroup which is strictly contracting is discrete , since the set of dilations accumulates only at @xmath17 ; so the point is to prove freeness . this follows from klein s ping - pong argument applied to a good disk @xmath26 and its translates . note that if @xmath72 generates a schottky semigroup , the centers of generators are distinct . indeed , a good disk @xmath26 must contain all of the centers , and since the generators map @xmath26 to disjoint disks , the centers must be distinct . thus for a schottky semigroup @xmath63 , the limit set is a cantor set , which is the intersection of the images of a good disk @xmath26 under elements of @xmath63 , and which can be identified ( topologically ) with the set of right - infinite words in the generators . thus , any two schottky semigroups with the same number of generators have topologically conjugate actions on their limit sets . in fact , we can say more : [ lemma : conjugacy ] any two isomorphic schottky semigroups @xmath63 , @xmath75 are topologically conjugate on their restriction to good disks @xmath26 , @xmath76 . if @xmath72 and @xmath77 are the generators of @xmath63 and @xmath75 , then choose any homeomorphism @xmath78 which extends a conjugacy on their boundaries , and extend to @xmath79 using @xmath80 and @xmath81 . then extend to @xmath82 by the canonical ( abstract ) isomorphism @xmath83 coming from the identification of these limit sets with the right - infinite words in the generators . note that schottky semigroups @xmath63 , @xmath75 are very rarely topologically conjugate on all of @xmath65 ; for , they are invertible on @xmath65 , and therefore any conjugacy would extend to a conjugacy between the _ groups _ they generate . but these are indiscrete , and indiscrete subgroups of @xmath84 are rarely topologically conjugate . for the remainder of the paper we focus almost entirely on semigroups generated by a pair of similarities with the same dilation @xmath1 . after conjugation by a similarity of @xmath65 we may assume that the two centers of the generators are at @xmath17 and @xmath19 respectively . thus the space of conjugacy classes of such semigroups is parameterized by @xmath85 . for @xmath85 , let @xmath86 denote the semigroup with generators @xmath87 and let @xmath10 denote the limit set of @xmath86 . we omit the subscript @xmath1 from @xmath16 and @xmath18 to lighten notation . for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] for various parameters . in each case , we show the decomposition of @xmath10 as the union of @xmath88 ( blue ) and @xmath89 ( orange ) . the points @xmath17 , @xmath20 and @xmath19 are marked in red . along the bottom from left to right , the parameters lie in @xmath90 , @xmath91 , and @xmath92 , respectively.,title="fig : " ] other normalizations have some nice features . barnsley and harrington @xcite , bousch @xcite and others use the normalization @xmath93 and solomyak @xcite uses @xmath94 our normalization has the convenient property that @xmath17 and @xmath19 are always in @xmath64 as the centers of the two generators , independent of @xmath1 . complex conjugation `` conjugates '' @xmath86 to @xmath95 . thus @xmath10 and @xmath96 are mirror images of each other . in particular , they are homeomorphic , and are therefore connected , simply connected etc . for the same values of @xmath1 . the semigroup @xmath86 has another basic symmetry : rotation through @xmath97 about the point @xmath20 interchanges the two generators @xmath16 and @xmath18 . thus the limit set @xmath10 is invariant under this symmetry : @xmath98 . on the other hand , by definition @xmath99 using the relation @xmath98 we obtain the identity @xmath100 which is the limit set of the semigroup @xmath101 with generators @xmath102 thus , although @xmath86 and @xmath101 are _ not conjugate _ ( not even topologically , and in general not even when restricted to @xmath10 ) , they have the same limit set . now , from the definition , the limit sets of @xmath101 and @xmath103 are similar . it follows that the same is true for @xmath86 and @xmath104 . we record this observation as a lemma : [ lemma : similar_limit_sets ] the limit sets @xmath105 and @xmath106 are similar or mirror images of each other . we now define three subsets in parameter space @xmath13 of our semigroups @xmath86 . these sets are the basic objects of interest in this paper @xmath0 is the set of @xmath1 such that @xmath10 is connected ; 2 . @xmath21 is the set of @xmath1 such that @xmath10 contains @xmath20 ; and 3 . @xmath92 is the set of @xmath1 such that @xmath10 is connected and full . recall that a set is _ full _ if its complement is connected . these sets are all closed . as far as we know , the set @xmath0 was first introduced by barnsley - harrington @xcite , and the set @xmath21 was first introduced by bousch @xcite . we are not aware of any previous explicit mention of @xmath92 , although bandt @xcite , solomyak @xcite and others have studied the ( closely related ) set of @xmath1 for which @xmath10 is a dendrite . figure [ mandelbrot ] is a picture of @xmath0 , and figure [ set_b ] is a picture of @xmath21 . the set @xmath92 is much less substantial , and it is harder to draw a good picture . drawn in @xmath13 . ] drawn in @xmath13 . ] [ proposition : c_in_b_in_a ] we have @xmath107 . it is straightforward to show ( see lemma [ lemma : disconnected_cantor ] ) that @xmath31 i.e. the limit set @xmath10 is connected if and only if @xmath108 intersects @xmath109 . since @xmath10 is rotationally symmetric about the point @xmath20 , it follows that @xmath21 is contained in @xmath0 . likewise , if @xmath10 is connected and simply - connected , then because it is rotationally symmetric about @xmath20 , it follows that @xmath10 contains @xmath20 . no two of these sets are equal ; see figure [ figure : limit_set_example ] . we will focus on the sets @xmath0 and @xmath21 for the remainder of the paper . we will show ( see theorem [ theorem : disconnected_is_schottky ] ) that @xmath1 is in the complement of @xmath0 if and only if @xmath86 is schottky . we have already observed that all schottky semigroups are topologically conjugate when restricted to good disks . the set of @xmath1 for which @xmath86 is schottky is evidently open . however , an examination of figure [ mandelbrot ] with a microscope reveals the apparent existence of tiny `` holes '' in @xmath0 , corresponding to `` exotic '' components of schottky space . one hole in @xmath0 is clearly visible in figure [ mandelbrot ] ; it is shaped approximately like a round disk except for two `` whiskers '' of @xmath0 along the real axis . but it turns out that there are also much smaller holes in @xmath0 , which can be thought of as exotic components of schottky space . this is in stark contrast to the situation of kleinian _ groups _ , where the ( teichmller ) spaces of ( quasifuchsian ) representations of a surface of fixed topological type are connected , as can be proved by means of the measurable riemann mapping theorem . figure [ hole ] shows a collection of holes in @xmath0 centered near the point @xmath110 , which we refer to colloquially as _ the diameter of the picture is approximately @xmath111 , so these holes are much too small to see in figure [ mandelbrot ] . it is one of the main goals of this paper to prove rigorously that infinitely many holes such as these really do exist in @xmath0 . near the point @xmath112 . the width of the figure is about @xmath111 . ] the sets @xmath0 and @xmath21 have a long history , and these sets ( and some close relatives ) were discovered independently several times by people working in quite different areas of mathematics . in fact , we ourselves did not learn of the work of bandt and solomyak until an advanced stage of our investigations . therefore we believe it would be useful to briefly mention some of the important papers on this subject that have appeared over the last 30 years , and say something about their contents . * in 1985 , barnsley and harrington @xcite initiated a ( mainly numerical ) study of @xmath0 . they discovered much structure evident in this set , most significantly the presence of apparent holes , whose rigorous existence they conjectured . another phenomenon they discovered was the real whiskers in @xmath0 , and they proved rigorously that @xmath0 is entirely real in some definite neighborhood of the endpoints @xmath113 of these whiskers : + [ theorem : barnsley_harrington_whiskers ] there is a neighborhood of the points @xmath113 in which @xmath0 is contained in @xmath114 . + let @xmath115 be the supremum of the real numbers @xmath116 for which @xmath0 intersects some neighborhood of @xmath117 $ ] only in real points . barnsley - harrington obtained a rigorous estimate @xmath118 but observed that this estimate is far from sharp . * in 1988 thierry bousch began a systematic study of @xmath0 and @xmath21 in his unpublished papers @xcite and @xcite . bousch proved many remarkable theorems about @xmath0 and @xmath21 , including the following : + [ theorem : bousch_connectivity ] @xmath0 and @xmath21 are both connected and locally connected . + bousch interpreted both sets as the zeros of power series with coefficients of a particular form ; we will return to this perspective in section [ section : regular ] . * in 1993 , odlyzko and poonen @xcite studied zeroes of polynomials with @xmath119 coefficients ( a set closely related to @xmath21 ) and showed the closure of this set is path connected ; their techniques are similar to those of bousch . they also noted the presence of apparent holes , and conjectured that they really existed . * in 2002 bandt @xcite developed some fast algorithms to draw accurate pictures of @xmath0 , and managed to rigorously prove the existence of a hole in @xmath0 , thus positively answering the conjecture of barnsely - harrington . bandt first realized the importance of understanding the set of interior points in @xmath0 , and made the following conjecture : + [ conjecture : bandt ] the interior of @xmath0 is dense away from the real axis . + which has been at the center of much subsequent work . note that the necessity to exclude the real axis from this conjecture is already implied by theorem [ theorem : barnsley_harrington_whiskers ] . + bandt s algorithm explicitly related @xmath120 to the dynamics of a @xmath121-generator semigroup @xmath122 , @xmath123 , @xmath124 which we denote @xmath101 , and remarked on the apparent similarity of @xmath0 and the limit set @xmath42 of @xmath101 at certain algebraic points on @xmath44 that he called _ landmark points_. * in 2003 solomyak @xcite and solomyak - xu @xcite made partial progress on bandt s conjecture , finding some interior points in @xmath0 with @xmath125 , and showing that interior points are dense in @xmath0 in some definite neighborhood of the imaginary axis . they also obtained strong results on the structure of the natural invariant measures on @xmath10 , relating this to the classical study of bernoulli convolutions , and were able to compute the hausdorff dimension and measure of the limit set for almost all @xmath1 . * in 2005 solomyak @xcite proved the asymptotic similarity of @xmath0 and @xmath42 at certain points @xmath1 which satisfy the condition that @xmath1 is a root of a rational function of a particular form . following solomyak , we refer to these points as _ landmark points_. then solomyak shows + [ theorem : solomyak ] if @xmath126 is a landmark point then @xmath0 is asymptotically similar at @xmath1 to the set @xmath42 at a certain specific point , and both of these sets are asympotically self - similar at these points . + here asymptotic similarity of two sets @xmath127 and @xmath128 at @xmath17 ( for simplicity ) means that the hausdorff distance between @xmath129 and @xmath130 restricted to balls of fixed radius ( and ignoring the boundary ) goes to zero as @xmath131 ; and asymptotic self - similarity means that there is a complex @xmath1 with @xmath132 so that the sets @xmath133 converge on compact subsets in the hausdorff topology to a limit . * in 2011 , thurston @xcite studied the set of galois conjugates of algebraic numbers @xmath134 where @xmath135 is the core entropy of a postcritically finite interval map @xmath136 , for which the parameter @xmath137 is taken from the main `` limb '' of the mandelbrot set ( the intersection of the mandelbrot set with @xmath114 ) . he asserted that the closure of this set of roots ( in @xmath65 ) is connected and path connected . in @xmath13 the closure of this set agrees with @xmath21 , and therefore the assertion generalizes theorem [ theorem : bousch_connectivity ] . for @xmath138 this assertion was verified in an elegant paper by giulio tiozzo @xcite , who also went on to plot galois conjugates associated with core entropies of postcritically finite maps @xmath136 , where @xmath137 comes from other limbs of the mandelbrot set ; these sets display a `` family resemblance '' to @xmath21 . these papers describe some remarkable connections related to the theory of postcritically finite interval maps , perron numbers , galois theory and so on . the richness and mathematical depth of these various sets has barely begun to be plumbed . we emphasize that the survey above is not exhaustive , and the papers cited contain a substantial amount beyond the part we summarize here . in this section we collect a few elementary estimates about the geometry of @xmath10 . recall our notation @xmath86 for the semigroup generated by @xmath139 and @xmath140 . the map @xmath16 fixes @xmath17 and the map @xmath18 fixes @xmath19 . any element @xmath141 of length @xmath51 acts as a similarity on @xmath65 with dilation @xmath142 and center some point of @xmath10 . we make some _ a priori _ estimates on the geometry of @xmath10 . [ lemma : diameter_bound ] the limit set @xmath10 is contained in the ball of radius @xmath143 centered at @xmath20 . let @xmath26 denote the ball of radius @xmath144 centered at @xmath20 . then @xmath145 and @xmath146 are the balls of radius @xmath147 centered at @xmath148 and @xmath149 respectively . so providing @xmath150 we have @xmath151 . but this means @xmath152 . [ lemma : distance_estimate ] let @xmath153 be words with a common prefix of length @xmath51 . let @xmath154 be contained in @xmath26 , the ball of radius @xmath143 centered at @xmath20 . then @xmath155 write @xmath156 and @xmath157 . then @xmath158 so @xmath159 by lemma [ lemma : diameter_bound ] . but the dilation of @xmath160 is @xmath161 , so the estimate follows . [ definition : compactification ] let @xmath162 denote the set of finite words in the alphabet @xmath9 , and let @xmath163 denote all right - infinite words in this alphabet , such that if a word contains @xmath164 , all successive letters are also @xmath164 . metrize @xmath163 with the metric @xmath165 where @xmath51 is the length of the biggest common prefix of @xmath166 and @xmath167 . the set @xmath168 decomposes naturally into the subset @xmath169 of words not containing the symbol @xmath164 , and words that do contain the symbol @xmath164 which are in natural bijection with @xmath162 , under the map that takes a finite word in @xmath170 to the infinite word obtained by padding with infinitely many @xmath164 symbols . [ lemma : compactification ] the space @xmath162 is compact . the subspace @xmath169 is homeomorphic to a cantor set , and @xmath162 is homeomorphic to a discrete set , whose accumulation points are precisely @xmath169 . this is immediate from the definition . there is a natural symmetry of @xmath171 interchanging the symbols @xmath16 and @xmath18 and fixing the symbol @xmath164 . note that @xmath171 is formally distinct from @xmath14 , which is the semi - group generated by compositions of the affine maps @xmath16 and @xmath18 . it s important to make the distinction between @xmath162 and @xmath14 as we are interested in how the semigroup changes as the parameter @xmath1 varies . there is an obvious map @xmath172 such that @xmath173 is the appropriate composition of the maps @xmath174 [ definition : wordaction ] let @xmath175 be a word of length @xmath51 , and ( by abusing notation ) , define the map @xmath176 given by @xmath177 we will also use the notation @xmath178 , and we often consider the map @xmath179 given by @xmath180 . the map @xmath160 is continuous in both @xmath1 and @xmath154 , which is evident in section [ section : regular ] . the map @xmath181 is defined by @xmath182 where @xmath183 is the prefix of @xmath160 of length @xmath51 , and @xmath11 is any point . by lemma [ lemma : distance_estimate ] , this limit is well - defined , independent of the point @xmath184 . [ lemma : word_composition ] for @xmath185 and @xmath186 , we have @xmath187 . that is , @xmath188 . for @xmath189 and @xmath190 , we have @xmath191 . obvious from the definitions . [ lemma : holder_estimate ] the map @xmath192 is hlder continuous with exponent @xmath193 , and the image is @xmath10 . evidently if @xmath166 is a periodic word @xmath194 then @xmath195 is the center ( i.e. the fixed point ) of @xmath196 ; since @xmath169 is compact , if @xmath97 is continuous , then the image is closed and is therefore equal to @xmath10 . so it suffices to show @xmath97 is hlder , and estimate the exponent . from the definition , if @xmath153 have a common maximal prefix of length @xmath51 then @xmath197 . on the other hand , by lemma [ lemma : distance_estimate ] we obtain @xmath198 for @xmath199 . the following result is proved in @xcite ; we include a proof for completeness . [ lemma : inner_outer ] @xmath0 ( the set of @xmath1 for which the semigroup @xmath86 has connected @xmath10 ) contains the region @xmath200 and is contained in the region @xmath201 . we shall see ( lemma [ lemma : disconnected_cantor ] ) that the limit set @xmath10 of the semigroup @xmath86 is disconnected if and only if @xmath202 is empty , in which case @xmath10 is a cantor set . in this case , the hausdorff dimension of @xmath10 can be computed from moran s theorem ( see @xcite , ch . 2 ) , as the unique @xmath203 for which @xmath204 in fact , this is easy to see directly : for a subset of euclidean space , the @xmath203-dimensional hausdorff measure transforms by @xmath205 when the set is scaled linearly by the factor @xmath135 . when @xmath10 is disconnected , it is the disjoint union of @xmath206 and @xmath207 , which are obtained ( up to translation ) by scaling @xmath30 by @xmath1 ; the formula follows . if @xmath208 then @xmath209 which is absurd , since @xmath10 is a subset of @xmath65 . thus @xmath208 is in @xmath0 , and since this set is closed , so is @xmath210 . conversely , if @xmath211 then the round disk @xmath26 of radius 1 centered at @xmath20 is a good disk , so @xmath86 is schottky , and @xmath10 is disconnected . the estimates in lemma [ lemma : inner_outer ] are sharp . taking @xmath212 , we see that @xmath213 , so @xmath214 ; in fact , @xmath215 $ ] in this case ( and for all @xmath216 ) . likewise , taking @xmath217 the rectangle @xmath144 with corners @xmath218 satisfies @xmath219 , so that @xmath220 , whereas for @xmath221 with @xmath222 the rectangle with corners @xmath223 is good and @xmath86 is schottky ; see figure [ figure : limit_set_example ] , left . solomyak xu @xcite thm . 2.8 show that the set of @xmath1 with @xmath125 for which the hausdorff dimension of @xmath10 is different from @xmath224 itself has hausdorff dimension less than @xmath225 . since ( as we shall show ) interior points are dense in @xmath226 away from @xmath114 , this implies that the simple formula for the hausdorff dimension of @xmath10 is valid on a dense subset of @xmath226 . finer results about the `` exceptions '' are known , but we do not pursue that here . the most interesting mathematical objects are those that can be defined in many different and apparently unrelated ways . the sets @xmath0 and @xmath21 can be defined in a way which is ( at first glance ) entirely unconnected to dynamics , namely as the closures of the set of _ roots _ of certain classes of polynomials . this connection is quite sensitive to the choice of normalization for our semigroups , and in fact the freedom to choose several different normalizations is itself of some theoretical interest . recall the normalization @xmath227 , @xmath228 . if @xmath229 is a word of length @xmath51 in @xmath16 and @xmath18 , we can express @xmath230 as a polynomial of a particularly simple form , namely @xmath231 where the @xmath232 are equal to @xmath19 or @xmath233 according to whether each successive letter of @xmath229 is equal to @xmath16 or @xmath18 . in particular , the limit set @xmath10 is precisely equal to the set of values of power series in @xmath1 with coefficients in @xmath234 . in this normalization , the center of symmetry is @xmath17 ( rather than @xmath20 as in our normalization ) , so we obtain the following characterization of @xmath21 : [ proposition : setb_power_series ] the set @xmath21 is the set of @xmath15 which are zeros of power series with coefficients in @xmath234 . similarly , the subsets @xmath206 and @xmath207 are the sets of _ values _ of power series with @xmath234 coefficients which start with @xmath19 and @xmath233 respectively . thus @xmath120 if and only if @xmath202 is nonempty , which happens if and only if @xmath1 is a root of a power series with coefficients in @xmath235 starting with @xmath236 . equivalently , after dividing such a power series by @xmath225 , we see that @xmath120 if and only if @xmath1 is a root of a power series with coefficients in @xmath237 starting with @xmath238 : [ proposition : seta_power_series ] the set @xmath0 is the set of @xmath15 which are zeros of power series with coefficients in @xmath237 with constant term @xmath239 . in either case , zeros of power series with prescribed coefficients can be approximated by zeros of _ polynomials _ with the same constraints on the coefficients . this suggests defining an `` extended '' @xmath0 ( resp . @xmath21 ) to be the closures of the set of _ all _ roots @xmath1 ( not just those in @xmath13 ) of polynomials with coefficients in @xmath237 ( resp . @xmath234 ) . reversing the order of the coefficients replaces a root by its reciprocal , so these extended sets are exactly the sets obtained by taking the union of @xmath0 ( resp . @xmath21 ) together with its image under inversion in the unit circle . using this interpretation of @xmath30 as the set of values of power series with @xmath234 coefficients , bousch noted an interesting relationship between @xmath0 and @xmath21 . we give the proof here for several reasons . firstly , bousch s paper is unpublished , and this argument is not easy to extract from the paper . secondly , it is short and illuminating . thirdly , it depends on a geometric fact which we use later in the proof of proposition [ proposition : iterated_seta_setb ] . [ proposition : square_set_b ] if @xmath240 then @xmath241 . consequently @xmath21 contains the annulus @xmath242 . let @xmath243 denote the set of power series with coefficients in @xmath234 . then for any @xmath244 we can write @xmath245 for unique @xmath246 . but this means that @xmath247 . now , in this normalization , limit sets all have rotational symmetry around @xmath17 . so if @xmath248 is connected but does nt contain @xmath17 , there is some symmetric innermost loop @xmath74 around @xmath17 . if @xmath10 does nt contain @xmath17 , then ( since @xmath249 ) it must be that @xmath248 and @xmath250 are disjoint , so that @xmath250 is contained in the disk bounded by @xmath74 , and similarly @xmath251 is contained in the disk bounded by @xmath252 , and therefore @xmath248 is disjoint from @xmath251 . but bousch shows this is absurd in the following way . write @xmath253 so that @xmath254 . by hypothesis , both @xmath255 and @xmath256 are compact and connected , so that @xmath257 intersects @xmath258 . but then @xmath255 must intersect @xmath256 , since if it is disjoint from them both , the union of @xmath255 with vertical rays from its top - most and bottom - most point to infinity separates @xmath257 from @xmath258 . now let s return to our normalization @xmath139 , @xmath259 . if we fix @xmath260 and vary @xmath1 , note that @xmath261 is a function of @xmath1 . for any fixed @xmath166 , we can express @xmath195 as a very simple power series in @xmath1 . in fact , the set of power series that can be obtained are precisely those whose coefficients , listed in order , are ( right - infinite ) words in an explicit regular language ( for an introduction to the theory of regular languages , see e.g. @xcite ) . [ proposition : power_series ] for any fixed @xmath262 of length @xmath263 there is a formula @xmath264 where each @xmath265 , and @xmath266 if @xmath166 ends with @xmath16 and @xmath267 if @xmath166 ends with @xmath18 . furthermore , the string of digits @xmath268 for @xmath269 can be recursively obtained as follows . read the letters of @xmath166 from left to right , and express this as a walk on the edges of the directed labeled graph in figure [ coefficient_graph ] , starting at the vertex labeled @xmath164 . @xmath164 at 50 150 @xmath19 at 150 125 @xmath233 at 250 125 @xmath17 at 200 200 @xmath17 at 200 50 @xmath16 at 125 200 @xmath18 at 105 150 @xmath18 at 155 177 @xmath18 at 155 73 @xmath16 at 245 177 @xmath16 at 245 73 @xmath16 at 200 244 @xmath16 at 200 155 @xmath18 at 200 95 @xmath18 at 200 6 are the vertex labels visited in order on the walk associated to a word.,title="fig : " ] the coefficients @xmath268 for @xmath269 ( in order ) are the labels on the vertices visited in this path , after the initial vertex . thus , the sequences that occur are precisely the sequences in which the nonzero coefficients alternate between @xmath19 and @xmath233 , starting with @xmath19 . similarly , for any fixed @xmath260 we can write @xmath270 where each @xmath265 , and the @xmath268 are obtained as the labels on the vertices associated to the right - infinite walk on the graph as above . this is immediate by induction . from proposition [ proposition : power_series ] we can quickly generate the formula for @xmath271 for any finite word @xmath166 . for example , taking @xmath272 , and writing @xmath233 as @xmath273 , we compute the sequence as follows : @xmath274 so that we obtain the formula @xmath275 we now show that in general families of semigroups of similarities with centers depending polynomially on the ( common ) dilation give rise to limit sets which are the values of power series with `` regular '' coefficients . very similar , but somewhat complementary observations were made by mercat @xcite . fix some finite alphabet @xmath72 of complex numbers , and fix a prefix - closed regular language @xmath276 . let @xmath277 denote the set of right - infinite words in @xmath72 whose finite prefixes are in @xmath255 . call a power series @xmath278 _ @xmath255-regular _ if the sequence @xmath279 . [ proposition : coefficient_language ] let @xmath280 for @xmath281 be a finite set of complex polynomials , and define @xmath282 to be the semigroup generated by contractions @xmath283 then there is a regular language @xmath255 in a finite alphabet of complex numbers so that a power series of the form @xmath284 is @xmath255-regular if and only if @xmath285 ; that is @xmath166 is an infinite composition of the generators @xmath286 , thought of as a function in @xmath1 . the effect of @xmath286 on some element of @xmath287 is to shift the sequence by one ( the @xmath288 part ) and to add the coefficients of @xmath280 to the first @xmath289 coefficients , where @xmath290 is the degree of @xmath280 . introduce the notation @xmath291 and pad coefficients up to @xmath292 where @xmath293 by defining @xmath294 if @xmath295 . then if we let @xmath296 be an arbitrary element of @xmath287 , the @xmath51th coefficient @xmath297 of the power series expansion of @xmath298 is given by the formula @xmath299 provided @xmath300 , and for @xmath301 we simply omit the terms @xmath302 for @xmath303 and @xmath304 . this coefficient depends only on the last @xmath305 letters visited in order , and a finite state automaton can store this information as a vertex . explicitly , we build a finite graph with @xmath306 vertices in bijection with words of length at most @xmath203 in the @xmath286 , and with an edge from each vertex corresponding to the word @xmath160 to the vertex corresponding to @xmath196 , with the edge labeled @xmath307 if @xmath308 has @xmath196 as a suffix of length @xmath309 . now at each vertex associated to a word @xmath160 of length @xmath310 of the form @xmath311 , put the coefficient @xmath312 we now relabel the edges in such a way that the new label on each edge is equal to the coefficient at the vertex it points to . the resulting directed graph is a _ nondeterministic _ finite state automaton in a finite alphabet ( the alphabet of possible coefficients ) , and the set of possible edge paths is some language @xmath255 . it is a standard theorem in the theory of automata due to kleene scott ( see @xcite , thm . 1.2.7 ) that there is a _ deterministic _ finite state automaton in the same alphabet recognizing @xmath255 ; this means ( by definition ) that @xmath255 is regular . moreover by construction , @xmath277 is precisely the language of coefficient sequences . we refer to the language of coefficient sequences as the _ coefficient language _ of the parameterized family @xmath282 , and denote it @xmath313 . in the special case that @xmath282 is generated by two elements @xmath314 , then at least on the subset where @xmath315 , the semigroup @xmath282 is conjugate to the semigroup generated by @xmath316 , @xmath140 that we have been studying up to now . which regular languages in a finite alphabet arise as @xmath317 for some @xmath318 ? [ example : differences ] let @xmath282 be a holomorphic family of semigroups of similarities , parameterized by @xmath1 , whose ifs has coefficient language @xmath313 . the set of differences @xmath319 is of the form @xmath320 for a suitable holomorphic family @xmath321 . in section [ section : differences ] we illustrate this difference operation in the context of our 2-generator ifss , obtaining a sequence of `` iterated mandelbrot sets '' , of which @xmath21 and @xmath0 are the first two terms . in this section we establish basic facts about the geometry and topology of @xmath30 , establishing quantitative versions of the fundamental dichotomy that either @xmath30 is ( path ) connected , or @xmath30 is a cantor set and @xmath14 is schottky . these facts lead to an explicit algorithm ( essentially due to bandt ) to ( numerically ) certify that a particular @xmath14 is schottky . it is important to describe this algorithm and its justification in some detail for several reasons . firstly , this algorithm powers our program schottky , which provided numerical certificates for many of the assertions we make in this paper ( and produced most of the pictures ! ) . secondly , understanding the _ theoretical _ behaviour of this algorithm , were it run on an ideal computer for infinite time , leads to some of the key theoretical insights that underpin our main theorems . recall that @xmath162 is the set of all finite words in @xmath16 and @xmath18 . for each @xmath322 define @xmath323 to be the set of words of length @xmath51 . because @xmath30 is minimal , for any @xmath324 , @xmath325 furthermore , the limit set is well - approximated by @xmath326 for any @xmath327 which is close to @xmath30 : [ lemma : orbits_are_uniformly_dense ] let @xmath328 . then @xmath329 where @xmath330 let @xmath331 be such that @xmath332 . we can write @xmath333 for @xmath334 . now let @xmath335 be given , and write @xmath336 for @xmath337 . let @xmath338 be the prefix of @xmath339 of length @xmath51 . consider @xmath340 . note @xmath341 , so @xmath342 . by lemma [ lemma : distance_estimate ] , @xmath343 , so by the triangle inequality , @xmath344 let @xmath345 be any compact set containing @xmath10 ( for example a disk of radius @xmath143 centered at @xmath20 ) . let @xmath346 ; this is a union of @xmath347 copies of @xmath345 scaled by the factor @xmath161 . we can construct @xmath10 as a descending intersection : [ lemma : intersection_of_disks ] we have @xmath348 observe that @xmath349 is a compact , nonempty invariant set . since @xmath10 is the unique such set , they must be equal . by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] by intersecting the unions of disks @xmath350 . the bottom right picture indicates how @xmath10 decomposes as a union of @xmath351 copies of itself centered at the indicated circles.,title="fig : " ] [ lemma : disconnected_cantor ] the following are equivalent : 1 . @xmath10 is disconnected ; 2 . @xmath10 is a cantor set ; or 3 . @xmath202 is empty . moreover , any of these conditions is implied by @xmath14 schottky . the implications ( 3 ) @xmath352 ( 2 ) @xmath352 ( 1 ) are obvious , and ( 1 ) @xmath352 ( 3 ) is a standard result in the theory of ifs . for a proof ( in exactly this context ) see @xcite , p.2 ( alternately , it follows from the estimates in lemma [ lemma : short_hop ] ) . @xmath14 schottky immediately proves ( 3 ) , since if @xmath26 is a good disk for @xmath14 , then @xmath26 contains @xmath64 , but then @xmath353 and @xmath354 contain @xmath206 and @xmath207 and are disjoint by the definition of a good disk . lemma [ lemma : disconnected_cantor ] implies that schottky semigroups have disconnected limit sets . the next lemma , although elementary , is the key to proving the converse . see figure [ figure : short_hop ] for an illustration of the paths produced by lemma [ lemma : short_hop ] . [ lemma : short_hop ] suppose that @xmath206 and @xmath207 contain points at distance @xmath355 apart . then the @xmath356 neighborhood of @xmath30 is path connected . since @xmath132 there is some @xmath51 so that for any two @xmath357 with a common prefix of length at least @xmath51 we have @xmath358 . suppose @xmath359 are words of length @xmath360 . write @xmath361 if there are right - infinite words @xmath362 with prefixes @xmath359 such that @xmath363 . then define @xmath364 to be the equivalence relation generated by @xmath365 . we claim that for all @xmath360 the equivalence relation @xmath364 has a single equivalence class ; i.e. that _ any _ two words of length @xmath360 can be joined by a sequence of words of length @xmath360 related by @xmath365 . evidently @xmath366 since we can choose right - infinite words @xmath367 and @xmath368 such that @xmath369 and @xmath370 are points in @xmath206 and @xmath207 respectively realizing @xmath371 . if @xmath372 for all words @xmath359 of length @xmath360 , then @xmath373 and @xmath374 for all words @xmath359 of length @xmath360 , since @xmath361 implies @xmath375 and @xmath376 . but if @xmath367 and @xmath368 are as above , and @xmath377 are the initial words of @xmath367 and @xmath368 of length @xmath378 then @xmath379 . so the claim is proved for all @xmath360 , by induction . taking @xmath380 and using the defining property of @xmath51 as above proves the lemma . neighborhood of @xmath10 by recursively `` jumping '' across the pair of points in @xmath88 and @xmath89 which are closest , as explained in the short hop lemma ( lemma [ lemma : short_hop]).,title="fig : " ] neighborhood of @xmath10 by recursively `` jumping '' across the pair of points in @xmath88 and @xmath89 which are closest , as explained in the short hop lemma ( lemma [ lemma : short_hop]).,title="fig : " ] neighborhood of @xmath10 by recursively `` jumping '' across the pair of points in @xmath88 and @xmath89 which are closest , as explained in the short hop lemma ( lemma [ lemma : short_hop]).,title="fig : " ] [ theorem : disconnected_is_schottky ] the semigroup @xmath14 has disconnected @xmath30 if and only if @xmath14 is schottky . it remains to prove that if @xmath30 is disconnected , then @xmath14 admits a good disk . since @xmath30 is disconnected , by lemma [ lemma : disconnected_cantor ] the distance from @xmath206 to @xmath207 is some positive number @xmath355 . by lemma [ lemma : short_hop ] it follows that the closed @xmath356 neighborhood @xmath381 of @xmath30 is path connected . so @xmath382 and @xmath383 are path connected . choose some @xmath384 with @xmath385 . let @xmath386 . by lemma [ lemma : orbits_are_uniformly_dense ] , there is an @xmath51 such that @xmath387 . let @xmath388 . by definition , each connected component of @xmath389 contains a point in @xmath10 , but @xmath390 , which is path connected , so there can only be a single connected component of @xmath389 containing @xmath10 , so @xmath389 is path connected . since @xmath389 is a finite union of closed disks which is path connected , it is homeomorphic to a disk with finitely many subdisks removed . furthermore , @xmath391 and @xmath392 are unions of round disks of radius @xmath393 around points of @xmath88 and @xmath89 , and therefore are contained in @xmath394 . because @xmath395 , @xmath391 and @xmath392 are disjoint . we now show that we can fill in the `` holes '' in @xmath389 ( if any ) and obtain a good disk . by construction @xmath389 has finitely many holes , so there is some hole of least diameter with boundary component @xmath74 . but then @xmath396 and @xmath397 have diameter strictly less than @xmath74 , and are contained in the interior of @xmath389 , so that they must bound subdisks of @xmath389 . so it follows that we can add to @xmath389 the subdisk bounded by @xmath74 to obtain a new closed set @xmath398 with @xmath399 and @xmath400 disjoint and contained in the interior of @xmath389 . add the bounded complementary components in this way one by one until we obtain a closed topological disk @xmath26 with @xmath353 and @xmath354 disjoint and contained in the interior of @xmath26 . in other words , @xmath26 is good for @xmath86 , so that @xmath86 is schottky . in this section , we describe a fast and practical algorithm to certify that the limit set @xmath10 is disconnected for a given parameter @xmath1 ( equivalently , to certify that @xmath86 is schottky ) . since this condition is open in @xmath1 , a careful analysis of this algorithm certifies that @xmath10 is disconnected on a definite open subset of parameter space . giving a rigorous numerical certificate that @xmath10 is _ connected _ , especially one valid in a definite open subset of parameter space , is more difficult , and is addressed in section [ section : interior ] . however practically speaking , the algorithm described in this section can be used to draw fast and accurate pictures of @xmath0 . the algorithm we describe differs only in inessential ways from that first discussed by bandt @xcite . we give some notation . let @xmath345 be a round disk centered at @xmath20 with the property that @xmath401 and @xmath402 are both contained in @xmath345 ; for example , we could take @xmath345 to be a disk of radius @xmath143 . let @xmath403 ; i.e. @xmath350 is the union of the images of @xmath345 under the set of words in @xmath162 of length @xmath51 . inductively , @xmath404 . by lemma [ lemma : intersection_of_disks ] , @xmath405 , so @xmath10 is disconnected if and only if @xmath350 is disconnected for some @xmath51 . @xmath350 is disconnected if and only if @xmath406 . obviously if @xmath406 , then @xmath350 is disconnected , so we must only show the converse . suppose @xmath350 is disconnected and @xmath407 . we can take @xmath51 to be minimal such that @xmath350 is disconnected and @xmath407 . since @xmath51 is minimal , for @xmath408 we have either @xmath409 is connected or @xmath410 . the latter is impossible , though , because @xmath411 , so @xmath412 . we conclude that @xmath409 is connected . but then @xmath413 and @xmath414 are connected , and @xmath407 , so @xmath350 is connected , a contradiction . naively , to check that @xmath413 is disjoint from @xmath414 would take exponential time , since we need to check the pairwise distances between elements of two sets , each with @xmath415 points . however , there is a great deal of redundacy : if @xmath160 and @xmath196 are words of length @xmath51 starting with @xmath16 and @xmath18 respectively , then if @xmath416 is disjoint from @xmath417 , then @xmath418 is disjoint from @xmath419 for all words @xmath420 . in fact , for any fixed @xmath160 and @xmath196 words of length @xmath51 , the images @xmath421 and @xmath422 are copies of @xmath350 , scaled by @xmath142 and translated relative to each other by @xmath423 . thus the relevant data to keep track of is the set of numbers @xmath424 ranging over @xmath425 of length @xmath51 where @xmath160 starts with @xmath16 and @xmath196 starts with @xmath18 , for which @xmath416 and @xmath417 intersect equivalently , for which there is an inequality @xmath426 . this discussion justifies algorithm [ algorithm : disconnectedness ] to test for connectedness of @xmath10 . we briefly explain the recursion in the context of the above observations . first , the algorithm initializes the set @xmath427 to contain the single number @xmath428 next , recall that for any word @xmath160 of length @xmath51 , we can write @xmath429 , where @xmath430 is a polynomial in @xmath1 . therefore , @xmath431 . so if we are given @xmath432 , we can compute ( for clarity , we write @xmath433 in place of @xmath434 ) : @xmath435 so given the set of differences of the form @xmath436 which are less than @xmath144 , where @xmath160 and @xmath196 may range over all words of length @xmath51 , we can compute the set of differences of words of length @xmath437 , discarding those which are larger than @xmath144 . @xmath438 @xmath439 @xmath440 @xmath441 @xmath442 @xmath443 @xmath444 @xmath445 true false if this algorithm returns true , then @xmath10 is disconnected . if it returns false , then @xmath10 might still be disconnected , but this would not be discovered without increasing the `` depth '' parameter . algorithm [ algorithm : disconnectedness ] is very fast , and has been implemented in our program schottky , available from @xcite . in practice , we can check connectedness to depths exceeding @xmath446 . the algorithm is faster in certain regions than others ; in particular , it is quite slow near the real axis . we follow this point up in section [ section : whiskers ] . in this section , we show how to construct paths inside the limit set @xmath10 and show that it is connected if and only if it is path connected . we will not explicitly need the results in this section ; however , it serves to further introduce the structure of @xmath10 , and we will use very similar ideas in section [ section : holes_in_m0 ] . these are not new results and can be derived from the general theory of ifss , but this direct approach is illuminating . note that this is essentially a continuous version of the short hop lemma [ lemma : short_hop ] . the following construction essentially appears in @xcite . suppose that @xmath447 , so there are @xmath448 with @xmath449 and @xmath450 and @xmath451 . then for any @xmath452 , we will construct a continuous path within @xmath10 between @xmath453 and @xmath454 . such that @xmath455 , we show an approximation of a path in @xmath10 between @xmath456 and @xmath457 for a given @xmath458 . the large red disk in the middle indicates a point in the intersection @xmath459 , and the other red disks show the image of this point under words of length less than or equal to @xmath351 . this method is completely analogous to figure [ figure : short_hop ] . ] first , let us narrate figure [ figure : path_in_lambda ] to explain the construction graphically . suppose that the ifs takes the disk of radius @xmath144 inside itself . we will use this fact to bound distances between points . suppose that @xmath460 and @xmath461 . note @xmath453 and @xmath454 are distance at most @xmath462 . consider the pair of words @xmath463 . by assumption since @xmath464 and @xmath465 , note that both @xmath453 and @xmath466 lie in @xmath88 , so they are distance at most @xmath467 apart . similarly , @xmath454 and @xmath466 lie in @xmath468 , so they are also distance at most @xmath467 apart . that is , the point @xmath466 coarsely interpolates between @xmath453 and @xmath454 . next consider the words @xmath469 and @xmath160 . for illustrative purposes , suppose @xmath470 and @xmath471 . then the pair @xmath472 interpolates between @xmath469 and @xmath160 : because @xmath367 agrees with @xmath469 to depth @xmath225 , and @xmath473 agrees with @xmath196 to depth @xmath225 , we have @xmath474 , and @xmath475 we can continue inductively producing points in @xmath10 coarsely between points in our path . figure [ figure : path_in_lambda ] shows the coarse path which is the result of stopping the construction after @xmath351 steps . in the limit , we will have a continuous path from the dyadic rationals into @xmath10 , and because @xmath10 is closed , this path extends continuously to @xmath476 $ ] . we now describe the construction precisely . given @xmath425 as above , we define an interpolation function on infinite words @xmath477 , as follows . given two right - infinite words @xmath478 , we may rewrite them as @xmath479 and @xmath480 , where @xmath229 is the maximal common prefix of @xmath481 and @xmath116 . then @xmath482 note that @xmath483 by lemma [ lemma : word_composition ] . furthermore if the maximal common prefix of @xmath478 has length @xmath51 , i.e. @xmath484 , then the maximal common prefix of @xmath481 and @xmath485 ( denoting the first coordinate ) is @xmath437 ; similarly , the maximal common prefix of @xmath116 and @xmath486 is also @xmath437 . now we define a set @xmath487 . the set @xmath488 will be indexed by the dyadic rational numbers , with @xmath489 denoting the element at @xmath490 ; that is @xmath491 . first set @xmath492 and @xmath493 . then recursively define @xmath494 in other words , to get the pair between @xmath495 and @xmath496 , apply the interpolation function @xmath497 to the second word at @xmath495 and the first word at @xmath496 . observe that @xmath498 is well - defined because @xmath499 . [ lemma : interpolate_lambda_props ] suppose that @xmath500 then for the construction above , we have the estimate @xmath501 by construction , @xmath502 and @xmath503 have a common prefix @xmath229 of length @xmath360 , and thus for all @xmath490 with @xmath504 , the pair of words comprising @xmath489 also has the prefix @xmath229 . therefore , the difference @xmath505 is given by a power series in @xmath1 with coefficients in @xmath506 whose first nonzero coefficient has degree at least @xmath378 . the estimate follows . [ proposition : path_in_lambda ] suppose @xmath448 with @xmath449 and @xmath450 and @xmath1 is such that @xmath451 . let @xmath452 , and let @xmath507 be the set of dyadic rational numbers . using @xmath508 as input , construct @xmath488 as above , and define the map @xmath509 given by @xmath510 . then @xmath229 extends continuously to @xmath476 $ ] , and satisfies @xmath511 ) \subseteq \lz.\ ] ] hence , @xmath10 is path connected iff it is connected iff @xmath447 . it follows from lemma [ lemma : interpolate_lambda_props ] that @xmath509 is continuous , and its image is contained in @xmath10 , so @xmath229 extends continuously as a map @xmath476\to \lz$ ] . to get the last assertion , observe that we have shown ( 3 ) @xmath512 ( 1 ) ; the other two implications ( 1 ) @xmath512 ( 2 ) and ( 2 ) @xmath512 ( 3 ) are obvious . as we saw in section [ section : topology_of_lambda ] , the topology and geometry of @xmath10 is controlled by the intersection @xmath513 . in particular , @xmath447 if and only if @xmath10 is connected . said another way , the limit set is path connected if and only if @xmath17 lies in the set of differences between points in @xmath10 . it turns out that the set of differences between points in @xmath10 is itself the limit set of an ifs . this limit set features prominently in @xcite and in section [ section : renormalization ] . define @xmath42 to be the limit set of the ifs generated by the three functions @xmath514 that is , the ifs generated by dilations by @xmath1 centered at the points @xmath515 . [ lemma : ifs_of_diffs ] we have @xmath516 . each point in @xmath30 is given by a power series in @xmath1 and associated with an infinite word in @xmath169 . thus , the set of differences of points in @xmath30 is associated with pairs of infinite words . given two words @xmath517 , it is straightforward to compute the power series giving the value @xmath518 recursively , as follows . suppose that @xmath154 begins with @xmath16 , so @xmath519 and @xmath339 begins with @xmath18 , so @xmath520 . then @xmath521 in other words , the difference associated to the pair of words @xmath522 and @xmath523 is obtained from the difference associated to the pair of words @xmath524 and @xmath525 by the transformation @xmath526 . similarly , prefixing the pair of words @xmath527 with the pair of letters @xmath528 , @xmath529 , @xmath530 transforms the differences by @xmath531 respectively . note that two of these transformations are the same . therefore , the limit set of the semigroup generated by these three transformations is precisely the set of differences in @xmath10 . but that limit set is @xmath42 . notice that the proof of lemma [ lemma : ifs_of_diffs ] effectively shows that the set of differences of any ifs generated from a regular language as in section [ section : regular ] is itself an ifs generated from a regular language . the set of differences between points in @xmath10 is @xmath42 . we can iterate this procedure by taking the set of differences in @xmath42 , and so on . let @xmath532 be the limit set of the ifs generated by @xmath533 , where @xmath286 is a dilation centered at @xmath360 . the set of differences of points in @xmath10 is @xmath534 . [ lemma : diffs_in_lambda ] the set of differences between points in @xmath535 is @xmath536 . this is just a computation in the generators analogous to the proof of lemma [ lemma : ifs_of_diffs ] . we note @xmath537 , so acting by the dilations at @xmath263 and @xmath51 on two points acts on their difference by the dilation at @xmath538 . if we define @xmath539 and @xmath540 then [ proposition : iterated_seta_setb ] @xmath541 . to see that @xmath542 , suppose that @xmath543 , so there is a pair of generators @xmath544 , @xmath545 of @xmath546 such that @xmath547 , and thus this holds for all @xmath51 , so the limit set @xmath548 is connected . conversely , if @xmath548 is connected , then @xmath549 intersects @xmath550 for some @xmath551 . but the images @xmath550 are translates of the same path connected set by multiples of the same vector , so it must be that @xmath549 intersects the translate which is closest , i.e. @xmath552 . hence @xmath543 . note that the last step in the proof of proposition [ proposition : iterated_seta_setb ] is essentially the same as bousch s proof of proposition [ proposition : square_set_b ] . in general , if our ifs is a set of dilations by @xmath1 at points @xmath553 , then the ifs which generates the differences in our ifs is the set of dilations by @xmath1 with centers at all differences of the @xmath554 . the fact that @xmath555 appears in the definition of @xmath556 ( as opposed to @xmath286 for another @xmath360 ) is natural because the number @xmath19 is always the generator of the lattice of centers . what sequences of sets arise as iterated differences ? what properties do these iterated ifs have ? we have already seen that @xmath0 contains many interior points ; in fact , the entire annulus @xmath557 is in @xmath0 . in this section we develop the method of _ traps _ to certify the existence of many interior points in @xmath0 , and examine the closure of the set of interior points . the result is quite surprising : the closure of the interior is all of @xmath0 _ except _ for some subset of the two real whiskers ! this assertion is theorem [ theorem : interior_dense ] below , which is the affirmation of bandt s conjecture ( i.e. conjecture [ conjecture : bandt ] ) . in section [ section : holes ] these techniques are used to certify the existence of ( infinitely many ) small holes in @xmath0 i.e. exotic components of schottky space . in this subsection we give a method to certify the existence of open subsets of @xmath0 . abstractly , to certify that @xmath1 is an interior point of @xmath0 is to give a proof that @xmath120 that depends on properties of @xmath1 which are stable under perturbation . showing that @xmath31 is equivalent to showing that @xmath206 intersects @xmath207 , so our strategy is to show that this intersection is inevitable for some _ topological reason _ ( depending on @xmath1 ) . proving that sets intersect in topology is accomplished by homology ( or , more crudely , separation or linking properties ) . but the homological properties of @xmath30 depend on its connectivity , which is what we are trying to establish ! so our method is first to consider precisely chosen _ neighborhoods _ of @xmath30 ( which may be presumed to be connected for some open set of @xmath1 ) , and then to consider homological properties of the configuration of the images of these neighborhoods under @xmath16 and @xmath18 which force an intersection . the key to our method is the existence of _ short hop paths _ and _ traps_. [ definition : short_hop_path ] let @xmath558 , let @xmath48 and let @xmath26 be a disk containing @xmath327 and @xmath559 . @xmath560-short hop path _ from @xmath327 to @xmath559 is a sequence @xmath561 in @xmath169 with @xmath562 and @xmath563 so that @xmath564 and @xmath565 for all @xmath360 . the existence of short hop paths is guaranteed by the short hop lemma ; in particular , we have : [ proposition : short_hop_paths_exist ] let @xmath160 and @xmath196 be right - infinite words with a common prefix @xmath229 of length @xmath51 , and suppose that there are points in @xmath88 and @xmath89 which are distance at most @xmath355 apart . let @xmath26 be a disk containing the @xmath566-neighborhood of @xmath567 . then there is a @xmath568-short hop path from @xmath569 to @xmath570 . let @xmath571 and @xmath572 , and let @xmath76 be any disk containing the @xmath356-neighborhood of @xmath10 . by lemma [ lemma : short_hop ] there is a @xmath573-short hop path from @xmath574 to @xmath575 . now apply @xmath229 to this short hop path . we now give the definition of a trap : [ definition : trap ] let @xmath26 be a closed topological disk containing @xmath10 in its interior . we say that a pair of words @xmath576 are a _ trap _ for @xmath577 if the following are true : 1 . @xmath160 starts with @xmath16 and @xmath196 starts with @xmath18 ; 2 . there are points @xmath578 in @xmath579 and @xmath580 in @xmath581 such that for some paths @xmath582 with endpoints @xmath578 and @xmath583 with endpoints @xmath580 the algebraic intersection number of @xmath115 and @xmath584 is nonzero ; and 3 . there are points in @xmath88 and @xmath89 within distance @xmath384 of each other , where the @xmath585 neighborhood of @xmath10 is contained in @xmath26 . the definition of a trap depends on a choice of paths @xmath115 and @xmath584 which intersect ; but a homological argument shows that the property does not depend on the choice : [ lemma : any_paths ] suppose @xmath425 are a trap for @xmath577 , and let @xmath586 and @xmath587 be as in definition [ definition : trap ] . then _ any _ paths @xmath582 with endpoints @xmath578 and @xmath583 with endpoints @xmath580 must intersect . any two paths @xmath588 joining @xmath578 and contained in @xmath589 are freely homotopic relative to endpoints in the complement of @xmath580 , and similarly for any two @xmath590 joining @xmath580 and contained in @xmath591 . thus the classes @xmath592 \in h_1(\c - q^\pm , p^\pm)$ ] and @xmath593 \in h_1(\c - p^\pm , q^\pm)$ ] are well - defined , and therefore so is their intersection product . [ example : trap ] figure [ figure : trap ] shows a trap in @xmath10 with @xmath594 which is visible to the naked eye . we have drawn @xmath595 for a disk @xmath26 with @xmath596 , so it is guaranteed that ( 1 ) @xmath88 and @xmath89 are contained inside the blue and orange sets , respectively and ( 2 ) there are points in @xmath10 inside every disk drawn . the computer also runs algorithm [ algorithm : disconnectedness ] to verify that @xmath595 is connected . these facts , and the ( visually evident ) fact that the highlighted disks satisfy the linking condition , proves the existence of points @xmath597 and @xmath598 inside the disks which give a trap . with @xmath594 . this picture proves the existence of a trap for this parameter , as explained in example [ example : trap ] . ] the next proposition shows that the existence of a trap for @xmath1 shows that @xmath1 is in the interior of @xmath0 . [ proposition : traps_a ] let @xmath425 be a trap for @xmath577 for some disk @xmath26 . then @xmath1 is in the interior of @xmath0 . suppose that @xmath355 is the distance from @xmath88 to @xmath89 . then any two points in @xmath10 can be joined by a @xmath599-short hop path . this is a sequence of points with gaps of size at most @xmath355 ; such a sequence is necessarily contained in the @xmath356-neighborhood of @xmath10 . it follows that @xmath600 can be joined to @xmath601 by a path @xmath115 in @xmath589 , every point on which is within distance @xmath602 of some point in @xmath603 , where @xmath160 has length @xmath51 . similarly , @xmath604 can be joined to @xmath605 by a path @xmath584 in @xmath591 , every point on which is within distance @xmath606 of some point in @xmath607 , where @xmath196 has length @xmath263 . but @xmath115 and @xmath584 must intersect , by the defining property of a trap , and lemma [ lemma : any_paths ] . thus the distance from @xmath603 to @xmath607 is at most @xmath608 . but @xmath609 because @xmath610 . this is contrary to the definition of @xmath355 unless @xmath611 . it is interesting to note that while the existence of a trap for @xmath1 certifies that @xmath1 is in the interior of @xmath0 and thus that @xmath447 , it is difficult to use it to algorithmically produce a point of intersection : as we decrease @xmath355 , the intersecting @xmath355-short hop paths need not converge or intersect `` nicely '' . in this subsection we demonstrate the theoretical utility of traps by proving that traps are dense in @xmath0 away from the real axis . since traps have nonempty interior , it follows that the interior of @xmath0 is dense in @xmath0 , again away from the real axis . this was conjectured by bandt in @xcite , p. 7 and some partial results were obtained by solomyak - xu @xcite , who proved the conjecture for points in a neighborhood of the imaginary axis . it is interesting that the proof depends on a complete analysis of the set of @xmath1 for which the limit set @xmath10 is convex ( lemma [ lemma : convex_zonohedra ] ) . it turns out that the @xmath1 with this property are exactly the union of dyadic `` spikes '' points of the form @xmath612 for coprime integers @xmath613 and @xmath490 real with @xmath614 . for @xmath615 these spikes are already in the interior of the solid annulus @xmath616 which is entirely contained in @xmath0 ; only the real `` whiskers '' protrude from this annulus , and this is why these are the only points in @xmath0 which are not in the closure of the interior . [ definition : cell_trap ] a compact connected subset @xmath617 is _ cell - like _ if its complement is connected . let @xmath127 be cell - like . a complex number @xmath229 is _ trap - like _ for @xmath127 if the following hold : 1 . the union @xmath618 is connected ( equivalently , @xmath127 intersects @xmath619 ) ; and 2 . there are 4 points in the outermost boundary of @xmath620 that alternate between points in @xmath621 and points in @xmath622 . [ lemma : nonconvex_cell_trap ] let @xmath127 be cell - like . there there is some trap - like vector @xmath229 for @xmath127 if and only if @xmath127 is nonconvex . if @xmath127 is convex , then the set of points in the boundary of @xmath623 in @xmath624 is connected , and similarly for those points in @xmath625 , so no @xmath229 is trap - like . conversely , suppose @xmath127 is nonconvex , and let @xmath626 be a supporting line such that @xmath627 both nonempty ( not necessarily connected ) , and separated by an open interval @xmath628 . the existence of such @xmath629 and @xmath630 is guaranteed precisely by the hypothesis that @xmath127 is not convex . after composing with an isometry of the plane , we can assume that @xmath626 is the horizontal axis , oriented positively , so that @xmath127 is on the side of @xmath626 with negative @xmath339 coordinates . let @xmath427 be a small open disk in @xmath631 containing the midpoint of @xmath628 , and choose @xmath632 on the side of @xmath626 with negative coordinates ( i.e. the side containing @xmath127 ) . let @xmath327 denote the point of @xmath629 with biggest @xmath154 coordinate . then @xmath633 is trap - like for @xmath127 . see figure [ figure : nonconvex_trap ] . at -4 45 at 31 48 at 70 51 .,title="fig : " ] this can be seen just by looking at the foliation of @xmath65 by vertical lines @xmath634 with @xmath154-coordinate @xmath116 , and for each line finding the point of @xmath620 with largest @xmath339-coordinate ( where this is nonempty ) . let @xmath635 be arbitrary , let @xmath116 be the maximum number such that @xmath636 is nonempty , and let @xmath490 be the point on @xmath637 with largest @xmath339-coordinate . then the four points @xmath638 are the highest points of @xmath620 on their respective vertical lines @xmath639 for @xmath640 , and alternate between the sets @xmath127 and @xmath619 . we would like to apply lemma [ lemma : nonconvex_cell_trap ] to the cell - like set @xmath641 one obtains from a limit set @xmath10 . thus , it is important to characterize @xmath1 for which the cell - like set @xmath641 obtained from @xmath10 is convex . [ lemma : convex_zonohedra ] let @xmath1 be in @xmath0 , and let @xmath641 be obtained from @xmath10 by filling in bounded complementary components , so that @xmath641 is the smallest cell - like set containing @xmath10 . then @xmath641 is convex if and only if @xmath642 for coprime integers @xmath613 and @xmath490 real with @xmath614 , in which case @xmath643 is a convex polygon . we make use of the following two facts : first , that @xmath641 has rotational symmetry of order 2 about the point @xmath20 ; and second , that @xmath10 is the union of @xmath88 and @xmath89 , obtained from @xmath10 by scaling by @xmath1 and translated relative to each other by @xmath644 . suppose @xmath641 is convex , and consider the collection of straight segments in the boundary of @xmath641 . this collection is nonempty ; for , if @xmath327 is an extremal point for @xmath641 tangent to the supporting line in the direction @xmath645 then @xmath646 and @xmath647 are extremal points for @xmath641 tangent to the same supporting line in the direction @xmath648 , and then the entire segment between these points is in the line . now , if @xmath649 is a straight segment in the boundary in the direction @xmath229 , then if @xmath650 , there is a straight segment in the boundary of the form @xmath651 or @xmath652 of length @xmath653 . it follows that there is a chain of straight segments @xmath654 where each @xmath655 is in a direction @xmath656 relative to @xmath657 , and has length @xmath658 . but then @xmath659 and @xmath660 must be in the @xmath644 or @xmath661 direction , so that their union is either equal to @xmath657 or the image of @xmath657 under the symmetry of order 2 . it follows that the argument of @xmath1 is of the form @xmath662 for some integers @xmath663 , and furthermore that @xmath664 . this proves one direction of the claim . the converse direction that limit sets @xmath10 for @xmath1 of this kind really are convex can be seen directly . in fact , these limit sets are zonohedra , the shadows of a linear semigroup acting in high dimensional space . let @xmath665 be the parallelepiped in @xmath666 consisting of vectors @xmath667 whose coordinates satisfy @xmath668 with indices taken mod @xmath559 . note this is simply a rectangular box inside the positive orthant with one corner at the origin and edges along the coordinate axes . let @xmath669 be the composition @xmath670 where @xmath671 means multiply the coordinates of @xmath196 by @xmath490 , and @xmath649 is the finite order rotation @xmath672 . so @xmath16 rotates and scales the box @xmath665 to another box along the coordinate axes . similarly , let @xmath673 where @xmath116 is the vector @xmath674 for which @xmath675 . the map @xmath18 acts in the same way as @xmath16 , except it translates the box up along the first coordinate by @xmath676 . the height of the box in the first coordinate is @xmath19 , and the heights of the acted - upon boxes @xmath677 and @xmath678 in the first coordinate are both @xmath679 . hence , providing @xmath680 , @xmath681 , so the parallelepiped @xmath665 is the limit set of the contracting semigroup @xmath682 . see figure [ figure : zonohedron ] . at 162 219 at 248 133 at 240 73 in the proof of lemma [ lemma : convex_zonohedra ] . the projection of @xmath665 to the plane is the limit set @xmath10 , so @xmath10 is convex and , in particular , a zonohedron.,title="fig : " ] projecting @xmath665 to the plane so that the vectors @xmath683 are projected to the @xmath684th roots of unity defines a semiconjugacy from this semigroup to @xmath86 where @xmath642 , taking @xmath665 to @xmath10 . [ example : hexagon ] take @xmath685 then @xmath64 is a hexagon with angles @xmath686 , and side lengths in the ratio @xmath687 . see figure [ hexagonal_limit_set ] . . ] [ lemma : surjective_perturb ] let @xmath688 be in @xmath0 , and let @xmath448 be such that @xmath689 . then for any @xmath48 there is @xmath54 and integer @xmath690 so that if @xmath691 denote the prefixes of length @xmath263 for any @xmath692 , and @xmath693 denotes the map @xmath694 then for any complex @xmath229 with @xmath695 there is @xmath696 with @xmath697 , and @xmath698 . the functions @xmath693 converge uniformly to the limit @xmath699 , which is holomorphic in @xmath1 . moreover , this limit could be constant only if @xmath700 and @xmath701 , in which case @xmath702 is identically @xmath17 and @xmath703 is identically @xmath19 , so @xmath704 ; however , @xmath705 . the limit is therefore nonconstant . thus @xmath706 takes the ball of radius @xmath384 about @xmath688 to a set containing the ball of radius @xmath707 about @xmath17 for some positive @xmath355 , and the conclusion of the lemma is satisfied for sufficiently big @xmath263 . [ corollary : surjective_perturb ] suppose @xmath708 and @xmath709 such that @xmath710 . then for any complex number @xmath229 , and any positive @xmath384 , we can find an @xmath263 and a @xmath696 with @xmath697 so that @xmath711 for sufficiently large @xmath263 , the map @xmath712 is surjective onto a neighborhood of @xmath17 , and the claim follows . we now complete the proof of bandt s conjecture : [ theorem : interior_dense ] the set of interior points is dense in @xmath0 away from the real axis ; that is @xmath713 let @xmath688 be in @xmath0 , and suppose the limit set @xmath714 is not convex . let @xmath715 be the region bounded by @xmath714 , so that @xmath715 is cell - like . since @xmath714 is not convex , neither is @xmath715 , and by lemma [ lemma : nonconvex_cell_trap ] there is some @xmath229 which is trap - like for @xmath715 . let @xmath716 , and let @xmath717 be the four points from part @xmath718 in definition [ definition : cell_trap ] . since @xmath719 , the points @xmath720 lie in @xmath714 . there is an @xmath384 so that the closed @xmath384-neighborhood of @xmath715 is connected , @xmath721 , and @xmath722 . furthermore , these conditions are open , so there is a @xmath54 such that they hold for @xmath641 for all @xmath1 with @xmath723 . now , since @xmath688 is in @xmath0 , there are @xmath724 starting with @xmath16 and @xmath18 respectively with @xmath689 . by corollary [ corollary : surjective_perturb ] , we can find some @xmath725 prefixes of @xmath160 and @xmath196 of length @xmath263 , and @xmath696 with @xmath726 so that @xmath727 . we obtain a trap for @xmath728 where @xmath729 . this follows from the three conditions above . we therefore find interior points of @xmath0 within distance @xmath355 of @xmath688 . since @xmath688 was arbitrary , we are done in the case that @xmath714 is not convex . if @xmath730 is convex and @xmath125 then @xmath1 is totally real , by lemma [ lemma : convex_zonohedra ] . if @xmath731 then we are already in the interior , by lemma [ lemma : inner_outer ] . this completes the proof . with @xmath10 convex ( in red ) overlaid on @xmath0 . the yellow circle indicates @xmath732 . ] figure [ convex_on_m ] shows the set of @xmath1 with convex @xmath10 overlaid on @xmath0 . the picture of @xmath0 in a neighborhood of the real axis is surprisingly complicated ; partial progress in understanding it was made by shmerkin - solomyak @xcite ; we describe some of their results in section [ subsection : whiskers_isolated ] , and explain how the method of traps can be modified to certify the existence of interior points in @xmath733 . in this section we rigorously certify the existence of holes in @xmath0 ( i.e. exotic components of schottky space ) . holes in @xmath0 were first observed experimentally by barnsley and harrington @xcite , and the existence of one hole was rigorously proved by bandt @xcite . however , our technique is quite different from bandt s and our proof of the existence of holes is new . furthermore , we shall show in section [ section : renormalization ] that our techniques generalize to prove the existence of _ infinitely many _ holes in @xmath0 . in this section , we give an example of an apparent hole in @xmath0 , an intuitive explanation of why the hole is truly an exotic component of schottky space , and the output of our program rigorously certifying the hole . in the next section , we give a careful justification of the algorithm . figure [ figure : certified_hole_apparent_hole ] depicts an apparent collection of holes in @xmath0 centered at @xmath734 . the diameter of the large hole is approximately @xmath735 . centered at @xmath734 . ] , left , and a zoomed view , right . the two components @xmath88 and @xmath89 ( blue and orange , respectively ) can not be unlinked with a rigid motion without intersecting . ] , left , and a zoomed view , right . the two components @xmath88 and @xmath89 ( blue and orange , respectively ) can not be unlinked with a rigid motion without intersecting . ] the limit set corresponding to a parameter inside the large hole is shown in figure [ figure : certified_hole_limit_set ] . the sets @xmath88 and @xmath89 are indeed disjoint , but they come very close . if one imagines that @xmath88 and @xmath89 are rigid , connected objects , then it is clear that one can not unlink them by a rigid motion without the two sets intersecting at some intermediate step . however , movement in parameter space does not produce exactly rigid motion of the limit set , so in order to prove that this `` hole '' in @xmath0 is not , in fact , part of the large component of schottky space , we need a more careful analysis . . a zoomed view of part of the loop shows how the program overlaps rigorous trap balls to produce a path inside the interior of @xmath0 . ] . a zoomed view of part of the loop shows how the program overlaps rigorous trap balls to produce a path inside the interior of @xmath0 . ] recall that the existence of a trap for parameter @xmath1 is an open condition there is some @xmath736 so that a trap for @xmath1 persists in @xmath737 . we call this a _ ball of traps_. our program certifies a putative hole in @xmath0 by producing overlapping balls of traps along a closed path encircling the hole , as shown in figure [ figure : certified_hole_trap_loop ] . this proves that the closed path lies completely inside the interior of @xmath0 . some technical remarks are in order . first , to complete the proof of the existence of a hole , we must certify some parameter @xmath1 on the inside of the loop as schottky . but since the connectedness of a pixel in figure [ figure : certified_hole_trap_loop ] is decided using algorithm [ algorithm : disconnectedness ] applied to some parameter inside that pixel , a white pixel is guaranteed to contain some parameter which is schottky , so this step is complete . also , we note that the loop of trap balls in figure [ figure : certified_hole_trap_loop ] appears to encircle many separate holes , but the output of this particular run of the program says nothing about whether these holes are actually distinct . we would need to run the program separately on loops encircling each of the holes we wished to rigorously separate . in section [ section : renormalization ] , we extend our algorithm to prove the existence of infinitely many holes . in this section , we describe our trap - finding algorithm in detail , including various numerical estimates . the details are important , because the output of a particular run of this algorithm serves as a rigorous certificate that there are multiple connected components of schottky space , or equivalently , holes in @xmath0 . the program will typically be required to produce a sequence of trap balls along a loop . thus , we will be interested in finding a large number of traps in a given small region of parameter space . the algorithm takes advantage of this by separating the work into two pieces : a more computationally intensive piece of one - time work to find _ trap - like balls _ , and a fast check to produce a single ball of traps . note that a trap - like ball ( of vectors ) and a ball of traps are not the same thing . this section is full of messy definitions and computations . these are necessary because we are in search of trap - like vectors similar to those found in definition [ definition : cell_trap ] but which work for _ all _ @xmath1 in a given region , so we need to carefully estimate how the limit set changes as we change @xmath1 . as a reward for this tedium , we get to compute these trap - like vectors ( which is hard ) only once , but we get to use them over an entire region . suppose that we will be searching for traps in a square region @xmath738 of parameter space centered at @xmath688 and with side length @xmath739 . let @xmath51 , the _ hull depth _ , be given . let @xmath740 ; this is the minimal radius such that a disk of radius @xmath741 centered at @xmath20 is mapped inside itself under both @xmath16 and @xmath18 . let @xmath742 denote the disk of radius @xmath327 centered at @xmath20 . typically , we compute @xmath743 , the union of images of @xmath744 under all words of length @xmath51 , to study @xmath10 . however , we need to control @xmath743 over all @xmath745 , so we need to understand how it changes as we vary @xmath1 . for this , we need some constants . the reader might consult lemma [ lemma : lambda_stays_inside ] and figure [ figure : lambda_stays_inside ] for motivation before working through the technical details . 1 . let @xmath318 be an upper bound for @xmath741 in @xmath46 . we can assume @xmath746 , so the value @xmath747 will always work . 2 . let @xmath748 be such that for any word @xmath749 and @xmath750 , we have @xmath751 . since @xmath160 can be expressed as a power series in @xmath1 with coefficients in @xmath752 , an upper bound for the derivative in terms of @xmath1 is given by @xmath753 , so a valid value of @xmath748 is given by @xmath754 . as previously mentioned , we can assume that @xmath46 lies within the disk of radius @xmath755 by lemma [ lemma : inner_outer ] , so the uniform value of @xmath756 will always work . 3 . let @xmath45 be an upper bound for @xmath757 over @xmath46 . because @xmath758 , we have @xmath759 . for the previous two constants , a uniform upper bound like this is acceptable . in this case , though , we will be raising @xmath45 to a large power , so it is critical to make @xmath45 as close to @xmath19 as possible . next set : @xmath760 and for @xmath750 , define @xmath761 [ lemma : lambda_stays_inside ] suppose that @xmath762 is connected . then for any @xmath750 , we have 1 . @xmath763 . 2 . @xmath764 contains an @xmath384-neighborhood of @xmath10 for some @xmath384 such that there are two points @xmath765 , @xmath766 such that @xmath767 . note that algorithm [ algorithm : disconnectedness ] shows that ( 2 ) implies @xmath764 is connected . at 60 44 at 75 53 at 64 73 at 20 72 at 199 47 at 260 46 at 135 48 at 144 69 at 250 67 at 173 33 at 227 33 just verifies that when we change the parameter @xmath688 to @xmath1 , each disk @xmath768 lies inside @xmath769 and @xmath10 still contains points that are close together . the figure on the right shows that we can prove ( 2 ) by proving that for each word @xmath160 , @xmath768 contains a @xmath770-neighborhood of @xmath771 . in the figure , @xmath772 and @xmath773.,title="fig : " ] the set @xmath764 is the union of disks of radius @xmath774 centered at the images @xmath434 over all words @xmath160 of length @xmath51 , and the set @xmath775 is a similar union of disks of radius @xmath776 centered at the images @xmath777 . we prove ( 1 ) by showing that for each @xmath778 , the disk of radius @xmath774 at @xmath434 lies inside the disk of radius @xmath776 at @xmath777 ; we just compute from the definition of @xmath779 : @xmath780 and by the definition of @xmath748 , we have @xmath781 , so ( 1 ) follows . to prove ( 2 ) , first note that since @xmath775 is connected , by algorithm [ algorithm : disconnectedness ] , there are words @xmath782 starting with @xmath170 , respectively , such that the disks of radius @xmath776 centered at @xmath783 and @xmath784 intersect . therefore , since these disks contain points in @xmath714 , there are right - infinite words @xmath425 starting with @xmath170 , respectively , such that @xmath785 . therefore , @xmath786 since @xmath569 and @xmath570 can each move by at most @xmath787 . so there are two points in @xmath10 which are closer than @xmath384 , where @xmath384 is the right hand side of the inequality . now we must show that @xmath764 contains an @xmath384-neighborhood of @xmath10 . we know that @xmath743 contains @xmath10 , so it suffices to show that the difference between the radii of the disks in @xmath743 and the disks in @xmath764 is at least @xmath384 . we compute @xmath788 where we have used @xmath789 and @xmath790 . also , because @xmath791 , we have @xmath792 . let @xmath793 be a component of the complement of @xmath775 inside the convex hull of @xmath775 . note that the boundary of @xmath793 contains a line segment along a supporting hyperplane for the convex hull ( the `` outside '' boundary of @xmath793 ) . there are two distinguished disks in @xmath775 which lie on either end of this line segment and are centered at images of @xmath20 under two words in @xmath323 . let these disks be centered at @xmath794 and @xmath795 . next , let @xmath559 be a point in @xmath793 which is distance @xmath796 from @xmath775 , and suppose that @xmath797 , where @xmath798 [ definition : trap_like_balls ] in the above notation , the balls @xmath799 and @xmath800 are _ trap - like balls _ for the region @xmath46 . intersects @xmath801 is two balls . vectors which translate these balls _ inside _ the convex hull but _ outside _ @xmath801 are trap - like ( left ) . translating by a trap - like vector moves @xmath801 transverse to itself and produces a trap ( right).,title="fig : " ] intersects @xmath801 is two balls . vectors which translate these balls _ inside _ the convex hull but _ outside _ @xmath801 are trap - like ( left ) . translating by a trap - like vector moves @xmath801 transverse to itself and produces a trap ( right).,title="fig : " ] that is , a trap - like ball for @xmath46 is a ball of vectors which translate a disk at a vertex of the convex hull of @xmath775 an appreciable amount into the region inside the convex hull but outside the set . see figure [ figure : trap_like_balls ] . one might wonder whether we should expect any trap - like balls to exist at all , since it s not immediately clear why @xmath115 should be positive . recall that @xmath802 , so for @xmath51 large enough , @xmath803 , and @xmath203 is probably tiny compared to the scale of @xmath775 ( which is approximately the limit set @xmath714 ) . [ lemma : trap_like_balls_are_symmetric ] if @xmath804 is a trap - like ball for @xmath46 , then @xmath805 is also a trap - like ball . the set @xmath775 is rotationally symmetric under a rotation of order @xmath225 about the point @xmath20 . a trap - like ball is taken to a trap - like ball under this rotation , and in the above notation , it will negate the vectors @xmath806 and @xmath807 . in this section , we show how to use the trap - like balls produced in the previous section to verify the existence of a ball of traps at @xmath1 . we fix notation as in the previous section , so we have a square region @xmath46 in parameter space with side length @xmath739 and centered at @xmath688 . we let @xmath318 be an upper bound for @xmath741 over @xmath46 and @xmath748 be such that @xmath781 for all @xmath778 and @xmath750 . [ lemma : ball_of_traps ] let @xmath808 be such that @xmath160 starts with @xmath16 and @xmath196 starts with @xmath18 and @xmath809 , where @xmath810 is a trap - like ball for @xmath46 . let @xmath811 be a lower bound for @xmath812 over @xmath46 . then there exists a trap for every @xmath813 , where @xmath814 we will check the 3 hypotheses of definition [ definition : trap ] on the words @xmath160 and @xmath196 with the topological disk @xmath764 . first , @xmath425 start with @xmath170 by construction , so the first condition is verified . the third condition , that there are points in @xmath88 and @xmath89 within distance @xmath384 , where the @xmath585 neighborhood of @xmath10 is contained in @xmath764 , is conclusion ( 2 ) of lemma [ lemma : lambda_stays_inside ] . now we need to verify condition ( 2 ) in the definition of a trap . this is the more difficult verification . after a suitable rescaling , the problem becomes more tractable . consider the unions @xmath815 and @xmath816 . these sets have a pair of intersecting paths as in condition ( 2 ) if and only if the original sets @xmath817 and @xmath818 do . furthermore , the sets @xmath815 and @xmath816 are exactly the same , up to translation , as @xmath764 and the translated set @xmath819 in other words , we translate the set @xmath764 off of itself by the vector @xmath820 . if we can find the interlocking paths of condition ( 2 ) , we are done . we start by considering @xmath775 and then thinking about what can happen as we change @xmath688 to @xmath1 . by hypothesis , @xmath229 lies in a trap - like ball for @xmath46 . these are four distinguished disks associated with @xmath229 , as follows . the vector @xmath229 is associated with a component of the complement of @xmath775 inside the convex hull of it . this component has one side which lies along a supporting hyperplane @xmath821 of the convex hull , and @xmath821 intersects two disks @xmath822 which sit on either side of the component . by the definition of the trap - like balls , the disk @xmath823 , which has radius @xmath776 , is translated by @xmath229 to a disk @xmath824 , which is distance at least @xmath825 away from both @xmath821 and @xmath775 . also note that @xmath822 are this same distance away from the translated set . let @xmath826 be a hyperplane perpendicular to @xmath229 , and translate it so it is supports @xmath775 . it lies tangent to some disk @xmath827 . now @xmath827 is translated by @xmath229 to a disk @xmath828 which is distance at least @xmath829 away from the slid @xmath826 , and thus that distance away from @xmath775 . the pairs of disks @xmath830 and @xmath831 are linked . see figure [ figure : trap_ball_movement ] . . if we change the parameter slightly , the marked balls @xmath832 can not move much and thus still give a trap . ] ( 0,0 ) ( -165,185)@xmath823 ( -92,163)@xmath833 ( -132,159)@xmath828 ( -35,140)@xmath821 ( -15,106)@xmath826 ( -15,23)@xmath828 now change the parameter from @xmath688 to @xmath1 , and consider @xmath764 . because @xmath834 , the disks @xmath832 can only shrink . and every disk , and the supporting hyperplanes @xmath835 , can move at most distance @xmath836 . therefore , these four disks are still disjoint from the opposing copy of @xmath764 , and each contains points in @xmath10 , and by lemma [ lemma : lambda_stays_inside ] , @xmath764 remains connected , so these points can be connected by paths with algebraic intersection number @xmath19 . this verifies condition ( 2 ) of the trap definition . this shows that there exists a trap for parameter @xmath1 , but recall that we desire a trap for every @xmath1 in @xmath837 . to see that this is true , observe that at the point @xmath1 , the disks @xmath832 are still distance at least @xmath838 away from the opposing copy of @xmath764 . so we can change @xmath1 again by at most @xmath384 while retaining this trap . all of this is contingent on the parameter @xmath1 remaining in @xmath46 , so the final ball of traps produced is @xmath839 . we now summarize this section . to certify a hole in @xmath0 which lies completely within some square region @xmath46 , we compute @xmath840 and the set @xmath775 for a reasonable - sized @xmath51 ( say , @xmath841 ) . then we compute the convex hull and some trap - like balls . then , at the initial point of a path @xmath74 encircling the hole , we apply lemma [ lemma : ball_of_traps ] to produce a ball @xmath842 of traps . then we go along @xmath74 to the edge of @xmath842 , and find another ball of traps @xmath843 , and so on . the balls overlap , so together they produce an open set inside set a containing @xmath74 . we do all computations to double precision , which has a precision of at least @xmath841 decimal digits . therefore , as long as no number in the computation ever requires more than , say , @xmath844 digits of precision , this is rigorous . in practice , this is never an issue . [ question : hole_combinatorics ] is there a combinatorial way to distinguish holes in @xmath0 ? in this section we describe a certain family of natural operators on the parameter plane which account for much of the observed self - similarity in the structure of @xmath0 and @xmath21 . similar ideas and some similar results already appear in the work of solomyak @xcite , although our approach is sufficiently different ( and enough of the results we obtain are new ) that it is worth including here . the first main result we obtain is the existence of _ infinitely many holes _ in @xmath0 , arranged in certain spirals . the proof of this fact does not technically need the theoretical apparatus of renormalization ; but the phenomenon is not properly explained without it . we defer the explanation until after a description and rigorous proof of the phenomenon , so that the techniques and definitions we then introduce are sufficiently motivated . numerical exploration of @xmath0 quickly reveals many interesting phenomena , of which one of the most interesting is the appearance of apparent `` spirals '' of holes . one of the most prominent is centered at the point @xmath37 . see figure [ figure : hexahole_spiral ] . the figure also illustrates part of @xmath21 ( in purple ) , and exhibits the limit as the `` tip '' of a spiral of @xmath21 . techniques of solomyak @xcite certify that @xmath21 is self - similar at the limit , and is asymptotically similar to the limit set @xmath845 . . ] the main theorem we prove in this section is the following : [ theorem : hole_limit ] let @xmath37 be the root of the polynomial @xmath38 with the given approximate value . then 1 . @xmath39 is in @xmath0 , @xmath21 and @xmath92 ; in fact , the intersection of @xmath40 and @xmath41 is exactly the point @xmath20 ; 2 . there are points in the complement of @xmath0 arbitrarily close to @xmath39 ; and 3 . there are infinitely many rings of concentric loops in the interior of @xmath0 which nest down to the point @xmath39 . thus , @xmath0 contains infinitely many holes which accumulate at the point @xmath39 . we refer informally to the holes accumulating on @xmath39 as _ hexaholes _ ( because of their approximate shape ) , and to @xmath39 itself as the _ hexahole limit_. the first step in the proof is to give a more combinatorial description of the hexahole limit @xmath39 , and prove the first bullet of theorem [ theorem : hole_limit ] . the set of @xmath1 for which @xmath846 are exactly the roots of @xmath38 . by proposition [ proposition : power_series ] , the power series associated to these two infinite words are actually finite polynomials ; equating them gives the identity @xmath847 so that @xmath848 . the next step in the proof requires us to certify the existence of points in the complement of @xmath0 , arbitrarily close to @xmath39 . any given value of @xmath1 can be numerically certified as being in the complement of @xmath0 by algorithm [ algorithm : disconnectedness ] , but we would like to apply this algorithm uniformly to an infinite collection of @xmath1 of a particular form . first recall the form of the algorithm : given @xmath1 as input , and a cutoff depth , we first load the number @xmath849 into a `` stack '' @xmath427 , and then recursively replace the content of the stack at each stage with a set of _ viable children_. more precisely , for each @xmath850 , there are three children @xmath851 , @xmath852 , and @xmath853 . a child is _ viable _ if its absolute value is less than a constant @xmath144 depending only on the initial value @xmath1 ( in fact , we can take @xmath144 to be fixed throughout a neighborhood of a given @xmath1 ) , and at each stage of the algorithm we replace each number in @xmath427 with the set of its viable children . the algorithm halts whenever the stack @xmath427 is empty ( in which case we certify that @xmath1 is in the complement of @xmath0 ) or if we exceed the `` run time '' ( i.e. the cutoff depth ) allocated in advance . let s imagine running our algorithm on an ideal machine without imposing a cutoff depth , so that the algorithm halts if and only if @xmath10 is disconnected . at each successive time step @xmath203 , the stack @xmath427 consists of a finite list of numbers . if it happens that the content of @xmath427 is eventually _ periodic _ ( and nonempty ) as a function of @xmath203 , then of course the algorithm never halts certifying that @xmath1 is in fact in @xmath0 . now , the numbers in @xmath427 at each finite stage are all laurent polynomials in @xmath1 of degree bounded by @xmath203 , so if @xmath427 is eventually periodic , then @xmath1 must be algebraic . if we apply the algorithm to the number @xmath39 defined above , then we indeed can certify that @xmath427 is eventually periodic , and in fact becomes _ constant _ after @xmath854 . for @xmath37 the root of @xmath848 we can take @xmath855 . unfortunately , the full stack over all @xmath856 steps is somewhat unwieldy , so we do not list it here . however , on step @xmath857 , the stack contains the number @xmath19 , and this is the only stack entry with viable descendants . the tree diagram of the algorithm at this point becomes periodic ; we show it in figure [ figure : periodic_stack ] . ( 0,0 ) ( -70,0)(a)@xmath858 ( -70,-15)(b)@xmath859 ( -70,-30)(c)@xmath860 ( -70,-45)(d)@xmath861 it is clear from the tree diagram that the stack becomes constant at step @xmath856 . by analyzing precisely which children have indefinitely viable descendents , we get a precise description of the intersection @xmath862 . in this case , we can readily observe that there is a unique pair of infinite words @xmath425 where @xmath160 starts with @xmath16 and @xmath196 with @xmath18 so that @xmath863 ; these words are in fact related under the canonical involution , so that the intersection consists exactly of the point @xmath20 . this proves the first bullet in theorem [ theorem : hole_limit ] . the second step in the proof of theorem [ theorem : hole_limit ] is to certify the existence of points in the complement of @xmath0 arbitrarily close to @xmath39 . these points will all be of the form @xmath864 for sufficiently big @xmath626 , and for a fixed constant @xmath865 . [ proposition : omega_limit ] for @xmath865 , there is @xmath626 so that point @xmath866 is schottky for sufficiently large @xmath626 . in order to prove proposition [ proposition : omega_limit ] , we are going to formally run the disconnectedness algorithm on @xmath1 and show that we can understand the contents of the stack as long as @xmath626 is large enough . the stack will essentially be the same as the stack for @xmath39 for a long time , followed by a uniformly bounded ( in @xmath626 ) number of steps which prove disconnectedness . this discussion is elementary , but it requires taking things to infinity in a careful order . we first prove a general lemma which provides the stack contents ; proving the proposition then reduces to doing a numerical computation for the given @xmath748 value . to set up the lemma , we need to do a computation . recall that when running the disconnectedness algorithm on @xmath39 , at every step there is a single stack entry ( i.e. `` @xmath19 '' ) which has infinitely many descendants . the true , unsimplified version of this entry in the stack at step @xmath51 is @xmath867 at every time step , there are @xmath868 other polynomials on the stack , which are the finitely many children of @xmath869 , @xmath870 , and @xmath871 which have not yet died . these polynomials are : @xmath872 it is important to note that obtaining these polynomials involves _ no _ simplification at any stage . thus , heuristically , for fixed @xmath51 and @xmath1 values close to @xmath39 , these polynomials should give the stack contents . this is the idea lemma [ lemma : contents_of_stack ] explores in detail . we compute the values of the @xmath873 polynomials at the point @xmath874 ( it is pedagogically helpful to split @xmath626 in proposition [ proposition : omega_limit ] into the two variables @xmath875 ) . this computation is just an expansion , simplified using the polynomial of which @xmath39 is a root . for compactness , we denote @xmath876 by @xmath877 . all of these polynomials have a large `` remainder '' term , which we will denote by @xmath878 now we list the polynomials : @xmath879 @xmath880 [ lemma : contents_of_stack ] for any @xmath748 , there are constants @xmath881 and @xmath690 such that for all @xmath882 and @xmath883 , the contents of the stack of the disconnectedness algorithm at step @xmath51 when run on @xmath884 is exactly the set of @xmath877 for @xmath885 . in order to prove this lemma , first consider running the algorithm on @xmath39 : the stack beyond step @xmath856 is constant at @xmath886 now think of varying the input from @xmath39 to @xmath884 . in order to prove that the contents of the stack are as claimed , we need to show two things ( 1 ) the polynomials @xmath877 stay on the stack for all @xmath887 and ( 2 ) every child which was discarded for @xmath39 through step @xmath51 still gets discarded . first note that for @xmath887 , @xmath888 and furthermore , @xmath889 converges to @xmath890 from below as @xmath891 . therefore , the absolute value @xmath892 is uniformly ( in @xmath51 ) bounded above by the `` worst case '' @xmath893 where @xmath894 : @xmath895 so for example , the variation @xmath896 is uniformly ( in @xmath51 ) bounded by the expression : @xmath897 there are similar expressions for @xmath898 for each @xmath360 . now if we make @xmath881 large and bound @xmath263 from below , we can make all these expressions as small as we like , and hence small enough so that the @xmath877 remain on the stack for all @xmath887 . to prove that these are the _ only _ things on the stack , we compute expressions for the children of @xmath877 and do exactly the same thing to prove that a large enough @xmath881 and @xmath263 make the worst - case deviation from the children of @xmath899 small for @xmath887 and hence these children will leave the stack exactly as the children of @xmath899 do . this computation is the same , so we omit it . we note that for a specific value of @xmath748 ( such at the one given in the proposition ) , it is possible to actually numerically compute @xmath881 . as an example , we show how to compute a value of @xmath881 which ensures @xmath900 remains on the stack for sufficiently large @xmath263 . as @xmath901 , the difference @xmath902 is bounded above by the limit @xmath903 hence if we compute @xmath904 and observe how far away it is from getting cut off the stack ( remember things get removed if their absolute value is too large ) , we can choose @xmath881 so that the expression above is small enough that @xmath900 remains on the stack for @xmath905 and @xmath887 , ( where @xmath690 can depend on @xmath881 ) . in order to compute a value of @xmath881 which actually works for lemma [ lemma : contents_of_stack ] , it is necessary to consider @xmath906 over all @xmath360 and all their children and make sure that @xmath881 is large enough to accept or reject them appropriately . doing the computation above for the specified value @xmath865 ( this is an exact value ) shows that lemma [ lemma : contents_of_stack ] holds with @xmath907 . therefore , for all @xmath263 sufficiently large , the contents of the stack at time @xmath887 will be as claimed in the lemma . by taking @xmath263 large , we can get the stack contents at step @xmath894 as close as we like to the limits , which we denote by @xmath908 . @xmath909 we want to continue running the disconnectedness algorithm at this point . recall we require a radius outside which we discard children . by taking @xmath263 large , we may assume this radius is the one for @xmath39 , i.e. @xmath910 and that the algorithm replaces @xmath115 with the three children @xmath911 now start this algorithm with the given ( numerical , with @xmath907 ) stack contents @xmath912 ; it terminates ( with an empty stack ) in @xmath913 steps . therefore , for @xmath907 , there is some @xmath690 such that for all @xmath882 , the disconnectedness algorithm run on the input @xmath914 certifies disconnectedness at step @xmath915 . this completes the proof . this proves the second bullet in theorem [ theorem : hole_limit ] . note that by means of this method we can numerically certify any @xmath916 for which the points @xmath917 are schottky for all sufficiently large @xmath51 . however , when this method of certification fails , we can not conclude that the corresponding points are all ( eventually ) in @xmath0 ; a different method is necessary for that . the last step in the proof of theorem [ theorem : hole_limit ] is to certify the existence of infinitely many rings of concentric loops in the interior of @xmath0 which nest down to the point @xmath39 . this depends on an analysis of how trap vectors transform under certain combinatorial and numerical operations . we discuss this in the remainder of the section . let @xmath918 be a small region containing @xmath39 . recall from section [ section : holes ] that we can produce a collection of trap - like balls for the region @xmath144 such that if @xmath919 and there exist @xmath920 starting with @xmath170 , respectively , such that @xmath921 lies in a trap - like ball , then there exists a trap at @xmath1 , and @xmath1 lies in the interior of @xmath0 . we will use this to show that for @xmath1 of the form @xmath917 , we can certify the existence of a trap for @xmath1 for _ all _ sufficiently large @xmath51 . given two words @xmath808 , not necessarily starting with @xmath170 , recall that we can write @xmath922 for some polynomials @xmath430 and @xmath923 in @xmath1 . for example , @xmath924 because @xmath925 . define words @xmath926 [ lemma : limit_trap ] in the above notation , if the vector @xmath927 lies in a trap - like ball @xmath928 for the region @xmath144 , then for sufficiently large @xmath51 , the words @xmath929 give a trap for @xmath917 . furthermore , if we let @xmath930 then for any compact subset @xmath72 of @xmath931 , there is an @xmath932 such that for any @xmath933 and @xmath934 , the words @xmath935 give a trap for @xmath936 . the proof is primarily a computation . by applying the definitions of @xmath16 and @xmath18 , we compute : @xmath937 @xmath938 so @xmath939 where @xmath940 . recall from the definition of @xmath39 that @xmath941 . since @xmath942 and @xmath943 have length @xmath944 , to show that this pair gives a trap - like vector for some @xmath1 , we ll be considering the expression @xmath945 we now show how to certify that this vector is trap - like for @xmath1 of the form @xmath917 , for sufficiently large @xmath51 . we therefore consider @xmath946 note that the right summand converges to @xmath947 as @xmath948 . we claim the left summand converges as well . to see this , we expand it out using the definition of @xmath949 : @xmath950 the first line is @xmath17 because , recall , @xmath941 , and it s straightforward to see that @xmath951 , so the last three lines converge to @xmath17 and the second line converges to @xmath952 . therefore , if the hypothesis of the lemma holds , then for sufficiently large @xmath51 , the words @xmath942 and @xmath943 are trap like for @xmath917 , as claimed . to get the last statement of the lemma , observe that the vector @xmath116 varies linearly with @xmath748 , so certainly for any @xmath953 , the hypotheses of the lemma are satisfied . but note that all the expressions above are uniformly continuous in @xmath748 on compact subsets , so given any compact subset , there is a uniform bound on the value of @xmath51 required . to complete the proof of theorem [ theorem : hole_limit ] , then , it suffices to exhibit a loop of overlapping balls output by lemma [ lemma : limit_trap ] encircling @xmath39 . because there are finitely many balls , there is a uniform @xmath932 such that for @xmath934 , there exists a trap for @xmath954 for every @xmath748 in every ball in this loop . in other words , the image of this loop under the map @xmath955 lies in the interior of @xmath0 for all sufficiently large iterates . figure [ figure : limit_trap_loop ] shows the loop of trap balls which we computed . . ] lemma [ lemma : limit_trap ] only states that this loop is _ eventually _ in the interior of @xmath0 ( under a large enough iterate of the map @xmath956 ) . however , experimentally , this loop lies in the interior for all iterates . the primary evidence for this is that a picture of limit traps near @xmath39 looks the same as a picture of regular traps . in this section , we place the above example in a more formal context and explain the relationship with the work of solomyak in @xcite . we first give a heuristic explanation of some of our definitions . we would _ like _ to define a renormalization operator @xmath957 such that @xmath958 is the parameter @xmath229 such that the limit set for @xmath229 is the same , in some sense , as the union @xmath959 . the right definition for this operator is elusive . however , we show below that we can understand what the fixed points of renormalization should be , and at these fixed points , there is a sensible definition of a limiting trap . for certain renormalization points , we give a new interpretation of a result of solomyak @xcite . let @xmath160 and @xmath196 be given words of the same length . these will be our prefixes . let @xmath481 and @xmath116 be two other words of the same length . we are interested in the appearance of the set @xmath960 and renormalization with respect to the words @xmath961 and @xmath962 as @xmath963 . as @xmath51 gets large , renormalization at @xmath961 , @xmath962 should converge to a locally - defined holomorphic function which , abusing notation , we ll call renormalization at ( the now infinite words ) @xmath964 , @xmath965 . parameters @xmath1 for which @xmath966 should be the fixed points of this renormalization . therefore , we say that a parameter @xmath1 is a _ renormalization point _ if there are words @xmath967 as above such that @xmath966 . we will show that there is a notion of a limit trap at a renormalization point and that this can sometimes give an asymptotic self - similarity . this section is essentially concerned with the behavior of the limit set @xmath10 at infinitesimal scales for renormalization points . that @xmath39 is a renormalization point means that @xmath968 and in fact there are two eventually periodic words @xmath425 so @xmath863 . we want to zoom in on this point of intersection . recall that for a finite ( or infinite ) word @xmath778 , we can write @xmath969 , where @xmath970 is a polynomial of degree @xmath51 ( if @xmath160 is infinite , @xmath430 is the power series @xmath569 ) . we take the convention that if @xmath160 has length @xmath17 , then @xmath971 . if @xmath972 , then @xmath973 and @xmath974 are translates of each other , and the displacement vector is @xmath975 . a more useful quantity turns out to be the displacement relative to the sizes of the sets @xmath973 and @xmath974 , that is @xmath976 we have already encountered this expression several times . as in the proof of theorem [ theorem : hole_limit ] , we will need to compute its value for parameters of the form @xmath917 for long words . this section contains a rather tedious computation which will be necessary for its generalization . [ lemma : sum_computation ] let @xmath425 have length @xmath469 ; let @xmath478 have length @xmath977 ; and let @xmath420 have length @xmath137 . let @xmath39 be a renormalization point for @xmath967 . write @xmath978 , so @xmath979 . then as @xmath963 , the quantity @xmath980 converges to @xmath981 first , some notation . write @xmath290 for the coefficients of the power series @xmath982 , so @xmath983 . note that @xmath290 is periodic with period @xmath977 for large enough @xmath360 ; write @xmath984 to mean the eventually periodic part of @xmath982 , shifted by @xmath469 , so @xmath985 where @xmath51 is taken large enough that the coefficients are constant in @xmath51 . if we take a finite power for @xmath481 and @xmath116 , the resulting polynomial ( which has degree @xmath986 ) will agree with @xmath982 to the term with degree @xmath987 , so define @xmath988 so that @xmath989 observation of the power series @xmath629 shows the facts ( the third following from the first two ) : @xmath990 we will soon encounter some rather large expressions , and it will be helpful to use some small notation . we denote the expression in the lemma by @xmath991 , so @xmath992 and we denote @xmath993 by @xmath994 . recall that @xmath995 . we expand using the fact that @xmath996 : @xmath997 the first part trivially converges to @xmath998 as @xmath948 . we will show that @xmath999 to do this , we expand the term @xmath1000 using the binomial theorem : @xmath1001 we handle these summand - by - summand . first , we rewrite ( 1 ) using the fact that @xmath979 so @xmath1002 , so @xmath1003 next , summand ( 2 ) : @xmath1004 finally , summand ( 3 ) . we will prove that it converges to @xmath17 . first , we bound the absolute value of the innermost sum . to do this , we pull out terms from the binomial coefficient to re - express it as a different binomial coefficient , so we can collapse the sum into a power . in the first line , we use the fact that @xmath1005 , and @xmath1006 : @xmath1007 so the entire summand ( 3 ) is bounded in absolute value by @xmath1008 let @xmath1009 . using the root test , it is easy to see that @xmath1010 is uniformly convergent for @xmath132 , so @xmath821 is uniformly convergent in a neighborhood of @xmath1011 . therefore , as @xmath963 , the above expression converges to @xmath1012 we have now shown that as @xmath948 @xmath1013 using the observations about @xmath629 at the beginning of the proof , this expression rearranges into the statement of the lemma . recall from section [ section : differences ] that the set of differences between points in @xmath10 is @xmath42 , the limit set generated by the three contractions @xmath1014 [ theorem : similarity ] suppose that @xmath39 is a renormalization point for @xmath967 , where @xmath478 have length @xmath977 . let @xmath978 . let @xmath1015 denote @xmath1016 , the translated , scaled copy of @xmath1017 1 . if @xmath1018 , then for all @xmath48 , there is a @xmath49 such that @xmath50 and for all sufficiently large @xmath51 , there is a trap for @xmath52 . if there is a unique pair of infinite words @xmath1019 such that @xmath1020 ( i.e. @xmath1021 , @xmath1022 ) , then there is @xmath54 such that for all @xmath1023 with @xmath56 , the limit set for the parameter @xmath57 is disconnected for all sufficiently large @xmath51 . a version of part ( 2 ) of theorem [ theorem : similarity ] still holds if there are finitely many such infinite @xmath1024 , as long as they are eventually periodic . in this case , we need to replace @xmath1015 with a union of multiple scaled , translated copies of @xmath1017 . we can think of theorem [ theorem : similarity ] as the verification of a kind of `` renormalized bandt s conjecture '' . it says that at a renormalizable point @xmath39 , there is an increasing union of open subsets of renormalizable traps , limiting to the asymptotically scaled copy of @xmath0 centered at @xmath39 . it implies ( but is stronger than ) one of the main consequences of theorem 2.3 from solomyak @xcite , that suitable neighborhoods of zero in @xmath1015 converge in the sense of hausdorff distance to suitably scaled neighborhoods of @xmath39 in @xmath0 . in contrast to solomyak , our argument is more closely expressed in the language of algorithms , since one of our aims was always to use this theorem to obtain numerical certificates of the existence of hole spirals . this is stated carefully in lemma [ lemma : limit_traps_general ] . [ lemma : limit_traps_general ] let @xmath425 have length @xmath469 ; let @xmath478 have length @xmath977 , and let @xmath420 have length @xmath137 . let @xmath39 be a renormalization point for @xmath967 . write @xmath978 . suppose that @xmath748 is such that the vector @xmath1025 is trap - like for @xmath39 . then the words @xmath1026 , @xmath1027 give a trap for @xmath57 for all sufficiently large @xmath51 . this is essentially immediate from lemma [ lemma : sum_computation ] , which says that the vector which determines whether @xmath1026 , @xmath1028 give a trap converges to the above expression as @xmath51 gets large . hence , if the above is trap like , we get a trap for @xmath1029 for all @xmath51 large enough . if lemma [ lemma : limit_traps_general ] holds for some point @xmath39 and @xmath748 , we say that @xmath748 admits a _ limit trap _ for @xmath39 . we first prove part ( 1 ) . let us be given @xmath1018 . by lemma [ lemma : limit_traps_general ] , if the vector : @xmath1030 is trap - like for @xmath39 , then @xmath318 admits a limit trap . let @xmath793 be a trap - like vector . then we can solve for the associated value @xmath49 which admits a limit trap : @xmath1031 as @xmath137 grows and @xmath154 and @xmath339 vary over all words of length @xmath137 , the first summand goes to zero , and the second two together converge ( in the hausdorff topology , say , but quite regularly ) to @xmath1015 . hence if @xmath1018 , then for any @xmath48 , there are words @xmath1032 so that @xmath49 admitting a limit trap as above has @xmath50 . this completes the proof of part ( 1 ) . now we prove part ( 2 ) . when we run algorithm [ algorithm : disconnectedness ] on @xmath39 , the stack entries at stage @xmath986 are exactly the scaled differences @xmath1033 between centers of words @xmath420 of length @xmath986 ( when these differences are small enough to remain on the stack ) . if there is a unique pair of words @xmath1024 such that @xmath1020 , then there is a single stack entry with infinitely viable children , and it is @xmath1034 . rewriting this as in the proof of lemma [ lemma : sum_computation ] , we see that by making @xmath51 large , this expression is as close as we d like to @xmath1035 . when we vary @xmath39 to @xmath1036 , and make @xmath51 large , then by lemma [ lemma : sum_computation ] , we can make this stack entry as close as we like to @xmath1037 therefore , there is a @xmath54 as in the statement of the theorem such that if @xmath1038 , then when we run the disconnectedness algorithm on the input @xmath1036 , the stack at step @xmath986 has the entry ( as close as we want to ) @xmath1039 , and every other stack entry has children which are eliminated in finite time . the value for @xmath355 can be found by checking how far the limiting entry @xmath1035 is from the cutoff ; then make @xmath355 small enough so that adding the term @xmath1040 does not push anything off of or onto the stack . now , compute all possible children after @xmath137 more steps ; by lemma [ lemma : sum_computation ] , we get @xmath1041 where @xmath420 vary over all words of length @xmath137 . we rearrange : @xmath1042 however , the fact that @xmath748 is not in @xmath1015 means that as we increase @xmath137 , the minimum value of quantity on the right above goes to infinity . thus , @xmath1043 . hence , at some finite @xmath137 , every one of these children has left the stack . recall the stack entries above are limits of the real stack entries we see for step @xmath1044 , but by choosing @xmath51 large enough , we can make the computation valid ( because @xmath137 is some finite number , so there are finitely many quantities to bring close to their limits ) . hence for @xmath51 large enough , the disconnectedness algorithm certifies that the limit set for @xmath993 is disconnected . figure [ figure : c_values_for_hexahole ] shows an example of @xmath1015 near @xmath17 for the renormalization point in theorem [ theorem : hole_limit ] . see also the pictures in @xcite . near @xmath17 ( left ) for @xmath1045 and set @xmath0 near @xmath39 ( on right).,title="fig : " ] near @xmath17 ( left ) for @xmath1045 and set @xmath0 near @xmath39 ( on right).,title="fig : " ] we end this section by proposing two ( related ) conjectures : [ conjecture : algebraic_points ] the algebraic points in @xmath44 are dense in @xmath44 . [ conjecture : limits_of_holes ] every point in @xmath44 not on the real axis is a limit of a sequence of holes with diameters going to zero . we believe that fixed points of renormalization are the key to both conjectures ; such fixed points are on the one hand algebraic , and on the other hand points where @xmath0 is asymptotically self - similar , and asymptotically similar to the limit set of a 3-generator ifs . it is very easy for a connected limit set of a 3-generator ifs to fail to be simply - connected : irregularities in the frontiers of the translates overlap each other in complicated ways , cutting off tiny holes . once there is one tiny hole , there will be infinitely many , accumulating densely in the boundary of the limit set ; thus one expects the corresponding point in @xmath0 to be a limit of tiny holes . the experimental evidence for conjecture [ conjecture : limits_of_holes ] is ambiguous . on the one hand , a computer - aided search ( using schottky ) will only reveal the holes at any scale that are big enough to see , so one must develop heuristics to identify promising regions for exploration . on the other hand , failure to find holes near some given frontier point does not rule out the possibility that they might exist , but be very elusive . in a private communication , boris solomyak suggested that there might be no tiny holes accumulating at the point @xmath1046 in @xmath44 ; this is an especially good candidate counterexample to conjecture [ conjecture : limits_of_holes ] , since although it is algebraic and in fact a fixed point of renormalization the limit set of the corresponding 3-generator ifs _ is _ full , and in fact _ convex_. thus one could not hope to prove the existence of a renormalization sequence of holes , certified by loops of limit traps , limiting to @xmath1046 . on the other hand , very small holes _ can _ be found by hand , as close to @xmath1046 as the resolution allows the ( numerically certified ) hole at @xmath1047 is a good example . in this section we discuss the subtle problem of the structure of @xmath0 and @xmath21 near the real axis . in light of theorem [ theorem : interior_dense ] it might be surprising that the structure of @xmath0 and @xmath21 near the real axis can be very complicated . in fact , as was already observed by barnsley - harrington @xcite , there is an open neighborhood of the points @xmath1048 in @xmath1049 $ ] in which @xmath0 is totally real . we give an elementary proof of this fact , using the description of the limit set @xmath10 as the values of certain power series in @xmath1 , as described in section [ subsection : coefficients ] . getting a better estimate depends on analyzing a real 2-dimension ifs introduced by shmerkin - solomyak @xcite which we discuss and study in section [ subsection:2d_ifs ] . [ lemma : whiskers_isolated ] there is some @xmath1050 so that the intersection of @xmath0 with some open subset of @xmath65 is equal to the interval @xmath1051 . recall that for @xmath1052 the image @xmath1053 is the value of the power series @xmath1054 where the coefficients @xmath1055 are determined recursively from the infinite word @xmath166 by the method in proposition [ proposition : power_series ] . the key point is that the nonzero coefficients alternate between @xmath19 and @xmath233 , starting with @xmath19 . let @xmath1056 be real , for some small positive @xmath384 . the limit set @xmath10 is exactly equal to the unit interval , and @xmath1057 $ ] , @xmath1058 $ ] so that the intersection is exactly the interval @xmath1059 $ ] . the words @xmath166 with @xmath195 in the overlap all start with @xmath1060 or @xmath1061 for some big @xmath51 ( depending only on @xmath51 ) so that the power series are of the form @xmath1062 or @xmath1063 depending whether @xmath166 starts with @xmath16 or @xmath18 , and in the second case @xmath1064 ( we include the possibility that @xmath1065 ) . in the first case , @xmath1066 , while in the second case @xmath1067 for big @xmath51 and any fixed @xmath1068 . since the derivative is holomorphic in @xmath1 , this means that if we perturb @xmath1 to @xmath1069 for some small positive @xmath355 , the imaginary part of @xmath195 becomes _ positive _ for @xmath166 beginning with @xmath16 , and _ negative _ for @xmath166 beginning with @xmath18 ( at least for @xmath195 close to the interval @xmath1059 $ ] ) , so that the two sets @xmath88 and @xmath89 are disjoint , and we are in the complement of @xmath0 . we will push this argument further by analyzing the pairs @xmath1070 and @xmath1071 for left - infinite words @xmath448 starting with @xmath16 and @xmath18 respectively , and showing that for all real @xmath1 in the interval @xmath1072 $ ] the pairs are disjoint . shmerkin - solomyak @xcite introduce a 2-dimensional _ real _ ifs acting on @xmath1073 whose limit set is precisely the pairs @xmath1074 for @xmath1075 . explicitly , for real @xmath1076 , define @xmath1077 and let @xmath1078 denote the limit set of the ifs generated by @xmath1079 and @xmath1080 ( the notation is supposed to suggest the action of our familiar @xmath16 and @xmath18 on 1-jets ) . analogous to our standard notation , we will write @xmath1081 for the action of the word @xmath1082 on @xmath1083 for a parameter @xmath1084 . also , we write @xmath1085 , where the limit does not depend on @xmath154 . [ lemma:2d_ifs ] let @xmath1084 and suppose @xmath1086 and @xmath1087 are disjoint . then @xmath0 is totally real in an open neighborhood of @xmath1 . this is the same argument as that in used in the proof of lemma [ lemma : whiskers_isolated ] . since this condition is open , it can be certified numerically . thus , if we define @xmath1088 to be the subset of @xmath1076 for which @xmath1078 is connected , then @xmath1089 . one can characterize @xmath1088 as the set of real numbers @xmath1 of absolute value at most @xmath19 for which there is some power series @xmath1090 where each @xmath1091 for which @xmath1092 . we discuss later the question of whether there are points in @xmath1088 which do not lie in the closure of the interior of @xmath0 . analogous to @xmath1088 , one can study the subset @xmath1093 consisting of @xmath1 for which @xmath1078 contains the point @xmath1094 , and then @xmath1095 . shmerkin - solomyak @xcite define @xmath115 to be the smallest positive real number in @xmath1088 , and @xmath1096 to be the smallest real number such that @xmath1097 . experimentally they obtained estimates @xmath1098 we improved the estimate of @xmath115 to @xmath1099 getting a rigorous estimate of @xmath1096 is much harder , but experimentally we obtain @xmath1100 . to obtain these estimates , we used an algorithm which is perfectly analogous to algorithm [ algorithm : disconnectedness ] , and is proved in essentially the same way . to describe this algorithm , we use the following shorthand : @xmath1101 furthermore , for a @xmath1102 column vector @xmath127 we say @xmath127 is _ small _ if @xmath1103 and @xmath1104 , where @xmath881 is an integer . note that for @xmath1 real with @xmath132 , this latter inequality reduces to the analysis of a small fixed number of cases for @xmath881 . in the regime in which we are interested , @xmath1 will be quite close to @xmath1105 , so the relevant cases are @xmath1106 and @xmath1107 , and in practice the inequality reduces to @xmath1108 . @xmath1109 @xmath439 @xmath440 @xmath1110 @xmath1111 @xmath1112 @xmath444 @xmath445 true false the justification for this algorithm is essentially the same as that of algorithm [ algorithm : disconnectedness ] . to ask whether @xmath1113 is connected is equivalent to asking whether @xmath1114 , which is equivalent to asking whether the set of differences contains @xmath17 just as in section [ section : differences ] , the set of differences between points in @xmath1113 is a limit set itself . we denote the set of differences by @xmath1115 , and we note it is the limit set of the ifs generated by the three maps @xmath1116 where @xmath1117 note @xmath1118 . we obtain these maps by looking at how pairs of maps @xmath528,@xmath1119 , @xmath530 , and @xmath529 act on differences of points ; there are only three distinct maps . since @xmath1120 consists of differences between points in @xmath1113 whose corresponding infinite words begin with @xmath1080 and @xmath1079 , respectively , to check whether @xmath1113 is connected it suffices to check whether @xmath1121 . to determine whether @xmath1121 , we start with a box @xmath144 centered at @xmath1122 , which is sent inside itself under the three maps @xmath1123 . we want to consider @xmath1120 , so first we apply @xmath1124 . next , we subdivide @xmath1125 into its three subboxes , which are @xmath1126 , @xmath1127 , and @xmath1126 , and discard those which can not contain @xmath17 . we then subdivide again , and so on . suppose that @xmath127 is the center ( image of @xmath1122 ) of an image of @xmath144 under a word of length @xmath51 . since the centers of @xmath1128 are at @xmath1129 , respectively , the centers of the children of @xmath127 will be at the points @xmath1130 , @xmath127 , @xmath1131 . for simplicity , it makes sense to rescale the problem at every step by @xmath1132 . hence , we initialize the algorithm with the rescaled @xmath811 , i.e. @xmath1133 . then we add @xmath1134 , and rescale by @xmath45 again , and so on . any child which lies too far from the origin can be discarded , which is exactly what the smallness condition guarantees . the precise constants in the smallness condition follow from an analysis of how the rectangle @xmath1135 , @xmath1136 behaves under the maps @xmath1137 . it is easy to see that the infinite strip @xmath1138 is sent inside itself , so @xmath1115 lies inside this strip . then if we consider the images of the four points @xmath1139 , @xmath1140 , @xmath1141 , @xmath1142 , we find that the image with the largest second coordinate is @xmath1141 under the word @xmath1143 , and this image has second coordinate @xmath1144 . therefore , if we find the @xmath881 maximizing that expression and set @xmath1145 , then the limit set must lie in the rectangle @xmath1146\times [ -b , b]$ ] . if we run algorithm [ algorithm:2d_ifs ] on our numerical value for @xmath115 , the output is quite interesting . for the correct theoretical value of @xmath115 , the set @xmath427 of children viable to each depth will never be empty , and the same must be true for our numerical approximation ( of course , this is how we find the approximation in the first place ) . but what is not obvious from the definition ( although it is intuitively plausible ) is the experimental fact that the size of @xmath1147 is uniformly bounded independently of the depth @xmath203 , and there is apparently a _ unique _ lineage viable to infinite depth . if we denote the children @xmath1148 , @xmath1149 , @xmath1150 of the vector @xmath127 by @xmath1151 respectively , then the ( numerically ) unique viable descendent of the initial vector @xmath1133 to 194 generations is of the form @xmath1152 i.e. the first few terms are @xmath1153 . one can think of the values of @xmath360 as analogs of the terms in the continued fraction expansion of a number . in fact , the analogy is quite good : if any viable sequence for an initial vector @xmath1154 is eventually periodic , we obtain an identity of the form @xmath1155 for distinct polynomials @xmath314 with coefficients in @xmath237 , and therefore deduce that @xmath1156 is a root of @xmath1157 and is therefore algebraic . the branching of the algorithm is shown in figure [ figure:2d_disconnected_branching ] . in view of our experimental evidence , it seems reasonable to make the following conjecture : [ conjecture : unique_lineage ] for @xmath115 as above , there is a unique child at every stage with viable descendents to all future depths . furthermore , this viable lineage consists of the initial segments in the sequence @xmath1158 for some sequence @xmath1159 as above , where the terms are uniformly bounded . in a similar vein , we define @xmath584 to be the smallest positive real number in @xmath1160 , and @xmath1161 to be the smallest real number such that @xmath1162 . using similar methods we obtain the following estimates @xmath1163 ( the same caveat about @xmath1161 applies ) . it is easy to modify algorithm [ algorithm:2d_ifs ] to determine , for a given real @xmath1 , when there are infinite words @xmath425 so that @xmath1164 and @xmath1165 ; we need only consider children @xmath1166 and @xmath1167 for each @xmath127 in the stack @xmath427 , and otherwise the algorithm runs in exactly the same way . figure [ numerical_alpha ] gives numerical plots of the subset of the intervals @xmath1168 \cap \xi_2 $ ] and @xmath1169 \cap \omega_2 $ ] . ( top ) and @xmath1088 ( bottom ) . ] this figure strongly suggests that @xmath1170 $ ] is totally disconnected , while @xmath1171 $ ] appears to contain many solid intervals . in fact , our method of traps can be easily adapted to this more complicated ifs , and in section [ subsection : intervals_omega_2 ] we give a method to certify interior points in @xmath1088 . recall from section [ subsection:2d_ifs ] that @xmath1088 is the set of positive real numbers @xmath1068 for which the ifs @xmath1172 generated by the affine linear maps @xmath1173 is connected . by abuse of notation , we denote these generators by @xmath16 and @xmath18 throughout this section . note that both generators have constant jacobian @xmath1174 throughout this section we restrict attention to real @xmath1 in the interval @xmath1175 $ ] . the analog of lemma [ lemma : short_hop ] is the following : [ lemma : affine_short_hop ] with @xmath1176 $ ] , suppose that @xmath1177 and @xmath1178 contain points at distance @xmath355 apart in the @xmath1179 metric on @xmath1073 . then for any word @xmath160 of length at least @xmath1180 , the @xmath1181 neighborhood of @xmath1182 in the @xmath1179 metric is path connected . the proof is identical to that of lemma [ lemma : short_hop ] , except that one must take into account the fact that @xmath1183 does not uniformly contract the @xmath1179 metric . however , for @xmath1 in the interval in question , @xmath1184 multiplies the @xmath1179 metric by at most @xmath1185 which is @xmath1186 for @xmath1187 . the analog of proposition [ proposition : traps_a ] is the following : [ proposition : affine_shrinking_trap ] suppose for some @xmath1188 that there are words @xmath425 beginning with @xmath16 and @xmath18 of length at least @xmath1180 so that @xmath1182 and @xmath1189 cross transversely . then @xmath1 is an interior point in @xmath1088 . since @xmath1182 and @xmath1189 cross transversely , the same is true for their @xmath384-neighborhoods , for some sufficiently small fixed @xmath384 . thus the same is true for the @xmath585-neighborhoods of @xmath1190 and @xmath1191 whenever @xmath1192 is small enough , depending on @xmath1 and @xmath384 . thus , we choose such a @xmath1193 , and suppose @xmath355 is the @xmath1179 distance from @xmath1194 to @xmath1195 , where @xmath1196 . then the @xmath1197 neighborhoods of @xmath1190 and @xmath1191 are path connected , so by transversality , there is some point within @xmath1179 distance @xmath1197 from both @xmath1190 and @xmath1191 , and consequently the @xmath1179 distance from @xmath1190 to @xmath1191 is at most @xmath1198 . but then @xmath1199 so that @xmath611 and @xmath1200 , as claimed . the visually evident trap on the right certifies that this point lies in the interior of @xmath1088.,title="fig : " ] . the visually evident trap on the right certifies that this point lies in the interior of @xmath1088.,title="fig : " ] such affine traps may be found and certified numerically ; for example @xmath1201 satisfies the proposition for the words @xmath1202 and we deduce that @xmath1203 is an interior point in @xmath1088 ; see figure [ 669 ] . one might hope to prove an analog of bandt s conjecture ( i.e. theorem [ theorem : interior_dense ] ) for the set @xmath1088 ; that is , that the interior is dense in @xmath1088 . however our proof of theorem [ theorem : interior_dense ] uses in several ways the fact that points in the limit set are holomorphic functions of the parameter , which of course can no longer be true for the real parameter @xmath1 . nevertheless , such a proof does not seem beyond reach , and we comfortably conjecture : [ conjecture : omega_2_interior ] affine traps are dense in @xmath1088 , and hence the interior of @xmath1088 is dense in @xmath1088 . recall that @xmath1089 . it is not known whether there are any points in @xmath1088 which do not lie in the closure of the interior of @xmath0 . however , the following lemma relates this to conjecture [ conjecture : omega_2_interior ] . for clarity , we write @xmath1204 . [ lemma:2difs_is_limit_of_nonreal ] for every @xmath1205 and @xmath1206 , we have that @xmath1207 the rate of this convergence does not depend on @xmath160 . consequently , the limit set @xmath1208 scaled vertically by @xmath1209 converges in the hausdorff topology to @xmath1210 . it is an easy calculus exercise to show the lemma if @xmath1211 approaches @xmath469 vertically , i.e. if @xmath154 is fixed at @xmath977 . however , we desire convergence in general , so we will need to look at power series . write @xmath1212 as the power series @xmath1213 . then @xmath1214 the terms which contribute to the imaginary part of this sum are exactly those for which @xmath551 is odd . hence @xmath1215 the top limit on the inner sum indicates that we should run the inner sum until @xmath1216 is larger than @xmath881 . also note we are recording the imaginary part of @xmath1217 , so the @xmath1218 terms have disappeared . therefore , @xmath1219 the entire sum is controlled in absolute value by @xmath1220 , which is uniformly convergent for @xmath1221 . therefore , as @xmath1222 , the entire sum converges , at a rate controlled independently of @xmath160 , to @xmath1223 . the point in @xmath1210 associated with @xmath160 has coordinates @xmath1224 in @xmath1073 , and the point in the vertically scaled copy of @xmath1208 has coordinates @xmath1225 , and the lemma follows . if affine traps are dense in @xmath1088 , then near any point in @xmath1088 , by lemma [ lemma:2difs_is_limit_of_nonreal ] there are nonreal parameters which have a trap and therefore lie in the interior of @xmath0 . so every point in @xmath1088 would be in the closure of the interior of @xmath0 ; i.e. conjecture [ conjecture : omega_2_interior ] implies @xmath1226 . @xmath21 is path connected @xcite , but bousch s proof is somewhat indirect . his strategy is to show that every point can be joined by a path to some parameter with absolute value close to @xmath19 . since @xmath21 contains an annulus around the unit circle , this gives path connectedness . he does not directly address what the paths in @xmath21 actually look like , or when a ( polygonal ) path near @xmath21 can be approximated by a path contained in @xmath21 . in this section , we show how to certify the existence of a point in @xmath21 in a neighborhood of a given point and how to certify a path in @xmath21 in a neighborhood of a given polygonal path . if we can certify paths , we can certify loops , and thus exotic holes in @xmath21 . as with @xmath0 , by a _ hole _ in @xmath21 , we mean a connected component of the complement which is distinct from the connected component of the complement which contains @xmath17 . just as @xmath0 closely resembles the limit set @xmath42 at many points , @xmath21 closely resembles @xmath10 . thus the methods in this section are closely related to the methods we developed in section [ subsection : lz_paths ] to construct paths in @xmath10 ( e.g. proposition [ proposition : path_in_lambda ] ) . in this section , we prove a lemma in complex analysis , but we first motivate it . suppose we have a holomorphic function @xmath1227 , and we find that @xmath1228 is quite close to a value we desire @xmath137 . we would like to conclude that there is a @xmath696 near @xmath688 so that @xmath1229 . if the derivative of @xmath1230 is bounded away from @xmath17 , and does not vary much near @xmath688 , then @xmath1230 can be well approximated by a linear function , and @xmath696 can be found . thus to certify the existence of such a @xmath696 , and to prove the validity of the certificate , is not technically difficult . however , lemma [ lemma : nearby_root ] is organized carefully to be of use to us later , and it can be confusing to read . one should understand the lemma as saying `` if there are four constants @xmath1231 which satisfy the hypotheses , then the conclusion holds '' . do not worry about where the constants come from at this stage . this lemma is very similar to lemma 3.1 in @xcite . [ lemma : nearby_root ] let @xmath1230 be a holomorphic function and @xmath1232 with @xmath1233 . suppose there are @xmath1234 and @xmath1235 such that @xmath1236 for all @xmath1 with @xmath1237 , and @xmath1238 then there exists a unique @xmath1239 with @xmath1240 such that @xmath1229 . write @xmath1242 for the affine part of the power series for @xmath1230 , centered at @xmath688 , i.e. @xmath1243 . under @xmath1244 , the circle @xmath1245 of radius @xmath203 is mapped to the circle @xmath1246 . therefore , if @xmath1247 , the image circle will enclose @xmath17 , and hence @xmath1248 , or equivalently @xmath1244 will have a zero within @xmath1249 , the ball of radius @xmath203 centered at @xmath688 . now consider @xmath1230 ; it might not be affine , and we record the remainder term as @xmath1250 : @xmath1251 suppose that there were a radius @xmath203 such that for all @xmath1252 , we had @xmath1253 . in other words , the error in the affine approximation is smaller than the radius of the affine image circle minus @xmath384 . then the image of the circle @xmath1245 under @xmath1230 would have to contain @xmath1254 , and hence @xmath1230 would have a zero in @xmath1249 . additionally , this follows immediately from rouche s theorem , which also gives the claimed uniqueness . to prove the lemma , then , it suffices to find a @xmath203 such that @xmath1255 for all @xmath1252 . from taylor s theorem and cauchy s derivative estimates , there is an inequality @xmath1256 where @xmath1257 , and the estimate is valid whenever @xmath1258 , and we can also estimate @xmath1259 . set @xmath1260 . rearranging the inequality in the hypothesis of the lemma , we have @xmath1261 since @xmath1262 , and plugging in @xmath1260 , we have @xmath1263 therefore , @xmath1264 satisfies the necessary inequality , so there is @xmath1265 with @xmath1229 . since making @xmath490 smaller maintains the validity of the bounds @xmath1266 for @xmath1267 , we may shrink @xmath490 until the inequality in the lemma is an equality , so the claimed bound on @xmath1268 holds . the hypotheses of lemma [ lemma : nearby_root ] may seem somewhat technical , but in fact they are not difficult to check in practice . we set @xmath490 to be quite small but still large compared to @xmath384 , and we get bounds on the derivative . then @xmath355 can be found by trial and error or any minimum - finding algorithm . in fact , mathematica produces an explicit formula for the @xmath355 which minimizes the expression on the right of the inequality for @xmath490 ; this formula is rather large and unedifying , so we omit it . one feature we will make use of is that lemma [ lemma : nearby_root ] can be checked for large collections of elements in @xmath169 at the same time , since two words with a large common prefix will satisfy the same @xmath1266 bounds with similar values of @xmath384 . lemma [ lemma : nearby_root ] requires good derivative bounds on @xmath1269 a given ball @xmath1270 . a naive way to approach this is to get a universal upper bound @xmath318 on the second derivative and then state that @xmath1271 on @xmath1270 . this is typically a bad estimate because @xmath490 can be large compared to the potential change in @xmath1269 . here is a better way . since @xmath1267 is holomorphic , its maximum will lie on the boundary of @xmath1270 . cover the boundary circle of @xmath1270 with many ( say , @xmath1272 ) small balls , use the naive approach on these small balls , and take the maximum . because the radius on which we apply the naive approach is now quite small , our error will be much less . we now give some initial observations about paths in @xmath21 to clarify the construction to follow . to each point @xmath1 in @xmath21 , there is a set of distinguished words in @xmath169 ; namely , the words @xmath154 such that @xmath1273 . therefore , if we have a path @xmath1274 \to \d^$ ] such that the image of @xmath74 lies in @xmath21 , there is a combinatorial map @xmath1275 \to \partial \sigma$ ] such that @xmath1276 . of course , @xmath135 is not uniquely defined , as there may be more than one word mapping to @xmath20 for a given parameter . in order to build paths in @xmath21 , we essentially go in the other direction , given two words @xmath452 , we first build a nice combinatorial path interpolating between @xmath469 and @xmath977 . then , provided we are close enough to @xmath21 , we show how apply lemma [ lemma : nearby_root ] to produce a path of parameters which drags this combinatorial path along @xmath20 . then for all @xmath1288 , there is a unique @xmath1289 such that @xmath1290 . consequently , there is a map @xmath1291 such that @xmath1290 . furthermore , @xmath811 is uniformly continuous and the image @xmath1292 is path connected . that the map @xmath811 exists and is well - defined ( single - valued ) follows immediately from lemma [ lemma : nearby_root ] , so the content of the lemma is the uniform continuity and path connectedness . we first address the former . this is with respect to the cantor metric , so it suffices to show that if two words @xmath1293 have a sufficiently long common prefix , then their images under @xmath811 are close ( independent of what the prefix is ) . let @xmath318 be equal to @xmath1294 . we claim that there exists a constant @xmath628 such that if @xmath1293 have a common prefix @xmath229 of length at least @xmath628 , then @xmath1295 we now prove this claim . we remark that @xmath160 is a prefix of @xmath229 , since @xmath1296 already have the common prefix @xmath160 . by lemma [ lemma : diameter_bound ] , for a given @xmath1 , the limit set @xmath10 is contained in a ball of radius @xmath1297 centered at @xmath20 , so if @xmath160 is a word of length @xmath51 , then @xmath973 is contained in a ball of size @xmath1298 centered at @xmath434 . in our situation , then , the limit set @xmath1299 is contained inside a ball of radius @xmath1300 . therefore , we have @xmath1301 we are going to apply lemma [ lemma : nearby_root ] to @xmath1302 and @xmath1303 to get nearby roots , but there is a slight subtlety . we have derivative bounds on all words in @xmath1304 and @xmath1305 , but to apply lemma [ lemma : nearby_root ] , we need derivative bounds in a ball centered at @xmath1306 . we can achieve these bounds in the following way . since @xmath1307 , the derivative bounds @xmath49 and @xmath748 must be valid over @xmath1308 . so if @xmath1309 for @xmath628 sufficiently long enough , then @xmath1310 so we can apply lemma [ lemma : nearby_root ] to the words @xmath1296 at the point @xmath1306 with radius @xmath1311 and @xmath1312 ; this gives nearby @xmath1313 so @xmath1314 and @xmath1315 . but @xmath811 is uniquely defined , so @xmath1316 and @xmath1317 , and hence @xmath1318 the claim that @xmath811 is uniformly continuous follows from the triangle inequality , and therefore the image of @xmath811 is compact . it remains to show that the image @xmath1292 is path connected . analogous to the set @xmath488 we constructed to build paths through @xmath10 in section [ section : topology_of_lambda ] , given any two words @xmath1319 , we will construct a combinatorial path through @xmath1304 interpolating between them , and then show that applying @xmath811 to this path gives a continuous path in @xmath21 . given a finite word @xmath229 , denote by @xmath1320 the word obtained from @xmath229 by swapping @xmath16 and @xmath18 . note that if @xmath229 is finite and there is a parameter @xmath1 such that @xmath1321 , then @xmath1322 , so @xmath1323 and @xmath1324 . additionally , for any infinite word @xmath1325 obtained by taking an infinite power of @xmath229 and swapping arbitrary copies of @xmath229 for @xmath1320 , we have @xmath1326 . therefore , @xmath1327 . now let @xmath821 be a set of pairs of elements of @xmath1304 indexed by the dyadic rationals and constructed inductively as follows . first set @xmath1328 and @xmath1329 . next , given @xmath1330 and @xmath1331 , let @xmath196 be the maximal common prefix of @xmath1332 and @xmath1333 , and let @xmath1334 that is , @xmath1335 is either @xmath1336 or @xmath1337 depending on the first letters of @xmath1332 and @xmath1333 after the initial prefix . by the observation above , the map @xmath811 is well - defined on the pairs in @xmath821 because each pair consists of two words of the form @xmath1338 for the same @xmath229 . by induction , if @xmath1339 , then @xmath1340 and @xmath1341 have a common prefix of length at least @xmath1342 . here we say @xmath1340 and @xmath1341 have a common prefix of length @xmath51 if at least one of the four possible pairings of a word in @xmath1340 and @xmath1341 has a common prefix of length @xmath51 . since @xmath811 is uniformly continuous , this means that @xmath1343 is continuous as a function of the dyadic rational @xmath490 , so @xmath1343 extends continuously to @xmath1344 $ ] , and @xmath1292 is compact , so the image @xmath1343 is contained in @xmath1292 and is a path beginning at @xmath1345 and ending at @xmath1346 , and the lemma is proved . by a _ hole _ in @xmath21 , we mean a connected component of the complement which is distinct from the `` obvious '' large connected component containing the point @xmath17 . lemma [ lemma : lambda_image_in_b ] shows how to find a map @xmath811 which takes a set of words @xmath1304 into @xmath21 in a nice way . in order to find a hole in @xmath21 , we will find words @xmath1347 satisfying lemma [ lemma : lambda_image_in_b ] , thus giving maps @xmath1348 . the images @xmath1349 are path connected , and we will show , for all @xmath360 with @xmath360 taken modulo @xmath51 , that we have @xmath1350 . thus , there is a path passing through each image in turn . furthermore , we ll show that the images encircle a point which is not in @xmath21 . this will complete the proof of the existence of a hole in @xmath21 . lemma [ lemma : lambda_image_in_b ] does not say what the images @xmath1349 will look like ; it only gives balls which are guaranteed to contain them . to get a more precise picture , we do the following : enumerate all words @xmath1351 of some length @xmath263 , and apply lemma [ lemma : lambda_image_in_b ] to @xmath1352 for every @xmath1353 . if all these computations succeed , we obtain @xmath1354 balls , and we know that ( 1 ) there is a point in @xmath1355 inside each ball and ( 2 ) these points are connected by paths inside @xmath1349 . after the discussion above , this proof reduces to showing the pictures shown in figure [ figure : hole_in_b ] and asserting that they were produced using the method above . note that this produces a _ loop _ in @xmath21 , and checking if a parameter is _ not _ in @xmath21 is rigorous , so it suffices to exhibit a single pixel in the middle of the putative hole which is not in @xmath21 . many such pixels are easily visible . for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] for @xmath1357 , and the red boxes indicate the zoomed regions shown in the following pictures . each picture is made up of many small disks guaranteed to contain points in @xmath21 . four linked disks are highlighted in each picture to show that the various images of @xmath811 must intersect , and each image is path connected , so there is a loop in @xmath21.,title="fig : " ] 99 c. bandt , _ on the mandelbrot set for pairs of linear maps _ , nonlinearity 15 * ( 2002 ) , no . 4 , 11271147 m. barnsley and a. harrington , _ a mandelbrot set for pairs of linear maps _ , phys . d. * 15 * ( 1985 ) , no . 3 , 421432 t. bousch , _ paires de similitudes _ , preprint , 1988 ; available from the author s webpage t. bousch , _ connexit locale et par chemins hlderiens pour les systmes itrs de fonctions _ , preprint , 1992 ; available from the author s webpage d. calegari and a. walker , schottky , software available from v. climenhaga and y. pesin , _ lectures on fractal geometry and dynamical systems _ , student mathematical library , * 52 * ams , providence , ri , 2009 . d. epstein , j. cannon , d. holt , s. levy , m. paterson and w. thurston , _ word processing in groups _ , jones and bartlett , boston , 1992 p. mercat , _ semi - groupes fortement automatiques _ , bull . france * 141 * ( 2013 ) , no . 3 , 423479 a. odlyzko and b. poonen , _ zeros of polynomials with @xmath119 coefficients _ , lenseignement math . * 39 * ( 1993 ) , 317348 p. shmerkin and b. solomyak , _ zeros of @xmath237 power series and connectedness loci for self - affine sets _ , experimental math . * 15 * ( 2006 ) , no . 4 , 499511 b. solomyak , _ mandelbrot set for a pair of line maps : the local geometry _ , analysis thy . appl . * 20 * ( 2004 ) , 149157 b. solomyak , _ on the ` mandelbrot set ' for pairs of linear maps : asymptotic self - similarity _ , nonlinearity * 18 * ( 2005 ) , no . 5 , 19271943 b. solomyak and h. xu , _ on the ` mandelbrot set ' for a pair of linear maps and complex bernoulli convolutions _ , nonlinearity * 16 * ( 2003 ) , no . 5 , 17331749 w. thurston , _ entropy in dimension one _ , boundarys in complex dynamics : in celebration of john milnor s 80th birthday , pages 339384 . princeton university press , 2014 g. tiozzo , _ galois conjugates of entropies of real unimodal maps _ , preprint , arxiv:1310.7647	in 1985 , barnsley and harrington defined a `` mandelbrot set '' @xmath0 for pairs of similarities this is the set of complex numbers @xmath1 with @xmath2 for which the limit set of the semigroup generated by the similarities @xmath3 is connected . equivalently , @xmath0 is the closure of the set of roots of polynomials with coefficients in @xmath4 . barnsley and harrington already noted the ( numerically apparent ) existence of infinitely many small `` holes '' in @xmath0 , and conjectured that these holes were genuine . these holes are very interesting , since they are `` exotic '' components of the space of ( 2 generator ) schottky semigroups . the existence of at least one hole was rigorously confirmed by bandt in 2002 , and he conjectured that the interior points are dense away from the real axis . we introduce the technique of _ traps _ to construct and certify interior points of @xmath0 , and use them to prove bandt s conjecture . furthermore , our techniques let us certify the existence of infinitely many holes in @xmath0 . [ subsection ] [ theorem]lemma [ theorem]corollary [ theorem]conjecture [ theorem]proposition [ theorem]question [ theorem]problem [ theorem]definition [ theorem]construction [ theorem]notation [ theorem]convention [ theorem]remark [ theorem]example h l p
the discrete energy spectra inside the vortex core in two - dimensional superconductors play an important role for transport properties in the mixed state especially in low temperatures . the energy dissipation due to the vortex motion results from the transition between these levels . the @xmath2-@xmath3 characteristic in superconductors has been studied by many authors . in the superclean case @xmath4 , where @xmath5 is the level separation and @xmath6 is the inverse scattering time , the dissipative conductivity has been obtained quasiclassically as @xmath7 . @xcite for the 2d superconductor in the superclean limit where there is only one impurity inside the vortex core , the energy dissipation due to the landau - zener transition between levels has been evaluated . the energy transfer takes place when the impurity passes through the `` dangerous region '' , where many landau - zener transitions contributing to the dissipation . @xcite there is also a perturbational approach for the moving impurity which is valid in the high velocity limit @xmath8 where @xmath9 is the vortex velocity and @xmath10 is the mean level spacing . @xcite the transition between time - dependent levels @xmath11 , where @xmath12 is the external perturbation ( the position of impurities ) are treated in ref . in the framework of random matrix theory , and almost the same result for @xmath13 as the quasi - classical one is obtained . the parametric level statistics for the energy spectra under various external perturbations provide a new perspective of transport properties in disordered quantum systems such as metallic grains or quantum dots . the thouless formula which relates the conductance to an equilibrium quantity such as level curvatures gave the first clue to investigations of universal behaviors of level statistics . @xcite the random matrix theory ( rmt ) and its classification of distributions of level spacing ( wigner - dyson statistics ) , which plays a role of a finger - print also for quantum chaotic systems , supplies a powerful tool for the study of disordered systems in the framework of level dynamics . @xcite there have been both theoretical and numerical studies within the framework of the rmt for the level dynamics of the disordered systems such as an ring subject to the aharonov - bohm flux and non - interacting electrons in background potentials . @xcite for the quantum chaotic systems , there are experimental studies of the microwave billiards where the gaussian distribution of level velocities has been observed . @xcite the motivation of the current study is to investigate the universal behavior of the level velocities for low energy excitation spectra inside a moving vortex core dragged through a layered superconductor over a wide region from moderately clean to dirty phase . in the presence of disorder , the hamiltonian for the scattering quasi - particles between the excitation levels inside the vortex core becomes the lie algebra @xmath14 , and it belongs to the class c of the classification of the atland - zirnbauer.@xcite the level statistics over the crossover region from the moderately clean to the dirty phase has been investigated in the framework of the rmt @xcite , and also verified numerically for various impurity concentration inside the core . for the energy levels far from the fermi energy @xmath15 in the dirty limit case , the level statistics have been shown to be identical with that of the gaussian unitary ensemble ( gue ) . @xcite in section ii , we investigate the universal distribution of the level velocities for the quasiparticle spectra within the moving vortex core for different regimes which are characterized by the parameter @xmath16 ( or the number of impurities within the core @xmath17 ) , with constant vortex velocity @xmath9 . we observe that in the case of @xmath18 the gaussian distribution of level velocities is observed . in the clean case , non - gaussian distribution with an enhanced peak at @xmath19 which is the characteristic of the systems with localized wave - functions @xcite is obtained . in the diffusive time - evolution of energy spectrum within the core , the vortex velocity @xmath9 plays an essential role . it should be large enough that pinning is negligible , since @xmath9 is taken as a constant . however it is assumed to be moderately small @xmath20 in order to satisfy the adiabaticity of transitions between levels . the parameter @xmath21 is determined by the elastic mean free path and it manifests the mesoscopic nature of the level dynamics . @xcite in section iii , the energy dissipation for the low velocity regime is investigated and nonlinear @xmath2-@xmath3 characteristics are studied . we follow the same procedure as in ref . in which the energy dissipation has been evaluated through iterative calculations of time - dependent spectra from the schrdinger equation for a perturbed hamiltonian . our result coincides with that obtained for the superclean case in ref . in the small @xmath9 regime . in the dirty phase ( @xmath18 ) , our numerical result is identical to that of the random matrix theory for the gaussian unitary ensemble with moderately large @xmath9 . the excitation spectrum inside the vortex core is obtained from the bogoliubov - de gennes equation , @xcite @xmath22 where @xmath23 we consider the case for the @xmath24-wave gap parameter @xmath25 where @xmath26 is the polar angle . with the absence of impurity ( @xmath27 ) , the excitation energy spectrum @xmath28 is obtained . we set @xmath29 hereafter and @xmath30e^{-2k(r)}\bigg/ \int_0^\infty dr e^{-2k(r ) } , \\ k(r ) & = & \int_0^r dr ' \delta(r')/v_f .\end{aligned}\ ] ] if we take @xmath31 then the level spacing becomes @xmath32 . the @xmath33-th eigenfunction is @xmath34 which we call caroli - de gennes - matricon ( cdm ) basis where @xmath35 denotes a bessel function and @xmath36 is a normalization constant . when the impurity scattering is present , the wave function becomes a linear combination of the cdm basis as @xmath37 where @xmath33-th column of the matrix @xmath38 is given by the first order perturbation as @xmath39 @xmath40 @xmath41/\lambda_f \xi,\ ] ] with the assumption that impurity potential @xmath42 to be a sum of short - range scattering sites @xmath43 @xmath44 the summation for @xmath45 in eq.([eq : first ] ) is taken for all impurity sites and @xmath46 is the coherence length . when the vortex is dragged within the sample due to the lorenz force , the hamiltonian @xmath47 is perturbed by changing realizations of the impurity positions within the core . the perturbation parameter @xmath48 is the vortex coordinate and the energy levels go up and down at different times ( fig.1 ) . then the hamiltonian becomes time dependent with the vortex velocity @xmath9 fixed as a constant . the universal gaussian distribution of the level velocities was studied for disordered metallic rings subject to the aharonov - bohm flux . @xcite the level dynamics become universal dependent only on the wigner - dyson ensemble after a rescaling of the energy @xmath49 and the controlling parameter @xmath12 as @xmath50 where @xmath10 is the mean level spacing and @xmath51 is the normalized dispersion of level velocities , @xmath52 the statistical average @xmath53 is taken over samples and @xmath12 , i.e. over time steps as we fix the vortex velocity @xmath9 . it was shown that the equilibrium quantity @xmath51 is directly related to the transport ( dissipative ) conductivity given by kubo formula . @xcite thus @xmath51 is also called the universal conductance . the density of states of quasi - particles inside the vortex core in dirty or moderately clean phase has been investigated theoretically . @xcite it was also shown from the explicit diagonalization of the hamiltonian matrix that in the dirty limit the wigner - dyson statistics of level distributions become identical to that of the gaussian unitary ensemble when the levels are far from the fermi energy . @xcite following the same procedure we evaluate the distribution of the rescaled level velocities @xmath54 where @xmath55 is written as @xmath56 for various numbers of impurities @xmath17 within the vortex core . the inverse scattering time is given by @xmath57 @xmath58 is the concentration of the impurities within the core of radius @xmath46 , where @xmath46 is the coherence length . in the numerical calculations we set @xmath59 and @xmath60 . thus from the relation @xmath61 we have @xmath62 the size of the hamiltonian matrix @xmath63 is taken as @xmath64 . the vortex motion is restricted in @xmath65-direction and periodicity in this direction for the configuration of moving impurities is assumed . the discretized shift @xmath66 of an impurity at discretized time step @xmath67 is @xmath68 . the rescaled level velocity @xmath55 for @xmath69 time steps are calculated and the @xmath70 for levels between @xmath71 to @xmath72 are averaged , where we denote @xmath73 for the fermi level ( @xmath74 ) . the sample averaging has been taken over @xmath75 ( @xmath76 for @xmath77 case ) different initial configurations of the impurities . in fig.2 we show averaged distribution of the level velocities @xmath78 for various numbers of impurities within the core . we set @xmath79 and @xmath80 . the inset shows the rescaled variance @xmath51 for each energy level . the condition for the dirty phase @xmath81 where @xmath82 is the half of the size of the hamiltonian is well satisfied in the case @xmath77 @xmath83 . in this dirty phase the hamiltonian can be described as the random matrix and the gaussian distribution of the level velocities are observed . in the clean case @xmath84 @xmath85 , the non - gaussian behavior is found . we obtain a sharp peak at @xmath86 which is similar to the result of the microwave experiment for the case when the local level dynamics are concerned . @xcite this non - gaussian behavior suggests that the wave functions for the scattering quasiparticles in this regime is still localized . we evaluate the evolution of the wave function @xmath87 from the time dependent schrdinger equation @xmath88 where @xmath89 or equivalently consider time evolution of the coefficient matrix @xmath38 which depends on time . @xmath90 the initial state @xmath91 is assumed to be the @xmath92-th eigenvector of @xmath63 which corresponds to the @xmath92-th energy level @xmath93 . @xmath94 then the initial wave functions are @xmath95 within the core are modified and the hamiltonian @xmath63 becomes time dependent . we investigate the time evolution of @xmath96 which is expressed in the basis @xmath97 of the instantaneous hamiltonian @xmath98 : @xmath99 as the vortex moves at a constant velocity @xmath9 , the time - dependent spectrum shows diffusive time - evolution . this is the characteristic behavior which suggests that consecutive landau - zener transitions occur incoherently . the energy transfer takes place at the points called avoided crossings , where the separation between levels is much less than the mean level spacing @xmath10 . we assume that the quantum interference term in the transition probability between the consecutive levels can be ignored . the full quantum coherent time - evolution , however , leads to localization in energy space . the condition for this assumption is that many landau - zener transitions contribute to the diffusion , so we may take enough statistical average . @xcite thus the occupation probability @xmath100 spreads diffusively as time . following ref . , we define the probability distribution @xmath101 as @xmath102 @xmath101 is averaged over samples ( over different configurations of initial impurity sites ) , and/or over many initial states @xmath92 . the diffusive behavior of @xmath103 for small @xmath104 is expected to be linearly dependent on time , @xmath105 solving the time dependent schrdinger equation numerically , we investigate the dependence of the diffusion constant @xmath106 on the rate of change of the external perturbation , i.e. on vortex velocity @xmath9 . the diffusive behavior is also dependent on the smoothed density of states @xmath107 . in this study we assume that @xmath107 is constant as @xmath108 where @xmath10 is the mean level spacing . there are two different regimes which are characterized by the small @xmath9 or the large @xmath9 respectively . the calculation of the diffusion constant @xmath106 for the levels far from the fermi energy reveals the nonlinear behavior of the energy dissipation of quasi - particles inside the vortex core in the low @xmath9 regime . in fig.3 we plot the result for the clean phase @xmath84 and the dirty phase @xmath109 for various values of @xmath110 and @xmath111 . the initial state is @xmath112 and @xmath113 , respectively near the hump of @xmath51 plot for @xmath114 case which is shown by the insets . the horizontal axis is the rescaled vortex velocity @xmath115 . in the calculation of the numerical solution of eq.([eq : ts ] ) , the discrete time step is taken as @xmath116 and the sample averaging has been done over @xmath117 different initial configurations of the impurities . the gradient of the dotted line which is the least square fit is @xmath118 . we find that @xmath119 where @xmath120 for small @xmath9 from the fit . since the vortex velocity is written as @xmath121 , this result is consistent with the behavior of the dissipative part of the current density @xmath122 which is obtained in ref . as @xmath123 where @xmath124 is the distance from the `` dangerous region '' to the vortex center and @xmath125 is the flux quantum . this formula is valid for @xmath126 and in the super - clean limit @xmath127 case . since @xmath51 is suppressed with the decrease of @xmath16 ( or increase of @xmath17 ) , the random matrix prediction is irrelevant as well as in the large @xmath17 case for this low @xmath9 regime as expected . we also note that for larger values of @xmath9 , we obtain larger values of @xmath128 in all dirty cases of @xmath129 and @xmath130 for which we have done the calculation . in the meanwhile for the large @xmath9 regime the random matrix theory predicts the diffusion constant @xmath106 for the perturbed hamiltonian whose matrix elements are taken from the gaussian unitary ensemble ( this is the relevant for the current study ) as @xmath131 where @xmath51 is given by eq.([c0 ] ) and @xmath132 . the vortex damping coefficient @xmath133 is defined as @xmath134 where l.h.s is the rate of the energy dissipation inside the core and @xmath133 is related to the dissipative conductivity as @xmath135 . from the quasi - classical result in the @xmath18 case , the damping coefficient @xmath133 is given by @xcite @xmath136 where @xmath137 is the density of electrons . thus the random matrix theory predicts @xcite @xmath138 in fig.4 we plot @xmath51 v.s . @xmath16 for various @xmath9 values . in the large @xmath9 case the relation @xmath139 is explicitely observed . in this study we have investigated numerically for the universal behavior of the distribution of the rescaled level velocities for the excitation spectra within the moving vortex core in layered superconductors . in the dirty phase where @xmath18 we obtain the gaussian distribution consistent with the result of the theoretical study for the nonlinear @xmath140 model in the various disordered quantum systems . contrary to this , the non - gaussian behavior which is considered as the characteristic property which manifests the localization of wave functions inside the core is observed within the clean phase @xmath141 . as to the energy dissipation within the core due to the landau - zener transition between levels which occurs at the points called `` avoided crossings '' , the nonlinear dependence of the @xmath13 on the external perturbation ( the vortex velocity ) is investigated . this is explicitly related to the nonlinear @xmath2-@xmath3 characteristics in the mixed state in low temperature region . the dissipation properties are classified into three different regimes characterized by two parameters , the number of impurities inside the core @xmath17 ( or equivalently the inverse of @xmath142 ) , and the vortex velocity @xmath9 . case i)large @xmath9 and large @xmath17 regime ; in this regime the prediction of the rmt is valid and at the same time the behavior becomes consistent with those obtained from the approach within the kubo - greenwood formula . case ii)small @xmath9 and large @xmath17 regime ; in this regime the nonlinear behavior of the @xmath13 is observed . since the variance of the scaled level velocity @xmath51 is suppressed as the increase of @xmath17 , the random matrix theoretical result is irreverent in this regime no matter how @xmath17 becomes large . however , we consider that a crossover behavior for the dissipation properties within the core should exist in this regime , which has been found both theoretically and numerically in the case of the density of states . thus the third regime should be the crossover regime denoted as iii)large @xmath17 and moderately large @xmath9 regime . we consider the result of large @xmath143 values which are obtained in fig . 4 in large @xmath9 case implicitly manifest this crossover behavior . the crossover properties in this regime from the super - clean result of larkin - ovchinikov @xcite to the dirty case where rmt result should become valid is the issue of future study . kopnin and v.e . kravtsov , zh . pisma red . * 23 * , 631 ( 1976 ) [ jetp lett . * 23,578 ( 1976 ) ] . larkin and yu . n. ovchinnikov , phys . b 57 * , 5457 ( 1997 ) . f. guinea and yu . pogorelov , phys . lett . * 74 * , 462 ( 1995 ) . feigelman and m.a . skvortsov , phys . * 78 * , 2640 ( 1997 ) . j. t. edwords and d.j . thouless , j. phys . c * 5 * , 807 ( 1972 ) . thouless , phys . rep . * 13 * , 93 ( 1974 ) . mehta , _ random matrices _ ( academic press , new york , 1991 ) . f. haake , _ quantum signatures of chaos _ ( springer - verlag , 2000 ) . e. akkermans and g. montambaus , phys .rev . lett . * 68 * , 642 ( 1992 ) . a. szafer and b.l . altshuler , phys . * 70 * , 587 ( 1993 ) . simons and b.l . altshuler , phys .rev . lett.70 , 4063 ( 1993 ) , + b.d . simons and b.l . altshuler , phys . b * 48 * , 5422 ( 1993 ) . y.v . fyodorov and a.d . mirlin , phys . rev . b * 51 * , 13403 ( 1995 ) . m. barth , u. kuhl , and h .- j . stckmann , phys . rev * 82 * , 2026 ( 1999 ) . a. altland and m.r . zirnbauer , phys . b * 55 * , 1142 ( 1997 ) . e. brzin , s. hikami and a.i . larkin , phys . b * 60 * , 3589 ( 1999 ) . a. fujita , phys . b * 62 * , 15190 ( 2000 ) . m. wilkinson , phys . rev . a * 41 * , 4645 ( 1990 ) ; + m. wilkinson and e.j . austin , phys . a * 46 * , 64 ( 1992 ) . c. caroli , p.g . de gennes , and j. matricon , phys . lett . * 9,307 ( 1964 ) . m. a. skvortsov , v.e . kravtsov and m.v . feigelman , jetp lett . 68 * , 84 ( 1998 ) . a.a . koulakov and a.i . larkin , phys . b * 60 * , 14597 ( 1999 ) .	we study parametric level statistics of the discretized excitation spectra inside a moving vortex core in layered superconductors with impurities . the universal conductivity is evaluated numerically for the various values of rescaled vortex velocities @xmath0 from the clean case to the dirty limit case . the random matrix theoretical prediction is verified numerically in the large @xmath0 regime . on the contrary in the low velocity regime , we observe @xmath1 which is consistent with the theoretical result for the super - clean case , where the energy dissipation is due to the landau - zener transition which takes place at the points called `` avoided crossing '' .
according to contemporary cosmology , the universe is made of about @xmath5 dark energy , @xmath6 dark matter , and @xmath7 baryonic ( visible ) matter @xcite . thus , the overwhelming preponderance of matter and energy in the universe is believed to be dark , i.e. unobservable by telescopes . the dark energy is responsible for the accelerated expansion of the universe . its origin is mysterious and presumably related to the cosmological constant or to some form of exotic fluid with negative pressure such as the chaplygin gas @xcite . on the other hand , dark matter is necessary to account for the observed flat rotation curves of galaxies @xcite . its nature is one of the most important puzzles in particle physics and cosmology . many candidates for dark matter have been proposed , the most popular ones being the axions and the weakly interacting massive particles ( wimps ) @xcite . dark matter is usually modeled as a cold classical collisionless gas with vanishing pressure . in the cold dark matter ( cdm ) model , primordial density fluctuations are generated during the inflation and become the seeds of the bottom - up structure formation model . the cdm model with a cosmological constant ( @xmath8cdm ) successfully describes the accelerated expansion of the universe , the temperature fluctuations of the cosmic microwave background ( cmb ) , and the large - scale structures of the universe @xcite . however , it seems to encounter many problems at the scale of galactic or sub - galactic structures . indeed , cdm simulations @xcite lead to @xmath9 cuspy density profiles at galactic centers ( in the scales of the order of @xmath10 kpc and smaller ) while most rotation curves indicate a smooth core density @xcite . on the other hand , the predicted number of satellite galaxies around each galactic halo is far beyond what we see around the milky way @xcite . these problems might be solved , without altering the virtues of the cdm model , if the dark matter is composed of quantum particles such as fermions ( e.g. massive neutrinos ) or bosons ( e.g. axions ) . the wave properties of the dark matter may stabilize the system against gravitational collapse providing halo cores instead of cuspy profiles . in these models , the formation of dark matter structures at small scales is suppressed by quantum mechanics . therefore , quantum mechanics could be a way to solve the problems of the cdm model such as the cusp problem and the missing satellite problem . some authors have proposed that dark matter is a gas of bosons at @xmath11 forming bose - einstein condensates ( becs ) . in this scenario , dark matter halos may be understood as the ground state of some gigantic bosonic atom where the boson particles are condensed in a single macroscopic quantum state @xmath12 . at the scale of galaxies , gravity can be treated with the newtonian framework so the evolution of the wave function @xmath13 is governed by the gross - pitaevskii - poisson ( gpp ) system . using the madelung @xcite transformation , the gross - pitaevskii ( gp ) equation @xcite turns out to be equivalent to hydrodynamic ( euler ) equations involving an isotropic pressure due to short - range interactions ( scattering ) and an anisotropic quantum pressure arising from the heisenberg uncertainty principle . at large scales , quantum effects are negligible and one recovers the classical hydrodynamic equations of the cdm model which are remarkably successful in explaining the large - scale structures of the universe . at small scales , gravitational collapse is prevented by the repulsive scattering of the bosons or by the uncertainty principle . this model could solve the cusp problem and the missing satellites problem . the possibility that dark matter could be in the form of becs has a long history @xcite . in some works @xcite , it is assumed that the bosons have no self - interaction . in that case , gravitational collapse is prevented by the heisenberg uncertainty principle which is equivalent to a quantum pressure . this leads to dark matter halos with a mass - radius relation @xmath14 @xcite . in order to account for the mass and size of dark matter halos ( typically @xmath15 and @xmath16 ) , the mass of the bosons must be extremely small , of the order of @xmath17 @xcite . ultralight scalar fields like axions may have such small masses ( multidimensional string theories predict the existence of bosonic particles down to masses of the order of @xmath18 ) . this corresponds to `` fuzzy cold dark matter '' @xcite . in other works @xcite , it is assumed that the bosons have a repulsive self - interaction measured by a scattering length @xmath19 . in that case , gravitational collapse is prevented by the pressure arising from the scattering . in the thomas - fermi ( tf ) approximation , which amounts to neglecting the quantum pressure , the resulting structure is equivalent to a polytrope of index @xmath20 . the radius of the halo is given by @xmath21 , independent on its mass @xmath22 @xcite . for @xmath23 , corresponding to the values of the scattering length observed in terrestrial bec experiments @xcite , the mass and size of dark matter halos are reproduced if the bosons have a mass @xmath24 @xcite . this mass is much larger than the mass @xmath17 required in the absence of self - interaction . this may be more realistic from a particle physics point of view . the general mass - radius relation of self - gravitating becs at @xmath11 with an arbitrary scattering length @xmath25 , connecting the non - interacting limit ( @xmath26 ) to the tf limit ( @xmath27 ) , has been determined analytically and numerically in @xcite . these papers also provide the general density profile of dark matter halos interpreted as self - gravitating becs at @xmath11 . however , the bec scenario encounters serious problems . in the non - interacting case , the mass of the bosons must be extremely small , of the order of @xmath17 , in order to reproduce the properties of dark matter halos . the existence of particles with such small masses remains dubious ( although not impossible a priori ) . furthermore , the mass of the halo decreases with the radius which is in contradiction with the observations that reveal that the mass increases with the radius . on the other hand , for self - interacting becs in the tf approximation , the radius of the halos turns out to be independent on their mass , and fixed by the properties of the bosons ( their mass and scattering length ) . this is a major drawback of the bec model because it implies that all the halos should have the same radius ( unless the characteristics of the bosons change from halo to halo ) , which is clearly not the case . it is possible that the bec model at @xmath11 describes only dwarf dark matter halos . in order to describe large halos , finite temperature effects should be taken into account ( see also footnote 4 ) . finite temperature effects in the self - gravitating bose gas have been studied in @xcite using different approaches . when temperature effects are included in the model , the system takes a core - halo structure with a small condensed core ( equivalent to a bec at @xmath11 ) surrounded by an extended isothermal atmosphere of non - condensed bosons . these structures may be more realistic to describe dark matter halos . another possible scenario is that dark matter is made of fermions ( such as massive neutrinos ) instead of bosons . this model also solves the cusp problem and the missing satellite problem . in that case , gravitational collapse is prevented by the pauli exclusion principle . the distribution function satisfies @xmath28 where @xmath29 is the spin multiplicity of the quantum states ( in the numerical applications , we shall take @xmath30 and @xmath31 ) . the fact that the distribution function is bounded implies that the density can not diverge . at @xmath11 , the halos are completely degenerate and , except for a matter of scales , they are similar to classical white dwarf stars where gravitational collapse is prevented by the quantum pressure of the electrons @xcite . their mass - radius relation is @xmath32 @xcite . this model could describe dwarf dark matter halos . however , in order to describe large halos , like in the case of the bosonic scenario , it may be necessary to consider the fermi gas at finite temperature . indeed , the mass of a self - gravitating fermi gas at @xmath11 decreases with its size which is not consistent with observations . a detailed study of phase transitions in the self - gravitating fermi gas at finite temperature has been performed by chavanis @xcite . this study shows how a degenerate compact object forms as the energy and the temperature are reduced . originally , the self - gravitating fermi gas at finite temperature with neutrino masses in the @xmath33 range was proposed as a model for dark matter halos ( e.g. @xmath34 and @xmath35 ) and clusters of galaxies @xcite . then , it was suggested that degenerate superstars composed of weakly interacting fermions in the @xmath36 range could be an alternative to the supermassive black holes that are reported to exist at the center of galaxies ( e.g. @xmath37 and @xmath38 in our galaxy ) @xcite . finally , it was argued that a weakly interacting fermionic gas at finite temperature could provide a self - consistent model of dark matter that describes both the center and the halo of the galaxies @xcite . in that model , the system has a core - halo structure with a small condensed core ( equivalent to a fermion ball at @xmath11 ) surrounded by an extended isothermal atmosphere . since the density of a self - gravitating isothermal gas decreases as @xmath39 at large distances @xcite , this model is consistent with the flat rotation curves of the galaxies . on the other hand , since the core is degenerate in the sense of quantum mechanics ( pauli exclusion principle ) , it leads to flat density profiles at the center and avoids the cusp problem of cdm models . in addition , the gravitational collapse of fermionic matter leads to a compact object ( fermion ball ) at the center of galaxies that could be an alternative to a central black hole @xcite . one difficulty with the finite temperature self - gravitating bose and fermi gases is to explain how the particles have thermalized and how they have reached a statistical equilibrium state . indeed , the collisional relaxation time of a self - gravitating halo is usually very large and exceeds the age of the universe by many orders of magnitude @xcite . to solve this timescale problem , we propose that dark matter halos can be treated as a collisionless gas having experienced a form of violent relaxation . this process , introduced by lynden - bell @xcite in stellar dynamics , and worked out in @xcite , leads to a distribution function similar to the fermi - dirac distribution function . the coarse - grained distribution function satisfies @xmath40 where @xmath41 is the initial value of the distribution function before mixing . in that case , the origin of `` degeneracy '' is due to dynamical constraints ( liouville s theorem ) instead of quantum mechanics ( pauli s principle ) . this theory was initially developed to describe collisionless stellar systems such as elliptical galaxies for which the non - degenerate limit may be the most relevant @xcite . however , this approach with dynamical degeneracy retained could also apply to dark matter halos @xcite . in that case , gravitational collapse is prevented by lynden - bell s type of exclusion principle . furthermore , this scenario provides a much more efficient relaxation mechanism than the fermionic scenario . indeed , the violent relaxation of collisionless systems ( leading to the lynden - bell statistics ) takes place on a few dynamical times while the collisional relaxation of fermions ( leading to the fermi - dirac statistics ) is very long and possibly exceeds the age of the universe by many orders of magnitude . therefore , it is not clear how the fermions have thermalized in the whole cluster . in addition , in the fermionic scenario , the thermodynamical temperature @xmath42 is expected to be very low so that the halos would be completely degenerate and would appear very different from what is observed ( except in the case of dwarf halos ) . by contrast , in lynden - bell s theory , the temperature is an effective out - of - equilibrium temperature @xmath43 that can be much larger than the thermodynamical temperature . this could account for the value of the temperature inferred from the rotation curves of the galaxies by using the virial theorem . therefore , the lynden - bell theory predicts a large effective temperature ( even if @xmath11 initially ) , a density profile decreasing as @xmath39 at large distances consistent with the flat rotation curves of galaxies , and an effective exclusion principle at short distances that could avoid the cusp problem and lead to fermion balls mimicking black holes , just like in the fermionic scenario @xcite . as a result , the lynden - bell theory has the same properties as the fermionic theory while solving the timescale problem and the temperature problem @xcite . this makes this scenario very attractive . described by the schrdinger - poisson equation undergoes gravitational collapse ( jeans instability ) , oscillates , and settles into a compact bosonic object through the radiation of a complex scalar field . as a result , the system reaches a qss made of a solitonic core surrounded by a halo made of scalar radiation . the halo is similar to a thermal halo so this process may explain how self - gravitating bosons can `` thermalize '' and acquire a large effective temperature @xmath43 even if @xmath11 formally . the presence of the radiative halo may also explain why the mass of the halos increase with their radius . in the analogy between bosons and fermions , the soliton corresponds to the fermion ball and the halo made of scalar radiation corresponds to the isothermal halo predicted by lynden - bell s theory . ] the dark matter halos formed by jeans instability and violent relaxation can merge and create bigger structures . this is called hierarchical clustering . this is also a process of violent relaxation . if dark matter is collisionless , a large halo should not evolve anymore after having reached a virialized state . as a result , it can not have very high densities . in order to be more general , and because very little is known concerning the nature of dark matter , we consider the possibility that the core of dark matter halos can be collisional @xcite . this seems to be necessary to explain the presence of black holes at the center of large dark matter halos as proposed by balberg _ when collisional effects are taken into account , dark matter halos behave similarly to globular clusters . however , the collisions between particles do not correspond to two - body encounters as in globular clusters but rather to collisions similar to those in a gas . on the other hand , in fermionic dark matter halos , the pauli exclusion principle must be taken into account . as a result , collisions tend to establish a fermi - dirac distribution at finite temperature . this distribution is not very different from lynden - bell s distribution but collisions allow the central concentration of the system to evolve in time towards large values . in a recent series of papers , de vega and sanchez @xcite compared the predictions of the finite temperature self - gravitating fermi gas with observations of dark matter halos . they argued that small halos are degenerate quantum objects while large halos are non degenerate classical objects . assuming that the smallest known halos are completely degenerate , they found that the mass of the fermions must be of the order of @xmath44 corresponding possibly to sterile neutrinos . concerning the rotation curves , they obtained encouraging results showing that the description of dark matter halos in terms of the fermi - dirac distribution may be a good starting point . @xcite at large distances while the density profile of dark matter halos decreases as @xmath4 @xcite . ] de vega and sanchez justify the fermi - dirac distribution by quantum mechanics ( for a system of fermions at statistical equilibrium ) although , as explained above , it may be due to lynden - bell s form of relaxation . we shall consider the two possibilities since they lead to similar distribution functions @xcite . actually , quantum degeneracy and lynden - bell s type of degeneracy compete with each other @xcite . is fundamentally fixed by quantum mechanics ( pauli ) . indeed , the maximum distribution function of the spatially homogeneous collisionless gas prior to violent relaxation is @xmath45 since the gas has a relativistic fermi distribution @xmath46 @xcite . therefore , the lynden - bell bound is equal to half the pauli bound : @xmath47 . ] in their study , de vega and sanchez use the usual fermi - dirac distribution . however , when coupled to gravity , this distribution has infinite mass so that it can not constitute a physical model . furthermore , this infinite mass problem precludes the possibility of studying the stability of the cluster ( except if we enclose the cluster within an artificial `` box '' ) . for these reasons , we propose , as a next step , to describe dark matter halos by the fermionic king model which is a truncated fermi - dirac distribution . this model was introduced independently by ruffini and stella @xcite and chavanis @xcite . it can be viewed as a generalization of the classical king model to the case of fermions . this model has a finite mass so it is more realistic than the fermi - dirac distribution . the fermionic king model can be derived @xcite from a kinetic equation ( the fermionic landau equation ) assuming that the particles leave the system when their energy overcomes a critical escape energy @xmath48 . this derivation is valid both for quantum particles ( fermions ) and for collisionless self - gravitating systems undergoing lynden - bell s form of violent relaxation . in the non - degenerate limit , the fermionic king model reduces to the classical king model . the classical king model @xcite was introduced in the context of stellar systems in order to describe globular clusters made of classical point mass stars . on the basis of thermodynamics , we would expect that a system of classical point mass stars in gravitational interaction reaches a statistical equilibrium state described by the boltzmann distribution . however , it is well - known that no statistical equilibrium state exist for self - gravitating systems because the boltzmann entropy has no maximum in an unbounded domain ( the isothermal sphere , corresponding to the boltzmann distribution coupled to the poisson equation , has infinite mass ) @xcite . therefore , the statistical mechanics of self - gravitating systems is essentially an out - of - equilibrium problem @xcite . the absence of statistical equilibrium state is related to the fact that self - gravitating systems such as globular clusters have the tendency to evaporate . however , evaporation is a slow process and a globular cluster can be found , for intermediate times , in a quasi stationary state close to the michie - king distribution @xcite which is a truncated boltzmann distribution with parameters slowly changing with time . the caloric curve of the king model was determined by katz @xcite . like in the case of box - confined isothermal spheres @xcite , equilibrium states exist only above a critical energy @xmath0 . these configurations are metastable but their lifetime is considerable since it scales as @xmath49 ( except close to the critical point ) @xcite . for globular clusters , for which @xmath50 , this lifetime is so large that metastable states can be considered as stable states . in this sense , we can say that self - gravitating systems with @xmath51 do reach a statistical equilibrium state described by a truncated boltzmann distribution ( even if there is no statistical equilibrium state in a strict sense ) . however , because of evaporation , the energy of a self - gravitating system slowly decreases . for @xmath2 , there is no equilibrium state anymore . therefore , when the energy passes below this threshold , the system undergoes a gravothermal catastrophe @xcite and experiences core collapse @xcite . this corresponds to a saddle - node bifurcation . for classical self - gravitating systems , such as globular clusters , core collapse leads to the formation of a binary star surrounded by a hot halo ( at the collapse time , the singular density profile has infinite central density but zero central mass ) @xcite . the binary can release sufficient energy to stop the collapse and even drive a re - expansion of the cluster in a post - collapse regime @xcite . this is followed by a series of gravothermal oscillations @xcite . it is estimated that about @xmath52 of globular clusters are described by the king model while @xmath53 have undergone core collapse @xcite . for self - gravitating systems made of fermions ( white dwarfs , neutron stars , dark matter halos ) , the collapse stops when the core of the system becomes degenerate in virtue of the pauli exclusion principle . in that case , we have to take quantum mechanics into account . to treat dark matter , we propose to use the fermionic king model @xcite . the fermionic king model was studied by ruffini and stella @xcite who determined the density profiles of dark matter halos for various values of the central potential . our series of papers is intended to complete their study in the following directions . we determine the caloric curves of the fermionic king model for arbitrary values of the degeneracy parameter and study in detail the phase transitions that may occur between a gaseous phase unaffected by quantum mechanics and a condensed phase dominated by quantum mechanics . in this way , we generalize the study of phase transitions in the self - gravitating fermi gas at finite temperature confined within a box performed by chavanis @xcite . this generalization is important since the fermionic king model provides a more relevant description of dark matter halos than box models . therefore , we obtain realistic caloric curves of dark matter halos . we also determine the density profile , the velocity dispersion profile , and the circular velocity profile of the different configurations in order to compare the predictions of the fermionic king model to the observations of dark matter halos . in the present paper ( paper i ) , we consider the non - degenerate limit corresponding to the classical king model . the non - degenerate limit is expected to be valid for large dark matter halos so that it is a good starting point . in our companion paper @xcite ( paper ii ) , we consider the fermionic king model for arbitrary values of the degeneracy parameter . degeneracy effects are expected to be important for dwarf and intermediate - size dark matter halos . a short account of our results is given in @xcite . the paper is organized as follows . in sec . [ sec_sm ] , we discuss models of dark matter halos based on statistical mechanics and we introduce the classical and fermionic king models . in sec . [ sec_gen ] , we formulate the general problem of determining the structure of a spherically symmetric self - gravitating system described by a distribution function of the form @xmath54 with @xmath55 , where @xmath56 is the individual energy of the particles . we introduce the `` generalized entropy '' associated to this distribution and discuss its physical interpretation . in secs . [ sec_kingclass]-[sec_improper ] , we apply this general formalism to the classical king model and compute several quantities of interest . we show that the king model leads to configurations with an isothermal core , an isothermal halo , and a polytropic envelope of index @xmath57 . in sec . [ sec_comp ] , we compare the predictions of the classical king model to the observations of dark matter halos . because of collisions and evaporation , the central density increases while the slope of the halo profile decreases until an instability takes place . we show that dark matter halos are relatively well - described by a king model at , or close to , the point of marginal microcanonical stability . at that point , the king model generates a density profile that can be approximated by the modified hubble profile @xcite . this profile has a flat core and decreases as @xmath4 at large distances , like the observational burkert @xcite profile . less steep halos are unstable . the flat core is due to finite temperature effects , not to quantum mechanics . on the other hand , the large distance behavior of the density profile is due to the polytropic nature of the king distribution at high energies that departs from the isothermal boltzmann distribution . we conclude that statistical mechanics provides a good description of dark matter halos when evaporation is taken into account . the agreement is very good in the core of the system that is well - relaxed . the discrepancies that remain in the halo may be interpreted as a result of an incomplete relaxation , like in the case of stellar systems . although our results are exposed in the context of dark matter halos , our study of the classical king model presented in this paper also applies to globular clusters . we consider the possibility that dark matter halos can be described by the fermi - dirac distribution @xmath58 where @xmath59 gives the mass density of particles with position @xmath60 and velocity @xmath61 , @xmath62 gives the mass density of particles with position @xmath60 , @xmath63 is the gravitational potential determined by the poisson equation @xmath64 , @xmath65 is the maximum accessible value of the distribution function , @xmath56 is the individual energy of the particles by unit of mass , @xmath66 is the inverse temperature , and @xmath67 is the chemical potential ( fermi energy ) . in the non - degenerate limit @xmath68 , we can make the approximation @xmath69 implying @xmath70 and the fermi - dirac distribution reduces to the boltzmann distribution @xmath71 as recalled in the introduction , the fermi - dirac distribution may have two origins : ( i ) it may describe a gas of fermions at statistical equilibrium in which case @xmath72 is the maximum accessible value of the distribution function fixed by the pauli exclusion principle ; ( ii ) it may result from the violent relaxation of a collisionless system of particles ( classical or quantum ) as described by lynden - bell @xcite and worked out in @xcite . in that case , eqs . ( [ sm1 ] ) and ( [ sm2 ] ) are valid for the coarse - grained distribution function ( usually denoted @xmath73 ) and @xmath65 is the maximum value of the fine - grained distribution function . that is a superposition of `` fermi - dirac '' distributions @xcite . the single `` fermi - dirac '' distribution function ( [ sm1 ] ) is obtained when the fine - grained distribution function takes only two values @xmath74 and @xmath75 . it may also provide an approximation of more general cases where @xmath65 represents the maximum value of the fine - grained distribution function . the fine - grained distribution function coincides with the initial distribution function before the system has mixed . ] we shall consider the two possibilities since the distribution functions are formally the same . in the quantum interpretation , @xmath76 where @xmath42 is the thermodynamical temperature . in lynden - bell s interpretation , @xmath77 where @xmath43 is a generalized ( out - of - equilibrium ) `` temperature '' . in order to unify the notations , we write @xmath78 where @xmath42 has the dimension of an energy by unit of mass . when coupled to the gravity through the poisson equation , the fermi - dirac distribution ( [ sm1 ] ) has an infinite mass since it reduces to the boltzmann distribution ( [ sm2 ] ) at large distances where the system is diluted ( non - degenerate ) . as a result , the density decreases as @xmath79 for @xmath80 @xcite which is not normalizable . in order to avoid the infinite mass problem , we shall use a truncated fermi - dirac distribution . specifically , we use the fermionic king model that we write as ( see paper ii ) : @xmath81 @xmath82 where @xmath48 is the escape energy above which the particles are lost by the system and @xmath83 is a dimensionless parameter that measures the importance of degeneracy . the chemical potential ( fermi energy ) is related to the escape energy by @xmath84 ( see paper ii ) . for @xmath85 , we can make the approximation @xmath86 and we recover the fermi - dirac distribution ( [ sm1 ] ) . the fermionic king model was introduced heuristically by ruffini and stella @xcite as a natural extension of the classical king model to fermions in order to describe dark matter halos made of massive neutrinos . this distribution function was independently introduced by chavanis @xcite where it was derived from a kinetic equation ( the fermionic landau equation ) assuming that the particles leave the system when they reach a maximum energy @xmath48 . the kinetic derivation given in @xcite is valid either for quantum particles ( fermions ) or for collisionless self - gravitating systems ( classical or quantum ) experiencing lynden - bell s type of relaxation . this derivation can also be extended to non - condensed bosons by simply replacing @xmath87 by @xmath88 in the kinetic equation . this leads to the bosonic king model @xmath89 @xmath90 in the non degenerate limit @xmath91 , we can make the approximation @xmath92 and we recover the classical king model @xmath93 @xmath94 for @xmath85 , we can make the additional approximation @xmath86 and we recover the boltzmann distribution ( [ sm2 ] ) . the classical king model describes globular clusters and , possibly , large dark matter halos for which degeneracy effects ( due to the pauli exclusion principle for fermions or due to the liouville theorem for collisionless systems undergoing violent relaxation ) are negligible . before studying specifically the classical king model in sec . [ sec_kingclass ] , we formulate here the problem for a general distribution function of the form @xmath54 with @xmath55 describing spherical clusters . this will allow us to extend our study to various situations in future works without having to recall the general formalism at each time . we emphasize that the scalings derived below ( for the energy , the temperature , the tidal radius ... ) are `` universal '' , i.e. they do not depend on the precise form of the considered distribution function . for any functional of the form @xmath95 where @xmath96 is a convex function ( i.e. @xmath97 ) , we consider the following maximization problems @xmath98\ , \|\ , e[f]=e,\quad m[f]=m\rbrace \label{vp2}\ ] ] and @xmath99=s[f]-\beta e[f]\ , \|\ , m[f]=m\rbrace , \label{vp3}\ ] ] where @xmath100 is the energy ( @xmath101 is the kinetic energy and @xmath102 is the potential energy ) and @xmath103 is the mass . the critical points of the maximization problem ( [ vp2 ] ) are determined by the variational principle @xmath104 where @xmath66 and @xmath105 are lagrange multipliers associated with the constraints @xmath106 and @xmath22 . the critical points of the maximization problem ( [ vp3 ] ) are determined by the variational principle @xmath107 where @xmath105 is a lagrange multiplier associated with the constraint @xmath22 . obviously , the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) have the same critical points ( canceling the first order variations ) . they are given by the equation @xmath108 where @xmath56 is the individual energy of the particles by unit of mass . since @xmath109 is convex , this equation can be reversed to give @xmath110 where @xmath111 . we note that @xmath112 , so that @xmath113 keeps the same sign everywhere . since @xmath114 is positive and vanishes at the escape energy @xmath48 , we must have @xmath55 close to the escape energy . therefore , @xmath55 everywhere and , consequently , @xmath115 . in conclusion , the temperature is positive and the distribution function decreases monotonically with @xmath116 until it vanishes at @xmath48 . for future convenience , we write @xmath117 , where @xmath118 is a constant with the dimension of a distribution function and @xmath119 is a dimensionless function ( it can still depend on @xmath118 and on other `` external '' parameters ) . this amounts to writing @xmath120 and @xmath121 . then , the critical points of the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) are given by @xmath122 we assume that both @xmath118 and @xmath119 ( or equivalently @xmath118 and @xmath123 ) are given . in the maximization problem ( [ vp2 ] ) , @xmath66 and @xmath105 must be related to @xmath106 and @xmath22 . in the maximization problem ( [ vp3 ] ) , @xmath66 is prescribed and @xmath105 must be related to @xmath22 . a distribution function of the form of eq . ( [ vp9 ] ) is a ( local ) maximum of @xmath124 at fixed @xmath106 and @xmath22 if , and only if , @xmath125 for all perturbations @xmath126 that conserve mass and energy at first order , i.e. @xmath127 . a distribution of the form of eq . ( [ vp9 ] ) is a ( local ) maximum of @xmath128 at fixed @xmath22 if , and only if , the inequality of eq . ( [ vp10 ] ) is satisfied for all perturbations @xmath126 that conserve mass , i.e. @xmath129 . the derivation of these results can be found in @xcite . to study the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) , we shall use a thermodynamical analogy . we call @xmath124 the entropy , @xmath128 the free energy so that @xmath130 . the function @xmath128 is sometimes called the massieu function . to simplify the terminology we will call it here the free energy . ] , @xmath78 the inverse temperature , and @xmath131 the chemical potential . the maximization problem ( [ vp2 ] ) in which the energy and the mass are fixed is associated to the microcanonical ensemble ( mce ) and the maximization problem ( [ vp3 ] ) in which the temperature and the mass are fixed is associated to the canonical ensemble ( ce ) . we shall be interested by local and global maxima of entropy at fixed mass and energy in mce , and by local and global maxima of free energy at fixed mass in ce . different interpretations of the variational problems ( [ vp2 ] ) and ( [ vp3 ] ) are discussed in appendix a of paper ii . the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) determine distribution functions of the form @xmath54 with @xmath55 . such distribution functions , that depend only on the individual energy @xmath116 of the particles , describe spherically symmetric self - gravitating systems @xcite . inversely , any distribution function of the form @xmath54 with @xmath55 is a critical point of the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) for a specific entropy of the form of eq . ( [ vp1 ] ) . in practice , it is convenient to prescribe a form of distribution function @xmath132 , determine the corresponding entropy @xmath124 , and consider the variational problems ( [ vp2 ] ) and ( [ vp3 ] ) . this is how we shall proceed in sec . [ sec_kingclass ] and in paper ii . however , for the moment , we remain very general . as we have seen , a distribution function @xmath54 with @xmath55 can always be written in the form of eq . ( [ vp9 ] ) . we assume furthermore that @xmath114 vanishes at some escape energy @xmath48 and that @xmath75 for @xmath133 . . if @xmath135 denotes the zero of @xmath136 , we have @xmath137 . this relation shows that @xmath48 is not a new parameter but that it is equivalent to the lagrange multiplier @xmath105 ( for a given value of @xmath66 ) . in the following , we shall work in terms of @xmath48 and @xmath66 instead of @xmath105 and @xmath66 . introducing the shifted function @xmath138 , satisfying @xmath139 , we can write @xmath114 in the form @xmath140 @xmath141 the local density is defined by @xmath142 substituting eqs . ( [ fde1 ] ) and ( [ fde2 ] ) in eq . ( [ fde3 ] ) , we get @xmath143 where @xmath144 is the local escape velocity . these expressions are valid only for @xmath145 , where @xmath146 is the radius of the cluster such that @xmath147 , i.e. @xmath148 . this is the distance at which the density vanishes : @xmath149 . for @xmath150 , we have @xmath151 . in the king model , @xmath146 represents the tidal radius . making the change of variables @xmath152 , we obtain @xmath153 defining @xmath154 and @xmath155 , where the index @xmath156 refers to the center of the cluster , the foregoing equation can be rewritten as @xmath157 at that point , it is convenient to introduce the family of functions @xmath158 for future reference , we note the identity @xmath159 that can be established by a simple integration by parts . in terms of these functions , the density profile can be written as @xmath160 the central density is @xmath161 therefore , we obtain @xmath162 substituting these results in the poisson equation @xmath163 and introducing the rescaled distance @xmath164 where @xmath165 is the core radius , we obtain the fundamental ordinary differential equation @xmath166 with the boundary conditions @xmath167 this differential equation is defined for @xmath168 where @xmath169 is the dimensionless radius of the cluster determined by the condition @xmath170 . the function @xmath171 decreases monotonically with @xmath172 . the differential equation ( [ fde13 ] ) defines a one - parameter family of density profiles with parameter @xmath173 ( the normalized central potential ) going from @xmath156 to @xmath174 . the dimensionless radius @xmath175 is a function of @xmath173 . we also note that @xmath176 . for a spherically symmetric distribution function @xmath114 , the local pressure is defined by @xmath177 substituting eqs . ( [ fde1 ] ) and ( [ fde2 ] ) in eq . ( [ ene2 ] ) , and introducing the variables defined in sec . [ sec_fde ] , we obtain @xmath178 we note that the density @xmath179 and the pressure @xmath180 are functions of @xmath181 and @xmath42 ( for a given @xmath118 ) : @xmath182 and @xmath183 . eliminating @xmath181 between eqs . ( [ fde8 ] ) and ( [ ene3 ] ) we find that the cluster is described by a barotropic equation of state @xmath184 parameterized by the temperature @xmath42 ( for a given @xmath118 ) . we note the universal scaling @xmath185 . furthermore , one can easily check that the condition of hydrostatic equilibrium @xmath186 is automatically satisfied for a system described by a distribution function of the form @xmath54 ( see appendix [ sec_hydro ] ) . therefore , the differential equation ( [ fde13 ] ) may be derived equivalently from the fundamental equation of hydrostatic equilibrium @xcite with the equation of state specified above ( see appendix [ sec_hydro ] ) . using eqs . ( [ fde10 ] ) and ( [ fde12 ] ) , the mass profile @xmath187 is given by @xmath188 combining this equation with the differential equation ( [ fde13 ] ) , we get @xmath189 applying this equation at @xmath190 , and using eqs . ( [ fde12b ] ) and ( [ fde15 ] ) , we obtain @xmath191 . \label{temp3}\ ] ] this relation can also be derived from the gauss theorem @xmath192 applied at @xmath190 . the parameter @xmath193 is the dimensionless inverse temperature normalized by the size @xmath146 of the system . this is the correct dimensionless parameter when we work in a box of fixed radius @xmath146 @xcite . however , in the present problem , the size of the configuration @xmath146 is not a fixed parameter . the fixed parameter is @xmath118 , not @xmath146 . we need therefore to normalize the inverse temperature by @xmath118 . combining eqs . ( [ fde9 ] ) and ( [ fde15 ] ) , we find that @xmath194 substituting this relation in eq . ( [ temp3 ] ) , we obtain @xmath195 this equation relates the normalized inverse temperature @xmath196 to the parameter @xmath173 . the total energy @xmath197 can be computed as follows . using the virial theorem @xmath198 we have @xmath199 so we just need to compute the kinetic energy . the kinetic energy can be written in terms of the pressure defined by eq . ( [ ene2 ] ) as @xmath200 substituting eq . ( [ ene3 ] ) in eq . ( [ ene4 ] ) , and using eqs . ( [ fde12 ] ) and ( [ fde15 ] ) , we obtain @xmath201 4\pi\zeta^2\ , d\zeta . \label{ene5}\ ] ] according to eqs . ( [ temp3 ] ) , ( [ ene1 ] ) , and ( [ ene5 ] ) , the total energy normalized by @xmath146 is @xmath202\zeta^2\ , d\zeta . \label{ene6}\ ] ] this is the proper normalization of the energy when we work in a box of fixed radius @xmath146 @xcite . however , in the present problem , as explained previously , we must normalize the energy by @xmath118 , not by @xmath146 . using eqs . ( [ temp3])-([temp5 ] ) , we obtain @xmath203\zeta^2\ , d\zeta . \label{ene7}\end{aligned}\ ] ] this equation relates the normalized energy @xmath204 to the parameter @xmath173 . according to eqs . ( [ temp4 ] ) and ( [ temp5 ] ) , the radius of the cluster normalized by @xmath118 is given by @xmath205 this equation relates the normalized radius @xmath206 to the parameter @xmath173 . the normalized distance is then given by @xmath207 the escape energy is related to the radius of the cluster and to its mass by @xmath208 using eq . ( [ temp3 ] ) , we find that @xmath209 . therefore , the chemical potential ( times @xmath210 ) is equal to @xmath211 . on the other hand , using eq . ( [ temp5 ] ) , the normalized escape energy varies along the series of equilibria according to @xmath212 according to eqs . ( [ fde8 ] ) and ( [ temp5 ] ) , the normalized density profile is given by @xmath213}{\tilde\beta^{3/2}(k)}. \label{centraldensity0}\ ] ] the normalized central density is related to @xmath173 by @xmath214 in general , the central density is a monotonically increasing function of @xmath173 . therefore , the parameter @xmath173 can be interpreted as a measure of the central density . we shall call it the concentration parameter . it can also be interpreted as the normalized central potential ( with the opposite sign ) . the circular velocity is defined by @xcite : @xmath215 using eq . ( [ temp2 ] ) , we obtain @xmath216 according to eqs . ( [ fde12b ] ) and ( [ temp5 ] ) , the normalized circular velocity profile is given by @xmath217 we also note that @xmath218 . the local velocity dispersion ( in one direction ) of a spherically symmetric distribution function @xmath114 is defined by @xmath219 using eqs . ( [ fde10 ] ) and ( [ ene3 ] ) we obtain @xmath220}{i_1[\chi(\zeta)]}. \label{dis2}\ ] ] the central velocity dispersion is therefore @xmath221 according to eq . ( [ temp5 ] ) , the normalized velocity dispersion profile is given by @xmath222}{i_1[\chi(\zeta)]}.\label{dis4}\ ] ] the normalized central velocity dispersion is related to @xmath173 by @xmath223 instead of working with @xmath173 it is sometimes convenient to work in terms of the parameter @xmath225 that is more directly accessible to observations and numerical simulations . for example , this parameter was used by katz @xcite and cohn @xcite in their studies of globular clusters . since @xmath155 with @xmath226 , we get @xmath227 where @xmath228 is given by eq . ( [ temp3 ] ) . combining this relation with eq . ( [ dis3 ] ) , we obtain @xmath229 for the classical king model , @xmath224 is a monotonically increasing function of @xmath173 so it can be used equivalently to parameterize the series of equilibria . for an extended classical king cluster ( large @xmath173 ) , we have @xmath230 and @xmath231 so that @xmath232 . if we define the kinetic temperature @xmath233 through the relation @xmath234 ( where @xmath101 denotes the kinetic energy ) , we find from the virial theorem ( [ ene1 ] ) that the kinetic caloric curve is simply given by @xmath235 . therefore , the kinetic specific heat is @xmath236 it has a constant negative value . however , @xmath233 is not the thermodynamic temperature in the present case . the thermodynamic temperature is @xmath237 and the thermodynamic specific heat is @xmath238 when the distribution function is non - boltzmannian , the kinetic and thermodynamic caloric curves @xmath239 and @xmath240 can be very different . as we shall see in sec . [ sec_kingclass ] for the king model , the thermodynamic specific heat is not constant and differs from eq . ( [ heat1 ] ) . the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) have the same critical points . they correspond to the distribution function ( [ vp9 ] ) . however , these maximization problems may not be equivalent . the stability of the distribution function ( [ vp9 ] ) may differ in mce and ce . as a result , the set of solutions of ( [ vp2 ] ) may not coincide with the set of solutions of ( [ vp3 ] ) . it can be shown that the solution of a maximization problem is always the solution of a more constrained dual maximization problem @xcite . therefore , a solution of ( [ vp3 ] ) with given @xmath66 is always a solution of ( [ vp2 ] ) with the corresponding @xmath106 . in the thermodynamical analogy , this means that `` canonical stability implies microcanonical stability '' : @xmath241 . ) is satisfied for all perturbations @xmath126 that conserve mass ( canonical stability criterion ) , it is _ a fortiori _ satisfied for all perturbations that conserve mass _ and _ energy at first order ( microcanonical stability criterion ) . ] however , the converse in wrong : a solution of ( [ vp2 ] ) is not necessarily a solution of ( [ vp3 ] ) . when this happens , we speak of ensemble inequivalence . ensemble inequivalence is generic for systems with long - range interactions but it is not compulsory . in order to determine the stability of a distribution function according to the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) we can use the theory of poincar on the linear series of equilibria @xcite . this is a powerful graphical method that just requires to determine the critical points of ( [ vp2 ] ) and ( [ vp3 ] ) and plot the series of equilibria @xmath242 . this theory uses the fact that @xmath243 in mce ( the inverse temperature is the conjugate of the energy with respect to the entropy ) and @xmath244 in ce ( minus the energy is the conjugate of the inverse temperature with respect to the free energy ) . it can be shown that a change of stability can occur only at a turning point or at a bifurcation point of the series of equilibria . in this paper and in paper ii , we shall only encounter the case of turning points . if we plot @xmath66 as a function of @xmath245 , we have the following results . in mce , a change of stability can only occur at a turning point of energy where @xmath246 . a mode of stability is lost if the curve rotates clockwise and gained if it rotates anti - clockwise . in ce , a change of stability can only occur at a turning point of temperature where @xmath247 . a mode of stability is lost if the curve rotates clockwise and gained if it rotates anti - clockwise . we refer to katz @xcite and chavanis @xcite for an application of the poincar theory to the case of self - gravitating systems . in this section , we apply the general formalism developed previously to the case of the classical king model . the classical king model is defined by @xmath248 @xmath249 where @xmath48 is the escape energy at which the particles leave the system ( @xmath75 ) . for @xmath250 , the king distribution reduces to the boltzmann distribution @xmath251 and , for @xmath252 , it reduces to @xmath253 corresponding to a polytropic distribution of index @xmath57 @xcite . therefore , the king model generically describes a cluster with an isothermal core , an isothermal halo , and a polytropic envelope of index @xmath57 . the proportion of these different regions depends on the concentration parameter @xmath173 as shown in the sequel . the distribution function @xmath114 is represented in fig . [ feps ] . in scaled variables showing the isothermal core , the isothermal halo , and the polytropic envelope . ] the king distribution is of the form of eqs . ( [ fde1 ] ) and ( [ fde2 ] ) with @xmath254 the corresponding entropy is given by eq . ( [ vp1 ] ) with ( see paper ii ) : @xmath255 for the king model , the functions @xmath256 defined in the general case by eq . ( [ fde7 ] ) can be written , after an integration by parts , as @xmath257 these functions may be expressed in terms of the error function . their asymptotic behaviors for small and large values of @xmath258 are easily obtained . for @xmath259 , we get @xmath260 so that @xmath261 and @xmath262 . for @xmath263 , we get @xmath264 so that @xmath265 and @xmath266 . the density profile of the king model is given by eq . ( [ fde10 ] ) where @xmath267 is the solution of the differential equation ( [ fde13])-([fde14 ] ) with the function @xmath268 defined by eq . ( [ df5 ] ) . the density profile vanishes at a radius @xmath190 corresponding to the tidal radius . the phase space portrait of the king model is represented in fig . [ pskmce ] . ( see below ) . the velocity is normalized by @xmath269 and the radial distance by @xmath270 . the envelope where @xmath75 , corresponding to @xmath271 , is given by @xmath272 in scaled variables . ] the equation of state @xmath273 of the king model is defined by the parametric equations ( [ fde8 ] ) and ( [ ene3 ] ) where @xmath268 and @xmath274 are given by eq . ( [ df5 ] ) . for @xmath275 , we find that @xmath276 leading to the isothermal equation of state @xmath277 this equation of state is valid at high densities . for @xmath278 , we find that @xmath279 leading to the polytropic equation of state @xmath280 this equation of state is valid at low densities . in scaled variables showing the isothermal core , the isothermal halo , and the polytropic envelope . ] for @xmath281 the density is related to the gravitational potential by the boltzmann distribution @xmath282 and for @xmath283 the density is related to the gravitational potential by the distribution @xmath284 corresponding to a polytrope of index @xmath57 . the relation @xmath285 is represented in fig . [ rhophi ] . in the limit @xmath286 , the function @xmath267 is always small , so we can use the approximation ( [ df6 ] ) everywhere . as a result , the king model is equivalent to a pure polytrope ( @xmath287 ) of index @xmath57 and polytropic constant @xmath288 . defining @xmath289 and @xmath290 , we find that the differential equation ( [ fde13 ] ) reduces to the lane - emden equation @xmath291 @xmath292 corresponding to a polytrope @xmath57 @xcite . solving this equation numerically , we obtain @xmath293 and @xmath294 . using the theory of polytropes , we can analytically obtain the mass - radius relation and the expression of the energy . this allows us to obtain an analytical expression of the series of equilibria @xmath295 for @xmath286 . we proceed as follows . using the virial theorem @xmath198 , the total energy @xmath197 is given by @xmath296 the potential energy of a polytrope of index @xmath297 is @xmath298 specializing to the index @xmath57 , and using eq . ( [ poly3 ] ) , we obtain @xmath299 the mass - radius relation of a polytrope of index @xmath297 is @xmath300 where @xmath301 . for @xmath57 , we have @xmath302 . using the expression of @xmath101 in terms of @xmath118 and @xmath66 given above , eq . ( [ poly6 ] ) takes the form @xmath303 with @xmath304 . combining eqs . ( [ poly5 ] ) and ( [ poly7 ] ) , we get @xmath305 introducing the dimensionless temperature , the dimensionless energy , and the dimensionless radius defined by eqs . ( [ temp5 ] ) , ( [ ene7 ] ) and ( [ rad1 ] ) , eqs . ( [ poly7 ] ) and ( [ poly9 ] ) lead to @xmath306 @xmath307 the radius and the energy are related by @xmath308 . according to eqs . ( [ temp5 ] ) and ( [ df6 ] ) , we also have @xmath309 from which we get @xmath310 and @xmath311 . these relations are valid for @xmath286 , hence for @xmath312 , @xmath313 , and @xmath314 . we also note that @xmath315 and @xmath316 . in the limit @xmath318 , the function @xmath267 is always large , except close to the tidal radius , so we can use the approximation ( [ df8 ] ) in almost all the cluster . as a result , the king model is almost equivalent to an isothermal sphere ( @xmath319 ) . defining @xmath320 and @xmath321 , we find that the differential equation ( [ fde13 ] ) reduces almost everywhere to the emden equation @xmath322 @xmath323 corresponding to the isothermal sphere @xcite . in this section , we develop the proper thermodynamic treatment of the king model associated to the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) . as explained in sec . [ sec_gen ] , in order to solve these maximization problems , we must work at fixed @xmath118 , not at fixed @xmath146 . therefore , the thermodynamical parameters must be normalized by @xmath118 , not by @xmath146 . accordingly , the thermodynamical parameters denoted @xmath66 , @xmath106 and @xmath146 in this section correspond to the dimensionless parameters @xmath324 , @xmath325 and @xmath326 defined by eqs . ( [ temp5 ] ) , ( [ ene7 ] ) and ( [ rad1 ] ) . on the other hand , @xmath124 and @xmath128 refer to @xmath327 and @xmath328 . in figs . [ aba ] and [ aea ] , we plot the inverse temperature @xmath66 and the energy @xmath245 as a function of the normalized central potential @xmath173 parameterizing the series of equilibria . for small @xmath173 , the system is equivalent to a polytrope of index @xmath57 and the functions @xmath329 and @xmath330 are approximately given by eqs . ( [ poly10 ] ) and ( [ poly11 ] ) represented as dashed lines in figs . [ aba ] and [ aea ] . for large @xmath173 , the system is similar to the isothermal sphere ( @xmath331 ) . as for a classical isothermal sphere confined within a box ( see , e.g. , figs . 3 and 5 in @xcite ) , the curves @xmath329 and @xmath330 present damped oscillations about some asymptotes @xmath332 and @xmath333 . for the king model , @xmath334 and @xmath335 . the temperature has a first peak at ( @xmath336 , @xmath337 ) and the energy has a first peak at ( @xmath338 , @xmath339 ) . for box - confined isothermal spheres , we have @xmath340 and @xmath341 where @xmath342 . instead of parameterizing the series of equilibria by @xmath173 , we can use the parameter @xmath224 defined in sec . [ sec_k ] . for the king model , the function @xmath343 was computed by katz @xcite . it is recalled in fig . [ kk ] for completeness . from this figure , we find that @xmath344 and @xmath345 . using the results of sec . [ sec_kingclass ] , we can easily establish that @xmath346 and @xmath347 for @xmath348 . as a function of the concentration parameter @xmath173 for the classical king model . the dashed line corresponds to the analytical formula obtained in the polytropic approximation . ] as a function of the concentration parameter @xmath173 for the classical king model . the polytropic approximation is valid for relatively low values of @xmath106 , before the inflection point of the curve @xmath330 occurring at about @xmath349 and @xmath350 . ] and the bullets indicate the limits of canonical and microcanonical stability . ] the series of equilibria @xmath351 is plotted in fig . this curve updates the one given by katz @xcite that was drawn by hand for large values of @xmath173 . this curve has a snail - like structure ( spiral ) similar to the series of equilibria of classical isothermal spheres confined within a box ( see , e.g. , fig . 1 in @xcite ) . we note , however , that the energy is always negative in the present case . this is a consequence of the virial theorem ( [ ene1 ] ) for a self - confined system . by contrast , for box - confined isothermal spheres , there is an additional term in the virial theorem due to the pressure against the boundary so the energy can be either positive or negative . the concentration parameter @xmath173 increases along the series of equilibria . for small @xmath173 , the system is equivalent to a polytrope of index @xmath57 and the function @xmath242 is approximately given by eq . ( [ poly10 ] ) represented as a dashed line in fig . this is valid for @xmath352 and @xmath353 . for large @xmath173 , the system approaches an isothermal sphere ( @xmath331 ) and the series of equilibria spirals about the limit point ( @xmath354 , @xmath355 ) . ) giving the inverse temperature @xmath66 as a function of the energy @xmath245 for the classical king model . ] in mce , where the control parameter is the energy @xmath106 , there exist equilibrium states only for @xmath51 with @xmath339 ( first turning point of energy ) . the critical energy @xmath0 is the equivalent of the emden energy for box - confined isothermal spheres @xcite . for @xmath2 there is no equilibrium state and the system undergoes a gravothermal catastrophe . for classical particles , this leads to a singularity corresponding to a tight binary surrounded by a hot halo . of a self - gravitating system at fixed mass and energy by approaching two particles at a very close distance to each other and redistributing the released energy in the halo in the form of kinetic energy ( see appendix a of @xcite ) . the binary has a small mass @xmath356 but a huge potential energy @xmath357 . since the total energy is fixed in mce , the kinetic energy ( temperature ) of the halo @xmath353 and , consequently , the entropy @xmath358 . since the halo is `` hot '' , it has the tendency to extend at large distances . it can be shown @xcite that the divergence of entropy is maximum when the mass in the core is the smallest , corresponding , in the discrete case , to a binary . we can also formulate this argument in terms of the density of states @xcite . the density of states of a self - gravitating system diverges for @xmath359 because we can form a pair with a binding energy tending to @xmath360 and transfer this energy to the kinetic energy of the other stars which grows to @xmath174 to maximize the volume of phase space they explore . ] in ce , where the control parameter is the temperature @xmath42 , there exist equilibrium states only for @xmath361 with @xmath362 ( first turning point of temperature ) . the critical temperature @xmath1 is the equivalent of the emden temperature for box - confined isothermal spheres @xcite . for @xmath3 there is no equilibrium state and the system undergoes an isothermal collapse . for classical particles , this leads to a singularity corresponding to a dirac peak containing all the mass . of a self - gravitating system at fixed mass by collapsing all the particles at the same point ( see appendix b of @xcite ) . it can be shown @xcite that the divergence of free energy is maximum when the mass in the core is the largest . we can also formulate this argument in terms of the partition function @xcite . the partition function of a self - gravitating system diverges for @xmath363 when all the particles are concentrated at the same point @xcite . ] we now investigate the stability of the classical king distributions according to the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) . the stable part of the series of equilibria in each ensemble defines the caloric curve . we first consider the canonical ensemble ( [ vp3 ] ) in which the control parameter is the temperature @xmath42 . for @xmath353 the system is stable in ce since it is equivalent to a polytrope with an index @xmath57 smaller than the critical value @xmath364 in ce @xcite . using the poincar theory , we conclude that the series of equilibria is stable until the first turning point of temperature ce and that it becomes unstable after that point . in other words , the king distribution is a maximum of free energy at fixed mass for @xmath365 and a saddle point of free energy at fixed mass for @xmath366 . since the series of equilibria always rotates clockwise , a mode of stability is lost at each turning point of temperature , so the system is more and more unstable as @xmath173 increases . we now consider the microcanonical ensemble ( [ vp2 ] ) in which the control parameter is the energy @xmath106 . for @xmath367 the system is stable in mce since it is equivalent to a polytrope with an index @xmath57 smaller than the critical value @xmath368 in mce @xcite . using the poincar theory , we conclude that the series of equilibria is stable until the first turning point of energy mce and that it becomes unstable after that point . in other words , the king distribution is a maximum of entropy at fixed mass and energy for @xmath369 and a saddle point of entropy at fixed mass and energy for @xmath370 . since the series of equilibria always rotates clockwise , a mode of stability is lost at each turning point of energy , so the system is more and more unstable as @xmath173 increases . accordingly , there exist a region of ensemble inequivalence between points ce and mce in fig . [ eba ] , i.e. for configurations with @xmath371 , where @xmath336 and @xmath338 ( we check , in passing , that @xmath372 since a canonical equilibrium is always a microcanonical equilibrium ) . this part of the series of equilibria is stable in mce ( entropy maxima at fixed mass and energy ) but unstable in ce ( saddle points of free energy at fixed mass ) . it corresponds to configurations with negative specific heat @xmath373 . we know that such configurations are forbidden in ce while they are allowed in mce . these results are very similar to those obtained for box - confined isothermal spheres ( see , e.g. , the reviews @xcite ) . since there is no global maximum of free energy at fixed mass for classical self - gravitating systems ( see footnote 16 ) , the configurations with @xmath365 in ce are only metastable ( local maxima of free energy at fixed mass ) . similarly , since there is no global entropy maximum at fixed mass and energy for classical self - gravitating systems ( see footnote 15 ) , the configurations with @xmath369 in mce are only metastable ( local maxima of entropy at fixed mass and energy ) . however , the probability to cross the barrier of free energy in ce , or the barrier of entropy in mce , and leave a metastable state , is a very rare event as it scales as @xmath374 @xcite . for self - gravitating systems with a large number of particles ( for example globular clusters contain about @xmath375 stars and the number of particles in dark matter halos is much larger ) this probability is totally negligible . therefore , in practice , metastable states are stable states @xcite . in this sense , self - gravitating systems described by the king model with @xmath365 in ce and with @xmath369 in mce can be considered to be at statistical equilibrium , even if there is no statistical equilibrium state in a strict sense . their lifetime is controlled by evaporation and gravitational collapse as discussed in the introduction . the physical caloric curve in ce corresponds to the part of the series of equilibria represented in fig . [ eba ] up to point ce and the physical caloric curve in mce corresponds to the part of the series of equilibria represented in fig . [ eba ] up to point mce . they are made of long - lived metastable states . in semi - log scales . ] in figs . [ esa ] and [ bja ] , we plot the entropy @xmath124 as a function of the energy @xmath245 in mce and the free energy @xmath128 as a function of the inverse temperature @xmath66 in ce . since @xmath376 ( for a fixed mass @xmath22 ) in mce , we find that @xmath377 is extremum when @xmath330 is extremum . similarly , since @xmath378 ( for a fixed mass @xmath22 ) in ce , we find that @xmath379 is extremum when @xmath329 is extremum . this explains the `` spikes '' observed in figs [ esa ] and [ bja ] . similar spikes are found for box - confined isothermal spheres in newtonian gravity ( see figs . 4 and 6 in @xcite ) and for box - confined self - gravitating systems described by a linear equation of state in general relativity ( see fig . 5 in @xcite ) . the series of equilibria becomes unstable after the first spike in each ensemble . this is in agreement with the fact that the states on the unstable branches ( after the first spike ) have lower entropy or lower free energy than the states on the stable branch ( before the first spike ) . in fig . [ betarayon ] , we plot the tidal radius @xmath146 as a function of the inverse temperature @xmath66 in ce . for @xmath353 the tidal radius tends to @xmath174 and it decreases as @xmath42 decreases . at the critical temperature @xmath1 the value of the tidal radius is @xmath380 . this is the minimum stable value of the radius in ce ( as we have seen previously , the part of the curve situated after the turning point of temperature is unstable ) . in fig . [ energyrayon ] , we plot the tidal radius @xmath146 as a function of the energy @xmath245 in mce . for @xmath352 the tidal radius tends to @xmath174 . as @xmath106 decreases , the radius first decreases up to the value @xmath381 ( reached at @xmath382 ) then increases . at the critical energy @xmath0 the value of the tidal radius is @xmath383 . in fig . [ kdensity0 ] , we plot the central density normalized by @xmath384 as a function of @xmath173 . this curve is monotonic so that the parameter @xmath173 can be considered as a measure of the central density . using the results of sec . [ sec_kingclass ] , we can easily establish that @xmath385 for @xmath286 and @xmath386 for @xmath318 . finally , even if we have not represented the curves @xmath387 and @xmath388 for brevity , we give their asymptotic values . we find that @xmath389 and @xmath390 for @xmath286 and we find that @xmath391 and @xmath392 for @xmath318 . for box - confined self - gravitating classical isothermal spheres , the temperature and the energy are normalized by the box radius @xmath146 ( see , e.g. , @xcite ) . this is the proper normalization in that context because the box radius is a fixed quantity . by analogy , we could normalize the temperature and the energy of the classical king model by the tidal radius @xmath146 ( the radius at which the density drops to zero ) . this normalization was considered by lynden - bell and wood @xcite and , more recently , by casetti and nardini @xcite . however , as already noted by katz @xcite , this normalization is not correct for a thermodynamical analysis . indeed , when we study the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) , we must consider that @xmath118 , not @xmath146 , is fixed . it is only under this condition that the theory of poincar applies and that the turning points of energy and temperature correspond to a change of thermodynamical stability in mce and ce according to the maximization problems ( [ vp2 ] ) and ( [ vp3 ] ) . if we fix @xmath146 instead of @xmath118 , the turning points of energy and temperature do not correspond to a change of thermodynamical stability in mce and ce . it is not clear whether these turning points signal another form of instability . we can give several arguments why @xmath118 should be kept fixed instead of @xmath146 ( see also the arguments given in the appendix of katz @xcite ) : ( i ) basically , we must fix @xmath118 because it explicitly enters in the expression of the entropy functional defined by eqs . ( [ vp1 ] ) and ( [ df4 ] ) . in order to apply the theory of poincar , all the parameters that appear in the entropy functional must be fixed along the series of equilibria ; they act as external parameters ; ( ii ) it is only when @xmath118 is regarded as given that a self - gravitating system described by a distribution function of the form ( [ fde1])-([fde2 ] ) has a well - defined barotropic equation of state @xmath273 as discussed in sec . [ sec_eosb ] ; ( iii ) according to eq . ( [ chem1 ] ) , fixing @xmath146 is equivalent to fixing @xmath48 . however , we expect that the tidal radius and the escape energy change along the series of equilibria as we vary the energy or the temperature . therefore , on a mathematical and physical point of view , it is more relevant to fix @xmath118 rather than @xmath146 . to make the difference between the two prescriptions clear , we consider in this section the effect of fixing @xmath146 instead of @xmath118 . accordingly , in this section , the thermodynamical parameters @xmath66 and @xmath106 correspond to the dimensionless parameters @xmath193 and @xmath116 defined by eqs . ( [ temp3 ] ) and ( [ ene6 ] ) . on the other hand , @xmath124 and @xmath128 still refer to @xmath327 and @xmath328 ( their expressions are unchanged whether @xmath118 or @xmath146 is fixed ) . in fig . [ ebr ] we plot the series of equilibria @xmath351 for fixed @xmath146 . this curve is obtained from eqs . ( [ temp3 ] ) and ( [ ene6 ] ) by varying @xmath173 from @xmath156 to @xmath174 . it starts from @xmath393 ( see eq . ( [ poly5 ] ) ) and makes a spiral for large values of @xmath173 about the point @xmath394 . the series of equilibria presents a first turning point of temperature at @xmath395 , @xmath396 and a first turning point of energy at @xmath397 , @xmath398 . however , as discussed above , the meaning of these turning points regarding the stability of the system is unclear . at least , they indicate that , when @xmath146 is fixed , there is no equilibrium below @xmath399 , or above @xmath400 . we also note that there is no possible equilibrium for @xmath401 when @xmath146 is fixed . for the classical king model when @xmath146 is fixed instead of @xmath118 . ] is fixed instead of @xmath118 . this figure reveals a paradox when @xmath146 is fixed because the configurations with @xmath402 ( that could be expected to be entropy maxima ) have a lower entropy than the configurations with @xmath403 ( that are unstable ) . ] is fixed instead of @xmath118 . ] , energy @xmath325 ( normalized by @xmath118 ) , and energy @xmath116 ( normalized by @xmath146 ) as a function of @xmath173 . ] since @xmath404 and @xmath405 when @xmath146 is fixed instead of @xmath118 , the entropy @xmath377 and the energy @xmath330 on the one hand , and the free energy @xmath379 and the inverse temperature @xmath329 on the other hand , do not have their extrema at the same values of @xmath173 . as a result , the curves @xmath406 and @xmath407 present turning points instead of spikes ( compare figs . [ esr ] and [ bjr ] to figs . [ esa ] and [ bja ] ) . fig . [ eets ] recapitulates the difference between fixing @xmath118 or @xmath146 . first , we note that the curve @xmath377 is the same in the two cases . when @xmath118 is fixed , the energy @xmath408 and the entropy @xmath377 have their extrema at the same points . when @xmath146 is fixed , the extrema of @xmath409 and @xmath377 are different . in order to compare a specific theoretical model of dark matter halos to observations , it is necessary to introduce quantities that are directly measurable . in the present section , we introduce such quantities . we define them for an arbitrary distribution function so they can be applied to various models in future works . then , we explicitly calculate these quantities for the classical king model and compare the results to observations . we consider a spherical cluster described by a distribution function of the form @xmath114 with @xmath55 . the density profile of the system is given by eq . ( [ fde10 ] ) . following de vega and sanchez @xcite , we define the halo radius @xmath410 such that @xmath411 . the dimensionless halo radius @xmath412 is therefore determined by the equation @xmath413 this is a function @xmath414 of the the variable @xmath173 parameterizing the series of equilibria . the halo radius is then given by @xmath415 where @xmath270 is defined by eq . ( [ fde12b ] ) . the radial distance normalized by the halo radius can be written as @xmath416 . the tidal radius normalized by the halo radius is given by @xmath417 the function @xmath418 is plotted in fig . [ rsurrh ] for the classical king model . for @xmath286 , we find that @xmath419 ( this asymptotic value can be directly obtained from the study of the lane - emden equation ( [ poly1 ] ) from which we get @xmath420 and @xmath293 ) and for @xmath318 , we find that @xmath421 . we note that the tidal radius normalized by the halo radius increases monotonically with @xmath173 while the tidal radius normalized by @xmath118 initially decreases with @xmath173 and finally makes damped oscillations ( see sec . [ sec_proper ] ) . as a function of @xmath173 . for @xmath422 , one finds that @xmath423 . for @xmath424 , one finds that @xmath425 . ] the density profile normalized by the central density is given by @xmath426 the normalized density profile @xmath427 corresponding to the classical king model is plotted as a function of the normalized radial distance @xmath428 in figs . [ densitylog ] and [ densitylin ] in logarithmic and linear scales respectively for different values of @xmath173 . roughly speaking , for a given value of @xmath173 , the core and the halo of the distribution are isothermal ( provided that @xmath173 is sufficiently large ) while the envelope is polytropic with an index @xmath57 . this is because the density is high in the core and the halo , and low in the envelope ( see sec . [ sec_eos ] ) . it is the polytropic nature of the envelope that confers to the system a finite radius . indeed , a purely isothermal system extends to infinity and has infinite mass . by contrast , a polytrope with index @xmath57 has a compact support . the proportion of the isothermal region with respect to the polytropic one depends on @xmath173 . for @xmath286 , the density profile almost coincides with that of a polytrope of index @xmath57 ( in figs . [ densitylog ] and [ densitylin ] , the king profile with @xmath422 is indistinguishable from a pure @xmath57 polytrope ) . in that case , the tidal radius @xmath146 is of the order of the halo radius @xmath410 . for @xmath318 , the tidal radius @xmath146 is rejected to @xmath174 and the density profile approaches the profile of the classical isothermal sphere except at very large distances @xmath429 where the density drops to zero , ensuring a finite mass . for @xmath430 , the density decreases algebraically as @xmath431 with @xmath432 . actually , the density profile exhibits damped oscillations about the @xmath39 profile and @xmath330 in figs . [ aba ] and [ aea ] leading to the spiral @xmath242 of fig . therefore , the onset of gravitational collapse in ce and mce ( associated with the turning points of temperature and energy ) can be traced back to the oscillations of the density profile . ] as for the classical isothermal sphere ( see , e.g. , fig . 7 in @xcite ) . however , our study shows that the profiles with @xmath433 are thermodynamically unstable . therefore , the oscillations of the density profile are not physically relevant . as @xmath173 decreases , the effective slope of the density profile increases . for @xmath424 the density profile decreases approximately as @xmath431 for @xmath430 with an effective slope @xmath434 . for @xmath435 the density profile has an effective slope @xmath436 . in mce , the king model is stable ( @xmath369 ) as long as the effective slope @xmath105 is approximately larger than @xmath437 . in ce , the king model is stable ( @xmath365 ) only when it is close to the @xmath57 polytrope . the dotted line represents the modified hubble profile ( see appendix [ sec_mh ] ) that has a slope @xmath438 @xcite . it fits well the core of the isothermal sphere for @xmath439 @xcite . it also fits well the king model with @xmath440 up to @xmath441 . the dashed - dotted line represents hnon s isochrone profile ( see appendix [ sec_isochrone ] ) that has a slope @xmath442 @xcite . it fits well the king model with @xmath443 up to @xmath444 . the dashed line represents the burkert profile corresponding to the observations of dark matter halos @xcite . it has a slope @xmath438 ( see sec . [ sec_dm ] ) . in mce , the isochrone profile is stable , the modified hubble profile is close to the limit of marginal stability , and the classical isothermal profile is unstable . in ce , the isochrone profile , the modified hubble profile , and the classical isothermal profile are all unstable . ( @xmath445 , @xmath446 ) , @xmath435 ( @xmath447 , @xmath448 ) , @xmath424 ( @xmath449 , @xmath450 ) , @xmath451 ( @xmath452 , @xmath453 ) , and @xmath454 ( @xmath455 , @xmath456 ) . dotted line : modified hubble profile . dashed - dotted line : isochrone profile . dashed line : burkert profile . ] , @xmath435 , @xmath457 , @xmath451 , and @xmath454 . ] . the point of microcanonical marginal stability @xmath338 corresponds to @xmath458 close to @xmath437 . the slope @xmath442 corresponds to @xmath459 . ] violent relaxation usually generates a density profile with a core - halo structure . the density in the halo decreases as @xmath431 with @xmath442 @xcite . this is the same exponent as hnon s isochrone profile . this also corresponds to a king model with a concentration parameter @xmath443 that is stable in mce ( but unstable in ce ) . we argue in appendix a of paper ii that the concentration parameter @xmath460 increases monotonically with time because of collisions and evaporation until an instability takes place at @xmath461 . such an evolution is shown numerically by cohn @xcite in the case of globular clusters . since the effective slope @xmath105 of the density profile decreases with @xmath173 ( see fig . [ kalpha ] ) , we conclude that @xmath462 decreases monotonically with time . in mce , the king profile is stable as long as @xmath463 ( i.e. @xmath464 ) and it becomes unstable afterwards . in ce , the king profile destabilizes at @xmath465 before even producing an effective power law . the circular velocity is defined by eq . ( [ circb3 ] ) . the value of the circular velocity at the halo radius is @xmath466 , where @xmath467 is the halo mass . using eq . ( [ circb4 ] ) , the circular velocity normalized by its value at @xmath468 is given by @xmath469 the normalized circular velocity @xmath470 corresponding to the classical king model is plotted as a function of the normalized radial distance @xmath428 in figs . [ vitesselog ] and [ vitesselin ] in logarithmic and linear scales respectively for different values of @xmath173 . we first note that the rotation curve does not sensibly depend on the value of @xmath173 in the range @xmath471 . by contrast , differences appear for @xmath472 . for @xmath286 , the system is close to a polytrope of index @xmath57 and the tidal radius @xmath146 is of the order of the halo radius @xmath410 . for @xmath150 , the density of the dark matter halo is equal to zero so the rotation curve has a keplerian profile ( not represented ) . for @xmath318 , the density decreases as @xmath39 at large distances , like for the classical isothermal sphere , leading to a flat rotation curve . actually , the rotation curve presents damped oscillations about the plateau ( due to the oscillations of the density profile ) that are clearly visible in logarithmic scales . however , the profiles with @xmath433 are thermodynamically unstable so these oscillations are not physically relevant . in addition , real halos do not extend at such large distances where these oscillations would appear ( if they were relevant ) . , @xmath473 , @xmath474 , @xmath475 , and @xmath476 . ] , @xmath473 , @xmath474 , @xmath475 , and @xmath476 . ] for smaller values of @xmath173 , the density decreases more rapidly than @xmath39 at large distances and consequently the rotation curve decreases with the distance . for @xmath440 , the rotation curve presents a maximum close to the halo radius @xmath410 before decreasing . this is in agreement with the observations ( see sec . [ sec_dm ] ) . the modified hubble profile fits relatively well the king profile with @xmath477 up to the tidal radius @xmath478 . similarly , the isochrone profile fits relatively well the king profile with @xmath435 up to the tidal radius @xmath479 . therefore , the fit is better on the rotation curves than on the density profiles . the local velocity dispersion of a spherically symmetric distribution function @xmath114 is defined by eq . ( [ dis1 ] ) . according to eq . ( [ dis2 ] ) , the velocity dispersion profile normalized by the central velocity dispersion is given by @xmath480 i_1(k)}{i_1[\chi(\zeta)]i_2(k)}. \label{dis4b}\ ] ] the normalized velocity dispersion profile @xmath481 corresponding to the classical king model is plotted as a function of the normalized radial distance @xmath428 in figs . [ temperaturelog ] and [ temperaturelin ] in logarithmic and linear scales respectively for different values of @xmath173 . for sufficiently large @xmath173 , these curves clearly show the isothermal region where the velocity dispersion is almost uniform ( coinciding with the temperature @xmath42 ) and the polytropic region where the velocity dispersion decreases rapidly with the distance . for @xmath286 , the system almost coincides with a polytrope of index @xmath57 and the velocity dispersion is far from being uniform . actually , it is related to the density profile by @xmath482 where @xmath101 is the polytropic constant defined in sec . [ sec_poly ] . as @xmath173 increases , the velocity dispersion becomes more and more uniform in the inner region of the distribution that extends at larger and larger radii . for @xmath318 , the system is almost isothermal except at very large distances , close to the tidal radius @xmath146 . for @xmath440 , the system is isothermal for @xmath483 and polytropic for @xmath484 . ( [ dis3 ] ) , we find that the ratio between the central velocity dispersion and the temperature behaves as @xmath485 for @xmath286 while @xmath486 for @xmath318 as expected . , @xmath473 , @xmath474 , @xmath475 , and @xmath476 . ] , @xmath473 , @xmath474 , @xmath475 , and @xmath476 . ] applying eq . ( [ temp2 ] ) at @xmath468 , we find that the halo mass @xmath489 normalized by @xmath490 is given by @xmath491}{\zeta_h(k)}\equiv f(k ) . \label{dim1}\ ] ] this is a function @xmath492 of the parameter @xmath173 parameterizing the series of equilibria . for the classical king model , this function is plotted in fig . its asymptotic values can be obtained analytically . for @xmath493 , the system reduces to a pure polytrope of index @xmath57 and one finds @xmath494 ( using @xmath420 and @xmath495 obtained from the study of the lane - emden equation ( [ poly1])-([poly2 ] ) ) . for @xmath348 , the system tends to the classical isothermal sphere and one finds @xmath496 ( using @xmath497 and @xmath498 obtained from the study of the emden equation ( [ iso1])-([iso2 ] ) ) . we see in fig . [ q ] that the function @xmath492 decreases monotonically between these two values . we note that this function does not change much as a function of @xmath173 so that it has an almost `` universal '' value @xmath499 . for @xmath477 , corresponding to the stability threshold in mce , we get @xmath500 . . ] . ] applying eq . ( [ fde12 ] ) at @xmath468 and using eq . ( [ fde12b ] ) , we obtain the expression of the temperature normalized by @xmath501 : @xmath502 combining eqs . ( [ dis3 ] ) and ( [ dim2 ] ) , we find that the central velocity dispersion normalized by @xmath501 is given by @xmath503 this is a function @xmath504 of the parameter @xmath173 parameterizing the series of equilibria . for the classical king model , this function is plotted in fig . [ f ] . as for the function @xmath492 , its asymptotic values can be obtained analytically . for @xmath286 , one finds @xmath505 and for @xmath348 , one finds @xmath506 . we see in fig . [ f ] that the function @xmath504 first decreases , reaches a minimum @xmath507 at @xmath508 , then increases towards its asymptote . we note that this function does not change much as a function of @xmath173 so that it has an almost `` universal '' value @xmath509 . for @xmath477 , we get @xmath510 . we note that @xmath511 . therefore , using eqs . ( [ dim1 ] ) and ( [ dim3 ] ) , we obtain @xmath512 this function varies between @xmath513 for @xmath286 and @xmath514 for @xmath318 . for @xmath477 , it takes the value @xmath515 . some measurable quantities of dark matter halos are the central density @xmath516 , the central velocity dispersion @xmath517 , the halo radius @xmath410 , and the halo mass @xmath489 . they are given for different types of galaxies ( dwarf and large ) in table 1 of ref . @xcite . on the other hand , the circular velocities @xmath518 are known with precision from galaxy observational data . the observation of the rotation curves of a large number of galaxies shows that the density of dark matter halos can be represented by a universal profile , called the burkert profile @xcite , given by the empirical law @xmath519 the burkert profile decreases at large distances as @xmath4 like the nfw profile @xcite . this leads to a mass profile @xmath520 diverging logarithmically with @xmath521 . however , contrary to the nfw profile , the burkert profile presents a flat core density for @xmath522 instead of exhibiting a @xmath9 density cusp . density cusps are not observed in dark matter halos unless they contain a massive central black hole . the rotation curve corresponding to the burkert profile is @xmath523 after normalization by the circular velocity at the halo radius , we get @xmath524 the halo mass is obtained by integrating eq . ( [ dm1 ] ) from zero to @xmath410 . this yields @xmath525 alternatively , using the observational data given in table 1 of @xcite , we find that @xmath526 the comparison between eqs . ( [ dm4 ] ) and ( [ dm5 ] ) shows that we should not give too much importance on the precise value of these quantities . we just note that the typical values of these quantities deduced from the burkert profile or directly from the observations are consistent with those obtained theoretically with the classical king model ( see sec . [ sec_dim ] ) . actually , it can be shown @xcite that many models of dark matter halos yield values of @xmath527 and @xmath528 that agree with the observational results . therefore , the comparison between theory and observations for these quantities is not very discriminatory . the density profiles and the circular velocity profiles obtained from the king model for different values of the concentration parameter @xmath173 are compared with the burkert profile in figs . [ densitylog]-[vitesselin ] . in the range @xmath471 , all the theoretical curves coincide , whatever the value of @xmath173 , and they are in good agreement with the burkert profile . for @xmath522 while the king density profiles behave as @xmath529 for @xmath522 which is the natural behavior of spherically symmetric systems . we recall that the burkert profile is purely empirical so we should not give too much credit to its precise behavior for @xmath522 ( its main property is to have a flat core ) . we also note that the difference of behavior between the burkert profile and the king profiles for @xmath522 is almost imperceptible on the rotation curves . ] this corresponds to the region where the distribution function is isothermal . this suggests that the core of dark matter halos is isothermal . actually , it can be shown @xcite that many models of dark matter halos yield the same results in this range of radial distances so the agreement with the burkert profile for @xmath530 can not be considered as a vindication of a particular theoretical model . by contrast , at larger distances @xmath531 , the theoretical rotation curves sensibly depend on @xmath173 and the comparison with the burkert profile gives more stringent constraints on the acceptable models . we note that the virial radius of dark matter halos is of the order of @xmath532-@xmath533 @xcite , so we have to compare the theoretical profiles with the burkert profile on distances greater than @xmath410 . for @xmath318 , we recover the classical isothermal profile . however , this profile does not agree with the observational burkert profile at large distances because the density decreases too slowly . for @xmath430 , the density decreases as @xmath39 instead of @xmath4 and the rotation curve forms a plateau while the observational rotation curves slightly decrease at large distances . therefore , the king profiles with a large value of the concentration parameter @xmath173 are not in agreement with the observations . this is consistent with our theoretical study since we find that the king models with @xmath433 are thermodynamically unstable . for @xmath286 , the king model is equivalent to a polytrope of index @xmath57 . this profile does not agree with the burkert profile because the density drops to zero too rapidly . in addition , the tidal radius is of the order of the halo radius while observational rotation curves extend well beyond the halo radius . therefore , the king models with a low value of the concentration parameter @xmath173 are not in agreement with the observations . this is consistent with our theoretical study since we argue in appendix a of paper ii that the concentration parameter @xmath534 increases with time as a result of collisions and evaporation so that sufficiently old halos should have relatively large values of @xmath173 . the best agreement with the burkert profile is achieved for the king models with @xmath440 . in that case , the density profile can be approximated by the modified hubble profile ( see appendix [ sec_mh ] ) that decreases as @xmath535 like the burkert profile . the prefactors are respectively @xmath536 and @xmath10 . the rotation curves corresponding to the modified hubble profile and to the burkert profile have a similar behavior . they achieve a maximum before decreasing . the maximum is located at @xmath537 for the modified hubble profile and at @xmath538 for the burkert profile . this difference is in the error bars of the observations ( at least @xmath53 ) . we also note that the tidal radius of the king model with @xmath477 is equal to @xmath539 which is of the same order of magnitude as the observational virial radius of dark matter halos . therefore , we conclude that the observations of dark matter halos , represented by the empirical burkert profile , can be relatively well - explained by a king model at , or close to , the limit of microcanonical stability . strictly speaking , the density profiles of dark matter halos are not universal since they depend on the concentration parameter @xmath173 , but it is natural to expect that most observed halos have a concentration parameter close to @xmath461 , which explain why their profile is _ quasi universal_. indeed , the concentration parameter can not be much smaller than @xmath461 since @xmath460 increases with time , and it can not be larger than @xmath461 since , at that concentration , the clusters become thermodynamically unstable and collapse . therefore , large dark matter halos that have not collapsed should have a concentration parameter of the order of @xmath440 , and this happens to be consistent with the observations . in conclusion , we propose to describe large dark matter halos by a classical king model at the point of marginal stability in mce . it can be approximated by a modified hubble profile with a slope @xmath438 . this profile approximately accounts for the observed rotation curves of dark matter halos up to the tidal radius @xmath478 which is of the same order of magnitude as the virial radius of dark matter halos ( @xmath540-@xmath533 ) . the fact that we observe dark matter halos with a slope @xmath438 instead of @xmath442 ( a typical outcome of collisionless violent relaxation @xcite ) may be an indication that dark matter is collisional . indeed , collisions and evaporation increase the concentration @xmath460 and decrease the slope @xmath462 from the initial state @xmath541 and @xmath542 ( @xmath543 hnon s isochrone profile ) to the final state @xmath544 and @xmath545 ( @xmath543 modified hubble profile ) . the same is true for globular clusters . it is an empirical fact that the surface density @xmath546 is approximately the same for all galaxies @xcite even if their sizes and masses vary by several orders of magnitudes ( see , e.g. , table 1 of @xcite ) . its typical value is @xmath547 . as a result , it is convenient to rewrite eqs . ( [ dim1 ] ) and ( [ dim3 ] ) in terms of @xmath548 instead of @xmath516 . we get @xmath549 considering @xmath548 as a constant , can reach values of the order of @xmath550 instead of @xmath547 ( see table 1 of @xcite ) . ] these equations exhibit the scalings @xmath551 and @xmath552 . introducing relevant scales , the foregoing relations may be rewritten as @xmath553 @xmath554 for the king model , the quantities @xmath492 and @xmath504 are plotted in figs . [ q ] and [ f ] . as we have seen in secs . [ sec_dim ] and [ sec_dm ] , these quantities do not change much with @xmath173 and have the typical values @xmath555 and @xmath556 respectively . furthermore , we have explained that the concentration parameter @xmath173 should be close to @xmath461 . this fixes the prefactors in eqs . ( [ law2 ] ) and ( [ law3 ] ) to the values @xmath557 and @xmath558 respectively . we emphasize that the scalings ( [ law2])-([law3 ] ) do not depend on the distribution function chosen to model dark matter halos . only the functions @xmath492 and @xmath504 depend on the model . furthermore , most models of dark matter halos give values of @xmath492 and @xmath504 that sensibly have the same order of magnitude @xcite . therefore , the observation of the scaling laws ( [ law2 ] ) and ( [ law3 ] ) can not be considered as a vindication of a particular theoretical model . _ remark : _ for large dark matter halos , which are non degenerate , the central velocity dispersion @xmath559 represents the ratio of the temperature of the cluster on the mass of the particles ( this is valid for sufficiently large @xmath173 ) . it is possible to determine @xmath517 observationally ( see table 1 of @xcite ) . however , since the temperature of the clusters is unknown , we can not determine the mass @xmath560 of the particles that compose them . assuming that dark matter halos are made of fermions , the mass of the fermions can be obtained only from the observation of dwarf dark matter halos that are degenerate @xcite . in a nice series of papers @xcite , de vega and sanchez propose to model dark matter halos as a self - gravitating gas of fermions at finite temperature described by the fermi - dirac - poisson system ( thomas - fermi approximation ) . this idea is not new since several works in the past already considered fermionic dark matter halos at finite temperature ( see the introduction ) . however , de vega and sanchez confront this model to observations and give convincing arguments that the mass of the fermions should be of the order of @xmath561 . this mass scale corresponds to warm dark matter ( wdm ) . the dark matter particle could be a sterile neutrino . in their first papers @xcite , they argue that the cusp problem and the satellite problem of cdm are solved by quantum mechanics ( pauli exclusion principle ) . this is valid for small halos ( @xmath562 ) for which quantum effects are important . in particular , the most compact known dwarf halo ( willman 1 ) with mass @xmath563 may be considered as a completely degenerate self - gravitating fermi gas at zero temperature stabilized by quantum mechanics . however , for large dark matter halos ( @xmath564 ) , which constitute most of the available observational data ( see table i of @xcite ) , quantum effects are negligible and the classical limit applies . therefore , in ref . @xcite , de vega and sanchez describe large dark matter halos by the boltzmann distribution . in that case , the cusp problem is solved by finite temperature effects , not by quantum mechanics . de vega and sanchez argue that large dark matter halos have a universal profile corresponding to the classical isothermal profile . this profile , which has a homologous structure @xcite , has been considered by many authors in the past . de vega and sanchez show that this profile agrees with the burkert profile for @xmath565 . however , if we continue the comparison at larger distances , up to the typical virial radius of dark matter halos ( @xmath540-@xmath533 ) , severe disagreements appear between the classical isothermal profile and the burkert profile ( see fig . [ comparaison ] ) as discussed in sec . [ sec_dm ] . in particular , the classical isothermal sphere ( in addition of being thermodynamically unstable ) leads to flat rotation curves while the observed circular velocity decreases with the distance . therefore , the classical isothermal profile can not correctly describe dark matter halos up to the virial radius . de vega and sanchez recognize this problem since they argue , in their last papers @xcite , that a deviation to isothermality must be present in the halo in order to account for the observations . to this aim , they introduce a family of empirical density profiles of the form @xmath566 and mention that these profiles with @xmath434 are appropriate to fit galaxy observations up to the virial radius . then , they use the eddington equation @xcite to determine the corresponding distribution function @xmath114 . they obtain a rather complicated expression but they manage to show that the distribution function is isothermal at low energies and non - isothermal at high energies . however , no justification of the density profiles ( [ empiric ] ) is given , so their approach remains essentially empirical . by using a different approach , we arrived at a similar conclusion from more physical considerations . we proposed to model dark matter halos by the fermionic king distribution . for large halos , we can neglect quantum effects and use the classical king distribution . we argued , on the basis of kinetic considerations and stability analysis , that dark matter halos should be described by the king distribution at , or close to , the point of marginal microcanonical stability . it turns out that , at this point , the king profile can be approximated by the modified hubble profile up to the tidal radius . the modified hubble profile corresponds to the profile of eq . ( [ empiric ] ) with @xmath438 ( see appendix [ sec_mh ] ) . our study provides therefore a physical justification of the empirical profile with @xmath434 considered by de vega and sanchez @xcite . furthermore , it shows that this profile comes from a distribution function that is close to the classical king distribution . this distribution is isothermal ( boltzmannian ) at low energies and polytropic ( with index @xmath57 ) at high energies . by contrast , the physical meaning of the distribution function @xmath114 that de vega and sanchez @xcite obtain from eq . ( [ empiric ] ) by using the eddington formula is less clear . therefore , our approach provides a new light on their results . in this paper , we have studied in detail the thermodynamical properties of the classical king model . this model was originally introduced to describe globular clusters @xcite but we have proposed to apply it also to the case of large dark matter halos . the king model is a physically motivated model that can be justified by statistical mechanics and kinetic theory once the evaporation of high energy particles is properly taken into account . this distribution has a finite mass , contrary to the classical isothermal distribution , so there is no need to enclose the system within an artificial box . this generalization is important because it allows us to plot the caloric curves and make a rigorous stability analysis of the different configurations . we have found that the system is canonically stable for @xmath567 and microcanonically stable for @xmath568 ( as previously obtained by katz @xcite ) where @xmath173 is the concentration parameter . the classical isothermal distribution , corresponding to the limit @xmath318 , is thermodynamically unstable . the marginal king distribution in mce turns out to be close to the modified hubble profile and the marginal king distribution in ce turns out to be close to a polytrope of index @xmath57 . we have then compared the classical king model to observations of dark matter halos . in particular , figs . [ densitylog]-[vitesselin ] compare the prediction of the king model with the burkert profile ( dashed line ) that fits a large variety of dark matter halos . the best agreement , up to the virial radius of about @xmath532-@xmath533 , is obtained for @xmath440 , that is to say for the king model at , or close to , the limit of microcanonical stability . this is physically natural since the concentration parameter @xmath460 increases with time as a result of collisions and evaporation until an instability takes place at @xmath477 @xcite . therefore , most halos that have not collapsed should have a concentration parameter close to its maximum stable value @xmath461 . at that value , the king distribution generates a density profile that is close to the modified hubble profile with a slope @xmath434 . the king models with @xmath569 are unstable . our study therefore provides a physical justification , from the king model , of the empirical density profiles with @xmath434 introduced by de vega and sanchez @xcite . furthermore , it shows that statistical mechanics properly modified to take evaporation into account may provide a good approach to understand the structure of dark matter halos . the agreement between the modified hubble profile and the burkert profile is very good in the core for @xmath530 . therefore , it appears that the core of dark matter halos is isothermal . this thermalization may be due to a collisionless violent relaxation or to a collisional relaxation , as explained in the introduction . we emphasize that finite temperature effects produce flat core densities and , therefore , solve the problems of the cdm model without the need to advocate quantum mechanics ( that is relevant only for dwarf halos ) . on the other hand , the agreement between the modified hubble profile and the burkert profile is less good in the halo for @xmath472 . we can have two points of view . we can argue that the difference lies in the error bars of observations so that the modified hubble profile provides an equally good , or even better , description of dark matter halos than the burkert profile . indeed , we have shown in sec . [ sec_dm ] that the two profiles are qualitatively similar , differing from each other by @xmath570-@xmath571 , which is in the error bars of the observations . of course , the superiority of the modified hubble profile over the burkert profile is that it can be justified physically as an approximation of the king model at the limit of microcanonical stability while the burkert profile is a purely empirical model ( fit ) deduced from the observations . alternatively , we can adopt a completely different point of view and argue that the king model is not relevant to describe dark matter halos . indeed , the difference between the modified hubble profile and the burkert profile may be considered to be too strong . in that point of view , we should recall that the king model is usually justified for collisional self - gravitating systems , governed by the landau equation , undergoing a continuous evaporation @xcite . if dark matter halos are purely collisionless , governed by the vlasov equation , we may argue that there is no continuous evaporation justifying the king distribution . in this point of view , the structure of dark matter halos may result from an incomplete collisionless violent relaxation @xcite , described by other types of distribution functions ( different from the king model ) , as in the case of stellar systems @xcite . although the cores of dark matter halos appear to be isothermal ( possibly justified by the statistical theory of violent relaxation @xcite ) , their halo is not totally relaxed ( in the sense of lynden - bell ) . the same observation is made for elliptical galaxies . models of incomplete violent relaxation are , unfortunately , difficult to develop @xcite . we note that the king model can also provide a model of incomplete violent relaxation @xcite . it is important to know if dark matter halos are collisionless or collisional . the fact that the density profiles of dark matter halos decrease as @xmath4 instead of @xmath572 ( the typical outcome of violent relaxation @xcite ) , and the presence of black holes at the center of the halos ( see below ) , suggests that collisions play a certain role in dark matter halos . this can be understood as a result of incomplete violent relaxation @xcite . if dark matter halos are collisionless we must explain why their density profile decreases as @xmath4 instead of @xmath572 . a possibility is that they are subjected to an external stochastic forcing due to their environment . this stochastic forcing may have an effect similar to collisions . it may generate a density profile decreasing as @xmath4 . it may also trigger the formation of a central black hole . ] we finally conclude on some speculations concerning the evolution of dark matter halos , assuming that they are collisional and described by the king model . for the classical king model , equilibrium states exist in mce only above a critical energy @xmath0 and for a concentration parameter @xmath369 . because of collisions and evaporation , the energy @xmath573 of a self - gravitating system slowly decreases during its evolution while its concentration parameter @xmath460 increases . when @xmath573 passes below @xmath0 , there is no equilibrium state anymore and the system undergoes a gravitational collapse ( gravothermal catastrophe ) . this corresponds to a saddle - node bifurcation . as explained previously , large dark matter halos that are observed in the universe are expected to be close to marginally stable king distributions with @xmath440 . however , some halos may have reached the instability threshold and have undergone gravitational collapse . if the halos are made of fermions , the collapse stops when their core becomes degenerate as a consequence of the pauli exclusion principle . therefore , complete collapse is arrested by quantum mechanics . to study the phase transition between a non - degenerate gaseous sphere and a degenerate compact object , we can use the fermionic king model . this is the subject of paper ii . it is shown that gravitational collapse leads to the formation of a degenerate compact object ( fermion ball ) with a much smaller mass and radius than the original halo , accompanied by the expulsion of a hot and massive envelope . indeed , by collapsing , the fermion ball releases an enormous energy that heats the envelope . as a result , the envelope is ejected , and dispersed , at very large distances so that , at the end , only the degenerate nucleus remains . this process is reminiscent of the formation of red - giants and to the supernova explosion phenomenon , but it occurs on a cosmological scale and is considerably much slower ( of the order of the hubble time ) . this could be a mechanism of formation of dwarf dark matter halos that are completely degenerate . one important result of our study in paper ii is that large dark matter halos can not harbor a fermion ball , unlike the proposition that has been made in the past @xcite , because the `` nucleus - halo '' structures that have been considered by these authors are unreachable : they correspond to saddle points of entropy at fixed mass and energy . therefore , it should not be possible to observe a large dark matter halo with a fermion ball . this may explain why black holes at the center of galaxies are favored over fermion balls @xcite . these black holes could be formed by the mechanism discussed by balberg _ @xcite if dark matter is collisional . because of collisions , the concentration parameter @xmath460 increases until the point of gravothermal catastrophe @xmath461 . during the gravothermal catastrophe , as the central concentration and central temperature increase , the system undergoes a dynamical ( vlasov ) instability of general relativistic origin and collapses into a black hole . during this process , only the core collapses . this creates a black hole of large mass hot halo ) that has an infinite density but zero mass @xcite . for strongly collisional systems such as the core of dark matter halos , the gravothermal catastrophe leads to a black hole with a large mass @xcite . this difference is important to emphasize . ] without affecting the structure of the halo . therefore , this process leads to large halos compatible with the burkert profile for @xmath574 but harboring a central black hole at @xmath575 . however , the fermionic scenario should not be abandoned . indeed , the structure of dark matter halos crucially depends on their size through the value of the degeneracy parameter @xmath576 as discussed in paper ii . several configurations are possible making the study of the fermionic king model very rich . the system can be non degenerate ( large halos ) , partially degenerate ( intermediate size halos ) , or completely degenerate ( dwarf halos ) . therefore , we can have core - halo configurations with a wide diversity of nuclear concentration . this may account for the diversity of dark matter halos observed in the universe . the modified hubble profile is given by @xcite : @xmath577 where @xmath578 is the king radius ( or core radius ) with @xmath579 . the modified hubble profile provides a good fit of the density profile of the isothermal sphere for @xmath580 @xcite . however , the profiles differ at larger distances . for @xmath80 , the modified hubble profile decreases as @xmath4 while the density of the isothermal sphere decreases as @xmath39 . the halo radius , defined in sec . [ sec_hr ] , is given by @xmath581 therefore @xmath582 . the modified hubble profile can be rewritten as @xmath583 this is a particular case of the family of density profiles defined by eq . ( [ empiric ] ) with @xmath438 . the corresponding rotation curve is @xmath584 after normalization by the circular velocity at the halo radius , we obtain @xmath585 the normalized density profile @xmath427 and the normalized circular velocity profile @xmath470 are plotted as a function of the normalized distance @xmath428 in figs . [ densitylog ] , [ densitylin ] , [ vitesselog ] and [ vitesselin ] . they are compared to the burkert profiles . the halo mass is obtained by integrating eq . ( [ mh4 ] ) from zero to @xmath410 . this yields @xmath586 we also have @xmath587 these values can be compared to those obtained in sec . [ sec_dim ] for the king model . they are relatively close to those corresponding to the marginal king model ( @xmath338 ) . actually , the modified hubble profile provides a good fit of the marginal king profile up to @xmath588 for the density and up to @xmath589 for the circular velocity . the density profile ( [ mh1 ] ) gives rise to a surface density profile that is similar to the hubble - reynolds law fitting the surface brightness of many elliptical galaxies . this is why it is called the modified hubble profile @xcite . this analytic profile was also introduced empirically by king @xcite to fit the observed profiles of globular clusters . for that reason , it is sometimes called the king profile . a few years later , king @xcite developed a more physical model of globular clusters from a kinetic theory leading to the distribution function ( [ df1 ] ) generating a one parameter family of density profiles . to avoid ambiguity , we refer to the profile ( [ mh1 ] ) as the modified hubble profile and we refer to the one - parameter family of profiles produced by the king distribution ( [ df1 ] ) as the king profiles . as we have seen , the modified hubble profile provides a good fit of the king profile at the point of marginal microcanonical stability . the density profile and the circular velocity profile of the isochrone cluster can be written as @xcite : @xmath590 @xmath591 where @xmath592 and @xmath593 is the core radius . the central density is given by @xmath594 . therefore , we obtain @xmath595 the halo radius , defined in sec . [ sec_hr ] , is determined by the condition @xmath596 we find @xmath597 . then , we obtain @xmath598 . finally , we can write @xmath599 equations ( [ isochrone4 ] ) and ( [ isochrone6 ] ) determine the normalized density @xmath427 as a function of the normalized distance @xmath428 . these equations are parameterized by @xmath600 . the normalized circular velocity profile is given by @xmath601 using @xmath602 , we find that the halo mass is given by @xmath603 numerically , we obtain @xmath604 . the isochrone cluster was introduced by hnon @xcite who determined the condition under which the orbital period of a star depends only on its energy . we note that the density profile of the isochrone cluster decreases as @xmath572 at large distances like the density profile of many elliptical galaxies @xcite . we also recall that the isochrone profile provides a good fit of the king profile for @xmath443 ( more precisely @xmath605 ) . we can also compare the isochrone cluster and the empirical density profile arising from the collapse of a cold uniform sphere @xcite : @xmath606 they both decay as @xmath572 at large distances . however , close to the center @xmath607 behaves as @xmath608 and as @xmath609 respectively . furthermore , the profile of eq . ( [ isochrone9 ] ) leads to @xmath610 . we consider a distribution function of the form @xmath54 with @xmath56 . the local pressure is defined by eq . ( [ ene2 ] ) . taking the gradient of this expression , we get @xmath611 this expression may be rewritten as @xmath612 integrating by parts , we obtain @xmath613 using the expression of the local density given by eq . ( [ fde3 ] ) , the foregoing equation is equivalent to the condition of hydrostatic equilibrium @xmath614 we also recall that a system described by a distribution function of the form @xmath54 has a barotropic equation of state @xmath615 ( see sec . [ sec_eosb ] ) . dividing eq . ( [ hydro4 ] ) by @xmath616 , taking its divergence , and using the poisson equation ( [ fde10b ] ) , we obtain @xmath617 this is the fundamental equation of hydrostatic equilibrium for a self - gravitating barotropic gas @xcite . we now show that eq . ( [ fde13 ] ) can be directly obtained from eq . ( [ hydro5 ] ) . taking the gradient of eq . ( [ ene3 ] ) , using the identity ( [ fde ] ) , and comparing the resulting expression with eq . ( [ fde8 ] ) , we obtain @xmath618 substituting this relation in eq . ( [ hydro5 ] ) , and using eq . ( [ fde8 ] ) , we get @xmath619 from eqs . ( [ fde9 ] ) and ( [ fde12b ] ) , we have @xmath620 therefore , eq . ( [ hydro7 ] ) can be rewritten as @xmath621 introducing the variable defined by eq . ( [ fde12 ] ) , we recover eq . ( [ fde13 ] ) .	we consider the possibility that dark matter halos are described by the fermi - dirac distribution at finite temperature . this is the case if dark matter is a self - gravitating quantum gas made of massive neutrinos at statistical equilibrium . this is also the case if dark matter can be treated as a self - gravitating collisionless gas experiencing lynden - bell s type of violent relaxation . in order to avoid the infinite mass problem and carry out a rigorous stability analysis , we consider the fermionic king model . in this paper , we study the non - degenerate limit leading to the classical king model . this model was initially introduced to describe globular clusters . we propose to apply it also to large dark matter halos where quantum effects are negligible . we determine the caloric curve and study the thermodynamical stability of the different configurations . equilibrium states exist only above a critical energy @xmath0 in the microcanonical ensemble and only above a critical temperature @xmath1 in the canonical ensemble . for @xmath2 , the system undergoes a gravothermal catastrophe and , for @xmath3 , it undergoes an isothermal collapse . we compute the profiles of density , circular velocity , and velocity dispersion . we compare the prediction of the classical king model to the observations of large dark matter halos . because of collisions and evaporation , the central density increases while the slope of the halo density profile decreases until an instability takes place . we show that large dark matter halos are relatively well - described by the king model at , or close to , the point of marginal microcanonical stability . at that point , the king model generates a density profile that can be approximated by the modified hubble profile . this profile has a flat core and decreases as @xmath4 at large distances , like the observational burkert profile . less steep halos are unstable . for large halos , the flat core is due to finite temperature effects , not to quantum mechanics . we argue that statistical mechanics may provide a good description of dark matter halos . we interpret the discrepancies as a result of incomplete relaxation like in the case of stellar systems .
in turbulence flow , helicity is an important physical quantity . it is widespread in the motions of the atmosphere , ocean circulation and other natural phenomena , and also found in leading edge and trailing vortices shed from wings and slender bodies @xcite . helicity can be defined as , @xmath0 , where @xmath1 and @xmath2 are the velocity and vorticity of the turbulence flow respectively , and _ h _ is a pseudoscalar quantity . similar with the status of energy in the dynamics of ideal fluids , helicity has the character of inviscid invariance . this physical property determines that helicity is an important quantity in turbulence research . recently , the researches on helical turbulence have developed greatly in theoretical research , experiment and numerical simulation . based on the helical decomposition of velocity , the mechanism of existing a joint forward cascade of energy and helicity has been explained in theory@xcite . cascades existing in helical turbulence have space scale and time scale@xcite , and the researches showed that the existing space scale of helicity cascade was large than energy cascade . in the inertial range , the joint cascade of energy and helicity was dominated by the energy cascade time scale in the low wave number and by the helicity cascade time scale in high wave number . using direct numerical simulation of isotropic turbulence , energy and helicity flux were studied , the research showed that helicity flux is more intermittent than the energy flux and the spatial structure were much finer @xcite . large eddy simulation ( les ) , as a important method , has been used to research helical turbulence . several kinds of sgs models have been proposed so far , such as eddy - viscosity model , dynamic model , vortex model , subgrid - scale estimation model , and even the developing constrained sgs model @xcite . although so many types of sgs model , there are few models on account of the character of helical turbulence . proposed a two - term dynamic mixed models ( dsh ) to do large eddy simulations of helical turbulence @xcite . compared the les results of the dsh model with other traditional models , the improvement of dsh model was not so reamarkable.in this paper , we first use the balance of helicity dissipation and helicity flux across the inertial range to get the eddy viscosity , then deduce a new helical model ( sr ) . using sr model and dynamic sr model ( dsr)to do large eddy simulation of isotropic helical turbulence , we find that sr and dsr models can predict energy and helicity well . contrast with dynamic smogorinsky model ( dsm ) and dsh model , the new helical model is a simple model and its results of les has get an obvious improvement . in isotropic helical turbulence , the forced helicity control equation can be get from the n - s equation , as below @xmath3 where @xmath4 is the strain rate tensor and @xmath5 is the symmetric vorticity gradient tensor . @xmath6 is the force , and @xmath7 is the flux term , which can be expressed as @xmath8 for les , the resolved helicity can be defined as @xmath9 , and we can get the control function of @xmath10@xcite through filtering the eq.(1 ) at scale @xmath11 as @xmath12 where @xmath13 is the sgs helicity dissipation rate , and @xmath14 is the sgs stress . the equation for the resolved helicity in statistically stationary forced isotropic helical turbulence can be obtained through taking the ensemble average of eq.(3 ) as @xmath15 where @xmath16 ( @xmath17 is the total helicity dissipation ) @xmath18 is the sgs helicity dissipation , and @xmath19 is the viscous helicity dissipation at scale @xmath11 . in the same way , we can also get the equation for the resolved energy in statistically stationary forced isotropic helical turbulence @xcitefrom the filtered ns equation as @xmath20 where @xmath21 ( @xmath22 is the total helicity dissipation ) , @xmath23 is the sgs energy dissipation and @xmath24 is the viscous energy dissipation at scale @xmath11 . from eq.(4 ) and eq.(5 ) , we can see that the helicity dissipation has the similar form and composition as the energy dissipation . in the inertial range of isotropic helical turbulence , the viscosity of the flow may be ignored , therefor we can get the helicity and energy relations@xcite @xmath25 and @xmath26 at the same time , the average of the energy flux across the inertial range is invariable and equals to the sgs energy dissipation , @xmath27 where @xmath28 is the energy flux across the inertial range . similar with energy , helicity has the same character in inertial subrange , @xmath29 where @xmath30 is the helicity flux across the inertial range . in large eddy simulation of isotropic helical turbulence , we take the eddy - viscosity model as @xmath31 where @xmath32 is the eddy viscosity . based on the assumption of eq.(9 ) and eq.(10 ) , we can obtain @xmath33 where @xmath34 is the sgs stress model . in this deduction , we consider the eddy viscosity @xmath32 as a global averaged value@xcite for the time being , thus @xmath35 in isotropic helical turbulence , we have @xmath36 and @xmath37 where @xmath38 and @xmath39 are the energy and helicity spectra functions , and @xmath40 is the cut off wavenumber . the energy and helicity spectra @xmath38 and @xmath39 have the inequality@xcite as @xmath41 in the inertial range , the energy and helicity spectra@xcite can be choosed as @xmath42 and @xmath43 @xmath44 is kolmogrov constant and @xmath45 is the coefficient of the helicity spectrum . substituting eq.(17 ) into eq.(14 ) , we can get @xmath46 from eq.(15 ) , we can see clearly that @xmath47 where @xmath48 is a dimensionless parameter , its value changes with @xmath49 , and @xmath50 . when the cut wavenumber is in the inertial range , from eq.(16 ) , eq.(17 ) and eq.(19 ) , we obtain @xmath51 taking the absolute value of both side of eq.(18 ) , and putting it into eq.(20 ) , we can get the expression @xmath52 then , from eq.(6 ) , ( 9 ) , ( 11 ) and ( 21 ) , we can obtain the expression of @xmath32 , @xmath53 where @xmath54 is the filter scale , @xmath55@xmath56 and the model coefficient @xmath57 . thus , we obtain a new helical sgs model as @xmath58 in the les model , @xmath48 is set to 2.0 and @xmath59 and @xmath45 is confirmed by the dns data about 1.35 . in this letter , the eddy viscosity @xmath32 of the model may be given a local value , and the expression of @xmath32 is @xmath60 where @xmath61 . thus we get the new helical model discussed in this letter ( sr model ) as @xmath62 where @xmath63 . the coefficient of helical model @xmath64 can also be decided dynamically , and then we get the dsr model in this letter . in this part , we will give _ a _ _ priori _ and _ a posteriori _ test of the les model , and do some comparison and analysis . in order to validate our model , a dns of three - dimensional incompressible homogeneous isotropic turbulence is introduced here . it solves the forced n - s equations using a pseudo spectral code in a cubic box with periodic boundary conditions , and the numerical resolution is @xmath65 . a guassian random field is the initial flow condition , and it has an energy spectrum as @xmath66 where @xmath67 and @xmath68 . the whole system is maintained by a constant energy input rate @xmath69 and @xmath70 in the first two wave number shells . have constructed the new helical models sr and dsr , we first test the validity of the models _ a priori_. in the inertial range , we have assumed the invariance of the energy and helicity sgs dissipations . in such a precondition , the invariance of @xmath71 and @xmath72 must hold , @xmath73 and @xmath74 where @xmath75 is the length scale varying in the inertial range . in fig.1 , we show @xmath71 and @xmath72 as a function of @xmath76 , and @xmath77 is the kolmogorov scale.it is easy to see that the values of the two functions in inertial range are almost constant , and it demonstrates the scale - invariance and the reasonableness of the model . we choose four les models to do some analysis and comparison here , the sr model , the dsr model , the dynamic smogorinsky model ( dsm ) and the dynamic mixed helical model ( dsh)@xcite . fig.2 and fig.3 display the energy and helicity spectra of dns and different models for steady isotopic turbulence respectively . the bolt solid line is for dns spectra , and the other lines are for the four models.it is easy to see that the sr and dsr show similar trends and predict the energy and helicity spectra better than dsm and dsh . in fig.2 , dsm and dsh underestimate the energy spectra near the cutoff wave - number and overestimate at the middle range of the wave-number.in fig.3 , dsm and dsh overestimate the helicity spectra at the middle range of the wave - number , and dsh underestimates it seriously near the cutoff wave - number . in fig.4 and fig.5 we show the time evolution of the dns and four models energy and helicity spectra for a decay problem starting from a fully developed statistical steady state respectively , where @xmath78 and @xmath79 . in fig.4 , similar with the steady prbolem , the sr and dsr also show similar trends and predict the energy spectra very well . dsm and dsh overestimate the energy spectra at the middle range of the wave - number and underestimate it near the cutoff wave - number . fig.5 show us the graphics different from fig.3 greatly.the helicity spectra have large fluctuation . because it is a free decaying course and helicity is a pseudoscalar quantity , the two factors cause such phenomenon . in fig.5 we can still see sr and dsr pridict the helicity spectra a little better than the other two models . fig.6 and fig.7 show us the sgs energy and helicity dissipations of dns and the four models distribute with @xmath76 for @xmath80 @xmath81 respectively , and they reflect case of the full developed steady turbulence . it is obvious from in the inertial range that the sgs energy and helicity dissipations from sr are closer to those of the dns than other models , and specially have the similar change trends with dns . in fig.8 and fig.9 , we show the decay of sgs energy and helicity dissipation with @xmath82(@xmath83 is the initial large eddy turnover time scale ) from a fully developed steady state for @xmath84 at the filter scale @xmath11 respectively . it is clear in fig.8 that the sgs energy dissipations from sr and dsr are closer to the dns result than dsm and dsh , and also have similar change trend . and also we can see in fig.9 that the sgs helicity dissipations from sr , dsr and dsm are closer to the dns result than dsh . in fig.10,we display the distribution of high - order velocity increment with @xmath11 , where @xmath85 is the separation distance and @xmath11 is the filter scale . fig.10 ( a ) , ( b ) and ( c ) denote the fourth - order , sixth - order and eighth - order structure functions of the longitudinal velocity increment of dns and the four models distribute with separation distance . we can from fig.10 that the result from sr is closest to the dns result very well in the range of @xmath86 , and the behavior of dsr is a little better than dsh and dsm . while in the range of @xmath87 , anyone of the models have still not given a rather good result . in this letter , we construct a new helical sgs stress model for large eddy simulation of the isotropic helical turbulence . different other eddy viscosity model , we conduct the helical model based on the balance of helicity dissipation and the average of helicity flux across the inertial range in helical turbulence . then we have tested the scale - invariance of the model for @xmath80 @xmath81 in inertial range , and make sure the validity of the helical model . we have given a constant coefficient helical model and a dynamic helical model here . have tested from @xmath80 @xmath81 and @xmath80 @xmath88 , the helical model is confirmed to predict energy and helicity spectra accurately , and also gives rather good simulating results in energy and helicity dissipation , @xmath89 @xmath90.besides the advantages above all , the helical model is a single model and can improve the computing efficiency greatly . as a sgs model on account of helical turbulence , it can also apply to the rotational turbulence . this work was supported by the national natural science foundation of china ( grant nos . ! ! ! ! ! ! ! ! ! ) . aipsamp d. k. lilly , _ the development and maintence of rotation inconvective storms _ in _ intense atmospheric vortices : proceedings of the joint symposium ( iutam / iugg ) held at reading _ , edited by l. bengtsson and j. lighthill ( springer - verlag , beilin , 1982 ) , pp . 149 - 60 . a. tsinober and e. levich , on the helical nature of three dimensional coherent structures in turbulent flows , " phys . lett . * * 99*a , 321 ( 1983 ) . h. k. moffatt and a. tsinober , helicity in laminar and turbulent flow , " annu fluid mech . 24 * , 281 ( 1992 ) . q. chen , s. chen , and g. l. eyink , the joint cascade of energy and helicity in three- dimensional turbulence , " phys . fluids * 15 * , 361 ( 2003 ) . p. d. diltevsen and p. giuliani , cascades in helical turbulence , " phys . e * 63 * , 036304 ( 2001 ) . p. d. ditlevsen and p. giuliani , dissipation in helical turbulence , " phys . fluids * 13 * , 3508 ( 2001 ) . r. h. kraichnan , inertial - range transfer in two - and three - dimensional turbulence , " j. fluid mech . * 47 * , 525 ( 1971 ) . s. kurien , m. a. taylor , and t. matsumoto , cascade time scales for energy and helicity in homogeneous isotropic turbulence , " phys . e * 69 * , 066313 ( 2004 ) . q. n. chen , s. y. chen , g. l. eyink , and d. d. holm , intermittency in the joint cascade of energy and helicity , `` phys . 90 * , 214503 ( 2003 ) . j. w. deardoff , '' the use of subgrid transport equation in a three dimensional model of atmospheric turbulence , " asme . * 26(6 ) , 669 ( 1973 ) . s. b. pope , _ turbulence flow _ ( cambridge university press , cambridge , 2000 ) j. smagorinsky , general circulation experiments with primitive equation , " monthly weather review , 91 * , 99 ( 1963 ) . p. moin and j. kim , numerical investigation of turbulent channel flow , " j. fluid mech . * 118 * , 341 ( 1982 ) . m. germano , u. piomelli , p. moin , and w. cabot , a dynamic subgridscale eddy viscosity model , " phys . fluids a * 3 * , 1760 ( 1991 ) . d. k. lilly , a proposed modification of the germano subgrid - scale closure method , " phys . fluids a , * 4 * , 3 ( 1992 ) . a. misra and d. i. pullin , a vortex - based subgrid stress model for large eddy simulation , " phys . fluids * 9 * , 2443 ( 1997 ) . j. a. domaradzki and e. m. saiki , a subgrid - scale model based on the estimation of unresolved scales of turbulence , " * 9 * , 2148 ( 1997 ) . j. a. langford and r. d. moser , optimal les formulations for isotropic turbulence , " j. fluid mech . * 398 * , 321 ( 1999 ) . y. mornishi and o. v. vasilyev , a recommended modification to the dynamic two - parameter mixed subgrid scale model for large eddy simulation of wall bounded turbulent flow , " phys . fluids * 13 * , 3400 ( 2001 ) . c. meneveau , statistics of turbulence subgrid - scale stresses ; necessary conditions and experimental tests , " phys . fluids * 6 * , 815 ( 1994 ) . s. ghosal , t. s. lund , p. moin , and k. akselvoll , a dynamic localization model for large - eddy simulation of turbulent flows , " j. fluid mech . * 286 * , 229 ( 1995 ) . y. li and c. meneveau , analysis of mean momentum flux in subgrid models of turbulence , " phys . fluids * 6 * , 3483 ( 2004 ) . y. shi , z. xiao , and s. chen,costrained subgrid - scale stress model for large sddy simulation , " phys . fluid * 20 * , 011701 ( 2008 ) . y. li , c. meneveau , s. chen , and g. l. eyink , subgrid - scale modeling of helicity and energy dissipation in helical turbulence , " phys . e * 74 * , 026 ( 2006 ) . c. meneveaua and t. s. lund , the dynamic smagorinsky model and scale - dependent coefficients in the viscous range of turbulence , `` phys . fluid * 9 * , 3932 ( 1997 ) . d. k. lilly , '' the representation of small - scale turbulence in numerical simulation experiments , " _ proceedings of the ibm scientific computing symposium on environmental sciences _ , 1967 , s. cerutti and c. meneveau , intermittency and relative scaling of subgrid - scale energy dissipation in isotropic turbulence , `` phys . fluid * 10 * , 928 ( 1998 ) . a. brissaud , u. frisch , j. leorat , m. lesieur , and a. mazure , '' helicity cascades in fully developed isotropic turbulence , " phys . fluids * 16 * , 1366 ( 1973 ) v. borue and s. a. orszag , spectra in helical three - dimensional homogeneous isotropic turbulence , " phys . e * 55 * , 7005 ( 1997 ) . y. zang , r. l. street , and j. r. koseff , a dynamic mixed subgrid - scale model and its application to turbulent recirculating flows , " phys . fluids a * 5 * , 3186 ( 1993 ) . s. w. liu , c. meneveau , and j. katz , on properties of similarity subgrid - scale models as deduced from measurements in a turbulent jet , " j. fluid mech . * 275 * , 83 ( 1994 ) . c. meneveau and j. katz , scale - invariance and turbulence models for large- eddy simulation , " annu . fluid mech . * 32 * , 1 ( 2000 ) . h. lu , c. j. rutland , and l. m. smith , a priori tests of one - equation les modeling of rotating turbulence , " j. turbulence . * 8 * , 37 ( 2007 ) . q. chen and g. l. eyink , the joint cascade of energy and helicity in three - dimensional turbulence , " phys . fluid * 15 * , 361 ( 2003 ) . c. meneveau and j. katz , scale - invariance and turbulence models for large - eddy simulation , " annu . fluid mech . * 32 * , 1 ( 2000 ) .	in isotropic helical turbulence , a new single helical model is suggested for large eddy simulation . based on the kolmogrov s hypotheses , the helical model is proposed according to the balance of helicity dissipation and the average of helicity flux across the inertial range , and the helical model is a kind of eddy viscosity model . the coefficient of the helical model is constant and can be also determined dynamically . numerical simulations of forced and decaying isotropic helical turbulence demonstrate that the helical model predicts the energy and helicity evolution well . the statistical character of the helical model is closer to the dns results.in _ a priori _ test , the energy and helicity dissipations of the helical model also show a good results . in general , the helical model has the advantage contrast with the dynamic smogorinsky and mixed models .
deep inelastic scattering ( dis ) of leptons off nucleons has played a fundamental role in understanding the structure of matter . hera the first electron - proton ( @xmath8 ) collider ever built is capable of investigating the two main contributions to dis , neutral current ( nc ) interactions , @xmath9 and charged current ( cc ) interactions , @xmath10 . in the standard model a photon ( @xmath11 ) or a @xmath6 boson is exchanged in a nc interaction , and a @xmath12 boson is exchanged in a cc interaction . dis can be described by the four - momentum transfer squared @xmath5 , bjorken-@xmath13 and inelasticity @xmath14 defined as : @xmath15 with @xmath16 and @xmath17 being the four - momentum of the incident ( scattered ) lepton and proton respectively . the centre - of - mass energy @xmath18 of the @xmath8 interaction is given by @xmath19 . in this paper measurements are presented of the nc and cc cross - sections at high @xmath20 for both electron and positron scattering from both the h1 and zeus experiments . the integrated luminosity in the @xmath1 data sample is @xmath21 and in the @xmath3 data sample is @xmath22 for the h1 ( zeus ) collaborations . the h1 and zeus detectors are described elsewhere @xcite,@xcite . both detectors are nearly hermetic multi - purpose apparatus built to investigate @xmath8 interactions . the measurements in both experiments rely primarily on the calorimetry , tracking and the luminosity detectors . in both experiments the charged current event selection is based on the observation of large @xmath23 , which is assumed to be the transverse momentum carried by the outgoing neutrino . for nc events it is based on the identification of a scattered electron ( positron ) , further fiducial ( nc ) and kinematic cuts ( cc and nc ) are then applied . details of the selection procedure can be found in @xcite . the electroweak born - level nc dis cross section @xmath24 for the reaction @xmath25 can be written as : @xmath26\ ] ] for unpolarised beams , the structure functions @xmath27 and @xmath4 can be decomposed taking into account @xmath6 exchange as : @xmath28 where @xmath29 is the mass of the @xmath6 , @xmath30 is a function of the weinberg angle @xmath31 and @xmath32 and @xmath33 are the vector and axial couplings of the electron to the @xmath6 . for unpolarised beams , @xmath27 is the same for electron and positron scattering , while @xmath4 changes sign . the nc `` reduced cross - section '' @xmath34 is defined from the measured as : @xmath35 the born double differential cc cross - section for @xmath36 can be written in leading order qcd as : @xmath37 \label{eqn : cc}\ ] ] for @xmath3 scattering the coupling is predominantly to the @xmath38 type quarks for @xmath1 scattering the electroweak coupling is predominantly to the @xmath39 type quarks . @xmath40\ ] ] the double differential cc `` reduced '' cross - section @xmath41 is defined as : @xmath42 the nc and cc single differential cross - sections and @xmath43 , @xmath44 are shown in fig [ fig : singdif ] , also shown are the standard model expectations given from nlo qcd fits to the data by both the cteq @xcite and the h1 @xcite collaborations . the standard model is seen to give a good description of the data . the measurement of he nc cross - section spans more than two orders of magnitude in @xmath5 and the cross - section falls with @xmath5 by about 6 orders of magnitude for both the @xmath1 and the @xmath3 cross section . for @xmath5 values above @xmath45 the @xmath3 cross section can be seen to be significantly above the @xmath1 cross section . this is consistent with the standard model picture in which one expects constructive interference of the photon and @xmath6 for @xmath3 interactions and destructive interference for @xmath1 interactions . due to the propagator mass term and the different coupling the cc cross - section is smaller than the nc cross - section and it falls less steeply , by about 3 orders of magnitude , between @xmath46 and @xmath47 for both the @xmath3 and @xmath1 cross section . the shape and magnitude of the nc and cc cross - sections are well described by the standard model expectation . the @xmath3 cross section is at relatively low @xmath5 to be approximately twice as big as the @xmath1 cross section , which is consistent with the standard model picture in which the @xmath38 quark density probed is expected to be approximately twice the @xmath39 quark density . shown in fig . [ fig : singdif ] is the single differential cc cross section @xmath48 for @xmath20 for @xmath1 scattering from the zeus collaboration @xcite . the ratio of the measured cross section @xmath48 . at high @xmath13 the @xmath1 cc cross - section depends mainly on the @xmath39 quark density which is less constrained than the @xmath38 quark density . all data agree well with the standard model expectation using pdfs extracted from cteq4d and those from the zeus nlo fit @xcite . the possibility of a larger @xmath49 ratio than previously assumed has been of interest in recent years , for example see @xcite . modification @xcite of pdfs with an additional term @xmath50 yields @xmath51 close to the nlo qcd fit . the nc cross - section @xmath52 for @xmath53 is shown in fig . [ fig : ddiff ] along with the expectations from the standard model with a @xmath6 mass of set to the pdg value , which gives a good description of the data . also shown is the expectation from the standard model with @xmath6 mass set to infinity , thereby effectively removing the weak interaction so that only photon ( @xmath11 ) exchange is possible . it can be seen the data clearly favours the need for the @xmath6 contribution . ( -140,175)(0,0)a ) ( -140,175)(0,0)b ) ( -130,130)(0,0)c ) ( -130,140)(0,0)d ) the double differential reduced nc cross - section is shown in fig . [ fig : ddiff ] for fixed @xmath13 as a function of @xmath5 for both @xmath54 and @xmath55 data . also shown are the expectations of the standard model which give a good description of the data . for @xmath5 values below @xmath56 the @xmath54 and @xmath55 cross - sections are within errors equal , for higher values of @xmath5 the cross sections are seen to deviate , with the @xmath3 cross - section seen to increase above the @xmath1 cross section . this difference is consistent with the change in sign of the parity violating structure function @xmath4 . the zeus collaboration go a step further and combine the @xmath54 and @xmath55 cross sections , where @xmath0 of @xmath1 data collected at a centre of mass energy of @xmath57 were combined with @xmath58 of @xmath1 data collected at a centre of mass energy of @xmath57 . to reduce statistical fluctuations several bins from the double differential binning @xcite . figure [ fig : ddiff ] shows @xmath4 at fixed values of @xmath5 as a function of @xmath13 . the standard model expectation evaluated with either the cteq4d or mrst(99 ) @xcite parton density functions are seen to give a good description of the data . the reduced cross sections as functions of @xmath13 and @xmath5 are displayed in figs . [ fig : ddiff ] along with the prediction of the h1 nlo qcd fit @xcite . the relative increase in the @xmath3 cross section over the @xmath1 cross section is consistent with a larger @xmath38 quark density relative to that of the @xmath39 quark . the absolute magnitude of the cc cross section , described by equation [ eqn : cc ] is determined by the fermi constant @xmath59 and the pdfs , while the @xmath5 dependance of the cc cross - section includes the propagator term @xmath60 $ ] which produces substantial damping of the cross section . the zeus collabiration have fitted the measured differential cross section , @xmath61 , treating @xmath59 and @xmath62 as free parameters , yields @xmath63 and @xmath64 the central values are obtained using the cteq4d pdfs . further fits to the data were also performed @xcite , the propagator - mass fit to the measured differential cross - section , @xmath61 , with @xmath59 fixed to the value @xmath65 yields the result @xmath66 using the standard model constraint @xmath67 , @xmath29 , and all fermion masses , other than the mass of the top quark , @xmath68 , are set to the pdg values @xcite . the central result of the fit was obtained with @xmath69 and the mass of the higgs boson @xmath70 . the @xmath71 function is evaluated along the line given by the sm constraint , gives the following result . from september 2000 , the luminosity of the hera collider will be increased by a factor of five . at the same time longitudinal lepton beam polarisation will be provided for the collider experiments h1 and zeus . over a six year running period it is anticipated that the total luminosity of @xmath73 will be delivered . this large data volume will allow @xmath74 to be extracted with an accuracy of @xmath75 over the kinematic range @xmath76 and @xmath77 . with this accuracy it is anticipated that it will be possible to determine @xmath78 from the scaling violations of @xmath74 with a precision of @xmath79 . the gluon distribution will also be determined from such a fit with a precision of @xmath75 for @xmath80 and @xmath81 . there will also be a significant benefit to the cc cross - section measurement , with the potential of a clean determination of the @xmath38 and @xmath39 quark densities . with the introduction of polarised beams it will be possible to obtain a measurement of the vector and axial - vector couplings of the @xmath38-quark , @xmath82 and @xmath83 respectively , obtained in a fit in which @xmath82 and @xmath83 are allowed to vary . with a luminosity of @xmath84 per charge , polarisation combination and taking the vector and axial - vector couplings of the @xmath38 and @xmath39-quarks as free parameters gives a precision of 13 % , 6 % , 17 % for @xmath82 , @xmath85 , @xmath86 and @xmath87 respectively . by comparing the nc couplings of the @xmath88 and @xmath89-quarks obtained at lep , it will be possible to test the universality of the nc couplings . the latest deep - inelastic neutral and charge current cross - sections have been presented from the h1 and zeus experiments . the standard model is seen to give a good description of the data in all cases . the high luminosity data at high @xmath5 from both experiments has enabled the first tests of electroweak effects in both neutral and charged current interactions . in nc interactions the data are seen to be consistent with the effect of @xmath90 interference and has allowed the first extraction of the parity violating structure function @xmath4 at high @xmath5 . with the factor of five increase in yearly luminosity expected from the hera upgrade further high precision tests of the strong and weak interaction will be made . the future cc measurements will allow precision determinations of the @xmath38 and @xmath39 quark densities and a determination of the vector and axial - vector couplings of the quarks from the nc interaction , providing important information for future high energy hadron colliders such as the lhc . the h1 detector h1 collab . , i. abt et al . meth . the zeus detector , status report 1993 , desy ( 1993 ) . * a336 * ( 1993 ) 460 . h1 collab . , c. adloff et al . , eur phys j * c 13 * ( 2000 ) 609 - 639 particle data group , c. caso et al . , eur . j. * c3 * ( 1998 ) 1 . lai et al . , phys . rev . * d55 * ( 1997 ) 1280 . m. botje , _ a qcd analysis of hera and fixed target structure function data _ , desy 99 - 038 , nikhef-99 - 011 ( in preparation ) . u.k . yang and a. bodek , _ phys . * 82 * ( 1999 ) 2467 . w. melnitchouk and a.w . thomas , _ phys . lett . _ * b377 * ( 1996 ) 11 ; + w. melnitchouk and j.c . peng , _ phys . lett . _ * b400 * ( 1997 ) 220 .	measurements of deep - inelastic neutral and charged current interactions are presented in lepton proton scattering at hera . the measurements are obtained from taken during 1996 to 1999 and consists of @xmath0 of @xmath1 and @xmath2 of @xmath3 data . the addition of the new high statistics electron data with the positron data allows the first extraction of the parity violating structure function @xmath4 and tests of high-@xmath5 electroweak effects of the heavy bosons @xmath6 and @xmath7 are observed and found to be consistent with the standard model expectation .
construction of chiral gauge theory is one of the long standing problems of lattice field theories . because of the fermion doubling problem , the lattice field theory discretized naively becomes non - chiral . several approaches on lattice have been proposed to overcome this difficulties , but none of them have been proven to be successful . recently kaplan has proposed a domain - wall model in order to construct lattice chiral gauge theories@xcite . the model consists of wilson fermion action in ( 2@xmath0 + 1 ) dimensions with a fermion mass term being the shape of a domain wall in the ( extra ) ( 2@xmath0 + 1)th dimension . in the case of free fermions it is shown for @xmath1 , where @xmath2 denotes the domain wall mass height , that a massless chiral state arises as a zero mode bound to the @xmath3-dimensional domain wall while all doublers have large masses of the lattice cutoff scale . in a way of introducing dynamical gauge fields two variants of this model have been proposed : the waveguide model@xcite and the overlap formula@xcite . however , it has been reported that the chiral zero mode disappears for these two variants even in the weak coupling limit , due to the roughness of gauge field@xcite . in the original model the roughness of the gauge field is replaced with the dynamical gauge field in the extra dimension , and in the weak coupling limit of the @xmath3 dimensional coupling , @xmath3 dimensional links @xmath4 = 1 ( @xmath5 ) while only extra dimensional links @xmath6 become dynamical . we would like to know whether the fermionic zero modes exist on the domain wall or not in this limit . this question has already been investigated in ( 2 + 1 ) dimensions@xcite , but the result , in particular , in the symmetric phase is not conclusive , due to peculiarity of the phase transition in the 2 dimensional u(1 ) spin model . therefore we numerically investigate this model with u(1 ) gauge field in ( 4 + 1 ) dimensions and report the result here . before numerical investigation we estimate the effect of the dynamical gauge field in the extra dimension using the mean - field analysis . in our mean - field analysis all link variables in the extra dimension are replaced with @xmath7-independent constant @xmath8 @xmath9 , so that the fermion propagator is easily obtained@xcite @xmath10 here was denoted @xmath11 @xmath12 in ref.@xcite ] @xmath13_{s , t } \nonumber , \end{aligned}\ ] ] where @xmath14 and @xmath15 . corresponding fermion masses are obtained from @xmath11 and @xmath16 in @xmath17 limit , and we found that fermionic zero modes exist only for @xmath18 . thus the critical value of domain wall mass is @xmath19 and therefore no zero mode survives if @xmath20 . we investigated the ( 4 + 1 ) dimensional u(1 ) model numerically , using a quenched approximation . at the zero physical gauge coupling the gauge field action of the model is reduced to the @xmath3 dimensional spin model with many copies : @xmath21,\ ] ] where @xmath22 . therefore there exists a phase transition which separates a broken phase from a symmetric phase . we calculated the order parameter @xmath23 using a rotational technique@xcite and found @xmath24 corresponds to the symmetric phase and @xmath25 to the broken phase . we calculated the fermion propagator over 50 configurations at @xmath25 and 0.29 on @xmath26 lattices with @xmath274 , 6 , 8 . for the fermion field we take periodic boundary conditions except the anti - periodic boundary condition in the 4th direction . the errors are estimated by the single elimination jack - knife method . at @xmath28 we obtained the inverse propagator @xmath29 and @xmath30 for @xmath31 as a function of @xmath32 . we obtain fermion mass squared , @xmath33 , by extrapolating @xmath29 and @xmath30 linearly to @xmath34 . in fig.[fig : broken ] we plotted @xmath35 in the broken phase ( @xmath25 ) as a function of @xmath2 . = 7.5 cm as seen from this figure , the finite size effect is small , and the left - handed modes are always massive , while the right - handed modes are massless if @xmath2 is larger than about @xmath36 . therefore , we conclude that chiral zero modes can exist in the broken phase . let us show the fermion mass in the symmetric phase ( @xmath24 ) in fig.[fig : symmetric ] . in the smallest lattice size the chiral zero modes seem to exist . = 7.5 cm however , for large lattices , the mass difference between the left- and right - handed modes becomes smaller . this suggest that the fermion spectrum becomes vector - like in the infinite volume limit . however , since the fermion mass near @xmath37 is so small , from this data alone , we can not exclude a possibility that the critical mass @xmath38 is very close to @xmath39 . to make a definite conclusion on the absence of chiral zero modes in the symmetric phase , we try to fit the fermion propagator using the form of the mean - field propagator with the fitting parameter @xmath8 . we show the quality of the fit in fig.[fig : propr ] . = 7.5 cm this figure shows that the fermion propagator is well described by the mean - field propagator . in fig.[fig : fittingr ] and fig.[fig : fittingl ] we plotted the parameter @xmath8 obtained the above fit as a function of @xmath40 = 7.5 cm = 7.5 cm the parameter @xmath8 s are almost independent of @xmath2 at the each @xmath40 except for the right - handed ones at @xmath41 . the solid circles represent the order parameter @xmath23 . the behaviors of @xmath8 at different @xmath2 are almost identical each other and are very similar to that of @xmath23 except the right - handed ones at @xmath41 . this suggest that @xmath8 can be identified with @xmath23 and therefore @xmath8 becomes zero as the lattice size goes to the infinity . if this is the case the fermion spectrum of this model becomes vector - like in the symmetric phase . we have carried out the numerical simulations of the u(1 ) original domain - wall model in ( 4 + 1 ) dimensions in the weak coupling limit of the 4-dimensional coupling . in the broken phase , there exist chiral zero modes on the domain wall for @xmath42 . the existence of the critical mass @xmath38 is predicted by the mean - field analysis . on the other hand , in the symmetric phase , the analysis using the mean - field propagator suggests that this model becomes vector - like . we should note , however , that the right - handed modes at @xmath41 behaves differently and the similar behavior was also found in the ( 2 + 1 ) dimensional model@xcite . therefore in the future , we must investigate this point in detail , for example , increasing the statistics . besides this point the results from both phases suggest that this model becomes vector - like in the continuum limit , which should be taken at the critical point (: the point between two phases ) . therefore , it seems also difficult to construct a chiral gauge theory on lattice via the original domain - wall model . 9 d.b . kaplan , phys . * b288 * , ( 1992 ) 342 . golterman , k. jansen , d.n . petcher , and j.c . vink , phys . * d49 * , ( 1994 ) 1606 ; m.f.l . golterman and y. shamir , phys . rev . * d51 * , ( 1995 ) 3026 . r. narayanan and h. neuberger , nucl . * b412 * , ( 1994 ) 574 . s. aoki and k. nagai , phys . * d53 * , ( 1996 ) 5058 .	we carry out a numerical simulation of a domain - wall model in ( 4 + 1 ) dimensions , in the presence of a quenched u(1 ) dynamical gauge field only in an extra dimension , corresponding to the weak coupling limit of a ( 4-dimensional ) physical gauge coupling . our numerical data suggest that the zero mode seems absent in the symmetric phase , so that it is difficult to construct a lattice chiral gauge theory in the continuum limit .
despite the rapid progress in designing fully autonomous systems , many systems still require human s expertise to handle tasks which autonomous controllers can not handle or which they have poor performance . therefore , shared autonomy systems have been developed to bridge the gap between fully autonomous and fully human operated systems . in this paper , we examine a class of shared autonomy systems , featured by switching control between a human operator and an autonomous controller to collectively achieve a given control objective . examples of such shared autonomy systems include robotic mobile manipulation @xcite , remote tele - operated mobile robots @xcite , human - in - the - loop autonomous driving vehicle @xcite . in particular , we consider control under temporal logic specifications . one major challenge for designing shared autonomy policies under temporal logic specifications is making trade - offs between two possibly competing objectives : achieving the optimal performance for satisfying temporal logic constraints and minimizing human s effort . moreover , human s cognition is an inseparable factor in synthesizing shared autonomy systems since it directly influences human s performance , for example , a human may have limited time span of attention and possible delays in response to a request . although finding an accurate model of human cognition is an ongoing challenging topic within cognitive science , markov models have been proposed to model and predict human behaviors in various decision making tasks @xcite . adopting this modeling paradigm for human s cognition , we propose a formalism for shared autonomy systems capturing three important components : the operator , the autonomous controller and the cognitive model of the human operator , into a stochastic _ shared - autonomy system_. precisely , the three components includes a markov model representing the fully - autonomous system , a markov model for the fully human - operated system , and a markov model representing the evolution of human s cognitive states under requests from autonomous controller to human , or other external events . the uncertainty in the composed system comes from the stochastic nature of the underlying dynamical system and its environment as well as the inherent uncertainty in the operator s cognition . switching from the autonomous controller to the operator can occur only at a particular set of human s cognitive states , influenced by requests from the autonomous controller to the operator , such as , pay more attention , be prepared for a possible future control action . under this mathematical formulation , we transform the problem of synthesizing a shared autonomy policy that coordinates the operator and the autonomous controller into solving a with temporal logic constraints : one objective is to optimize the probability of satisfying the given temporal logic formula , and another objective is to minimize the human s effort over an infinite horizon , measured by a given cost function . the trade - off between multiple objectives is then made through computing the pareto optimal set . given a policy in this set , there is no other policy that can make it better for one objective than this policy without making it worse for another objective . in literature , pareto optimal policies for s have been studied for the cases of long - run discounted and average rewards @xcite . the authors in @xcite proposed the weighted - sum method for s with multiple temporal logic constraints by solving pareto optimal policies for undiscounted time - bounded reachability or accumulated rewards . these aforementioned methods are not directly applicable in our problem due to the time unboundness in both satisfying these temporal logic constraints and the accumulated cost / reward . to this end , we develop a novel two - stage optimization method to handle the multiple objectives and adopt the so - called _ tchebychev scalarization method _ @xcite for finding a uniform coverage of all pareto optimal points in the policy space , which can not be computed via weighted - sum ( linear scalarization ) methods @xcite as the latter only allows pareto optimal solutions to be found amongst the convex area of the pareto front . finally , we conclude the paper with an algorithm that generates a pareto - optimal policy achieving the desired trade - off from user - defined weights for coordinating the switching control between an operator and an autonomous controller for a stochastic system with temporal logic constraints . we provide necessary background for presenting the results in this paper . a vector in @xmath0 is denoted @xmath1 where @xmath2 are the components of @xmath3 . we denote the set of probability distributions on a set @xmath4 by @xmath5 . given a probability distribution @xmath6 $ ] , let @xmath7 be the set of elements with non - zero probabilities in @xmath8 . [ def : labeledmdp ] a _ labeled _ is a tuple @xmath9 where @xmath4 and @xmath10 are finite state and action sets . @xmath11 is the initial probability distribution over states . the transition probability function @xmath12 $ ] is defined such that given a state @xmath13 and an action @xmath14 , @xmath15 gives the probability of reaching the next state @xmath16 . @xmath17 is a finite set of atomic propositions and @xmath18 is a labeling function which assigns to each state @xmath19 a set of atomic propositions @xmath20 that are valid at the state @xmath21 . @xmath22 is a reward function giving the immediate reward @xmath23 for reaching the state @xmath16 after taking action @xmath24 at the state @xmath21 and @xmath25 is the reward discount factor . @xmath26 in this context , @xmath27 gives a probability distribution over the set of states . @xmath28 and @xmath29 both express the transition probability from state @xmath21 to state @xmath16 under action @xmath24 in @xmath30 . a _ path _ is an infinite sequence @xmath31 of states such that for all @xmath32 , there exists @xmath33 , @xmath34 . we denote @xmath35 to be a set of actions enabled at the state @xmath21 . that is , for each @xmath36 , @xmath37 . a _ randomized policy _ in @xmath30 is a function @xmath38 that maps a finite path into a probability distribution over actions . a deterministic policy is a special case of randomized policies that maps a path into a single action . given a policy @xmath39 , for a measurable function @xmath40 that maps paths into reals , we write @xmath41 $ ] ( resp . $ ] ) for the expected value of @xmath40 when the starts in state @xmath21 ( resp . an initial distribution of states @xmath43 ) and the policy @xmath39 is used . a policy @xmath39 induces a probability distribution over paths in @xmath30 . the state reached at step @xmath44 is a random variable @xmath45 and the action being taken at state @xmath45 is also a random variable , denoted @xmath46 . we use @xcite to specify a set of desired system properties such as safety , liveness , persistence and stability . in the following , we present some basic preliminaries for specifications and introduce a product operation for synthesizing policies in s under constraints . a formula in is built from a finite set of atomic propositions @xmath17 , @xmath47 , @xmath48 and the boolean and temporal connectives @xmath49 and @xmath50 ( always ) , @xmath51 ( until ) , @xmath52 ( eventually ) , @xmath53 ( next ) . given an formula @xmath54 as the system specification , one can always represent it by a @xmath55 where @xmath56 is a finite state set , @xmath57 is the alphabet , @xmath58 is the initial state , and @xmath59 is the transition function . the acceptance condition @xmath60 is a set of tuples @xmath61 . the run for an infinite word @xmath62w[1]\ldots \in ( 2^{{\mathcal{ap}}})^\omega$ ] is an infinite sequence of states @xmath63 where @xmath64 and @xmath65)$ ] . a run @xmath66 is accepted in @xmath67 if there exists at least one pair @xmath68 such that @xmath69 and @xmath70 where @xmath71 is the set of states that appear infinitely often in @xmath72 . we define a product operation between a labeled and a . [ def : product ] given a labeled @xmath73 and the @xmath74 , the _ product _ is @xmath75 , with components defined as follows : @xmath76 is the set of states . @xmath10 is the set of actions . @xmath77 $ ] is the initial distribution , defined by @xmath78 where @xmath79 . @xmath80 $ ] is the transition probability function . given @xmath81 , @xmath82 , @xmath83 and @xmath84 , let @xmath85 . the reward function is defined as @xmath86 where given @xmath87 , @xmath83 , @xmath33 , @xmath88 for @xmath33 . the acceptance condition is @xmath89 . @xmath26 the problem of maximizing the probability of satisfying the formula @xmath54 in @xmath30 is transformed into a problem of maximizing the probability of reaching a particular set in the product @xmath90 , which is defined next . @xcite the _ end component _ for the product @xmath91 is a pair @xmath92 where @xmath93 is a non - empty set of states and @xmath94 is a randomized policy . moreover , the policy @xmath39 is defined such that for any @xmath95 , for any @xmath96 , @xmath97 ; and the induced directed graph @xmath98 is strongly connected . here , @xmath99 is an edge in the directed graph if @xmath100 for some @xmath101 . an is an end component such that @xmath102 and @xmath103 for some @xmath104 . @xmath26 let the set of s in @xmath90 be denoted @xmath105 and the set of _ accepting end states _ be denoted by @xmath106 . note that , by definition , for each @xmath107 , by exercising the associated policy @xmath39 , the probability of reaching any state in @xmath108 is 1 . due to this property , once we enter some state @xmath109 , we can find at least one accepting end component @xmath107 such that @xmath95 , and initiate the policy @xmath39 such that for some @xmath110 , all states in @xmath111 will be visited only a finite number of times and some state in @xmath112 will be visited infinitely often . the set @xmath105 can be computed by algorithms @xcite in polynomial time in the size of @xmath90 . we aim to synthesize a shared autonomy policy that switches control between an operator and an autonomous controller . the stochastic system controlled by the human operator and the autonomous controller , gives rise to two different s with the same set of states @xmath4 , the same set @xmath17 of atomic propositions and the same labeling function @xmath113 , but possibly different sets of actions and transition probability functions . * autonomous controller : @xmath114 where @xmath115 $ ] is the transition probability function under autonomous controller . * human operator : @xmath116 where @xmath117 $ ] is the transition probability function under human operator . let @xmath118 $ ] be the initial distribution of states , same for both @xmath119 and @xmath120 . for the same system , the set of physical actions can be the same for both the autonomous controller and the human . we can add subscript to distinguish whose action it is . the models @xmath119 and @xmath120 can be constructed either from prior knowledge or from experiments by applying a policy that samples each action from each state a sufficient amount of times @xcite . in the shared autonomy system , the interaction between the autonomous controller and the operator is often made through a dialogue system @xcite . the controller may send a request of attention , or some other signal to the operator . the operator may grant the request , or respond to signals , depending on his current workload , level of attention . admitting that it is not possible to capture all aspects of an operator s cognitive states , we have the following model to capture the evolution of the modeled cognitive state . the operator s cognition in the shared autonomy system is modeled as an @xmath121 where @xmath122 represents a finite set of cognitive states . @xmath123 is a finite set of events that trigger changes in cognitive state . $ ] is the initial distribution . @xmath125 $ ] is the transition probability function . @xmath126 is the cost function . @xmath127 is the cost of human effort for the transition from @xmath128 to @xmath129 under event @xmath130 . @xmath131 is the discount factor . @xmath132 is a subset of states at which the operator can take over control . @xmath26 this cognitive model can be generalized to accomodate different model of operator s interaction with the autonomous controller . the set @xmath123 of events can be requests sent by the autonomous controller to the operator , a workload that the operator assigns to himself , or any other external event that influences the operator s cognitive state . this model generalizes the model of operator s cognition in @xcite , in which an event is a request to increase , decrease , or maintain the operator s attention in the control task . in particular , it is assumed that in a particular set of states , transitions from the autonomous controller to the operator can happen . for instance , for tele - operated robotic arm or semi - autonomous vehicle , operator may take over control only when he is aware of the system s state and not occupied by other tasks @xcite . the model is flexible and can be extended to other cognitive models in shared autonomy . in this paper , we assume the model of operator for the given task is given . one can obtain such a model by statistical learning @xcite . we illustrate the concepts using the robotic arm example . consider a robot manipulator having to pick up the objects on a table and place it into a box . there are two types of objects , small and large . for small and large objects , the probabilities of a successful pick - and - place maneuver performed by the autonomous controller is @xmath133 and @xmath134 respectively . the for the controller is shown in figure 1a . with an operator tele - operating the robot , the probabilities of a successful pick - and - place maneuver is @xmath135 and @xmath136 respectively . the operator s cognitive model includes two cognitive states : @xmath137 represents the state when the human does not pay any attention to the system ( at the attention level @xmath137 ) , and @xmath138 represents the state when he pays full attention ( at the attention level @xmath138 ) . the set of events in @xmath139 is the requests of human attention to the task , @xmath140 where @xmath141 represents the current requested attention level is @xmath130 . for any @xmath142 , @xmath141 , let @xmath143 for @xmath144 , otherwise @xmath145 . the transition probability function of @xmath139 is shown in figure 1b . represents there are @xmath146 small objects and @xmath147 large objects remaining to be picked . the available actions are @xmath24 and @xmath148 for picking up small and large objects , respectively . the for the robotic arm tele - operated by the human can be obtained by changing the probabilities on the transitions . ( b ) the @xmath139 for modeling the dynamics of human s attention changes.[],scaledwidth=45.0% ] given two s , @xmath119 for the controller and @xmath120 for the operator , and a cognitive model for the operator @xmath139 , we construct a _ shared autonomy stochastic systems _ as an as follows . @xmath149 where @xmath150 is the set of states . a state @xmath151 includes a state @xmath152 of the system and a cognitive state @xmath128 of human . @xmath153 is the set of actions . if @xmath154 , the system is controlled by the autonomous controller and the event affecting human s cognition is @xmath130 . if @xmath155 , the system is controlled by human operator and the event affecting human s cognition is @xmath130 . @xmath156 $ ] is the transition probability function , defined as follows . given a state @xmath157 and action @xmath158 , @xmath159 , which expresses that the controller acts and triggers an event that affects the operator s cognitive state . given a state @xmath151 for @xmath160 , and action @xmath161 , @xmath162 , which expresses that the operator controls the system and an event @xmath130 happens and may affect the cognitive state . $ ] is the initial distribution . @xmath164 , for all @xmath13 , @xmath165 . @xmath166 is the labeling function such that @xmath167 . @xmath168 is a cost function for human effort defined over the state and action spaces and @xmath169 . @xmath131 is the discount factor , the same in @xmath139 . slightly abusing the notation , we denote the cost function in @xmath170 the same as the cost function in @xmath139 and the labeling function in @xmath170 the same as the labeling function in @xmath30 . note that , although the cost of human effort only contains the cost in his cognitive model , it is straightforward to incorporate the cost of human s actions into the cost function . we construct @xmath170 in figure [ fig : samdp ] for the robotic arm example . for example , @xmath171 , which means the probability of the robot successfully picking up a small object and placing it into the box while the human changes his cognitive state to 1 ( fully focused ) upon the robot s request is @xmath172 . also it is noted that from the states @xmath173 and @xmath174 , no human s action is enabled . the cost function is defined such that @xmath175 if @xmath144 , otherwise @xmath145 . for robotic arm example ( note only a subset of states and transitions are shown ) . subscripts @xmath176 and @xmath122 distinguish actions performed by the autonomous controller ( @xmath176 ) and the human ( @xmath122 ) , respectively . , scaledwidth=45.0% ] the main problem we solve is the following . [ def : problem ] given a stochastic system under shared autonomy control between an operator and an autonomous controller , modeled as s @xmath120 and @xmath119 , a model of human s cognition @xmath139 , and an specification @xmath54 , compute a policy that is pareto optimal with respect to two objectives : maximizing the discounted probability of satisfying the specification @xmath54 and minimizing the discounted total cost of human effort over an infinite horizon . the definition of pareto optimality in this context is given formally at the beginning of section [ sec : multiobj ] . by following a pareto optimal policy , we achieve a balance between two objectives : it is impossible to make one better off without making the other one worse off . given an @xmath177 and a @xmath178 , the product following definition [ def : product ] is @xmath179 . recall that the policy maximizing the probability of satisfying the specification is obtained by first computing the set of s in @xmath90 and then finding a policy that maximizing the probability of hitting the set @xmath180 of states contained in s ( see section [ sub : mdpltl ] ) . for quantitative objectives , for example , maximizing the probability of satisfying an formula , or a discounted reward objective over an infinite horizon , a memoryless policy in the product suffices for optimality @xcite . in the following , by policies , we mean memoryless ones in the product . problem [ def : problem ] is in fact a multi - objective optimization problem for which we need to balance the cost of human s effort and satisfaction for constraints . however , the solutions for multi - objective s can not be directly applied due to the constraint that once the system runs into an of @xmath90 , the policy should be constrained such that all states in that are visited infinitely often . based on the particular constraint , we divide the original problem into a two - stage optimization problem : the policy synthesis for s is separated from solving a multi - objective formulated before reaching a state in an . the first stage is to balance between a quantitative criterion for a temporal logic objective and a criterion with respect to the cost of human effort before a state in the set @xmath180 is reached . remind that @xmath180 is the union of states in the accepting end components of @xmath90 . we formulate it as an . however , for objectives of different types , such as , discounted , undiscounted , and limit - average . the scalarization method for solving s does not apply . thus , we consider to use the discounted reachability property @xcite for the given specification , as well as discounted costs for the human attention , with the same discount factor @xmath181 specifying the relative importance of immediate rewards . for an specification , discounting in the state sequence before reaching the set @xmath180 means that the number of steps for reaching @xmath180 is concerned @xcite . without discounting , as long as two policies have the same probability of reaching the set @xmath180 , they are equivalent regardless of their expected numbers of steps to reach @xmath180 . with discounting though , a policy has smaller expected number of steps in reaching @xmath180 is considered to be better than the other . [ def : momdpvaluefunc ] given the product @xmath90 , for a state @xmath182 in @xmath90 , the _ discounted probability _ for reaching the set @xmath180 under policy @xmath183 is @xmath184\ ] ] where the reward function @xmath185 is defined such that @xmath186 if and only if @xmath187 and @xmath188 , otherwise @xmath189 . the discounted total reward with respect to human attention for a policy @xmath190 and a state @xmath182 is @xmath191,\ ] ] where the reward function @xmath192 is defined such that @xmath193 if and only if @xmath194 and @xmath195 , @xmath196 if @xmath187 and @xmath188 , and @xmath197 otherwise . here , @xmath198 is the discounted cost of human attention for remaining in an accepting end components under the optimal policy for the second stage . @xmath26 the _ discounted value profile _ , at @xmath182 for policy @xmath39 , is defined as @xmath199 . we denote @xmath200 as the vector of reward functions . the function @xmath198 is computed in the next section . @xcite given an @xmath201 and a vector of reward functions @xmath202 , for a given state @xmath203 , policy @xmath39 _ pareto - dominates _ policy @xmath204 at state @xmath182 if and only if @xmath205 and for all @xmath206 . a policy @xmath39 is _ pareto optimal _ in a state @xmath207 if there is no other policy @xmath204 pareto - dominating @xmath39 . for a pareto - optimal policy @xmath39 at state @xmath182 , the corresponding value profile @xmath208 is referred to as a _ pareto - optimal point ( or an efficient point)_. the set of pareto - optimal point are called the _ pareto set_. @xmath26 a pareto optimal policy @xmath39 for a given initial distribution is defined analogously by comparing the expectations of value functions under the initial distribution . we employ tchebycheff scalarization method @xcite to find pareto optimal policies for user specified weights . first , we solve a set of single objective s , one for each reward function . let @xmath209 be the value function of the optimal policy @xmath210 with respect to the @xmath211-th reward function . the ideal point @xmath212 is then computed as follows : for @xmath213 , @xmath214 . given a weight vector @xmath215 where @xmath216 is the weight for the @xmath211-th criterion such that @xmath217 , a pareto optimal policy associated with the weight vector @xmath218 can be found with the following nonlinear program : @xmath219 where @xmath220 is a small positive real that can be chosen arbitrarily , @xmath221 is interpreted as the expected discounted frequency of reaching the state @xmath182 and then choosing action @xmath24 , @xmath222 , and @xmath223 is a positive weighting vector computed from a weight @xmath218 , the ideal points and the nadir points @xcite for all reward functions ( detailed in appendix ) . the nonlinear programming problem can then be formulated into a linear programming problem in the standard way by setting a new variable @xmath224 . the pareto optimal policy @xmath225 is defined such that @xmath226 which selects action @xmath24 with probability @xmath227 from the state @xmath182 , for all @xmath203 , @xmath228 . continue with the robot arm example . given the discount factor @xmath229 , for the simple objective ( @xmath138st objective ) as quickly as possible of reaching a state at which all objects are in the box , the optimal strategy @xmath230 is shown in the first row of table [ tbl : pickplacepareto ] . intuitively , the robot starts by requesting the operator to increase his level of attention and wants to switch control to human as soon as possible as the latter has higher probability of success for a pick - and - place maneuver . alternatively , the optimal policy with respect to minimizing the cost of human effort ( @xmath231nd objective ) , is to let the robot pick up all the objects since by doing so , eventually all the objects will be collected into the box . the strategy @xmath232 is shown in the second row of table [ tbl : pickplacepareto ] . now suppose that a user gives a weight @xmath233 for the first objective and @xmath234 for the second objective , through normalization , the new weight vector @xmath235 , is obtained with the method in appendix . by solving the linear programming problem in , we obtain a pareto - optimal policy @xmath236 shown in the third row of table [ tbl : pickplacepareto ] . noting that the difference of @xmath236 and @xmath237 is that when it comes to the small object , if the current human attention is high , the robot will request the human to decrease his attention level and therefore , if the object fails to be picked up through tele - operation , the autonomous controller will take over for picking up the small object . whileas in @xmath237 , the robot prefers the human operator to pick up all objects , no matter it is a big one or a small one . figure [ fig : pickplacepareto ] shows the state value for the initial state @xmath238 with respect to reward functions @xmath239 , under the policies @xmath237 , @xmath232 and a subset of pareto optimal policies , one for each weight vector @xmath218 in the set @xmath240 . [ tbl : pickplacepareto ] , under policies @xmath237 , @xmath232 and a set of pareto optimal policies @xmath236 , one for each weight vectors in the set @xmath241 . the @xmath242-axis and @xmath243-axis represent the values of the initial state under the 1st and 2nd criteria , respectively.,scaledwidth=50.0% ] though the pareto optimal policy for @xmath244 is deterministic in this example . it may generally need to be randomized for a given weight vector . so far we have introduced a method for synthesizing pareto optimal policies before reaching a state in one of the accepting end components . next , we introduce a constrained optimization for synthesizing a policy that minimize the expected discounted cost of staying in an and visiting all the states in that infinitely often . for a state @xmath182 in @xmath180 , one can identify at least one @xmath107 such that @xmath95 . it is noted that the policy @xmath245 is a randomized policy that ensures every state in @xmath108 is visited infinitely often with probability 1 @xcite . however , there might be more than one that contains a state @xmath182 , and we need to decide which to stay in such that the expected discounted cost of human effort for the control execution over an infinite horizon is minimized . we consider a constrained optimization problem : for each @xmath107 where @xmath246 and @xmath247 , solve for a policy @xmath248 such that the cost of human effort for staying in that is minimized . the constrained optimization problem is formulated as follows . @xmath249\\ & \text{subject to : } \forall v\in w , { \mathrm{pr}}^g(\forall t , \exists t'>t , x_{t'}=v ) = 1 , \text { and } \\ & \forall v \in w , \forall a \notin { \gamma}(v ) , g(v)(a)=0 , \end{split}\end{aligned}\ ] ] where the term @xmath250 measures the probability of infinitely revisiting state @xmath182 under policy @xmath251 . the linear program formulated for solving can be obtained as follows : @xmath252\\ & \text{subject to : for } v\in w , \\ & \sum_{a\in { \gamma}(v ) } x(v , a)= \eta(v ) + \gamma \sum_{v'\in v } \sum_{a'\in { \gamma}(v ' ) } \delta(v',a ' , v ) \cdot x(v ' , a'),\\ & \forall v\in w , \forall a \in \sigma , x(v , a ) \ge 0,\\ & \forall v \in w , \sum_{a\in { \gamma}(v ) } x(v , a ) > = \varepsilon , \text { and } \\ & \forall v \in w , \forall a \notin { \gamma}(v ) , x(v , a ) = 0 , \end{split}\end{aligned}\ ] ] where @xmath253 is an arbitrarily small positive real . @xmath254 $ ] is the initial distribution of states when entering the set @xmath108 . because for single objective optimization the optimal state value does not depend on the initial distribution @xcite , @xmath255 can be chosen arbitrarily from the set of distributions over @xmath108 . the physical meaning of @xmath256 is the discounted frequency of visiting the state @xmath182 , which is strictly smaller than the frequency of visiting the state @xmath182 as long as @xmath257 . by enforcing the constraints @xmath258 , we ensure that the frequency of visiting every state in @xmath108 is non - zero , i.e. , all states in @xmath108 will be visited infinitely often . the solution to produces a memoryless policy @xmath259 that chooses action @xmath24 at a state @xmath182 with probability @xmath260 . using policy evaluation @xcite , the state value @xmath261 for each @xmath95 under the optimal policy @xmath262 can be computed . then , the terminal cost @xmath263 is defined as follows . @xmath264 and the policy after hitting the state @xmath182 is @xmath251 such that @xmath265 . we now present algorithm [ alg : twostage ] to conclude the two - state optimization procedure . [ [ remark ] ] remark + + + + + + although in this paper we only considered two objectives , the methods can be easily extended to more than two objectives for handling specifications and different reward / cost structures in synthesis for stochastic systems , for example , the objective of balancing between the probability of satisfying an formula , the discounted total cost of human effort , and the discounted total cost of energy consumption . we apply algorithm [ alg : twostage ] to a robotic motion planning problem in a stochastic environment . the implementations are in python and matlab on a desktop with intel(r ) core(tm ) processor and 16 gb of memory . figure 4a shows a gridworld environment of four different terrains : pavement , grass , gravel and sand . in each terrain , the mobile robot can move in four directions ( heading north ` n ' , south ` s ' , east ` e ' , and west ` w ' ) . there is onboard feedback controller that implements these four maneuver , which are motion primitives . using the onboard controller , the probability of arriving at the correct cell is @xmath135 for pavement , @xmath266 for grass , @xmath136 for gravel and @xmath267 for sand . alternatively , if the robot is operated a human , it can implement the four actions with a better performance for terrains grass , sand and gravel . the probability of arriving at the correct cell under human s operation is @xmath135 for pavement , @xmath268 for grass , @xmath133 for gravel and @xmath266 for sand . the objective is that either the robot has to visit region @xmath269 and then @xmath270 , in this order , or it needs to visit region @xmath271 infinitely often , while avoiding all the obstacles . formally , the specification is expressed with an formula @xmath272 . figure 4b is the cognitive model of the operator , including three states : @xmath273 , @xmath30 and @xmath122 represent that human pays low , moderate , and high attention to the system respectively . the costs of paying low , moderate and high attention to the system are @xmath138 , @xmath145 , and @xmath274 , respectively . action ` @xmath275 ' ( resp . @xmath276 ) means a request to increase ( resp . decrease ) the attention and action @xmath277 means a request to maintain the current attention . the operator takes over control at state @xmath122 . gridworld , where the disk represents the robot , the cells @xmath269 , @xmath270 , and @xmath271 are the interested regions , the crossed cells are obstacles . we assume that if the robot hits the wall ( edges ) , it will be bounced back to the previous cell . different grey scales represents different terrains : from the darkest to the lightest , these are `` sand , '' `` grass , '' `` pavement '' and `` gravel . '' ( b ) the @xmath139 of the human operator.,scaledwidth=45.0% ] during control execution , we aim to design a policy that coordinates the switching of control between the operator and the autonomous controller , i.e. , onboard software controller . the policy should be pareto optimal in order to balance between maximizing the expected discounted probability of satisfying the formula @xmath54 , and minimizing the expected discounted total cost of human efforts . figure [ fig : gridworldpareto ] shows the state value for the initial state with respect to reward functions @xmath278 for the formula and @xmath279 for the cost of human effort , under the single objective optimal policy @xmath237 and @xmath232 , and a subset of pareto optimal policies , one for each weight vectors @xmath218 in the set @xmath280 . for the specification , all policies are randomized . , under policies @xmath237 , @xmath232 and a set of pareto optimal policies @xmath236 , for each @xmath281 . the @xmath242-axis represents the values of the initial state for discounted probability of satisfying the specification . the @xmath243-axis represents the values of the initial state with respect to the cost of human effort.,scaledwidth=40.0% ] we developed a synthesis method for a class of shared autonomy systems featured by switching control between a human operator and an autonomous controller . in the presence of inherent uncertainties in the systems dynamics and the evolution of humans cognitive states , we proposed a two - stage optimization method to trade - off the human effort for the system s performance in satisfying a given temporal logic specification . moreover , the solution method can also be extended for solving multi - objective s with temporal logic constraints . in the following , we discuss some of the limitations in both modeling and solution approach in this paper and possible directions for future work . we employed two s for modeling the system operated by the human and for representing the evolution of cognitive states triggered by external events such as workload , fatigue and requests for attention . we assumed that these models are given . however , in practice , we might need to learn such models through experiments and then design adaptive shared autonomy policies based on the knowledge accumulated over the learning phase . in this respect , a possible solution is to incorporate joint learning and control policy synthesis , for instance , pac - mdp methods @xcite , into multi - objective s with temporal logic constraints . another limitation in modeling is that the current cognitive model can not capture all possible influences of human s cognition on his performance . consider , for instance , when the operator is bored or tired , his performance in some tasks can be degraded , and therefore the transition probabilities in @xmath120 are dependent on the operator s cognitive states . in this case , we will need to develop a different product operation for combining the three factors : @xmath119 , a set of @xmath120 s for different cognitive states , and @xmath139 , into the shared autonomy system . despite the change in modeling the shared autonomy system , the method for solving pareto optimal policies developed in this paper can be easily extended . consider a multiobjective @xmath282 where @xmath283 is a vector of reward functions and @xmath284 is the discount factor , let @xmath285 be the vectorial value function optimal for the @xmath211-th criterion , specified with the reward function @xmath286 . an approximation of the nadir point for the @xmath211-th criterion is computed as follows , @xmath287 where @xmath288 is a vector value function obtained by evaluating the optimal policy for the @xmath289-th criterion with respect to the @xmath211-th reward function . the weight vector after normalization is defined as @xmath290 b. pitzer , m. styer , c. bersch , c. duhadway , and j. becker , `` towards perceptual shared autonomy for robotic mobile manipulation , '' in _ ieee international conference on robotics and automation _ , may 2011 , pp . 62456251 . k. kinugawa and h. noborio , `` a shared autonomy of multiple mobile robots in teleoperation , '' in _ proceedings of ieee international workshop on robot and human interactive communication _ , 2001 , pp . 319325 . s. gnatzig , f. schuller , and m. lienkamp , `` human - machine interaction as key technology for driverless driving - a trajectory - based shared autonomy control approach , '' in _ ieee international symposium on robot and human interactive communication _ , sept 2012 , pp . 913918 . w. li , d. sadigh , s. sastry , and s. seshia , `` , '' in _ _ , ser . lecture notes in computer science , e. brahm and k. havelund , eds.1em plus 0.5em minus 0.4emspringer berlin heidelberg , 2014 , vol . 8413 , pp . 470484 . k. chatterjee , r. majumdar , and t. a. henzinger , `` markov decision processes with multiple objectives , '' in _ symposium on theoretical aspects of computer science_.1em plus 0.5em minus 0.4emspringer , 2006 , pp . 325336 . v. forejt , m. kwiatkowska , and d. parker , `` pareto curves for probabilistic model checking , '' in _ proceedings of 10th international symposium on automated technology for verification and analysis _ , ser . lncs , s. chakraborty and m. mukund , eds . 7561.1em plus 0.5em minus 0.4emspringer , 2012 , pp . 317332 . p. perny and p. weng , `` on finding compromise solutions in multiobjective markov decision processes , '' in _ proceedings of the 19th european conference on artificial intelligence_.1em plus 0.5em minus 0.4emios press , 2010 , pp . 969970 . i. das and j. e. dennis , `` a closer look at drawbacks of minimizing weighted sums of objectives for pareto set generation in multicriteria optimization problems , '' _ structural optimization _ 14 , no . 1 , pp . 6369 , 1997 . k. chatterjee , m. henzinger , m. joglekar , and n. shah , `` symbolic algorithms for qualitative analysis of markov decision processes with bchi objectives , '' _ formal methods in system design _ , vol . 42 , no . 3 , pp . 301327 , 2013 . d. henriques , j. g. martins , p. zuliani , a. platzer , and e. m. clarke , `` statistical model checking for markov decision processes , '' in _ 9th international conference on quantitative evaluation of systems _ , 2012 , pp . 8493 . mouaddib , s. zilberstein , a. beynier , l. jeanpierre , _ et al . _ , `` a decision - theoretic approach to cooperative control and adjustable autonomy . '' in _ european conference on artificial intelligence _ , 2010 , pp .	in systems in which control authority is shared by an autonomous controller and a human operator , it is important to find solutions that achieve a desirable system performance with a reasonable workload for the human operator . we formulate a shared autonomy system capable of capturing the interaction and switching control between an autonomous controller and a human operator , as well as the evolution of the operator s cognitive state during control execution . to trade - off human s effort and the performance level , e.g. , measured by the probability of satisfying the underlying temporal logic specification , a two - stage policy synthesis algorithm is proposed for generating pareto efficient coordination and control policies with respect to user specified weights . we integrate the tchebychev scalarization method for multi - objective optimization methods to obtain a better coverage of the set of pareto efficient solutions than linear scalarization methods .
strongly correlated superconductors such as the cuprates,@xcite heavy fermions,@xcite and organic charge transfer salts,@xcite share many phenomena not found in weakly correlated metals and superconductors . examples of such effects include , superconductivity in close proximity to the mott transition,@xcite metallic states not described by landau s theory of fermi liquids,@xcite small superfluid stiffnesses in the superconducting state@xcite and pseudogap phenomena.@xcite important questions about these materials include : do these phenomena have a common origin and how similar are the phenomena observed in the different families of materials ? one common feature of the materials discussed above is that they display strong antiferromagnetic spin fluctuations.@xcite this can be seen in the spin lattice relaxation rate , @xmath5 , which is very different from those found in weakly correlated materials ( where @xmath6 cf . ref ) . in many cuprates , heavy fermion materials , and organic superconductors @xmath7 is strongly temperature dependent : as the temperature is lowered from room temperature @xmath7 increases until it reaches a maximum at a temperature we label @xmath8 . below @xmath8 a suppression of spectral weight is manifest in @xmath7 due to the opening of a superconducting gap or a pseudogap . throughout this paper , when we refer to the pseudogap regime of the organic superconductors we mean to indicate the temperatures between @xmath8 and @xmath9 , the superconducting critical temperature , where there is a loss of spectral weight evident in the nmr , but no bulk superconductivity . it has recently been shown that in the high temperature regime ( i.e. , where there is neither a pseudogap or superconductivity ) simple scaling relations , based on a phenomenological two - fluid picture , describe the spin lattice relaxation rate , @xmath5 , in a large number of high temperature superconductors@xcite and heavy fermion materials.@xcite further , it has been shown that this two fluid model can describe many other experiments on the heavy fermion materials.@xcite therefore , given the similarities between these materials and the organic charge transfer salts,@xcite it is natural to ask whether similar scaling behaviours describe the behaviour of the organic charge transfer salts . in this paper we focus on this question in the context of the @xmath0-(et)@xmath10 family of organic charge transfer salts , however , we also make some comments on ( tmtsf)@xmath1clo@xmath4 . an important difference between the cuprates and heavy fermion materials , and the @xmath0-(et)@xmath10 salts is that the @xmath0-phase organics there is a single half - filled band , whereas the other materials are multiband and/or doped . the mott transition in @xmath0-(et)@xmath10 , where @xmath11 is a monovalent anion , can be driven by applying a hydrostatic pressure or varying the anion , which is often referred to as chemical pressure . it is believed that this corresponds to increasing the ratio of @xmath12 in the single band hubbard model description of these systems ( cf . [ fig : aniso]).@xcite the behaviour of the spin degrees of freedom in are particularly interesting . whereas the insulating phases of other @xmath0-(et)@xmath10 salts order antiferromagnetically , the insulating phase of shows no signs of magnetic order to the lowest temperatures investigated ( @xmath1332 mk).@xcite as the exchange energy extracted@xcite from fits to the high temperature bulk susceptibility is @xmath13250 k this has been taken as evidence that is a spin liquid.@xcite suffers from stronger geometrical frustration than other @xmath0-(et)@xmath10 salts : tight binding calculations@xcite indicate that the band structure of is close to that of an equilateral triangular lattice ( i.e. , @xmath14 ; cf . [ fig : aniso ] ) , whereas other @xmath0-(et)@xmath10 materials are described by an anisotropic triangular lattice . ( @xmath15);@xcite more importantly , fits of high temperature series expansions to the bulk susceptibility show that @xmath16 in but that @xmath17 in .@xcite a pressure of greater than @xmath130.3 gpa drives the ground state of from a spin liquid to a superconductor . to date very little is known experimentally about this superconducting state . it has recently been argued@xcite that the frustration will drive changes in the spin fluctuations which will lead to a superconducting state with broken time reversal symmetry , i.e. , a @xmath18 ( or more strictly @xmath19 ) state . therefore it is important to understand the nature of the spin fluctuations in in both the insulating and the metallic phases . in order to investigate the spin fluctuations we present , below , a theoretical analysis of previously published nmr experiments . in section [ sect : hight ] we show that although nmr experiments on the organic charge transfer salts are rather similar to those on the cuprates and heavy fermion materials there are some important differences . a detailed analysis shows that , in the @xmath0-phase organics , the pseudogap energy scale is set by the spin fluctuations , while in ( tmtsf)@xmath1clo@xmath4 the superconducting gap is the same size as the characteristic spin fluctuations . in section [ sect : coh ] we compare the pseudogap energy scale with the temperature at which coherence emerges in the intralayer transport . we find that they are the same to within experimental error and propose a phenomenological interpretation of this . in section [ sect : highp ] we discuss the behaviour of the organic charge transfer salts under hydrostatic pressures and raise some important issues about the pseudogap that have not yet been addressed experimentally . finally , we draw our conclusions in section [ sect : conc ] . in the two - fluid model proposed by pines _ et al_. @xcite the spin lattice relaxation rate is given by @xmath20 where @xmath0 and @xmath21 are material dependent constants , the latter measuring the proximity to a putative quantum critical point , and @xmath8 is the temperature where there is a maximum in @xmath7 . the two fluids are proposed to be a fermi liquid component and a spin liquid component . we plot previously published@xcite nmr data for both insulating and metallic phases of organic charge transfer salts in this ` scaling ' form in fig . [ fig : scaling ] . for the metallic salts it can be seen that , while there is a broad trend in the data , the data do not collapse onto a single curve as they do in the cuprates and the heavy fermion materials.@xcite this suggests that two fluid model may not be relevant to the organic charge transfer salts , but also that there is something to be learnt from fig . [ fig : scaling ] . for the insulating states the spin lattice relaxation is very different for those compounds that order magnetically ( and fully dueterated \{henceforth } ) from that in the spin liquid . indeed the data for is remarkably similar to the data for the metallic salts . important evidence that a two - fluid model is required in the heavy fermion materials comes from comparing the knight shift , @xmath22 , to the bulk susceptibility @xmath23 . according to the two fluid model @xmath24\chi_{sl } , \label{eqn:2f - chi}\ ] ] where @xmath25 is the fraction of electrons in the fermi liquid , @xmath26 is the susceptibility of the fermi liquid , which is basically independent of temperature , and @xmath27 is the susceptibility of the spin liquid . further , the two - fluid model predicts that @xmath28\chi_{sl } , \label{eqn:2f - ks}\ ] ] where @xmath29 is the hyperfine coupling constant between the itinerant electrons and the probe nuclei and @xmath30 is hyperfine coupling of the spin liquid to the nuclei . hence , if @xmath31 then @xmath22 will not be proportional to @xmath23 . systematic differences between temperature dependence of @xmath22 and that of @xmath23 are indeed found in the heavy fermion materials.@xcite in the high temperature ( @xmath32 ) regime @xmath33 in ( refs . ) , ( ref . ) or ( ref . ) . we are not aware of any reports of @xmath22 in either or , or data for @xmath22 at high temperatures in ( tmtsf)@xmath1clo@xmath4 . there is only a single band that plays an important role in the organics , so one might reasonably argue that @xmath34 . nevertheless the comparison of @xmath22 and @xmath23 does not force one to consider a two fluid model and we now move on to discuss another possible explanation of the data in fig . [ fig : scaling ] . moriyas@xcite theory of spin fluctuations in nearly antiferromagnetic metals , particularly in the phenomenological form pioneered by millis , monien , and pines,@xcite has been shown to give good agreement with nmr experiments in the cuprates,@xcite heavy fermion materials,@xcite and the organics.@xcite therefore , it is natural to ask whether this can explain the similarities between the rescaled data for the various materials in fig . [ fig : scaling ] . the simplest assumption@xcite for the temperature dependence of the spin correlation length is @xmath35 . here the temperature dependence of the spin fluctuations is controlled by a single parameter , @xmath36 , thus one finds@xcite that , in the limit of strong spin fluctuations , @xmath37 where @xmath38 ^ 2 $ ] , @xmath39 is the lattice spacing , and @xmath40 is a material specific constant . for @xmath41 , i.e. , if the magnetic correlation length is large and the temperature is low , one can ` rescale ' this to @xmath42 so that material specific part cancels and we find that @xmath43 thus the spin fluctuation model predicts that , for nearly antiferromagnetic metals , @xmath44 is linear in @xmath45 . this is in good agreement with the data for the metallic materials in fig [ fig : scaling]a . further , the spin fluctuation theory for a nearly antiferromagnetic metal constrains the gradient of the data to lie between 0 ( @xmath46 , for a material with neither a pseudogap nor superconductivity ) and 1 ( @xmath47 , for material with no spin fluctuations ) , cf . ( [ scaling ] ) . indeed , we find that @xmath48 for all of the metallic organic charge transfer salts . this last result is an important conclusion , as it suggests that the pseudogap energy scale is set by that of the spin fluctuations . note that in ( tmtsf)@xmath1clo@xmath4 the maximum in the @xmath7 occurs at the superconducting critical temperature , i.e. , @xmath49 . this suggests that spin fluctuations may determine the superconducting critical temperature and thus the spin fluctuations may mediate the superconductivity in ( tmtsf)@xmath1clo@xmath4 . this is consistent with the unconventional superconducting state observed in this material.@xcite for materials that order antiferromagnetically , one expects , within mean field theory ( whence the correlation length critical exponent @xmath50 ) , that @xmath51 , where @xmath52 is the nel temperature . thus @xmath53 where @xmath54 ^ 2 $ ] and @xmath39 is the lattice spacing . rescaling we find that , @xmath55 note that the maximum in @xmath7 does not occur at @xmath52 due to short range correlations . rather , @xmath7 has a infinite tangent at @xmath52 , as does the bulk susceptibility.@xcite therefore on very general grounds@xcite one expects @xmath56 . a sharp decline in @xmath7 below @xmath8 is observed in the data for ( ref . ) and ( ref . ) , consistent with these expectations . thus eq . ( [ scalingafm ] ) provides an excellent description of the data for and ( shown in fig . [ fig : scaling]b ) . note in particular that eq . ( [ scalingafm ] ) , correctly , predicts a steep gradient and a negative intercept in agreement with the data , and in contrast to eq . ( [ scaling ] ) and the data for the non - magnetic salts . the rescaled nmr relaxation rate in metallic ( data at 0.35 and 0.4 gpa ) are remarkable similar to those in and which are much more weakly frustrated . this can be understood because @xmath57 depends only the the integral over the brillouin zone of the dynamic susceptibility . the spin fluctuation theory , which gives a good description of all of the data , only assumes that the dynamic susceptibility has a peak _ somewhere _ in the brillouin zone and that the peak is away from the origin . the predicted scaled @xmath58 is independent of the location of the peak . thus this data does not indicate which @xmath59 has the strongest magnetic fluctuations , but does show that the magnetic fluctuations are just as strong in metallic as they are in the more weakly frustrated @xmath0-(et)@xmath10 salts . the rescaled nmr relaxation rate in insulating ( data at 0 and 0.3 gpa ) is very different from that in the insulating phases of and . instead the data for the insulating phase of is well described by eq . ( [ scaling ] ) , which describes the data in the metallic materials considered above . thus it appears that the central difference between the scaled nmr relaxation rates is not whether the material is insulating or metallic but whether the material orders magnetically or not . we note that both data sets fit remarkably well to the prediction of eq . ( [ scaling ] ) for @xmath60 in the insulating phases of . this brings to mind the weak coupling bcs form formula for the pairing temperature in a superconductor . however , at this stage we have no evidence that this is any more than a numerical coincidence . we now turn our attention to how the pseudogap energy scale , @xmath61 , is related to the other energy scales in the @xmath0-phase organic charge transfer salts . fig . [ fig : phase ] shows that , for the weakly frustrated salts \ { , , and } @xmath61 coincides with the temperature below which the resistivity varies quadratically , @xmath62 , and the temperature at which a broad , deep minimum is observed in the ultrasonic velocity , @xmath63 . further , when these quantities are plotted against the superconducting transition temperature , @xmath64 , all three materials show the same trend . [ fig : phase ] represents a significant quantitative success for the chemical pressure hypothesis , which holds that the major effect of changing the anions is to alter size of the unit cell and therefore control the strength of the electronic correlations and which underpins a great deal of the thinking on the organic charge transfer salts . however , the chemical pressure hypothesis clearly fails for , which does not share the same scaling of temperature scales as the other salts investigated . further , @xmath8 is clearly rather different from @xmath65 in . this suggests that the metallic state of is rather different from the metallic states of the more weakly frustrated @xmath0-phase organic charge transfer salts , consistent with the idea that the spin fluctuations in this material differ from those in its less frustrated cousins in important ways.@xcite nevertheless the observation of a quadratic temperature dependence of the resistivity at temperatures slightly above @xmath9 shows that charge transport is coherent in the metallic state of under pressure . the observation that the crossover to a quadratic temperature dependence of the resistivity approximately coincides with the minimum in the ultrasonic attenuation has been understood@xcite in terms of dynamical mean field theory ( dmft ) . dmft predicts a crossover from a fermi liquid at low temperatures to a ` bad metal ' ( characterised by the absence of quasiparticles and the drude peak , and a resistivity that exceeds the mott - ioffe - regal limit ) as the temperature is increased above a coherence temperature , @xmath66 . thus it is believed that @xmath67 . however , while dmft gives an adequate description of the nuclear spin relaxation rate above @xmath68 , below @xmath69 dmft predicts a fermi liquid and thus a constant @xmath7 ; this is clearly _ not _ what is observed.@xcite this shows that dmft does not capture all of the relevant physics below @xmath69.@xcite it is also interesting to note that dmft does describe the observed behaviour of probes of charge degrees of freedom , such as the resistivity , and only fails for probes of the spin degrees of freedom , such as @xmath7 . as dmft is a purely local theory a reasonable hypothesis it that the relevant physics , not described by dmft , involves short - range spin correlations . further , the enhanced korringa ratios observed@xcite at temperatures slightly above @xmath9 suggest that antiferromagnetic correlations remain important at in the coherent transport regime.@xcite the spin degrees of freedom in the pseudogap regime of the @xmath0-phase organics behave in much the same way as the spin degrees of freedom in the pseudogap regime of the underdoped cuprates . there has been significant debate as to whether , and if so how , the pseudogap in the cuprates is related to the other exotic phenomena seen in the normal state , such as the linear temperature dependence of the resistivity . in this context it is interesting to note that in the region of the phase diagram of the organics where the pseudogap is found , the resistivity varies quadratically with temperature@xcite and the magnitude of the coefficient of the quadratic term in the resistivity is as expected from fermi liquid theory given the observed effective mass.@xcite further , clear evidence of quasiparticles is seen via quantum oscillation experiments.@xcite therefore , in , , and the loss of spectral weight in the pseudogap is not associated with non - fermi liquid behaviour . an important difference between the cuprates and the organics is that the organics are half filled whereas the cuprates are more strongly correlated , doped systems . in this context it is worth noting that a linear resistivity has recently been reported in an organic charge transfer salt with an anion layer that has a lattice constant that is incommensurate with the lattice constant of the organic layer.@xcite the authors argued that this non - stoichiometric organic charge transfer salt is effectively doped away from half filling . an important question in understanding the phenomenology of the @xmath0-(et)@xmath10 salts is : do the two energy scales , @xmath69 and @xmath8 remain equal as pressure is increased and we move further from the mott transition ? this question is difficult to answer at present because there is little experimental data for high pressures ( including high chemical pressure , i.e. , materials with low @xmath9 s ) . there is however tantalising evidence that something rather interesting happens to the superconducting state at high pressures.@xcite in particular while the materials near the mott transition have a superfluid stiffness , @xmath70 , within a factor of two or so of the prediction of bcs theory , at high pressures the superfluid penetration depth , @xmath71 , increases as @xmath72 ( ref . ) . with some materials having superfluid stiffnesses ( @xmath73 ) that are an order of magnitude smaller than the prediction of bcs theory.@xcite we are only aware of one nmr experiment at high pressures in these materials . ref . reports data for at 3 kbar ( which leads to @xmath74 k)@xcite and 4 kbar ( @xmath75 k ) . at these pressures strong spin fluctuations are not observed , and the @xmath7 looks quite conventional . yet for the 3 kbar data there is small but noticeable decrease in @xmath7 below @xmath76 k. is this the last vestige of the pseudogap ? if so , it suggests that at high pressures the pseudogap and the coherent intralayer transport energy scales are different . either way more experiments are clearly required to understand where the pseudogap vanishes . on the basis of the above discussion we propose that a number of new features should be included in the phase diagram of these materials , which we sketch in fig . [ fig : new_phase ] . we stress that this phase diagram is relevant to the weakly frustrated materials for which the chemical pressure hypothesis holds , and therefore does not include which would necessitate an additional axis to include the effects of frustration . we have included the nernst region where nam _ et al_. @xcite have observed a large nernst effect above @xmath9 in at ambient pressure , which they find to be absent in at ambient pressure . note that we have drawn the @xmath8 and the @xmath9 lines both supressed to zero at the same pressure . this is deliberately provocative . as we have stressed above there is insufficient experimental data to determine the relative order in which the superconductivity and the pseudogap disappear as pressure is increased . the issue of where the superconductivity and pseudogap vanish is related to an ongoing debate in the cuprates . a recent review of a wide range of experimental data in a wide variety of cuprates suggested that the pseudogap and superconductivity both vanish at the same critical doping.@xcite similarly , more detailed experimental and theoretical studies of the organics in the vicinity of the pressure above which superconductivity vanishes might give important insights into how the pseudogap is related to superconductivity . the possibility of a quantum critical point somewhere in the vicinity of the pressure where the superconducting critical temperature goes to zero may have important consequences for the observation that the materials with the lowest superconducting critical temperatures have extremely small superfluid stiffnesses and are very different from bcs superconductors.@xcite finally , we sketch an explanation of the observed phase diagram ( fig . [ fig : new_phase ] ) . we have noted above that dmft describes the competition between the insulating , bad metal , and fermi liquid phases.@xcite but , dmft fails to predict either the pseudogap or the unconventional superconductivity , which suggests that these effects involve non - local physics . however , it has recently been argued that the resonating valence bond ( rvb ) theory can describe the superconducting state of the quasi - two - dimensional organic charge transfer salts,@xcite this theory also predicts a pseudogap with the approximately the right energy scale . therefore , we propose that dmft captures the crossover from incoherent to coherent charge transport , but is insufficient to describe the behaviour in the coherent regime because short - range spin correlations play a significant role here . the simplest theory that can capture the low temperature physics of is the rvb theory . however , the rvb theory does not capture the loss of coherence as the temperature is raised , which we argue leads to a cut off of the pseudogap phenomena . if this speculation is correct , the challenge is then to produce a single theory capable of describing all of the physics , including the large nernst effect above @xmath9 in ( ref . ) , which neither the dmft nor the rvb theory predicts . therefore cellular dmft calculations , which can describe both short - range spin fluctuations and the loss of intralayer coherence , may have an important role to play in understanding the organic superconductors . we have argued that a two - fluid description is not required for the organic charge transfer salts . it is interesting to speculate why this is . one possibility is that the difference between the organic salts discussed above and many other strongly correlated superconductors is that the organics are stoichiometric , while , for example , the metallic cuprates are doped systems and the heavy fermions show a subtle hybridisation between almost localised states and conduction electrons . therefore it would be interesting to measure the nuclear relaxation rate in the recently discovered non - stoichometric organic superconductors . we have seen that there are strong similarities in the rescaled spin lattice relaxation across the non - magnetic organic charge transfer salts . these similarities can be understood in terms of spin fluctuations . further , our analysis suggests that the energy scale for the spin fluctuations may set the energy scale for the pseudogap in the organic charge transfer salts . in the weakly frustrated metals [ all the metals studied bar ] the pseudogap opens at the same temperature as coherence emerges in the ( intralayer ) transport . we argued that this is because spin correlations , which are responsible for the pseudogap , are cut off by the loss of intralayer coherence at high temperatures . in these weakly frustrated materials the data can be compared across materials quite reliably , consistent with the chemical pressure hypothesis . in contrast the metallic state of , which is highly frustrated , was shown to be rather different from those of the other materials . finally , we have also shown that in ( tmtsf)@xmath1clo@xmath4 the characteristic temperature scale of the spin fluctuations is the same as @xmath9 . this suggests that spin fluctuations may mediate the superconductivity in the bechgaard salts . we would like to thank s. brown , k. kanoda , j. merino , and h. monien for useful conversations . this work was supported in part under australian research council s discovery projects funding scheme ( project dp0557532 ) . is supported by the australian research council under the queen elizabeth ii scheme ( project dp0878523 ) . is supported by the australian research council under the apf scheme ( project dp0877875 ) . y. j. uemura , l. p. le , g. m. luke , b. j. sternlieb , w. d. wu , j.h . brewer , t. m. riseman , c. l. seaman , m. b. maple , m. ishikawa , d. g. hinks , j. d. jorgensen , g. saito and h. yamochi , phys . lett . * 66 * , 2665 ( 1991 ) . powell and r.h . mckenzie , phys . lett . * 94 * , 047004 ( 2005 ) ; j. y. gan , y. chen , z. b. su , and f. c. zhang , _ ibid_. * 94 * , 067005 ( 2005 ) ; j. liu , j. schmalian , and n. trivedi , _ ibid_. * 94 * , 127003 ( 2005 ) . c. strack , c. akinci , v. paschenko , b. wolf , e. uhrig , w. assmus , m. lang , j. schreuer , l. wiehl , j. a. schlueter , j. wosnitza , d. schweitzer , j. mller , and j. wykhoff , phys . b * 72 * , 054511 , ( 2005 ) .	we show that there are strong similarities in the spin lattice relaxation of non - magnetic organic charge transfer salts , and that these similarities can be understood in terms of spin fluctuations . further , we show that , in all of the @xmath0-phase organic superconductors for which there is nuclear magnetic resonance data , the energy scale for the spin fluctuations coincides with the energy scale for the pseudogap . this suggests that the pseudogap is caused by short - range spin correlations . in the weakly frustrated metals @xmath0-(bedt - ttf)@xmath1cu[n(cn)@xmath1]br , @xmath0-(bedt - ttf)@xmath1cu(ncs)@xmath1 , and @xmath0-(bedt - ttf)@xmath1cu[n(cn)@xmath1]cl ( under pressure ) the pseudogap opens at the same temperature as coherence emerges in the ( intralayer ) transport . we argue that this is because the spin correlations are cut off by the loss of intralayer coherence at high temperatures . we discuss what might happen to these two energy scales at high pressures , where the electronic correlations are weaker . in these weakly frustrated materials the data is well described by the chemical pressure hypothesis ( that anion substitution is equivalent to hydrostatic pressure ) . however , we find important differences in the metallic state of @xmath0-(bedt - ttf)@xmath1cu@xmath1(cn)@xmath3 , which is highly frustrated and displays a spin liquid insulating phase . we also show that the characteristic temperature scale of the spin fluctuations in ( tmtsf)@xmath1clo@xmath4 is the same as superconducting critical temperature , which may be evidence that spin fluctuations mediate the superconductivity in the bechgaard salts . 3@xmath0-(et)@xmath1cu@xmath1(cn)@xmath3 3@xmath0-(cn)@xmath3 8pbr@xmath0(d8)-br c
one of the most fascinating phenomena in quantum mechanics is a quantum tunneling through a barrier . the tunneling shows up in zener tunneling in biased semiconductors , quantum devices like diodes , scanning tunneling microscopy , @xmath0 decay in heavy nuclei , etc . in general , however , the tunneling rate (: tunneling power ) is very small or the tunneling time is very long . even in the case of resonant tunneling of electrons through hetero - structures @xcite , there is room for research on the tunneling time @xcite . therefore it is desirable to invent a protocol of accelerating the tunneling . masuda and nakamura @xcite investigated a way to accelerate quantum dynamics with use of a characteristic driving potential determined by the additional phase of a wave function . one can accelerate a given quantum dynamics to obtain a target state in any desired short time . this kind of acceleration is called the fast forward @xcite of quantum dynamics , which constitutes one of the promising ways of the shortcut to adiabaticity @xcite . the relationship between the fast forward and the shortcut to adiabaticity is nowadays clear@xcite . the idea of the tunneling seems to be incompatible with that of fast forward . but here one can combine these two ideas , that is , one can conceive a theory to accelerate the tunneling dynamics through the high barrier and complete , in any desired short time , the tunneling phenomena which originally needs a long tunneling time . before embarking upon the main part of the text , we briefly summarize the previous theory of the fast - forward of quantum dynamics by masuda and nakamura @xcite . the schrdinger equation in standard time with a nonlinearity constant @xmath1 ( appearing in macroscopic quantum dynamics ) is represented as @xmath2 @xmath3 is a known function of space @xmath4 and time @xmath5 under a given potential @xmath6 and is called a standard state . for any long time @xmath7 called as a standard final time , we choose @xmath8 as a target state that we are going to generate . let @xmath9 be a fast - forwarded state of @xmath10 as defined by @xmath11 with @xmath12 where @xmath5 is a new time variable distinct from the standard one . @xmath13 is a magnification time - scale factor defined by @xmath14 we consider the fast - forward dynamics with a new time variable which reproduces the target state @xmath15 in a shorter final time @xmath16 defined by @xmath17 since the generation of @xmath18 requires an anomalous mass reduction , @xmath18 as it stands can not be a candidate for the fast - forward state@xcite . but one can obtain the target state by considering a fast - forwarded state @xmath19 which differs from @xmath20 by an extra phase as @xmath21 where @xmath22 is a real function of @xmath4 and @xmath5 and is called the additional phase . with use of a new time variable @xmath5 that appeared in eq.([ff - t ] ) , the schrdinger equation for fast - forward state @xmath23 is supposed to be given as @xmath24 with the driving scalar potential @xmath25 . if we shall choose @xmath26 as in eq.([alpha ] ) , the additional phase can vanish at the final time of the fast - forward @xmath27 and we can obtain the exact target state @xmath28 the explicit expression for @xmath26 in the fast - forward range ( @xmath29 ) is proposed by masuda and nakamura@xcite as : @xmath30 where @xmath31 is the mean value of @xmath26 and is given by @xmath32 . besides the time - dependent scaling factor in eq.([nonunif - scale ] ) in the fast - forward range , we can also have recourse to the uniform scaling factor : @xmath33 which will be useful in the quantitative analysis of fast forward . substituting eqs.([mn - standeq]),([ff - t ] ) and ( [ add - f ] ) into eq.([mn - ffeq ] ) and taking its real and imagery parts , we obtain a pair of equations for @xmath34 and @xmath35 , which are solvable . while the above idea guarantees the exact target state at @xmath36 , in the intermediate time range @xmath37 it accelerates only the amplitude of the wave function and fails to accelerate its phase because of the non - vanishing additional phase @xmath34 in eq.([add - f ] ) on the way . if one wish to accelerate the time - dependent current , one must innovate the theory to recover the phase exactly in the intermediate time range until @xmath36 , which will be done below . our theory will hold to both quantum dynamics described by the schrdinger equation and macroscopic quantum dynamics described by the nonlinear schrdinger equation . section [ new appro ] is concerned with a new framework of the exact fast forward of both quantum and macroscopic quantum dynamics . sections [ ff - tunnel - wp ] and [ moshinsky ] are devoted to application to quantum tunneling ( @xmath38 ) and section [ macro ] treats the macroscopic quantum tunneling which includes the nonlinearity ( @xmath39 ) . the schrdinger equation for the wave function @xmath3 of a charged particle in the presence of the scalar potential @xmath40 is the same as in eq.([mn - standeq ] ) . for any long time @xmath7 called as a standard final time , we choose @xmath41 as a target state that we are going to generate . in contrast to the previous works @xcite , the fast - forward wave function here does not include the additional phase factor throughout the fast forwarding time range until @xmath42 in eq.([t - tff ] ) and is given by @xmath43 here @xmath44 is the same as in eq.([ff - t ] ) . we shall try to realize @xmath23 by applying the electro - magnetic field @xmath45 and @xmath46 . let s assume @xmath23 to be the solution of schrdinger equation for a charged particle in the presence of additional vector @xmath47 and scalar @xmath48 potentials , as given by @xmath49 for simplicity , however , we shall hereafter employ the unit of velocity of light @xmath50 and the prescription of a positive unit charge @xmath51 . note : @xmath35 in eq.([ffeq ] ) is introduced independently from a given potential @xmath40 , in contrast to the one in eq.([mn - ffeq ] ) which included @xmath40 . the driving electro - magnetic field is related by , @xmath52 substituting eqs ( [ mn - standeq ] ) , ( [ ffwf ] ) into eq.([ffeq ] ) and taking its real and imaginary parts , we obtain a pair of equations @xmath53\textbf{a}_{ff}\nonumber\\ & + & \hbar(\alpha-1 ) \textrm{im}\left[\frac{\nabla^{2}\tilde{\psi}_{0}}{\tilde{\psi_{0}}}\right]=0\end{aligned}\ ] ] and @xmath54\nonumber\\ & + & \frac{\hbar}{m}\textbf{a}_{ff}\textrm{im}\left[\frac{\nabla\tilde{\psi}_{0}}{\tilde{\psi_{0}}}\right ] -\frac{1}{2m}\textbf{a}_{ff}^{2}+(\alpha-1)v\nonumber\\ & -&(\alpha-1)c_0\|\tilde{\psi_0}\|^2.\end{aligned}\ ] ] now we write @xmath18 as @xmath55 with use of the real amplitude @xmath56 and phase @xmath57 defined by @xmath58 then , using in eqs.([aff ] ) the equalities @xmath59=\frac{\nabla\rho}{\rho},\quad\textrm{im}\left[\frac{\nabla\tilde{\psi}_{0}}{\tilde{\psi_{0}}}\right]=\nabla\eta , \quad\textrm{re}\left[\frac{\nabla^{2}\tilde{\psi}_{0}}{\tilde{\psi_{0}}}\right]=\frac{\nabla^{2}\rho}{\rho}-(\nabla\eta)^{2},\quad \textrm{im}\left[\frac{\nabla^{2}\tilde{\psi}_{0}}{\tilde{\psi_{0}}}\right]=2\frac{\nabla\rho}{\rho}\nabla\eta+\nabla^{2}\eta$ ] , one finds that @xmath60 satisfies eq.([aff ] ) . thanks to eq.([mn - standeq ] ) with the variable @xmath5 being replaced by @xmath44 , @xmath61 $ ] can be re - expressed as @xmath62=\frac{2m}{\hbar } \left(\frac{\partial \eta}{\partial \lambda(t)}+\frac{1}{\hbar}v \right)-\frac{2mc_0}{\hbar^2}\rho^2.\end{aligned}\ ] ] then @xmath35 can be expressed only with use of @xmath57 as @xmath63 with use of the driving vector @xmath64 and scalar @xmath35 potentials in eqs.([aff - sol ] ) and ( [ vff - sol ] ) , we can obtain the fast - forwarded state @xmath23 in eq.([ffwf ] ) which is now free from the additional phase factor @xmath34 in eq.([add - f ] ) used in masuda and nakamura s framework@xcite . logically , @xmath64 and @xmath35 , which prove to be both independent of the amplitude @xmath56 , serve to compensate the additional phase @xmath34 in eq.([add - f ] ) in their framework . the electro - magnetic field introduced in refs . @xcite is designed to guarantee the equality in eq.([tff - t ] ) at @xmath36 , and fails in removing the additional phase in the fast - forward time range @xmath65 . two points should be noted : 1 ) the above driving potentials do not explicitly depend on the nonlinearity coefficient @xmath1 : eqs.([aff - sol ] ) and ( [ vff - sol ] ) work for the nonlinear schrdinger equation as well . 2 ) the magnetic field @xmath66 is vanishing , because a combination of eqs . ( [ eb - av ] ) and ( [ aff - sol ] ) leads to @xmath67 . therefore only the electric field @xmath45 is required to accelerate a given dynamics . with use of eqs . ( [ eb - av ] ) , ( [ aff - sol ] ) and ( [ vff - sol ] ) , @xmath45 is given explicitly by @xmath68 a remarkable issue of the present scheme is the enhancement of the current density @xmath69 . using a generalized momentum which includes a contribution from the vector potential in eq.([aff - sol ] ) , we see : @xmath70\nonumber\\ & = & \frac{\hbar}{m}\alpha(t)\rho^2(\textbf{x},\lambda(t))\nabla\eta(\textbf{x},\lambda(t)),\end{aligned}\ ] ] under the prescription of a positive unit charge . noting the current density in the standard dynamics : @xmath71\nonumber\\ & = & \frac{\hbar}{m}\rho^2(\textbf{x},t)\nabla\eta(\textbf{x},t),\end{aligned}\ ] ] we find @xmath72 thus the standard current density becomes both squeezed and magnified by a time - scaling factor @xmath26 in eq . ( [ nonunif - scale ] ) or eq.([unif - scale ] ) as a result of the exact fast forwarding which enables acceleration of both amplitude and phase of the wave function throughout the time evolution . finally in this section , we shall evaluate the expectation of energy of a particle in fast - forward dynamics and compare it with the corresponding expectation in standard dynamics . we can formally rewrite the schrdinger equations in eq.([mn - standeq ] ) and eq.([ffeq ] ) , respectively as @xmath73 @xmath74 where @xmath75 and @xmath76 are taken as corresponding hamiltonian operators . we can write the expectation of energy in two cases as : @xmath77 and @xmath78 where the integration is over full space ( @xmath79 ) . substituting eqs.([tenergy1s ] ) and ( [ tenergy1f ] ) into eqs.([tenergy2s ] ) and ( [ tenergy2f ] ) respectively , we obtain @xmath80 and @xmath81 @xmath26 comes from @xmath82 in eq.([ff - t ] ) . comparing eq.([tenergy3f ] ) with eq.([tenergy3s ] ) , we have the relation between the expectations of energy between standard and fast - forward dynamics : @xmath83 which is similar to eq.([stand - ff - rel ] ) and will play a vital role in the fast forward of quantum tunneling . now we shall apply the present scheme to several tunneling phenomena in quantum mechanics . as for @xmath26 , we shall choose a non - uniform factor in eq.([nonunif - scale ] ) in the fast - forward time region , except when stated otherwise . confining to the one - dimensional ( 1-d ) motion , we now investigate the time evolution of a localized wave packet when it runs through the delta - function barrier . the initial wave packet centered at @xmath84 and having the width @xmath85 and momentum @xmath86 is expressed as @xmath87 @xmath88 satisfies the normalization condition @xmath89 . therefore , @xmath90 and @xmath91 at @xmath92 . while the wave function in eq.([in - wp ] ) is non - differentiable at @xmath93 , it does not generate a discontinuity in physical quantities like probability amplitude , current density , energy density , etc . time - dependent schrdinger equation with delta function barrier at @xmath94 is given by @xmath95\psi_0(x , t)=v(x)\psi_0(x , t)\end{aligned}\ ] ] with @xmath96 . in order to simplify the notation , we shall use `` natural unit '' ( @xmath97 ) from now on . the time evolution of @xmath98 for @xmath99 follows from @xmath100 the first term on the r.h.s . of eq.([method ] ) describes the time evolution of the free @xmath101 wave packet . here @xmath102 the free - particle propagator given by @xmath103 @xmath104 is moshinsky function@xcite defined in terms of the complementary error function by @xmath105 which is interpreted as the wave function of a monochromatic particle that is confined to the left half- space @xmath106 at @xmath92 . the explicit solution @xmath107 for @xmath99 was given by elberfeld and kleber@xcite as @xmath108\nonumber\\+v_{0}\sqrt\beta[s(x_{0},\lambda^{};t)-s(x_{0},-\lambda;t)\nonumber\\+e^{-\lambda x_{0}}[s(0,-\lambda;t)+s(0,\lambda;t)]],\end{aligned}\ ] ] where @xmath109 and @xmath110 is defined by : @xmath111[m(\|x\|+\xi ;- iv_{0};t)\nonumber\\-m(\|x\|+\xi ;- i\beta;t)].\end{aligned}\ ] ] the tunneling current evaluated just behind the barrier at @xmath94 is : @xmath112_{x=+0 } .\end{aligned}\ ] ] the current is continuous , i.e , @xmath113 owing to a nonabsorbing potential barrier . these are results of the standard tunneling , i.e. , tunneling dynamics on standard time scale . now we analyze the fast forward of tunneling of the wave packet , and find the current density . by extracting the space - time dependent phase @xmath57 of the wave function in eq.([wp - evol ] ) in standard tunneling , one can obtain both vector and scalar potentials in eqs.([aff - sol ] ) and ( [ vff - sol ] ) . here @xmath57 is available only numerically because @xmath98 in eq.([wp - evol ] ) is a linear combination of special functions . under these driving potentials , one can generate the fast - forward state of tunneling of the wave packet through the barrier : @xmath114\nonumber\\ & + & v_{0}\sqrt\beta[s(x_{0},\lambda^{};\lambda(t))-s(x_{0},-\lambda;\lambda(t))\nonumber\\ & + & e^{-\lambda x_{0}}[s(0,-\lambda;\lambda(t))+s(0,\lambda;\lambda(t))]],\nonumber\\\end{aligned}\ ] ] which accelerates both amplitude and phase of eq.([wp - evol ] ) exactly . it should be noted that , without having recourse to @xmath57 , @xmath23 in eq.([psiff ] ) is obtained from @xmath98 by the definition itself in eq.([ffwf ] ) . from eq.([stand - ff - rel ] ) , the current density evaluated at @xmath115 for the fast - forward 1-d tunneling phenomenon is : @xmath116 in our numerical analysis in sections [ ff - tunnel - wp ] and [ moshinsky ] , we shall use typical space and time units like @xmath117 _ the linear dimension of a device _ and @xmath118 _ the phase coherent time _ , besides the natural unit ( @xmath97 ) . then , any length , wave number and time are scaled by @xmath119 , @xmath120 and @xmath121 , respectively . in this scaling , we choose @xmath122 and @xmath123 . we shall show the standard dynamics up to the standard final time @xmath124 and its fast - forward version up to the shortened final time @xmath125 with the mean acceleration factor @xmath126 . ) . in the scaling described below eq.([ff - tun - curr ] ) , @xmath127 and @xmath126 . the barrier height here is @xmath128 : ( a ) standard dynamics given by @xmath129 for @xmath130 ; ( b ) fast - forward dynamics given by @xmath131 for @xmath29 . in figs . [ wp - tunnel - current]-[asymp - curr ] , the same scaling as in this figure is used . ] in fig . [ bifurc - wf ] , we see the exponential wave function partly go through the barrier and is partly reflected back after its collision with the barrier . the dynamics up to @xmath7 on the standard time scale is reproduced in the fast - forward dynamics up to @xmath42 : the phenomena in the latter is just the squeezing ( along time axis ) of the one in the former . the minor discrepancy in the similarity of figures between upper and lower panels in time axis is due to the non - uniform time - scaling factor in eq.([nonunif - scale ] ) , while the wave function amplitude w.r.t . to space axis is exactly reproduced at each time of the fast forward dynamics . figure [ wp - tunnel - current ] shows the tunneling current at the position just behind the barrier ( @xmath115 ) as a function of time @xmath5 . here we choose @xmath132 and @xmath126 . we find the the temporal behavior of the current @xmath133 in the standard tunneling is both squeezed and amplified in that of the current @xmath134 in the fast - forward tunneling . ) for the tunneling wave function starting from the exponential wave packet in eq.([in - wp ] ) . @xmath132 and @xmath126 . the barrier height is @xmath128 : a ) @xmath135 ; b ) @xmath136 . note : scales of the horizontal and vertical axes differ between upper and lower panels . ] as a function of @xmath137 and @xmath5 : a ) top view ; b ) side view . ] let us define the tunneling rate (: tunneling power ) as @xmath138 where @xmath7 here means the final time that the tunneling phenomenon is almost completed . the corresponding rate for the fast - forwarded case is : @xmath139 in fig.[wp - tunnel - current ] we see @xmath140 and @xmath141 in the standard and fast - forward tunnelings , respectively . noting eq.([ff - tun - curr ] ) , we find the numerator of r.h.s of eq.([ff - tunnel - rate ] ) is equal to that of eq.([tunnel - rate ] ) . then one can conclude : @xmath142 thus the fast forwarding lets the standard tunneling rate enhanced by a factor of the mean magnification time scale . this is a great advantage of the fast forward of quantum tunneling . figure [ tunafvf ] shows a 1-d version of the driving electric field @xmath45 in eq.([ele - fld ] ) to generate the exact fast forward of the tunneling dynamics , which is given by @xmath143 the bottom panel of figure [ tunafvf](c ) shows a 3-d plot of @xmath144 on @xmath145 plane , while the top and middle panels are its cross sections in @xmath5 and @xmath137 directions , respectively . in si unit for electric field , our dimensionless @xmath144 corresponds to @xmath146 where @xmath147 and @xmath148 are electron mass , electron charge , velocity of light , frequency of laser light and its wave length , respectively . typical value @xmath149 in ordinates in fig . [ tunafvf ] in case of ir lasers of wave length @xmath150 1@xmath151 m means @xmath152 . the driving electric field shown in fig . [ tunafvf ] can be implemented using for instance a rapidly moving laser beam to create a possibly dynamic time - averaged optical dipole potential@xcite . before closing this section , we should note : the standard tunneling here is autonomous ( i.e. , @xmath153 ) and thereby the total energy of the electron @xmath154 , while the corresponding fast - forward tunneling is non - autonomous with the time - dependent total energy @xmath155 . @xmath155 satisfies @xmath156 as a special case of eq.([tenergy4 ] ) . in the tunneling phenomenon of standard dynamics , we always see the inequality @xmath157(barrier height ) . then we can define @xmath158 and choose the time scaling factor @xmath13 as @xmath159 which guarantees the inequality @xmath160 . in conclusion , so long as eq.([amax ] ) is satisfied , the fast - forwarded dynamics here is also the tunneling phenomenon keeping the particle s energy @xmath161 below the barrier height @xmath162 throughout the time evolution and the time scaling works more effectively for the particle with lower incident energy . the same assertion as above will hold in the following sections . now let s investigate the dynamics of a monochromatic beam of noninteracting particles of mass @xmath163 and energy @xmath164 moving parallel to the @xmath137 axis from the left to the right . until @xmath165 , the beam is assumed being stopped by the shutter at @xmath94 perpendicular to the beam . if at @xmath92 the shutter is opened , the transient particle current is observed at a distance @xmath137 from the shutter . this problem was first solved by moshinsky@xcite , and then received a renewed attention by elberfeld and kleber@xcite , who introduced a delta - function barrier with a finite height at @xmath94 and considered the tunneling through it . the shutter acts as a perfect absorber . then , the wave function that represents a particle of the beam is initially given by @xmath166 with the step function @xmath167 and @xmath168 for @xmath169 and for @xmath170 , respectively . the time - dependent schrdinger equation with a delta function barrier is the same as in eq.([tdsch - barr ] ) . by applying the same method as in eq.([method ] ) , the solution satisfying the initial condition in eq.([in - step ] ) was obtained @xcite as : @xmath171\nonumber\\\times [ m(\|x\| ; -iv_{0};t)-m(\|x\|;k;t)].\end{aligned}\ ] ] the tunneling current just behind the barrier is evaluated using eq.([s - tun - curr ] ) . * * a. fast forward of moshinsky shutter in the presence of delta - function barrier and @xmath5 . the density distribution illustrates the tunneling dynamics of the semi - infinite wave train ( @xmath172 ) penetrating through the delta - barrier with the height @xmath128 located at @xmath115 : ( a ) standard tunneling dynamics until the final time @xmath124 ; ( b ) fast - forward tunneling dynamics until @xmath173 with the average time scaling factor @xmath126 . ] and @xmath126 : a ) @xmath174 with @xmath175 ; b ) @xmath176 with @xmath177 . note : scales of the horizontal and vertical axes differ between upper and lower panels . ] as a function of @xmath137 and @xmath5 : a ) top view ; b ) side view . ] we now analyze the fast forwarded tunneling for moshinsky shutter . one can evaluate the phase @xmath57 of the wave function solution in eq.([st - tun - shutt ] ) , which is used to determine both driving vector and scalar potentials in eqs.([aff - sol ] ) and ( [ vff - sol ] ) . by applying these driving potentials , we obtain the exact fast forwarded state , which is given by replacing @xmath5 by @xmath44 in eq.([ff - t ] ) as : @xmath178\nonumber\\ & \times & [ m(\|x\| ; -iv_{0};\lambda(t))-m(\|x\|;k;\lambda(t ) ) ] . \nonumber\\\end{aligned}\ ] ] concerning the relation between @xmath23 and @xmath57 , one should recall the notion just before and after eq.([psiff ] ) . the fast - forwarded tunneling current is given by eq.([ff - tun - curr ] ) . using the same units as described below eq.([ff - tun - curr ] ) , we choose @xmath179 , @xmath124 and @xmath173 and show in fig.[3d - mosh - tun ] the density profiles of wave functions in both cases of the standard and fast - forward ( or accelerated ) dynamics . we find that the time evolution of @xmath180 is squeezed in @xmath181 by the factor @xmath182 . interference between the incoming and reflected waves leads to an oscillatory profile for @xmath169 . for @xmath170 the outgoing wave shows a rather smoothly - varying density profile . in the case of @xmath183 , we show in fig.[mosh - tun - curr ] the tunneling current just behind the barrier in wider time range with use of @xmath175 and @xmath184 . the temporal behavior of the current @xmath174 in the standard dynamics , which shows a very slow decrease w.r.t . @xmath5 , is squeezed and enhanced in that of @xmath176 in the fast - forward dynamics . as noted in the previous section , eq.([ff - tun - curr ] ) leads to the equality : @xmath185 in case of fig.[mosh - tun - curr ] , a rough estimate for the r.h.s of eq.([integrate ] ) is @xmath186 , which agrees with @xmath187 ( : average fast - forward current ) evaluated for l.h.s . therefore @xmath188 , as seen in fig.[mosh - tun - curr ] . in general , the tunneling current in the standard dynamics greatly decreases when the barrier height @xmath162 becomes much larger than unity . but , a suitable fast - forward mechanism recovers the current for the case of @xmath128 , which will be described by asymptotic argument in the next subsection . the fast forward state in eq.([ff - mosh - wave ] ) can be generated as a solution of the time - dependent schrdinger equation in eq.([ffeq ] ) for the charged particle in the presence of vector @xmath189 and scalar @xmath190 potentials . figure [ mosh - vff - aff ] shows the corresponding electric field @xmath144 evaluated by eq.([ele - fld-1d ] ) as a function of @xmath137 and @xmath5 . * * b. asymptotic approach in case of a very high barrier in the limit of a very high barrier , the tunneling current becomes negligibly small . but by a suitable choice of the time scaling @xmath26 , one can recover the standard magnitude of the tunneling current , which we shall show below . we first rewrite the wave function solution on the r.h.s . of the barrier ( i.e. for @xmath170 ) in eq.([st - tun - shutt ] ) as @xmath191 here @xmath192 where @xmath104 is the moshinsky function defined in eq.([a - moshin ] ) . with use of a new variable @xmath193 we see @xmath194 and @xmath195 we shall now concentrate on the asymptotic region given by @xmath196 with @xmath197 and @xmath198 , which leads to @xmath199 . then we find @xmath200 and @xmath201 noting @xmath202 and @xmath203 , we have @xmath204 and @xmath205 , which results in @xmath206 in a similar way , we find : @xmath207 and @xmath172 . in the fast - forwarded case , the uniform time scaling with @xmath208 is employed during the period between @xmath209 and @xmath42 with @xmath210 . time range depicted is @xmath211 . ( a ) standard current available from eqs . ( [ s - tun - curr ] ) and ( [ st - tun - shutt ] ) ; ( b ) fast - forwarded exact current in eqs . ( [ ff - tun - curr ] ) and ( [ ff - mosh - wave ] ) ; ( c ) fast - forwarded asymptotic current in eq.([asym - uff - cur ] ) . the rapid oscillation in the fast - forward time region comes from the 2nd term in the last expression of eq.([asym - uff - cur ] ) which includes an un - cancelled factor @xmath212 of moshinsky function in eq.([a - moshin ] ) ; d ) 3-d plot of the driving electric field to realize the fast - forwarded exact current in case of b ) . ] using the decomposition in eq.([st - tun - shutt2 ] ) , the standard current just behind the barrier can be expressed as @xmath213 \nonumber\\ & = & \frac{k^{2}}{v^{2}_{0}+k^{2}}\textrm{im}(m_{1}^{\ast}\partial_{x}m_{1})\|_{x=0}\nonumber\\ & + & \frac{kv_{0}}{v^{2}_{0}+k^{2}}\textrm{re}(m_{1}^{\ast}\partial_{x}m_{2})\|_{x=0}\nonumber\\ & + & \frac{kv_{0}}{v^{2}_{0}+k^{2}}\textrm{re}(m_{2}^{\ast}\partial_{x}m_{1})\|_{x=0}\nonumber\\ & + & \frac{v_0^{2}}{v^{2}_{0}+k^{2}}\textrm{im}(m_{2}^{\ast}\partial_{x}m_{2})\|_{x=0}.\nonumber\\\end{aligned}\ ] ] noting the asymptotics in eqs.([m2prod ] ) and ( [ m12prod ] ) , one sees that the first and third terms give dominant contributions of @xmath214 and other terms give minor contributions of @xmath215 . then @xmath216 becomes asymptotically : @xmath217 the tunneling current in the standard dynamics has proved to be of @xmath214 , which is very small . however , the idea of fast forward can recover the current in the case of @xmath218 . in fact , by applying the driving vector and scalar potentials in eqs.([aff - sol ] ) and ( [ vff - sol ] ) , we can realize the exact fast - forward state in the time domain @xmath219 with @xmath42 in eq.([t - tff ] ) , and its corresponding fast - forward current is given by eq.([ff - tun - curr ] ) . therefore , if we shall use a large enough magnification time - scaling factor @xmath13 with its mean value @xmath220 , the current in the case of the small barrier ( @xmath218 ) will be recovered . to make the quantitative argument , let s employ the uniform scaling factor in eq.([unif - scale ] ) . then the fast - forward current is given by @xmath221 noting @xmath222 in the fast - forward time domain , eq.([asym - uff - cur ] ) shows that if we shall choose @xmath223 the tunneling current in the high barrier case ( @xmath224 ) will recover the value in the low barrier case ( @xmath218 ) . figure [ asymp - curr ] shows : the fast forwarding with use of the driving electric field makes a negligible tunneling current for the case of a very high barrier with @xmath196 increased to the value for the case of a standard barrier with @xmath225 . in fact , in case of @xmath226 , @xmath227 in fig.[asymp - curr](a ) , but @xmath228 in figs . [ asymp - curr](b ) , ( c ) in the fast - forward time region @xmath229 . ( vertical direction ) as a function of @xmath137 and @xmath5 : ( a ) standard tunneling dynamics of the soliton where @xmath107 satisfies eq.([sol - standeq ] ) with @xmath230 ; ( b ) fast - forward tunneling dynamics of the soliton where @xmath231 satisfies eq.([ffsol - split ] ) under the non - uniform time - scaling factor @xmath26 with its mean @xmath126 in eq.([nonunif - scale ] ) . @xmath232 are the same as in fig.[split - barrless](a ) ; ( c ) the same as in fig.[split - barrless](b ) except for the uniform time scaling @xmath233 . in figs . [ bec - curr ] and [ sol - drive ] , the same space and time units as in this figure are used . ] the theory of fast - forward can also be applied to the macroscopic quantum mechanics . we shall consider the fast - forwarded tunneling of a solitonic wave packet in 1-d bose - einstein condensates ( bec ) governed by nonlinear schrdinger equation in eq.([mn - standeq ] ) with the barrier at origin , @xmath234 . the standard dynamics for @xmath235 is described by : @xmath236 on the other hand , the governing equation for the fast - forward function @xmath23 is given by ( see the notice just below eq.([ffeq ] ) ) : @xmath237 : ( a ) standard tunneling current ; ( b ) fast - forward tunneling current . note : scales of the horizontal and vertical axes differ between upper and lower panels . ] below , besides the natural unit ( @xmath97 ) we shall employ the same units as used in the previous sections on microscopic quantum dynamics . namely , space and time are scaled by @xmath117 _ the system size _ and @xmath238 _ the dissipation time _ , respectively , and we put the nonlinearity constant @xmath1 ( scaled by @xmath239)@xmath240 . then eqs . ( [ sol - standeq ] ) and ( [ ffsol - split ] ) become dimensionless , which we shall analyze . if there is no barrier , the solution of eqs.([sol - standeq ] ) is a travelling zakharov - shabat s soliton @xcite given by @xmath241e^{i\phi_0+ivx + i(a^2-v^2)t/2}\end{aligned}\ ] ] with @xmath242 and @xmath243 for the amplitude and propagation velocity , respectively . as a function of @xmath137 and @xmath5 : a ) top view ; b ) side view . ] in the presence of the barrier , we numerically solve eq.([sol - standeq ] ) with use of an initial profile : @xmath244 which stands for the soliton with @xmath245 and initial position @xmath246 for center of mass . in fig.[split - barrless](a ) , we show the amplitude @xmath247 of the soliton as a function of @xmath137 and @xmath5 in standard time . the soliton located at @xmath93 moves to the right and after the time @xmath248 that it collides with the barrier at @xmath94 . then it splits into two parts : reflected and transmitted ones which are moving to left and right , respectively . the result accords with the one by holms et al.@xcite . increase of the barrier height @xmath249 diminishes the transmitted part , namely decreases the tunneling rate . according to the idea of the fast forward , the same wave function patterns as seen in the time domain @xmath250 can be realized in the shortened time domain @xmath65 ( see eq.([t - tff ] ) ) with use of @xmath251 and @xmath35 in eq.([aff - sol ] ) and eq.([vff - sol ] ) , respectively . using the non - uniform time - scaling factor @xmath13 with its mean @xmath126 in eq.([nonunif - scale ] ) , we have solved eq.([ffsol - split ] ) . in fig.[split - barrless](b ) , the amplitude @xmath252 of the soliton is shown . at the shortened time @xmath253 the soliton collides with the delta - barrier at @xmath94 and the splitting process is also shortened . in fig.[split - barrless](c ) , we show @xmath252 by solving eq.([ffsol - split ] ) with use of the uniform scaling factor @xmath126 in eq.([unif - scale ] ) . in this case we can see the exact time - squeezed version of soliton dynamics in fig.[split - barrless](a ) . the soliton reaches the barrier at @xmath254 and transmitted and reflected patterns are shortened by the constant time scaling @xmath31 . now we shall compute the tunneling current @xmath174 and @xmath176 at @xmath115 in eqs.([s - tun - curr ] ) and ( [ ff - tun - curr ] ) , respectively . figures [ bec - curr ] shows standard and fast - forward ( with a non - uniform time - scaling factor with its mean @xmath126 ) cases , respectively . standard tunneling current has a peak at @xmath255 , when the soliton almost reaches the barrier . the fast - forward tunneling current is a squeezed and enhanced version of the standard one . since the soliton arrives at the barrier earlier than standard arriving time , the peak of the current is realized at time @xmath256 . figure [ bec - curr ] also shows the enhancement of tunneling rate by @xmath126 as indicated by eq.([ratio ] ) . in fig.[sol - drive ] the driving electric field @xmath144 necessary for the fast - forwarding of the soliton is evaluated by eq.([ele - fld-1d ] ) and is depicted as a function of @xmath137 and @xmath5 . we developed a theory of fast - forwarding of quantum dynamics for charged particles , which exactly accelerates both amplitude and phase of the wave function throughout the fas - forward time range . we elucidated the nature of the driving electro - magnetic field together with vector and scalar potentials to guarantee these exact fast forwarding . the theory is applied to the tunneling phenomena through a tunneling barrier . typical examples described here are : 1 ) the initially - exponential wave packet moving through the delta - function barrier ; 2 ) the opened moshinsky shutter with a delta - function barrier just behind the shutter . standard ( non - accelerated ) dynamics in these examples is known to be exactly solvable . we see the remarkable squeezing and enhancement of the tunneling current density , caused by the fast - forwarding of quantum tunneling . we find : even if the barrier height will be increased , one can generate a recognizable tunneling current by using a large enough time - scaling factor @xmath26 . at the same time , we have shown : so long as @xmath26 is less than @xmath158 with the barrier height @xmath162 and incident energy @xmath257 in the standard tunneling , the corresponding fast - forwarded dynamics is also the tunneling phenomenon keeping the particle s energy @xmath155 below the barrier height @xmath162 throughout the time evolution , and the time scaling works more effectively for the particle with lower incident energy . the analysis is also carried out on the acceleration of macroscopic quantum tunneling with use of the nonlinear schrdinger equation which accommodates a delta - function barrier . finally we should note that this work is inside a broader concept to enhance the visibility of quantum transient phenomena ( post - exponential decay@xcite , quantum backflow@xcite , diffraction in time@xcite , as well as quantum tunneling ) which are predictable by quantum mechanics but hardly detectable because the detection number of particles is very small . the general theory in section [ new appro ] will be an alternative vehicle to optimize the visibility parameters to improve those feeble observations . + _ acknowledgments . _ one of the authors ( k. n. ) is grateful to s. masuda , a. del campo , m. nakayama and y. musakhanov for enlightening discussions and comments in various stages of this work . 99 l. esaki , long journey into tunneling , proc . of the ieee * 62 * , 825 ( 1974 ) . a. m. steinberg , time and history in quantum tunneling , superlattices and microstructures * 23 * , 823 ( 1998 ) . s. masuda and k. nakamura , fast - forward problem in quantum mechanics , phys . rev . a * 78 * , 062108 ( 2008 ) . s. masuda and k. nakamura , fast - forward of adiabatic dynamics in quantum mechanics , proc . a * 466 * , 1135 ( 2010 ) . s. masuda and k. nakamura , acceleration of adiabatic quantum dynamics in electromagnetic fields , phys . a * 84 * , 043434 ( 2011 ) . this terminology means to reproduce a series of events or a history of matters in a shortened time scale , like a rapid projection of movie films on the screen . the fast forward contains as special limits the shortcut to adiabaticity and acceleration of particle motions . m. demirplak and s a. rice , adiabatic population transfer with control fields , j. phys . a * 107 * , 9937 ( 2003 ) . m. demirplak and s. a. rice , assisted adiabatic passage revisited , j. phys . b * 109 * , 6838 ( 2005 ) . m. v. berry , transitionless quantum driving , j. phys . a : math . theor . * 42 * , 365303 ( 2009 ) . h. r. lewis and w. b. riesenfeld , an exact quantum theory of the time - dependent harmonic oscillator and of a charged particle in a time - dependent electromagnetic field , j. math . phys . * 10 * , 1458 ( 1969 ) . x. chen , a. ruschhaupt , s. schmidt , a. del campo , d.gu lery - odelin , and j. g. muga , fast optimal frictionless atom cooling in harmonic traps : shortcut to adiabaticity , phys . lett . * 104 * , 063002 ( 2010 ) . e. torrontegui , s. ibanez , m. martinez - garaot , m. modugno , a. del campo , d. guery - odelin , a. ruschhaupt , xi chen and j. g. muga , shortcuts to adiabaticity , adv . at . 62 * , 117 ( 2013 ) . s. masuda , k. nakamura and a. del campo , high - fidelity rapid ground - state loading of an ultracold gas into an optical lattice , phys . lett . * 113 * , 063003 ( 2014 ) . e. torrontegui , s. martinez - garaot , a. ruschhaupt and j. g. muga , shortcuts to adiabaticity : fast - forward approach , phys . a * 86 * , 013601 ( 2012 ) . k. takahashi , fast - forward scaling in a finite - dimensional hilbert space , phys . rev . a * 89 * , 042113 ( 2014 ) . kiely , j. p. l. mcguinness , j. g. muga and a. ruschhaupt , fast and stable manipulation of a charged particle in a penning trap , j. phys . b * 48 * , 075503 ( 2015 ) . m.moshinsky , diffraction in time , phys.rev.88 , 625 ( 1952 ) ; diffraction in time and the time - energy uncertainty relation , am.j.phys . * 44 * , 1037 ( 1976 ) . v. manko , m. moshinskiy and a. sharma , diffraction in time in terms of wigner distributions and tomographic probabilities , phys . a , * 59 * , 1809 ( 1999 ) . w. elberfeld and m. kleber , time - dependent tunneling through thin barriers : a simple analytical solution , am.j.phys , * 56 * , 154 ( 1988 ) . k. henderson , c. ryu , c. maccormick and m. g. boshier , experimental demonstration of painting arbitrary and dynamic potentials for bose - einstein condensates , new j. phys . * 11 * , 043030 ( 2009 ) . zakharov and a.b . shabat , exact theory of two - dimensional self - focusing and one - dimensional self - modulation of waves in nonlinear media , sov . * 34 * , 62 ( 1972 ) . j. holmer , j. marzuola and m. zworski , soliton splitting by external delta potentials , j. nonlinear sci . , * 17 * , 349 ( 2007 ) . e. torrontegui , j. g. muga , j. martorell , and d. w. l. sprung , enhanced observability of quantum postexponential decay using distant detectors , phys . a * 80 * , 012703 ( 2009 ) . a. del campo , long - time behavior of many - particle quantum decay , phys . a * 84 * , 012113 ( 2011 ) . m. palmero , e. torrontegui , j. g. muga , and m. modugno , detecting quantum backflow by the density of a bose - einstein condensate , phys . a * 87 * , 053618 ( 2013 ) . m. m. ristic g. b. poparic , and d. s. belic , excitation of the a 3@xmath258 state of co by electron impact , phys . a * 83 * , 042714 ( 2011 ) . s. cordero and g. garcia - calderon , diffraction in time for tunneling invisibility in quantum systems , phys . a * 88 * , 052118 ( 2013 ) .	we propose a scheme of the exact fast - forwarding of standard quantum dynamics for a charged particle . the present idea allows the acceleration of both the amplitude and phase of the wave function throughout the fast - forward time range and is distinct from that of masuda - nakamura ( e.g. , proc . r. soc . a * 466 * , 1135 ( 2010 ) ) which enabled acceleration of only the amplitude of the wave function on the way . we shall apply the proposed method to the quantum tunneling phenomena and obtain the electro - magnetic field to ensure the rapid penetration of wave functions through a tunneling barrier . typical examples described here are : 1 ) an exponential wave packet passing through the delta - function barrier ; 2 ) the opened moshinsky shutter with a delta - function barrier just behind the shutter . we elucidate the tunneling current in the vicinity of the barrier and find the remarkable enhancement of the tunneling rate (: tunneling power ) due to the fast - forwarding . in the case of a very high barrier , in particular , we present the asymptotic analysis and exhibit a suitable driving force to recover a recognizable tunneling current . the analysis is also carried out on the exact acceleration of macroscopic quantum tunneling with use of the nonlinear schrdinger equation which accommodates a tunneling barrier .
the model underpinning our calculations successfully gives a @xmath2 behavior in high energy @xmath3 scattering @xcite . quarks and antiquarks are described by complex scalar fields @xmath4 , and gluons as the scalar field @xmath5 coupled through the lagrangian @xmath6 the quarks are massless , while the gluons are assigned a mass @xmath7 in order to avoid infra - red divergences . since in qcd the quartic interaction between gluons leads to terms which are sub - leading in ln @xmath8 , it is not included here . the two labels on each of the fields refer to the direct product of two color groups which is necessary to make the three gluon vertex symmetric under the exchange of two bosonic gluons . to describe non - equilibrium phenomena , we use the green functions of the schwinger - keldysh formalism @xcite . the equations of motion for the wigner transforms of the green functions , the so - called transport and constraint equations , are derived @xcite and read @xmath9 and @xmath10 where @xmath11 is a generic green function , @xmath12 for quarks and @xmath13 for gluons . @xmath14 is the appropriate mass , and @xmath15 is the collision term , @xmath16 @xmath17 and @xmath18 are terms containing retarded and advanced components : @xmath19 in eqs.([e : collision ] ) to ( [ e : pim ] ) , @xmath20 is a generic self - energy ( @xmath21 for quarks and @xmath22 for gluons ) and the operator @xmath23 is given by @xmath24 for the green functions , we use the quasiparticle approximation : @xmath25 @xmath26 is obtained from the expression for @xmath11 by replacing @xmath27 with @xmath28 and vice versa . @xmath29 is the quark ( @xmath30 ) or the gluon ( @xmath31 ) distribution function , @xmath32 is an abbreviation for @xmath33 . then an evolution equation for the parton distribution function is obtained by integrating eq.([e : transport ] ) over an interval @xmath34 that contains the energy @xmath35 . to lowest order in an expansion that sets @xmath36 , the integral for the collision term is performed in the next section . for the collision integral , one has @xmath37 _ i.e. _ the off - diagonal quasiparticle self - energies are required to be calculated on shell . in the following , only the results for the quarks are presented , the generalization to the gluons is then obvious . the self - energies are organized now according to their number of loops . the hartree self - energies ( self - energy with a quark or a gluon loop ) vanish due to their color factors as it is the case for real qcd . the fock self - energy gives a contribution to the loss term of the collision integral in eq.([e : collint ] ) given by @xcite @xmath38 here , @xmath39 means the scattering amplitude of the process @xmath1 of order @xmath40 which is pictured in fig.1 a ) . processes as @xmath41 , @xmath42 or @xmath43 are kinematically forbidden and therefore do not occur in this collision integral . the gain term can be obtained by exchanging @xmath27 with @xmath28 in eq.([e : lossint ] ) . the two loop self - energy graphs are pictured in fig.2 . according to their topology , they are called rainbow ( r ) , ladder ( l ) , cloud ( c ) , exchange ( e ) and quark - loop ( ql ) graphs . let us first consider the contribution of the seven diagrams ra ) , la ) , ca ) , cb ) , ea ) , eb ) and qla ) to the collision integral . the first five of these diagrams lead to the ( lowest order ) scattering amplitudes of the processes @xmath44 and @xmath45 , while the latter two lead to the ( lowest order ) scattering amplitudes of the processes @xmath46 and @xmath47 @xcite . thus , to obtain _ all _ possible @xmath48 processes it is necessary to consider all types of diagrams . but an analysis of the color factors shows that the quark - loop diagrams are subleading as it is the case in real qcd . therefore , in an additional expansion in the inverse number of colors , @xmath49 , one can neglect the processes @xmath46 and @xmath47 . that means that gluon production is favored . the three diagrams rd ) , ld ) and qld ) vanish due to the momentum structure of their propagators @xcite . to see the purpose of the remaining ten diagrams @xcite , i.e. rb ) , rc ) , lb ) , lc ) , cc ) , cd ) , ec ) , ed ) , qlb ) and qlc ) , we investigate first the process @xmath1 which is shown up to order @xmath50 in fig.1 . @xmath51 was given in lowest order by the fock term . to construct it up to order @xmath52 , one has to take the hermitian conjugate of the scattering amplitude of fig.1a ) and multiply it with each of the scattering amplitudes of fig.1b)-g ) . it is also necessary to take the hermitian conjugate of each of these products , too . using a symbolical notation , we write these products as @xmath53 . then a detailed analysis shows that each of these products , with the exception of @xmath54 and @xmath55 , is provided by one of the above mentioned ten self - energy diagrams . the diagram g ) of fig.1 does not enter into the collision integral , as it is a renormalization diagram for the incoming quark , for which the momentum @xmath56 is fixed externally . so far , we have constructed the collision integral out of self - energy diagrams with graphs up to two loops . let us now return to the transport equation ( [ e : transport ] ) . if we integrate the left hand side over the interval @xmath34 and neglect the contributions of @xmath57 and @xmath58 on the right hand side , then we get for the transport equation up to order @xmath52 @xmath59.\end{aligned}\ ] ] the generalization to higher orders is then obvious . including , e.g. , the three loop self - energy into the collision term leads to cross sections of all kinematically allowed processes with two ( three ) partons in the initial state and three ( two ) partons in the final state . furthermore , they give all additional corrections of order @xmath60 to the process @xmath1 , e.g. the product of scattering amplitudes of the diagrams b)-f ) in fig.2 , and also corrections to the scattering processes of two partons into two partons . one may speculate that pinch singularities could enter into the collision integral up to the two loop level . in fact this is not so , and has been explicitly demonstrated @xcite . 9999 j.r . forshaw and d.a . ross , _ quantum chromodynamics and the pomeron _ ( cambridge university press , cambridge , 1997 ) . j. schwinger , j. math . phys . * 2 * , 407 ( 1961 ) ; l.v . keldysh , jetp * 20 * , 1018 ( 1965 ) . s. mrwczyski and u. heinz , _ ann . ( n.y . ) * 229 * , 1 ( 1994 ) . klevansky , a. ogura , and j. hfner , _ ann . * 261 * , 261 ( 1997 ) . isert and s.p . klevansky , hep - ph/9905289 . isert and s.p . klevansky , hep - ph/9912203 .	a model of scalar quarks and scalar gluons is used to derive transport equations for quarks and gluons . in particular , the collision integral is studied . the self - energy diagrams are organized according to the number of loops . a generalized boltzmann equation is obtained , which involves at the level up to two loops all possible two @xmath0 two parton scattering processes and corrections to the process @xmath1 . 0.35 cm 0.45 cm -1 cm
in what follows , we denote by @xmath2 an integer partition of @xmath3 and @xmath4 the number of parts of @xmath5 . if @xmath6 is the number of parts of @xmath5 that are equal to @xmath7 ( by convention @xmath8 ) , then we write @xmath5 as @xmath9 and let @xmath10 . also , if @xmath11 , let @xmath12 and @xmath13 be the partitions of @xmath14 @xmath15 and @xmath16 respectively . let @xmath17=\{1,\ldots , m\}$ ] and @xmath18)$ ] be the symmetric group on @xmath19 elements , and let @xmath20 be the conjugacy class in @xmath21 of permutations @xmath22 with cycle type , @xmath23 , @xmath24 . we look at * perfect pairings * of the set @xmath25\cup [ \widehat{n}]=\{1,\ldots n , \widehat{1},\ldots , \widehat{n}\}$ ] of * non - hat * or * hat * numbers which we view as fixed point free involutions in @xmath26\cup [ \widehat{n}])$ ] . note that the disjoint cycles of the product @xmath27 have repeated lengths , @xmath28 . also , given @xmath29 , let @xmath30 be the pairing @xmath31 . let @xmath32 be the hyperoctahedral group which we view as the centralizer in @xmath33 of the involution @xmath34 . then @xmath35 , and it is well known that the double cosets of @xmath32 in @xmath33 are indexed by partitions @xmath0 of @xmath3 and consist of permutations @xmath29 such that the cycle type of @xmath36 is @xmath37 ( * ? ? ? if we denote such a double coset by @xmath38 and pick from it a fixed element @xmath39 , then let @xmath40 be the number of ordered factorizations @xmath41 of @xmath39 where @xmath42 and @xmath43 . , @xmath44 we provide a combinatorial formula for @xmath40 when @xmath1 ( say @xmath45 ) by interpreting these factorizations as * locally orientable unicellular ( partitioned ) hypermaps. we give this formula using symmetric functions : for @xmath11 , we use the monomial symmetric function @xmath46 which is the sum of all different monomials obtained by permuting the variables of @xmath47 , and the power symmetric function @xmath48 , defined multiplicatively as @xmath49 where @xmath50 . [ thm : main ] let @xmath51 be the number of ordered factorizations @xmath41 of @xmath52 where @xmath42 and @xmath53 . if @xmath54 and @xmath55 are the power and monomial symmetric functions then @xmath56 where , @xmath57 is the set of @xmath58-tuples @xmath59 of tridimensional arrays of non negative integers indexed by @xmath60 with @xmath61 , @xmath62 , and @xmath63 r & = & \sum_{i , j , k}(j+k)(p_{ijk } + p'_{ijk } ) , & r&=&\sum_{i , j , k}(j+k)(q_{ijk } + q'_{ijk } ) , \\[2 mm ] q ' & = & \sum_{i , j , k}j(p_{ijk } + p'_{ijk } ) , & p ' & = & \sum_{i , j , k}j(q_{ijk } + q'_{ijk}).\\ \end{array}\ ] ] where @xmath64 and if @xmath65 , @xmath66 , \end{gathered}\ ] ] otherwise if @xmath67 , then @xmath68 . there are special cases of the formula detailed in appendix [ app ] . by abuse of notation , let the double coset @xmath38 also represent the sum of its elements in the group algebra @xmath70 . then @xmath38 form a basis of a commutative subalgebra of @xmath70 ( the _ hecke algebra _ of the _ gelfand pair _ @xmath71 ) and one can check that @xmath72 . thus , @xmath73 are the connection coefficients * of this double coset algebra . we use @xmath74 to denote the * zonal polynomial * indexed by @xmath5 which can be viewed as an analogue of the schur function @xmath75 ( for more information on these polynomials see ( * ? ? ? vii ) ) . in terms of @xmath76 : @xmath77 where @xmath78 , @xmath79 are the irreducible characters of the symmetric group ; and @xmath80 where @xmath81 . the following formula for the connection coefficients in terms of @xmath82 was given in ( * ? ? ? * lemma 3.3 ) : @xmath83 where @xmath84 ( * ? ? ? * lemma 2.1 ) , and @xmath85 is the product of all the _ hook - lengths _ of the partition @xmath13 . let @xmath86 so that @xmath87 is the lhs of . equation immediately implies that @xmath88 . moreover , if for an @xmath89 matrix @xmath90 we say that @xmath91 , then in ( * ? ? ? 3.5 ) it was shown that @xmath92 is also the expectation of @xmath93 over @xmath94 , where @xmath94 are @xmath89 matrices whose entries are independent standard normal random _ real _ variables and @xmath95 are arbitrary but fixed real symmetric matrices ( @xmath96 denotes transpose ) . note that the quantities @xmath97 determine the eigenvalues of @xmath98 . from a topological point of view , * locally orientable hypermaps * of n edges can be defined as a connected bipartite graph with black and white vertices . we view each edge as being a _ fat _ edge composed of two edge - sides both connecting the two incident vertices ( \(4 ) at ( 0,0 ) [ wv ] ; ( 5 ) at ( 1,0 ) [ bv ] ; ( 0 ) at ( 0,-0.1 ) ; ( 1 ) at ( 1,-0.1 ) ; ( 2 ) at ( 0,0.1 ) ; ( 3 ) at ( 1,0.1 ) ; ( 0 ) ( 1 ) ; ( 2 ) ( 3 ) ; ) . we call these edge - sides * half edges. this graph is embedded on a locally orientable surface such that if we cut the graph from the surface the remaining part consists of connected components called faces or cells , each homomorphic to an open disk . note that two half edges can be parallel or cross in the middle . a crossing or a twist of two half edges ( \(4 ) at ( 0,0 ) [ wv ] ; ( 5 ) at ( 1,0 ) [ bv ] ; ( 0 ) at ( 0,-0.1 ) ; ( 1 ) at ( 1,-0.1 ) ; ( 2 ) at ( 0,0.1 ) ; ( 3 ) at ( 1,0.1 ) ; ( 0 ) ( 3 ) ; ( 2 ) ( 1 ) ; ) indicates a change of orientation in the map and that the map is embedded on a non orientable surface ( projective plan , klein bottle , moebius strip , ... ) . in @xcite , it was shown that such hypermaps admit a natural formal description involving triples of perfect pairings @xmath99 on the set of half edges where : @xmath100 associates half edges of the same edge , @xmath101 associates immediately successive ( i.e. with no other half edges in between ) half edges moving around the white vertices , and @xmath102 associates immediately successive half edges moving around the black vertices . formally we label each half edge with an element in @xmath25\cup [ \widehat{n}]=\{1,\ldots , n,\widehat{1},\ldots,\widehat{n}\}$ ] and define @xmath99 as perfect pairings on this set . combining the three pairings gives the fundamental characteristics of the hypermap since : the cycles of @xmath103 give the succession of edges around the white vertices . if @xmath104 then the degree distribution of the white vertices is @xmath105 ( counting only once each pair of half edges belonging to the same edge ) , the cycles of @xmath106 give the succession of edges around the black vertices . if @xmath107 then the degree distribution of the black vertices is @xmath108 ( counting only once each pair of half edges belonging to the same edge ) , the cycles of @xmath109 encode the faces of the map . if @xmath110 then the degree distribution of the faces is @xmath0 in what follows , we consider the number @xmath111 of unicellular * , or one - face , locally orientable hypermaps with face distribution @xmath112 , white vertex distribution @xmath105 , and black vertex distribution @xmath108 . let @xmath101 be the pairing @xmath113 and @xmath114 . we have @xmath115 . then by the above description , one can see that @xmath116\cup [ \widehat{n } ] ) \mid f_3\circ f_1 \in \mathcal{c}_{\lambda\lambda } , f_3\circ f_2 \in \mathcal{c}_{\mu\mu } \ } \mid.\ ] ] moreover , the following relation between @xmath117 and @xmath118 holds : ( * ? ? ? * cor 2.3 ) @xmath119 thus we can encode the connection coefficients as numbers of locally orientable hypermaps and @xmath120 figure [ fig : example ] depicts a locally orientable unicellular hypermap in @xmath111 with @xmath121 and @xmath122 ( do not pay attention to the geometric shapes at this stage ) where where @xmath123 next , we consider locally orientable hypermaps where we partition the white vertices ( black resp . ) . in terms of the pairings , this means we `` color '' the cycles of @xmath103 ( @xmath106 resp . ) allowing repeated colors but imposing that the two cycles corresponding to each white ( black resp . ) vertex have the same color . definition [ def : phm ] below makes this more precise . we will use @xmath124 to denote a set partition of @xmath25\cup[\widehat{n}]$ ] with blocks @xmath125 . we will only work with set partitions whose blocks have even size . we say that these set partitions @xmath124 have * half - type * , @xmath126-@xmath127 , @xmath5 if the cardinalities of the blocks are given by a the integer partition @xmath13 of @xmath14 . [ def : phm ] we consider the set of triples , @xmath128 where @xmath100 is a pairing on @xmath25\cup [ \widehat{n}]$ ] , @xmath129 and @xmath130 are set partitions on @xmath25\cup [ \widehat{n}]$ ] with blocks of even size and of respective half - types , @xmath126-@xmath127 , @xmath105 and @xmath108 , with the constraint that @xmath131 @xmath132 is stable by @xmath133 and @xmath100 . any such triple is called a * locally orientable partitioned hypermap * of type @xmath134 . furthermore , let @xmath135 . the analogous notion of partitioned or colored map is common in the study of orientable maps , see @xcite,@xcite . recently , bernardi @xcite extended the approach in @xcite to find a correspondence between locally orientable partitioned maps and orientable maps that have a certain distinguished spanning subgraph ( called a _ planar - rooted map _ ) . this correspondence does not fully apply to count locally orientable hypermaps ( locally orientable bipartite maps ( * ? ? ? * sect . the number of hat numbers in a block is equal to the number of non hat numbers if non hat number @xmath7 belongs to block @xmath136 then @xmath137 belongs as well to @xmath136 . same argument applies to blocks of @xmath130 with @xmath138 . as an example , the locally orientable hypermap on figure [ fig : example ] is partitioned by the blocks : @xmath139 ( blocks are depicted by the geometric shape , all the vertices belonging to a block have the same shape ) . let @xmath140 be the number of unordered partitions @xmath141 of the set @xmath142 $ ] such that @xmath143 for @xmath144 . then for the monomial and power symmetric functions , @xmath145 and @xmath146 , we have : @xmath147 ( * ? ? ? * prop.7.7.1 ) . we use this to obtain a relation between @xmath148 and @xmath149 [ prop1 ] for partitions @xmath150 we have @xmath151 , where @xmath105 and @xmath108 are refinements of @xmath0 and @xmath152 respectively . let @xmath153 . if @xmath104 and @xmath107 then by definition of the set partitions we have that @xmath105 and @xmath108 are refinements of @xmath126-@xmath154 and @xmath126-@xmath155 respectively . thus , we can classify the elements of @xmath156 by the cycle types of @xmath103 and @xmath106 . , @xmath157 where @xmath158 if @xmath159 then it is easy to see that @xmath160 . by the change of basis equation between @xmath146 and @xmath145 , this immediately relates the generating series @xmath161 and the generating series for @xmath162 in monomial symmetric functions , @xmath163 let @xmath164 be the set , with cardinality @xmath165 , of locally orientable partitioned hypermaps with @xmath3 edges where @xmath166 are tridimensional arrays such that for @xmath167 : @xmath168 ( resp . @xmath169 ) is the number of blocks of @xmath129 of half size @xmath7 such that : either @xmath170 belongs to the block or its maximum * non - hat * number is paired to a * hat * number by @xmath100 ( resp . blocks of @xmath129 not containing @xmath170 such that the maximum * non - hat * number of the block is paired to a * non - hat * number by @xmath100 ) , the block contains @xmath171 pairs @xmath172 where @xmath173 is the maximum * hat * number of a block of @xmath130 such that @xmath174 is also a * hat * number , and , the block contains as a whole @xmath175 pairs @xmath176 where both @xmath177 and @xmath178 are * non - hat * numbers . @xmath179 ( resp . @xmath180 ) is the number of blocks of @xmath130 of half size @xmath7 such that : the maximum * hat * number of the block is paired to a * non - hat * ( resp . * hat * ) number by @xmath100 , the block contains @xmath171 pairs @xmath172 where @xmath173 is the maximum * non - hat * number of a block of @xmath129 non containing @xmath170 and such that @xmath174 is also a * non - hat * number , and the block contains as a whole @xmath175 pairs @xmath176 where both @xmath177 and @xmath178 are * hat * numbers . as a direct consequence , for @xmath165 to be non zero @xmath181 has to check the conditions of theorem [ thm : main ] . furthermore : @xmath182 the partitioned hypermap on figure [ fig : example ] belongs to @xmath164 for @xmath183 , @xmath184 , @xmath185 , @xmath186 where @xmath187 , the elementary tridimensional array with @xmath188 and @xmath189 elsewhere . one can notice that a hypermap is orientable if and only if @xmath174 is a hat number when @xmath173 is a non hat number ( we go through each edge of the map in both directions and there are no changes of direction during the map traversal ) . as a result , a hypermap in @xmath164 is orientable if and only if : @xmath190 and @xmath191 if @xmath192 and/or @xmath193 . in this particular case , we have the following values for @xmath194 and @xmath195 : @xmath196 @xmath197 if we denote @xmath198 the number of such orientable maps , by theorem [ thm : main ] , equation , lemma [ lem : lp ] and relation we recover the following result from ( * ? ? ? 1 ) : @xmath199 note that @xmath200 are as well connection coefficients of the symmetric group , that count the number of ordered factorizations @xmath201 of the long cycle @xmath202 in @xmath203 where @xmath204 and @xmath205 . we show that locally orientable partitioned hypermaps admit a nice bijective interpretation in terms of some recursive forests defined as follows : [ def : forest ] in what follow we consider the set @xmath206 of permuted rooted forests composed of : a bicolored identified ordered * seed tree * with a white root vertex , other bicolored ordered trees , called * non - seed trees * with either a white or a black root vertex , each vertex of the forest has three kinds of ordered descendants : * tree - edges * or edges ( connecting a white and a black vertex ) , * thorns * ( halves of tree - edges connected to only one vertex ) and * loops * connecting a vertex to itself . the two _ extremities _ of the loop are part of the ordered set of descendants of the incident vertex and therefore the loop can be intersected by thorns , edges and other loops as well . the forests in @xmath206 also have the following properties : the root vertices of the non - seed trees have at least one descending loop with one extremity being the rightmost descendant of the considered vertex , the total number of thorns ( resp . loops ) connected to the white vertices is equal to the number of thorns ( resp . loops ) connected to the black vertices , there is a bijection between thorns connected to white vertices and the thorns connected to black vertices . the bijection between thorns will be encoded by assigning the same symbolic _ labels @xmath207 to thorns associated by this bijection , there is a mapping that associates to each loop incident to a white ( resp . black ) vertex , a black ( resp . white ) vertex @xmath208 such that the number of white ( resp . black ) loops associated to a fixed black ( resp . white ) vertex @xmath208 is equal to its number of incident loops . we will use symbolic _ greek _ labels @xmath209 to associate loops with vertices except for the maximal loop of a root vertex @xmath210 of the non - seed trees . in this case , we draw an arrow ( ( 0,0 ) to ( 0.8,0 ) ; ) outgoing from the root vertex @xmath210 and incoming to the vertex associated with the loop . arrows are non ordered , and the ascendant / descendant structure defined by the edges of the forest and the arrows defined above is a tree structure rooted in the root of the seed tree . finally the degree @xmath181 of the forest is given in the following way : @xmath168 ( resp @xmath169 ) counts the number of non root white vertices ( including the root of the seed tree ) ( resp . white root vertices excluding the root of the seed tree ) of degree @xmath7 , with @xmath171 incoming arrows and a total of @xmath175 loops . @xmath179 ( resp @xmath180 ) counts the number of non root black vertices ( resp . black root vertices ) of degree @xmath7 , with @xmath171 incoming arrows and total @xmath175 loops . as an example , figure [ fig : exrecons ] depicts two permuted forests . the one on the left is of degree @xmath59 for @xmath211 , @xmath184 , @xmath212 , and @xmath186 while the one on the right is of degree @xmath213 for @xmath214 , @xmath215 , @xmath216 , and @xmath217 . using the lagrange theorem for implicit functions one can show : @xmath218 * reformulation of the main theorem * + the next sections are dedicated to the proof of the following stronger result that will imply theorem : there is a bijection @xmath219 ) and so @xmath220 . we proceed with the description of the bijective mapping @xmath221 between locally orientable partitioned hypermaps and permuted forests of degree @xmath181 . let @xmath222 be a partitioned hypermap in @xmath164 . the first step is to define a set of white and black vertices with labeled ordered half edges such that : each white vertex is associated to a block of @xmath129 and each black vertex is associated to a block of @xmath130 , the number of half edges connected to a vertex is half the cardinality of the associated block , and the half edges connected to the white ( resp . black ) vertices are labeled with the non hat ( resp . hat ) integers in the associated blocks so that moving clockwise around the vertices the integers are sorted in increasing order . then we define an ascendant / descendant structure on the vertices . a black vertex @xmath223 is the descendant of a white one @xmath22 if the maximum half edge label of @xmath223 belongs to the block of @xmath129 associated to @xmath22 . similar rules apply to define the ascendant of each white vertex except the one containing the half edge with label @xmath170 . if black vertex @xmath224 ( resp . white vertex @xmath225 ) is a descendant of white vertex @xmath226 ( resp . black vertex @xmath227 ) and has maximum half edge label @xmath19 such that @xmath228 is the label of a half edge of @xmath226 ( resp . @xmath227 ) , i.e. @xmath229 is a non hat ( resp . hat ) number , then we connect these two half edges to form an edge . otherwise @xmath228 is a hat ( resp . non hat ) number and we draw an arrow ( ( 0,0 ) to ( 0.8,0 ) ; ) between the two vertices . note that descending edges are ordered but arrows are not . + [ lem : tree]the above construction defines a tree structure rooted in the white vertex with half edge @xmath170 . let black vertices @xmath230 and @xmath231 associated to blocks @xmath232 and @xmath233 be respectively a descendant and the ascendant of white vertex @xmath22 associated to @xmath234 . we denote by @xmath235 , @xmath236 and @xmath237 their respective maximum half edge labels ( hat , hat , and non hat ) and assume @xmath238 . as @xmath239 is stable by @xmath101 , then @xmath240 is a non hat number in @xmath234 not equal to @xmath170 . it follows @xmath241 . then as @xmath233 is stable by @xmath102 , it contains @xmath242 and @xmath243 . putting everything together yields @xmath244 . in a similar fashion , assume white vertices @xmath245 and @xmath246 are descendant and ascendant of black vertex @xmath223 . if we note @xmath247 , @xmath248 and @xmath249 their maximum half edge labels ( non hat , non hat , and hat ) with @xmath250 , one can show that @xmath251 . finally , as @xmath252 , the black vertex with maximum half edge @xmath253 is descendant of the white vertex containing the half edge label @xmath170 . using the hypermap of figure [ fig : example ] we get the set of vertices and ascendant / descendant structure as described on figure [ fig : verttree ] . next , we proceed by linking half edges connected to the same vertex if their labels are paired by @xmath100 to form loops . if @xmath7 and @xmath254 are the labels of a loop connected to a white ( resp . black ) vertex such that neither @xmath7 nor @xmath254 are maximum labels ( except if the vertex is the root ) , we assign the same label to the loop and the black ( resp . white ) vertex associated to the block of @xmath130 ( resp . @xmath129 ) also containing @xmath7 and @xmath254 . as we assign at most one label from @xmath209 to a given vertex , several loops may share the same label . [ lem : loops ] the number of loops connected to the vertex labeled @xmath255 is equal to its number of incoming arrows plus the number of loops labeled @xmath255 incident to other vertices in the forest . the result is a direct consequence of the fact that in each block the number of hat / hat pairs is equal to the number of non hat / non hat pairs as a final step , we define a permutation between the remaining half edges ( thorns ) connected to the white vertices and the one connected to the black vertices . if two remaining thorns are paired by @xmath100 then these two thorns are given the same label from @xmath256 . all the original integer labels are then removed . we denote by @xmath257 the resulting forest . we continue with the hypermap from figure [ fig : example ] and perform the final steps of the construction as described on figure [ fig : loopsperm].(note that the geometric shapes are here for reference only , they do not play any role in the final object @xmath257 ) . as a direct consequence of definition [ def : forest ] , @xmath257 belongs to @xmath258 . we show that mapping @xmath259 is indeed one - to - one . we start with a forest @xmath257 in @xmath206 and show that there is at most one triple @xmath222 in @xmath164 such that @xmath260 . the first part is to notice that within the construction in @xmath221 , the original integer label of the leftmost descendant ( thorn , half loop or edge ) of the root vertex of the seed tree is necessarily @xmath170 ( this root is the vertex containing @xmath170 and the labels are sorted in increasing order from left to right ) . + assume we have recovered the positions of integer labels @xmath261 , for some @xmath262 , non hat number . then four cases can occur : @xmath7 is the integer label of a thorn of latin label @xmath263 . in this case , @xmath254 is necessarily the integer label of the thorn connected to a black vertex also labeled with @xmath263 . but as the blocks of @xmath130 are stable by both @xmath100 and @xmath102 then @xmath264 is the integer label of one of the descendants of the black vertex with thorn @xmath263 . as these labels are sorted in increasing order , necessarily , @xmath265 labels the leftmost descendant with no recovered integer label . @xmath7 is the integer label of a half loop of greek label @xmath255 . then , in a similar fashion as above @xmath265 is necessarily the leftmost unrecovered integer label of the black vertex with symbolic label @xmath255 . @xmath7 is the integer label of a half loop with no symbolic label ( i.e , either @xmath7 or @xmath254 is the maximum label of the considered white vertex ) . then , @xmath265 is necessarily the leftmost unrecovered integer label of the black vertex at the other extremity of the arrow outgoing from the white vertex containing integer label @xmath7 . @xmath7 is the integer label of an edge . @xmath265 is necessarily the leftmost unrecovered integer label of the black vertex at the other extremity of this edge . finally , using similar four cases for the black vertex containing the descendant with integer label @xmath265 and the fact that blocks of @xmath129 are stable by @xmath100 and @xmath101 , the thorn , half loop or edge with integer label @xmath266 is uniquely determined as well . we continue with the procedure described above up until we fully recover all the original labels @xmath25\cup [ \hat{n}]$ ] . according to the construction of @xmath257 , the knowledge of all the integer labels uniquely determines the blocks of @xmath129 and @xmath130 . the pairing @xmath100 is as well uniquely determined by the loops , edges and thorns with same latin labels . assume the permuted forest @xmath257 is the one on the right hand side of figure [ fig : exrecons ] . the steps of the reconstruction are summarized in figure [ fig : recons ] . we get that the unique triple @xmath222 such that @xmath260 is : @xmath267 to prove that @xmath221 is surjective , we have to show that the reconstruction procedure of the previous section always finishes with a valid output . + assume the procedure comes to an end at step @xmath7 before all the integer labels are recovered ( where @xmath7 is for example non hat , the hat case having a similar proof ) . it means that we have already recovered all the labels of vertex @xmath268 identified as the one containing @xmath265 ( or @xmath269 ) prior to this step . this is impossible by construction provided @xmath268 is not the root vertex of the seed tree since the number of times a vertex is identified for the next step is equal to its number of thorns , plus its number of edges , plus twice the number of loops that have the same greek label as @xmath268 , plus twice the incoming arrows . using lemma [ lem : loops ] , we have that the sum of the two latter numbers is twice the number of loops of @xmath268 . as a consequence , the total number of times the recovering process goes through @xmath268 is exactly ( and thus never more than ) the degree of @xmath268 . if @xmath270 is the root vertex of the seed tree , the situation is slightly different due to the fact that we recover label @xmath170 before we start the procedure . to ensure that the procedure does not terminate prior to its end , we need to show that the @xmath271-th time the procedure goes through the root vertex is right after all the labels of the forest have been recovered . again , this is always true because : the last element of a vertex to be recovered is the label of the maximum element of the associated block . consequently , all the elements of a vertex are recovered only when all the elements of the descending vertices ( through both arrows and edges ) are recovered . lemma [ lem : tree ] states that the ascendant / descendant structure involving both edges and arrows is a tree rooted in @xmath270 . as a result , the procedure goes the @xmath270-th time through @xmath270 only when all the elements of all the other vertices are recovered . in the orientable case , one can use schur symmetric functions and the irreducible characters of the symmetric group to prove the identity in equation ( see @xcite ) . this requires : ( i ) ( @xmath272 ) the murnaghan - nakayama rule ( * ? ? ? 7.17.3 ) , ( ii ) ( @xmath273 ) finding the number of semistandard young tableaux of hook shape @xmath274 and type @xmath108 which is just @xmath275 , and ( iii ) using inclusion exclusion . one could try to replicate this on @xmath276 and obtain an algebraic proof of theorem [ thm : main ] . we show the outcome after step ( i) using ( * ? ? ? steps ( ii) and ( iii) appear quite less tractable . @xmath277 . for the formula of theorem [ thm : main ] to be complete , we need to observe the following cases : when @xmath65 and @xmath282 , which can only occur if @xmath283 ( we assume @xmath284 , the elementary tridimensional array with @xmath188 and @xmath189 elsewhere ) the whole summand @xmath285 on the rhs of equation reduces to : @xmath286\\ & \times\frac{(n-2r)!p'!q'!\left ( r - p ' \right)!\left ( r- q ' \right)!}{2^{2r - p'-q'}}\binom{t}{u , v , u+v}\prod_{i , j , l}{\binom{i-1}{j , k , j+k}}^{q_{ijk}}{\binom{i-1}{j , k , j+k-1}}^{(p'+q')_{ijk}}\end{aligned}\ ] ]	this paper is devoted to the evaluation of the generating series of the connection coefficients of the double cosets of the hyperoctahedral group . hanlon , stanley , stembridge ( 1992 ) showed that this series , indexed by a partition @xmath0 , gives the spectral distribution of some random real matrices that are of interest in random matrix theory . we provide an explicit evaluation of this series when @xmath1 in terms of monomial symmetric functions . our development relies on an interpretation of the connection coefficients in terms of locally orientable hypermaps and a new bijective construction between locally orientable partitioned hypermaps and some permuted forests .
the power of discovering high redshift galaxies via the broad - band drop - out technique , ie . lyman break galaxies ( lbgs ) , is now well established ( steidel et al . 1999 ; 2000 ) . using this technique , over a thousand galaxies have now been detected at @xmath9 . detailed studies show that these galaxies have stellar masses between 10@xmath10 and 10@xmath11 @xmath5 , and a ( comoving ) volume density at @xmath0 of @xmath12 mpc@xmath13 ( giavalisco 2002 ) . a number of issues are still being investigated for lbgs . one important issue is deriving the total star formation rate . besides model parameters such as the imf and star formation history , the dust correction in the uv remains under investigation . steidel et al . ( 1999 ) originally estimated a typical uv dust attenuation factor of about 5 , based on optical spectroscopy . adelberger & steidel ( 2000 ) applied a similar method to a larger sample of lbgs , as well as observations at other wavebands ( radio , submm ) , and conclude : `` ... the mean extinction at 1600 for lbgs , a factor of 6 in our best estimate , could lie between a factor 5 and a factor of 9 . '' more recently , reddy & steidel ( 2004 ) have derived the average dust correction factors for uv - selected galaxies at @xmath14 based on deep x - ray , radio , and optical spectroscopic studies of galaxies in the goods north field . they find that , for galaxies with total star formation rates @xmath15 @xmath5 year@xmath6 , the dust attenuation factor is between 4.4 and 5.1 . a second issue for lbgs is the agn fraction . shapley et al . ( 2003 ) found that @xmath16 of lbgs at z@xmath173 have optical emission line spectra that are consistent with agn . the cosmic evolution survey ( cosmos ) , covering 2 , is a comprehensive study of the evolution of galaxies , agn , and dark matter as a function of their cosmic environment . the cosmos field has state - of - the - art multiwavelength observations , ranging from the radio through the x - ray ( scoville et al . 2007 ; capak et al . a major part of this study is to identify the largest samples of lbgs to date . in section 2 we describe the lbg samples from the cosmos survey based on u , b , and v - band drop - out searches . in this paper we consider the radio properties of these large samples of lbgs , similar to the study of @xmath18 lyman-@xmath19 emitting galaxies by carilli et al . observations of the cosmos field have been done at 1.5@xmath20 resolution ( fwhm ) at 1.4 ghz with the very large array ( schinnerer et al . we employ the new vla image that has added integration time in the central @xmath21 ( schinnerer in prep . ) , giving an rms at the field center of @xmath22 @xmath23jy beam@xmath6 . in section 3.1 we search for radio counterparts to individual lbgs , while in section 3.2 we perform a stacking analysis to derive the statistical properties of the samples . in section 4 we discuss the implications these observations have on the questions of dust extinction and the agn fraction in lbgs . we adopt a standard concordance cosmology . we have identified large samples of u , b , and v - band drop - out galaxies using the cosmos photometric data ( capak et al . 2007 ) . for an initial , robust analysis of the radio properties of high z lbgs , we have adopted a series of conservative criteria to identify lbgs at the different redshifts . the u , b , and v - band drop selection criteria are give in table 1 , where u = cfht u , i = subaru i , and r = subaru r ( lee et al . in prep . ) . it is well documents that lbg samples can include contaminants ( see steidel et al . 2000 , capak et al . 2004 , bouwens et al . 2007 ) . in the u and b - band drop - out selection , the contamination comes from a and m stars along with @xmath24 star forming galaxies , while the v - band drop - out selection is contaminated with m and t stars along with @xmath25 obscured galaxies . since these contaminating objects can be sources of radio emission , we take several steps to produce a clean sample of high redshift objects . we begin by removing all acs point sources with a star- or agn - like spectral energy distribution using the star / galaxy / agn classifier described in robin et al . in addition we remove all stars identified by a bzk diagram ( daddi et al . 2007 ) at @xmath26 , this removes very red stars which may not be identified by the robin et al . ( 2007 ) classifier . we then use the bpz photometric redshifts from mobasher & mazzei ( 2000 ) to remove objects with a best fit photo - z at @xmath27 , @xmath28 , and @xmath29 for the u , b , and v - band drop - outs , respectively . finally we remove objects with a @xmath30% probability of being at @xmath31 . the resulting sample is largely clear of contaminants . redshifts were obtained with deimos on keck - ii using the 830l / mm grating and 3.5 - 4h of integration time for a sample of color selected objects with @xmath32 and @xmath33 or @xmath34 ( capak et al . in prep ) . the spectroscopic sample matches 2 u - band drop - outs , 18 b - band drop - outs , and 7 v - band drop - outs in our cleaned lbg sample . of these objects , only 1 v - band drop - out object , which is a faint m star , is not in the expected redshift range , yielding a contamination fraction of @xmath35% . however , the u - band drop - outs sample was not specifically targeted , so the constraint on the contamination fraction for these objects is weak . of the remaining 26 objects at @xmath36 , one b - band drop - out shows an agn signature , implying a residual agn fraction of @xmath35% . although our sample is clean , it is not complete . in general , the phot - z cut creates a bias towards galaxies which have good photometry , and closely match the expected uv - optical spectral templates used for the photometric redshifts . objects with unusual spectra , such as agn / galaxy composites , or objects with reddening that does not conform to the calzetti et al . ( 1997 ) or milky way laws , will likely be excluded . furthermore , we have explicitly excluded optically bright broad line agn , and the probability requirements systematically bias us against faint ( @xmath37 ) objects . the flux bias occurs because faint objects have broad probability distributions due to their poor photometric measurements . we are also potentially biased against heavily obscured objects which will have fainter uv fluxes , and hence poor phot - z fits . an effort to use intermediate band photometry ( ilbert et al . in prep ) and a systematic deep spectroscopic follow - up ( lilly et al . 2007 , capak et al . in prep ) is underway to fully characterize this sample and understand the biases . table 2 summarizes the current samples . column 2 indicates the redshift range implied for a given drop - out band . column 3 shows the number of sources in the sample . even with these conservative selection criteria , we identify thousands of high redshift lbgs due to the very wide cosmos field . in this initial study , we feel it important to work with such a conservative sample of sources , in order to make robust conclusions concerning the radio properties of lbgs . for reference , the radio luminosity density at a rest frame frequency of 1.4 ghz , for a source with @xmath38jy , assuming a spectral index of @xmath39 ( condon 1992 ) , at @xmath0 is : @xmath40 ergs s@xmath6 hz@xmath6 . for comparison , the nearby radio agn source m87 has @xmath41 erg s@xmath6 hz@xmath6 , while the luminous starburst galaxy arp 220 has @xmath42 erg s@xmath6 hz@xmath6 . in m87 , the radio emission is powered by a relativistic jet emanating from the supermassive black hole with @xmath43 @xmath5 ( gebhardt et al . 2000 ) , while in arp 220 the radio emission is driven by star formation with a total star formation rate @xmath44 @xmath5year@xmath6 ( carilli & yun 2000 ; yun & carilli 2001 ) . even at the sensitivity of the cosmos survey , we are limited to detecting either radio agn sources , or extreme starburst galaxies , ie . galaxies with star formation rates @xmath45 a few thousand @xmath5 year@xmath6 . we have searched for radio emission from each of the lbgs in the cosmos field . at each position we determine the surface brightness , and the rms noise in the region . the median rms in the lbg fields was @xmath46jy beam@xmath6 for all three samples , implying a median @xmath47 detection limit of 60@xmath23jy beam@xmath6 . our radio images have a synthesized beam fwhm = 1.5 , hence we have searched for radio sources brighter than 4@xmath48 , within 1@xmath49 radius of the lbg optical position . column 4 in table 2 lists the number of 4@xmath48 radio counterparts in each sample . roughly , for each drop - out band , we detect radio counterparts for @xmath50 of the uv sources . we need to correct for the number of detections we expect by chance in such a survey , either due to image noise , or faint sources unrelated to the lbg . we calculate this correction factor by performing searches with identical criteria as for the lbg positions , but at random positions selected to be @xmath51 away from the coordinates of the lbg samples . this should give us a random sampling of the cosmos field , with noise statistics that match the lbg regions . we find 19 4@xmath48 random detections in the u - band drop - out fields , 4 for the b - band drop - outs , and 3 for the v - band drop - outs . these correspond to 0.29% , 0.28% , and 0.49% , respectively . we apply these corrections in column 5 of table 2 . overall , we detect radio counterparts to @xmath50 of the uv sources , with the expected number of detections at random being @xmath52 . these results imply that only a small fraction , @xmath53 , of the lbgs , are radio sources brighter than @xmath54 , where , on average @xmath55jy . we investigate the statistical radio properties of the lbgs through a stacking analysis . we adopt the median stacking method used by white et al . ( 2007 ) in their stacking analysis based on the first radio survey . white et al . perform detailed calculations that show that a median stacking analysis is superior to a mean stacking , since it is robust to small numbers of bright sources , and it does not require any maximum allowed flux density cut - off prior to stacking . moreover , they show that : ... in a limit where almost all of the values in the sample are small compared to the noise , it is straight - forward to interpret the median stack measurements as representative of the mean for the population. the cosmos sample obeys this criterion . figure 1 shows the images derived by a median stacking of all the sources in each drop - out band . these images are derived by extracting sub - images centered around each lbg with a size of @xmath56 . we then stack these sub - images , aligned at the center ( on - source ) position , and derive the median surface brightness at each pixel position . the off - source pixels serve as test positions for a random stacking analysis . we find that the mean values for the off positions equal zero , to within the noise , as expected for random positions . we also derive the statistical noise of the analysis from these off - positions . column 6 in table 2 lists the peak surface brightnesses derived at the stacked lbg position , plus the rms off - source noise , for each sample . we obtain a clear detection of radio emission in the u - band drop - out sample , with a value of @xmath57jy . the other bands show marginal detections , although at @xmath58 value , and hence we consider these upper limits . our results are consistent with the upper limit found by ivison et al . ( 2007 ) , who performed a 1.4 ghz stacking analysis of 107 @xmath59 lbgs in the extended groth strip ( the aegis survey ) , for which they obtained a median value of @xmath60jy . the stacked source is unresolved , with a size @xmath2 . the implied physical size is @xmath3kpc , hence the typical source must be galaxy size , or smaller . at the sensitivity limits of current deep radio surveys , such as cosmos , detection of individual sources at @xmath9 is limited to either extreme starbursts ( star formation rates @xmath61 @xmath5 year@xmath6 ) , or radio jet sources with luminosities comparable to m87 . after correcting for random detections , we obtain a detection rate of @xmath53 . shapley et al . ( 2003 ) find an agn fraction in lbg samples of @xmath16 , based on optical spectroscopy . if all of our detected sources are radio agn , then the implied radio loud fraction is @xmath62 . interestingly , a value of 10% is the canonical value for radio loud agn based on studies of nearby galaxies ( eg . ivezic et al . 2002 ; petric et al . 2007 ) . we emphasize caution in interpreting our results on the radio loud fraction of agn at high redshift , for a number reasons . first , in a recent comparison of the first and sdss surveys , jiang et al . ( 2007 ) find that the radio loud fraction likely depends on both optical luminosity and redshift . this study is not directly comparable to our cosmos study , since they were limited to much more luminous sources at high redshift , by about two orders of magnitude in the radio . second , the above analysis does not consider the possibility that some of the 0.3% of the radio detections are extreme , highly obscured starbursts , comparable to the submm galaxies . indeed , in a companion paper , we present the discovery of a submm galaxy in our radio - selected lbg cosmos sample ( capak et al . the radio - detected lbg has a spectroscopic redshift of @xmath63 , and a radio flux density of @xmath64jy . thermal emission from warm dust is detected from this galaxy at 250ghz using the mambo bolometer camera , with a flux density of @xmath65mjy . this source is the first submm galaxy yet identified ( spectroscopically ) at @xmath66 ( capak et al . 2008 ) . the unprecedented size of the cosmos lbg sample has allowed us to reach sub-@xmath23jy sensitivity levels in the stacking analysis of the lbg samples . we obtain a clear statistical detection of @xmath67jy for the u - band ( @xmath0 ) drop - outs . the implied rest frame 1.4 ghz luminosity density is @xmath68 erg s@xmath6 hz@xmath6 , assuming a spectral index of @xmath39 ( condon 1992 ) . assuming the radio emission is driven by star formation , we can derive a total star formation rate ( 0.1 to 100 @xmath5 ) , from the rest frame 1.4 ghz luminosity density . there are a number of recent calibrations of this relationship for nearby galaxies using , eg . the iras and nvss surveys . these studies adopt the star formation rates derived from the far - ir emission based on the relations in kennicutt ( 1998 ) , assuming a salpeter imf , and then calibrate the radio conversion factor using the tight radio to far - ir correlation for star forming galaxies . we adopt the conversion factor from the study of yun , reddy , and condon ( 2001 ) : where @xmath70 is in erg s@xmath6 hz@xmath6 . from this , we calculate a mean star formation rate of @xmath4 @xmath5 year@xmath6 . note that the conversion factor above is within 10% of that derived by bell ( 2003 ) . the mean observed uv luminosity at a rest frame wavelength of 1600 of the u - band drop - out sample is : @xmath71 erg s@xmath6 hz@xmath6 ( capak et al . 2008 in prep ) . using equation ( 1 ) in kennicutt ( 1998 ) for the relationship between uv luminosity and total star formation rate , we derive a total star formation rate of 17 @xmath5 year@xmath6 , uncorrected for dust attenuation . the ratio of radio derived star formation rate to uv derived star formation rate is @xmath72 . the simplest conclusion is that the uv emission is attenuated by dust by a factor of 1.8 . however , this factor is considerably smaller than the standard factor @xmath7 adopted for lbgs ( see section 1 ) . we consider some possible reasons for this difference . one possibility is that the dust attenuation factor for the @xmath0 cosmos lbg sample is indeed smaller than for other lbg samples . most of the studies of dust attenuation of lbg galaxies have been at @xmath73 ( see section 1 ) , and hence it is possible that at higher redshift the dust attenuation factor decreases . an argument against this decrease is the recent x - ray and optical study of a sample of lbgs at @xmath0 by nandra et al . ( 2002 ) , who also derive a dust attenuation factor of about 5 . we also re - emphasize that the cosmos lbg sample analyzed herein de - selected sources with poor phot - z fits , which can occur for very heavily obscured objects ( section 2 ) . on the other hand , wilkins , trentham , & hopkins ( 2008 ) present a detailed comparison of the build up of stellar mass in galaxies versus the cosmic star formation rate density . they conclude that there is a discrepancy between the rate of stellar mass creation , and the star formation rates derived from the uv luminosities , at high redshift , if one assumes the standard factor @xmath7 dust attenuation . the discrepancy is in the sense that the uv derived star formation rates are too high . they find that the peak in the discrepancy occurs at @xmath74 , and they suggest that the uv - derived star formation rates at this redshift may be over - estimated by a factor @xmath75 . they also point out that ... this large deviation at high redshift offers an explanation for why the integrated star formation history implies a local stellar mass density in excess of that measured. a second possibility is that the conversion factor of radio luminosity to star formation rate is higher in the @xmath0 lbg cosmos sample than has been derived for low redshift galaxies . radio studies of 24@xmath23 m selected galaxies by appleton et al . ( 2004 ) imply that the radio conversion factor is constant out to @xmath76 , ie . that the radio - fir correlation is constant out to this redshift . reddy & steidel ( 2004 ) extend this conclusion out to @xmath14 in their extensive study of uv selected galaxies . most recently , ibar et al . ( 2008 ) conclude that the radio conversion factor remains constant out to @xmath0 , using a 24@xmath23 m source sample with redshifts from the sxdf . however , given the need for individual source detections in the radio , their @xmath0 radio sources have star formation rates about two orders of magnitude larger than the stacking results presented herein , and hence a direct comparison is problematic . one physical reason why we might expect the radio conversion factor to diverge at the highest redshifts is increased relativistic electron cooling due to inverse compton scattering off the cosmic microwave background ( cmb ) . the ratio of relativistic electron energy losses due to synchrotron radiation , to energy losses due to inverse compton radiation , equals the ratio of the energy density in the magnetic field to that in the photon field . the energy density in the cmb increases as : @xmath77 ergs cm@xmath13 . figure 2 shows a comparison of @xmath79 with the typical energy densities in the magnetic fields in different regions in galaxies . we show the range of fields considered typical for spiral arms ( @xmath17 few @xmath23 g ) , and for starburst galaxy nuclei ( of order 100@xmath23 g ; beck et al . 1994 ; see review by beck 2005 ) . the important point is that ic losses off the cmb will dominate synchrotron losses in a typical ism at @xmath80 , and dominate in starburst nuclei at @xmath81 . we note that inverse compton losses will not affect the thermal electrons responsible for free - free emission from star forming galaxies . such free - free emission may dominate the total radio emission from galaxies at rest frequencies between roughly 40 ghz and 100 ghz ( condon 1992 ) , and hence become an important factor in radio continuum studies of very high redshift galaxies . a depressed radio luminosity for a given star formation rate could also arise if the @xmath0 lbgs are systematically smaller galaxies . the standard model that produces the radio - fir correlation ( condon 1992 ) requires a cosmic ray processing box size @xmath82 kpc . it has been observed that dwarf galaxies at low redshift depart from the radio - fir correlation by about a factor of 2 , in the sense of being radio under - luminous . the hypothesis is that the cosmic rays diffuse out of the galaxy on timescales shorter than required to maintain the standard radio - fir correlation ( yun et al . giavalisco ( 2002 ) describes high @xmath83 lbgs as having typical half - light radii of 4 to 7 kpc , significantly larger than typical dwarf galaxies . however , he points out that : `` .. frequently the galaxies have disturbed or fragmented morphologies , with one bright core , or multiple knots embedded in diffuse nebulosity , reminiscent of merger events . '' in this paper , we report the first robust statistical ( median ) detection of sub-@xmath23jy radio emission from lbg galaxies at @xmath59 . this detection was made possible by the very wide area , and depth , of the cosmos field . while our physical interpretation of the result remains inconclusive , there are a number of future studies we are pursuing to address the interesting questions raised concerning the uv attenuation factor for lbgs , and the the radio luminosity to star formation rate conversion factor , at @xmath8 . the most important study involves obtaining optical spectra of a large sample of lbgs from the cosmos sample . spectra will elucidate the nature of the sources , and allow for a study of the dust correction factor as a function of galaxy type ( agn , starburst , elliptical ... ) . also , the selection criteria for this lbg sample were very conservative ( section 2 ) . we will further refine ( and increase ) our lbg samples using the cosmos multiband photometry , and perform statistical analyses of the radio properties as a function of eg . stellar mass , or @xmath84 . we are also exploring the ir properties of these samples with spitzer . such ir studies have particular relevance in the light of the results of ivison et al . ( 2007 ) , who find that for the seven ir luminous @xmath59 lbgs in their aegis sample ( @xmath85jy ) , the median 1.4 ghz flux density is 44@xmath23jy , comparable to that expected for sub - mm galaxies . lastly , unambiguous identification of radio agn using eg . optical spectra , will provide input into faint radio source population models that are being generated in the context of planning for future , large area radio telescopes ( wilman et al . the results presented herein indicate that future deep ( sub-@xmath23jy ) , wide field surveys with the expanded very large array , will detect routinely the radio emission from individual lbgs out to @xmath0 . the hst cosmos treasury program was supported through nasa grant hst - go-09822 . we gratefully acknowledge the contributions of the entire cosmos collaboration consisting of more than 70 scientists . more information on the cosmos survey is available + at * http://www.astro.caltech.edu / cosmos*. cc thanks the max - planck - gesellschaft and the humboldt - stiftung for support through the max - planck - forschungspreis . we thank the referee for helpful comments . sources & corrected # of sources & median flux density + & & & & & @xmath23jy + u & 2.53.5 & 6457 & 43 ( 0.67% ) & 24 ( 0.38% ) & @xmath95 + b & 3.54.5 & 1447 & 8 ( 0.55% ) & 5 ( 0.34% ) & @xmath96 + v & @xmath45 4.5 & 614 & 5 ( 0.81% ) & 2 ( 0.33% ) & @xmath97 +	we present an analysis of the radio properties of large samples of lyman break galaxies ( lbgs ) at @xmath0 , 4 , and 5 from the cosmos field . the median stacking analysis yields a statistical detection of the @xmath0 lbgs ( u - band drop - outs ) , with a 1.4 ghz flux density of @xmath1jy . the stacked emission is unresolved , with a size @xmath2 , or a physical size @xmath3kpc . the total star formation rate implied by this radio luminosity is @xmath4 @xmath5 year@xmath6 , based on the radio - fir correlation in low redshift star forming galaxies . the star formation rate derived from a similar analysis of the uv luminosities is 17 @xmath5 year@xmath6 , without any correction for uv dust attenuation . the simplest conclusion is that the dust attenuation factor is 1.8 at uv wavelengths . however , this factor is considerably smaller than the standard attenuation factor @xmath7 , normally assumed for lbgs . we discuss potential reasons for this discrepancy , including the possibility that the dust attenuation factor at @xmath8 is smaller than at lower redshifts . conversely , the radio luminosity for a given star formation rate may be systematically lower at very high redshift . two possible causes for a suppressed radio luminosity are : ( i ) increased inverse compton cooling of the relativistic electron population due to scattering off the increasing cmb at high redshift , or ( ii ) cosmic ray diffusion from systematically smaller galaxies . the radio detections of individual sources are consistent with a radio - loud agn fraction of 0.3% . one source is identified as a very dusty , extreme starburst galaxy ( a submm galaxy ) .
in the late fifties it was pointed out by k. gottfried that after making a few approximations , the cross section for photoinduced two - nucleon ( @xmath7 ) knockout @xmath8 can be related to quantities that are sensitive to the relative and center - of - mass motion of nucleon pairs in the nuclear system @xcite . in this pioneering work , it was predicted that @xmath7 knockout cross sections are proportional to the so - called pair function @xmath9 , which is related to the probability of finding a nucleon pair with c.m . momentum @xmath10 in a finite nuclear system . momentum @xmath10 can be determined in terms of the momenta of the two escaping nucleons ( @xmath11 and @xmath12 ) and the momentum transferred by the photon ( @xmath13 ) : @xmath14 it should be noted that as soon as one goes beyond the approximations that are at the basis of the gottfried approach , the differential cross sections can no longer be formally factorized in terms of the pair function and factors that depend on the relative motion of the active pair . this is for example the case when considering more realistic wave functions than plane waves for the outgoing nucleons . nevertheless , the dominant role of the pair function @xmath9 in the a@xmath15 and a@xmath16 processes has been experimentally confirmed @xcite . this observation raises confidence in that the @xmath17 reaction mechanism can be sufficiently kept under control to gain also empirical information about the relative motion of nucleon pairs in the medium . it is worth noting that also the measured @xmath3c@xmath18 @xcite and @xmath2o@xmath18 @xcite cross sections have recently been shown to scale in terms of the pair function @xmath9 . pioneering two - nucleon knockout studies with moderate energy resolution were performed at tokyo and bonn @xcite . in recent years , the pip - tof collaboration has collected a vast amount of 2n knockout data at the tagged photon facility of the 800 mev mainz electron accelerator . in comparison with earlier work , these measurements are characterized by an improved energy resolution and cover a wide photon energy region ( @xmath19 mev ) . measurements with polarized photon beams on @xmath20he and @xmath3c are in the process of being analyzed . other experiments with polarized photon beams were performed for the target nuclei @xmath21he @xcite and @xmath2o @xcite at the legs facility located at the brookhaven national laboratory . in these experiments , the photon energy ranged from 220 up to 305 mev . the @xmath22he(@xmath23 asymmetries turned out to be remarkably similar to those obtained from d(@xmath24,@xmath25)@xmath26 , provided that one selects that part in the proton - neutron phase space @xmath27 for which the residual proton @xmath25 can almost be guaranteed to be at rest . with this restriction , only two nucleons are involved in the reaction process and the residual proton is a spectator . recently , it was pointed out by sandorfi and leidemann @xcite that the mere assumption of proton pairs predominantly moving in a relative @xmath28 state has major implications for the linear - polarization @xmath22he(@xmath29 asymmetries . along the same lines , wilhelm , niskanen and arenhvel @xcite pointed out that relatively simple forms for the @xmath16 cross sections and polarization observables can be obtained when assuming photoabsorption on @xmath28 diprotons and dominance of any two of the three multipoles @xmath30 in the transition matrix elements . turning to finite nuclei , a number of models to deal with two - nucleon knockout have been developed over the last couple of years . the models fall into different categories . the one developed by the valencia group @xcite aims at describing all the different aspects of the photon - nucleus coupling and pion final - state interactions in a unified diagrammatic approach . this includes all the ( multi-)pion and ( multi-)nucleon production channels . a very large amount of contributing mechanisms are included in the description of the pion and nucleon photoproduction channels . all this at the expense of the nuclear structure aspects of the reaction which are dealt with in a nuclear - matter approach . the local - density approximation ( lda ) is applied to obtain results for finite nuclei . the valencia calculations have been carefully compared to the @xmath31 data collected in recent years @xcite . generally a fair description of the missing energy spectra is obtained , an exception made for the @xmath16 channel for which the calculations tend to overestimate the data @xcite . a major conclusion from these comparisons is that for the photoabsorption mechanism to be two - body in nature one has to select these processes for which the @xmath32 fragment is created at low missing energies @xcite . at higher missing energies the two - nucleon knockout strength is predicted to be essentially due to initial @xmath33 production and subsequent pion final - state interactions . this implies that the higher missing energy range is excluded when studying the properties of dinucleons in the medium . indeed , in these studies it is essential to guarantee that only two nucleons participate in the reaction process . the models developed in pavia @xcite and gent @xcite are less ambitious as far as the number of included reaction mechanisms is concerned . they concentrating on the two - body photoabsorption mechanisms and do not include the pion production channels . in the light of the previous discussion , these models are therefore restricted to the low missing energy part of the two - nucleon knockout spectra . in comparison with the valencia approach , the gent and pavia models put more emphasis on a proper description of the nuclear structure aspects of the 2n emission reaction process . in both models , the shell - model framework is the starting point to account for the nuclear structure aspects . in the light of 2n emission processes being a possible probe to study ground - state correlations , the mean - field wave functions are corrected for ( central ) jastrow correlation effects . here we report on calculations that aim at investigating the dynamics of dinucleons in the medium with the aid of photoinduced two - nucleon knockout . more in particular , we concentrate on the additional degrees of freedom created by using ( linearly ) polarized photon beams and recoil nucleon polarimetry . the organization of this paper is as follows . in sect . [ sec : ang ] the definitions and conventions for the @xmath31 cross sections and polarization observables are given . [ sec : model ] is devoted to a description of the different assumptions which are at the basis of the model calculations presented here . a factorized ( subsect . [ sec : fac ] ) and an unfactorized ( subsect . [ sec : unfac ] ) model are sketched . rather than elaborating on the technical details , we make an attempt to describe the basic ideas and assumptions behind the models . section [ sec : results ] contains the results of the numerical calculations . we concentrate on proton - proton and proton - neutron knockout from the target nuclei @xmath2o and @xmath3c in several kinematical conditions and consider photon energies from 100 up to 500 mev . finally , sect . [ sec : conclus ] contains the conclusions . using standard techniques the four - fold differential cross section for the photoinduced two - nucleon knockout process @xmath34 in the lab frame ( @xmath35 ) reads @xmath36 where @xmath37 denotes the appropriate averaging over the inital states and sum over the final states . the variables @xmath38 ( @xmath39 ) and @xmath40 ( @xmath41 ) refer to the three - momentum and total energy of the escaping nucleons . further , @xmath42 denotes the photon circular polarization . the @xmath43 is the transition matrix element for the reaction under study and is defined as @xmath44 where @xmath45 and @xmath46 are the distorted waves of the outgoing nucleon @xmath47 and @xmath48 respectively . the @xmath49 determines the wave function of the residual @xmath32 system . its excitation energy @xmath50 is expressed relative to the ground - state energy of the bound @xmath32 nucleus . in all forthcoming derivations we assume that the residual nucleus is created in a specific state characterized by an excitation energy @xmath50 and angular momentum @xmath51 . this does not imply , however , that our considerations are restricted to the discrete part of the @xmath32 spectrum . indeed , when the situation occurs that the @xmath32 nucleus is created in the continuous part of its spectrum a summation over the momenta @xmath51 has to be carried out . in all further considerations , the z axis is chosen along the direction of the incoming photon momentum . the @xmath52 plane is defined by @xmath13 and @xmath11 . the y axis is then along the direction of the vector @xmath53 . the photon asymmetry @xmath54 is given by @xmath55 where @xmath56 is the differential cross section for photons linearly polarized parallel ( perpendicular ) to the @xmath52 reaction plane . the polarization for the nucleon @xmath47 is defined according to @xmath57 where @xmath58 is the differential cross section for nucleon @xmath47 polarized along ( against ) the @xmath59-axis . remark that the choice of the reference system is such that the nucleon on which recoiling nucleon polarimetry is performed is labelled as @xmath47 and that the spin is measured along the direction normal to the reaction plane containing @xmath60 and @xmath11 . the unpolarized differential cross section and asymmetry can be written in terms of structure functions and read @xcite @xmath61 with @xmath62 \label{wt } \\ w_{tt } ( q _ { \gamma},k_a , k_b,\theta _ a,\theta _ b,\phi _ a,\phi _ b ) & = & 2 re \biggl [ \sum_{m_{s_{a}},m_{s_{b}},m_r } \left(m_f^{fi}(\lambda=-1)\right)^ * \left(m_f^{fi}(\lambda=+1)\right ) \biggr ] \;. \label{wtt}\end{aligned}\ ] ] at this point it is worth remarking that the unpolarized @xmath63 cross section depends on both the @xmath64 and @xmath65 structure function . apart from these , the @xmath66 differential cross section is further determined by the @xmath67 and @xmath68 structure function , reflecting the longitudinal degree of freedom in electron scattering . in any case , with the aid of polarized real photon beams one can gain control over the transverse channels which have been shown to contribute substantially to triple coincidence reactions of the @xmath69 type @xcite . the outgoing nucleon polarization can be written in the following form @xmath70 \ ; .\end{aligned}\ ] ] direct two - nucleon knockout following photoabsorption is dominated by two - body current operators @xmath71 } ( { \vec q})$ ] . in our model calculations , an effective lagrangian approach is adopted to construct the two - body currents that determine the coupling of the photon field to the nuclear system . in selecting the current operators or , equivalently , the contributing reaction mechanisms , we have been led by the observation that the dominant role of ( virtual ) pion production in photoinduced two - nucleon knockout has been established in several photoabsorption calculations @xcite . the most elementary process that allows gaining control over the role of the pion in photoinduced experiments is pion photoproduction on the nucleon @xmath72 . it is very well known that for photonenergies below 0.5 gev the cross sections and polarization observables for these reactions are fairly well understood in an effective langrangian approach retaining a selected class of diagrams @xcite . the diagrams which are commonly included in photopion production studies are shown in fig . [ fig : pion ] . we now transfer all elementary pion production processes of fig . [ fig : pion ] to the nuclear medium and assume that the created pion is reabsorbed . as a result of such a process a second nucleon can be brought on the mass shell and one obtains the diagrams of fig . [ fig : twon ] all feeding the two - nucleon knockout channel . these diagrams represent the processes which are believed to be the major contributors to the @xmath31 reaction process at low missing energies . in figure [ fig : twon ] , diagrams ( a ) , ( b ) , ( d ) and ( e ) correspond with genuine two - body photoabsorption , respectively related to meson - exchange ( ( a ) and ( b ) ) and isobaric currents ( ( d ) and ( e ) ) . diagrams ( c ) and ( f ) are characterized by one - body photoabsorption with subsequent two - nucleon knockout and refer to final - state interactions and ground - state correlations respectively . the two - body operators related to fig . [ fig : twon](a ) and ( b ) are the standard pion - exchange current operators that can be derived from the one - pion exchange part of the nucleon - nucleon interaction through minimal substitution @xcite . we use the standard @xmath73 pseudo - vector coupling with @xmath74 = 0.079 and regularize the @xmath73 vertices with monopole form factors using a cut - off mass of @xmath75 . the one - nucleon photoabsorption mechanism of figure [ fig : twon](f ) would not contribute in a shell - model ( or independent particle ) picture of the target nucleus and refers to the situation in which one - body photoabsorption on a correlated pair forces two nucleons to escape . in principle , this process could be treated on the same footing as the other contributing diagrams . this would imply , however , that solely ground - state correlations induced by the one - pion exchange part of the nucleon - nucleon interaction are included . this is at odds with the widely spread belief that heavier meson exchange is the major source of ground - state correlations . generally , short - range effects are difficult to deal with in a diagrammatic approach . nevertheless , various techniques have been developed to cope with the complexity of the ground - state of an interacting fermion system . here , we resort to a semi - phenomenological approach to relate the results of these theoretical ground - state investigations with the cross sections and polarization observables for @xmath31 reactions . it consists of correcting the mean - field slater determinant @xmath76 with a jastrow - like correlation function : @xmath77 such an expansion mocks up two- , three- , a - body correlations . in the actual calculations only the central correlations of the two - body type are retained . as a matter of fact , such an approximation amounts to correcting the product of the single - particle bound - state functions of the active nucleon - pair @xmath78 with a central correlation function : @xmath79 in an attempt to find out about the sensitivity of the @xmath31 angular cross sections and polarization observables to the ground - state correlations we have used several types of correlation functions in the calculations . the results of this paper will confirm the importance of the @xmath80 resonance in photoinduced two - nucleon knockout for photonenergies below 0.5 gev . as we consider photon energies in the resonance region special care must be taken in constructing the isobaric current operators related to intermediate @xmath81 creation of the type @xmath82 . a detailed account of the isobaric currents in the context of two - nucleon knockout reactions was given in ref . @xcite . as our model calculations start from the assumption that the pion is the major mediator between several nucleons , the @xmath83 interaction which enters in the construction of the isobaric currents is the pion exchange interaction . we now consider a simplified model for @xmath17 . two approximations are necessary for the angular ( @xmath84 ) cross section to factorize . first , a plane wave description for the escaping particles should be adopted . in addition , a severe assumption regarding the relative motion of the pairs has to be made in the sense that the decoupling of the c.m . and relative motion of the pair can formally only be achieved after the photon is assumed to couple to dinucleons residing in relative @xmath6 waves . this simplification is known as the `` quasi - deuteron '' approximation . this terminology might be a bit confusing in the sense that is used for both proton - proton and proton - neutron knockout . furthermore , the range of the relative @xmath6 wave is assumed to be small in comparison with all other mechanisms that play a role in the reaction process . from the technical point of view , this allows replacing the relative @xmath6 wave by a delta function ( known as the `` zero - range approximation '' ) , thus enormously simplifying the theoretical calculations @xcite . the `` zero - range approximation '' somehow implies that one simplifies the fermi motion in the relative wave function of the pair . the factorized scheme can be equally well applied to @xmath15 and @xmath16 reactions . the fact that solely the terms related to the ground - state correlations and a few pieces from the isobaric current contribute to the direct proton - proton knockout mechanism makes the factorized model to be particularly attractive as rather simple analytical expressions for the asymmetry and cross sections can be derived . the procedure for arriving at factorized expressions for the structure functions was outlined in ref . @xcite . for the sake of completeness we rewrite the derived expressions for the @xmath64 and @xmath65 structure functions at the real photon point @xmath85 \nonumber \\ & + & \frac { 256 } { 81 } \left ( \frac { f_{\gamma n \delta } f_{\pi n \delta } f_{\pi nn } } { m_{\pi } ^3 } \right)^2 g_{\delta } ^2 \left ( \vec{q } _ { \gamma } \times \left ( \frac { \vec{k}_a - \vec{k } _ b } { 2 } \right ) \right)^2 \times \left [ \vec{k}_+ \frac { 1 } { k_+^2+m_{\pi}^2 } - \vec{k}_- \frac { 1 } { k_-^2+m_{\pi}^2 } \right]^2 \biggr\ } \nonumber \\ w_{tt } & = & f_{hh'}(p ) \biggl\ { \frac { - 2 \mu _ p ^2 e^2 q_{\gamma}^2 } { m_p^2 } \left ( g(k_+ ) - g(k_- ) \right)^2 \nonumber \\ & - & \frac { e^2 } { m_p^2 } \left [ \left ( k_{a , x } g(k_- ) + k_{b , x } g(k_+ ) \right)^2 - \left ( k_{a , y } g(k_- ) + k_{b , y } g(k_+ ) \right)^2 \right ] \nonumber \\ & - & \frac { 256 } { 81 } \left ( \frac { f_{\gamma n \delta } f_{\pi n \delta } f_{\pi nn } } { m_{\pi } ^3 } \right)^2 g_{\delta } ^2 \left [ \left ( \vec{q } _ { \gamma } \times \left ( \frac { \vec{k}_a - \vec{k } _ b } { 2 } \right ) \right)_x ^2 - \left ( \vec{q } _ { \gamma } \times \left ( \frac { \vec{k}_a - \vec{k } _ b } { 2 } \right ) \right)_y ^2 \right ] \nonumber \\ & & \times \left [ \vec{k}_+ \frac { 1 } { k_+^2+m_{\pi}^2 } - \vec{k}_- \frac { 1 } { k_-^2+m_{\pi}^2 } \right]^2 \biggr\ } \ ; , \label{eq : wtwtt}\end{aligned}\ ] ] with @xmath86 the function @xmath87 appearing in the above expression is the fourier transform of the central correlation function @xmath88 and @xmath89 is the probability of finding a proton pair with quantum numbers @xmath90 , c.m . momentum @xmath10 and small internucleon separations in the target nucleus . the @xmath91 is the sum of the `` resonant '' ( direct @xmath92 term corresponding with fig.[fig : twon](d ) ) and non - resonant @xmath92 propagator ( crossed @xmath92 term corresponding with fig.[fig : twon](e ) ) . @xmath93 where @xmath941232 mev , @xmath95 is the free @xmath96 decay width and @xmath97 is the @xmath92 selfenergy in the medium . for light nuclei the imaginary part of this selfenergy is established to be 40 mev at full nuclear density @xcite . the @xmath98 and @xmath99 are the invariant energies of the @xmath92 in respectively the diagram of fig . [ fig : twon](d ) and ( e ) . in deriving the analytical expressions of eq.([eq : wtwtt ] ) the @xmath92-current terms in ( @xmath100 ) have been neglected . the first two terms in the expressions for @xmath64 and @xmath65 refer to one - body photoabsorption on the magnetization and convection current respectively . retaining only the contribution from @xmath92 current to the above structure functions and using the indentity ( [ eq : sigma ] ) one readily obtains the following expression for the @xmath16 asymmetry @xmath101 where ( @xmath102,@xmath103 ) are the polar and azimuthal angle of the escaping nucleons . for planar kinematics the isobar contribution to @xmath64 and @xmath65 is equal and in the absence of ground - state correlations the factorized model predicts that the @xmath104 asymmetry is exactly -1 . at first sight this conclusion seems to be at odds with the fact that symmetry observations require @xmath105 for @xmath106 . under these circumstances , however , also the @xmath92 current contributions to the @xmath64 and @xmath65 structure functions are predicted to be exactly zero for both photon polarizations , leaving the photon asymmetry @xmath54 undetermined . remark that the major contribution from the ground - state correlations , namely the one originating from photoabsorption on the one - body magnetization current @xcite , does equally contribute to @xmath64 and @xmath65 . moreover , in both cases it has the same relative sign with respect to the isobaric terms . for that reason the factorized scheme suggests that no particular sensitivity of the @xmath16 asymmetries to ground - state correlations should be expected . the above expression further suggests a sort of universal behaviour for the asymmetry of direct @xmath16 processes , which is independent from the shell - model structure of the active pair or target mass number a and only involves the momenta of the two ejected protons . we did not succeed in deriving a simple form for the @xmath107 . in order to reach scaling in terms of the pair function @xmath9 one relies on closure properties and such a procedure seems to be excluded when nucleon spin observables are probed . in refs . @xcite an unfactorized model for calculating two - nucleon knockout cross sections was presented . in comparison with the factorized model outlined in previous subsection it provides a more realistic description of the escaping nucleon wave functions . moreover , no assumptions are made regarding the nature of the relative wave function of the active nucleon pair . as a matter of fact , the unfactorized model does not make the separation between relative and center - off - mass motion as this can only be achieved in a harmonic oscillator basis . instead of working in a harmonic oscillator basis , we use mean - field quantities ( wave functions , potentials and phase shifts ) from a hartree - fock calculation with the effective skyrme force ske2 @xcite . we deem these single - particle wave functions to be realistic , as we could obtain a good description for the exclusive quasi - elastic @xmath108 cross sections using these wave functions and modern optical potentials @xcite . here , we solely summarize the physical ideas behind the unfactorized model and refer to refs . @xcite for the technical details . the principal idea is that following the absorption of a photon by the target nucleus , two nucleons are excited from a bound into a continuum eigenstate of a mean - field potential . by making a proper partial wave expansion with products of these continuum eigenstates , an antisymmetrized wave function can be constructed that is characterized by two asymptotically escaping nucleons and a residual @xmath32 fragment . in order to ensure the antisymmetrization of the final wave function we found it appropriate to start from a basis set in which the residual fragment is created with a particular angular momentum @xmath51 . within the outlined approach , the essential quantities that enter the calculation of the two - nucleon knockout cross sections are the reduced transition matrix elements of the type @xmath109 where @xmath110 ( @xmath111 ) are the bound state wave functions for the orbits from which the nucleons are escaping and @xmath25 ( @xmath112 ) the continuum eigenfunctions of the mean - field potential . the @xmath113 stands for the magnetic and electric transition operator which is obtained after making a multipole decomposition of the current operators . for the results presented here we reached convergence after including all electric and magnetic multipoles up to @xmath114=5 . each of the diagrams of figure [ fig : twon ] will contribute to the transition operator @xmath113 . for the meson - exchange and isobaric currents the expressions for the matrix elements are given in the appendix of ref . it should be noted that in the latter reference , which concentrated on photonenergies below the resonance region , the static limit for the @xmath92 propagators were considered . here an updated version that uses dynamic @xmath92 propagators and accounts for both the direct and crossed @xmath92 terms has been used . the corresponding current operator was described in detail in refs . this current operator could fairly well describe the photon energy dependence of the @xmath3c(@xmath1 ) and @xmath3c(@xmath0 ) cross sections through the @xmath92 resonance region @xcite . the procedure for including the ground - state correlation effects of figure [ fig : twon](f ) was outlined in ref . @xcite , where also the expression for the corresponding reduced matrix elements are given . we would like to stress that the unfactorized model can be used with plane wave outgoing nucleon waves . it suffices to use bessel functions for the continuum single - particle states @xmath115 and @xmath116 in the above matrix element . the fact that the same hamiltonian is used to construct the mean - field and the bound - state wave functions , together with the transverse nature of the isobaric current , makes the @xmath16 calculations gauge invariant . in what follows , predictions for @xmath2o and @xmath3c(@xmath117,@xmath118 ) differential cross sections and polarization observables will be presented . the sensitivity of the measurable quantities to the various aspects of the reaction mechanism will be discussed . the role of the final state interaction can be assessed by considering that the unfactorized model can be applied with either distorted or plane wave outgoing nucleon wave functions , keeping all other ingredients of the calculations exactly equal . the extent to which proton - proton knockout is dominated by @xmath6 wave absorption , will be estimated by comparing the results obtained within the context of the full unfactorized model with the predictions from the factorized scheme . indeed , the latter model is explicitly based on the dominance of @xmath28 proton pairs whereas no such restrictions are made for the unfactorized approach . in the course of this section we will suggest another way of gaining insight into the role of the different combinations for the relative wave function in the photoabsorption mechanism . the method is based on the selective nature of the photoabsorption mechanism when the final state is created with a specific angular momentum @xmath51 . the reason for concentrating on p - shell nuclei is partly inspired by the observation that the bulk of the experimental activities are concentrated in this part of the mass table . for the heavier target nuclei the transparency , which is related to the probability for nucleons to escape , is considerably smaller @xcite . accordingly , the ambiguities with respect to the fsi in two - nucleon emission reactions are expected to become increasingly worrying with increasing target mass number a. we consider two types of kinematical conditions that create favourable conditions to study the dynamics of pairs in the medium and present exclusive cross sections for ( groups of ) states in the low missing - energy range . in this subsection we are presenting cross sections in so - called `` quasi - deuteron '' ( qd ) kinematics which is constrained by imposing the condition that the missing momentum @xmath119 equals zero . accordingly , one is considering photoabsorption on dinucleons which are at rest ( this of course in an ideal world in which final state interactions could be completely ignored ) . for in - plane kinematics , the qd condition leads to a unique solution for @xmath120 , @xmath121 and @xmath122 for each given set of ( @xmath60 , @xmath123 , @xmath50 ) . the effect of the c.m . motion on the angular cross sections is then minimized as the pair function @xmath124 remains constant . accordingly it is hoped that a maximized sensitivity to the dynamics of the pair relative motion can be achieved . quasi - deuteron kinematics further allows comparing a@xmath15 observables with results obtained in photodisintegration of the deuteron . this opens perspectives to study in how far proton - neutron pairs in the medium resemble quasi - deuteron properties and to learn more about possible medium modifications of the photon - nucleus coupling . from the experimental point of view qd kinematics has the obvious advantage of sampling that region of the phase space for which the largest cross sections are expected . it is worth reminding that we have used realistic single - particle wave functions in our calculations and do not rely on an expansion in terms of relative and c.m . harmonic - oscillator wave functions in order to calculate the cross sections . nevertheless , it turns out that considering the different relative and c.m . combinations in the harmonic oscillator limit is very helpful in interpreting the general behaviour of the angular cross sections and polarization observables . in tables [ tab : pp ] and [ tab : pn ] we have collected all the possible quantumnumbers of the pair relative and c.m . wave functions for proton - proton and proton - neutron knockout from the different s- and p - shell combinations . the combinations are made as if the bound state wave functions are of the simple harmonic - oscillator type . after making a separation in the relative and c.m . motion through a moshinsky transformation the antisymmetric two - body wave functions read @xmath125 \ ; , \end{aligned}\ ] ] where @xmath126 , @xmath127 ( @xmath6 ) is the total isospin ( spin ) of the pair and @xmath128 ( @xmath129 ) the angular momentum of the relative ( c.m . ) pair wave function . in the above expression we have used the conventions of ref . @xcite for the angular momentum coupling coefficients . for direct two - nucleon knockout from a @xmath130 target nucleus the _ total _ angular momentum of the pair ( which is the sum of the c.m . angular momentum , relative angular momentum and total spin of the pair ) determines the quantum number @xmath51 of the residual fragment . by studying the cross section for the different @xmath51 s , which could experimentally be achieved in a high - resolution experiment @xcite , one can then study the behaviour of the observables as a function of the various types of relative pair wave functions . in qd kinematics , pair c.m . wavefunctions with an angular momentum @xmath131 are unlikely to contribute substantially to the differential cross sections . assuming that the inital photoabsorption occurs on a relative @xmath6 state , then knockout from the ( 1p@xmath132)@xmath133 configuration is expected to populate a @xmath134 ( proton - proton case ) and @xmath135 ( proton - neutron case ) . we have calculated the observables for proton - proton and proton - neutron knockout from @xmath3c considering qd kinematics and four representative photon energies . the results are summarized in figs . [ fig : seapp ] and [ fig : seapn ] and refer to the situation in which the @xmath32 fragments @xmath136b and @xmath136be are created in a ( 1p@xmath132)@xmath137 two - hole state . the figures show the contribution for the individual angular momentum components @xmath51 of the a-2 fragment as well as the result ( solid lines ) which is obtained after incoherently adding all the components . accordingly , the solid lines reflect the situation in which an integration over the missing - energy range for p - shell knockout would be made . the results of figs . [ fig : seapp ] and [ fig : seapn ] are in conformity with the above predictions in the sense that the major contribution to the respective cross sections comes indeed from @xmath51=0@xmath138 ( proton - proton case ) and @xmath51=1@xmath138 ( proton - neutron case ) . these are the states that one would expect to be populated assuming absorption on dinucleons in relative @xmath6 waves and c.m . angular momentum @xmath129=0 . the latter type of components are naturally favoured in qd kinematics . nevertheless , the calculations produce substantial contributions from configurations that fall beyound this qd picture . the population of the @xmath130 state in the @xmath15 reaction with a non - negligible cross section could be interpreted as a manifestation of photoabsorption on @xmath139=1 proton - neutron pairs which is a slightly unbound configuration in the free proton - neutron system . the next two important @xmath51 s contributing to the @xmath15 observables are @xmath134 and @xmath140 . the @xmath141 cross section is very small . it should be stressed that all above statements regarding the relative population of the states apply to qd kinematics and should not be considered as general . when moderate and larger values of the missing momentum @xmath10 are probed , configurations with c.m . angular momenta @xmath142 are expected to start playing a major role and the relative population of the different angular momentum states of the @xmath32 fragment is expected to be different . moving out of qd kinematics is expected to make the role of the c.m . configurations with @xmath143more important . at the same time , the cross sections will become smaller as larger values of the missing momentum are probed . when comparing the corresponding proton - proton and proton - neutron results of figs . [ fig : seapp ] and [ fig : seapn ] one observes that the @xmath15 asymmetries are generally smaller than the @xmath16 ones . a similar remark holds for the polarizations . it is also worth remarking that the shapes of the proton - proton and proton - neutron differential cross sections are considerably different at corresponding photon energies . the angular dependency and magnitude of the polarization observables is noticed to be remarkably close to the ones for the @xmath134 ( @xmath16 ) and @xmath135 ( @xmath15 ) component . even more than the cross sections the polarization observables seem to be dominated by photoabsorption on @xmath144 pairs in qd kinematics . this seems to be particularly the case for the higher end of the photon energies considered here and is more pronounced in the @xmath15 channel . in an attempt to investigate whether @xmath6 wave absorption automatically implies that the ( @xmath1 ) observables exhibit a deuteron like behaviour we have compared the @xmath3c(@xmath1 ) asymmetries with d(@xmath145)n data at corresponding photon energies . at e@xmath146=100 mev the deuteron asymmetry has very little resemblance with the @xmath3c predictions . this points towards different underlying reaction mechanisms . whereas , one - body photoabsorption plays an important role for the d@xmath147n observables at low photon energies @xcite , the calculations predict that @xmath3c(@xmath117,pn ) is dominated by pion exchange currents . as one moves into the @xmath92 resonance region , the calculated @xmath3c asymmetries move closer to the deuteron data . at e@xmath146=300 and 400 mev the agreement of the @xmath3c asymmetries with the deuteron data is even remarkable . at these energies , also the @xmath3c proton polarizations @xmath107 turn out to have the same sign and similar magnitude than what was obtained in d(@xmath148)n measurements . we now invesigate the sensitivity of the ( @xmath1 ) observables to the different ingredients which enter the calculations . [ fig : figa ] shows the calculated @xmath2o(@xmath1 ) results for e@xmath146=150 mev in qd kinematics . we consider the situation whereby the two nucleons are emitted from the 1p shell . two different types of shell - model configurations were considered : ( @xmath149)@xmath137 and ( @xmath149)@xmath150(@xmath151)@xmath150 . to obtain the curves of fig . [ fig : figa ] the contributions from the @xmath51 components to the various cross sections ( @xmath152 ) were incoherently added . the sensitivity to the various terms in the photoabsorption process is illustrated by comparing the dashed and dotted line . relative to a calculation that solely accounts for the seagull pion - exchange diagram ( fig . [ fig : twon](a ) ) , the inclusion of the pion - in - flight term ( fig . [ fig : twon](b ) ) reduces the cross section and does even switch the sign of the asymmetry . this illustrates the sensitivity of the latter to the different contributing terms in the pion exchange part of the photoabsorption mechanism . remark that even at photonenergies as low as 150 mev the effects from the isobaric current are sizeable . the dot - dashed line uses the full photoabsorption operator but plane waves for the escaping proton and neutron wave function . comparing these results with the solid line , which is obtained with exactly the same current operator but using distorted outgoing nucleon waves , one can estimate the role of the final - state interaction . the asymmetry is hardly affected by the outgoing nucleon distortions , whereas the differential cross section is roughly reduced by a factor of two . the effect of the distortions on the proton polarizations is large . its magnitude is of the same order as the deuteron results at corresponding photon energies @xcite . in comparing the @xmath15 results of fig . [ fig : figa ] for both shell - model configurations one gets a feeling about the nuclear structure dependency of the cross sections and polarization observables . in this context it is worth remarking that in the hartree - fock basis that we are using the radial dependence of the @xmath149 and @xmath151 single - particle wave functions differs . the general trends for all observables are rather uniform for both configurations . the subtle differences between the two situations are mainly a manifestation for the importance of mechanisms that go beyond @xmath153 absorption . indeed if there would be solely quasi - deuteron like absorption on ( @xmath153(t=0),@xmath129=0 ) pairs both shell - model configurations would be dominated by the @xmath51=1@xmath138 contribution . predictions for the @xmath3c@xmath16 cross section within the simple factorized model are shown in fig . [ fig : overpp ] and are compared with the full ( unfactorized ) model calculations . at low photon energies , where the range of the photons is really too large to expect mere photoabsorption on short - ranged @xmath6-waves , the factorized model largely undershoots the full model calculations . with increasing photon energy , the factorized predictions seem to get closer to the full model predictions . the full model calculations , however , produce differential cross sections which are not as sharply peaked . referring to the results of fig . [ fig : seapp ] , which shows the different @xmath51 contributions for the distorted wave ( solid line ) calculation of fig . [ fig : overpp ] , this is mainly due to excitation of the @xmath51=1@xmath138 state , which is somehow excluded when considering ( @xmath28,@xmath129=0 ) `` quasi - deuteron '' absorption . remark further that the proton angle dependence for the @xmath51=0@xmath138 contribution in the unfactorized model ( fig . [ fig : seapp ] ) , which is the configuration that dominant s - wave photoabsorption would select , bears a strong resemblance with the factorized model s prediction . as a consequence , it can be concluded that a major source of deviations between the results produced by the factorized scheme and the full model calculations has to be ascribed to photoabsorption mechanisms that go beyond a quasi - deuteron like mechanism . the effect of the final state interaction can be estimated by comparing the solid and dashed line in fig . [ fig : overpp ] . as is usually the case , the final - state distortions tend to widen the peaks of the angular cross sections obtained in the plane - wave approximation . the effect of the outgoing nucleon distortion is particularly large for the left panel of fig . [ fig : overpp ] . the typical outgoing nucleon kinetic energy is 20 - 50 mev , which makes this result not that surprising . remark that for this low photonenergy the effect of the fsi on the @xmath54 is even sizeable . for the right panel , corresponding with outgoing nucleon kinetic energies ranging from 60 - 210 mev , the role of the outgoing nucleon distortions is minor but not negligible . a plane wave calculation with a hermitian current operator would produce an outgoing nucleon polarization which is exactly zero . two mechanisms of completely different origin can make @xmath107 different from zero . first , final state interaction effects and secondly , the fact that the @xmath92 propagator @xmath154 of eq . ( [ eq : propa ] ) entering the isobaric current , contains imaginary parts . at low photon energies , where the damping effects of the @xmath92 propagator can only play a marginal role , one expects a plane wave calculation to produce @xmath107=0 . this fundamental property can be exploited to check the numerical accuracy of the calculations . it speaks in favour of the calculations that the plane wave result at @xmath155=100 mev does indeed obey this criterion for the full range of proton emission angles . at @xmath155=100 mev , the polarization is thus fully determined by the fsi mechanisms . in the resonance region , the occurrence of the @xmath92 width would make a plane wave calculation already produce a @xmath1560 . the @xmath107 is then a measure for the medium - dependent damping mechanism of intermediate @xmath92 creation and the fsi effects . this is confirmed by the results in the right panel of fig . [ fig : overpp ] , where neither of the two mechanisms is observed to be dominant . now we address the question in how far @xmath16 polarization observables could be helpful in discriminating between the different model predictions for the ground - state correlation effects . within the context of the present model this amounts to investigating the sensitivity to the different choices for the central correlation function ( sometimes referred to as defect function ) . in order to minimize the contributions from the isobaric currents , we have considered quasi - deuteron kinematics and photon energies on either side of the @xmath81 resonance : ( a ) e@xmath146=100 mev and ( b ) e@xmath146=400 mev . referring to eq . ( [ eq : wtwtt ] ) the effect of the ground - state correlations , which primarily comes through the magnetization current , is predicted to have a @xmath157 dependence , which make the src effects more likely to manifest themselves at higher photon energies . the left panel of fig . [ fig : wasrc ] , which refers to e@xmath146=100 mev , illustrates that even at lower photon energies the ground - state correlation effects do not overshoot the ( suppressed ) contribution from isobaric currents and exception made for the hard core correlation function of ref . @xcite ( dot - dashed line ) which is generally considered not to be very realistic . as a matter of fact , it is obvious from fig . [ fig : wasrc ] that dedicated ( @xmath0 ) experiments could unambiguously settle the unrealistic character of hard - core correlation functions . remark that there are relatively little uncertainties regarding the @xmath92 current operator at low photon energies as the static limit for the isobaric current operator would be fully justified . using realistic soft - core correlation functions , the ground - state correlation effects are relatively more visible in the differential cross section and proton polarization than in the photon asymmetry . at e@xmath146=400 mev the strength from the isobaric currents remains sizeable and the effect of the jastrow correlations on the observables is of the same size than for @xmath155=100 mev . the results of fig . [ fig : wasrc ] illustrate that the mere fact of having the photon absorbed on a @xmath28 diproton pair does not guarantee that the reaction process is dominated by ground - state correlations . coplanar and symmetrical kinematics refer to a specific situation for which the following kinematical conditions are obeyed : ( @xmath158 = @xmath159 , @xmath160 , @xmath161 , @xmath162 ) . for each nucleon emission angle @xmath102 and photonenergy , energy - momentum conservation will provide a unique solution for the momentum @xmath163 . in the considered kinematics the c.m . momentum p varies rapidly with the polar angle @xmath123 and within the factorized model of subsect . [ sec : fac ] we have @xmath164 = @xmath165 . so , in many respects coplanar and symmetrical kinematics is complementary to qd kinematics . whereas qd kinematics would rather be used to compare the relative motion of bound pairs with the deuteron , coplanar and symmetrical kinematics is intrinsically meant to explore the fact that in the medium nucleon pairs are characterized by c.m . degrees of freedom besides their relative motion . the strong dependence of the cross sections on the pair function @xmath9 , makes the absolute magnitude of the cross sections to vary dramatically as a function of the escaping nucleon angle in coplanar and symmetrical kinematics . within the factorized @xmath16 model of subsection [ sec : fac ] , the major contribution from the central correlations to the cross section , which comes from a coupling of the photon field to the one - body magnetization current , is exactly zero for @xmath166 = @xmath167 . accordingly , an abstraction made of the ( small ) contribution from the convection current , both the @xmath64 and @xmath65 structure functions from eq . ( [ eq : wtwtt ] ) are predicted to be dominated by the isobaric current when facing coplanar and symmetrical kinematics . figure [ fig : sunday ] displays the calculated @xmath2o@xmath16 and @xmath2o@xmath15 cross sections and photon asymmetries versus the proton emission angle and photon energy . we have considered a ( 1p@xmath168)@xmath137 configuration for the residual a-2 nucleus , so that only @xmath51=0@xmath138 can contribute for the @xmath169 case , whereas the @xmath170 channel can feed two angular momenta states , namely @xmath51=0@xmath138 and 1@xmath138 . for both channels and all emission angles considered a clear resonance at @xmath171260 mev is observed for the cross section . the resonance is not very visible in the asymmetry . the @xmath170 channel is characterized by a wider photon energy and emission angle dependence than the @xmath169 breakup channel . in both cases , however , the pair function @xmath9 makes the strength to reside in these parts of the phase space for which the missing momentum @xmath10 is small . at a fixed photon energy , the peak in the differential cross sections , which corresponds with the situation that the missing momentum is approximately zero , is localized at proton emission angles around 60 - 70 degrees in the lab frame . remark that @xmath172 corresponds with zero missing momentum and the maximum of the cross section for the low end of the photon energies considered here . at higher photon energies ( e@xmath173 300 mev ) the peak does gradually shift to 60 degrees in the lab frame . the @xmath174 situation in coplanar and symmetrical kinematics obeys also the qd conditions . from the right panel of fig . [ fig : sunday ] it becomes clear that the @xmath15 asymmetry is characterized by rather strong variations in both the photon energy and emission angle dependence . the smallest values are reached for the low photon energies . with increasing @xmath155 the isobaric currents gain in relative importance and as was already observed for the @xmath15 results of fig . [ fig : figa ] they tend to increase the asymmetry relative to the values that one would get including solely the meson - exchange contributions . the attention is drawn to the fact that the @xmath16 results of the left panel in fig . [ fig : sunday ] represent a rather unique case in the sense that the selected ( 1p@xmath168)@xmath137 final state can only feed the j@xmath175=0@xmath138 state which gets its dominant contribution through ( @xmath28,@xmath129=0 ) photoabsorption . an exception made for the lower photon energies , the proton - proton asymmetry is close to -1 for the full range of ( e@xmath146,@xmath176 ) covered in fig . [ fig : sunday ] . we remind that the asymmetry would be exactly -1 within the factorized model of subsect . [ sec : fac ] . this is another indication for the predictive power of the factorized model when @xmath28 absorption is the dominant configuration . our findings also confirm the calculations of ref . @xcite , where it was pointed out that even after including the c.m . motion of the pair , simple @xmath28 diproton breakup produces @xmath22he(@xmath177 ) asymmetries that are large and negative . with the aim of studying in greater detail the impact of the different relative angular momentum states on the @xmath104 asymmetries we display model predictions for a ( ( 1p@xmath132)@xmath137;@xmath178 ) and a ( ( 1p@xmath132)@xmath150(1p@xmath168)@xmath150;@xmath135 ) final state in fig . [ fig : monday ] . in comparison with the ( ( 1p@xmath168)@xmath137;@xmath134 ) situation , the considered shell - model configurations are not as selective as far as the different contributing relative angular momentum states is concerned . indeed , the ( 1p@xmath132)@xmath137 configuration can be excited in proton - proton knockout through @xmath6 , @xmath10 and @xmath179 absorption . the ( ( 1p@xmath132)@xmath150(1p@xmath168)@xmath150;@xmath135 ) configuration is unique in that @xmath6-wave absorption is excluded and only relative @xmath10 waves would contribute in a harmonic oscillator model @xcite . from the right panel of fig . [ fig : monday ] it becomes clear that this produces a doubly - bumped structure in the @xmath176 dependence with a minimum around @xmath174 . at the same time , the asymmetry is positive in most of the covered ( @xmath180 ) region . the left panel of fig . [ fig : monday ] shows features that turn out to be a hybrid mixture of characteristics related to @xmath6 and @xmath10 wave absorption . the @xmath10 wave admixture makes the asymmetry to be substantially smaller than what was obtained in the left panel of fig . [ fig : sunday ] where the @xmath6 wave dominates . rather strong variations of @xmath54 in e@xmath146 and @xmath181 are observed and there is very little resemblance with a flat structure wiggling slightly above the @xmath54=-1 plane as was noticed for the @xmath16 results in fig . [ fig : sunday ] . so , rather than fsi effects or fermi motion it turns out that photoabsorption on relative @xmath10 diprotons has a very large impact on the ( @xmath24,@xmath169 ) asymmetries . the present results seem to suggest that an admixture of @xmath10-wave absorption might also help in interpreting the measured @xmath21he(@xmath177 ) asymmetries ( ref . @xcite ) which were shown to be incompatible with @xmath182 photoabsorption in ref . @xcite . in an attempt to study the photonenergy dependence of the @xmath16 variables more profoundly and estimate the effect of the ground - state correlations , we display in fig . [ fig : coppsea ] some cuts along fixed values of the opening angle in the three - dimensional plot of fig . [ fig : sunday ] . the opening angles are chosen to lie in that part of the phase space where the cross sections are reasonably large . four values of the opening angles have been considered . the sensitivity of the cross sections to the ground - state correlations is found to be relatively small . so is the effect on the asymmetry for most of the phase space covered . however , the spike observed at @xmath176=90@xmath183 and lower photon energies ( see also fig . [ fig : sunday ] ) has to be fully ascribed to the ground - state correlations . it is maybe worth stressing that @xmath176=90@xmath183 corresponds with the situation that the largest values of the relative pair momentum are reached in coplanar and symmetrical kinematics . the effect of the short - range effects on the proton polarization is large . this might however be very characteristic for the chosen shell - model configuration in the final state . indeed , the proton polarization turns out to be zero when considering only the isobaric contribution and a 0@xmath138 final state . the corresponding @xmath15 results for the proton - proton results in fig . [ fig : coppsea ] are contained in fig . [ fig : coppnsea ] . for all nucleon emission angles considered , a wide @xmath92 resonant structure is visible in the cross section . surprisingly , the resonant behaviour is not pronounced for the polarization observables . this confirms the conclusion that the asymmetry is extremely sensitive to the interference terms between the different contributing two - body operators . remark that the predicted effect of central ground - state correlations on the @xmath15 observables is marginal . in this paper we have explored the possibilities of using polarized photon beams and outgoing nucleon polarimetry to learn more about the dynamics of bound nucleon pairs with the aid of @xmath16 and @xmath15 reactions . as it is often the case , the asymmetry was shown to be less dependent on the outgoing nucleon distortions than the corresponding differential cross sections . the effect of the outgoing nucleon distortions on the polarization , however , is rather large . the @xmath107 was further shown to be a sensitive observable to investigate the mechanisms related to the short lifetime of the @xmath92 resonance in the medium . indeed , the latter reflect themselves in imaginary parts entering the @xmath92 selfenergy and produce sizeable contributions to the outgoing nucleon polarization . it was pointed out that by studying the differential cross sections for excitation of the @xmath32 fragment in a state with particular angular momentum @xmath51 one can deduce information about the nature of the initial pair wave function . this does not imply , however , that the experiments would have to resolve each residual state . observables for different shell combinations , which would typically feed the residual system in a certain range of missing energies , would already be very instructive to learn about the different possible pair combinations . we have shown that the observables are characterized by a selective sensitivity to mechanisms that go beyound photoabsorption on a @xmath28 proton - proton and @xmath153 proton - neutron pair . this was found to be already the case in so - called quasi - deuteron kinematics where the missing momentum is constrained to be zero , so that the influence of the pair c.m . motion is minimized and `` deuteron - like '' conditions are created . in this type of kinematics , the major deviations from s - wave absorption are visible in the angular cross sections . the polarization observables seem to be far less affected . the fact that s - wave absorption plays a predominant role does not guarantee , however , that the @xmath15 asymmetries exhibit deuteron - like properties in quasi - deuteron kinematics . in the @xmath92 resonance region a strong similarity between the d@xmath184 and the @xmath3c@xmath184 asymmetries is observed . the situation changes for the lower photon energies where the @xmath3c predictions deviate substantially from the measured deuteron asymmetries . for all these reasons , we deem that the exclusive @xmath31 channel is an ideal probe to study the limitations of the qd approach and to reach a better level in our understanding of the dynamics of proton - neutron pairs in the medium relative to the deuteron . within the factorized model , a general behaviour for the a@xmath104 asymmetries is predicted in that they would be exactly -1 as long as one is dealing with @xmath28 absorption in coplanar kinematics and the dominant role of the isobar current is guaranteed . neither fsi effects , nor a more sophiscated treatment of the pair relative @xmath6 wave function seem to change this very much . the slightest admixture of mechanisms going beyond @xmath28 absorption are noticed to induce major changes to the asymmetry . whereas @xmath185=-1 for @xmath186 photoabsorption , the calculations predict a completely different behaviour as soon as relative @xmath187 start playing a role . the extent to which mechanisms beyound @xmath6 wave absorption play a role is very much dependent on the kinematics and the nuclear structure of the a-2 fragment . the sensitivity of the ( @xmath0 ) asymmetries to ground - state correlations was shown not to be very large and was considerably smaller than the effects on the angular cross sections and outgoing nucleon polarizations . such a behaviour could already be inferred from the expressions derived within the context of the factorized model . all this , however , is illustrating the usefulness of having simplified models for predicting the major trends and sensitivities in the different observables . it would be of interest to see to which degree the deviations from qd - like absorption and deuteron - like behaviour can be experimentally confirmed . in any case , high or moderate resolution data in the low - energy part of the missing energy spectrum would provide invaluable information to test wether the dynamics of the pairs in the medium complies with mean - field like behaviour or whether there are important deviations in ( some of ) the relative wave function combinations . in order to minimize the uncertainties regarding the final state interaction polarization observables , and in particular the photon asymmetries , will be of very great help in these studies . this work was supported by the fund for scientific research - flanders ( fwo ) and in part by the nato through the research grant crg970268 . one of us ( j.r . ) thanks the institute for nuclear theory at the university of washington for partial support during the completion of this work . 99 k. gottfried , nucl . phys . * 5 * , 557 ( 1958 ) . s. homma , m. kanazawa , m. koike , y. murata , h. okuno , f. soga , n. yoshikawa and a. sasaki , phys . lett . * 52 * , 2026 ( 1984 ) . s.n . dancer _ lett . * 61 * , 1170 ( 1988 ) . et al . _ , phys . lett . * b380 * , 247 ( 1996 ) . p. grabmayr _ et al . _ , b370 * , 17 ( 1996 ) . k.i . blomqvist _ et al . _ , submitted for publication . lett . * 78 * , 4893 ( 1997 ) . m. kanazawa _ et al . _ , phys . rev . * c35 * , 1828 ( 1987 ) . j. arends _ et al . _ , z. phys . * a298 * , 103 ( 1980 ) . d.j . tedeshi _ lett . * 73 * , 408 ( 1994 ) . h. baghaei _ _ , proceedings of the second workshop on electromagnetically induced two - nucleon emission ( gent , may 17 - 20 , 1995 ) , eds . j. ryckebusch and m. waroquier , ( university of gent publication , 1995 ) , 195 and r. lindgren _ et al . _ , submitted for publication . a.m. sandorfi and w. leidemann , phys . c * 53 * , 1506 ( 1996 ) . p. wilhelm , j.a . niskanen and h. arenhvel , nucl . a597 * , 613 ( 1996 ) . carrasco , m.j . vicente vacas and e. oset , nucl . phys . * a570 * , 701 ( 1994 ) . et al . _ , a593 * , 463 ( 1995 ) . et al . _ , proceedings of the second workshop on electromagnetically induced two - nucleon emission ( gent , may 17 - 20 , 1995 ) j. ryckebusch and m. waroquier , ( university of gent publication , 1995 ) , 179 . et al . _ , z. phys . a * 355 * , 1 ( 1996 ) . c. giusti and f.d . pacati , nucl . phys . * a585 * , 618 ( 1995 ) . j. ryckebusch , m. vanderhaeghen , l. machenil and m. waroquier , nucl . phys . * a568 * , 828 ( 1994 ) . s. boffi , c. giusti , f.d . pacati and m. radici , nucl . phys . * a564 * , 473 ( 1993 ) . c. giusti and f.d . pacati , nucl . phys . * a615 * , 673 ( 1997 ) . j. ryckebusch , v. van der sluys , k. heyde , h. holvoet , w. van nespen , m. waroquier and m. vanderhaeghen , nucl . a ( 1997 ) , in print . m. wakamatsu and k. matsumoto , nucl . phys . * a392 * , 323 ( 1982 ) . htt , a.i . milstein and m. schumacher , a618 * , 483 ( 1997 ) . t. ericson and w. weise , _ in _ pions and nuclei ( oxford university press , new york , 1988 ) . m. vanderhaeghen , k. heyde , j. ryckebusch and m. waroquier , nucl . phys . * a595,219 ( 1995 ) . towner , phys . rep . 155 * , 263 ( 1987 ) . l. boato and m.m . giannini , j. phys . g. * 15 * , 1605 ( 1989 ) . j. ryckebusch , l. machenil , m. vanderhaeghen and m. waroquier , phys . lett . * b291 * , 213 ( 1992 ) . j. ryckebusch , phys . lett . * b383 * , 1 ( 1996 ) . m. vanderhaeghen , l. machenil , j. ryckebusch and m. waroquier , nucl . * a580 * , 551 ( 1994 ) . e. oset and l.l . salcedo , nucl . a468 * , 631 ( 1987 ) . m. waroquier , j. ryckebusch , j. moreau , k. heyde , n. blasi , s.y . van der werf and g. wenes , phys . rept . * 148 * , 249 ( 1987 ) . v. van der sluys , k. heyde , j. ryckebusch and m. waroquier , phys . rev . c * 55,1982 ( 1997 ) . et al . _ , submitted for publication . g. garino _ et al . _ , c 45 * , 780 ( 1992 ) . pandharipande and steven c. pieper , phys . c * 45 * , 791 ( 1992 ) . k. heyde , the nuclear shell - model ( springer verlag , berlin heidelberg , 1994 ) . l. isaksson , ph.d . thesis , department of physics , university of lund ( 1996 ) . h. arenhvel and m. sanzone , _ photodisintegration of the deuteron _ , _ in _ few body systems -supplementum 3 , ( springer verlag , wien , 1991 ) . f. ritz , h. arenhvel and t. wilbois , university of mainz preprint mkph - t-97 - 19 , nucl - th/9707045 . levchuk , few - body systems * 19 * , 77 ( 1995 ) . gearhart , phd thesis , university of washington ( 1994 ) unpublished , and w. dickhoff , private communication . g. blanpied _ et al . c * 52 * , r455 ( 1995 ) and legs data release l1 - 3.0 available from http://www.legs.bnl.gov/ legs/. v.g . gorbenko , yu.v . zhebrovskij , l.ya . kolesnikov , a.l . rubashkin and p.v . sorokin , nucl . a381 * , 330 ( 1982 ) . v.b . ganenko _ _ , z. phys . a. * 341 * , 205 ( 1992 ) . i. arias de saaverdra _ et al . _ , nucl . phys . * a605 * , 354 ( 1996 ) . clark , lecture notes in physics * 138 * , eds . r. guardiola and j. ros ( springer verlag , berlin , 1981 ) p 184 . .possible configurations for proton - proton knockout from the s- and p - shell combinations . the @xmath129 ( l ) denotes the c.m . ( relative ) angular momentum of the pair . the j@xmath188 is the total angular momentum of the pair ( including relative and c.m . the separation in c.m . and relative motion is done as if we are dealing with ho single - particle wave functions . [ cols="<,^,^,^,^ " , ]	the polarization degrees of freedom in photoinduced two - nucleon knockout from finite nuclei are studied . it is pointed out that they open good perspectives to study the dynamics of dinucleons in the medium in detail . the ( @xmath0 ) and ( @xmath1 ) angular cross sections , photon asymmetries and outgoing nucleon polarizations are calculated for the target nuclei @xmath2o and @xmath3c and photonenergies ranging from 100 up to 500 mev . it is investigated to which degree the two - nucleon emission reaction is dominated by photoabsorption on @xmath4 proton - neutron and @xmath5 proton - proton pairs in the nuclear medium . the calculations demonstrate that dominance of @xmath6 wave photoabsorption in the ( @xmath1 ) channel does not necessarily imply that the reaction mechanism is similar to what is observed in deuteron photodisintegration .
to date , most of our knowledge about the relativistic phase of the interstellar medium ( ism ) outside of the galaxy , consisting of relativistic charged particles and magnetic fields , has been obtained indirectly through the detection of synchrotron emission via multi - frequency radio observations ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . synchrotron emission arises from cosmic - ray ( cr ) electron energy losses as these particles are accelerated in the magnetic fields of galaxies . although the energy density in cr electrons is only @xmath21% of that for cr nuclei , the similarity between the spatial distributions of gamma - ray and synchrotron emission within the galaxy suggests that cr electrons and cr nuclei are fairly well mixed on the scales of a few hundred parsecs ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? the spatial distribution of a galaxy s synchrotron emission is a function of a galaxy s cr electron and magnetic field distributions . thus , radio synchrotron maps provide only limited insight on the source distribution of the cr electrons as well as the distances the particles may have traveled before ending up in their current location of emission . massive stars ( @xmath48 @xmath5 ) are the progenitors of supernovae ( sne ) whose remnants ( snrs ) , through the process of diffusive shock acceleration @xcite , appear to be the main acceleration sites of cr electrons responsible for a galaxy s observed synchrotron emission . these same young massive stars are often the primary sources for dust heating as they emit photons which are re - radiated at far - infrared ( fir ) wavelengths . this shared origin between the fir and radio emission of galaxies is thought to be the foundation for the observed fir - radio correlation among ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and within galaxies ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? coupling the shared origin of a galaxy s fir and radio emission with the fact that the mean free path of dust - heating photons ( @xmath2100 pc ) is significantly shorter than the expected diffusion length of cr electrons ( @xmath21 - 2 kpc ) led @xcite to conjecture that the radio image of a galaxy should resemble a smoothed version of its infrared image . consequently , it appears that the close spatial correlation between the fir and radio continuum emission within galaxies can be used to characterize the propagation history of cr electrons . this prescription has been shown to hold for galaxies observed at the `` super resolution '' ( @xmath01@xmath6 ) of _ iras _ hires data @xcite and , more recently , for high resolution ( @xmath218@xmath7 ) _ spitzer _ 70@xmath8 imaging ( * ? ? ? * hereafter m06a ) . this phenomenology has been further corroborated on scales @xmath450 pc by @xcite who find synchrotron haloes around individual star - forming regions are more extended than fir - emitting regions within the large magellanic cloud . * hereafter , m06b ) recently studied how the spatial distributions of a galaxy s fir and radio emission vary as a function of the intensity of star formation . they concluded that cr electrons are , on average , younger and closer to their place of origin within galaxies having higher amounts of star formation activity compared with more quiescent galaxies . using a wavelet - based image decomposition , we extend this work by attempting to characterize separately cr electron populations associated with a galaxy s diffuse disk and its star - forming complexes . we carry out this study for a sample of galaxies observed as part of the _ spitzer _ infrared nearby galaxies survey ( sings ; * ? ? ? * ) and the westerbork synthesis radio telescope ( wsrt - sings ; * ? ? ? * ) for which we have the spatial resolution to resolve physical scales @xmath9 kpc . the complete study can be found in @xcite . observations at 24 , 70 , and 160 @xmath1 were obtained using the multiband imaging photometer for _ spitzer _ ( mips ; * ? ? ? * ) as part of the sings legacy science program . a detailed description of the sings observational strategy can be found in @xcite . radio continuum imaging at 22 cm was performed using the westerbork synthesis radio telescope ( wsrt ) as part of the wsrt - sings survey . a complete description of the radio observations and image processing steps can be found in @xcite . we match the resolution of the mips and radio images using gaussian psfs rather than the mips psfs , which suffer from significant power in their side - lobes . we compute the radiation field energy density ( @xmath10 ) of each galaxy using its tir ( @xmath11 ) surface brightness since this parameter is sensitive to the diffusion of cr electrons . using the deprojected area of elliptical apertures ( @xmath12 ) , we calculate tir surface brightnesses @xmath13 along with estimates of @xmath10 for radiation emitted near the surface of a semi - transparent body such that @xmath14 where @xmath15 is a galaxy s bolometric surface brightness , @xmath16 is the speed of light , and all quantities are given in cgs units . the parenthetical term in equation [ eq - urad ] provides a correction for non - absorbed uv emission that was empirically derived by @xcite using archived fir and uv data for a sample of more than 200 galaxies . using a phenomenological image - smearing model , first presented by @xcite , m06b studied how the spatial distributions of fir and radio emission varied as a function of star - formation intensity within 12 spiral galaxy disks . the basic procedure used here is similar to that presented in m06a , b . we calculate and minimize the residuals between the free - free corrected radio and observed infrared images after convolving the infrared maps by a parameterized kernel @xmath17 . the new element is that we now , using a wavlet - based technique , decompose each observed infrared image , into two sub - images : ( 1 ) a _ structure _ image containing features with spatial scale smaller than 1 kpc ; ( 2 ) a _ disk _ image containing all structures with characteristic spatial scales larger than or equal to 1 kpc , largely constituting a galaxy s diffuse disk . the _ structure _ and _ disk _ images are smeared independently , summed , and then compared to the radio image . for a full description of the phenomenological model we refer the reader to @xcite . by plotting the smearing scale - length which best matches the infrared image to the observed radio map versus tir surface brightness ( see figure [ fig-1 ] ) , we find that the fir and non - thermal radio morphologies are more similar to each other for galaxies having higher radiation field energy densities compared to galaxies with lower radiation field energy densities . following the interpretation of m06b , our results indicate that cr electrons are , on average , closer to their place of origin in galaxies having higher star formation activity . as cr electrons propagate through the ism of galaxies they lose their energy due to a number physical processes including synchrotron , inverse - compton ( ic ) scattering , bremsstrahlung , ionization , and adiabatic expansion losses . in normal galaxies synchrotron and ic scattering processes are likely the most significant energy loss terms for cr electrons associated with 1 ghz emission @xcite ; the other terms listed will become non - negligible , however , for galaxies hosting extreme episodes of star - formation like starbursting ultra - luminous infrared galaxies ( ulirgs ) ( e.g. * ? ? ? a cr electron having energy @xmath18 will emit most of its energy at a critical frequency @xmath19 where @xmath20 from this , we can express the effective cooling timescale for cr electrons due to synchrotron and ic losses as @xmath21 in simple diffusion models , the propagation of cr electrons is usually characterized by an empirical , energy - dependent diffusion coefficient , @xmath22 ( e.g. * ? ? ? the value of @xmath22 has been found to be around @xmath23 for @xmath0gev crs by fitting diffusion models with direct measurements of cr nuclei ( i.e. secondary - to - primary ratios like boron - to - carbon ) within the solar neighborhood ( e.g. * ? ? ? we will assume , @xmath24 now , neglecting escape and using a simple random - walk equation , cr electrons will diffuse a distance @xmath25 before losing all of their energy to synchrotron and ic losses . by combining equations [ eq - nube ] and [ eq - de ] , we can express @xmath22 as a function of @xmath26 for a fixed @xmath19 such that , for cr electrons having energies @xmath271 gev , @xmath28 using the above equations , we derive simple , order - of - magnitude estimates to determine whether diffusion and cooling of cr electrons in a steady - state star formation model are able to account for our observations . taking the mean value of @xmath10 for those galaxies plotted in figure [ fig-1]b ( i.e. @xmath29erg @xmath30 ) , and assuming @xmath31 , which has been shown to be a reasonable assumption for a large sample of spiral galaxies @xcite , we find from equation [ eq - cool ] that the average cooling time for a 1.4 ghz emitting cr electron is @xmath2@xmath32 yr . inserting this value into equation [ eq - diff ] , we measure a diffusion scale - length of @xmath26.8 kpc ; this value is clearly off of the scale shown in figure [ fig-1]b . on the other hand , if we instead assume a fixed , typical magnetic field strength of 9 @xmath33 g @xcite , and that @xmath34 erg @xmath30 , the average cooling time for a 1.4 ghz emitting cr electron is @xmath2@xmath35 yr with a diffusion scale - length of @xmath22.6 kpc . while this value for @xmath10 is much higher than what we infer from the average tir surface brightness of the sample , it must apply near bright star - forming structures , whose tir surface brightnesses are much greater . even so , this diffusion scale - length is much larger than any value we find for the best - fit structure scale - lengths . from these simple order of magnitude estimates it appears that particle fading due to cooling by inverse compton and synchrotron processes alone can not explain the structural differences between the fir and radio maps ; the best explanation seems to be differences in the cr electron population ages . in order to provide a physical interpretation for the location of galaxies in the best - fit scale - length@xmath10 diagram , we describe their general locations , designated by roman numerals , in figure [ fig-2 ] . while the placement of ngc 925 , ngc 4725 , ngc 4559 in figure [ fig-1]a may be the result of a signal - to - noise effect , we will now consider alternative physical explanations . the irregular galaxies in our sample ( holmb ii , ic 2574 , and ngc 4236 ) have also been excluded thus far in our discussion due to their lack of a disk component in the fir and radio . the placement of these objects in figure [ fig-2 ] and their behavior will now be put into the context of our phenomenological picture . a. in this region of the diagram we assume a galaxy has just reached a peak in its surface brightness after a recent episode of enhanced star formation ; the episode must have occurred within the last few myr to ensure that snrs are still relatively young and that the morphologies of the radio and infrared emission are similar because cr electrons have not had time to diffuse very far . b. after reaching a peak in global star formation activity , those cr electrons associated with the recent enhancement of star formation will dominate the cr population . as they spread through the galaxy they will begin to lose their energy to synchrotron and ic processes . a galaxy s position along this trend can not be due to variations in ism parameters alone ; rather , its star formation history is the dominant parameter ( see figure [ fig-1]b ) . galaxies may move along this part of phase space as star formation activity becomes enhanced and then fades . c. galaxies in this region are characterized by low star formation activity and have best - fit scale - lengths which are shorter than expected given the trend found in region ii . while this situation can be explained by a signal - to - noise effect , it is can also be explained by a significant amount of fir emission arising from an old stellar population . d. we find no galaxies occupying this region of phase space in which relatively short global scale - lengths would be measured for a moderate values of @xmath10 . one way for a galaxy to populate this part of the diagram is if , after a long period of quiescence , it begins to form stars at a moderate rate . the galaxy would then pass through this region very quickly ( @xmath010@xmath36 yr ) before shifting into region ii . the lack of such galaxies suggests that , at least in the local universe , star formation in spirals does not completely cease for long periods of time . e. the irregulars behave markedly different than the sample spirals . we do not believe this discrepant behavior to be the result of signal - to - noise effects . the morphologies and ism of these galaxies do not seem consistent with keeping their cr electrons bound outside of their initial clouds around snrs . after the cr electrons leave this cloud , the lack of a dense ism and magnetic field to keep them trapped through multiple scatterings off of magneto - hydrodynamic ( mhd ) waves allows them to easily escape the system and enter intergalactic space . using a two - component image - smearing analysis , we have separated the signatures of cr electron diffusion at spatial scales corresponding to star - forming structures ( @xmath371 kpc ) and galaxy disks ( @xmath271 kpc ) within 18 galaxies observed as part of sings and wsrt - sings . our results and conclusions can be summarized as follows : 1 . the best - fit global scale - lengths decrease as a function of increasing star formation activity as measured by the infrared surface brightness of a galaxy . our interpretation is that a galaxy s cr electrons are closer to their place of origin within galaxies having intense star formation activity . 2 . the trend of decreasing best - fit _ global _ scale - length with increasing radiation field energy density is due to higher surface brightness galaxies having undergone a recent enhancement of star formation rather than variations in other ism parameters . for sufficiently large enhancements , these galaxies are observed within @xmath210@xmath3 yr of the onset of the most recent star formation episode . 3 . unlike spirals , irregular galaxies lack any well defined diffuse disk component at either 70 @xmath1 , or especially at 22 cm . presumably , the cr electrons escape these galaxies soon after leaving their parent star - forming regions due the absence of a dense ism which would keep large - scale interstellar magnetic field locked into place . this conclusion helps to explain why these galaxies have global fir / radio ratios systematically greater than canonical values . 4 . as infrared surface brightness increases , the characteristic diffusion scale - length of a galaxy s cr electron population begins to transition at @xmath38 from being biased by cr electrons making up its diffuse disk to being biased by those recently injected near star - forming structures . from this we conclude that a galaxy s cr electron population transitions from being dominated by old cr electrons to being dominated by young cr electrons as a function of star formation intensity . the two - component analysis works better than smearing with a single smoothing kernel for spiral galaxies of type sb or later which have high amounts of ongoing star formation activity ( i.e. @xmath240% of the sample ) . this result suggests that star formation must be intense and highly structured for the two - component analysis of these data to differentiate properly between the different cr electron populations . e.j.m would also like to thank other members of the _ spitzer _ infrared nearby galaxies survey ( sings ) team for their invaluable contributions to the work presented here , especially those of r.c.kennicutt , jr . , d.calzetti , k.d.gordon , c.w.engelbracht , and g.bendo . as part of the _ spitzer _ space telescope legacy science program , support was provided by nasa through contract number 1224769 issued by the jet propulsion laboratory , california institute of technology under nasa contract 1407 .	we present results on the interstellar medium ( ism ) properties of 29 galaxies based on a comparison of _ spitzer _ far - infrared and westerbork synthesis radio telescope radio continuum imagery . of these 29 galaxies , 18 are close enough to resolve at @xmath01 kpc scales at 70 @xmath1 and 22 cm . we extend the @xcite approach of smoothing infrared images to approximate cosmic - ray ( cr ) electron spreading and thus largely reproduce the appearance of radio images . using a wavelet analysis we decompose each 70 @xmath1 image into one component containing the star - forming _ structures _ and a second one for the diffuse _ disk_. the components are smoothed separately , and their combination compared to a free - free corrected 22 cm radio image ; the scale - lengths are then varied to best match the radio and smoothed infrared images . we find that late - type spirals having high amounts of ongoing star formation benefit most from the two - component method . we also find that the disk component dominates for galaxies having low star formation activity , whereas the structure component dominates at high star formation activity . we propose that this result arises from an age effect rather than from differences in cr electron diffusion due to varying ism parameters . the bulk of the cr electron population in actively star - forming galaxies is significantly younger than that in less active galaxies due to recent episodes of enhanced star formation ; these galaxies are observed within @xmath210@xmath3 yr since the onset of the most recent star formation episode . the sample irregulars have anomalously low best - fit scale - lengths for their surface brightnesses compared to the rest of the sample spirals which we attribute to enhanced cr electron escape .
ukidss is the ukirt infrared deep sky survey @xcite , carried out using the wide field camera ( wfcam ; * ? ? ? * ) installed on the united kingdom infrared telescope ( ukirt ) . data acquisition for the survey started in 2005 may . the early data release ( edr ; * ? ? ? * hereafter d06 ) , a prototype dataset , and the first data release ( dr1 ; * ? ? ? * hereafter w07 ) , the first release of survey - quality data , took place in 2006 . this paper describes the second data release ( dr2 ) . the data are available from the wfcam science archive ( wsa ) at http://surveys.roe.ac.uk/wsa . ukidss is a programme of five imaging surveys that each uses some or all of the broadband filter complement @xmath3 , and that span a range of areas , depths , and galactic latitudes . there are three high galactic latitude surveys , providing complementary combinations of area and depth ; the large area survey ( las ) , will cover 4000 deg@xmath0 to @xmath1 , the deep extragalactic survey ( dxs ) , 35 deg@xmath0 to @xmath2 , and the ultra deep survey ( uds ) , 0.8 deg@xmath0 to @xmath4 . there are two other wide surveys to @xmath1 , aimed at targets in the milky way ; the galactic plane survey ( gps ) will cover 1900 deg@xmath0 , and the galactic clusters survey ( gcs ) 1100 deg@xmath0 . the complete ukidss programme is scheduled to take seven years , requiring @xmath51000 nights on ukirt . the current implementation strategy is focused on completing an intermediate set of goals , defined by the ` 2-year plan ' detailed in d06 . all magnitudes quoted in this paper use the vega system described by @xcite . depths , where not explicitly specified , are the total brightness of a point source for which the flux integrated in a @xmath6 diameter aperture is detected at @xmath7 . a set of five baseline papers provides the relevant technical background information for the surveys . the overview of the programme is given by @xcite . this sets out the science goals that drove the design of the survey programme , and details the final coverage that will be achieved , in terms of fields , areas , depths , and filters . the camera is described in detail by @xcite , and the _ zyjhk _ photometric system is characterised by @xcite , who provide synthetic colours for a wide range of types of star , galaxy , and quasar . details of the data pipeline and data archive will appear in irwin et al . ( 2007 , in prep . ) and hambly et al . ( 2007 ) , respectively . each ukidss data release includes the data contained in previous releases , together with new data . the instrument is scheduled in blocks of time of typically three to six months duration . current policy is that each observing block results in a data release . dr1 comprised data from the 2005a and 2005b observing blocks . dr2 includes the relatively short 2006a observing block , which ran over 2006 may to july . dr3 will be much larger , including the consecutive 2006b and 2007a blocks , running over 2006 november to 2007 may . each ukidss data release is accompanied by a paper summarising the contents of the release , and detailing procedural changes since the previous release . the edr provided a small prototype datset , and represented a step towards regular release of survey - quality data.the edr paper , d06 , is a self - contained summary of all information relevant to understanding the contents of the edr . it also serves as the baseline paper for technical details for all releases . besides a summary of the contents of the edr database , d06 includes relevant details of the camera design , the observational implementation ( integration times , microstepping ) , the pipeline and calibration , data artifacts , and the quality control procedures , as well as a brief guide to querying the archive . subsequent papers summarise the contents of the particular release , together with a description of changes to the implementation , pipeline , calibration , quality control , and archive procedures . the dr1 paper , w07 , also included plots summarising the distributions of seeing , sky brightness , and airmass over the dataset . the present paper covers dr2 . dr2 includes new data for the las , gps , gcs , and dxs , but not the uds which was unobservable in 2006a . the uds therefore appears in dr2 in identical form to dr1 . details concerning the uds may be found in w07 , and no further mention is made here . dr2 is appeared in two stages . the first stage , released on 2007 march 1 , was complete except for the gps . the gps data were released on 2007 april 12 . in section [ update ] we detail changes between dr1 and dr2 to the pipeline , calibration , and archive procedures . in section [ summary ] we summarise the contents of dr2 . d06 contains details of the implementation , pipeline , calibration , quality control , and archive procedures applied to the edr data , and a glossary of technical terms . w07 details changes to these procedures betwen the edr and dr1 . in this section we detail further changes since dr1 . these include mostly minor changes to the pipeline , calibration , and archive procedures . the implementation and quality control procedures are virtually unchanged . during 2006a , fields were not observed when the moon was within 30deg , in order to avoid moon ghosts ( d06 , w07 ) . a baffle was installed after the end of 2006a that appears to have eliminated moon ghosts altogether , and the restriction has been removed for 2006b . for dr2 all 2005a wfcam data were reprocessed from scratch from the raw data . the 2005a data were first released in the edr , and were not revised for dr1 . the main changes in the image processing aspects between edr and dr1 , and used in processing the 2005b data , were an improved sky correction strategy more closely linked to the observing block structure ( msbs ) , and the application of the cross - talk supression algorithm ( w07 ) . further minor improvements have been made to the sky correction strategy since dr1 . therefore the 2005a and 2006a data benefit from these improvements . the 2005b data have not been reprocessed since dr1 however . further small changes to the cataloguing software , mainly related to improving the object detection filter , have also been made since dr1 and were used in the catalogue ( re)generation for the 2005a data . all of the new 2006a wfcam data were processed with the latest versions of the pipeline software . several minor improvements to the photometric calibration procedure have also been implemented since dr1 in an attempt to reduce any dependence of the derived photometric zero - points on galactic coordinates , due to , for example , varying dwarf - giant ratios , extinction and extreme colour objects . as previously , each frame is calibrated using 2mass stars in the frame , converting the 2mass photometry to the wfcam system using appropriate colour equations . the zero point for each detector is now provided as the attribute photzpcat in the table multiframedetector . the attribute photzp in the table multiframe has been deprecated . the updated wfcam calibration uses a restricted ( extinction - corrected ) colour range of @xmath8 ( the subscript @xmath9 standing for 2mass ) to help exclude late - type giants , unusual objects and heavily reddened stars . also , as a result of discovering a correlation of the derived @xmath10- and @xmath11-band zero points with galactic extinction , particularly in regions of heavy extinction , an extra extinction - dependent calibration term has been included . the revised colour equations are as follows : + @xmath12 + @xmath13 + @xmath14 + @xmath15 + @xmath16 here @xmath17 is the reddening computed using the prescription of @xcite , from the data of @xcite . the corrections for reddening are small in the _ jhk _ bands and in practice also for the @xmath10 band which is only used in the las where the extinction is low . the reddening corrections for the _ z _ band , only used in the gcs , are large in fields with large reddening , and should be treated with caution . the revised colour equations produce a more stable calibration for the @xmath10 and @xmath11 bands ( but generally have little impact on the _ jhk _ calibration ) . in addition , errors on the individual 2mass - derived zero - points are now included together with measures summarising the photometric quality of the night and the overall nightly average zero - point . w07 note that there is evidence for an overall zero - point offset with respect to the vega system for the @xmath10 and @xmath11 bands . an analysis of this issue is deferred to dr3 . in previous releases two databases were provided , one containing all the data ( called edr+ , dr1 + ) , and a second , a subset of the larger database , containing data for all fields observed with the full filter complement for the particular survey ( called edr , dr1 ) . for dr2 there is only one database , dr2 + . nevertheless _ views _ are provided of the source tables for each survey , e.g. _ lasyjhksource _ in order to provide the equivalent functionality . a significant improvement over dr1 is the provision of quality error bit flags for each detected source . in the detection tables the attribute is pperrbits , and in the source tables the appropriate filter name is appended , e.g. ypperrbits , j@xmath181pperrbits . the attribute is a 32-bit integer where each bit can be set to 1 to flag up to 32 different quality issues . the quality issues are ranked in severity , with the more severe issues assigned to higher bits . currently only five bits are set as follows : bit 4 = the source is the offspring of deblending , bit 6 = a bad pixel exists in the default aperture ( @xmath19 radius ) , bit 16 = the source is likely to contain saturated pixels , bit 19 = the source is likely to be either cross - talk or a real source affected by cross - talk ( see d06 and w07 for details on cross - talk ) , bit 22 = the source is likely to include incomplete data as a consequence of lying close to the frame boundary . therefore by converting pperrbits to binary it is possible to filter out sources with particular quality issues . alternatively a threshold can be set e.g. pperrbits@xmath20 to select sources with only very minor quality issues . in addition the seaming algorithm has been improved since dr1 . the seaming algorithm deals with sources that are detected in separate ( two or more ) multiframes that overlap . to decide which detection will be designated primary a sequence of discriminants is considered , until a preference is established : first the number of filters in which the source is detected , second the count from pperrbits , and finally the distance of the source from the detector edge . ukidss dr2 is crossmatched with sdss dr5 @xcite , updated from sdss dr3 for ukidss dr1 . note also that new crossmatching with catalogues from the millenium galaxy catalogue @xcite and the nrao vla sky survey @xcite is provided for dr2 . ccccccccc las & 282 & - & 341 & 505 & 477 & 476 & - & 685 + gcs & 57 & 80 & 81 & 83 & 116 & 478 & - & 496 + gps & 236 & - & - & 249 & 252 & 657 & 70 & 676 + total & 575 & 80 & 422 & 837 & 845 & 1611 & 70 & 1857 + cccc @xmath11 & - & 20.54 & - + @xmath10 & 20.23 & 20.16 & - + @xmath21 & 19.61 & 19.59 & 19.77 + @xmath22 & 18.90 & 18.87 & 19.00 + @xmath23 & 18.23 & 18.20 & 18.05 + for the shallow surveys , las , gcs , gps , figs [ coverage_dr2 ] and [ coverage_dr2 + ] plot the survey coverage in dr2 . fig . [ coverage_dr2 ] illustrates the sky coverage of fields where the filter set for the particular survey is complete , while fig . [ coverage_dr2 + ] shows all fields in which any data have been taken . the new las data are confined to the region @xmath24 the new observations in the northern block , near @xmath25 , are j - band only , and are first epoch observations for proper motions in fields that will later be reobserved in the full filter complement , after a minimum interval of two years ( d06 ) . details of the summed area covered , by filter , in each of the shallow surveys , las , gcs , and gps , are provided in table [ tab_shallow_coverage ] . for each survey , the table provides the area covered in a particular filter , the area covered by all filters , and the area covered by any filter . table [ tab_shallow_depth ] provides the median @xmath7 depth ( as defined in d06 ) achieved in each band in each of the two surveys . these values are slightly deeper than those for dr1 , provided in w07 . a quirk persists from dr1 ( see section 4.1 in w07 ) which results in a small number of las sources recorded in the archive as detected in _ yj _ only , coincident with distinct sources ( in reality the same sources ) recorded as detected in _ hk _ only . lccccccccccc xmm - lss & 1.00 & 36.5752020 & -4.7496444 & 6400 & 22.195 & 0.88 & 0.05 & 7040 & 20.845 & 0.76 & 0.08 + xmm - lss & 1.10 & 36.5803542 & -4.5294528 & 7680 & 22.265 & 0.86 & 0.06 & 8320 & 20.935 & 0.74 & 0.09 + xmm - lss & 1.01 & 36.8012000 & -4.7496444 & 7680 & 22.245 & 0.86 & 0.05 & 7680 & 20.885 & 0.77 & 0.08 + xmm - lss & 1.11 & 36.8012000 & -4.5294528 & 7040 & 22.275 & 0.86 & 0.05 & 5760 & 20.735 & 0.75 & 0.09 + xmm - lss & 2.00 & 35.7098417 & -4.7496444 & - & - & - & - & 6400 & 20.675 & 0.93 & 0.08 + xmm - lss & 2.10 & 35.7098417 & -4.5294528 & - & - & - & - & 7040 & 20.715 & 0.91 & 0.07 + xmm - lss & 2.01 & 35.9306875 & -4.7496444 & - & - & - & - & 7040 & 20.695 & 0.91 & 0.08 + xmm - lss & 2.11 & 35.9306875 & -4.5294528 & - & - & - & - & 6400 & 20.675 & 0.89 & 0.07 + xmm - lss & 3.00 & 35.7094541 & -3.8688833 & - & - & - & - & 640 & 19.325 & 1.17 & 0.11 + xmm - lss & 3.01 & 35.9300625 & -3.8688833 & - & - & - & - & 640 & 19.375 & 1.12 & 0.10 + lockman hole & 1.00 & 163.3623000 & 57.4753556 & 1280 & 21.565 & 1.06 & 0.10 & 12600 & 20.995 & 0.93 & 0.09 + lockman hole & 1.10 & 163.3623000 & 57.6955472 & 1280 & 21.575 & 1.06 & 0.08 & 13960 & 21.055 & 0.98 & 0.09 + lockman hole & 1.01 & 163.7756170 & 57.4753556 & 640 & 21.085 & 1.16 & 0.06 & 13820 & 21.025 & 0.99 & 0.08 + lockman hole & 1.11 & 163.7756170 & 57.6955472 & 1280 & 21.465 & 1.15 & 0.08 & 11820 & 20.965 & 0.98 & 0.08 + lockman hole & 6.00 & 161.6707620 & 59.1223000 & - & - & - & - & 8840 & 20.765 & 1.06 & 0.08 + lockman hole & 6.10 & 161.6707620 & 59.3424917 & - & - & - & - & 9840 & 20.855 & 1.05 & 0.08 + lockman hole & 6.01 & 162.1040380 & 59.1223000 & - & - & - & - & 9340 & 20.815 & 1.05 & 0.06 + lockman hole & 6.11 & 162.1040380 & 59.3424917 & - & - & - & - & 8700 & 20.745 & 1.06 & 0.08 + elais n1 & 1.00 & 242.5994000 & 54.5031333 & 6560 & 21.785 & 1.01 & 0.06 & 8000 & 20.675 & 0.99 & 0.08 + elais n1 & 1.10 & 242.5994000 & 54.7233250 & 6060 & 21.785 & 0.96 & 0.07 & 8500 & 20.795 & 0.98 & 0.08 + elais n1 & 1.01 & 242.9817370 & 54.5031333 & 8480 & 21.855 & 0.99 & 0.07 & 9000 & 20.845 & 0.98 & 0.09 + elais n1 & 1.11 & 242.9817370 & 54.7233250 & 9480 & 21.845 & 1.04 & 0.07 & 8000 & 20.725 & 0.97 & 0.09 + elais n1 & 2.00 & 241.1580080 & 54.5031333 & 8320 & 22.125 & 0.92 & 0.05 & 9400 & 20.815 & 0.89 & 0.08 + elais n1 & 2.10 & 241.1580080 & 54.7233250 & 7680 & 22.075 & 0.93 & 0.05 & 7620 & 20.595 & 0.91 & 0.09 + elais n1 & 2.01 & 241.5403460 & 54.5031333 & 8960 & 22.115 & 0.94 & 0.05 & 9260 & 20.785 & 0.89 & 0.08 + elais n1 & 2.11 & 241.5403460 & 54.7233250 & 6400 & 21.955 & 0.94 & 0.05 & 8400 & 20.715 & 0.87 & 0.08 + elais n1 & 3.00 & 241.1364170 & 55.3170222 & - & - & - & - & 500 & 19.165 & 1.08 & 0.05 + elais n1 & 3.10 & 241.1364170 & 55.5372139 & - & - & - & - & 500 & 19.215 & 1.09 & 0.05 + elais n1 & 3.01 & 241.5266580 & 55.3170222 & - & - & - & - & 500 & 19.275 & 0.96 & 0.09 + elais n1 & 3.11 & 241.5266580 & 55.5372139 & - & - & - & - & 500 & 19.215 & 0.95 & 0.12 + elais n1 & 4.00 & 242.6111380 & 55.3170222 & - & - & - & - & 500 & 19.405 & 0.83 & 0.09 + elais n1 & 4.10 & 242.6111380 & 55.5372139 & - & - & - & - & 500 & 19.415 & 0.81 & 0.11 + elais n1 & 4.01 & 243.0013830 & 55.3170222 & - & - & - & - & 500 & 19.405 & 0.86 & 0.09 + vimos 4 & 1.00 & 334.2667460 & 0.1698000 & 8820 & 22.175 & 0.89 & 0.04 & 10400 & 20.945 & 0.83 & 0.06 + vimos 4 & 1.10 & 334.2667460 & 0.3899917 & 10740 & 22.285 & 0.86 & 0.04 & 12820 & 21.085 & 0.79 & 0.06 + vimos 4 & 1.01 & 334.4869460 & 0.1698000 & 10100 & 22.355 & 0.84 & 0.04 & 10900 & 21.055 & 0.80 & 0.06 + vimos 4 & 1.11 & 334.4869460 & 0.3899917 & 6900 & 22.165 & 0.88 & 0.05 & 10260 & 20.975 & 0.85 & 0.06 + vimos 4 & 2.00 & 335.1420250 & 0.1817444 & 5120 & 21.855 & 0.92 & 0.04 & 8960 & 20.945 & 0.75 & 0.06 + vimos 4 & 2.10 & 335.1420250 & 0.4019361 & 4480 & 21.845 & 0.93 & 0.03 & 10240 & 20.965 & 0.74 & 0.06 + vimos 4 & 2.01 & 335.3622250 & 0.1817444 & 5120 & 21.845 & 1.02 & 0.04 & 10240 & 20.945 & 0.74 & 0.06 + vimos 4 & 2.11 & 335.3622250 & 0.4019361 & 5760 & 21.925 & 0.94 & 0.04 & 9600 & 20.885 & 0.75 & 0.06 + vimos 4 & 3.00 & 335.1420790 & 1.0559111 & 11520 & 22.325 & 0.85 & 0.04 & 10240 & 20.795 & 0.94 & 0.06 + vimos 4 & 3.10 & 335.1420790 & 1.2761028 & 11520 & 22.375 & 0.87 & 0.04 & 11520 & 20.905 & 0.98 & 0.05 + vimos 4 & 3.01 & 335.3623330 & 1.0559111 & 9600 & 22.315 & 0.88 & 0.04 & 11520 & 20.885 & 0.98 & 0.05 + vimos 4 & 3.11 & 335.3623330 & 1.2761028 & 11520 & 22.365 & 0.88 & 0.05 & 10880 & 20.895 & 0.93 & 0.06 + vimos 4 & 4.00 & 334.2668040 & 1.0559111 & 1920 & 21.535 & 0.98 & 0.04 & 10240 & 20.915 & 0.85 & 0.06 + vimos 4 & 4.10 & 334.2668040 & 1.2761028 & 2560 & 21.725 & 0.93 & 0.03 & 12160 & 20.975 & 0.87 & 0.05 + vimos 4 & 4.01 & 334.4870580 & 1.0559111 & 1920 & 21.515 & 1.00 & 0.04 & 14720 & 21.085 & 0.86 & 0.06 + vimos 4 & 4.11 & 334.4870580 & 1.2761028 & 1280 & 21.285 & 0.93 & 0.04 & 12800 & 21.035 & 0.87 & 0.06 + vimos 4 & 5.00 & 333.3940250 & 1.0559111 & - & - & - & - & 640 & 19.345 & 1.03 & 0.12 + vimos 4 & 5.10 & 333.3940250 & 1.2761028 & - & - & - & - & 640 & 19.235 & 1.15 & 0.13 + vimos 4 & 6.00 & 333.3939708 & 0.2064666 & - & - & - & - & 640 & 19.285 & 1.15 & 0.10 + vimos 4 & 6.10 & 333.3939708 & 0.4266583 & - & - & - & - & 640 & 19.385 & 1.06 & 0.05 + vimos 4 & 6.01 & 333.6141708 & 0.2064666 & - & - & - & - & 640 & 19.215 & 1.10 & 0.03 + vimos 4 & 6.11 & 333.6141708 & 0.4266583 & - & - & - & - & 640 & 19.365 & 1.05 & 0.03 the dxs is targeting the following four fields : xmm - lss ( 2@xmath26,@xmath27 ) , the lockman hole ( 10@xmath28,+57@xmath29 ) , elais n1 ( 16@xmath30,+54@xmath31 ) , and vimos 4 ( 22@xmath32,+0@xmath33 ) . the fields are plotted as small quadrilaterals in figs [ coverage_dr2 ] and [ coverage_dr2 + ] . eventually each field will be tessellated by 12 tiles , each of four pointings . the observing strategy @xcite aims to reach full depth in a tile , @xmath2 , @xmath34 , before moving to the next tile . depth is built up by a series of observations of length 640s ( 500s in 2005a ) , known as intermediate stacks . all the intermediate stacks for any pointing are combined to create _ multiframes ( d06 ) . all intermediate stack and _ deepleavstack _ multiframes are included in a release . in dr2 the numbers of _ deepleavstack _ multiframes are 55 in @xmath23 and 32 in @xmath21 . details of all the dxs _ deepleavstacks _ are provided in table [ tab_dxs_deep_stacks ] . the first column provides the field name , and the second is a code identifying the sub - field , in the form _ tile.xy_ , where _ xy _ are binary coordinates specifying one of four multiframe positions that make up a tile . columns three and four provide the coordinates of the base position , which is the centre of the field of view ( which is a point not imaged by the detectors ) . the remaining columns give the total integration time @xmath35 of the frames contributing to the stack ( i.e. excluding deprecated frames ) , the @xmath7 depth , the seeing , and the ellipticity , for the _ j _ and _ k _ frames .	the ukirt infrared deep sky survey ( ukidss ) is a set of five large near infrared surveys , covering a complementary range of areas , depths , and galactic latitudes . the ukidss second data release ( dr2 ) includes the first data release ( dr1 ) , with minor improvements , plus new data for the las , gps , gcs , and dxs , from observations made over 2006 may through july ( when the uds was unobservable ) . dr2 was staged in two parts . the first part excluded the gps , and took place on 2007 march 1 . the gps was released on 2007 april 12 . dr2 includes 282 deg@xmath0 of multicolour data to ( vega ) @xmath1 , complete in the full _ yjhk _ set for the las , 57 deg@xmath0 in the _ zyjhk _ set for the gcs , and 236 deg@xmath0 in the _ jhk _ set for the gps . dr2 includes nearly 7 deg@xmath0 of deep _ jk _ data ( dxs , uds ) to an average depth @xmath2 . in addition the release includes a comparable quantity of data where coverage of the filter set for any survey is incomplete . we document changes that have occurred since dr1 to the pipeline , calibration , and archive procedures . the two most noteworthy changes are presentation of the data in a single database ( compared to two previously ) , and provision of additional error flags for detected sources , flagging potentially spurious artifacts , corrupted data and suspected cross - talk sources . we summarise the contents of each of the surveys in terms of filters , areas , and depths .
the causal character of a singularity has a well defined meaning within the theory of conformal boundaries @xcite . the knowledge of this causal character is fundamental since whenever the spacetime possesses timelike or past null singularities there are always null geodesics which are past incomplete . if such a singularity could develop from a generic gravitational collapse in the framework of general relativity theory , this would mean that the theory would lose its predictability . the question on whether general relativity contains a built - in safety feature that precludes this possibility was put forward by penrose in 1969 @xcite and gave rise to what is known as the _ cosmic censorship conjecture _ many counterexamples to the ccc have been proposed as well as many arguments in its favour ( see , for example , @xcite and references therein ) so that the question of the ccc remains open . in this article we will deal with probably the most interesting type of singularities in spherically symmetric spacetimes : the _ zero - areal - radius singularities _ , i.e. , given the _ areal radius _ @xmath1 defined such that the area of a 2-sphere is @xmath2 , we will be interested in singularities _ at _ @xmath0 . in the current literature the study of the causal character of the @xmath0 singularities has been carried out for important particular solutions . in a few simple cases the singular conformal boundary has been obtained by using a conformal compactification ( see , for instance , ) , while in most cases there is not an analytical compactification and , as an alternative method , the causal character of the singularities has been studied through the analysis of radial null geodesics around them ( see , for example , ) . in addition to the analysis of particular cases , this last technique allows some _ general _ approaches for studying zero - areal - radius singularities . in particular , it has led to show @xcite that the causal character of the singularity is related to the _ mass function _ . more general studies on the formation of _ naked _ singularities in spherically symmetric spacetimes along these lines can be found in @xcite , @xcite and , by using ad hoc devised procedure , in @xcite . furthermore , in @xcite , by using the techniques of the qualitative behaviour of dynamic systems on the differential equations satisfied by the radial null geodesics we were able to present the most comprehensive scheme so far to try to find out their causal characterization taking into account , and analyzing , the possible limitations of the approach . however , this work was carried out in specific coordinates ( the so called _ radiative coordinates _ ) , so that its results were restricted to models described in these coordinates . our aim in this article is to show that the causal character of the zero - areal - radius ( @xmath0 ) singularity in spherically symmetric models is related with some specific invariants . apart from being an interesting result from a theoretical point of view , this coordinate independent approach means that , if some assumptions are satisfied , one could find out the causal character of a model s singularity algorithmically through the computation of these invariants in arbitrary coordinates . in order to try to reach our goal we will base our approach in an analysis of the results in our previous article @xcite . we will show that our previous results admit an interpretation and rewriting in terms of some invariants and we will analyze and explicitly state the limits for the applicability of our results coming from our specific approach . on the other hand , throughout the article we will use a geometrical approach requiring only the existence of a spacetime , but not the fulfillment of einstein s equations . thus , we just try to discover the possibilities allowed by this geometrical approach which includes the classical as well as the semiclassical framework . the paper has been divided as follows : in section [ basis ] we revise well - known properties of spherically symmetric spacetimes and of the radial null geodesics , but emphasizing the corresponding degrees of differentiability for each defined object ( what will be an important aspect for the later development of the work ) . in section [ coordchange ] we analyze the relationship between general coordinate systems and the coordinate system used in @xcite . the different cases that appear when treating the causal character of singularities are treated from section [ m_neq0 ] on . in particular , section [ m_neq0 ] is devoted to the analysis of _ singularities with non - null mass function_. section [ sec_m=0 ] deals with the preliminaries required to the study of _ singularities with null mass function_. finally , sections [ sechyper ] , [ secnonhyper ] and [ secnoniso ] analyze every _ null mass function _ subcase in detail . let us consider a simply connected open set @xmath3 in a spherically symmetric spacetime and such that a part of its boundary consists of a @xmath0 interval . the metric line element of an oriented spherically symmetric spacetime can be ( and for practical purposes it is usually ) written in the local chart endowed with coordinates @xmath4 in the form @xmath5 where @xmath6 is an oriented two - dimensional lorentzian metric ( @xmath7=@xmath8 ) , @xmath9 and . in the lorentzian two - surface orthogonal to the 2-spheres two nonvanishing null vector fields may be defined such that they are linearly independent at each point . if some differentiability requirements are satisfied in @xmath3 the integral curves of the two null vector fields provide us with two families ( @xmath10 and @xmath11 ) of affinely parametrized null geodesics called the _ radial null geodesics_. take , for example , the geodesics belonging to the @xmath10 family satisfying @xmath12 where @xmath13 is their affine parameter . the theory of ordinary differential equations @xcite together with the definition of the christoffel symbols @xmath14 guarantees the existence and uniqueness of the affinelly parametrized geodesics provided that @xmath15 is at least @xmath16 and @xmath17 . from now on we will guarantee the existence and uniqueness of affinely parametrized null geodesics by assuming that @xmath15 is @xmath18 , where @xmath19 means @xmath20 and that the @xmath21th - order derivatives are locally lipshitz . on the other hand , @xmath18 with @xmath22 means @xmath23 . ] , and that and @xmath17 . under these assumptions the theory of odes also guarantees that the solution of the geodesic equation will be @xmath24 . since the radial null tangent vector field to the @xmath10 family @xmath25 has an associated covector satisfying @xmath26}=0 $ ] , then it can be written as the differential of a @xmath24 function @xmath27 : @xmath28 . the curves @xmath29constant define the trajectories of the @xmath10 family of null geodesics in @xmath3 . taking into account that the scalar invariant @xmath30 is a @xmath18 function we can define @xmath31 which is a scalar invariant @xmath32 function . this invariant is related to the @xmath32 invariant _ mass function _ through @xmath33 in order to investigate the causal characterization of a @xmath0 interval we will consider the radial null geodesics around a point @xmath34 in this interval . as we shown in @xcite this procedure requires that the radial null geodesics from at least one family , say @xmath10 , _ reach _ ( or _ leave _ ) every point in the interval , what will be assumed in the next sections of this article . on the other hand , let us comment that provided that a @xmath0 interval is not _ reached _ ( or _ left _ ) by radial null geodesics of any family then the causal characterization of the interval is straightforward since this interval can not be translated into a piecewise @xmath35 interval in the conformal boundary of the spacetime @xcite . it can only be translated into a _ point _ where the boundary is not a @xmath35 curve and , thus , where there is not tangent vector properly defining its causal character . as we mentioned in the introduction , the main goal of this article is to extend the coordinate dependent results presented in @xcite so that they can be used independently of the coordinate system chosen to work with , i.e. , to provide the causal characterization of @xmath0-singularities in a invariant manner . in order to do this , in this section we will deal with the connection between a general coordinate system of the type used for ( [ mi ] ) and the coordinate system used in @xcite . [ coocha ] under the assumptions that the metric ( [ mi ] ) is @xmath18 , with @xmath19 , and @xmath36 ( or , equivalently , @xmath37 ) there exists a coordinate change @xmath38 , where @xmath1 is the areal coordinate and @xmath39 is a null coordinate , such that the metric ( [ mi ] ) can be locally written as @xmath40 where @xmath41 and @xmath42 , @xmath43 are @xmath32 functions . this lemma is based in the fact that , since @xmath39 is @xmath24 and @xmath1 is @xmath18 , there is a class @xmath18 map @xmath44 . the inverse function theorem ( see , for example , @xcite ) guarantees the existence of @xmath18 functions @xmath45 and @xmath46 such that @xmath47 and @xmath48 provided that the jacobian determinant @xmath49 is not null in @xmath3 . along this work we will denote the open set @xmath50 by @xmath51 . it follows that we can write the function @xmath52 as @xmath53 , where the chain rule guarantees that @xmath54 will also be at least a @xmath32 function in the variables @xmath55 . on the other hand , the condition @xmath56 or , equivalently , @xmath37 implies that the vectors associated with these one - forms can not be parallel : @xmath57 is @xmath58 . in this way , taking into account that @xmath59 will be @xmath32 ( @xmath19 ) in @xmath3 we can define the invariant non - null constant @xmath60 note that it is invariant under future directed reparametrizations of @xmath61 . if @xmath62 ( or @xmath63 ) , the expansion of the null geodesics with tangent vector @xmath61 is positive ( negative , respectively ) in every point of @xmath51 . if we perform the coordinate change @xmath38 then , due to the light - like character of the coordinate @xmath39 , the metric of the spacetime ( [ mi ] ) will take the form : @xmath64 where @xmath65 and @xmath66 . the general future directed and affinely parametrized null vector @xmath61 and the future directed null vector @xmath67 tangent to the @xmath68 family satisfying @xmath69 can be written as @xmath70 where @xmath71 depends on the affine parameter chosen for @xmath61 . clearly , sign@xmath72sign@xmath73 , so that we can rewrite @xmath73 as @xmath74 with @xmath42 . if we state explicitly the relationship between the component @xmath75 of the metric tensor in these new coordinates with regard to the old ones @xmath76 where , as usual , @xmath6 should be understood as @xmath77 , and taking into account that both @xmath78 and @xmath79 are @xmath18 functions in their respective variables , then the chain rule theorem implies that @xmath80 is at least a @xmath32 function . on the other hand , if one evaluates the function @xmath52 with this form of the metric and uses ( [ bbeta ] ) one finds @xmath81 thus , taking into account ( [ bbeta ] ) and ( [ achi]),we obtain the required form for the metric ( [ metrad ] ) with the degree of differentiability stated in the lemma . [ corm ] under the assumptions in lemma [ coocha ] the mass function @xmath82 is a @xmath32 function . let us remark here that we do not need to find the explicit form ( [ metrad ] ) since we want to work in the original coordinates @xmath83 . in particular , the function @xmath84 can be written with the help of ( [ beta ] ) , and the appropriate labeling"@xmath85 . ] , as a function of @xmath86 : @xmath87 in this section we will discuss the causal character of @xmath0 in a point @xmath89 such that @xmath88 , where @xmath90 . this condition implies that there is a scalar curvature singularity at @xmath34 @xcite , so that @xmath34 does not belong to the spacetime , but to its singular boundary . if one radial null geodesic of , say , the @xmath10 family _ reaches _ @xmath34 then we have the following [ t_m_neq0 ] in case the spacetime metric ( [ mi ] ) is @xmath16 and there is a radial null geodesic reaching ( either toward its past or its future ) a @xmath0-singularity at @xmath34 with a value of its affine parameter @xmath91 then : * if @xmath92 , there is a spacelike singularity at @xmath34 , * if @xmath93 , there is a timelike singularity at @xmath34 . this result can be found in our previous article @xcite where it was shown in specific coordinates . however , taking into account that @xmath94 is a scalar invariant , the proposition is true for other coordinate systems and it is , thus , an invariant result . let us reiterate that the requirement of , at least , a @xmath16 metric is a minimum assumption for the existence and uniqueness of radial null geodesics in the spacetime . the theorem only requires the existence of the _ directional _ limit along the radial null geodesic . as a corollary we have the following more applicable result : [ c_m_neq0 ] in case the spacetime metric is @xmath16 and @xmath88 , where @xmath34 is such that @xmath95 , then the causal character of the @xmath0-singularity at @xmath34 is defined by the sign of the invariant mass function as the function approaches @xmath34 : * if @xmath96 then there is a spacelike singularity at @xmath34 , * if @xmath97 then there is a timelike singularity at @xmath34 this is the most involved case . it is known that if @xmath98 the causal character of @xmath0 at @xmath34 admits any possibility : spacelike , lightlike or timelike @xcite . therefore , the question is : are in this case other invariants which define the causal character of @xmath0 independently of the coordinate system used ? in order to answer this , let us introduce in this section some new definitions and lemmas . for instance , in spherical symmetry there is an invariantly defined vector , the kodama vector @xcite , which is also known to possess very interesting properties ( see , for instance , @xcite and references therein ) : @xmath99 where @xmath100 denotes the volume form associated with the two - metric @xmath6 . it satisfies @xmath101 , so that if @xmath102 the orientation of @xmath100 can be chosen in such a way that @xmath103 is a future directed timelike vector . kodama s vector characterizes the spherically symmetric directions tangent to the hypersurfaces @xmath104constant and provides and invariantly defined direction in which the area of the two - spheres remains constant @xcite . for the metric ( [ metrad ] ) @xmath103 takes the form : @xmath105 which satisfies @xmath106 in the local chart . we have seen that , if the assumptions in lemma [ coocha ] are satisfied , the functions @xmath107 and @xmath80 are @xmath32 functions defined in an open set @xmath51 . we will now see that these functions can be extended beyond the open set provided that some conditions are fulfilled . in particular , we are mainly interested in an extension of the functions around @xmath0 that , while it has not any physical meaning , will allow us to apply the theory of the qualitative behaviour of dynamic systems in a open set centered in a point of @xmath0 @xcite . [ extens ] provided that there is a natural number @xmath108 such that the limits of the functions @xmath94 and @xmath84 and of their ith - order derivatives , for all @xmath109 , as every point in the boundary of the open set is approached , exist ( and are finite ) then the functions @xmath94 and @xmath84 admit a @xmath110 extension @xmath111 and @xmath112 . in order to show this it suffices to define the extended function and its derivatives for all points @xmath113 in the boundary of the open set @xmath51 as @xmath114 for all integers @xmath115 , @xmath116 such that @xmath117 . in this way the lemma can be considered as a simple case of whitney s extension theorem @xcite and the existence of @xmath110 extensions @xmath111 and @xmath112 that coincide with @xmath94 and @xmath84 in @xmath51 is guaranteed . we will denote the extended domain of definition by @xmath118 . on the other hand , we want to work with the original coordinates @xmath119 and infer from here the extendibility of the functions @xmath107 and @xmath80 . we can do this by using the chain rule applied to the derivatives of @xmath120 and @xmath121 as the _ boundary points _ are approached . for example , if one is looking for a @xmath110 extension , according to lemma [ extens ] one needs , among others , the limit of @xmath122 in the boundary , but @xmath123 taking into account that @xmath124 can be written as a function of @xmath125 ( @xmath126 ) and @xmath127 ( @xmath128 ) , the existence of the limit is guaranteed if we require that the limits of @xmath129 , @xmath130 , and @xmath131 as every point in the boundary of the open set is approached exist ( and are finite ) and we also require that the limit of @xmath49 when approaching the same points is not zero . this can be formalized and generalized for @xmath110 extensions similarly : [ extenscn ] if the limits @xmath132 as every point in the boundary of the open set is approached exist ( and are finite ) for all integers @xmath115 , @xmath116 such that @xmath117 and the limit of @xmath49 as the same points are approached exists and is not zero , then the functions @xmath107 and @xmath80 admit a @xmath110 extension @xmath133 and @xmath134 . in order to analyze the singular conformal boundary of a spherically symmetric spacetime it is possible to perform just the conformal compactification of the two - dimensional surface orthogonal to the _ 2-spheres _ retaining all the important information . this is so because , by means of a coordinate change , the induced lorentzian metric or _ first fundamental form _ of the two - dimensional surface can be brought into a conformally flat form @xmath135 or , equivalently , @xmath136 where @xmath137 are lightlike coordinates ( @xmath138 , @xmath139 ) . in this way , it can be naturally embedded in an _ unphysical _ two - dimensional minkowskian spacetime ( see , for instance , @xcite ) . despite an _ conformal compactification can only be found for certain simple particular cases some information can be extracted without fully following the procedure . for instance , with regard to the differentiability of the singular boundary we have the following [ teoc3 ] if @xmath6 and @xmath1 are @xmath140 functions , @xmath141 in @xmath3 and the assumptions in lemma [ extenscn ] ( for the existence of @xmath142 extensions ) are satisfied and , in particular , they are satisfied in a connected open interval @xmath143 of the singular boundary @xmath0 in which @xmath144 , then @xmath143 will be @xmath140 in the unphysical spacetime . in order to show this we will study the differentiability of @xmath0 in the unphysical spacetime by analyzing the coordinate change which takes the metric to the form ( [ dobnul ] ) . to begin with , note that , under the assumptions in the theorem we can perform a first coordinate change from coordinates @xmath86 to @xmath55 such that @xmath107 and @xmath80 will be @xmath145 functions . that this is satisfied follows from the requirement of a @xmath140 metric in the theorem and lemma [ coocha ] together with corollary [ corm ] . furthermore , requiring the fulfillment of the assumptions in lemma [ extenscn ] implies that there will be @xmath142 extensions @xmath133 and @xmath134 . we know from ( [ metrad ] ) that the metric of the lorentzian surface can be written in coordinates @xmath55 as @xmath146 where @xmath147 . in order to rewrite this in double - null coordinates @xmath137 we will look for an integrating factor @xmath148 such that @xmath149 the integrability condition for ( [ dv ] ) takes the form of the following first - order linear inhomogeneous pde in @xmath118 : @xmath150 where the functions @xmath151 , @xmath152 and @xmath153 are defined as @xmath154 our assumptions imply that @xmath151 and @xmath152 are @xmath145 functions , while @xmath153 is @xmath155 . we now choose the curve @xmath143 ( which can be described as @xmath156 , where @xmath157 is a parameter defined in a connected open real interval ) as the curve of initial conditions for the pde and , in addition , we choose the initial condition on @xmath143 to be @xmath158 , where @xmath159 is an arbitrary @xmath142 function . then the tangent vector to @xmath143 ( @xmath160 ) and the characteristic direction have distinct projections on the @xmath39,@xmath1-plane ( @xmath161 in @xmath162 ) . this together with the fact that the pde is nondegenerate ( @xmath163 in an open neighborhood of @xmath143 since @xmath164 ) implies that the initial value problem has one and only one solution which will be a @xmath145 function @xmath165 . in this way ( [ dv ] ) can now be used to evaluate the slope of the singular boundary @xmath166 in the unphysical spacetime : @xmath167 since there are only @xmath145 functions in the right hand side of this differential equation , its solution @xmath168 will be a @xmath140 curve in the @xmath39,@xmath169-plane . qed as a corollary , under these assumptions , @xmath143 admits a tangent vector whose causal character determines the causal character of @xmath0 . in other words , if @xmath170 the slope of the curve is negative so that the curve in the unphysical spacetime ( therefore , the singularity ) will be spacelike . likewise , if @xmath171 the singularity will be timelike . this just reiterates the results in theorem [ t_m_neq0 ] , which were then shown under less restrictive assumptions . the so far untreated case @xmath172 in which the singularity can be timelike , lightlike or spacelike and in which the degree of differentiability of the singular boundary can be just @xmath173 under the differentiability assumptions on this subsection will be the subject of the rest of the article . [ m0_mu ] in case @xmath98 and @xmath174 , where @xmath34 is a point in the singular boundary @xmath0 reached or left by a single null geodesic of the @xmath10 family , the causal character of @xmath0 around @xmath34 can be obtained through the invariants @xmath175 , @xmath176 provided that @xmath6 and @xmath1 are @xmath177 functions in @xmath3 and that the assumptions in lemma [ extenscn ] ( for the existence of @xmath35 extensions ) are satisfied . then the causal characterization around @xmath34 is inferred from @xmath175 , @xmath178 according to figure [ hyper ] . _ sketched _ characterization of the @xmath0 singularity when @xmath98 and @xmath179 , where the @xmath0singularity at @xmath34 is chosen to be reached by the radial null geodesic @xmath180 . as explained in the text , @xmath181 , @xmath182 and @xmath183 . in this case we have @xmath184 for @xmath185 . in this way , if the requirements in theorem [ teoc3 ] are satisfied the singular boundary should be @xmath186 for @xmath185 . ( in these sketches we just draw straight lines for @xmath185 instead of curves ) . however , in @xmath180 , where @xmath187 the singular boundary can be just @xmath173 . consider , for example , the first sketch in the @xmath62 column where a @xmath0-_spacelike _ singularity for @xmath188 must be abruptly followed by a lightlike singularity for @xmath180 . ] note that the invariant @xmath189 is independent of the parametrization of @xmath190 . in @xcite 5 ) the causal characterization in this case was obtained in radiative coordinates provided that @xmath107 and @xmath80 were @xmath16 functions . that this requirement is satisfied follows from the requirement of a @xmath177 metric in the theorem and lemma [ coocha ] together with corollary [ corm ] . another requirement was that there should be @xmath35 extensions @xmath133 and @xmath134 , hence our requirement that the assumptions on lemma [ extenscn ] for the existence of @xmath35 extensions should be satisfied . on the other hand , it was shown that , in radiative coordinates , the causal characterization depends on @xmath191 where we have chosen to analyze the @xmath0-singularity around @xmath180 which describes the null geodesic reaching or leaving @xmath34 ( what is always allowed through a redefinition of @xmath192contant ) . on the other hand , using the expression ( [ lk ] ) for @xmath190 and the kodama vector ( [ kodrad ] ) it is easy to verify that , in radiative coordinates , the invariant @xmath193 can be written as @xmath194 while the invariant @xmath189 is simply @xmath195 . therefore , taking the limit @xmath196 , we can rewrite the quantities @xmath197 , @xmath198 and @xmath199 in an _ explicit invariant form _ as @xmath200 finally , in @xcite it was shown that the causal character can be read from a table ( figure [ hyper ] ) depending on these quantities . in this way , one first computes @xmath94 , @xmath103 and @xmath190 in arbitrary coordinates in order to get from them the invariants @xmath175 , @xmath201 and @xmath202 . then , if the assumptions in the theorem are satisfied , these invariants provide us with the causal characterization of the singularity for this case . [ m0_mu_mr ] in case @xmath98 , @xmath204 and @xmath205 , where @xmath34 is a point in the singular boundary @xmath0 reached or left by a single null geodesic of the @xmath10 family , the causal character of @xmath0 is given by the invariants @xmath175 , @xmath206 provided that @xmath6 and @xmath1 are @xmath18 functions in @xmath3 ( @xmath207 ) and that the assumptions in lemma [ extenscn ] ( for the existence of @xmath32 extensions ) are satisfied . then @xmath175 , @xmath208 where @xmath209 . ] whereas @xmath210 for @xmath211 ( i.e. , we assume that there is a finite @xmath212 such that it is the lowest value satisfying @xmath209 ) , provide us with the causal characterization of the singular boundary around @xmath34 according to figure [ semihyper ] . characterization of the singularity when @xmath98 , @xmath204 and @xmath205 . ] in @xcite ( sec . 6 ) the causal characterization in this case was obtained in radiative coordinates provided that @xmath107 and @xmath80 were @xmath32 ( @xmath207 ) functions admitting @xmath32 extensions . that this is satisfied follows , as in the previous theorem [ m0_mu ] , first , from the requirement of a @xmath18 metric in this theorem and lemma [ coocha ] together with corollary [ corm ] and , second , from the requirement that the assumptions in lemma [ extenscn ] for the existence of @xmath32 extensions @xmath133 and @xmath134 are satisfied . on the other hand , in @xcite it was shown that , in radiative coordinates , the causal characterization in this case depends on provides just an extra factor exp(@xmath213 to @xmath214 which does not affect ( [ deln ] ) . ] @xmath215 where @xmath216 for @xmath211 ( @xmath207 ) . but again these quantities are invariant . the second quantity ( [ mr ] ) has already been shown to correspond with @xmath202 . with regard to the first one , let us consider the simplest case @xmath217 . in radiative coordinates the invariant @xmath218 is @xmath219.\ ] ] since @xmath220 , in the limit as @xmath221 we will have @xmath222 likewise , in case @xmath223 for @xmath211 and @xmath224 , we would have for @xmath207 @xmath225 therefore we can write the two quantities ( [ deln ] ) and ( [ mr ] ) in an explicit invariant form @xmath226 since these quantities define the causal character of @xmath0 in this case according to figure [ semihyper ] @xcite , then now the causal character has been determined invariantly . [ m0_mu_mr0 ] in case @xmath98 , @xmath204 and @xmath227 , where @xmath34 is a point in the singular boundary @xmath0 reached or left by a single null geodesic of the @xmath10 family , the causal character of @xmath0 is given by the invariants @xmath175 , @xmath228 provided that @xmath6 and @xmath1 are @xmath18 functions and that the assumptions in lemma [ extenscn ] ( for the existence of @xmath32 extensions ) are satisfied . then the causal characterization can be found according to figures [ nilpotentodd ] and [ nilpotenteven ] through the computation of @xmath229 where @xmath230 whereas @xmath231 for @xmath232 , @xmath233 whereas @xmath234 for @xmath235 and @xmath236max@xmath237 is supposed to exist ant to be finite . characterization of the singularity when @xmath98 , @xmath204 and @xmath227 and @xmath238 is odd . in the _ extra - conditions _ we use @xmath239 and @xmath240 from ( [ lambda ] ) . ] characterization of the singularity when @xmath98 , @xmath204 and @xmath227 and @xmath238 is even . in the _ extra - conditions _ we use @xmath241 . ] in @xcite ( sec . 6 ) the causal characterization in this case was obtained in radiative coordinates provided that @xmath107 and @xmath80 were @xmath32 ( @xmath207 , @xmath242max@xmath237 , where @xmath238 , @xmath243 are defined in ( [ sak ] ) and ( [ sbn ] ) , respectively ) functions admitting @xmath32 extensions . that this is satisfied follows , first , from the requirement of a @xmath18 metric in this theorem and lemma [ coocha ] together with corollary [ corm ] and , second , from the requirement that the assumptions in lemma [ extenscn ] for the existence of @xmath32 extensions @xmath133 and @xmath134 are satisfied . according to @xcite the causal characterization in this case and for radiative coordinates depends on sign(@xmath244 ) , sign(@xmath245 ) and @xmath246 , where ) appear since in this work we do not choose @xmath247 . it has to be taken into account that if one now wishes to obtain @xmath244 and @xmath245 as in @xcite there is a corresponding slight modification in the change of variables used there for this _ nilpotent case _ : @xmath248exp(@xmath249 . ] @xmath250 with @xmath251 for @xmath232 and @xmath252 for @xmath235 . but again these quantities are invariant . the first quantity sign(@xmath244 ) has already been treated in the previous theorem where we showed that it corresponds to sign@xmath253 . with regard to the second one sign(@xmath245 ) , let us consider the simplest case @xmath254 . in radiative coordinates the invariant @xmath255 can be written as @xmath256 which in the limit as @xmath221 provide us with @xmath257 which is clearly related to @xmath258 ( [ sbn ] ) . if @xmath259 we should consider @xmath260,\ ] ] which in the limit as @xmath221 provide us with @xmath261 likewise , for @xmath262 we will have @xmath263 to be compared with ( [ sbn ] ) . in this way we can write the two quantities ( [ sak ] ) and ( [ sbn ] ) and , thus , their signs in an explicit invariant form @xmath264 finally , taking into account its definition , @xmath240 can also be written in an explicit invariant form @xmath265 since these quantities define the causal character of @xmath0 in this case according to figures [ nilpotentodd ] and [ nilpotenteven ] , then now the causal character has been determined invariantly for this case . [ noniso ] if @xmath98 @xmath267 in a connected open interval @xmath162 of @xmath0 then the causal character of @xmath34 is determined by the invariant @xmath268 provided that @xmath6 and @xmath1 are @xmath177 functions in @xmath3 , that @xmath94 and @xmath84 are such that they admit @xmath35 extensions according to lemma [ extens ] and that the limit @xmath269 exists ( and is finite ) for all @xmath270 . then the causal characterization around @xmath34 is inferred from * if @xmath271 then @xmath0 is timelike in this interval . ( this case includes both the possibility of a singular or a _ regular _ interval @xmath143 ) . * if @xmath272 then @xmath0 is spacelike in this interval . * if @xmath273 then @xmath0 is lightlike in this interval . in @xcite ( sec . 7 ) the causal characterization in this case was obtained in radiative coordinates provided that @xmath107 and @xmath80 were @xmath16 functions admitting @xmath35 extensions @xmath133 and @xmath134 . that these requirements are satisfied thanks to the assumptions in the theorem has already been shown for theorem [ m0_mu ] . on the other hand , in @xcite it was shown that , in radiative coordinates , the causal characterization depends on @xmath195 that we have shown can be written in an explicit invariant form as @xmath189 . in this way , the relationship between its value and the causal characterization follows directly from the results in @xcite ( sec . in this article we have shown that , provided some assumptions are satisfied , the causal character of the @xmath0singularities in spherically symmetric spacetimes depends on some specific invariants . this allows to deduce the causal character of the singularity algorithmically . basically , one starts with the knowledge of the areal radius @xmath1 , the mass function @xmath94 , a tangent vector field to the radial null geodesics @xmath61 and the kodama vector field @xmath103 . from here one should compute the invariants @xmath274 where @xmath275 and the exact last value to be computed is determined by the values of the lowest order invariants as the singularity is approached ( @xmath276 and @xmath277 , respectively ) in the following manner : * if @xmath278 this value suffices to characterize the singularity according to theorem [ t_m_neq0 ] provided the spacetime is at least @xmath16 ( i.e. , if the existence and uniqueness of radial null geodesics is guaranteed ) . * if @xmath279 in an isolated point @xmath34 _ in _ @xmath0 then different cases appear : * * if @xmath280 then we will also need @xmath202 . the causal characterization around the singular point can be inferred from figure [ hyper ] , if the assumptions in theorem [ m0_mu ] are satisfied . * * if @xmath281 and @xmath282 then one needs to compute @xmath276 , where we demand the existence of a finite @xmath283 such that it is the lowest value satisfying @xmath230 . the causal characterization around the singular point can then be inferred from figure [ semihyper ] , if the assumptions in theorem [ m0_mu_mr ] are satisfied . * * if @xmath281 and @xmath284 then one needs to compute @xmath276 , where we demand the existence of a finite @xmath283 such that it is the lowest value satisfying @xmath230 , and to compute @xmath277 , where we demand the existence of a finite @xmath243 such that it is the lowest value satisfying @xmath233 . the causal characterization around the singular point can be inferred from figures [ nilpotentodd ] and [ nilpotenteven ] , if the assumptions in theorem [ m0_mu_mr0 ] are satisfied . * if @xmath285 in an open interval of @xmath0 then the causal characterization of the singular interval can be deduced from sign(@xmath286 ) according to theorem [ noniso ] , if the assumptions in the theorem are satisfied . note that for every case some assumptions must be satisfied . these assumptions come mainly from the fact that our results are based on the use , in @xcite , of the qualitative theory of dynamic systems to the differential equations satisfied by the radial null geodesics . the application of the appropriate theorems to the analysis of these differential equations requires some degree of differentiability for the functions @xmath287 and @xmath288 . ( more details on this issue can be found in @xcite ) . likewise , the reader can consult @xcite for some applications of this technique to the study of the generation of naked singularities or black hole evaporation . we would like to thank j.m.m . senovilla and conan wu for helpful discussions . we would also like to acknowledge the _ generalitat de catalunya _ ( grant 2009sgr-00417 ) for financial support . 99 penrose r 1963 _ phys . rev . lett . _ * 10 * 66 garca - parrado a and senovilla j m m 2005 _ class . quantum grav . _ * 22 * r1 fayos f and torres r 2011 _ class . quantum grav . _ * 28 * 215023 penrose r 1969 _ riv . nuovo cimento _ * 1 * 252 penrose r 1999 _ j. astrophys . astr . _ * 20 * 233 hawking s w and ellis g f r 1973 _ the large scale structure of space - time . _ cambridge : cambridge university press volovich i v , zagrebnov v a and frolov v p 1976 _ teoret fiz . _ * 29 * 191 hiscock w a , williams l g and eardley d m 1982 _ phys . d _ * 26 * 751 kuroda y 1984 _ prog . theor . phys . _ * 72 * 63 eardley d m and smarr l 1979 _ phys . rev . * 19 * 2239 christodoulou d 1984 _ commun . phys . _ * 93 * 171 ori a and piran t 1990 _ phys . d _ * 42 * 1068 hayward s a 1996 _ phys . * 53 * 1938 misner c w and sharp d h 1964 _ phys . b _ * 136 * 571 lake k 1992 _ phys . lett . _ * 68 * 3129 singh t p 1999 _ class . quantum grav . _ * 16 * 3307 giambo r , giannoni f , magli g and piccione p 2003 _ class . quantum grav . _ * 20 * l75 plebaski j and krasiski a 2006 _ an introduction to general relativity and cosmology _ cambridge : cambridge university press arnold v i 1992 _ ordinary differential equations _ berlin heidelberg : springer - verlag hernndez w c and misner c w 1966 _ astrophys . j. _ * 143 * 452 cahill m e and mcvittie g c 1970 _ j. math . phys . _ * 11 * 1382 zannias t 1990 _ phys . rev . _ * d41 * 3252 boothby w m 1986 _ an introduction to differentiable manifolds and riemannian geometry _ florida : academic press inc . kodama h 1980 _ prog . * 63 * 1217 abreu g and visser m 2010 _ phys . d _ * 82 * 044027 bengtsson i and senovilla jmm 2011 _ phys . d _ * 83 * 044012 whitney h 1934 _ trans . * 36 * 63 courant r and hilbert d 1989 _ methods of mathematical physics _ new york : john wiley & sons	the causal character of singularities is often studied in relation to the existence of naked singularities and the subsequent possible violation of the cosmic censorship conjecture . generally one constructs a model in the framework of general relativity described in some specific coordinates and finds an _ ad hoc _ procedure to analyze the character of the singularity . in this article we show that the causal character of the zero - areal - radius ( @xmath0 ) singularity in spherically symmetric models is related with some specific invariants . in this way , if some assumptions are satisfied , one can ascertain the causal character of the singularity algorithmically through the computation of these invariants and , therefore , independently of the coordinates used in the model .
using long - baseline interferometers in both the mid - ir and near - ir , we are probing the innermost dusty structure in the putative torus in active galactic nuclei ( agns ) at unprecedented spatial resolutions in the ir . one of the first goals of our study has been to directly map the _ radial distribution _ of the accreting material . we currently focus on type 1 agns , where our line of sight is thought to be more or less face - on and the innermost region can directly be studied without significant effects from inclination and obscuration . we aim to summarize below the status of our systematic type 1 study in both the mid - ir and near - ir . let us start with addressing a simple , but naive , luminosity scaling that should be carefully considered in our study . the mid- and near - ir radiation in agns is generally thought to be the thermal emission from dust grains that are radiatively heated by the uv / optical emission of the central engine . effectively , we are using these dust grains to trace the inner structure . when the temperature of the observed dust grains is simply determined by the central engine s _ direct _ illumination and the dust thermal re - emission , the radial distance of the dust grains at a given temperature will be proportional to @xmath3 . here we assume that the incident spectral shape does not change with the uv luminosity @xmath1 , which we discuss more below . then our observations at a fixed wavelength would find the size of the structure to also scale with @xmath3 , since they would be sensing the dust grains of the same temperature range . this simple @xmath3 scaling may not realize due to various quantities and properties potentially changing with luminosity . however , at least for the innermost dust sublimation region , this seems to hold the dust sublimation radius @xmath2 as probed by the near - ir reverberation measurements is known to be proportional to @xmath3 @xcite . here the relation is found for optical luminosity , rather than the more relevant uv luminosity , implying the same uv optical spectral shape over the range of luminosities at least approximately . therefore , we assume a generic spectral shape and define the uv luminosity @xmath1 as a scaled optical luminosity . this innermost @xmath3 scaling is in fact supported also by the results from our near - ir interferometry , although there is a distinction between our interferometric size measurements and the reverberation measurements , as we discuss in sect.[sect - nir ] . based on this empirical relationship , which matches the naive expectation , it would be physically insightful if we normalize the radial extent in our study by the dust sublimation radius @xmath2 given by the @xmath3 fit to the reverberation measurements . this will remove the simple @xmath3 scaling and let us investigate what is left behind it . each interferometric measurement gives a visibility at a certain spatial frequency , or its reciprocal , spatial wavelength , which roughly corresponds to the spatial scale probed by that particular configuration of the interferometer ( see more in sect.[sect - mir ] ) . we normalize this probed spatial scale by @xmath2 for each observed object ( in this case , spatial frequency will have the units of cycles per @xmath2 ) . differences in distance , as well as in luminosity , will be folded into the different angular sizes of @xmath2 , where visibility curves as a function of the spatial wavelength in units of @xmath2 for different objects can be compared uniformly @xcite . the normalization gives a physical meaning to the probed spatial scale , which makes the interferometric observables more interpretable physically and directly . this is quite important since our @xmath5 converage is normally poor and we still have to deal with direct observables in fourier space ( i.e. visibility ) , rather than quantities converted to those in image space . we note that the dust sublimation radii from the reverberation measurements are systematically smaller by a factor of @xmath03 @xcite than that expected for a representative property of dust grains in the galactic interstellar medium for each given @xmath1 @xcite . we have specifically discussed this issue in @xcite , including the case for large grains as well as anisotropic radiation and absorption . the case for large grains in the nuclear region has long been suggested on different grounds ( e.g. @xcite ) , while the effect of anisotropy is extensively studied recently by @xcite . ( a)m in the rest frame of each object , shown as a function of normalized spatial frequency ( or its reciprocal , normalized spatial wavelength shown in the upper axis ) . the data are plotted using different symbols for different objects ( see legend ) , and color - coded using the uv luminosity of the central engine in each object ( see color bar ) , labeled with the position angle and length of the projected baseline . dotted curves are the visibility functions for power - law brightness distributions with designated half - light radii in units of @xmath2 . dashed and dot - dashed gray curves are for gaussians with hwhm 20 and 5 @xmath2 , respectively . ( b ) the same plot as ( a ) but with @xmath6-axis in linear scale . both panels are from @xcite.,title="fig:",scaledwidth=50.0%](b)m in the rest frame of each object , shown as a function of normalized spatial frequency ( or its reciprocal , normalized spatial wavelength shown in the upper axis ) . the data are plotted using different symbols for different objects ( see legend ) , and color - coded using the uv luminosity of the central engine in each object ( see color bar ) , labeled with the position angle and length of the projected baseline . dotted curves are the visibility functions for power - law brightness distributions with designated half - light radii in units of @xmath2 . dashed and dot - dashed gray curves are for gaussians with hwhm 20 and 5 @xmath2 , respectively . ( b ) the same plot as ( a ) but with @xmath6-axis in linear scale . both panels are from @xcite.,title="fig:",scaledwidth=50.0% ] figure [ fig_midir_vis]@xmath7 is such a uniform comparison plot . it shows the visibility amplitudes measured with midi / vlti for 6 type 1 objects at the rest - frame wavelength range of 1012 @xmath8 m as a function of normalized spatial frequency in log scale ( or normalized spatial wavelength ; see upper axis ) . figure [ fig_midir_vis]@xmath9 is for the reader who prefers to see the same plot in linear scale . first , _ if _ the structure scales only with @xmath3 , i.e. if we see intrinsically the same distribution ( only with the radius - temperature scaling discussed above ) , the visibility curves should look all the same in the normalized units . however , this is _ not _ the case . visibilities at a given normalized spatial frequency , or at a given spatial scale probed , are systematically different : within the small sample , which nevertheless spans over @xmath02.5 orders of magnitudes in uv luminosity , the overall mid - ir emission size in higher luminosity objects ( shown in bluer colors ) seems more compact than that in lower luminosity ones ( redder colors ) , when seen in units of @xmath2 . here we need to be careful when we talk about the size . in fig.[fig_midir_vis]@xmath7 , we also have plotted visibility curves of gaussians with hwhm of 20 and 5 @xmath2 . in this log - scale plot , the visibility curve for a gaussian with another hwhm will simply be shifted in the horizontal direction with no change in shape , with the shift amount proportional to the hwhm ratio . this conveniently tells us which visibility data points probe approximately which spatial scales , but this also immediately tells us that a single gaussian seems quite inadequate for describing the mid - ir visibility curve for each object . this is also evident in fig.[fig_midir_vis]@xmath9 where none of the objects seem to follow gaussian - like visibility curves . rather , the observed visibility curves show that the mid - ir flux originates from a much wider range of spatial scales in each object . the dotted curves in both panels of fig.[fig_midir_vis ] are for power - law brightness distributions with different steepnesses , where the steepness is characterized as a half - light radius , within which the half of the total integrated light at a given wavelength is contained ( with the inner cut - off at radius @xmath2 ; the outer radius is set to 100@xmath2 here ) . each object seems to follow a power - law brightness distribution of a different steepness , with higher luminosity objects looking steeper , while lower luminosity ones being more extended . furthermore , since the radial temperature run of the heated ( illuminated ) dust grains as a function of the normalized radius is expected to be roughly the same for these face - on targets , the steepness of the brightness distribution would correspond to that of the surface density of these heated dust grains . thus we can infer that the higher luminosity objects have steeper dust density distributions . in the near - ir , the first measurement of the k - band ( 2.2 @xmath8 m ) visibility was made for the brightest type 1 agn ngc4151 with the keck interferometer ( ki ) , which showed squared visibility @xmath10 at a projected baseline @xmath9 of 83 m @xcite . the result was recently confirmed by us @xcite and also by @xcite . furthermore , we measured the k - band visibility for in total 8 type 1 agns @xcite . all of the targets showed high visibilities @xmath11 0.8 0.9 at @xmath12 m. we interpret this as an indication of the partial resolution of the dust sublimation region . first , we have marginally detected a decrease of visibility over increasing projected baselines for ngc4151 ( fig.[fig_keck_ngc4151 ] ) @xcite , which would be robust evidence that we are resolving a structure . secondly , from this visibility curve , we derived a thin - ring radius for ngc4151 , and also for the other agns from the observed visibilities . these interferometric ring radii turned out to be quite comparable to the dust sublimation radii from the near - ir reverberation measurements ( fig.[fig_nearir_vis]a ) @xcite , which strongly suggests that we are resolving the dust sublimation region . of ngc4151 as a function of projected baseline lengths observed in two epochs ( see legend ) from @xcite . the possible , perhaps conservative , systematic uncertainty in the @xmath13 calibration of up to @xmath00.03 is shown as dashed bar in the left inset ( see sect.3.2 of @xcite ) . in the right inset , ring radii corresponding to each data point are plotted along the position angle of each projected baseline . the dotted line in each panel corresponds to the best - fit thin - ring model with a radius of 0.45 mas for our may 2009 data.,scaledwidth=60.0% ] a further close look at the ring radii and reverberation radii shows that the former are slightly and systematically larger than the latter on average . this is not surprising since the reverberation measurements are known to be most sensitive to the smallest responding radii @xcite , meaning that they are most probably probing a radius very close to the inner boundary of the dust distribution ( thus adequately adopted as the dust sublimation radius ) . on the other hand , the interferometric ring radii are rather brightness - weighted , effective radii of the overall near - ir brightness distribution . therefore , we can postulate that the ratio of the ring radius to the reverberation radius @xmath2 , i.e. ring radius in units of @xmath2 , can approximately describe the steepness of the structure , in the sense that the ratio is closer to unity when the structure is steeper , while much larger than unity when it is more extended ( fig.[fig_nearir_vis]a ) . m for 8 type 1 agns ( filled circles @xcite ) as a function of uv luminosity @xmath1 . the near - ir reverberation radii are shown as gray plus signs , and the dotted line is the @xmath4 fit ( @xcite ) , which defines the dust sublimation radius @xmath2 we adopt here . the ratio of the ring radius to @xmath2 would indicate the steepness of the brightness distribution , and thus the darker and lighter shades correspond to steeper and shallower structures , respectively . ( b ) the ring radii in units of @xmath2 are plotted against radio ( 5 ghz ) to optical luminosity ratio @xcite . ] this description of course has a limitation . as the power - law becomes more shallow and extended , the single ring radius description becomes more inadequate , as we have seen in the mid - ir visibility curves . here , the underlying assumption is that the near - ir brightness distribution is steep enough to be approximately described by a single ring radius , independent of baseline lengths ( though still at relatively low spatial frequencies ) . currently , the near - ir structure is not well constrained yet , and this should be followed up with multiple , longer baseline observations . under this approximate description of the inner radial structure , do we see any relationship between the inner radial structure and the central engine ? in fact , there is tentative evidence that jet - launching objects tend to have a steeper structure ( fig.[fig_nearir_vis]@xmath9 ) @xcite . however , the statistical significance is not high and this should be investigated further with a larger sample . m , while dotted curves are the visibilities only for the former power - law component . ( b ) the total flux spectrum in the mid - ir with midi and host - galaxy - subtracted total flux spectrum in the near - ir are shown in black . the two mid - ir data points slightly above the midi spectrum are visir measurements with a slightly larger aperture , and gray data points are the spitzer measurements with even larger apertures . the correlated flux given by the midi and ki data are shown beneath the total fluxes , color - coded to the corresponding spatial wavelength ( reciprocal of spatial frequency ) in units of @xmath2 ( see the color bar at the top ) . the model curves for each midi baseline configuration are also shown in the same color - coding . see @xcite for more details . ] we discussed above the mid - ir and near - ir emission regions separately . obviously , we do want to know whether these two are nicely interconnected . can we think these two are coming from a single component ? the answer does not seem to be affirmative , or at least not in a very obvious way . certainly we do see that the structure looks more and more compact toward shorter wavelengths . figure [ fig_vis_corrflx ] shows the data for ngc4151 , and some of their aspects are shared by other objects . figure [ fig_vis_corrflx]@xmath7 shows visibilities both in the mid- and near - ir , where the color of each data point represents the observing wavelength as indicated by the color bar at the top . toward shorter wavelengths , the overall visibility becomes larger quite quickly , and thus the corresponding structure becomes smaller . however , when we look into the total flux spectrum , shown in black in figure [ fig_vis_corrflx]@xmath9 , it can be easily seen that if we try to extrapolate the mid - ir power - law component toward shorter wavelengths , its color is too red to consistently explain the near - ir part of the spectrum . instead , the near - ir part makes up a somewhat distinct component on top of the mid - ir power - law component . as we have seen in the previous section , the near - ir size scale is quite close to the dust sublimation radius @xmath2 , and the near - ir total flux spectrum shows that its color temperature is around 1400 k , quite generically for other objects as well ( e.g. @xmath012 @xmath8 m sed fit in fig.2 of @xcite , and fig.3 of @xcite ) . from the measured size and the absolute flux level of this near - ir component , its brightness temperature is estimated to be quite close to the color temperature ( see fig.11 of @xcite ) , meaning that the surface filling factor is quite close to unity and the radiation is almost like that of a black body . thus the near - ir flux seems to originate from an optically - thick , bright rim region of the dust grains at around @xmath2 and at the sublimation temperature of @xmath01400 k. figure [ fig_vis_corrflx]@xmath9 also shows the correlated flux spectra at progressively longer baseline with different colors ( see the color bar , coded for spatial wavelength ) . for the visibilities at relatively low spatial frequencies in the first lobe ( i.e. before the first null ) and in the case of a roughly centro - symmetric object ( which we would expect for type 1s ) , the correlated flux approximately shows the spectrum of the part of the source unresolved by the interferometer . the correlated flux at longer baselines excludes the flux from the outer region , leaving the spectrum only for the inner core part of the source . since we also see that the visibility curves look relatively flat at long baselines ( fig.[fig_vis_corrflx]@xmath7 ) , they probably show the flux fraction of an almost unresolved source , with the corresponding correlated flux showing the spectrum of this unresolved core . the correlated flux spectrum at long baselines in fig.[fig_vis_corrflx]@xmath9 then indicates that the spectrum of this unresolved core does not seem to be as blue as the mid - ir tail of the near - ir - emitting hot rim component we inferred above it rather suggests a core at a much lower temperature . in this sense , this lower temperature core seems to co - exist with the near - ir hot bright rim at the spatial scale of @xmath2 . this hybrid description seems also consistent with the mid - ir and near - ir data for the other 5 targets , having a difference in the steepness of the mid - ir brightness distribution @xcite . all analyses above involve _ almost no modeling _ we can essentially infer all these conclusions from the direct observables . however , we can also make our arguments a little more quantitative by introducing a very simple , physical parameterization . to describe the face - on structure physically , we simply use power - law distributions of temperature and surface density , as we have already seen evidence for the distribution resembling a power law . in addition , we approximate the near - ir rim emission as a ring with a certain effective radius . the accretion disk component , which seems to become important only at wavelengths @xmath14 1 @xmath8 m , is assumed to be unresolved and have the near - ir spectral shape of @xmath15 , which is supported both theoretically and observationally @xcite . m in the rest frame of each object , shown in linear scale , using the best - fit parameters for the temperature and surface density indices @xcite , with spatial scales shown in units of @xmath2 . higher luminosity objects are put toward the right.,scaledwidth=100.0% ] in pc from best - fit models at rest - frame 13 and 8.5 @xmath8 m ( red and green , respectively ) , plotted against uv luminosity . thin - ring radii in pc at 2.2 @xmath8 m are plotted in purple for the ki - observed sample . the near - ir reverberation radii ( @xcite and references therein ) are indicated as gray plus signs with the @xmath3 fit shown as a black dotted line ( which is our definition of @xmath2 ) . the mid - ir part of this figure is more model - dependent than , but essentially consistent with , that of fig.8 in @xcite . the fits to @xmath16 are shown in red and green dotted lines with quoted slopes , where the normalizations are [email protected] and [email protected] pc at @xmath18(5500)=10@xmath19 erg / sec for 13 and 8.5 @xmath8 m , respectively . , scaledwidth=60.0% ] this parameterization can describe the whole data sets at least approximately . figure [ fig_vis_corrflx ] includes the fitted model curves for ngc4151 , and those for the whole sample are found in figs.1 and 2 of @xcite . using the best - fit parameters , we can construct the model images which visually illustrate what we have discussed . the panels of figure [ fig_image_rin ] are such images at 13 @xmath8 m in the rest frame of each object , with spatial scales in units of @xmath2 , showing that the structure becomes steeper for higher luminosity objects ( to the right ) . we can also show this structural change back in unnormalized , physical spatial scales , e.g. in pc , by looking at the change of characteristic radii at different wavelengths . figure [ fig_model ] shows the mid - ir half - light radii in pc for the 6 mid - ir targets at the rest - frame 13 and 8.5 @xmath8 m derived from the model fits above , together with the thin - ring radii at 2.2 @xmath8 m for the 8 near - ir targets ( with two overlapping targets ) . if the surface density structure were the same over the sample , the ratios of the mid - ir to near - ir size would just be the same over the sample . however , since mid - ir radii in pc increase with luminosity @xmath1 _ much slower _ than @xmath3 ( specific values are given in fig.[fig_model ] ) , the difference between radii at different wavelengths becomes smaller toward higher luminosities , meaning that the radial density structure becomes steeper at higher luminosities . note that figure [ fig_model ] is essentially the same as fig.8 of @xcite which shows the mid - ir half - light radii measured with a simple geometrical power - law brightness at each wavelength . the half - light radii from the temperature-/density - gradient model are certainly more model - dependent , but utilize more information for each object including the total flux spectrum . the results are essentially consistent within the errors . the structural steepening could potentially mean that , since the accretion disk luminosity is directly linked to the mass accretion rate , the radial density distribution might become steeper in higher accretion rate objects ( here we refer to the absolute rate , not normalized by the eddington rate ) . we now even have quantitative estimations for the radial surface density index of the heated dust distribution , which ranges from @xmath0@xmath20 to @xmath0@xmath21 for the luminosities covered . we plan to further investigate correlations with different quantities such as the eddington ratio using a larger sample in order to advance our physical understanding . we have summarized the results of our first systematic study of type 1 agns with ir interferometry , where we attempted to map the radial structure of the inner dusty accreting material . for the warm mid - ir power - law - like component , we seem to see steeper density distributions in objects with higher luminosities , or higher mass accretion rates . for the near - ir compact component in the dust rim region , we see a tentative correlation of steeper structures with more radio - loud central engines . we plan to further scrutinize all the results with a larger sample .	with mid - ir and near - ir long - baseline interferometers , we are now mapping the radial distribution of the dusty accreting material in agns at sub - pc scales . we currently focus on type 1 agns , where the innermost region is unobscured and its intrinsic structure can be studied directly . as a first systematic study of type 1s , we obtained mid-/near - ir data for small samples over @xmath034 orders of magnitudes in uv luminosity @xmath1 of the central engine . here we effectively trace the structure by observing dust grains that are radiatively heated by the central engine . consistent with a naive expectation for such dust grains , the dust sublimation radius @xmath2 is in fact empirically known to be scaling with @xmath3 from the near - ir reverberation measurements , and this is also supported by our near - ir interferometry . utilizing this empirical relationship , we normalize the radial extent by @xmath2 and eliminate the simple @xmath3 scaling for a direct comparison over the samples . we then find that , in the mid - ir , the overall size in units of @xmath2 seems to become more compact in higher luminosity sources . more specifically , the mid - ir brightness distribution is rather well described by a power - law , and this power - law becomes steeper in higher luminosity objects . the near - ir flux does not seem to be a simple inward extrapolation of the mid - ir power - law component toward shorter wavelengths , but it rather comes from a little distinct brightness concentration at the inner rim region of the dust distribution . its structure is not well constrained yet , but there is tentative evidence that this inner near - ir - emitting structure has a steeper radial distribution in jet - launching objects . all these should be scrutinized with further observations .
it is a great pleasure to thank my colleagues in the mirabelle , bebc and delphi experiments with whom many results considered in this paper were obtained and discussed . t. sjstrand , comp . phys . comm . * 82 * ( 1994 ) 74 . chliapnikov , ( 1999 ) 341 . chliapnikov , ( 2000 ) 263 . anisovich and v.m . shekhter , ( 1973 ) 455 . anisovich , v.a . nikonov and j. nuiri , . fiz . * 64 * ( 2001 ) 877 , phys . * 64 * ( 2001 ) 812 . r. barate et al . , phys . * 294 * ( 1998 ) 1 . ( 1995 ) 543 . ( 1995 ) 207 . ( 1996 ) 61 . ( 1998 ) 585 . ( 1999 ) 364 . r. akers et al . , ( 1994 ) g. alexander et al . , ( 1995 ) 162 . r. akers et al . , ( 1995 ) . g. alexander et al . , ( 1997 ) 569 . k. ackerstaff et al . , ( 1998 ) 19 . k. ackerstaff et al . , ( 1998 ) 411 . sld collab . , k. abe et al . , ( 1999 ) 052001 . chliapnikov , a.g . tomaradze , v.a . uvarov and e.a . de wolf , ( 1983 ) 432 . france - ussr and cern - ussr collab . chliapnikov et al . , ( 1980 ) 303 . kniazev et al . , . fiz . * 40 * ( 1984 ) 1460 . cern - ussr collab . ajinenko et al . , ( 1984 ) 307 . ajinenko et al . , ( 1980 ) 51 . garuchava , v.a . uvarov and p.v . chliapnikov , pisma zh . * 40 * ( 1988 ) 121 . e.a . de wolf et al . , ( 1986 ) 13 . brussels - cern - genova - mons - nijmegen - serpukhov collab . , m. barth et al . , ( 1981 ) 39 . brussels - cern - genova - mons - nijmegen - serpukhov collab . , m. barth et al . , ( 1984 ) 431 . o. biebel , _ http://home.cern.ch/b/biebel/www/rp00 _ yi - jin pei , ( 1996 ) 39 . ( 1997 ) 271 . k. ackerstaff et al . , ( 1997 ) 437 . k. ackerstaff et al . , ( 1997 ) 210 . g. abbiendi et al . , ( 2000 ) 61 . france - ussr and cern - ussr collab . ajinenko et al . , ( 1980 ) 177 . b. andersson , g. gustafson , g. ingelman and t. sjstrand , phys . rep . * c97 * ( 1983 ) 31 . a. capella and j. tran thanh van , ( 1981 ) 249 ; ( 1982 ) 450 .	the ratios of the production rates @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , @xmath6 and their @xmath7-dependence obtained from results of the lep and sld experiments in @xmath8 hadronic decays are analyzed . the corresponding ratios for promptly produced mesons are estimated at @xmath9 . a comparison of the lep results with those from the mirabelle and bebc @xmath10 experiments at 32 and 70 gev/@xmath11 shows striking similarity in fragmentation of the @xmath12 valence quark of the incident @xmath13 and strange quarks produced in @xmath8 decays . the jetset model describes the lep , mirabelle and bebc results . the model of pei is consistent with the data for mesons , but presumably underestimates the fractions of primary octet baryons . the quark combinatorics model of anisovich et al . is incompatible with the data . ollab . , the vector and pseudoscalar mesons , or decuplet and octet baryons , differ in the relative orientation of the quark spins . therefore the ratios of their production rates , @xmath14 and @xmath15 , provide us with important information on spin dependence in fragmentation . hadronization models predict @xmath14 and @xmath15 for promptly produced particles , not resulting from decays of other particles or resonances . therefore apart from the models based on monte - carlo generators , such as the jetset model @xcite , these predictions are difficult to test experimentally . in the nonrelativistic quark model , the mass difference of the vector and pseudoscalar mesons , decuplet and octet baryons is explained by the hyperfine mass splitting . the production rates exhibit a strong mass dependence . therefore the vector - to - pseudoscalar and decuplet - to - octet suppressions could also be explained , at least qualitatively , by the hyperfine mass splitting @xcite provided that the rates of promptly produced particles were reliably estimated . the simplest spin model of fragmentation , such as the model suggested in @xcite and subsequently developed by anisovich and his collaborators in many other papers , is a random combination of quark spin states giving @xmath16 and @xmath17 . recently anisovich et al . @xcite compared these predictions with the lep data . they analyzed the @xmath14 ratios obtained from the aleph data @xcite at large @xmath18 , where a contribution of the resonance decays is supposed to be strongly suppressed , and concluded that the experimental results are consistent with @xmath16 . this conclusion is in contradiction with the results obtained in @xcite . besides , the lep results on hadron production , in general , and aleph results @xcite , in particular , agree well with the jetset model @xcite . in jetset , the probability to produce a vector meson is controlled by the parameter @xmath19 . this parameter pertaining to mesons directly produced in hadronization is much smaller than 0.75 , the value expected from the quark - combinatorics prediction . in order to resolve these problems and to get better insight into the nature of the vector - to - pseudoscalar and decuplet - to - octet suppressions we investigate in this paper the ratios of the production rates ^(892)@xmath20/ @xmath21 , @xmath22/ @xmath21 , ^0 @xmath23/@xmath24 , @xmath25/@xmath24 , ^++(1232 ) @xmath26/ @xmath27 , ^(1385 ) @xmath28/@xmath29 and ^- @xmath30/@xmath29 obtained from results of the lep [ 6 - 17 ] and sld @xcite experiments . we estimate the values of these ratios for promptly produced hadrons by studying their @xmath7-dependence and comparing the fragmentation of strange quark in @xmath31 decays and ^+p @xmath10 reactions . indeed , a relatively strong suppression of multi - strange final states in kaon induced reactions at moderate energies offers good possibilities to trace the flow of the incident strange valence flavour among the reaction debris . the ^0 @xmath32 , ^0(892)@xmath33 , @xmath22 , \|@xmath34 , ( 1385 ) @xmath35 and @xmath36 in the mirabelle [ 19 - 24 ] and bebc [ 25 - 27 ] ^+p @xmath10 experiments at 32 and 70 gev/@xmath11 are dominantly produced on the @xmath12 valence quark of the incident ^+ @xmath13 this is clearly seen from their feynman-@xmath37 spectra exhibiting a prominent leading particle effect and , consequently , from a relatively small production of these particles in the central region . this allows to obtain the reliable estimates of @xmath14 and @xmath15 ratios for promptly produced hadrons . therefore an interesting and sensible test of the lep results , as well as model predictions , on @xmath14 and @xmath15 ratios for promptly produced hadrons is feasible with kaon beams . in studying the ratios ^(892)@xmath20/ @xmath21 , @xmath22/ @xmath21 , ^0 @xmath23/@xmath24 , @xmath25/@xmath24 , ^++(1232 ) @xmath26/ @xmath27 , ^(1385 ) @xmath28/@xmath29 and ^- @xmath30/@xmath29 , the differential cross - sections , @xmath38 , for the resonances and ^- @xmath30 were taken from the lep experiments [ 6 - 17 ] . in some of them , the inclusive spectra have been presented as a function of @xmath39 . the difference between @xmath7 and @xmath40 , important only at very small values of @xmath7 and @xmath40 , was ignored . the differential cross - sections for the @xmath21 , @xmath24 , @xmath27 or @xmath29 in the same @xmath7-intervals were taken either from the same experiments or from combined @xmath7-spectra of corresponding particles measured by aleph @xcite , delphi @xcite , opal @xcite and sld @xcite and compiled in @xcite . these spectra were fitted by a sum of two exponentials , and cross - sections in the corresponding @xmath7-intervals were calculated using the results of these fits . such combined @xmath7-spectra for the @xmath41 and @xmath42 with the results of their fits by exponentials are illustrated , for example , in fig . 1 . ( a ) and @xmath41 ( b ) differential cross - sections , @xmath38 , for @xmath43 measured by aleph , delphi , opal and sld . the _ curves _ are the results of the fit by a sum of two exponentials . ] ( a ) and @xmath41 ( b ) differential cross - sections , @xmath38 , for @xmath43 measured by aleph , delphi , opal and sld . the _ curves _ are the results of the fit by a sum of two exponentials . ] the ^0(892)@xmath33/(3^+ @xmath13 ) ratios as a function of @xmath7 are presented in fig . 2a . the ^0(892)@xmath33 differential cross - sections were taken from the aleph @xcite , delphi @xcite and opal @xcite experiments . the results of three lep experiments are well consistent . the ^0(892)@xmath33/(3^+ @xmath13 ) ratio increases with @xmath7 , presumably approaching the value for the promptly produced mesons at @xmath9 . the ratios @xmath44 and @xmath45 for the promptly produced mesons measured in the mirabelle @xcite and bebc @xcite ^+p @xmath10 bubble chamber experiments at 32 and respectively 70 gev/@xmath11 are also shown in fig . 2a production cross - section @xcite not accounted for in @xcite . ] . they are in excellent agreement with a trend of the lep data for @xmath9 . they are also well consistent with the predictions of the jetset @xcite and pei @xcite models for the promptly produced mesons at lep also shown in fig . 2a . on the other hand , and contrary to conclusion of ref . @xcite based on the same analysis of the aleph data , the quark - combinatorics prediction , ^(892)@xmath20/(3 @xmath21 ) = 1 , for the promptly produced mesons is ruled out by the lep , mirabelle and bebc data . the experimental values at the largest @xmath7 are a factor of 3 lower than the prediction and deviate from it by at least 11 standard deviations . /(3^+ @xmath13 ) ( a ) and @xmath22/(3^+ @xmath13 ) ( b ) ratios obtained from the lep experiments as a function of @xmath7 ; the ^0(892)@xmath33/(3 ^ 0 @xmath32 ) ratios for the promptly produced particles from the mirabelle and bebc ^+p @xmath10 experiments at 32 and 70 gev/@xmath11 ( a ) ; the @xmath46 ratio as a function of feynman-@xmath37 from mirabelle ( b ) . the jetset and pei model predictions for the promptly produced particles at lep are also shown . the quark - combinatorics model predicts @xmath47 and @xmath48 . here and in the subsequent figures , some data points were slightly shifted to avoid overlap . ] /(3^+ @xmath13 ) ( a ) and @xmath22/(3^+ @xmath13 ) ( b ) ratios obtained from the lep experiments as a function of @xmath7 ; the ^0(892)@xmath33/(3 ^ 0 @xmath32 ) ratios for the promptly produced particles from the mirabelle and bebc ^+p @xmath10 experiments at 32 and 70 gev/@xmath11 ( a ) ; the @xmath46 ratio as a function of feynman-@xmath37 from mirabelle ( b ) . the jetset and pei model predictions for the promptly produced particles at lep are also shown . the quark - combinatorics model predicts @xmath47 and @xmath48 . here and in the subsequent figures , some data points were slightly shifted to avoid overlap . ] the @xmath22/(3^+ @xmath13 ) ratios , with the @xmath22 differential cross - sections measured by aleph @xcite , delphi @xcite and opal @xcite , are presented as a function of @xmath7 in fig . 2b . the data from three lep experiments are consistent within errors and exhibit a clear rise with increasing @xmath7 , very similar to the behaviour seen in fig . 2a . the @xmath49 ratio as a function of feynman variable @xmath50 , where @xmath51 is the longitudinal momentum in the centre - of - mass system , from mirabelle @xcite is also presented in fig . 2b for @xmath52 . it is strikingly similar to the behaviour of the lep data . the jetset and pei model predictions are somewhat lower than the experimental data at large @xmath7 , but consistent with them within one - two errors . the quark - combinatorics model @xcite predicts @xmath48 , where @xmath53 is the strangeness suppression parameter . for usually accepted value @xmath54 , this prediction is a factor of more than two higher than the experimental values at large @xmath7 . it deviates from the delphi value of @xmath55 at @xmath56 by 7 standard deviations or even more for @xmath57 suggested in @xcite . the ^0 @xmath23/(3^+ @xmath58 ) ratios , with the ^0 @xmath23 differential cross - sections measured by aleph @xcite and delphi @xcite , are presented as a function of @xmath7 in fig . 3 . the results of these experiments are consistent within errors for @xmath59 . for @xmath60 , the aleph data point is significantly higher than the delphi data points . the @xmath25/(3@xmath24 ) ratios measured by aleph @xcite and opal @xcite are also presented in fig . 3 . they are consistent with the ^0 @xmath23/(3^+ @xmath58 ) ratios for @xmath59 . for @xmath61 , the ratio @xmath25/(3 ^ 0 @xmath62 ) measured by aleph is close to the ^0 @xmath23/(3^+ @xmath58 ) ratio measured by delphi . the ^0 @xmath23/(3 ^ - @xmath63 ) ratio as a function of feynman-@xmath37 taken from the mirabelle experiment @xcite ( and convoluted around @xmath64 ) is also presented in fig . 3 . it is well consistent with the lep results for @xmath65 and lies between the aleph and delphi data points at @xmath7 = 0.7 - 1.0 . thus , in spite of some inconsistency of the lep results at the largest @xmath7 , the general tendency of the lep and mirabelle data in fig . 3 is quite similar to the one observed in fig . 2 . however , in this case the jetset and especially pei model predictions for the ratios of promptly produced mesons at lep appear to be underestimated . the quark - combinatorics prediction , @xmath66 , is not consistent with the data , contrary to conclusion of ref . @xcite based on the same analysis of the aleph data . it is again a factor of 2 to 5 higher than the experimental values at large @xmath7 . the values @xmath67 at @xmath7 = 0.6 - 0.8 @xcite and @xmath68 at @xmath7 = 0.5 - 1.0 @xcite deviate from this prediction by 16 and respectively 11 standard deviations . in principle , the experimental values of the vector - to - pseudoscalar ratios in the fragmentation region at @xmath9 could be reconciled with the quark combinatorics predictions in case of strong spin alignment of vector mesons , since in this case the expected value would be @xmath69 . the spin density matrix elements for the ^0 @xmath23 , @xmath70 , @xmath25 , @xmath22 and ^0(892)@xmath33 have been measured by delphi @xcite and opal [ 31 - 33 ] ( a summary of the results can be found in @xcite ) . the helicity density matrix elements @xmath71 of the ^0 @xmath23 , @xmath70 and @xmath25 are found compatible , over the entire @xmath7 range , with 1/3 corresponding to a statistical mix of helicity -1 , 0 and + 1 states , with no evidence for spin alignment . for the ^0(892)@xmath33 and @xmath22 produced in the small @xmath7 region @xmath72 , the values of @xmath73 and @xmath74 @xcite are also well consistent with no spin alignment . for the @xmath22 produced in the high @xmath75 region , some indication on unequal population in the three helicity states is observed , with @xmath76 @xcite and @xmath77 @xcite . for the ^0 @xmath78 , delphi @xcite measured @xmath79 at @xmath80 and opal @xcite obtained @xmath81 at @xmath75 . it is interesting that the lep results on the ^0 @xmath78 spin alignment are again very similar to the results of the mirabelle ^+p @xmath10 experiment @xcite . thus in spite of some preference for occupation of the helicity zero state observed for the ^0 @xmath78 and @xmath22at large @xmath7 , this can not explain the failure of the quark combinatorics model . the kinematics of the decuplet baryon decays does not allow to obtain information about ratios of promptly produced decuplet and octet baryons at large @xmath7 as have been possible for mesons . still a study of @xmath7-dependence of the decuplet - to - octet ratios allows to make several important observations . the ^++(1232 ) @xmath26/ @xmath27 ratio as a function of @xmath7 is presented in fig . 4 . the ^++(1232 ) @xmath26 differential cross - section was measured by delphi @xcite and opal @xcite . the @xmath7-dependence of the ^++(1232 ) @xmath26/ @xmath27 ratio obtained from the delphi measurements is essentially flat , as expected . the opal results are well consistent with those of the delphi for @xmath43 , but show quite unexpected behaviour at the smallest @xmath7 . the discrepancy in the ^++(1232 ) @xmath26 total rate in two experiments is obviously due to those suspicious opal data points at the smallest @xmath7 , where the extraction of the ^++(1232 ) @xmath26 signal is the most difficult . this is supported by comparison of @xmath7-dependences for the ^++(1232 ) @xmath26/ @xmath27 and ^+(1385 ) @xmath82/@xmath29 . the latter is also presented in fig . 4 . the data for the ^+(1385 ) @xmath82 were taken from aleph @xcite , delphi @xcite and opal @xcite . the ^+(1385 ) @xmath82/@xmath29 ratio as a function of @xmath7 is essentially flat . this behaviour is very similar to the one observed for the ^++(1232 ) @xmath26/ @xmath27 ratio by delphi in all @xmath7 range and by opal for @xmath43 . the ratios of the total rates ( @xmath83 + @xmath84)/(2\|@xmath34 ) = 0.086 0.016 , @xmath84/\|@xmath34 = 0.102 0.044 obtained respectively from mirabelle @xcite and bebc @xcite ( and also shown in fig . 4 ) agree very well with the lep results for the ^+ @xmath85/@xmath29 and delphi results for the ^++ @xmath86/ @xmath27 . the jetset and pei model predictions for the ratio ^+ @xmath85/@xmath29 of total rates at lep are also indicated in fig . 4 . they are well consistent with the lep , mirabelle and bebc data . finally we compared the ratios ^- @xmath30/@xmath29 obtained from the lep @xcite , mirabelle @xcite and bebc @xcite experiments . with the ^- @xmath30 and @xmath29 total rates averaged over results of all lep experiments @xcite , one gets @xmath87 . from the @xmath36 @xcite and obtains @xmath88 , in good agreement with the lep value . the corresponding value , @xmath89 , from bebc @xcite is consistent within errors with the lep and mirabelle results . the jetset and pei model predictions for the ratio of the total rates at lep , 0.090 and 0.072 , respectively , are in good agreement with the lep , mirabelle and bebc results . in view of good agreement between the lep and bubble chamber experiments on the ratios ^+ @xmath85/@xmath29 and ^- @xmath30/@xmath29 of the total rates , it is reasonable to assume that the corresponding ratios for the promptly produced baryons in these experiments might also be the same . the @xmath83 , @xmath84 and @xmath36 in the mirabelle ^+p @xmath10 experiment at 32 gev/@xmath11 can presumably be safely considered as promptly produced . indeed , even at lep , the fractions of promptly produced ^+ @xmath85 and ^- @xmath30 equal 0.92 and 0.75 in the jetset , and 0.91 and 0.41 in the pei models . the cross - section of the promptly produced \|@xmath34 in mirabelle can be estimated from the \|@xmath34 total rate @xcite after subtraction of \|@xmath34 from the @xmath83 , @xmath84 and @xmath90 decays cross - section is equal to the averaged value of the @xmath83 and @xmath84 cross - sections . ] and small fraction of the centrally produced \|@xmath34 @xcite . this gives _ prompt = 0.14 0.04 , [ ( @xmath84 + @xmath83)/2(\|@xmath34 + @xmath91)]_prompt = 0.14 0.03 where we took into account that one can not separate the prompt jetset and pei models predict @xmath92_{prompt } = 0.11 $ ] and 0.14 , respectively , in good agreement with the mirabelle result . for the @xmath93_{prompt}$ ] , the same models predict 0.13 and 0.41 , respectively . the jetset estimate is in good agreement with the mirabelle result . the pei model prediction is significantly higher , since the fractions of primary produced @xmath29 and ^0 @xmath94 are smaller in this model than in jetset by the factors of 3.7 and 2.2 , respectively . notice that good agreement of the pei model with the mirabelle result for the @xmath92_{prompt}$ ] and disagreement for the @xmath93_{prompt}$ ] is not contradictory , since the fraction of the promptly produced ^- @xmath30 in this model is also smaller by a factor of 1.8 than in jetset . one can also notice that the pei model predicts essentially the same ratios ^+ @xmath85/@xmath29 and ^++ @xmath86/ @xmath27 for the promptly produced baryons . therefore if the mirabelle result for the @xmath93_{prompt}$ ] ratio is not biased and indeed represents a good estimate of the corresponding ratio at lep , as indicated by all results of this paper , this implies that the fractions of promptly produced octet baryons in the pei model might be underestimated . the quark - combinatorics @xcite predicts @xmath92_{prompt } = \lambda/2 $ ] and @xmath93_{prompt } = 1 $ ] . the former just tests the strangeness suppression factor and agrees with the usually accepted @xmath54 . the latter is a factor of 7 higher than the mirabelle value and deviates from it by 29 standard deviations . in conclusion , we have shown that a study of the vector - to - pseudoscalar ratios at lep as a function of @xmath7 allows to obtain estimates of these ratios for promptly produced mesons at @xmath9 . these estimates , as well as the @xmath36/(\|@xmath34 + @xmath91 ) and ( @xmath84 + @xmath83)/2(\|@xmath34 + @xmath91 ) ratios for the total rates , have been found in very good agreement with the results of the mirabelle and bebc ^+p @xmath10 experiments . this shows that the fragmentation properties of the @xmath12 valence quark of the incident kaon and strange quarks produced in e^+e^- @xmath95 annihilation are very similar . in fact , this interesting experimental observation is not unexpected . in the lund string model @xcite ( implemented in the jetset 7.4 generator ) , quark - antiquark pair in an @xmath96 event is produced in a colour force field stretched between the @xmath97 and the @xmath98 . the hadron production is viewed as a breaking of the string which can be interpreted as virtual @xmath99 pair production in a colour flux - tube . a soft hadronic collision is also considered in this model as a colour separation mechanism whereby valence quarks of incident meson act as borders of colour strings , analogous to the @xmath100 field in e^+e^- @xmath95 annihilation . therefore it is not surprising that many features in fragmentation of the @xmath12 valence quark of the incident ^+ @xmath13 and strange quarks in e^+e^- @xmath95 annihilation must be similar . other fragmentation models for soft , low-@xmath101 hadron - hadron interactions such as the dual sheet models based on the dual topological unitarization ( dtu ) ( see @xcite , for example ) differ from the lund model in their prescriptions for the number and topology of colour strings and in the role attributed to valence quarks ( as discussed , for example , in @xcite ) . therefore , a comparison of meson induced reactions with e^+e^- @xmath95 data might allow to get better understanding of the fate of meson valence quarks and to discriminate among different approaches . the predictions of the jetset model have been found in good agreement with the results of the lep and ^+p @xmath10 experiments on the vector - to - pseudoscalar ratios for the promptly produced mesons as well as with the @xmath36/(\|@xmath34 + @xmath91 ) and ( @xmath84 + @xmath83)/2(\|@xmath34 + @xmath91 ) ratios for the total and direct rates . the pei model predictions are also in reasonable agreement with the data , apart from the ^0 @xmath23/@xmath24 ratio and significant disagreement with the mirabelle result on the ( @xmath84 + @xmath83)/2(\|@xmath34 + @xmath91 ) ratio for the direct rates . this indicates that the pei prediction about significantly smaller fraction of promptly produced octet baryons in comparison with jetset might be too strong . the simplest spin models of fragmentation based on spin counting and , in particular , the quark - combinatorics model of anisovich et al . are incompatible with the results of the lep and ^+p @xmath10 experiments .
weak maass forms are maass forms which allow exponential growth at cusps . we denote the space of weak maass forms of weight @xmath0 and eigenvalue @xmath1 for a congruence subgroup @xmath2 and a character @xmath3 as @xmath4 . and harmonic weak maass forms are weak maass forms whose eigenvalue is zero . we denote the space of harmonic weak maass forms of weight @xmath0 for a congruence subgroup @xmath2 and a character @xmath3 as @xmath5 . harmonic weak maass forms are related to ramanujan s mock theta functions . a mock theta function was introduced by ramanujan in his last letter to hardy . he gave some examples and definition . in 2001 , zwegers@xcite discovered that mock theta functions are holomorphic parts of nonholomorphic modular forms of weight @xmath6 . more generally , zagier@xcite defined mock modular forms . and it turns out that mock modular forms are holomorphic parts of harmonic weak maass forms of any weight . + in this paper , we construct even - integral weight weak maass forms from half - integral weight weak maass forms . to construct maps , we shall use the indefinite theta series . this theta series was used by shintani , niwa and cipra . using this theta series , shintani@xcite constructed an inverse map of shimura correspondence and niwa@xcite gave another method to construct even - integral weight cusp forms from half - integral weight cusp forms and proved the conjecture about the level in the shimura correspondence . and cipra@xcite extended niwa s lifting to all holomorphic modular forms of all positive , half - integral weight . and we will regularize theta integral to make theta liftings well - defined in the case of weak maass forms . borcherds@xcite used a regularized theta integral and constructed theta liftings on weakly holomorphic modular forms . and bruinier@xcite defined theta lifting on harmonic weak maass forms of weight @xmath6 . + let @xmath7 be the upper half - plane and let @xmath8 . following cipra s method in @xcite for given @xmath9 and a character @xmath3 for @xmath10 we obtain theta functions @xmath11 where @xmath12 . this theta function has an important property : @xmath11 is a nonholomorphic modular form of weight @xmath13 for @xmath10 and @xmath3 as a function of @xmath14 and @xmath15 is a nonholomorphic modular form of weight @xmath16 for @xmath17 and @xmath18 as a function of @xmath19 . for a given weak maass form @xmath20 of weight @xmath13 for @xmath10 and @xmath3 we define a function on @xmath7 @xmath21 and @xmath22 where @xmath23 for @xmath24 and @xmath25 . here @xmath26 is a regularized integral which was used by borcherds @xcite . then this gives a lifting from weak maass forms of weight @xmath13 to weak maass forms of weight @xmath16 . the following is the precise statement : [ main1 ] let @xmath0 and @xmath27 be any integers and let @xmath28 . assume that @xmath29 . then @xmath30 is a weak maass form of weight @xmath16 and eigenvalue @xmath31 for @xmath32 and @xmath18 . but it may have singularities at heegner points of the form @xmath33 for @xmath34 , while @xmath35 does not have any singularities in the case when @xmath20 is a usual maass form and @xmath36 . shimura correspondence is the map from cusp forms of weight @xmath13 to cusp forms of weight @xmath37 . unlike the shimura correspondence , we can choose @xmath0 and @xmath27 independently . and @xmath0 and @xmath27 have any sign . here @xmath38 is a subspace of @xmath39 the space of weak maass forms , whose precise definition will be given in section 2 . and @xmath40 . in particular , harmonic weak maass forms are weak maass forms whose eigenvalues are zero . the space of harmonic weak maass forms are denoted by @xmath41 , that is , @xmath42 . and we have subspaces @xmath43 and @xmath44 of @xmath45 , which will be defined in section 2 . note that if @xmath46 then @xmath20 has a fourier expansion of the form @xmath47 where @xmath48 for @xmath49 and @xmath50 . [ main2 ] let @xmath51 and @xmath52 . assume @xmath29 and that @xmath53 if @xmath54 and @xmath3 is a principal character . and let @xmath55 . if @xmath56 has a fourier expansion as above then @xmath57 with the same possible singularities at heegner points as in theorem [ main1 ] . moreover if @xmath58 then we have the following results : 1 . if @xmath59 then we have @xmath60 and if @xmath54 then we have @xmath61 2 . if @xmath59 then we have @xmath62 @xmath63 and if @xmath54 then we have @xmath64 where @xmath65 and @xmath66 . then this is the analogous result with niwa@xcite and cipra@xcite . so this is the extension of niwa s theta lifting to harmonic weak maass forms . if @xmath20 is a usual maass form and @xmath36 , then we do not need to regularize integral because @xmath67 is rapidly decreasing at all cusps ( see theorem 2.6 in @xcite ) . we consider a holomorphic eisenstein series ( see proposition 2.10 in @xcite ) : let @xmath59 . define @xmath68 then it is a maass form of weight @xmath13 with eigenvalue @xmath69 for @xmath10 and @xmath3 . assume that @xmath70 is primitive mod @xmath71 . recall that @xmath72 . then @xmath73 where @xmath74 then it is a maass form of weight @xmath75 with eigenvalue @xmath76 for @xmath17 and @xmath18 . let @xmath20 be the same as that in theorem [ main2 ] . if @xmath77 and we write its fourier expansion as follows @xmath78 then we have @xmath79 so we have the similar result with shimura s correspondence in @xcite but we have extra terms . in this section , we review weak maass forms and harmonic weak maass forms , which were introduced in @xcite and @xcite . for @xmath80 and @xmath81 , let @xmath82 where @xmath83 or @xmath84 as @xmath85 or @xmath86 , and @xmath87 is the quadratic residue symbol as defined in @xcite . if @xmath88 , and @xmath89 is a function on @xmath90 , define @xmath91 when @xmath92 is odd and @xmath93 , define @xmath94 let @xmath89 be a smooth function on @xmath90 and let @xmath0 be an integer or half - integer . and let @xmath2 be a congruence subgroup of @xmath95 and let @xmath3 be a character for @xmath2 . and let @xmath96 if @xmath97 then we say that @xmath89 is a nonholomorphic modular form of weight @xmath0 for @xmath2 and @xmath3 . define @xmath98 . this is the usual hyperbolic laplace operator in weight @xmath0 . + a twice continuously differentiable function on @xmath7 is called a maass form of weight @xmath0 and eigenvalue @xmath1 for @xmath2 and @xmath3 if 1 . @xmath99 for all @xmath100 , 2 . @xmath101 for @xmath102 , 3 . for all @xmath103 , @xmath104 as @xmath105 for some @xmath106 . we denote the vector space of those * maass forms * by @xmath107 . if we change the third condition , which is a growth condition , into 1 . for all @xmath103 , @xmath108 as @xmath105 for some @xmath106 . then @xmath89 is called a * weak maass form. especially , if @xmath109 , then it is called a harmonic weak maass form. + let @xmath110 denote respectively the spaces of weak maass forms and harmonic weak maass forms . if @xmath89 is a weak maass form for @xmath2 , then it satisfies @xmath111 since @xmath112 for some positive integer @xmath113 . hence there is a fourier expansion @xmath114 then we define @xmath115 this is the subspace of @xmath4 . in particular , @xmath116 can be characterized by the differential operator @xmath117 , which is studied by bruinier and funke@xcite : we define a differential operator @xmath118 . this gives a map @xmath119 where @xmath120 is the space of weakly holomorphic modular forms of weight @xmath0 for @xmath2 and @xmath3 . it is easy to see that @xmath121 . here @xmath122 is the space of holomorphic modular forms of weight @xmath0 for @xmath2 and @xmath3 . the fourier expansion of any @xmath123 gives a unique decomposition @xmath124 , where @xmath125 and @xmath48 for @xmath49 and @xmath50 . here @xmath126 and @xmath127 are called the holomorphic part and nonholomorphic part of @xmath89 , respectively . note that @xmath89 has a fourier expansion of the form ( [ fourier ] ) at all cusps . and the fourier polynomial @xmath128 is called the principal part of @xmath89 . we define another subspace @xmath129 of @xmath130 as the inverse image of @xmath131 under the map @xmath117 . so it has a fourier expansion @xmath132 in summary we have introduced the following notations : @xmath133 @xmath134 in this section we define theta functions by using cipra s method in @xcite . the most of results in this section are in @xcite . we begin by defining the weil representations on @xmath135 on the space of schwartz functions @xmath136 . let @xmath137 be a rational symmetric matrix of signature @xmath138 . for @xmath139 , define the inner product @xmath140 for a matrix @xmath141 and a schwartz function @xmath142 , define the weil representation @xmath143f(ax ) & \text{if $ c=0$},\\ \|\det q\|^{-1/2}\|c\|^{-n/2}\int_{\mathbb{r}^n}e[\frac{a < x , x>-2<x , y>+d < y , y>}{2c}]f(y)dy & \text{if $ c\neq0$}. \end{cases}\ ] ] [ matrix ] let @xmath144 and @xmath145 for @xmath146 . define @xmath147 by @xmath148 , and let @xmath149 then 1 . @xmath150 , 2 . @xmath151 . proof * this is proposition 1.3 of @xcite . [ second ] for @xmath152 , and @xmath153 , let @xmath154 . let @xmath155 be as before . then @xmath156 and @xmath157 * proof * this follows since @xmath158 . let @xmath159 be an even lattice in @xmath160 and let @xmath161 be the dual lattice . denote by @xmath162 the volume of a fundamental parallelotope of @xmath159 in @xmath160 : @xmath163 let @xmath164 be a @xmath165-basis for @xmath159 , and define @xmath166 . 1 . we say that a function @xmath167 has the * first permutation property * for @xmath10 with a character @xmath3 if it satisfies 1 . @xmath168 if @xmath169 , 2 . @xmath170 for @xmath171 , + where @xmath3 is a character mod @xmath71 . we say that a function @xmath142 has the * first spherical property * for weight @xmath13 if it satisfies @xmath172 for all @xmath173 where @xmath92 and for @xmath174 @xmath175 let @xmath176 and define , for @xmath177 @xmath178 then , by shintani@xcite : [ shintani0 ] let @xmath80 . then @xmath179 where @xmath180 & \text{if $ c\neq0$}. \end{cases}\ ] ] take @xmath89 having the first spherical property for weight @xmath13 , and let @xmath181 have the first permutation property for @xmath10 with character @xmath3 . define @xmath182 and @xmath183 [ shintani ] [ shintani ] let @xmath184 . then @xmath185 where @xmath186 as in proposition [ shintani0 ] . * proof * this is theorem 1.5 in @xcite . [ shintani ] [ first ] let @xmath187 . then @xmath188 where @xmath189 with the hilbert symbol @xmath190 let @xmath191 be the orthogonal group of @xmath137 : @xmath192 . let @xmath193 denote the connected component of the identity in @xmath191 , consisting of those matrices @xmath20 with @xmath194 . we define a unitary representation of @xmath193 on @xmath195 by letting @xmath196 . by definition of @xmath193 , @xmath197 commutes with the weil representation ( see @xcite page 64 ) : @xmath198 now we introduce theta kernel as in @xcite . take the following special q : let @xmath199 be a matrix with signature @xmath200 . let @xmath201 . then @xmath202 . also , @xmath203 and @xmath204 . as a quadratic form , @xmath137 is given by the determinant of a matrix , for @xmath205 @xmath206 and there is a map from @xmath135 to @xmath193 : @xmath207 this map gives an isomorphism of @xmath193 with @xmath208 . 1 . let @xmath209 be a discrete subgroup of @xmath193 which leaves @xmath159 invariant . let @xmath210 be the ( normal ) subgroup of @xmath209 which fixes @xmath211 . let @xmath3 be a character of @xmath209 which is trivial on @xmath210 . we say that @xmath167 has the * second permutation property * for @xmath209 with character @xmath3 if it satisfies @xmath212 2 . let @xmath213 . we say that a function @xmath214 has the * second spherical property * for weight @xmath16 if it satisfies , by identifying @xmath155 as an element of @xmath193 @xmath215 let us define hermite polynomials : for @xmath216 , define @xmath217 [ spherical ] let @xmath27 and @xmath218 be integers . then for every positive integer @xmath219 such that @xmath220 , there is a function @xmath221 such that @xmath222 has the * first spherical property * for weight @xmath223 , and the * second spherical property * for weight @xmath16 . the function @xmath221 is defined by @xmath224 where @xmath225 for any choice of @xmath226 and @xmath227 such that @xmath228 . in particular , we may take @xmath229 * proof * this is theorem 2.1 in @xcite . let @xmath230 for some fixed @xmath219 as in theorem [ spherical ] where @xmath231 . consider theta function associated with @xmath232 : for @xmath233 and given character @xmath3 for @xmath10 we define a theta function @xmath234 where @xmath70 , @xmath235 , and @xmath236 , the dual lattice to @xmath237 . + the above theta function is a two - variable nonholomorphic modular form : 1 . @xmath238 as a function of @xmath14 , 2 . @xmath239 as a function of @xmath19 . * proof * basically we follow the proof of theorem 2.3 of @xcite . let @xmath240 for @xmath241 . we define @xmath242 where @xmath243 . with the notation of corollary [ first ] , @xmath244 with @xmath245 defined by 1 . @xmath246 if @xmath247 , 2 . @xmath248 if @xmath249 , where @xmath250 and its dual is @xmath251 . note that @xmath181 has the first permutation property for @xmath10 with character @xmath252 , and the second permutation property for @xmath253 with character @xmath18 ( see proposition 2.2 in @xcite ) . note that @xmath254 has the first spherical property of weight @xmath13 since @xmath255 commutes with the weil representation ( see ( [ commute ] ) ) . then by the corollary [ first ] , @xmath256 as a function of @xmath14 . + if we use equation ( [ p ] ) then we see that for @xmath257 @xmath258 where @xmath259 and @xmath260 . note that @xmath261 . so if we use the second spherical property of @xmath232 and second permutation property of @xmath181 then we get the desired transformation property for the second variable @xmath19 . + we define the fricke involution @xmath262 : @xmath263 let @xmath264 act on the variable @xmath14 , and @xmath265 act on @xmath19 ( see @xcite ) . then @xmath266 and hence transformation formulas of @xmath11 follow . + this theta function also has a good property about differential operators . [ pde ] the theta function @xmath11 , defined in ( [ theta ] ) , satisfies the pde @xmath267\theta(z , w;f_{k , m})\ ] ] @xmath268\theta(z , w;f_{k , m}).\ ] ] * proof * this is proposition 2.13 of @xcite . in this section we explain how to regularize the integral and we define theta lifting using that regularization . we will use borcherds * regularized integral. throughout we use the setup of @xcite . if we use the weak maass form , the integral in the theta lifting may be divergent . so it has to be regularized as follows : we integrate over the region @xmath269 , where @xmath270 is the usual fundamental domain of @xmath271 and @xmath269 is the subset of @xmath272 of points @xmath14 with @xmath273 . suppose that @xmath274 exists for @xmath275 and can be continued to a meromorphic function defined for all complex @xmath1 . then we define @xmath276 to be the constant term of the laurent expansion of the function ( [ limit ] ) at @xmath109 . if we use this regularized integral , we can define the theta lifting even though we use weak maass forms . + this is a regularized integral for @xmath95 but we need a regularized integral for @xmath10 . note that for a modular form @xmath277 @xmath278 where @xmath279 for @xmath24 . for @xmath280 define a function on @xmath7 @xmath281 + and for square - free , positive integer @xmath282 and @xmath283 let @xmath284 . then @xmath285 . define @xmath286 but in this case , we use @xmath11 such that its level is @xmath287 and its character is @xmath66 and use a regularized integral for @xmath288 . first we prove the convergence of @xmath35 for @xmath289 . since @xmath35 is defined by using @xmath290 , it is enough to show the convergence of @xmath291 . note that since @xmath292 is a finite sum , we can exchange sum and integration . and since @xmath293 , we only need to check @xmath294 for each @xmath295 . and if @xmath296 has a fourier expansion as @xmath297 then @xmath298 so we need to check the fourier coefficients of the constant term of @xmath296 . for this we need fourier expansions of @xmath299 and @xmath300 with respect to @xmath14 . by computing explicitly representations in the definition of @xmath11 in ( [ theta ] ) it turns out that @xmath301 where @xmath302 is a polynomial of @xmath303 and @xmath19 and @xmath304 . this gives a fourier expansion with respect to @xmath305 . next we will see the fourier expansion of @xmath300 for general @xmath295 . let @xmath306 this is just the original @xmath137 in ( [ q ] ) when @xmath307 . likewise let @xmath308 be a function @xmath232 in ( [ ex1 ] ) with @xmath307 . then @xmath308 satisfies the first and second spherical properties for the weights @xmath13 and @xmath16 respectively . now let @xmath309 and define @xmath167 by @xmath310 then @xmath181 has the first permutation property for @xmath10 with character @xmath311 and the second permutation property for @xmath253 with character @xmath312 . direct computation shows that , up to constant multiple , @xmath313 by theorem [ shintani ] we see that for each @xmath295 , @xmath300 can be written as @xmath314 where @xmath315 is another polynomial for @xmath303 and @xmath19 . and we know that @xmath316 has the fourier expansion of the form ( [ fourier0 ] ) . so the constant term of @xmath296 is a sum of terms of the form @xmath317 where @xmath318 . by the definition of @xmath319 , @xmath320 for some @xmath106 . so every term is exponentially decreasing as @xmath105 except the case of @xmath321 . so if @xmath322 for all @xmath323 then the constant term of @xmath296 goes to zero when @xmath324 goes to @xmath325 . so ( [ convergence ] ) converges and hence @xmath326 is well defined where @xmath322 for all @xmath323 . and it may have singularities where @xmath327 for some @xmath323 . since @xmath29 , we do nt need to consider the case of @xmath328 . so the singularities may occur where @xmath33 for @xmath34 . and transformation properties of @xmath326 come easily from the fact that @xmath15 is a nonholomorphic modular form of weight @xmath16 for @xmath17 and @xmath18 as a function of @xmath19 . we have maass differential operators on smooth functions from @xmath90 to @xmath329 ( see @xcite page 97 ) : @xmath330 for any smooth function @xmath331 and @xmath332 it is well known that @xmath333 the operator @xmath334 can be expressed in terms of @xmath335 and @xmath336 by @xmath337 we will use @xmath338 as the laplace operator with respect to @xmath19 and @xmath339 as the laplace operator with respect to @xmath14 . to prove that @xmath326 is an eigenfunction of @xmath338 we need following lemmas , which are essentially lemma 4.2 and lemma 4.3 in @xcite . [ previous ] let @xmath340 and @xmath341 . then @xmath342_{v = t}du . \end{aligned}\ ] ] proof * the assumptions imply that @xmath343 is a @xmath271 -invariant @xmath344-form on @xmath90 . by stokes theorem we have @xmath345 in the integrand over @xmath346 on the left hand side the contributions from @xmath271-equivalent boundary pieces cancel . thus @xmath347_{v = t}du.\ ] ] this implies the assertion . [ bruinier ] let @xmath348 . then @xmath349_{v = t}du\\ & -&\int^{1/2}_{-1/2}[v^{k/2}\sum_{\alpha\in \gamma_0(4n)\setminus { { \mathrm{sl}}}_2(\mathbb{z})}(l_{k/2}(f_\alpha))(z)\overline{g_\alpha(z)}]_{v = t}du . \end{aligned}\ ] ] * proof * we write @xmath350 and apply lemma [ previous ] twice . observe that @xmath351 . we have @xmath352 the last equality comes from proposition [ pde ] . and by lemma [ bruinier ] , @xmath353 by the same argument in the proof of the convergence , we see that two integrals @xmath354 vanish . and note that @xmath355 transforms under the group @xmath356 . so @xmath357 transforms under the group @xmath32 . let @xmath72 . we consider the lifting from @xmath358 to @xmath359 with singularities . in this case we take @xmath360 . then @xmath67 is the same theta function which was used by niwa@xcite and cipra@xcite . + as a special case of theorem [ main1 ] , if @xmath58 then @xmath361 with singulairites . now we will compute the image of the lifting when @xmath51 by using the unfolding method . to do that we need to rewrite theta functions as follows : now we will observe the behavior of @xmath326 at @xmath364 . the following computations are essentially done in theorem 2.12 in @xcite . assume that @xmath51 . let @xmath365 using the lemma [ unfolding ] , we have @xmath366\\ & & \times ( \eta / y)^{1-\nu}\sum_{m=-\infty}^\infty \chi_1(m)m^{\lambda-\nu}e^{-2\pi^2m^2(\eta / y)^2}dy / y \end{aligned}\ ] ] with @xmath367 . as @xmath368 , the only non - negligible term is that one involving @xmath369 : @xmath370 if @xmath59 then we invert the theta function using poisson summation @xmath371 with @xmath372 . we can now actually sum over @xmath373 : @xmath374 so @xmath375 with @xmath376 . the integral gives @xmath377 , and @xmath378 . so @xmath379 where @xmath380 and @xmath381 . and if we use functional equations for the @xmath159-series , we get the result about the constant term . r. e. borcherds , automorphic forms with singularities on grassmannians , invent . 3 , 491 - 562 . k. bringmann and k. ono , the @xmath400 mock theta function conjecture and partition ranks , invent . 165 ( 2006 ) , no . 2 , 243 - 266 . j. h. bruinier , borcherds products on @xmath401 and chern classes of heegner divisors , springer lecture notes in mathematics 1780 , springer - verlag ( 2002 ) , 152 pages . j. h. bruinier and j. funke , on two geometric theta lifts , duke math . j. 125 ( 2004 ) , no . 1 , 45 - 90 . j. h. bruinier , g. van der geer , g. harder , and d. zagier , the 1 - 2 - 3 of modular forms , universitext , springer - verlag , berlin , 2008 , lectures at a summer school in nordfjordeid , norway . b. cipra , on the niwa - shintani theta - kernel lifting of modular forms , nagoya math . j. 91 ( 1983 ) , 49 - 117 . s. niwa , modular forms of half integral weight and the integral of certain theta - fuctions , nagoya math . j. , 56 ( 1975 ) , 147 - 161 . g. shimura , on modular forms of half integral weight , ann . of math . , 97 ( 1973 ) , 440 - 481 . t. shintani , on construction of holomorphic cusp forms of half integral weight , nagoya math . j. , 58 ( 1975 ) , 83 - 126 . d. zagier , ramanujan s mock theta functions and their applications , sminaire bourbaki , 60me anne , @xmath402 . s. zwegers , mock theta functions , ph.d thesis , universiteit utrecht , 2002 .	we construct theta liftings from half - integral weight weak maass forms to even integral weight weak maass forms by using regularized theta integral . moreover it gives an extension of niwa s theta liftings on harmonic weak maass forms . and we obtain the similar results to those by niwa .
in this paper , we consider an equation named after two important topics in the field of integrable systems the kp hierarchy and the ultradiscrete systems . the kp hierarchy @xcite is a sequence of equations generated from a bilinear identity and plays an important role in the construction of a unified theory for integrable systems . by applying a transformation for the independent variables called the miwa transformation , the kp hierarchy is rewritten in the form of the discrete kp hierarchy @xcite . the fundamental equation in the discrete kp hierarchy is a trinomial equation known as the hirota - miwa equation @xcite . it is known that all equations in the discrete kp hierarchy are generated by such trinomial equations @xcite and that most discrete integrable equations are obtained by taking some reduction for this equation . ultradiscrete systems are difference equations in which only @xmath0 and @xmath1 operators appear and which are obtained by a limiting procedure @xcite from canonical difference equations . the remarkable point of this procedure is that it preserves the good properties of integrable systems , although the dependent variables only take discrete values . the most famous example is the box and ball system ( bbs ) @xcite , which is a cellular automaton consisting of an infinite sequence of boxes and a finite amount of balls . the bbs has solitons and an infinite amount of conserved quantities and is obtained by the ultradiscretization of the kdv equation . it is an interesting problem to try to obtain the structure of the solutions to ultradiscrete soliton equations , as opposed to those of ordinary soliton equations . in previous papers @xcite , we proposed a recursive representation which corresponds to the notion of vertex operators , to several ultradiscrete analogues of soliton equations including the ultradiscrete kp equation . as an analogue of determinant - type solutions , the ultradiscretization of signature - free determinants is discussed in @xcite and the relationship between this type of solution and ultradiscrete soliton equations is discussed in @xcite . the establishment of relationships to other mathematical topics is also studied , for example , algebro - geometrical @xcite and combinatorial theories @xcite . recently , finding exact solutions to specific ultradiscrete systems became an active field of research . for example , negative soliton solutions were found in the analysis of the bbs with generalized dependent variables @xcite . we previously solved the initial value problem of this equation including negative solitons @xcite . soliton solutions with periodic phase term are presented in @xcite . such solutions do not appear in discrete integrable equations . in this paper , we first propose yet another solution of the ultradiscrete kp equation . we secondly extend this solution to that of the ultradiscrete kp hierarchy and the non - autonomous one . we next discuss ultradiscrete specific solutions called backgrounds to this equation . finally , as an application , we consider the background solutions of the bbs with carrier ( bbsc ) , express this cellular automata as a lax form and discuss its compatibility condition . it is expected that the ultradiscrete kp hierarchy can be the master equation of the ultradiscrete as well as discrete systems . we define @xmath2 by @xmath3 for @xmath4 given by @xmath5 . here , @xmath6 and @xmath7 are parameters satisfying @xmath8 @xmath9 , @xmath10 and @xmath11 are given by @xmath12 where @xmath13 , @xmath14 . [ mainthm ] @xmath2 satisfies equations : @xmath15 the first equation ( [ ukphier1 ] ) does not depend on @xmath16 and @xmath17 . then , by regarding @xmath16 and @xmath17 as parameters , this equation is the ultradiscrete kp equation we proposed in @xcite . we shall call this equation the ultradiscrete kp equation for the variables ( @xmath18 ) . by focusing on variables @xmath16 , @xmath17 and @xmath19 and regarding the other variables as constants , the function ( [ defkphtau ] ) is equal to the solutions we proposed previously in @xcite . therefore , it solves the equation @xmath20 we call this equation the ultradiscrete kp equation for the variables ( @xmath16 ; @xmath17 , @xmath19 ) . this corresponds to ( [ ukphier1 ] ) in the case where @xmath21 and @xmath22 are larger than @xmath6 and @xmath7 respectively and the function depends on the form of @xmath23 and @xmath24 for independent variables @xmath16 , @xmath17 , @xmath25 and @xmath26 . we note that this function also solves @xmath27 by interchanging the roles of ( @xmath16 , @xmath25 ) and ( @xmath17 , @xmath19 ) . we call this equation the ultradiscrete kp equation for the variables ( @xmath28 ) . we also note that if the function ( [ defkphtau ] ) solves ( [ ukphier1 ] ) , it also solves @xmath29 by combining arguments above . we call this equation the ultradiscrete kp equation for variables the ( @xmath30 ) . before proceeding with the proof , we introduce some lemmas . [ lem1 ] let @xmath31 for @xmath32 such that @xmath33 then , it holds that @xmath34 we omit the proof because it is the same as that of lemma 2 in @xcite . [ lem2 ] when one requires that @xmath35 are solutions of ( [ ukphier1 ] ) and ( [ ukphier2 ] ) , one then has @xmath36 for @xmath37 . we prove ( [ lem2 - 1 ] ) by induction . by employing @xmath38 , we obtain @xmath39 adding the inequalities yields @xmath40 where @xmath41 . the third argument is less than @xmath42 because of ( [ ukphier1 ] ) for @xmath43 . then , all arguments in the maximum are less than @xmath42 by the assumption of the induction and lemma [ lem1 ] . the proofs of ( [ lem2 - 2 ] ) and ( [ lem2 - 3 ] ) are the same . now , let us prove theorem [ mainthm ] . we employ induction on n. it is trivial that @xmath44 solves these equations and we assume that @xmath45 satisfies ( [ ukphier1 ] ) and ( [ ukphier2 ] ) . let us prove ( [ ukphier1 ] ) first . by definition ( [ defkphtau ] ) , one has @xmath46 by virtue of lemma [ lem1 ] , the third argument of ( [ ukph1 ] ) can not yield the maximum because it is always less than the second . therefore , we can rewrite ( [ ukph1 ] ) as @xmath47 by employing the same reasoning , we obtain @xmath48 by the assumption , we have to prove that @xmath45 satisfies the equation : @xmath49 here , we consider each case for the values of @xmath9 , @xmath10 and @xmath11 . we note that in the case @xmath50 , @xmath10 should be equal to @xmath7 because of @xmath51 . therefore , we should consider six case @xmath52 or @xmath21 and @xmath53 , @xmath54 or @xmath55 . however , in the case @xmath52 and @xmath56 , ( [ ukphier1 ] ) reduces to ( [ ukpnnpm ] ) , which is already proven . therefore , it suffices to consider the remaining four cases . ( i ) : : the case @xmath57 and @xmath58 . in this case , @xmath45 depends on the form of @xmath59 for @xmath16 , @xmath26 , and @xmath19 . we arrange all of the shifts for the variables @xmath26 and @xmath19 to @xmath17 and omit these variables . then , ( [ mideq1 ] ) is rewritten as @xmath60 then , we should prove @xmath61 however , these inequalities are satisfied because @xmath62 is a solution of the ultradiscrete kp equation for @xmath63 . ( ii ) : : the case @xmath57 and @xmath64 . in this case , @xmath45 depends on @xmath24 for @xmath17 and @xmath26 . then , ( [ mideq1 ] ) is rewritten @xmath65 therefore , we should prove @xmath66 and @xmath67 now , ( [ mideq - a - ii-1 ] ) is ( [ ukphier1 ] ) for @xmath43 and variables @xmath68 and ( [ mideq - a - ii-2 ] ) is a special case of ( [ lem2 - 1 ] ) . ( iii ) : : the case @xmath52 and @xmath53 . @xmath45 depends on @xmath23 for @xmath16 , @xmath25 . then , ( [ mideq1 ] ) is rewritten @xmath69 therefore , we should prove @xmath70 and @xmath71 here , ( [ mideq - a - iii-1 ] ) is a special case of ( [ ukphier2 ] ) for @xmath43 and ( [ mideq - a - iii-2 ] ) is also a special case of ( [ lem2 - 2 ] ) . ( iv ) : : the case @xmath57 and @xmath53 . ( [ mideq1 ] ) is reduced to ( [ ukphier2 ] ) for @xmath43 . let us next prove ( [ ukphier2 ] ) . by virtue of a method similar to that in the proof of ( [ ukphier1 ] ) , we should prove @xmath72 now , by employing lemma [ lem2 ] , this equation is reduced to @xmath73 similar to the proof of ( [ ukphier1 ] ) , we should consider each possible case for the parameters . the case where @xmath52 and @xmath56 is already proven in @xcite . therefore , we should prove the remaining four cases . ( i ) : : the case @xmath57 and @xmath58 . @xmath45 depends on @xmath74 for @xmath75 . now , by virtue of the relation @xmath76 , the equation is rewritten as @xmath77 therefore , we should prove @xmath78 this inequality is obtained by interchanging the roles of the variables between ( @xmath16 , @xmath25 ) and ( @xmath17 , @xmath19 ) in the case @xmath79 for ( [ lem2 - 2 ] ) . ( ii ) : : the remaining cases : ( [ mideq - b ] ) are reduced to ( [ ukphier1 ] ) for @xmath43 . since the dependent variables @xmath25 , @xmath26 and @xmath19 do not change in the recursion ( [ defkphtau ] ) in the proof in the last section , we can extend the above discussion in two ways . one way is to increase the number of independent variables . by introducing integers @xmath80 and @xmath81 satisfying @xmath82 , and variables @xmath83 and @xmath84 , we define @xmath85 as @xmath86 where @xmath87 here , we omit subscripts which do not change in the equation . applying the proof to the variables ( @xmath88 ; @xmath89 , @xmath90 ) for @xmath91 , we obtain the ultradiscrete kp equation for ( @xmath88 ; @xmath89 , @xmath90 ) : @xmath92 by interchanging the roles of variables @xmath88 and @xmath89 , we also obtain the ultradiscrete kp equation for ( @xmath88 ; @xmath89 , @xmath90 ) . these equations are ultradiscrete analogues of the basic relation of the discrete kp hierarchy . the other way is by extending to non - autonomous systems . let us introduce functions @xmath93 and @xmath94 satisfying @xmath95 and @xmath96 for @xmath97 , @xmath98 and @xmath99 instead of parameters @xmath100 and @xmath101 . then , by denoting @xmath102 and @xmath103 , we can replace @xmath104 and @xmath105 with @xmath106 and @xmath107 respectively , in ( [ eta ] ) . here @xmath108 means @xmath109 combining these two generalizations yields the following theorem : the function written as @xmath110 for @xmath4 given by @xmath111 solves the non - autonomous ultradiscrete kp hierarchy : @xmath112 where the tuple of variables ( @xmath113 , @xmath114 , @xmath115 ; @xmath116 , @xmath117 , @xmath118 ) is ( @xmath88 , @xmath89 , @xmath90 ; @xmath119 , @xmath120 , @xmath101 ) or ( @xmath89 , @xmath88 , @xmath121 ; @xmath101 , @xmath122 , @xmath119 ) for @xmath123 and @xmath124 . in this section , we extend the start of the recursion ( [ defkphtau ] ) . first , we consider the case @xmath125 and @xmath126 for all @xmath127 , @xmath128 and @xmath129 . in this case , @xmath130 depends only on functions in the form of @xmath131 and @xmath132 . we denote this ( two variable ) function as @xmath133 , i.e , @xmath134 here , by virtue of lemma [ lem2 ] , one has @xmath135 then , @xmath133 solves the ultradiscrete kp hierarchy ( [ ukphier3 ] ) and ( [ ukphier4 ] ) becomes a trivial identity and all necessary inequalities to prove theorem [ mainthm ] are generated by the inequalities ( [ condbg0a ] ) and ( [ condbg0b ] ) . we note that we can also append a new soliton parametrized as @xmath136 and @xmath137 by the recursion ( [ defkphtau2 ] ) . therefore , we can choose arbitrary @xmath133 satisfying these inequalities for the start of the recursion , i.e. , we can choose a @xmath133 that satisfies @xmath138 and @xmath139 and @xmath140 . however , it should be noted that we do not need to restrict @xmath141 and @xmath142 to @xmath143 , i.e. , we can construct solutions even if the suprema of the left hand sides of ( [ condbg1 ] ) and ( [ condbg2 ] ) are negative . this solution is called a background " , because all solitons evolve over it . different from soliton solutions , backgrounds can keep their structure under addition because the left hand sides of ( [ condbg1 ] ) and ( [ condbg2 ] ) are linear for @xmath133 . now , note that the solutions of the ultradiscrete kp equation in @xcite are special cases of the solution defined in ( [ defkphtau ] ) . the backgrounds are further special cases of the solutions in @xcite . it is known that the standard bbs has several degrees of freedom , for example , the capacity of balls for boxes at each site and for carriers at each time . by setting reductions properly , we can obtain exact solutions of this system with backgrounds . we consider the case @xmath144 , @xmath145 and assume that @xmath130 depends on @xmath146 for @xmath16 , @xmath17 , @xmath147 and @xmath148 , i.e. , we consider the case where @xmath149 and @xmath150 are sufficiently large . we now introduce new independent variables @xmath151 , @xmath152 and @xmath153 and new dependent variables @xmath154 , @xmath155 , @xmath156 and @xmath157 . note that because of the constraints on @xmath158 , @xmath159 of section 3 , we have that @xmath160 , which means that the capacity of the carriers is greater than that of the boxes . then , the non - autonomous ultradiscrete kp equation for ( @xmath147 ; @xmath148 , @xmath161 ) and ( @xmath147 ; @xmath148 , @xmath162 ) is rewritten as @xmath163 here , we denote @xmath164 and set the boundary condition : @xmath165 for @xmath166 . due to these equations , @xmath167 equals @xmath168 . however , by virtue of the boundary condition , we also express this as @xmath169 therefore , one has @xmath170 which is the time evolution rule of the box and ball system with carrier ( bbsc ) . solutions to this equation are obtained by setting parameters @xmath171 for @xmath172 . soliton solutions for this system were first presented in @xcite and the direct relationship to the discrete case is shown in @xcite by ultradiscretizing soliton solutions of the discrete modified kdv equation . we note that we can also construct solutions of the bbsc multi - kinds of boxes and balls by properly setting different reduction conditions . due to the specialization of parameters @xmath6 and @xmath7 , the background solution satisfying ( [ condbg1 ] ) and ( [ condbg2 ] ) is expressed as @xmath173 where @xmath174 is a function satisfying @xmath175 . [ cols="^ " , ] we next observe the behavior of @xmath176 numerically by setting parameters @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 for @xmath184 , @xmath185 for @xmath186 . figure [ pic2 ] depicts @xmath176 in the range @xmath187 . two blocks of balls travel with at the same speed @xmath188 when @xmath189 . but when @xmath190 , a ball is injected the former block and passes the latter one finally . this phenomenon can be explained using the ultradiscrete miura transformation of @xcite . the dependent variables @xmath191 and @xmath176 satisfy the relationship : @xmath192 the following ultradiscrete - closed proof is originally presented in @xcite . by subtracting left hand side from right hand side , we obtain @xmath193 here , by employing ( [ backlund1 ] ) and ( [ backlund2 ] ) , one can rewrite @xmath194 which is identically equal to @xmath195 because of @xmath160 . due to the ultradiscrete miura transformation ( [ udmiura ] ) , the case where @xmath196 is achieved only if @xmath197 and @xmath165 . therefore , the length of a block increases at @xmath190 because of an increment of capacity of the carrier and the velocity of blocks in the bbsc ( [ bbsc ] ) . figure [ pic3 ] depicts the time evolution of @xmath191 which is obtained by ( [ udmiura ] ) from @xmath176 in figure [ pic2 ] . the contribution of @xmath198 is hidden in the degrees of freedom of solutions to ( [ ukdvnonlinear ] ) . to end this section , we stress that @xmath176 does _ not _ satisfy the ultradiscrete kdv equation known as the time evolution rule for the standard bbs : @xmath199 because this equation is obtained from by taking reductions @xmath200 ( and @xmath176 is the same ) from the ultradiscrete kp equation for ( @xmath16 ; @xmath17 , @xmath19 ) . however , any @xmath176 that satisfies ( [ standardbbs ] ) also solves ( [ ukdvnonlinear ] ) , because one can represent @xmath201 in two ways . in this paper , we proposed a recursive representation of solutions to an ultradiscrete analogue of the discrete kp hierarchy and its background solutions as a starting point for the recursion . for the equation , we took two different types of independent variables @xmath88 and @xmath89 and chose one variable from one type and two variables from the other . in the proof , we employed the properties of two variables which belong to the same type . however , solutions no longer solve the equation if we choose all three variables from one type . the difference of type plays an important role in the ultradiscrete equations . equation ( [ ukphier2 ] ) is considered to be obtained by ultradiscretizing a combination of some equations in the discrete kp hierarchy . in fact , we employ the reduction of ( [ ukphier2 ] ) as a lax form for the ultradiscrete kdv equation in section 6 . different from the discrete equations , ultradiscrete equations are not easily combined because of the @xmath0 operators . employing this reduction approach , we can obtain variations of ultradiscrete soliton equations which have been previously been obtained from discrete equations by an ultradiscrete , closed , approach . as discussed in section 3 , we can consider that the recursion ( [ defkphtau ] ) does not depend on any other independent variables except @xmath16 and @xmath17 . in this case however , equation ( [ ukphier1 ] ) is essentially not a difference equation but just an ordinary algebraic equation . since the number of variables is greater than that of the equations , its solutions form a 1-parameter family . if this parameter satisfies a relation corresponding to ( [ ukphier2 ] ) , we can obtain new solutions by applying the recursion to these solutions . we believe that this can be explained by some elementary approach , for example , a geometrical one , which should represent the essence of the ultradiscrete soliton equations . the author would like to thanks professors t. tokihiro and r. willox for helpful comments . this work was supported by platform for dynamic approaches to living system from the ministry of education , culture , sports , science and technology , japan .	we propose a recursive representation of solutions to an ultradiscrete analogue of the discrete kp hierarchy , which is the master equation of discrete soliton equations . we also propose a class of solutions which can be used to start the recursion . finally , as an application , we discuss the compatibility condition of ultradiscrete soliton equations . _ keywords _ : integrable systems ; solitons ; discrete systems ; cellular automaton ; discrete kp hierarchy
recently , stochastic partial differential equations ( spdes ) provide a quantitative description for a lot of mathematical models in areas such as physics , engineering , biology , geography and finance . many specialists exert a strong interest in spdes and develop their mathematical theories and analytical techniques . however , it is a difficult target to obtain the analytical solutions of spdes . thus , how to estimate the numerical solutions of spdes becomes a fast growing research area . a lot of modern numerical tools are devised to solve spdes such as the stochastic collocation method @xcite , the it taylor expansions method @xcite and the finite element method combining with monte carlo and quasi - mote carlo method @xcite . but there are still great many unsolved problems of the numerical solutions of spdes . the books @xcite show that the kernel - based approximation method to solve deterministic high - dimensional pdes , and moreover , this method can be also applied into the stochastic models mentioned in @xcite . it offers us a new idea to apply the kernel - based approximation method ( kernel - based collocation method , meshfree approximation method ) to obtain numerical solutions of high - dimensional spdes given in our recent paper and doctoral thesis @xcite . its approximate basis induced by the positive definite kernel ( reproducing kernel ) is different from the polynomial basis of the finite element method , which means that the construction of the kernel basis does not need an underlying triangular mesh . furthermore , the data points can be flexibly chosen for the use of either deterministic or random design , e.g. , halton points or sobol points . after the discussions of this fresh numerical method with many mathematicians and engineers , we determine to renew and improve the analytical results and numerical algorithms of our previous papers by their nice and helpful suggestions . we want to let this numerical tool for spdes be well readable in the interdisciplinary fields of both computational mathematics and statistics . in section [ sec : gauss - rk - pb ] , we extend [ theorem 3.1 , @xcite ] and [ theorem 7.2 , @xcite ] into theorem [ t : gauss - pdk - pb ] such that we can replace integral - type kernels by more general positive definite kernels to construct gaussian fields in the classical @xmath0-based sobolev spaces instead of the reproducing kernel hilbert spaces . we will employ these gaussian fields to introduce the kernel - based approximate solutions of spdes similar as the techniques of @xcite . their approximate kernel bases are set up by the positive definite kernels with the related differential and boundary operators defined as in the equation , and the covariance matrixes of the gaussian fields at the collocation points given in the formula are used to compute their expansion random coefficients . the blocks of the covariance matrixes are corresponding to the pairwise of collocation points . the covariance matrixes can be seen as the generalization forms of the traditional kernel - based interpolation matrixes discussed as in @xcite . section [ sec : ell - spde - ker - sol ] complements the construction processes and the proofs of the kernel - based approximation method , and we even give the new kernel - based approximate results for solving a system of high - dimensional linear elliptic spdes driven by various kinds of right - hand - side random noises , which are unsolved in the final remark section of @xcite . the kernel - based approximate solution is obtained for fitting the observation values simulated by the elliptic spdes . the kernel - based approximate solution of the elliptic spdes is a linear combination of the kernel basis , and moreover , its expansion random coefficients are solved by a random linear system whose random parts are simulated by the elliptic spdes ( see the equations ( [ eq : ker - based - sol - linear - ell]-[eq : ker - based - sol - coef - linear - ell ] ) ) . proposition [ p : prob - convergence - linear - ell - spde ] shows that the errors of the kernel - based estimators can be bounded by the fill distances in the probability sense . the fill distance denotes the radius of the largest ball which is completely contained in the space domain and which does not contain the chosen collocation points . this means that the kernel - based approximate solutions are convergent to the exact solutions of the elliptic spdes in probabilities and distributions . we present more detail of the error analysis for the elliptic spdes than in @xcite , and we discuss the convergent analysis by the knowledge of statistical learning replacing the maximum error bound to the confident interval in a different way of the classical kernel - based approximation method for the deterministic pdes . in section [ sec : par - spde - ker - sol ] , we use the implicit euler scheme to discretize the parabolic spde driven by time and space lvy noises at time in order to transform it to serval elliptic spdes at each discretization time step . next we solve these elliptic spdes by the kernel - based approximation method . we also briefly discuss the convergence of the kernel - based approximate solutions of the parabolic spdes undone in @xcite . we will consider many other time - stepping schemes and their convergent rates in our future research . the numerical examples for the sobolev - spline kernels and the compact support kernels show that the approximate probability distributions are well - behave for the second - order parabolic spdes driven by the time and space poisson noises ( see figure [ fig : par - spde - num - exa ] ) . since the covariance matrixes of the compact support kernels are sparse , we can solve the related linear systems as fast as the finite element method . more numerical examples will be posed in the author s personal webpage . in our next paper , we will use the extended theoretical results given in this article to set up the kernel - based estimators for the nonlinear spdes driven by lvy noises . in our previous papers and doctoral thesis of spdes , the kernel - based approximation method is also called the kernel - based collocation method . but some people may confuse the original name with another different method as stochastic collocation . in the same way as in the book @xcite we recall this numerical method its general name , and moreover , its estimator is said the kernel - based approximate solution or the kernel - based solution in this article . the kernel - based approximation method , the kernel - based collocation method and the meshfree approximation method are the same in all our spdes papers . we want to make a convenience for the engineers and the computer scientists to understand the kernel - based approximation method and avoid its technical details and proofs . in the beginning we give a traditional example of the parabolic spde to explain the kernel - based approximate processes and algorithms in a simple way . let @xmath1 be a regular bounded open domain of @xmath2 and @xmath3 be a @xmath0-based sobolev space of degree @xmath4 . suppose that @xmath5 is a poisson noise in @xmath3 with the form @xmath6 where @xmath7 is an orthonormal subset of @xmath8 , @xmath9 is a positive sequence and @xmath10 are the independent scalar poisson processes with parameter @xmath11 for all @xmath12 . we consider the second - order parabolic spde driven by @xmath5 @xmath13 where @xmath14 is a laplace differential operator and @xmath15 . suppose that this parabolic spde is well - posed and its solution @xmath16 almost surely such that @xmath17 . the proposed numerical method for solving the spde can be described as follows : 1 . discretize the spde in time by the implicit euler scheme at equally space time points , i.e. , @xmath18 where @xmath19 , @xmath20 and @xmath21 . let @xmath22 . we choose a finite collection of predetermined pairwise distinct collocation points @xmath23 and a symmetric positive definite kernel @xmath24 to construct the basis @xmath25 and the covariance matrix ( interpolating matrix ) @xmath26 here @xmath27 and @xmath28 mean that we differentiate the kernel function @xmath29 with respect to its first and second arguments , respectively , i.e. , @xmath30 . we can also compute that @xmath31 and @xmath32 . 3 . because the white noise increment @xmath33 at each time instance @xmath34 is independent of the solution @xmath35 . the noise term @xmath36 is well - defined and we can simulate @xmath37 at @xmath38 , i.e. , @xmath39 let @xmath40 for @xmath41 . combing the equation with the dirichlet boundary condition we obtain the elliptic spde @xmath42 where @xmath43 is seen as an unknown part , and @xmath44 and @xmath37 are viewed as given parts . the kernel - based solution of the spde can be written as @xmath45 and its random expansion coefficients @xmath46 and @xmath47 are computed by the random linear equation @xmath48 where @xmath49 . + this means that @xmath50 is approximated by @xmath51 for all @xmath52 . 4 . repeat ( s3 ) for all @xmath53 . we can also create an algorithm to obtain the sample paths of the spde : + initialize * @xmath54 and @xmath55 are given in the equations ( [ eq : ker - basis - exa ] ) and ( [ eq : cov - matrix - exa ] ) for @xmath19 . * @xmath56 . * @xmath57 . * @xmath58 . repeat @xmath53 * simulate @xmath59 for all @xmath12 . * @xmath60 for all @xmath52 . * @xmath61 . * in this section we want to extend the theoretical results of the integral - type kernels and the related reproducing kernel hilbert spaces given in [ lemma 2.2 and theorem 3.1 , @xcite ] and [ lemma 7.1 and theorem 7.2 , @xcite ] so that we can apply many various kinds of the positive definite kernels and the related sobolev spaces to create gaussian fields . these gaussian fields and positive definite kernels are used to introduce the kernel - based solutions of spdes in the following sections . [ d : pdk ] a continuous symmetric kernel @xmath63 is called _ positive definite _ on @xmath64 if , for all @xmath65 and all sets of pairwise distinct centers @xmath66 , the quadratic form @xmath67 which is equivalent to the matrix @xmath68 is positive definite . let @xmath3 be the @xmath0-based sobolev space of degree @xmath4 defined as in section [ sec : diffbound ] and @xmath69 be the borel @xmath70-algebra on @xmath3 . suppose that @xmath1 is a _ regular _ bounded open domain of @xmath2 for @xmath71 , and the symmetric kernel function @xmath29 is a positive definite kernel on @xmath1 . further suppose that @xmath24 . according to the discussions in section [ sec : rk - rkhs ] , the kernel function @xmath29 has the positive eigenvalues @xmath72 and the continuous eigenfunctions @xmath73 such that @xmath74 and it possesses the absolutely and uniformly convergent representation @xmath75 for simplifying the notations , we denote @xmath76 for the linear differential and boundary operators @xmath77 and @xmath78 denoted in the equations ( [ eq : diff]-[eq : bound ] ) . here @xmath79 , @xmath80 and @xmath81 , @xmath82 mean that we differentiate the kernel function @xmath29 with respect to its first and second arguments , respectively . the kolmogorov s extension theorem guarantees the existence of countable independent standard normal random variables @xmath83 on some probability space @xmath84 , i.e. , @xmath85 . since @xmath86 for all @xmath87 , the stochastic process @xmath88 is well - defined on @xmath84 . according to [ theorem a.19 , @xcite ] , the random variable @xmath89 is normal for all @xmath87 . we can also compute the mean @xmath90 and the covariance @xmath91 for all @xmath92 . therefore @xmath37 is a gaussian field with mean @xmath93 and covariance kernel @xmath29 ( see definition [ d : gaussian ] ) . [ l : gauss - sobolev ] almost all sample paths of the gaussian field @xmath37 defined in the formula belong to @xmath3 , i.e. , @xmath94 for almost all @xmath95 . denote that @xmath96 . since @xmath24 , we have @xmath97 in the @xmath98-norm . we fix any @xmath99 . according to lemma [ l : rk - diff - bound ] , the eigenfunctions @xmath100 and the convergent representation @xmath101 is absolute and uniform on @xmath102 . since @xmath1 is bounded , the map @xmath103 and the representation @xmath104 is also convergent in the @xmath105-norm . thus , the sequence @xmath106 is a cauchy sequence in the hilbert space @xmath107 because @xmath108 when @xmath109 . this ensures that there exists a @xmath110 such that @xmath111 in the @xmath107-norm . combining the above results with lemma [ l : sobolev - diff - bound ] , we can conclude that @xmath112 which indicates that @xmath94 for almost all @xmath113 . let @xmath114 be the reproducing kernel hilbert space of a sobolev spline kernel @xmath115 with degree @xmath116 and shape parameter @xmath117 defined in example [ ex : sobolev - spline ] . since @xmath114 is equivalent to the sobolev space @xmath3 , the measurable spaces @xmath118 , where @xmath119 and @xmath69 are the borel @xmath70-algebras on @xmath114 and @xmath3 . thus @xmath120 belongs to @xmath121 for almost all @xmath95 . [ l : gauss - pdk ] suppose that @xmath1 is a regular bounded open domain of @xmath2 and the symmetric positive definite kernel @xmath24 for @xmath4 . then , for any fixed @xmath122 , there exists a probability measure @xmath123 defined on the measurable space @xmath124 such that the stochastic process @xmath125 is a gaussian field with mean @xmath126 and covariance kernel @xmath29 on the probability space @xmath127 . according to lemma [ l : gauss - sobolev ] almost all the sample paths of the gaussian field @xmath37 defined in the formula belong to @xmath121 . lemma [ l : gauss - rkhs ] provides that the probability measure induced by @xmath37 @xmath128 is well - defined on @xmath129 , and @xmath130 has the same probability distribution as @xmath37 . this shows that @xmath130 is a gaussian field with mean @xmath93 and covariance kernel @xmath29 on the probability space @xmath131 . because of the fact that @xmath122 , we have @xmath132 and @xmath133 for all @xmath134 . we can set up the another probability measure @xmath135 by shifting @xmath136 for @xmath126 , i.e. , @xmath137 since @xmath138 and @xmath139 on @xmath131 for @xmath92 , the stochastic process @xmath130 is a gaussian field with mean @xmath126 and covariance kernel @xmath29 on the probability space @xmath140 . using lemma [ l : gauss - pdk ] we can set up many other kinds of gaussian fields with respect to differential and boundary operators on the above probability space @xmath140 . [ t : gauss - pdk - pb ] suppose that @xmath1 is a regular bounded open domain of @xmath2 and the symmetric positive definite kernel @xmath24 for @xmath4 . let @xmath77 and @xmath78 be the linear differential and boundary operators of orders @xmath141 and @xmath142 , respectively , ( see the equations ( [ eq : diff]-[eq : bound ] ) ) . then , for any fixed @xmath143 , there exists a probability measure @xmath123 ( independent of @xmath77 and @xmath78 ) defined on the measurable space @xmath124 such that the stochastic processes @xmath144 are the gaussian fields with means @xmath145 , @xmath146 and covariance kernels @xmath147 , @xmath148 on the probability space @xmath127 , respectively . denote that @xmath149 where @xmath130 is the gaussian field given in lemma [ l : gauss - pdk ] . because @xmath150 and @xmath151 . if we can verify that @xmath152 and @xmath153 are the gaussian fields with means @xmath93 and covariance kernels @xmath147 and @xmath148 , then the proof is completed . since @xmath154 is a gaussian field with mean @xmath93 and covariance kernel @xmath29 defined on the probability space @xmath127 , the karhunen representation theorem provides that @xmath155 , where the random variables @xmath85 on @xmath127 . according to lemma [ l : sobolev - diff - bound ] , @xmath152 and @xmath153 have the representations @xmath156 and @xmath157 . combining the expansions of @xmath152 and @xmath153 with lemma [ l : rk - diff - bound ] , we can solve the means and covariance kernels of @xmath152 and @xmath153 , i.e. , @xmath158 and @xmath159 since @xmath24 , we have @xmath160 and @xmath161 . according to the above deductions , we can conclude that @xmath152 and @xmath153 are gaussian with means @xmath93 and covariance kernels @xmath147 and @xmath148 , respectively . the construction of gaussian fields in theorem [ t : gauss - pdk - pb ] is analogous to the form of wiener measure defined on the measurable space @xmath162 , called canonical space , such that the coordinate mapping process @xmath163 is a brownian motion . why do we need these kinds of gaussian fields ? because they help us to produce the normal random variables associated to the differential and boundary operators of spdes and the given collocation points located in the space domain and its boundary . using their joint probability density functions and conditional probability density functions , we can obtain the kernel - based solutions to fit the observation values which are simulated by the spdes . we choose any linear differential and boundary operators @xmath79 , @xmath81 and @xmath80 , @xmath82 of orders @xmath164 , @xmath165 and @xmath166 , @xmath167 , respectively . in the same manner of the proof of theorem [ t : gauss - pdk - pb ] , we can compute their covariances : @xmath168 , @xmath169 and @xmath170 . next we consider the vector linear differential and boundary operators @xmath171 composed of the finite linear differential and boundary operators of the orders @xmath172 and @xmath173 , respectively , where their orders are denoted by @xmath174 given the pairwise distinct collocation points @xmath175 we can use the gaussian fields @xmath176 and @xmath177 to create the related multi - normal vector @xmath178 on @xmath140 by theorem [ t : gauss - pdk - pb ] , where @xmath179 the mean and covariance matrix of @xmath180 can be computed by the same method as above . [ c : mean - cov - spbx ] the multi - normal vector @xmath180 given in the equation defined on the probability space @xmath140 has the mean @xmath181 and the covariance matrix @xmath182 where @xmath183 we let the linear operator @xmath184 be a differential operator @xmath77 of order @xmath141 or a boundary operator @xmath78 of order @xmath142 , i.e. , @xmath185 or @xmath186 . we fix any data point @xmath87 or @xmath187 corresponding to the operator @xmath184 , i.e. , when @xmath188 then @xmath87 , or when @xmath189 then @xmath187 . since theorem [ t : gauss - pdk - pb ] shows that the random variable @xmath190 and the random vector @xmath180 are both normal on @xmath140 , the conditional probability density function @xmath191 of @xmath190 given @xmath180 has the the explicit form @xmath192 , where @xmath193 and @xmath194 are the joint probability density functions of @xmath195 and @xmath180 . [ c : conpdf - lsx - spbx ] the conditional probability density function of the random variable @xmath190 given the random vector @xmath180 defined on the probability space @xmath140 ( discussed as in the above paragraph ) has the form @xmath196 for @xmath197 and @xmath198 , with the mean @xmath199 and the variance @xmath200 where @xmath201 and @xmath202 ( here @xmath203 and @xmath55 are the mean and covariance matrix of @xmath180 solved in corollary [ c : mean - cov - spbx ] . ) in particular , for the given observation values @xmath204 , the probability density function of @xmath190 given @xmath205 is equal to @xmath206 . since the covariance matrix @xmath55 is always semi - positive definite , its pseudo - inverse @xmath207 is well - behaved . we observe that @xmath54 is the kernel basis of kernel - based solutions , and the variance function @xmath208 , which is equal to the power function , is used to estimate the error bound . according to theorem [ t : gauss - pdk - pb ] and fubini s theorem we can use the gaussian field @xmath130 defined on @xmath140 to obtain that @xmath209 for all @xmath99 which indicates that @xmath210 the markov s inequality provides that @xmath211 according to the construction of the gaussian field @xmath130 defined on @xmath212 , we can introduce the following corollary . [ c : pdk - gauss - norm ] suppose that the probability space @xmath140 defined same as in lemma [ l : gauss - pdk ] . for any @xmath213 , the subsets @xmath214 have the probabilities @xmath215 in this section , we employ the similar techniques as in @xcite to introduce the kernel - based solutions of the elliptic spdes and their convergent rates . let @xmath1 be regular bounded open domain of @xmath2 , @xmath4 and the vector linear differential and boundary operators @xmath216 and @xmath217 of the orders @xmath172 and @xmath173 , whose elements defined as in the equations ( [ eq : diff]-[eq : bound ] ) with the coefficients @xmath218 and @xmath219 are linear independent . denote that @xmath220 such that @xmath221 is bijective for any fixed @xmath197 and @xmath222 . suppose that the vector noise @xmath223 composes of finite independent stochastic processes @xmath224 defined on the probability space @xmath225 for @xmath226 . we consider a system of elliptic spdes driven by @xmath227 as follow : @xmath228 where @xmath229 and @xmath230 are the deterministic functions for @xmath226 and @xmath231 . suppose that the spde is well - posed for its differential and boundary operators . thus , when @xmath232 and @xmath233 are convergent to the left - hand sides of the spde , then the estimator @xmath234 is convergent to the exact solution @xmath235 of the spde in the same rate , e.g. , the maximum principle for the laplace differential operator gives @xmath236 for the heat spdes with dirichlet boundary conditions . ( here the notation means that @xmath237 if there is a positive constant @xmath238 such that @xmath239 . ) further suppose that the solution @xmath235 belongs to the sobolev space @xmath3 almost surely . we choose the sets of pairwise distinct collocation points @xmath240 from the domain @xmath1 and its boundary @xmath241 , i.e. , @xmath242 the _ fill distance _ of @xmath243 for @xmath1 and @xmath241 is denoted by @xmath244 where @xmath245 let @xmath246 be the joint probability density functions of the random vectors @xmath247 for all @xmath226 . same as the techniques in section [ sec : simulation - random ] , we can be simulated @xmath248 by their joint probability density functions @xmath249 . since @xmath248 are independent and @xmath250 is a bijective function for any fixed @xmath251 and @xmath222 , we can use @xmath249 to obtain the joint probability density function of @xmath252 . next we find an appropriate positive definite kernel @xmath63 to introduce the kernel - based solutions of the spdes . the covariance matrix @xmath55 given in the equation ( see corollary [ c : mean - cov - spbx ] ) will be used to evaluate the coefficients of the kernel - based solutions later . because we need @xmath55 to be nonsingular . we suppose that the symmetric positive definite kernel @xmath24 satisfies the condition @xmath253 where @xmath254 is the point evaluation functional at @xmath255 . according to [ theorem 16.8 , @xcite ] , the condition ensures that all covariance kernels @xmath256 and @xmath257 are positive definite on @xmath1 and @xmath241 , respectively , for @xmath226 and @xmath231 . the linearly independent condition even indicates that the covariance matrix @xmath55 is positive definite . so the inverse of @xmath55 exists and @xmath258 . @xcite show that that @xmath259 can contain enough more polynomials . one technique to verify the condition is to find a polynomial @xmath260 such that @xmath261 if and only if @xmath262 for any finite pairwise distinct collocation points @xmath243 . as a special case , the kernel function @xmath29 has the form @xmath263 for a positive definite function @xmath264 and all coefficients of the differential and boundary operators of the spde are scalars , then [ corollary 16.12 , @xcite ] provides that the condition is always true . the main reason of the condition is to let the covariance matrix @xmath55 be always nonsingular so that the system of linear equations is uniquely solvable for any choice of collocation points @xmath243 . this means that the condition can be replaced to choose well - distributed collocation points @xmath243 dependent of the differential and boundary operators of the spde such that @xmath55 is nonsingular . actually , we may not need the condition to obtain the kernel - based solutions because we could solve the linear system by the least square method for the semi - positive definite matrix @xmath55 . because we want to employ the theorems of the power functions given in @xcite to introduce the convergent rate directly . in this paper we always assume that the kernel function @xmath29 satisfies the condition in order to avoid the technical discussions and reproof of the similar theorems in @xcite . but we can still introduce the similar convergent rate as the techniques of the proof of [ theorem 14.4 and 14.5 , @xcite ] without the condition . since @xmath24 , [ theorem 10.45 , @xcite ] shows that the reproducing kernel hilbert space @xmath265 . using theorem [ t : gauss - pdk - pb ] , for any fixed @xmath266 , we can create a probability measure @xmath135 on the measurable space @xmath267 such that the stochastic processes @xmath268 are the gaussian fields with means @xmath269 , @xmath270 and covariance kernels @xmath256 , @xmath257 defined on the probability space @xmath127 , for @xmath226 and @xmath231 . since we have two different kinds of probability spaces @xmath127 and @xmath271 , it is necessary for us to combine them into a new product probability space . we define the tensor product probability space @xmath272 and let all the original random variables be extended in the natural way : if the random variables @xmath273 and @xmath274 are defined on @xmath127 and @xmath271 , respectively , then their extensions @xmath275 the extensions @xmath276 and @xmath277 preserve the original probability distributions , and they are also independent on the product probability space @xmath278 . this means that the gaussian fields induced by the chosen positive definite kernels and the noise terms of spdes can be extended to the product probability space while preserving the original probability distributional properties , and moreover , their extensions are independent , e.g. , the extensions of the gaussian field @xmath279 and the noise @xmath280 for @xmath281 are independent on @xmath278 . in addition , since the solution @xmath235 of spdes can be seen as a mapping from @xmath282 into @xmath283 , we have @xmath284 for all @xmath285 . for any @xmath286 , we let @xmath287 we simulate the observation values of the right - hand side of the spde at the collocation points @xmath243 , i.e. , @xmath288 and @xmath289 denote that @xmath290 we let @xmath291 where the random vector @xmath180 induced by the gaussian fields @xmath292 and @xmath293 at the collocation points @xmath243 is defined in the equation . the approximate probability measure @xmath294 is used to set up the kernel - based solution @xmath234 to approximate the exact solution @xmath235 of the spde . let @xmath295 for @xmath197 and @xmath87 . using the same techniques as in @xcite , the _ kernel - based approximate solution _ @xmath296 is a global maximizer of the conditional probability , i.e. , @xmath297 where @xmath298 is the conditional probability density function of the random variable @xmath299 given @xmath180 defined on @xmath278 ( see corollary [ c : conpdf - lsx - spbx ] ) , and the kernel basis @xmath54 is given in the equation . here we can think @xmath300 and @xmath301 as the fixed values to find the approximate mean . the kernel - based solution @xmath234 can be also written as the linear combination of the kernel basis @xmath302 and its random expansion coefficients @xmath303 are solved by the random linear equations @xmath304 it is obvious that @xmath234 satisfies the interpolation conditions at the collocation points almost surely , i.e. , @xmath305 and @xmath306 , and moreover , @xmath307 for all @xmath285 . because the random part of @xmath296 is only associated to its random expansion coefficients uniquely solved by the random vector @xmath300 and the deterministic vector @xmath301 . we can formally rewrite @xmath308 into @xmath309 . this means that @xmath296 can be seen as a random variable defined on the finite - dimensional probability space @xmath310 in order that its probability distribution is the same as the original version , where the probability measure @xmath311 and @xmath312 is the joint probability density function of @xmath300 . now we propose to describe the convergence of the kernel - based solutions . let @xmath255 belong to @xmath1 or @xmath241 corresponded to the linear operator @xmath184 , which is equal to the differential operator @xmath313 for @xmath314 or the boundary operator @xmath315 for @xmath222 , i.e. , when @xmath316 then @xmath87 , or when @xmath317 then @xmath187 . we define @xmath318 where the kernel - based solution @xmath234 and the observation values @xmath319 induced by the linear spde are the same as in section [ sec : lin - ell - spde - ker - sol ] . we fix any @xmath320 . according to section [ sec : powerfun ] , the conditional mean @xmath321 of the conditional probability density function @xmath322 stated in corollary [ c : conpdf - lsx - spbx ] has the uniformly representation independent of @xmath255 , i.e. , @xmath323 when the fill distance @xmath324 is enough small , then the chebyshev s inequality provides @xmath325 where @xmath326 is the direct delta function at @xmath301 and @xmath312 is the joint probability density function of @xmath300 . since the variance function @xmath327 is equal to the power function @xmath328 stated in section [ sec : powerfun ] , we have @xmath329 because @xmath330 if and only if @xmath331 for all @xmath332 . we can find a @xmath333 and a @xmath334 such that @xmath335 for all @xmath226 and @xmath231 . this indicates that @xmath336 since the spde is well - posed for its differential and boundary operators , we conclude that : [ p : prob - convergence - linear - ell - spde ] the kernel - based approximate solution @xmath234 given in the equations ( [ eq : ker - based - sol - linear - ell]-[eq : ker - based - sol - coef - linear - ell ] ) is convergent to the exact solution @xmath235 of the spde in the probability @xmath337 for all @xmath266 when the fill distant @xmath324 tends to @xmath93 , i.e. , for any @xmath266 and any @xmath338 , @xmath339 where @xmath340 . [ r : error - linear - spde ] more precisely , the convergent rate can be represented as @xmath341 which indicates that one choice of the optimal designs of @xmath38 and @xmath342 has the form @xmath343 . actually the error bounds of the kernel - based estimators can be also described in terms of the number @xmath344 of the collocation points @xmath243 and the dimension @xmath345 of the domain space @xmath1 , and moreover , the worse - case errors for some special kernel functions can be even dimension - independent and decay as a polynomial in terms of @xmath344 as done in @xcite . for the deterministic problem , the maximum error of the kernel - based estimator is bounded by the power function . for the stochastic problem , we discuss the convergence of the kernel - based estimators by using probability measures , which means that the deterministic error bound is replaced by the confidence interval . the confidence interval can be computed by the variance function and it is employed to predict the error in the probability sense . the power function and the variance function have the same forms but they represent different mathematical meanings . we can also obtain the analogous stable error estimate for the both deterministic and stochastic problems . because the convergence in the probability implies the convergence in the distribution , i.e. , @xmath346 is pointwise convergent to @xmath347 when @xmath348 , where @xmath346 and @xmath347 are the cumulative distribution functions of @xmath296 and @xmath349 . according to proposition [ p : prob - convergence - linear - ell - spde ] the distributions of @xmath235 can be estimated by the distributions of @xmath234 . [ c : prob - convergence - linear - ell - spde ] let @xmath234 and @xmath235 be the same as discussed in proposition [ p : prob - convergence - linear - ell - spde ] . if @xmath350 is a continuous and bounded function defined on @xmath351 , then @xmath352 in particular , if @xmath353 , then @xmath354 let @xmath1 be a regular bounded open domain of @xmath2 , and @xmath355 be a levy process in the sobolev space @xmath3 for @xmath4 defined on the probability space @xmath356 with the form @xmath357 where @xmath7 is an orthonormal subset of @xmath8 , @xmath9 is a positive sequence and @xmath358 are the independent scalar lvy processes with triples @xmath359 for all @xmath12 such that @xmath360 ( see @xcite ) . when @xmath355 is a wiener process , then @xmath355 can be constructed by infinite countable orthonormal bases and independent standard scalar brownian motions , e.g. , @xmath355 is a wiener process in @xmath3 with mean zero and spatial covariance kernel function @xmath350 given by @xmath361 . suppose that the linear differential operator @xmath77 of order @xmath141 and the linear boundary operators @xmath315 of orders @xmath362 for all @xmath231 defined as in the equations ( [ eq : diff]-[eq : bound ] ) with the coefficients @xmath218 and @xmath219 are linear independent . given an initial condition @xmath363 , we consider a parabolic spde driven by @xmath355 @xmath364 where @xmath365 and @xmath366\times\partial{\mathcal{d}}\to{\mathbb{r}}$ ] are the deterministic functions for @xmath231 . suppose that the spde is well - posed and that its exact solution @xmath367 belongs to @xmath368 for all @xmath369 such that @xmath370 . we transform the parabolic spde into several elliptic spdes by the implicit euler schemes and solve these elliptic spdes using the kernel - based approximation method as in section [ sec : ell - spde - ker - sol ] . 1 . we discretize the equation by the implicit euler schemes at time @xmath371 , i.e. , @xmath372 where @xmath373 and @xmath374 for all @xmath53 . 2 . because the levy noise increment @xmath375 at each time instance @xmath34 is independent of the solution @xmath35 . we can view that @xmath376 is an unknown part and @xmath44 is a given part computed at the previous step . the equation together with the corresponding moving boundary condition becomes a well - posed elliptic spde of the form @xmath377 where @xmath40 and @xmath378 . + for solving the spde , we select the well - distributed pairwise distinct collocation points @xmath66 and @xmath379 . in addition , we simulate the noise @xmath380 at @xmath38 , i.e. , @xmath381 + next we choose a positive definite kernel @xmath24 which satisfies the condition related to @xmath382 and @xmath383 defined same as in section [ sec : ell - spde - ker - sol ] . the kernel - based solution @xmath384 of the spde is computed by the equations ( [ eq : ker - based - sol - linear - ell]-[eq : ker - based - sol - coef - linear - ell ] ) , i.e. , @xmath385 where @xmath386 here the kernel base @xmath54 and the covariance matrix @xmath55 induced by the kernel @xmath29 with @xmath387 and @xmath388 are defined in the equations and . + this means that @xmath389 is approximated by the kernel - based solution @xmath384 . 3 . repeat ( s2 ) for all @xmath53 . we briefly discuss the convergence of the above algorithm for the spde similar as done in @xcite . suppose that the distances of all discretization time steps are equal to @xmath390 , and that the collocation points @xmath243 and the positive definite kernel @xmath29 are chosen to be the same at each time step @xmath34 . according to theorem [ t : gauss - pdk - pb ] , we firstly set up a probability measure @xmath123 on the measurable space @xmath391 for @xmath266 . in the same ways of section [ sec : ell - spde - ker - sol ] we define a new product probability space such that the extensions of the white noise @xmath392 and the solution @xmath367 of the spde preserve the original probability distributions in the product probability space , i.e. , @xmath393 now we consider the global error bound of the estimator @xmath394 defined in the equation in the probability @xmath337 . because of @xmath15 , the it formula @xcite provides the local truncation error bound of the implicit euler scheme in the probability @xmath337 , i.e. , @xmath395 for all @xmath53 , where @xmath396 is dependent of the scalar lvy processes @xmath397 . moreover , proposition [ p : prob - convergence - linear - ell - spde ] provides the errors for each local elliptic spde @xmath398 where @xmath340 . denote the global errors @xmath399 combining the both local errors given in the equations and , we have @xmath400 where @xmath401 by induction we can deduce that @xmath402 if @xmath403 then the spectral radius @xmath404 of the matrix @xmath405 satisfies @xmath406 , where @xmath407 is dependent of @xmath29 and @xmath77 ( see @xcite ) . thus @xmath408 this indicates that @xmath409 is convergent to @xmath410 in the probability @xmath337 when both @xmath390 and @xmath324 tend to @xmath93 , for all @xmath411 . in this paper , we focus mainly on the step ( s2 ) to solve the elliptic spde same as in @xcite . the numerical analysis of deterministic parabolic pdes for the kernel - based approximation method is a delicate and nontrivial question , only recently solved in @xcite . we will address this question in the case of many other parabolic spdes in our future research . now we do a numerical test for the two - dimensional parabolic spde driven by the poisson noises with dirichlet boundary conditions a typical case of the spde . let the domain @xmath412 , and the poisson noise @xmath5 have the form @xmath413 where @xmath414 are the independent scalar poisson process with parameter @xmath11 for all @xmath415 . we use this poisson noise @xmath5 to create the parabolic spde @xmath416 where @xmath417 and @xmath418 . according to the same algorithm given in section [ sec : alg - tradition - par - spde ] we can compute the kernel - based solutions of the spde by two kinds of positive definite kernels : compact support kernel induced by @xmath419-compact support radial basis function @xmath420 and sobolev - spline kernel induced by @xmath419-matrn function @xmath421 where @xmath422 and the cutoff function @xmath423 when @xmath424 otherwise @xmath425 . as collocation points we select halton points @xmath38 in @xmath1 and evenly space points @xmath342 on @xmath241 . using the kernel - based approximation method , we can obtain thousand numerical sample paths @xmath426 for @xmath427 by the algorithm given in section [ sec : alg - tradition - par - spde ] . moreover , we can compute its sample means and sample standard deviations by these estimate sample paths , i.e. , @xmath428 observing figure [ fig : par - spde - num - exa ] , we find that the approximate means and the approximate standard deviations for both kernels are symmetric with the line @xmath429 because the time and space poisson noises @xmath5 are symmetric with @xmath430 and @xmath431 in space . compact support kernel , @xmath432 sobolev - spline kernel , @xmath433 in our current numerical experiments , the distribution of collocation points and the shape parameter are chosen empirically and based on the authors experiences . actually , different choices of the shape parameters will affect the convergent rate and the stability of the algorithms . the convergent rate will decrease when the shape parameter becomes large , but the algorithm will be unstable when the shape parameter becomes small . how to select the best shape parameter is still an open problem . we will try to solve this problem using the probability measures of the kernel functions in our future research . in this paper we present how to employ the kernel - based approximation method to estimate the numerical solutions of spdes driven by the lvy noises . we transform the parabolic spde into the elliptic spde by the implicit euler time scheme at each time step . the kernel - based solution of the elliptic spde is a linear combination of the kernel basis with the related differential and boundary operators centered at the chosen collocation points . here we only consider the elliptic spdes with right - hand - side random noises . actually , the kernel - based approximation method can be even applied into solving the spde driven by the random differential and boundary operators as done in @xcite . the main idea of this paper is the same as in our recent paper @xcite but we give the new contents and extensions to improve the previous theoretical results as follow : * we extend [ theorem 3.1 , @xcite ] into theorem [ t : gauss - pdk - pb ] in order that we can apply more general positive definite kernels to set up the kernel - based approximate solutions of spdes instead of the integral - type kernels , and we only need to assume that the exact solutions of spdes belong to the classical sobolev spaces rather than the reproducing kernel hilbert spaces . * we obtain the kernel - based approximate solutions of a system of linear elliptic spdes driven by various kinds of random noises undone before . * we complete the discussions of the error analysis of the kernel - based approximation method for spdes and provide its precise convergent rate in terms of the fill distances ( or possible the time distances ) . comparing the kernel - based approximation method for solving the deterministic and stochastic elliptic pdes similar as in @xcite : * the kernel - based solution for the deterministic case is to minimize the reproducing kernel norm for interpolating the data values induced by the pdes , and the kernel - based solution for the stochastic case is to maximize the probability conditioned on the observation values simulated by the spdes . * the estimate error is bounded by the power function for the deterministic case , and the confident interval is computed by the variance function for the stochastic case , and moreover , the formulas of the power function and the variance function are equal when the pde and the spde have the same differential and boundary operators . we also discuss the side - by - side differences of the kernel - based approximation method and the finite element method for the elliptic spde stated as in @xcite : * for the kernel - based approximation method we transfer the original spde probability space to the tensor product probability space such that the extension of the noise preserves the same probability distributions . for the finite element method , we approximate the noise by its truncated noise which means that we truncate the original spde probability space to the finite dimensional probability space . * the bases of the kernel - based solution are set up by the positive definite kernels with the differential and boundary operators of the spde and the collocation points , while the the bases of the finite element solution are the finite element polynomials induced by the triangular meshes . * we can simulate the noise at the collocation points by its probability structure to compute the random coefficients of the kernel - based solution , but we can simulate the random part of the finite element solution on the truncated probability space . * the convergent rate of the kernel - based solution is only dependent of the fill distances . however , the convergent rate of the finite element solution depends on the maximum mesh spacing parameter of the triangulation and the truncation dimension of the original probability space . in our future work we will try to solve the open problems of the kernel - based approximation method for spdes : * we will solve the kernel - based estimators of the nonlinear elliptic spdes based on the theoretical results given in this paper . * we will try many other time - stepping schemes to create the kernel - based solutions of the parabolic spdes , and introduce the precise rates of their convergence . * we will design the best choice of the collocation points and the optimal kernel function by maximizing the conditional probability measure dependent of the observation values simulated by the spdes analogous as the maximum likelihood estimation method . in this section , we review some classical materials of linear differential and boundary operators mentioned in @xcite . let @xmath1 be a regular bounded open domain of @xmath2 , e.g. , it satisfies the uniform @xmath434-regularity condition which implies the strong local lipschitz condition and the uniform cone condition . this means that @xmath1 has a regular boundary @xmath435 . we call the the partial derivative of order @xmath436 @xmath437 and denote its degree @xmath438 . the _ test function space _ is chosen to be @xmath439 composed of all functions @xmath440 with compact support in @xmath1 . for any fixed @xmath441 , if there exist a function @xmath442 such that @xmath443 then @xmath444 is said the _ weak derivative _ @xmath445 of @xmath446 . the @xmath0-based sobolev space of degree @xmath447 is defined by @xmath448 equipped with the inner product @xmath449 because the weak derivative @xmath450 can be seen as a linear bounded operator from @xmath451 when @xmath452 or a linear bounded operator from @xmath453 when @xmath454 according to the boundary trace embedding theorem . in this paper all linear differential and boundary operators are the linear combinations of weak derivatives with uniformly continuous coefficients . we define that the linear differential operator @xmath455 for all @xmath456 has the order @xmath457 and the linear boundary operator @xmath458 for all @xmath459 has the order @xmath460 it is easy to check that @xmath77 and @xmath78 are linear and bounded . [ l : sobolev - diff - bound ] suppose that a sequence @xmath461 and a function @xmath462 in the @xmath8-norm . if there exist a function @xmath463 such that @xmath464 in the @xmath8-norm for any fixed @xmath99 , then @xmath465 . this indicates that @xmath466 in the @xmath8-norm and @xmath467 in the @xmath468-norm for any differential and boundary operators @xmath77 and @xmath78 of orders @xmath141 and @xmath142 , respectively . fixing @xmath99 , we have @xmath469 which shows that @xmath444 is equal to the weak derivative @xmath445 of @xmath446 . therefore @xmath465 and @xmath470 in the @xmath3-norm . most of the detail presented in this section can be found in the monograph @xcite . for the reader s convenience we repeat here what is essential to the kernel - based approximation method . [ d : rkhs ] a hilbert space @xmath259 consisting of functions @xmath471 is called a _ reproducing kernel hilbert space _ and a kernel function @xmath63 is called a _ reproducing kernel _ for @xmath259 if @xmath472 for all @xmath473 and @xmath474 , where @xmath475 is used to denote the inner product of @xmath259 . according to [ theorem 10.4 , @xcite ] the reproducing kernel @xmath29 is always _ semi - positive definite_. moreover , [ theorem 10.10 , @xcite ] guarantees the existence of the reproducing kernel hilbert space with the positive definite kernel @xmath29 . suppose that the symmetric positive definite kernel @xmath476 and @xmath1 is a regular bounded open domain of @xmath2 . since @xmath477 is compact , the mercer s theorem shows that there exists a countable set of positive _ eigenvalues _ @xmath478 and continuous _ eigenfunctions _ @xmath479 such that @xmath74 and the kernel @xmath29 has the absolutely and uniformly convergent representation @xmath480 . furthermore , @xmath73 is an orthonormal basis of @xmath8 and @xmath481 . [ l : rk - diff - bound ] if the symmetric positive definite kernel @xmath24 , then its eigenfunctions @xmath482 . moreover , the convergent representations @xmath483 and @xmath484 are absolute and uniform on @xmath102 and @xmath485 for any differential and boundary operators @xmath77 and @xmath78 of orders @xmath141 and @xmath142 defined as in equations ( [ eq : diff]-[eq : bound ] ) , respectively . because of @xmath24 , @xmath486 for any @xmath487 . this indicates that the convergent representation @xmath488 is absolute and uniform on @xmath102 for any @xmath99 . now we show a special class of reproducing kernels whose reproducing kernel hilbert spaces are equivalent to the sobolev spaces . [ ex : sobolev - spline ] we consider the matrn function of degree @xmath4 and shape parameter @xmath422 @xmath489 where @xmath490 is the modified bessel function of the second kind of order @xmath491 . according to the theoretical results given in @xcite , we can check that the sobolev spline kernel induced by the matrn function @xmath492 @xmath493 is a positive definite kernel on @xmath2 and its reproducing kernel hilbert space @xmath494 is equivalent to the sobolev space @xmath495 , i.e. , @xmath496 let @xmath1 be a regular domain of @xmath2 . according to [ theorem 6 , @xcite ] , the reproducing kernel hilbert space @xmath114 of the sobolev spline kernel @xmath115 restricted on @xmath1 is endowed with the reproduction norm @xmath497 and it is also equivalent to the sobolev space @xmath3 , i.e. , @xmath498 the papers @xcite also show many other general kinds of reproducing kernels and their related reproducing kernel hilbert spaces are introduced by green functions and generalized sobolev spaces . let @xmath499 be linear differential operators of orders no more than @xmath116 , whose coefficients @xmath218 , and @xmath383 be linear boundary operators of orders no more than @xmath500 , whose coefficients @xmath219 . denote that @xmath216 and @xmath501 . we choose the set of pairwise distinct collocation points @xmath502 from a regular bounded open domain @xmath1 and its boundary @xmath241 , i.e. , @xmath503 and @xmath504 . suppose that the symmetric positive definite kernel @xmath24 satisfies the condition related to @xmath505 and @xmath506 denoted as in section [ sec : ell - spde - ker - sol ] . according to [ theorem 16.8 , @xcite ] , @xmath507 belongs to the dual space @xmath508 of the reproducing kernel hilbert space @xmath259 , where @xmath184 is a differential or boundary operator of order @xmath509 and @xmath510 and @xmath254 is the point evaluation functional at @xmath87 or @xmath187 . the _ power function _ induced by the positive definite kernel with the differential and boundary operators at the collocation points is defined by @xmath511 where the matrix @xmath55 and the vector @xmath512 are denoted in the equations and respectively . we can observe that the power function @xmath328 is equal to the formula of the variance @xmath513 defined in the equation . [ theorem 16.11 , @xcite ] provides that @xmath514 as the discussion of [ section 16.3 , @xcite ] we have @xmath515 where @xmath516 and @xmath517 . combining [ theorem 11.3 and 16.9 , @xcite ] the power functions can be also bounded by the fill distances of the collocation points @xmath38 and @xmath342 , respectively , @xmath518 i.e. , @xmath519 where @xmath520 and @xmath521 are the positive constants independent of @xmath255 . the precise form of the fill distance of @xmath241 is equal to @xmath522 using the distance defined on the manifold surface @xmath241 . since the boundary @xmath241 is regular , we have @xmath523 . for convenience we does not consider the manifold distance in this article . in this section we discuss the basic relationship of gaussian fields and reproducing kernel hilbert spaces given in @xcite . [ d : gaussian ] let @xmath64 . a stochastic process @xmath524 is said to be _ gaussian _ with mean @xmath525 and covariance kernel @xmath63 on a probability space @xmath526 if , for any pairwise distinct points @xmath66 , the random vector @xmath527 is a multi - normal random variable on @xmath528 with mean @xmath529 and covariance matrix @xmath530 , i.e. , @xmath531 where @xmath532 and @xmath533 . suppose that @xmath259 is a reproducing kernel hilbert space and @xmath534 is the borel @xmath70-algebra on @xmath259 . let @xmath535 be a gaussian field defined on @xmath526 . if the sample paths @xmath120 belong to @xmath259 for almost all @xmath536 , then @xmath37 can be seen as a mapping from @xmath537 into @xmath538 . [ l : gauss - rkhs ] suppose that @xmath535 is a gaussian field on a probability space @xmath526 with almost all sample paths in a reproducing kernel hilbert space @xmath259 . then the probability measure given by @xmath539 is well - defined on the measurable space @xmath540 such that the stochastic process defined by @xmath541 is a gaussian field on the probability space @xmath542 and @xmath130 has the same probability distribution as @xmath37 , i.e. , the both gaussian fields @xmath130 and @xmath37 have the same mean and covariance kernel . ( here the reproducing kernel @xmath29 may be different from the covariance kernel . ) lemma [ l : gauss - rkhs ] shows that we can transfer the original probability space @xmath543 into the new probability space @xmath542 so that the original gaussian field @xmath37 has the invariant element @xmath130 defined on the new probability space . for the kernel - based collocation methods , we need to simulate the spde noise term @xmath544 defined on the spde probability space @xmath225 at the collocation points @xmath545 . for example , if the noise @xmath37 is equal to the product of a poisson random variable @xmath546 with parameter @xmath11 and a deterministic function @xmath547 , i.e. , @xmath548 , then @xmath549 for all @xmath52 , which can be simulated by the monte carlo methods ( see @xcite ) . furthermore , we can simulate many other kinds of noises by the monte carlo methods if the joint probability density function @xmath550 of the random vector @xmath551 is known . if there is an one - to - one function @xmath552 such that @xmath553 where @xmath554 is the jacobian matrix of the inverse @xmath555 of @xmath556 evaluated at @xmath557 , then we can simulate @xmath558 by the independent strand uniform random variables @xmath559 , i.e. , @xmath560 for all @xmath411 . the vector function @xmath561 can be computed by @xmath562 where @xmath563 is the cumulative distribution function of @xmath564 and @xmath565 is the cumulative distribution function of @xmath566 given @xmath567 , etc . , i.e. , @xmath568 the author would like to express his gratitude to prof . igor cialenco ( chicago ) and my advisor , prof . gregory fasshauer ( chicago ) for their guide and assistance of this research topic at illinois institute of technique , chicago . the author would also like to thank the following people for their helpful suggestion and discussion : prof . uday banerjee ( syracuse ) , prof . michael griebel ( bonn ) , prof . klaus ritter ( kaiserslautern ) , prof . ian sloan ( sydney ) and prof . xu sun ( wuhan ) . g. e. fasshauer , f. j. hickernell and h. woniakowki , average case approximation : convergence and tractability of gaussian kernels , _ monte carlo and quasi - monte carlo methods 2010 _ , eds . l. plaskota and h. woniakowski ( springer - verlag , 2012 ) , pp . 329344 . g. e. fasshauer and q. ye , reproducing kernels of sobolev spaces via a green kernel approach with differential operators and boundary operators , _ adv . _ * 38 * ( 2013 ) 891921 . g. e. fasshauer and q. ye , kernel - based collocation methods versus galerkin finite element methods for approximating elliptic stochastic partial differential equations , in _ meshfree methods for partial differential equations vi _ , eds . m. griebel and m. a. schweitzer ( springer - verlag , 2013 ) , pp . 155 - 170 . g. e. fasshauer and q. ye , a kernel - based collocation method for elliptic partial differential equations with random coefficients , in _ monte carlo and quasi - monte carlo methods 2012 _ , eds . j. dick , f. y. kuo , g. w. peters and i. h. sloan ( springer - verlag , 2013 ) , to appear . f. y. kuo , c. schwab and i. h. sloan , quasi - monte carlo finite element methods for a class of elliptic partial differential equations with random coefficient , _ siam j. numer . * 50 * ( 2012 ) 33513374 .	in this paper , we improve and complete the theoretical results of the kernel - based approximation ( collocation ) method for solving the high - dimensional stochastic partial differential equations ( spdes ) given in our previous papers . according to the extended theorems , we can use more general positive definite kernels to construct the kernel - based estimators to approximate the numerical solutions of the spdes . because a parabolic spde driven by lvy noises can be discretized into serval elliptic spdes by the implicit euler scheme at time . we mainly focus on how to solve a system of elliptic spdes driven by various kinds of right - hand - side random noises . the kernel - based approximate solution of the elliptic spdes is a linear combination of the positive definite kernel with the differential and boundary operators of the spdes centered at the chosen collocation points , and its random coefficients are obtained by solving a system of random linear equations , whose random parts are simulated by the elliptic spdes . moreover , we introduce the error bounds confident intervals of the kernel - based approximate solutions of the elliptic ( parabolic ) spdes in terms of fill distances ( or possible time distances ) in the probability sense . we also give a well coding algorithm to compute the kernel - based solutions of the second - order parabolic spdes driven by time and space poisson noises . the two - dimensional numerical experiments show that the approximate probability distributions of the kernel - based solutions are well - behave for the sobolev - spline kernels and the compact support kernels . * keywords : * kernel - based approximation ( collocation ) method ; meshfree approximation method ; high dimension ; stochastic partial differential equation ; positive definite kernel ; gaussian field ; lvy noise ; poisson noise . ams subject classification : 46e22 , 65d05 , 60g15 , 60h15 , 65n35
the decay constant of the @xmath2 meson , @xmath3 , is of fundamental importance to tests of the standard model because it is needed in the determination of the ckm matrix . while one of the simplest weak matrix elements to study , it is not very well known and dominates the uncertainty in extracting @xmath6 , one of the smallest ckm elements , from @xmath7 mixing . the best chance of an experimental measurement of @xmath3 lies in the decay @xmath8 . however , this decay has yet to be observed , and @xmath3 appears in combination with @xmath9 , which is only known to within a factor of 2 . theoretical calculations using for example qcd sum rules within the framework of heavy quark effective theory ( hqet ) @xcite and lattice field theory have been actively pursued over the last few years with the current world average standing at @xmath10 mev @xcite . as a first principles approach , lattice simulations offer ultimately the most reliable calculation of @xmath3 . the initial problem of simulating a @xmath11 quark with compton wavelength of the order of the typical lattice spacing presently achievable was first solved by using the static approximation @xcite , where the heavy quark is treated as a static colour source within the heavy - light meson . simulations using conventional relativistic methods around the @xmath12 meson showed the corrections to the static approximation in this region , and hence also at the @xmath2 meson , are much larger than naively expected . this has contributed to the sizable uncertainty remaining in @xmath3 , shown above . the determination of @xmath3 directly at @xmath13 on the lattice is required , and this is now possible with the development of non relativistic qcd ( nrqcd ) @xcite . using this approach the error in the decay constant can be significantly reduced with the computational power currently available . initial calculations have already been performed by hashimoto @xcite and davies @xcite . this analysis forms the second stage of a project applying nrqcd to heavy - light systems and studying heavy quark symmetry at the @xmath2 meson . our previous papers have dealt with the @xmath2 meson spectrum in both the quenched approximation @xcite and partially including the effects of sea quarks @xcite . we now present our results for the decay constants of the heavy - light mesons from dynamical configurations . note that the combination @xmath14 , where @xmath15 is the mixing parameter for the @xmath2 system , is the experimentally relevant quantity , and thus a calculation of @xmath15 is also needed . the paper is organised as follows . in section [ simdet ] we describe the simulation details including the systematic errors associated with the calculation . the meson operators required to compute the decay constant to @xmath1 are set out in section [ mesop ] and we discuss computing the decay constant consistently to this order in terms of the evolution equation used . next , we examine the predictions of heavy quark symmetry for the behaviour of the decay constant in the heavy quark mass limit . the contributions from the interactions of the heavy quark at @xmath1 are analysed and combinations of decay constants which isolate these individual contributions are discussed . section [ results ] illustrates our results and fitting analysis with a subset of the data . the heavy quark mass dependence of the decay constants are investigated in section [ hq_decay ] ; the individual non - perturbative @xmath1 coefficients of the slope of the decay constant are extracted . finally we present our conclusions in section [ conc ] . the simulations were performed using 100 @xmath16 gauge configurations at @xmath17 with two flavours of staggered dynamical sea quarks with a bare quark mass of @xmath18 . these configurations were generously made available by the collaboration ; more details can be found in @xcite . we fixed the configurations to the coulomb gauge . the light quark propagators were generated using the wilson fermion action , without an @xmath19 improvement term , at two values of the hopping parameter , @xmath20 and @xmath21 . the former corresponds to a quark mass close to strange , where @xmath22 from @xmath23 , while @xmath21 is somewhat lighter , with @xmath24 . with only two values of light quark mass it is not possible to perform a trustworthy chiral extrapolation . in addition , the systematic errors associated with the light quarks , discussed below , affect the determination of the bare strange quark mass ; @xmath25 extracted from pseudoscalar mesons , using the lowest order chiral mass dependence to find the ` experimental mass ' of the pure @xmath26 pseudoscalar , differs from @xmath25 obtained from the vector meson @xmath27 . thus , most of the results will be presented for @xmath28 . in this simulation we truncate the nrqcd series at @xmath29 , where @xmath30 is the bare heavy quark mass , and the action takes the form : @xmath31 where @xmath32 tadpole improvement of the gauge links is used throughout i.e. they are divided by @xmath33 , where we use @xmath34 measured from the plaquette , and the hyperfine coefficient is given the tree - level value @xmath35 . we use the standard clover - leaf operator for the b field . the heavy quark propagators were computed using the evolution equation @xcite : @xmath36 where @xmath37 is the stabilising parameter . in the static limit this reduces to @xmath38 we generated heavy quark propagators at 11 values of ( @xmath39,n ) corresponding to ( 0.8,5 ) , ( 1.0,4 ) , ( 1.2,3 ) , ( 1.7,2 ) , ( 2.0,2 ) , ( 2.5,2 ) , ( 3.0,2 ) , ( 3.5,1 ) , ( 4.0,1 ) , ( 7.0,1 ) and ( 10.0,1 ) , and the static limit . this roughly corresponds to a range of meson masses from @xmath40 to @xmath41 and is sufficient for a reasonable investigation of heavy quark symmetry . however , it is not possible to simulate the @xmath12 meson on this lattice using nrqcd ; a larger lattice spacing is required , @xmath42 or @xmath43 . details of the construction of meson operators , the smearing functions used and our fitting analysis can be found in @xcite . as discussed in this reference there are large systematic errors associated with this simulation arising from the use of wilson light fermions . the emphasis of our analysis will be on determining the heavy quark mass dependence of the decay constants and extracting the nonperturbative coefficients of the heavy quark expansion to @xmath1 . the systematic error arising from the light quark in these @xmath44 coefficients is naively estimated to be a relative error of @xmath45 for this ensemble ( where we take @xmath46 for these configurations ) . the truncation of the nrqcd series and the use of the tree - level value for the hyperfine coefficient introduce absolute errors of @xmath47 and @xmath48 respectively . both correspond to approximately @xmath49 errors in the value and @xmath50 errors in the slope of the decay constant at @xmath13 . the coefficients of the quadratic term in the heavy quark expansion can not be correctly determined in this analysis . the size of the systematic errors induced by the light quark action are indicated by the large uncertainty in the lattice spacing ; we use @xmath51 gev . this gives @xmath52 as corresponding to the bare @xmath11 quark mass . we are currently repeating our analysis using tadpole - improved clover light fermions @xcite . the pseudoscalar decay constant is defined by @xmath53 where @xmath54 at tree level and @xmath55 and @xmath56 represent 4-component light and heavy quark fields respectively . in this convention the experimental value of @xmath57 is @xmath58 mev . similarly , the vector decay constant is given by @xmath59 where @xmath60 at tree level . the calculation of these matrix elements , and thus decay constants , to @xmath1 in nrqcd requires several operators . at tree level these can be obtained by relating @xmath56 of full qcd to the 2-component nrqcd fields , @xmath61 and @xmath62 for heavy quarks and heavy antiquarks respectively , using the foldy - wouthuysen transformation . to @xmath1 , @xmath63 with the convention that our meson states be built out of a heavy quark and a light antiquark , the @xmath62 field will not contribute to the above matrix elements through @xmath1 , and we obtain @xmath64 where @xmath65 and @xmath66 for the axial current , and @xmath67 and @xmath68 for the vector current . @xmath69 and @xmath70 denote the upper and lower two components of @xmath55 respectively . our euclidean space @xmath71 matrices are chosen as , @xmath72 beyond tree level all operators with the same quantum numbers can mix under renormalisation . thus at @xmath73 the basis of operators for the axial current , is given by @xmath74 using translation invariance , the zero momentum currents involving @xmath75 and @xmath76 are identical on the lattice and comparing with equation [ foldycurr ] it is sufficient to calculate matrix elements of the tree level operators @xmath77 and @xmath75 . similarly , the basis of operators for the vector current , @xmath78 is reduced , for the numerical computation , to @xmath77 , @xmath79 and @xmath80 , where @xmath81 matrix properties have also been used . we computed the pseudoscalar and vector matrix elements corresponding to this minimal set of operators by inserting a smeared interpolating field with the same quantum numbers as @xmath77 at the source and all combinations of local currents at the sink for each meson . the perturbative calculation of the renormalisation factors needed to relate the matrix elements in nrqcd to the currents in full qcd has not yet been completed @xcite . thus , we use the static renormalisation factors @xcite and are restricted to the tree level @xmath1 corrections ; we omit @xmath82 for the vector meson . since this correction is of @xmath48 it is roughly the same size as that induced by unknown perturbative matching corrections in the coefficient @xmath83 , which are not included in the analysis . the renormalisation factors in nrqcd depend on the heavy quark mass and hence using the static value introduces an @xmath84 error into the @xmath1 coefficients extracted from our results . it is possible that the power divergences which appear in nrqcd may lead to significant non - perturbative contributions to the renormalisation factors @xcite . no clear indication of ` renormalon ' effects has been found in other quantities calculated perturbatively in nrqcd @xcite , however , we intend to compute the renormalisation factors non - perturbatively to check this . strictly , our analysis is not fully consistent to @xmath1 . note that the evolution equation given in equation [ evol ] induces an @xmath85 systematic error in the currents . this is because it does not apply the relativistic correction operator @xmath86 on the initial timeslice . since all subsequent timeslices include this operator the transfer matrix is consistent to @xmath1 . thus , it is only in the amplitude of the meson correlator , i.e. the decay constant , that is missing this @xmath85 contribution . a better but computationally more costly evolution equation is given by @xmath87 for all timeslices . since our use of the wilson action for light quarks also induces errors of this order of magnitude , we used equation [ evol ] in this project . in a comparison of these evolution equations using clover fermions for the light quarks @xcite we found the insertion of @xmath86 on the source timeslice introduces @xmath88 correction to @xmath89 and @xmath90 correction to @xmath91 around @xmath92 . considering the systematic errors inherent in this calculation and our statistical accuracy of @xmath93 , the omission of this correction will not significantly affect our findings . in the heavy quark limit the decay constant can be parameterised in terms of an expansion in @xmath0 : @xmath94 the coefficient @xmath95 is determined by nonperturbative contributions arising from the hyperfine interaction ( @xmath96 ) and the kinetic energy of the heavy quark ( @xmath97 ) , which appear in the nrqcd action , and the corrections to the current arising from the foldy - wouthuysen transformation . from first order perturbation theory in @xmath0 about the static limit : @xmath98 where @xmath99 and @xmath100 for pseudoscalar and vector mesons respectively . the integer factors in this equation are introduced for convenience when comparing with hqet . @xmath101 @xmath102 @xmath103 where @xmath104 represents the meson in the limit of infinite heavy quark mass , and equations [ gkin]-[cooreqn ] are tree level expressions . note that the kinetic and hyperfine terms contribute to the decay constant through the correction to the meson wavefunction . equation [ slope ] suggests that the individual contribution from each @xmath1 term can be obtained separately by taking appropriate combinations of the pseudoscalar and vector decay constants , @xmath105 and @xmath106 respectively . for clarity below , we denote @xmath107 with similar quantities defined for the vector particle by replacing ps with v. we also denote the spin - average of the decay constants , @xmath108 .expansions about the static limit to @xmath1 of combinations of the pseudoscalar and vector decay constant , with and without the current corrections . [ expan ] [ cols="^,^,^",options="header " , ] we presented a lattice study , partially including the effects of dynamical quarks , of the heavy - light pseudoscalar and vector decay constants using wilson light quarks and nrqcd heavy quarks . all @xmath1 terms in the nrqcd action and matrix elements were included ; an @xmath109 wave - function renormalization induced by the evolution equation is the same order as the light quark discretization errors . we find consistency between the extrapolation of the nrqcd result and the static case , as expected , although the static result is not very well determined . there are significant @xmath110 corrections to @xmath111 around @xmath13 , and this leads to a large systematic error in extracting @xmath3 ; the use of wilson light quarks introduces a systematic error of roughly the same magnitude . it is necessary to go to the next order in nrqcd and use an improved light quark action in order to reliably extract @xmath3 . we found the slope of the decay constants with @xmath0 to be significantly larger than that found in previous lattice simulations . we believe this is because we are closer to the static limit . in particular , for the first time we obtained the three @xmath1 contributions to the linear component of the slope separately . we found that @xmath97 dominates and thus is responsible for the large slope . good agreement is found between our results for @xmath97 and physical combinations of @xmath96 and @xmath112 and a prediction of qcd sum rules . the results show that the kinetic energy of the heavy quark gives rise to contributions to @xmath111 much greater than @xmath113 , contrary to naive expectations . the computations were performed on the cm-2 at scri . we thank the hemcgc collaboration for use of their configurations and light quark propagators . we would like to thank t. bhattacharya , r. gupta and a. kronfeld for useful discussions . this work was supported by shefc , pparc and the u.s . we acknowledge support by the nato under grant crg 941259 and the eu under contract chrx - ct92 - 0051 . 99 j. m. flynn and n. isgur , j. phys . * g18 * 1627 ( 1992 ) . m. neubert , phys . reports * 245 * 259 ( 1994 ) . c. allton , plenary talk presented at the _ international symposium on lattice field theory , melbourne . australia , 11 - 15 july 1995 _ , to appear in nucl . * b * ( proc . suppl . ) . s. narison , talk presented at the _ cracow epiphany conference on heavy quarks _ , montpellier preprint no : pm 95/05 , hep - ph-9503234 . e. eichten and b. hill , phys . lett . * b234 * 511 ( 1990 ) . a. duncan et al , phys . * d51 * 5101 ( 1995 ) . b. a. thacker and p. lepage , phys . rev * d43 * 196 ( 1991 ) . p. lepage et al , phys d 46 * 4052 ( 1992 ) . s. hashimoto , phys . * d50 * 4639 ( 1994 ) . c. t h. davies , nucl * b34 * ( proc . suppl . ) 437 ( 1994 ) a. ali khan et al , glasgow preprint no : gutpa/95/12/1 , scri preprint no : fsu - scri-95 - 121 , ohio preprint no : ohstpy - hep - t-95 - 026 , hep - lat-9512025 . to be published in phys . * d53. s. collins et al , scri preprint no : fsu - scri-96 - 13 , glasgow preprint no : gutpa/96/2/8 , ohio preprint no : ohstpy - hep - t-96 - 002 , edinburgh preprint no : edinburgh 96/1 , hep - lat-9602028 . submitted to phys . rev . d. note that the fit parameters for the multi - exponential fits differ slightly ( @xmath114 ) from the results presented in this reference . in that analysis a bootstrap procedure was used in calculating the covariance matrix . it was found , however , that the advantages of this procedure did not outweigh the computational effort involved and it is not implemented here . k. m. bitar et al , phys . d46 * 2169 ( 1992 ) . c. t. h. davies et al , phys . d * 50 * , 6963 ( 1994 ) . s. collins et al , in preparation . c. morningstar and j. shigemitsu , work in progress . e. eichten and b. hill , phys . lett . * b240 * 193 ( 1990 ) . m. crisafulli et al , nucl . * b457 * 594 ( 1995 ) . m. neubert , phys . * d46 * 1076 ( 1992 ) . ukqcd collaboration r. baxter et al , phys . rev . * d49 * 1594 ( 1994 ) . c. bernard et al , phys . * d49 * ( 1994 ) 2536 . t. onogi and j. n. simone , nucl . * b * ( proc . suppl . ) * 42 * 434 ( 1995 ) . m. j. booth , florida preprint no : ufift - hep-94 - 10 , hep - ph-9412228 . a. ali khan et al , in preparation .	we present a lattice investigation of the heavy - light meson decay constants using wilson light quarks and nrqcd heavy quarks , partially including the effects of dynamical sea quarks . we calculate the pseudoscalar and vector decay constants over a wide range in heavy quark mass and are able to perform a detailed analysis of heavy quark symmetry . we find consistency between the extrapolation of the nrqcd results and the static case , as expected . we find the slope of the decay constants with @xmath0 is significantly larger than naive expectations and the results of previous lattice calculations . for the first time we extract the non - perturbative coefficients of the slope arising from the @xmath1 heavy quark interactions separately and show the kinetic energy of the heavy quark is dominant and responsible for the large slope . in addition , we find that significant systematic errors remain in the decay constant extracted around the @xmath2 meson mass due to truncating the nrqcd series at @xmath1 . we estimate the higher order contributions to @xmath3 are approximately @xmath4 ; roughly the same size as the systematic errors introduced by using the wilson action for light quarks . fsu - scri-96 - 43 + gutpa/96/4/1 + ohstpy - hep - t-96 - 009 @xmath2 decay constants from nrqcd with dynamical fermions + * s. collins , u. m. heller , and j. h. sloan * + scri , florida state university , tallahassee , fl 32306 - 4052 , usa + * j. shigemitsu * + the ohio state university , columbus , oh 43210 , usa + * a. ali khan and c. t. h. davies@xmath5 * + university of glasgow , glasgow , g12 8qq , scotland pacs numbers : 12.38.gc , 12.39.hg , 13.20.he , 14.20.nd
the idea that newton s theory of gravitation , can be formulated in curved space has been recently analyzed by abramowicz and collaborators @xcite@xmath0 . in addition to considering newtonian gravitation in curved space , an equation linking the geometry of the 3-space with the matter sector was put forward @xcite@xmath0 . this is called an enhanced newtonian gravitation . in particular , this modification to newton s theory allows the construction of a wormhole space @xcite@xmath0 . wormholes in general relativity have been studied in several works see @xcite for example . in this paper , we report on the construction of a spherically symmetric wormhole @xcite in this enhanced newtonian gravitation . we further analyze the tidal effects that emerge from the gravitational forces and the curvature of space in this spherically symmetric newtonian wormhole space . the outline of the paper is as follows : in sec . [ fund ] we start by writing the equations that define the enhanced newtonian theory of gravitation . in sec . [ newtonworm ] we construct a static spherically symmetric newtonian wormhole , we find the gravitational field , the gravitational potential and the pressure of the fluid that supports the wormhole . then , we study test particle s motion in the wormhole geometry and gravitational field . finally we analyze the tidal forces exerted by the gravitational field and the curvature of space in two nearby particles . in sec . [ conc ] , we conclude and speculate on the possible existence of truly newtonian black holes in this enhanced newtonian theory of gravitation . the classical formulation of newton s theory of gravitation was constructed for an absolute 3-dimensional euclidean space . we see , however , that the set of equations that comprise newton s theory of gravitation are also well defined for curved space . indeed , abramowicz et al . @xcite recently proposed a formulation of newton s theory of gravitation in curved space . poisson s equation in curved space is given by @xmath1 where @xmath2 is the curved space metric , @xmath3 is the covariant derivative induced by the metric , the indices @xmath4 run as @xmath5 , and @xmath6 is the density of the matter . for static systems the continuity equation is trivially verified and the euler equation is simplified to @xmath7 where @xmath8 is the fluid s pressure . now , a possible enhancement to newton s theory of gravitation was proposed in @xcite where a relation between the geometry of space and the matter was introduced and given by @xmath9 where @xmath10 is the ricci scalar and @xmath11 is an arbitrary constant . the equations of motion of a test particle with mass @xmath12 subjected to a gravitational potential @xmath13 are given by newton s second law , @xmath14 where it was assumed that the inertial and gravitational mass of the test particle are equal . - define an enhanced newtonian gravitation . our purpose is to make use of eqs . - to construct a static , spherical symmetric wormhole space . let us then generically write the metric of the space as @xmath15 such that the ricci scalar @xmath10 is @xmath16}{r^{2}a\left(r\right){}^{2}}\,.\label{eq:5}\ ] ] now , to proceed we consider the expression for the mass density to be @xmath17 where @xmath18 has dimensions of mass density and @xmath19 of distance . substituting eqs . and in eq . we find the general form of the metric , for the imposed mass density , @xmath20 where @xmath21 is an integration constant and @xmath22 having found the generic form of the metric , we have now to restrict the values of the parameters @xmath21 , @xmath18 and @xmath11 to have a wormhole geometry . for such , we will follow @xcite and use an embedding diagram . so , in the euclidean embedding space the axially symmetric embedded surface can be , using cylindrical coordinates @xmath24 , uniquely described by a function @xmath25 . identifying the coordinates @xmath26 of the embedding euclidean space with the coordinates @xmath27 of the wormhole space , we find the following relation @xmath28 which can be used to study the properties of the space . from eq . and imposing the throat condition we find that the integration constant @xmath21 is given by @xmath29 and the range of values of the parameter @xmath30 is constrained to be @xmath31 . the same restriction to the parameter @xmath23 is found from the flare - out condition . substituting eq . in eq . we find @xmath32 where @xmath33 another restriction to have a wormhole geometry is that although near the wormhole s throat , @xmath34 , the radial coordinate @xmath35 is ill behaved , the proper radial distance must be finite . this condition implies a new restriction to @xmath23 @xmath36 where the critical value of @xmath23 is given by @xmath37 $ ] , and @xmath38 represents the infimum of the function in the specified range . this gives @xmath39 up to the third decimal place . comparing eq . with the restriction found from the throat condition , @xmath40 , we conclude that the product @xmath41 is then restricted by eqs . and . due to the range of values that the parameter @xmath23 might take , various distinct cases could be considered . here we shall be interested in the class of wormholes whose parameters @xmath18 , @xmath11 and * * @xmath19 obey @xmath42 together with @xmath43 , and so @xmath44 . now that we have found the metric of the newtonian wormhole we have to remove the coordinate singularity at @xmath34 . considering a new coordinate @xmath45 , defined as @xmath46 , such that @xmath47 , one finds that the metric of the space can be rewritten as @xmath48 it will also be useful to write the mass density in the new coordinates . it is given by the expression @xmath49 having defined the geometry of the space we can now study the gravitational potential of the newtonian wormhole by solving eq . . it is , however , considering the symmetries of the system , straightforward to solve eq . by using gauss s law for gravity and integrating this equation over a volume whose boundary are the surfaces of constant gravitational potential . defining the gravitational force field as @xmath50 with a comma denoting a partial derivative , we find @xmath51 where the mass within radius @xmath45 is given by @xmath52 defining the magnitude of the gravitational field as @xmath53 we give in fig . [ fig:1 ] its behavior as a function of the radial coordinate @xmath45 for various values of the parameter @xmath23 . magnitude of the gravitational force field for various values of the parameter @xmath54 as a function of the radial coordinate @xmath45 ( here assume @xmath55 ) . when @xmath56 the magnitude of the gravitational force field tends to infinity at some value of @xmath57 . ] using the expression for the gravitational field , eq . together with eq . , the gravitational potential @xmath13 can be computed at a given point using numerical methods , provided the wormhole parameters are given . moreover , these equations can also be used to find the pressure of the fluid that covers the wormhole space . numerically solving eqs . and we find that the pressure is positive throughout the space showing that the wormhole is hold against gravitational collapse by pressure @xcite@xmath0 . from eq . ( [ gravforce ] ) one sees that the gradient of the potential @xmath13 is the gravitational force field @xmath58 , and from eq . ( [ eulereq ] ) the gradient of the pressure @xmath8 goes with minus the gradient of the gravitational potential @xmath13 . thus one can infer from fig . [ fig:1 ] , and the help of eqs . ( [ eulereq ] ) , ( [ gravforce ] ) , and ( [ mag2 ] ) , that at the wormhole s center @xmath59 , i.e. , at the throat , the pressure is maximal , as one would expect . in the limiting case that the wormhole has @xmath60 a newtonian event horizon develops at some radial coordinate @xmath61 , in the sense that any particle inside the sphere defined by @xmath61 can only affect an outside observer if it has infinite acceleration , and thus infinite velocity . thus , since no particles can come out of this inner region one is in the presence of a newtonian black hole . however , for @xmath60 the pressure support at the throat goes to infinity , and so , in this limit , the wormhole eventually collapses . now we use eq . to study the motion of a test particle in the gravitational field of the wormhole space . consider that a particle s path is described by a curve @xmath62 , whose components @xmath63 are given by @xmath64 such that the particle s velocity is @xmath65 , where a dot means a derivative with respect to time @xmath66 . the right hand side of eq . is the gravitational field given in eq . and the left hand side the acceleration of the particle , defined as @xmath67 . it is possible to show that one can treat the problem of the motion of a test particle by considering pure equatorial orbits , i.e. , @xmath68 , @xmath69 and @xmath70 . gathering these results the equations of motion are @xmath71}{2l\left(b^{2}+l^{2}\right)\left[b\ , e^{\frac{l^{2}}{b^{2}}}\left(e+\beta\right)-\sqrt{b^{2}+l^{2}}\left(e^{1+\frac{l^{2}}{b^{2}}}+\beta\right)\right]}\right]\,\,\dot{l}{}^{2}+\\ \hspace{70bp}+\left[\frac{b\sqrt{b^{2}+l^{2}}\left(e+\beta\right)-\left(b^{2}+l^{2}\right)\left(e+\beta\ , e^{-\frac{l^{2}}{b^{2}}}\right)}{e\ , l}\right]\,\,\dot{\varphi}{}^{2}=\\ \hspace{70bp}=-\frac{g\ , m\left(l\right)}{2\sqrt{e}\,\left(b^{2}+l^{2}\right)l}\sqrt{\left(b^{2}+l^{2}\right)\left(e+\beta\ , e^{-\frac{l^{2}}{b^{2}}}\right)-b\sqrt{b^{2}+l^{2}}\left(e+\beta\right)}\ , , \end{aligned } \label{eq : ldotdot_simp}\ ] ] @xmath72 @xmath73 having found the equations of motion of a test particle , we have now to solve them . let us consider the simpler problem of a test particle describing pure circular motion in the newtonian wormhole space . in this case the coordinate @xmath45 is a constant , @xmath74 , say , so @xmath75 and @xmath76 . from this , eq . can be solved for @xmath77 and obtain @xmath78 eq . relates the radial position @xmath74 and the angular velocity @xmath77 for a particle to describe a circular orbit in the wormhole space . now , except the case of pure circular motion , to solve the equations of motion - for a more generic motion , numerical methods must be used , provided the initial position and velocity of the test particle and the wormhole parameters * @xmath19 * and @xmath18 are given . we now analyze the relative separation of two test particles in the gravitational field of the spherical newtonian wormhole space . the two test particles are initially considered to be infinitesimally close , subjected only to the gravitational field of the newtonian wormhole , such that the acceleration of a test particle at a certain point is given by eq . and eqs . - . this relative separation , i.e. , the tidal deviation equation , is in part given by the variation of the gravitational force field acted on each particle and another part given by the geodesic deviation due to the spatial curvature through the equation , @xmath79 where @xmath80 is the separation vector between two infinitesimally close particles , @xmath58 is the gravitational force field generated by the wormhole mass , given by eq . , @xmath81 is the velocity of the fiducial test particle and @xmath82 represents the riemann tensor . let us simplify the calculations , by remarking that , as was mentioned in sec . ( [ sub : test_particle_dynamics ] ) , the velocity of the fiducial test particle can be assumed to be along the equator , @xmath83 ( see eq . ) . hence , in eq . , we might take all the terms in @xmath84 to zero . notice , however , that this does not mean that , in general , the acceleration of the component @xmath85 of the separation vector is zero since the motion of the second particle may not be along the equator . now , considering the metric of the spherical newtonian wormhole space , eq . , and assuming @xmath83 and @xmath86 , we find from eq . that the deviation equations for two infinitesimally close test particles in the spherically symmetric newtonian wormhole space are @xmath87\,,\label{dev_theta}\\ \frac{d^{2}n^{\varphi}}{dt^{2}}= & -\gamma_{\varphi l}^{\varphi}\mathcal{g}^{l}n^{\varphi}+r_{ll\varphi}^{\varphi}v^{l}\left[n^{\varphi}v^{l}-n^{l}v^{\varphi}\right]\,.\label{eq : dev_phi}\end{aligned}\ ] ] where the christoffel symbols that appear in eqs - have the form @xmath88}{2b^{2}l\left(b^{2}+l^{2}\right)\left[b(\beta+e)e^{\frac{l^{2}}{b^{2}}}-\sqrt{b^{2}+l^{2}}\left(e^{\frac{l^{2}}{b^{2}}+1}+\beta\right)\right]}\,,\label{gammalll}\ ] ] @xmath89 the other non - zero christoffel symbols necessary to calculate the riemann tensor components are @xmath90 , @xmath91 , and @xmath92 . at the plane @xmath83 one has @xmath93 and @xmath94 , the remaining christoffel symbols being zero or given by symmetry of the ones just calculated . the riemann tensor components appearing in eqs . - , assuming @xmath83 , are @xmath95 these equation complete the geodesic deviation equations - . now , looking at eqs . - , in general , it is not possible to find analytical solutions since the fiducial test particle s velocity must also be taken into account . so , to solve this system of differential equations , we must first solve the equations of motion , eqs . - , and in general there is no analytical solution . we can however work out some features of the tidal effects in the case of pure radial motion of two test particles such that throughout the motion the line that connects the two particles is purely radial , i.e. , the case where @xmath85 , @xmath96 , @xmath84 , and @xmath97 are zero throughout the particles motion . in this case eq . simplifies to @xmath98 and eqs . and yield @xmath99 and @xmath100 , respectively . although it is still not possible to solve analytically eq . we can instead make a qualitative analysis to infer the general behavior of two initially close test particles describing pure radial motion . plot of the quantity @xmath101 as a function of the radial coordinate @xmath45 ( here assume @xmath55 ) . ] defining the right hand side of eq . as @xmath102 we present in fig . [ fig:2 ] its behavior . we see that the sign of the function @xmath103 changes at some value of the coordinate @xmath45 , @xmath104 , say . now , when @xmath105 the function @xmath106 is positive , @xmath107 , which implies that the separation vector has positive acceleration . so , suppose two test particles describing pure radial motion , such that the line that connects them is purely radial . one particle has radial coordinate @xmath108 , the other @xmath109 , and let us assume , @xmath110 and @xmath111 with @xmath112 for simplicity . in this case , since @xmath107 , the particle with coordinate @xmath108 is accelerating more than the particle with coordinate @xmath109 , and they fly apart . on the other hand , for @xmath113 the function @xmath106 is negative , @xmath114 , so the separation vector has negative acceleration . assuming now @xmath115 , and still @xmath110 , this means that the particle with coordinate @xmath108 is accelerating less , but still towards the wormhole s throat , than the particle with coordinate @xmath109 . numerically solving eq . for two initially close test particles in the regions @xmath105 and @xmath113 , indeed , we verify this conclusions . notice that as we consider particles further away from the wormhole s throat we recover the tidal behavior expected from newtonian gravitation since the space is asymptotically flat . a static , spherically symmetric wormhole in an enhanced newtonian theory of gravitation was constructed . the newtonian wormhole s mass density is positive , the gravitational field of the wormhole is well - behaved and the matter that sustains it has positive pressure . test particles motion in the wormhole gravitational field and tidal effects were studied . we have also argued about the possibility of having true newtonian black holes , i.e. , newtonian objects that have regions from which any particle must have infinite acceleration and thus infinite velocity to escape to the outside of it , in this enhanced newtonian theory of gravitation . m. abramowicz , g. f. r. ellis , j. hork , and m. wielgus , `` the perihelion of mercury advance and the light bending calculated in ( enhanced ) newton s theory '' , gen . gravit * 46 * , 1630 ( 2014 ) ; arxiv:1303.5453 [ gr - qc ] ( 2013 ) . j. ehlers , `` contributions to the relativistic mechanics of continuous media '' , gen . relativ . gravit * 25 * , 1225 ( 1993 ) , ( translation of the proceedings of the mathematical - natural science section of the mainz academy of science and literature * 11 * , 792 ( 1961 ) ) . g. f. r. ellis and h. van elst , `` cosmological models '' , in proceedings of the nato advanced study institute on theoretical and observational cosmology , cargse , nato science series , series c , mathematical and physical sciences * 541 * , 1 , ed . m. lachize - rey ( kluwer academic , boston 1999 ) ; arxiv : gr - qc/9812046 . a. b. balakin , j. p. s. lemos , and a. e. zayats , `` nonminimal coupling for the gravitational and electromagnetic fields : traversable electric wormholes '' , phys d * 81 * , 084015 ( 2010 ) ; arxiv:1003.4584 [ gr - qc ] .	a spherically symmetric wormhole in newtonian gravitation in curved space , enhanced with a connection between the mass density and the ricci scalar , is presented . the wormhole , consisting of two connected asymptotically flat regions , inhabits a spherically symmetric curved space . the gravitational potential , gravitational field and the pressure that supports the fluid that permeates the newtonian wormhole are computed . particle dynamics and tidal effects in this geometry are studied . the possibility of having newtonian black holes in this theory is sketched .
the idea of mobile ad - hoc networks ( manets ) spontaneous wireless networks , operating without a backbone infrastructure , whose users / nodes relay packets for each other in order to enable multihop communications continues to inspire practitioners and poses interesting theoretical questions regarding its performance capabilities . vehicular ad - hoc networks ( vanets ) may well be currently one of the most promising incarnations of manets . promoters of vanets believe that these networks will both increase safety on the road and provide value - added services . numerous challenging problems must however be solved in oder to be able to propose these services e.g. efficient and robust physical layers , reliable and flexible medium access protocols , routing schemes and optimized applications . an almost ubiquitous stochastic assumption in the theoretical studies of these problems in manets is that the nodes of the network are distributed ( at any given time ) as points of a planar ( 2d ) poisson point process . in conjunction with the aloha medium access ( mac ) scheme simpler but less efficient than csma usually considered in this context by practitioners 2d poisson manet models allow for quite explicit evaluation of several performance metrics ; cf . section [ s.rw ] . however , they mostly regard local ( one - hop ) transmissions . in fact , introducing routing even to the simplest 2d poisson - aloha model is very difficult and it is hard to obtain rigorous theoretical results regarding truly multihop performances of manets ; cf . again section [ s.rw ] . one reason for this is that , while the source - node can be considered as a typical node in the manet and the powerful palm theory of point process can be used in the analysis of the first hop , further relay nodes on a given path ( traced by the dijkstra algorithm or any reasonable local routing on a 2d poisson manet ) can not be seen as typical nodes of the manet . in fact , the route followed by a packet is a random subset of the manet s point pattern ( depending on the routing algorithm ) and the typical point `` seen '' by the packet on a long route is not the typical point of the whole manet in the sense of palm theory , cf _ routing paradox _ in @xcite . in order to overcome this impasse and to propose an analytically tractable multihop model , in this paper we propose to `` decouple '' a tagged route from the remaining ( external to the route ) part of the manet . more precisely , we consider two stochastically independent point patterns to model respectively : some given route on which packets of a tagged flow are relayed and nodes which are only sources of interference for packet transmissions on this route . moreover , we assume that the tagged route is modeled by a linear ( 1d ) stationary point process . these assumptions allow us to distinguish the typical node of the route from the typical point of the manet , and appropriately use palm theory to manipulate the two objects . the advantage of the 1d - modeling of the tagged route is that packets relayed on it through the nearest neighbor transmissions `` see '' , at any relaying node , the typical ( 1d palm ) distribution of the whole route . this is a well - known _ point - shift invariance property of the palm distribution in 1d _ , which does not have a natural extension in higher dimensions . another advantage of our decoupling of the two parts of the manet is that we can go beyond poisson assumptions in both of them . specifically , we will also consider a _ poisson route appended with a lattice structure of relaying nodes _ , which turns out to be crucial to improve the routing performance on long routes in the presence of external interference and/or noise . regarding external interference , we are able to study the impact of the clustering of nodes , considering e.g. _ poisson - line manet _ ( to be explained in what follows ) . before describing our main results , let us further justify our route - in - manet model . firstly , it is a quite natural scenario in vanets , where vehicles are randomly located on a road and subject to external sources of interference . in the context of a general 2d manet , the linearity ( 1d - pattern ) of routing is clearly a simplification . in this case we think of packets as being relayed in a given direction , e.g. in a strip as shown on the left sketch in figure [ fig.lineproj ] . we can approximate the `` real '' route by a `` virtual '' one by taking the orthogonal projections of the real nodes on the line joining the source and the destination . such a situation represents an approximation of a geographic routing . last but not least , in this paper we consider a 2d _ poisson - line manet _ model , in which all the nodes are randomly located on a ( background ) poisson process of lines , and form the so - called _ doubly - stochastic poisson process _ ( or cox process ) see on the right in figure [ fig.lineproj ] . in this case our decoupled , tagged route is rigorously ( in the sense of palm theory ) the typical route in this poisson - line manet . we consider slotted aloha mac to be used both by the nodes of the tagged route and of the interfering part of the manet and the signal - to - interference - and - noise ratio ( sinr ) capture / outage condition , with the power - law path loss model and rayleigh fading . we assume that all the nodes of the manet are backlogged , i.e. , they always have packets in their buffers to transmit . moreover , we are interested in the routing performance of a tagged packet relayed by successive nodes on the route with priority in all queues on the route ( cf . a discussion of these assumptions in section [ s.rw ] ) . the main results of this paper are the following : * in the absence of external ( to the route ) noise and interference we evaluate the _ mean local delay _ on the poisson route , i.e. the expected number of time slots required for the packet to be transmitted by the typical node of the route to its nearest neighbor in a given direction . the inverse of this delay is intrinsically related to the _ speed of the packet progression on asymptotically infinite routes _ , which can also be related to the _ route transport capacity _ ( number of bit - meters pumped per unit length of the route ) . * the local mean delay is minimized ( equivalently , the progression speed is maximized ) at a unique value of the aloha medium access probability @xmath0 . moreover , the routing is unstable the delay is infinite ( the speed is null ) for @xmath0 larger than some critical value . this observation is fully compliant with the phase - transition phenomenon of the mean local delay discovered in the 2d poisson manet model in @xcite . * we evaluate the _ mean end - to - end delay _ of the packet routing between two given nodes of the route . it shows a cut - off phenomenon , namely that _ routing on distances shorter than some critical value is very inefficient in terms of exploiting the route transport capacity_. * next , we study the impact of noise and external interference on our previous observations . confirming the theoretical findings of @xcite in 2d poisson manets , we observe that the routing on a poisson route over long distances is unfeasible : _ the speed of packet progression on such routes is null _ due to the existence of large hops in random ( poisson ) paths . evaluating the end - to - end delay on finite distances with noise , we identify another , double cut - off phenomenon : the existence of _ critical values of end - to - end distance and noise power , beyond which the speed of packet progression is close to 0 _ , again making the routing inefficient in terms of exploiting its transport capacity . * in order to allow efficient routing over long routes one can complete the poisson route with a fixed lattice of ( equidistant ) relaying nodes . for this model , we evaluate the mean local delay and show how the route transport - capacity can be maximized by an optimal choice of the inter - relay distance of the lattice structure . * we explicitly evaluate the poisson - line manet model and compare it to that in the ( basic ) poisson - line - in - poisson - field model . we show that the poisson - line exhibits a larger coverage probability but also a larger end - to - end delay . this confirms previous general observations regarding the impact of clustering in interference ; @xcite . * regarding applications to vanets , we evaluated delays of some emergency and broadcast communications . local ( one - hop ) characteristics in 2d poisson manets , such as sinr outage probability , the related density of packet progress , mean local delay and many others , have been extensively studied in the literature ; cf . e.g. among others @xcite . poisson models also allow one to discover some intrinsic theoretical limitations of manets , regarding e.g. the scaling of the capacity of dense and/or extended networks @xcite , or the speed of packet progression on long routes @xcite . in @xcite the authors study delay and throughput in a model with multihop transmissions with relays which are placed equidistantly on the source - destination line . the model is very similar to ours but with the following differences . in the model presented in @xcite the topology of relaying nodes is regular ( they are equidistant ) and the pattern on interferers is re - sampled at each slot . the delays include both the service times and waiting times in the buffers on the given route . a combined tdma / aloha mac protocol with intra - route tdma and inter - route aloha is employed . in contrast , we use simple aloha , ignore queueing , and focus on the performance issues caused by the _ randomness of the topology of relaying nodes_. we are interested in intrinsic limitations of the performance of _ long - distance routing _ in wireless networks with an irregular topology . we believe that the observed limitations remain valid for networks employing csma . this is because they primarily depend on the existence of ( arbitrarily ) long hops . csma , which copes better with interference , can not improve upon this situation . in contrast , we show that using a regular structure of relay nodes superposed with irregular manet routes leads to better performance . this solution is also evidently necessary for the stability of queuing processes ( not covered in our paper ) . _ the remaining part of this paper _ is organized as follows . in section [ s.linearnearest ] , we present our nearest neighbor routing model with slotted aloha . in section [ s.deterministic ] we compute routing delays in deterministic networks . in section [ s.endtoend ] we study the end - to - end delays on a poisson route when the interference is limited to the interfering nodes on the poisson route . in section [ s.noise ] we study the impact of external noise and interference . section [ s.conclusion ] concludes the paper . let us denote by @xmath1 , with @xmath2 , the locations of nodes participating in the routing of some _ tagged flow of packets _ from @xmath3 to @xmath4 . this route is assumed not to change on the time scale considered in this paper . the following assumptions regarding @xmath5 will be considered . although it is not the main scenario of this paper , we begin by considering a deterministic , fixed , finite pattern of nodes @xmath6 . in this scenario we suppose that @xmath7 forms a poisson point process of intensity @xmath8 , on the line @xmath9 . in this model the notational convention is such that @xmath10 and packets are sent by any given node @xmath3 to its nearest - to - the - right neighbor @xmath4 . note that the poisson assumption means that the 1-hop distances @xmath11 are independent ( across @xmath12 ) exponential random variables with some given mean @xmath13 . we will call this scenario _ nearest neighbor ( nn ) routing on poisson line_. we will also consider a version of this model , where the packet is transmitted to the _ nearest ( available ) receiver ( nr ) _ to the right on @xmath5 . this is an opportunistic routing allowing for longer hops , as will be explained in section [ ss.nnvsnr ] . in this model , the tagged route consists of a superposition @xmath14 , where the poisson route @xmath15 is completed with equidistantly located `` fixed '' relay nodes @xmath16 ; @xmath17 is some fixed parameter and @xmath18 is a uniform random variable on @xmath19 , independent of @xmath15 , making @xmath5 stationary . in this model we also consider nn routing , i.e. , packets are always transmitted to the nearest neighbor to the right in @xmath5 . we consider @xmath5 as some `` tagged '' route obtained in a manet by some routing mechanism . a simple way of extending @xmath5 to a 2d manet model consists in embedding @xmath5 in an external field of nodes @xmath20 , on the plane @xmath21 . when doing so we will always consider that @xmath5 and @xmath22 are independent . the following assumptions regarding @xmath22 will be considered . one may consider a _ fixed , deterministic _ pattern @xmath20 , although , once again , it is not the main scenario considered in this paper . in this model , we assume that @xmath22 is a poisson point process of some given intensity @xmath23 on @xmath21 . the poisson linear route @xmath7 embedded in a poisson process of interferers will be our default _ poisson - line - in - poisson field _ model . we will also consider a case where the interferers are located on a _ poisson process of lines _ ( roads ) on @xmath21 of rate @xmath24 representing the total line - length per unit of surface . assuming that on each line of this process there is an independent poisson process of points of intensity @xmath25 nodes per unit of line length , we obtain a doubly stochastic poisson point process on @xmath21 with intensity @xmath26 nodes per unit of surface ; see @xcite . in particular , assuming @xmath27 , one can rigorously consider the poisson liner route @xmath7 embedded in such a poisson line process of interferers , as the typical route of the _ poisson - line manet _ , see figure [ fig.lineproj ] on the right - hand side . we assume that the _ all _ nodes of @xmath5 and @xmath22 try to access the channel ( transmit packets ) according to the aloha scheme , in perfectly synchronized time slots @xmath28 . each node in each time slot independently tosses a coin with some bias for heads , which will be referred to as the _ aloha medium access probability _ ( aloha map ) . nodes whose outcome is heads transmit their packets , the others do not transmit . we denote by @xmath0 the map of nodes in @xmath5 , and by @xmath29 the map of nodes in @xmath22 . the above situation will be modeled by marking the points @xmath30 with random , bernoulli , medium access indicators @xmath31 equal to 1 for the nodes which are allowed to emit in the slot @xmath32 and 0 for the nodes which are not allowed to emit . we have @xmath33 for all @xmath34 . when there is no ambiguity , we will skip the time index @xmath32 in the notation and also use the notation @xmath35 . similarly , we mark the interfering nodes by independent bernoulli , medium access indicators with parameter @xmath29 . at each time slot @xmath32 , aloha splits @xmath5 into two point processes @xmath36 of emitters ( having a mac indicator @xmath37 equal to 1 at time @xmath32 ) and ( potential ) receivers @xmath38 . it is known that when @xmath7 is a poisson process of intensity @xmath8 then @xmath39 and @xmath40 are independent poisson processes with intensity @xmath41 and @xmath42 , respectively . each transmitting node uses the same transmission power , which without loss of generality is assumed to be equal to 1 . the signal - power path - loss is modeled by the power - law function @xmath43 where @xmath44 are some constants and @xmath45 is the distance between the transmitter and the receiver . signal - power is also perturbed by random fading @xmath46 which is independently sampled for each transmitter - receiver pair at each time slot @xmath32 . thus , the actual signal - power received at @xmath47 from @xmath48 at time @xmath32 is equal to @xmath49 . in this paper we will restrict ourselves to an important special case where @xmath46 is _ exponentially distributed _ , which corresponds to the situation of independent _ rayleigh fading_. by renormalization of @xmath50 , if required , we can assume without loss of generality that @xmath46 has mean 1 . when a node located at @xmath48 transmits a signal to a node located at @xmath47 , then successful reception depends on the signal - to - interference - and - noise ratio ( sinr ) @xmath51 where @xmath52 is the shot - noise process of @xmath53 : @xmath54 representing the interference created by the nodes of route @xmath5 and @xmath55 represents an external ( to @xmath5 ) noise . this external noise can be a constant ambient noise @xmath56 or a random field . in particular , @xmath55 may comprise the interference created by the external field of interferers @xmath22 , which do not belong to the route @xmath5 . in this case : @xmath57 where @xmath58 is the subset of nodes of @xmath22 transmitting at time @xmath32 . throughout this paper we adopt the common assumption throughout this paper is that the _ noise process @xmath55 is independent of the route - interference process @xmath59 _ and that _ the process @xmath55 is stationary in @xmath32_. in this paper we assume a fixed bit - rate coding , i.e. , @xmath47 successfully receives the signal from @xmath48 if @xmath60 where @xmath61 is given by ( [ sinr ] ) and @xmath62 is the sinr - threshold related to the bit - rate given some particular coding scheme . recall that in our nn routing model each transmitter in @xmath39 transmits to the nearest node on its right in @xmath5 , without knowing whether it is authorized to transmit at this time . successful reception requires that this selected node is _ not _ authorized by aloha to transmit ( i.e. , it is in @xmath40 ) , and that the sinr condition ( [ eq : sinr ] ) for this transmitter - receiver pair is satisfied . this corresponds to the usual separation of the routing and mac layers . in contrast , nr routing consists in transmitting to the nearest node ( in the given direction ) on @xmath5 _ which , at the given time slot , is not authorized by aloha to transmit_. the nr model hence might be seen as an opportunistic routing , which requires some interplay between mac and the routing layer . as we shall see , both routing schemes allow for quite explicit performance analysis in our basic poisson - line - in - poisson - field model . the default assumption in our numerical examples throughout the paper is @xmath63 nodes per meter in the poisson line model , i.e. the mean distance between two consecutive nodes on the line is 100 m. we will also use @xmath64 , @xmath65 and @xmath66 . the fading @xmath46 is rayleigh , f is exponentially distributed of mean 1 , @xmath67 . unless otherwise specified , in this section we assume that the locations of _ _ the nodes in the network are known and fixed ( deterministic ) . the only source of randomness is aloha mac and independent rayleigh fading between any two given nodes . in what follows we present a simple computation which allows us to express coverage and routing delays in such networks . our observations will be useful in the remaining part of this paper , when we will study random routes in random manets . let us consider a transmitter located at @xmath68 communicating to a receiver located at @xmath69 in the presence of a field @xmath70 of interferers , with all nodes subject to aloha mac with map @xmath0 . we assume that @xmath71 . we denote by @xmath72 the _ probability of successful transmission from @xmath48 to @xmath47 in one time slot_. this probability accounts for the likelihood of @xmath48 being authorized by aloha to transmit , @xmath47 not to being authorized to transmit and , given such circumstances , the probability of achieving the sinr larger than @xmath62 at the receiver . we define two functions : @xmath73 for reasons which will become clear in what follows , we call @xmath74 the _ interference factor _ and @xmath75 the _ noise factor_. [ l.pi ] we have @xmath76 [ r.pi-recur ] let us note that @xmath77 satisfies the following recursion when adding an interferer @xmath78 to the field @xmath22 : @xmath79 in other words , the external noise reduces the probability of successful transmission over the distance @xmath45 by the factor @xmath80 , which in the absence of noise and interference is equal to @xmath81 ( because of the aloha scheme ) . moreover , adding an interfering node @xmath78 to @xmath22 causes a decrease in the successful transmission probability by the factor @xmath74 , where @xmath82 is the distance of the new interferer to the receiver . we have @xmath83\end{aligned}\ ] ] and @xmath84 = ( 1-p)+p { { \mathbf e}}\big[e^{-tf_{(z , y ) } l(\|x - y\|)/l(\|z - y\|)}\big ] = 1-\frac{p}{\frac1{t } \frac{\|z - y\|^{\beta}}{\|x - y\|^{\beta}}+1}\,,\ ] ] where we use the assumption that @xmath46 is an exponential variable . the proof follows by induction . @xmath72 is the probability that node @xmath48 can successfully send a tagged packet to node @xmath47 in a single transmission . however , a single transmission is not sufficient in aloha . hence , after an unsuccessful transmission , the transmitter will try to retransmit the packet with aloha , possibly several times , until the packet s reception . we denote by @xmath85 the expected number of time slots required to successfully transmit a packet from @xmath48 to @xmath47 ( considering the previous scenario of @xmath68 transmitting to @xmath69 in the presence of a field @xmath70 of interferers ) . under our assumptions this number is a geometric random variable of parameter @xmath72 and hence its expected value is equal to : @xmath86 we call @xmath85 the _ local ( routing ) delay_. by lemma [ l.pi ] @xmath87 and an analogous recurrence to this of remark [ r.pi-recur ] holds for @xmath88 with the reciprocals of the noise and interference factors . let us now consider a route consisting of nodes @xmath89 in the field of interfering nodes @xmath22 ( we assume that @xmath90 ) . the average delay to send a packet from @xmath91 to @xmath92 using the route @xmath5 is simply the sum of the local delays on successive hops @xmath93 note that for the hop from @xmath94 to @xmath95 other nodes of the route @xmath96 act as interferers . we call @xmath97 the _ ( routing ) delay _ on @xmath5 . using ( [ e.l ] ) we obtain @xmath98 we also denote by @xmath99 the mean speed of packet progression on the route @xmath5 . the above expressions allow for a relatively simple , explicit analysis of routing delays in fixed networks , i.e. in networks where the locations of the nodes are fixed and known . in the case when such information is not available , one adopts a stochastic - geometric approach averaging over possible geometric scenarios regarding @xmath5 and/or @xmath22 . for example , assume that the route @xmath5 is given and fixed , but the number and precise locations of interferers are not given . assuming some statistical hypothesis regarding the distribution of @xmath22 ( which becomes random point process ) one can calculate the average route delay @xmath100 $ ] where the expectation regards the distribution of @xmath22 . specifically , for a given couple @xmath101 we define a function @xmath102 we denote by @xmath103 the laplace transform of @xmath22 , i.e. , for a given ( say non - negative ) function @xmath104 defined on @xmath21 , @xmath105 $ ] . we then have : @xmath106 & = p(1-p ) w(\|x - y\|){{\mathcal{l}}}_\psi(h_{x , y})\,,\\ { { \mathbf e}}_\psi[l(x , y,\psi ) ] & = \frac1{p(1-p ) } w^{-1}(\|x - y\| ) { { \mathcal{l}}}_\psi(-h_{x , y})\label{e.epsil}\end{aligned}\ ] ] and @xmath107= \frac1{p(1-p ) } \sum_{k=0}^{n-1 } { { \mathcal{l}}}_\psi(-h_{x_k , x_{k+1}})\!\!\!\ ! \prod_{z \in { { \mathcal{r}}}\setminus \{x_k , x_{k+1}\ } } h^{-1}(\|z - x_{k+1}\|,\|x_k - x_{k+1}\| ) w^{-1}(\|x_k - x_{k+1}\|)\,.\ ] ] in the next section we will consider a scenario when the route @xmath5 is also modeled as a point process . in this section we assume poisson route @xmath7 and no external interferers ( @xmath108 ) . we also assume that the external noise process @xmath109 is negligible with respect to the interference created by the nodes participating in the routing . in this scenario we will consider nn and nr routing schemes ; cf [ ss.pl ] . we begin with a simple calculation of the capture probability . we consider a typical node on the route @xmath5 , that is of the poisson point process @xmath15 . by the slivnyak theorem , it can be seen as an `` extra '' node located at the origin @xmath110 , with the other nodes of the route distributed according to the original stationary poisson process @xmath15 . all marks ( bernoulli mac indicators , fading , etc ) of these extra nodes are independent of the marks of the points of @xmath15 . we denote by @xmath111 and @xmath112 the probability of successful transmission of the typical node @xmath110 in a given time slot to the receiver prescribed by the nn and nr routing schemes , respectively . in the capture probability @xmath113 and @xmath114 in the two routing schemes considered , we assume that the typical node is authorized to transmit . obviously @xmath115 and @xmath116 . for arbitrary @xmath117 we denote @xmath118 and @xmath119 moreover , for given @xmath62 and @xmath120 let us denote : @xmath121 [ p.pnn ] the probability of successful transmission by a typical node of poisson route @xmath7 authorised by aloha to transmit to its relay node in the nn routing model without noise is equal to : @xmath122 and similarly in the nr receiver model @xmath123 it is easy to see that @xmath124 and hence @xmath125 , i.e. , the opportunistic choice of the receiver pays off regarding the probability of successful transmission . note directly from the form of the sinr in ( [ sinr ] ) that @xmath113 and @xmath114 do not depend on @xmath50 . hence in the remaining part of the proof we take @xmath64 . we consider first the nn model . the scenario with the typical user @xmath110 located at the origin corresponds to the palm distribution @xmath126 of the poisson route @xmath127 . by the known property of the poisson point process , the distance @xmath128 from @xmath129 to its nearest neighbor to the right , @xmath130 , has an exponential distribution with parameter @xmath8 . moreover given @xmath131 , all other nodes of the poisson route @xmath5 form a poisson point process of intensity @xmath8 on @xmath132 . conditioning on @xmath131 we can thus use ( [ e.epsipi ] ) with @xmath133 ( recall that @xmath134 is a poisson process of intensity @xmath41 of nodes transmitting in a given time slot ) to obtain : @xmath135 using the well known formula for the laplace transform of the poisson p.p . ( see e.g. , ( * ? ? ? * ( 16.4 ) ) ) we obtain : @xmath136 inserting this expression in ( [ e.pnn-cond ] ) we obtain @xmath137 which boils down to the right - hand side of ( [ e.pnnd ] ) . in order to obtain expression ( [ e.pnrd ] ) for the @xmath138 model , we follow the same process , with the following modifications . the distance to the nearest _ receiver _ to the right has the exponential distribution of parameter @xmath42 . moreover the distribution of the point process of emitters is poisson with parameter @xmath41 and independent of the location of this receiver . note also that by the very choice of the receiver , it is not authorized to emit in the given time slot , hence there is no @xmath139 factor in the numerator . let us denote by @xmath140 the _ local delay _ at the tagged node in the nn routing scheme , i.e. the number of consecutive time slots needed for the tagged node to successfully transmit a given packet to the receiver designated by the given routing scheme . in what follows we will give the expressions for the expected local delay @xmath141 $ ] at the typical node of the poisson route , which we call the _ average local delay _ , where the averaging regards all possible poisson configurations of other nodes @xmath15 . as we will see , this expectation is finite only if @xmath0 is sufficiently small ( for any given @xmath62 and @xmath120 ) . let us denote @xmath142 [ p.emergency ] under the assumptions of proposition [ p.pnn ] , the mean local delay in the nn routing model is equal to @xmath143 provided @xmath144 and @xmath145 otherwise . conditioning on the distance to the nearest neighbor as in the proof of proposition [ p.pnn ] and using ( [ e.epsil ] ) , we have : @xmath146 with @xmath147 consequently , we have : @xmath148 it is easy to see that this integral is infinite if @xmath149 and @xmath150 otherwise . this completes the proof . similar phase transition ( existence of the critical @xmath0 ) for the mean local delay was discovered in 2d manets in @xcite . note that @xmath151 with the equality attained when @xmath152 . hence , for small @xmath0 we have @xmath153 , and the critical @xmath0 for which the average expected delay explodes is approximately @xmath154 . expression ( [ e.el ] ) can be used to tune @xmath0 to the value which minimizes the mean local delay . in the next section we will see that this value of @xmath0 also _ optimizes the performance of nn routing on long routes , for almost any realization of the poisson route_. we call long - distance speed the speed at which a packet progresses over a long distance . let us denote by @xmath155 , @xmath156 the local delay of the packet on its hop from @xmath3 to @xmath4 . more specifically , then @xmath157 is the end - to - end delay on @xmath158 hops on the route that starts at @xmath110 . we denote by @xmath159 the _ long - distance speed of the packet progression _ ( expressed in the units of distance per slot ) . note also that @xmath160 multiplied by the number of bits carried by one packet corresponds to the _ route transport capacity _ defined as the number of bit - meters `` pumped '' by the network per slot . [ p.ldspeed ] under the assumptions of proposition [ p.pnn ] , the mean long - distance speed of the transmission of packets in nn routing model is equal to : @xmath161 provided @xmath162 and 0 otherwise . the mean empirical speed during the first @xmath158 hops is : @xmath163 by the ergodic property when @xmath158 tends to @xmath164 the numerator tends almost surely to @xmath165 and the denominator tends almost surely to @xmath166 $ ] . this completes the proof . figure [ fig.speed1 ] shows the mean long - distance speed for @xmath63 , @xmath65 and @xmath167 . we observe that there is an optimal value of @xmath0 which maximizes the long - distance speed of packet progression and that this speed drops to 0 for @xmath0 larger than some critical value . [ rem.critp ] existence of the critical value of @xmath0 can be attributed to hops , traversed by a tagged packet on the infinite route , being statistically too long . indeed , analysing the expression in ( [ e.pnn1 ] ) one sees that the `` rate '' at which hops of length @xmath45 occur it is equal to @xmath168 while the rate at which the packet passes them is @xmath169 . thus , when @xmath170 the packet delay becomes infinite . some manet models ( in particular the bipolar one ) use the _ mean density of progress _ @xmath171 to optimize the performance of the model with respect to @xmath0 ( cf e.g. @xcite ) . recall that in a 2d manet @xmath171 is defined as the expected total progress of all the successful transmissions per unit of surface . in our 1d model , by campbell s formula @xmath171 can be expressed as : @xmath172\,.\ ] ] the density of progress , can be also seen as quantifying the number of bit-(or packet)-meters `` pumped '' per unit length of a route . [ p.dnn ] under the assumptions of proposition [ p.pnn ] the density of progress in the nn receiver model is equal to : @xmath173 and is maximized for @xmath0 equal to : @xmath174 moreover , @xmath175 . using campbell s formula , the density of progress in the nn model can be expressed as @xmath176\ ] ] with the notation as in the proof of proposition [ p.pnn ] . following the same arguments as in this latter proof we obtain : @xmath177 which is equal to the right - hand side of ( [ e.dnnd ] ) . taking the derivative of the latter expression in @xmath0 we find that its sign is equal to that of the polynomial @xmath178 . note that @xmath179 , @xmath180 , @xmath181 and @xmath182 . hence in the interval @xmath183 , @xmath184 has a unique root : @xmath185 which maximizes @xmath186 . the explicit expression for this root follows from the general formulas for the roots of cubic equations . note that density of progress is a `` static '' quantity , calculated with respect to one slot . it can also be easily evaluated for the nr model in contrast to the mean local delay . however , it fails to discover the existence of the critical value of @xmath0 for the performance of the manet revealed by the analysis of the local delay and packet progression speed . also , one may be tempted to approximate this speed by @xmath187 . let us note , however , in figure [ fig.speed1 ] , that the optimization of this quantity in @xmath0 can not be directly related to the maximization of the packet progression speed . for this reason we will not consider the density of progress any further in this paper . to node @xmath188 in this section we will studythe issues of delays and the speed of packet progression on finite segments of routes . in this regard , we assume that in addition to the node located at the origin @xmath189 ( also called @xmath190 ) , a second node , being the final destination of a given packet is located at a fixed position @xmath191 on the line . mathematically , this corresponds to the palm distribution @xmath192 of the poisson pp , given these two fixed points . we want to compute the mean end - to - end delay for a given packet to leave the node @xmath193 and reach node @xmath194 following nn routing . the situation proposed is shown in figure [ fig.delays ] . we denote this end - to - end delay by @xmath195 . [ p.delaysuv ] under the assumptions of proposition [ p.pnn ] the mean end - to - end delay in the nn routing on the distance @xmath194 is equal to @xmath196}\\ & = \frac{1}{p(1-p ) } \bigg ( e^{-\lambda m } e(m)\label{a}\tag{a}\\ & \hspace{1em}+\int_0^{m } \lambda e^{-\lambda r } e(r ) g_m(0,r ) { \mathrm{d}}r \label{b}\tag{b}\\ & \hspace{1em } + \lambda \int_0^{m } \int_0^{m - s } e(r ) g_0(s , r ) g_m(s , r)\lambda e^{-\lambda r}{\mathrm{d}}r { \mathrm{d}}s \label{c}\tag{c}\\ & \hspace{1em}+ \lambda \int_0^{m } e(m - s ) g_0(s , m - s ) e^{-\lambda ( m - s ) } { \mathrm{d}}s \bigg ) \label{d}\tag{d } \end{aligned}\ ] ] with @xmath197 , @xmath198 and @xmath199 . the mean sum of the delays from the source node 0 to the destination node @xmath200 under @xmath201 can be expressed using campbell s formula as @xmath202= { { \mathbf e}}^{0m}[d(0)]+ \lambda \int_{0}^m{{\mathbf e}}^{0ms}[d(s ) ] { \mathrm{d}}s\,,\ ] ] where the first term corresponds to the average exit time from node @xmath193 and @xmath203 $ ] is the average exit time from a `` current '' node located at @xmath204 . these two expectations differ from @xmath205 $ ] , expressed in ( [ e.el ] ) , because there is a fixed node at @xmath194 which acts as an additional interferer but which also limits the hop length . moreover , for the transmission from the current node @xmath204 , the node at 0 also acts as an additional interferer . we will show how the terms in ( [ a])([d ] ) of ( [ e0 m ] ) reflect these circumstances . let us first remark first from ( [ e.el ] ) that the function @xmath206 is the expected exit time from a typical node , given its receiver is located at the distance @xmath45 and given no additional ( non - poisson ) interfering nodes . now , it is easy to conclude that the term ( [ a ] ) gives the expected end - to - end delay when it is necessary to make the direct hop from 0 to @xmath194 ( when there is no poisson relay node between them ) . the term ( [ b ] ) gives the expected exit time from 0 to its nearest receiver when it is located at some @xmath207 : @xmath208}\\ & = \frac{1}{p(1-p)}\int_0^{m } \lambda e^{-\lambda r}e^{\lambda p t { \,{\mathcal{d}}}_1(p ) } h(m - r , r)^{-1 } { \mathrm{d}}r\,,\end{aligned}\ ] ] where the factor @xmath209 ( not present in ( [ e.el ] ) ) is due to the fact that @xmath194 acts as an additional interferer for the transmission from 0 to @xmath210 . similarly , the term ( [ d ] ) corresponds to the direct hop from the running node at @xmath82 to @xmath194 ( when there is no poisson relay node between them ) and term ( [ c ] ) corresponds to the delay of going from @xmath82 to its nearest receiver @xmath211 . the node at 0 interferes with both these transmissions , which is reflected by the factors @xmath212 and @xmath213 in the terms ( [ d ] ) and ( [ c ] ) , respectively . the node at @xmath194 also interferes with the transmission from @xmath82 to @xmath211 , whence @xmath214 in ( [ c ] ) . [ p.speeduv ] the average speed of the packet progression on distance @xmath194 is equal to @xmath215\bigr)^{-1}\,,\ ] ] where @xmath216 is given by proposition [ p.delaysuv ] . [ r.flight ] in figure [ fig.varm ] we present the speed of the packet progression on distance @xmath194 assuming @xmath167 and @xmath65 , for @xmath194 varying from @xmath217 to @xmath218 . the calculation is performed for the value of @xmath0 that maximizes the asymptotic long - distance speed calculated in proposition [ p.ldspeed ] , presented as the horizontal line in figure [ fig.varm ] . the existence of the destination node in a finite horizon has a double impact on the packet progression . on the one hand it `` attracts '' the packet reducing the negative impact of long hops ( cf remark [ rem.critp ] ) . indeed , no hop can be longer than the direct hop to @xmath194 . on the other hand , it `` repels '' the packet , because it creates an additional interference , which is more significant when the packet is close to the destination . figure [ fig.varm ] shows two phases related to these two `` actions '' . one can interpret these observations by saying that _ routing on distances shorter than some critical values ( @xmath219 ) is very inefficient in terms of exploiting of the route transport capacity_. with respect to @xmath194 in the nn model and long - distance speed @xmath160 . ] in the previous section we assumed that the external noise is negligible ( @xmath220 ) . in this section we study the impact a non - null external noise field @xmath55 . recall that we assume that the noise field @xmath221 is independent of the route process @xmath5 and that it is stationary in @xmath32 ; i.e. , @xmath222 is equal in distribution to @xmath223 . we denote by @xmath224 the laplace transform of @xmath225 with @xmath226 . in is easy to extend the results of propositions [ p.pnn ] and [ p.dnn ] to the case of an arbitrary noise field @xmath55 . [ p.pnnw ] the probability of successful transmission by a typical node of poisson route @xmath7 authorised by aloha to transmit to its relay node in the nn routing model with noise field @xmath227 is equal to @xmath228 the proof is straightforward and follows the same lines as in the proof of proposition [ p.pnn ] . the following results show a very negative impact of arbitrarily small noise on the performance of nn routing on long routes . [ p.emergencyw ] the mean local delay in the poisson @xmath7 nn routing model with a noise field satisfying @xmath229 for some @xmath230 , is infinite @xmath166=\infty$ ] . consequently the speed of packet progression on long routes is almost surely null . as in the proof of proposition [ p.emergency ] we can show that @xmath231\!\!\!&=&\!\!\!\frac\lambda{p(1-p)}\int_0^\infty\!\!\!\ ! e^{-\lambda r}e^{\lambda p r { \,{\mathcal{d}}}_1(p)}{{\mathcal{l}}}_{w(r)}(-tw(ar)^{\beta})\,{\mathrm{d}}r\\ & \ge&\frac{\epsilon\lambda}{p(1-p ) } \int_0^\infty e^{-\lambda r(1-p{\,{\mathcal{d}}}_1(p))+tw(ar)^{\beta } } \,{\mathrm{d}}r\,=\infty . \end{aligned}\ ] ] the reason why the long distance speed is @xmath193 lies in the fact that the hops in a poisson route can be arbitrarily long and that such hops slow down ( to zero ) the packet progression in the presence of noise . one method to cope with overly long hops consists in adding `` fixed '' points to the poisson route @xmath5 . in fact , we will show in section [ ss.short ] that adding only the fixed destination node @xmath194 is sufficient to make the end - to - end delay finite in the presence of noise . for infinite routes one needs , however , a `` fixed '' regular grid of relaying nodes ; cf section [ ss.lattice ] . our setting is now as in section [ ss.delays ] , i.e. , we consider the poisson route @xmath15 under palm with fixed origin and destination nodes @xmath193 and @xmath194 , and we compute the end - to - end delay assuming some external noise field of the form ( [ e.w ] ) created by a stationary pattern of interfering nodes @xmath22 . we assume that this field is stationary and stochastically independent from the poisson route @xmath15 . moreover , we assume that the nodes of @xmath22 use aloha mac with probability @xmath29 . we denote by @xmath232 and @xmath233 the laplace transform of the point process @xmath22 taken for the function @xmath234 given by ( [ eq.h ] ) with @xmath0 replaced by @xmath29 . [ p.delaysuv_w ] under the assumptions of proposition [ p.pnn ] and in the presence of the external noise of the form ( [ e.w ] ) , as explained above , the mean end - to - end delay from @xmath193 to @xmath194 in poisson nn routing is given by ( [ e0 m ] ) with @xmath235 replaced by @xmath236 . observe first that the conditional capture probability given the receiver in @xmath15 at distance @xmath45 , and given the interfering pattern @xmath22 , which we denote by @xmath237 , is equal to @xmath238 cf the proof of proposition [ p.emergency ] . hence , by independence , the mean local delay , given the distance to the receiver @xmath131 , is equal to @xmath239= \frac{e(r)e'(r)b(r)}{p(1-p)}\,.\ ] ] the remaining part of the proof follows the same lines as the proof of proposition [ p.delaysuv ] . the function @xmath240 ( corresponding to the laplace transform of @xmath22 ) can be explicitly given in the case of poisson and poisson - cluster point processes . this allows us to study the impact of the clustering of interference in a manet . in particular , we have the following expressions . for poisson field of interfering nodes of @xmath22 with intensity @xmath23 on the plane we have @xmath241 for a poisson - line interferers ( cf section [ sss.poisson-line ] ) we have @xmath242 both expressions follow from the evaluation of the laplace transforms of the respective point processes . for a poisson process this gives @xmath243 and can be evaluated to the required expression ( which also appeared in ( * ? ? ? * equation 17.43 ) . @xcite gives the following formula for the laplace transform of the poisson - line field of interferers @xmath244 from which our expression for @xmath245 follows . [ poisson_ppl ] note that @xmath246 hence , for the same intensities of points @xmath26 , @xmath247 . this implies that the mean end - to - end delay is larger in the poisson - line field of interferers than in the homogeneous poisson field . as the former process shows more clustering , we can conclude that _ clustering in the interfering field increases end - to - end delays_. [ r.flightw ] the average speed of the packet progression on the distance from @xmath190 to @xmath194 can be expressed as in corollary [ p.delaysuv ] . we now present some numerical illustrations how this speed depends on the model s parameters . we first assume only constant noise @xmath248 . figure [ fig.varwm10000 ] shows the speed of packet progression with respect to @xmath248 . we have assumed @xmath167 , @xmath65 and @xmath249 , the value which maximizes the average long - distance speed without noise . the curves correspond to different routing distances @xmath250 . figure [ fig.varmw ] shows the same speed of packet progression with respect to @xmath194 for @xmath251 . for @xmath250 we keep @xmath249 the value which maximizes the average long - distance speed without noise . both figures [ fig.varmw ] and [ fig.varwm10000 ] show a sharp cut of phenomenon : the speed of packet progression drops rapidly to zero beyond some critical noise level and/or route length . the impact of the noise is stable below its critical value ( for a given distance ) . comparing figure [ fig.varm ] to figure [ fig.varmw ] we see that after the `` fast acceleration '' the packet reaches a constant speed which can be maintained on routes shorter than some critical length depending on the level of noise . figure [ fig.chlambda ] shows the mean end - to - end speed with respect to the spatial density of the field of poisson interferers @xmath25 for two values of the aloha parameter @xmath252 and @xmath253 used by the interferers . as in the case of constant noise , the speed rapidly falls from an almost constant value ( corresponding to the case without interferers ) to zero . figure [ fig.ppl ] shows the mean end - to - end speed with respect to the spatial density of the field of thepoisson - line process of rate @xmath24 and @xmath25 for @xmath252 and @xmath253 . we observe that this speed is smaller than the mean speed in a field of poisson interferers with a comparable spatial density @xmath254 ( cf . figure [ fig.chlambda ] ) ; this confirms remark [ poisson_ppl ] . in the nn model . ] in the nn model . ] , @xmath255 ] of poisson - line process , @xmath255 ] as previously mentioned , in order to solve the problem of long hops slowing down the packet progression in the presence of external noise , one can add fixed relay stations , thus allowing efficient routing on long poisson routes . in this section we will study this problem . we assume regularly spaced `` fixed '' relay nodes @xmath16 , where @xmath17 is some fixed distance . these nodes are added to the poisson route @xmath15 , they use the same aloha parameter @xmath0 and the nn routing is considered on @xmath256 . we wish to compute the mean local delay on @xmath256 in the presence of noise , and thus characterise the speed of packet progression on long routes equipped with fixed relay stations . this will allow us to optimise the performance of the network in the parameter @xmath257 . it should be noted that a small value of @xmath257 makes the fixed relay nodes too dense , creating a lot of interference and making the routing hops very small . on the other hand , a large value of @xmath257 leads to long hops which slow down the packet progression . [ p.delaysld_w ] assume @xmath258 to be a poisson route equipped with fixed relay stations with the inter - relay distance @xmath257 , as described above . the mean local delay in the nn routing on this model in the presence of external noise of the form ( [ e.w ] ) as in proposition [ p.delaysuv_w ] is equal to : @xmath259}\\ & = & \frac\lambda{p(1-p)(\epsilon+\lambda ) } \bigg ( \frac{1}{\delta } \int_0^{\delta } e(z)e'(z ) e^{-\lambda z } e^{h(0,z ) } { \mathrm{d}}z\nonumber\\ & & + \frac{\lambda}{\delta } \int_0^{\delta } \int_0^{z}e(r)e'(z ) e^{-\lambda r } e^{h(z - r , r)+\log(h(z - r , r ) ) } { \mathrm{d}}r { \mathrm{d}}z \bigg ) \nonumber \\ & & \hspace{-1em}+ \frac\epsilon { p(1-p)(\epsilon+\lambda ) } \bigg ( \frac{\lambda}{\delta } \int_0^{\delta } e(z)e'(z ) e^{-\lambda z } e^{h(-z , z ) } { \mathrm{d}}z\nonumber\\ & & \hspace{1em } + e ^{-\lambda\delta } b(\delta ) e(\delta ) e^{h(0,\delta)-\log(h(\delta,\delta ) ) } \bigg ) \nonumber\end{aligned}\ ] ] with @xmath260 and @xmath261 . let us suppose that a typical node 0 of @xmath262 is at @xmath189 . we know that this typical node belongs to @xmath22 with probability @xmath263 and to @xmath264 with probability @xmath265 . thus we have @xmath266=\frac\lambda{\epsilon+\lambda } { { \mathbf e}}^{0}[l\|0\in \psi]+ \frac\epsilon{\epsilon+\lambda } { { \mathbf e}}^{0}[l\|0\in { \,{\mathcal{g}}}].\ ] ] let us now assume that the typical node @xmath193 is in @xmath22 and we use the fact that the nodes in @xmath22 are independent of the nodes in @xmath264 . if its closest node ( towards the right ) is in @xmath264 , the average delay to leave node @xmath193 is : @xmath267 . conditioning by the fact there is no node of @xmath22 in @xmath268 $ ] ( that occurs with probability @xmath269 ) , we find the first integral in the first line of ( [ el ] ) . we still assume that the typical node is in @xmath22 but now we also assume that the closest node ( towards the right ) is in @xmath22 . computing the average delay in this case , we find the second double integral of the first line of ( [ el ] ) ; @xmath270 is actually the density of the right - hand neighbor s position . the contribution @xmath271 corresponds to the fact that the node in @xmath264 located at @xmath78 is not taken into account by @xmath272 since the summation in @xmath273 is for @xmath274 . the second part of the contribution for @xmath275 $ ] corresponds to the case where node @xmath193 is in @xmath264 . if we assume that the right - hand neighbor of @xmath193 is in @xmath22 , the contribution of the delay corresponds to the integral of the proposition in the second line of @xmath275 $ ] . there is no correction needed for @xmath276 since the typical node @xmath193 is in @xmath264 and does not contribute to the delay as it is in @xmath276 . the last term of the second line in ( [ el ] ) corresponds to the case where node @xmath193 and its right - hand neighbor are in @xmath264 . this occurs with probability @xmath277 . the node in @xmath264 located at @xmath257 does not contribute to the delay since we analyze the delay from @xmath193 and its right - hand neighbor in @xmath264 ( thus located at @xmath257 ) . thus @xmath278 must be corrected by @xmath279 . in figure [ fig.grw ] , we plot the mean long - distance speed @xmath280)$ ] with respect to @xmath257 with fixed noise of levels @xmath281 and @xmath282 . comparing figure [ fig.varmw ] to figure [ fig.grw ] one can observe that the optimal inter - relay distance @xmath257 of the fixed structure corresponds to the smallest length @xmath194 of the route on which the mean speed of packet progression attains its maximum without the fixed infrastructure . in other words , the relay stations should `` break down '' long routes into segments of the length optimal for the end - to - end routing given the noise level . for @xmath281 and @xmath282 . we have studied performances of end - to - end routing in manets , using a linear nearest - neighbor routing model embedded in an independent planar field of interfering nodes using aloha mac . we have developed numerically tractable expressions for several performance metrics such as the end - to - end delay and speed of packet progression . they show how the network performance can be optimized by tuning aloha and routing parameters . in particular , we show a need for a well - tuned lattice structure of wireless relaying nodes ( not back - boned ) which help to relay packets on long random routes in the presence of a non - negligible noise . our analysis does not take into account packet queuing at the relay nodes , which we believe can be introduced into our model in future work .	in our basic model , we study a stationary poisson pattern of nodes on a line embedded in an independent planar poisson field of interfering nodes . assuming slotted aloha and the signal - to - interference - and - noise ratio capture condition , with the usual power - law path loss model and rayleigh fading , we explicitly evaluate several local and end - to - end performance characteristics related to nearest - neighbour packet relaying on this line . we study how these metrics depend on the density of relaying nodes and interferers , tuning of aloha and on the external noise level . we consider natural applications of these results in a _ vehicular ad - hoc network _ , where vehicles are randomly located on a straight road . we also propose to use this model to study a `` typical '' route traced in a ( general ) planar ad - hoc network by some routing mechanism . such a decoupling of a given route from the rest of the network in particular allows us to quantitatively evaluate the non - efficiency of long - distance routing in `` pure ad - hoc '' networks , previously observed in @xcite , and the need for a well - tuned structure of `` fixed '' relaying nodes . we consider several extensions of our basic poison - line - in - poisson field - model , notably as _ poisson - line ad - hoc network _ in which all nodes ( including interfering ones ) are randomly located on a poisson process of lines ( routes ) . in this case our analysis rigorously ( in the sense of palm theory ) corresponds to the typical route of this network . manet , vanet , aloha , sinr , mac , routing , end - to - end delay , packet speed , poisson , poisson - line process , doubly - stochastic poisson process , lattice .
in recent years there is a substantial amount of literature dealing with the dislocation pileups in crystals within the continuum approach ( see , e.g. , @xcite and the references therein ) . the proposed models turn out to be quite useful as they can predict the size effect for the yield stress , which agrees quite well with the experimental data @xcite but however does not confirm the well - known empirical law formulated by @xcite and @xcite . a traditional explanation of the hall - petch relation , based on the classical dislocation pileup model considered in @xcite , is that dislocation pile - ups serve to enhance the stress felt at grain boundaries . however , the leibried s dislocation pileup model differs from the contemporary continuum models @xcite in two important aspects : firstly the absence of a threshold stress for dislocation nucleation , and secondly the absence of a finite - sized dislocation - free region . this is the motivation for us to reconsider the leibfried s model in order to resolve this discrepancies . leibfried s one - dimensional theory of dislocation pileups is based on the force equilibrium : the peach - koehler resultant force acting on a dislocation produced by applied external stress and by other dislocations must vanish @xcite . within the continuum approximation one can deduce from here the well - known integral equation @xcite , provided the dislocation density is not zero . this equation need not be satisfied in a dislocation - free zone . in his now classical paper @xcite mentioned that the natural way to find the stable equilibrium distribution of dislocations if such a zone occurs is to use the variational principle of minimum energy : among all admissible distributions of dislocations the true stable distribution minimizes energy of the crystal ( cf . with the leds - hypothesis formulated in @xcite ) . to the best of the author s knowledge , the one - dimensional continuum model of dislocation pileups based on the energy minimization taking the self - energy of dislocations into account has not yet been proposed . the aim of this short paper is to fill this gap . we start from the expression for the energy of crystal containing an array of dislocations which includes also the self - energy of dislocations . the self - energy density is proportional to the absolute value of the dislocation density , so this does not change the resultant force acting on the dislocation as well as the integral equation except the forces acting at the tails of the pileups . however , this self - energy influences the stable equilibrium distribution of dislocations essentially . we will show that the density of dislocations is identically zero if the applied stress is found below some critical value called the yield ( or threshold ) stress . besides , the number of dislocations depends on the applied stress non - linearly , in contrast to the classical theory . the results of the proposed theory agree quite well with those of the continuum models of dislocation pileups obtained in @xcite . a continuum model of dislocation pile - ups taking into account the frank - read source proposed in @xcite ( see also @xcite ) leads to the similar results although it does not have the energetic structure . in the next section we present the continuum model of dislocation pileup accounting for the self - energy of dislocations . in section 3 the energy minimization problem is solved and the related property of the solution is discussed . finally , section 4 concludes the paper . consider the plane strain problem of an infinite crystal which is uniformly loaded by a shear stress @xmath0 applied at infinity . under this loading condition a linear array of equal number of positive and negative edge dislocations may occur on the slip line which is chosen to be the @xmath1-axis ( see fig . [ fig : a ] ) . the dislocation lines are parallel to the @xmath2-axis , while their burgers vectors are directed along the @xmath1-axis . we assume that there are two obstacles ( like two inclusions or grain boundaries ) at @xmath3 so that dislocations are confined to stay in the interval @xmath4 of the @xmath1-axis . in the continuum limit we may replace the sum of many closure failures induced by dislocations in form of step functions by a smooth function @xmath5 ( see fig . [ fig : pileup3 ] ) . the obstacles impose the following constraints on this function @xmath6 now we present the resultant inverse plastic distortion @xcite in the form @xmath7 with @xmath8 being the dirac - delta function . differentiating this equation with respect to @xmath1 we obtain @xmath9 the interpretation of is quite simple : if we integrate this equation over a circle @xmath10 with the middle point at @xmath11 and the radius @xmath12 , then @xmath13 thus , we get the closure failure of an amount @xmath5 which should be equal to the net burgers vector of all dislocations within the interval @xmath14 . denoting the dislocation density by @xmath15 , we get @xmath16 so , it is extremely useful to think of this double pileup of dislocations as a mode ii crack and to interpret the closure failure @xmath5 as the crack opening which is shown schematically in fig . [ fig : pileup3 ] . the energy of this system can be found directly from the energy of crystal containing continuously distributed dislocations in the interval @xmath17 of the @xmath1-axis . indeed , for the plane strain state the energy functional per unit depth of the crystal reads @xmath18=\int_{a}\phi ( \varepsilon _ { \alpha \beta } -\varepsilon ^p_{\alpha \beta } ) \ , dxdy-\int_{\partial a}\tau _ \alpha u_\alpha \ , ds,\ ] ] where @xmath19 is the cross - section area of the crystal in form of a cylinder , @xmath20 the free energy density , @xmath21 the symmetric part of the plastic distortion , while @xmath22 is the external traction acting at the boundary @xmath23 . substituting @xmath24 into this energy functional and integrating the term @xmath25 by parts using the equilibrium equation and the boundary conditions , we can show that this term is canceled out with the last term , so @xmath26 let us first exclude the self - stress ( causing the self - energy of dislocations ) and neglect the influence of the boundary of crystal by assuming that @xmath19 occupies the whole @xmath27-plane . then we can present the shear stress @xmath28 on the @xmath1-axis in the form @xmath29 the first term in this formula gives the stress field induced by the continuous distribution of dislocations excluding those in the neighborhood of @xmath1 ( the integral @xmath30 in is defined as cauchy s principal value ) , the second term corresponds to the stress field caused by the external shear stress @xmath0 . substituting into the energy functional and integrating the first term by parts , we obtain finally @xmath31=-\frac{d}{2}\int_l\int_l \ln \|x-\xi \|\varphi ^\prime ( x)\varphi ^\prime ( \xi ) \,d\xi \ , dx -\int_l \tau \varphi ( x)\ , dx.\ ] ] the double integral in is the symmetric form of the quadratic functional which should be understood in the following sense @xmath32 it is interesting to note that this quadratic functional is positive definite and obeys an inequality similar to that of wirtinger ( see @xcite ) . note also that , since the left - hand side is invariant with respect to the scaling @xmath33 due to the boundary condition , the arguments in the log - kernel can be regarded as dimensionless . functional is the energy functional of the classical continuum dislocation theory in which , as we have seen from the previous derivation , the self - stress of dislocations is excluded . this means that the self - energy of dislocations is neglected . let us now take the self - energy of dislocations into account by the following deliberations . it is well - known that the self - energy of one dislocation ( per unit length ) does not depend on its sign and equals @xmath34 , where @xmath35 , with @xmath36 being the distance of dislocation to the boundary of crystal and @xmath37 the poisson ratio @xcite . since the dislocation density is @xmath38 and since the self - energy does not depend on the sign of dislocations , the total self - energy of dislocations is given by @xmath39 adding this energy to the functional we get finally @xmath40=-\frac{d}{2 } \int_{l } \int_{l } \ln \|x-\xi \|\ , \varphi ^\prime ( x ) \varphi ^\prime ( \xi ) dx d\xi -\int_{l}\tau \varphi ( x ) dx+\kappa \mu b\int_{l } \|\varphi^\prime ( x)\| dx,\ ] ] we formulate the following variational principle : the stable equilibrium distribution of dislocations corresponds to the minimizer @xmath5 of functional under the constraints . the existence of minimizer is guaranteed by the above mentioned inequality for the quadratic functional provided in @xcite . symmetry of the problem implies that @xmath5 is even and @xmath41 is odd . it can be shown that the presence of @xmath42 in the last term of the energy functional causes the minimizer to have the dislocation - free zone as observed in @xcite . therefore we assume that there is a length @xmath43 such that @xmath44 thus , there are two boundary layers where positive and negative dislocations pile up against the obstacles and the dislocation - free zone in - between . let us first fix the constant value of @xmath5 in @xmath45 , @xmath46 , as well as the length @xmath47 . varying functional and taking into account we get for the first variation @xmath48 \delta \varphi ( x)\ , dx.\ ] ] there is no contribution of the last integral in due to the fact that it is equal to @xmath49 which is fix . since @xmath50 may be chosen arbitrarily for @xmath51 , @xmath5 should satisfy the singular integral equation @xmath52 now , it is not difficult to show that the variation of @xmath47 leads to the boundary conditions @xmath53 which guarantee the continuity of the dislocation density . thus , equation is subjected to the boundary conditions and . the solution of , , and , found in @xcite ( see also @xcite ) , reads @xmath54 from here one easily finds @xmath5 in terms of the elliptic integrals ( see @xcite , p. 276 ) @xmath55 & \text{for $ x\in l_\pm $ } , \\ \frac{\tau } { \pi d}[ce(q)-\frac{a^2}{c}k(q ) ] & \text{otherwise } , \end{cases}\ ] ] where @xmath56 in formula @xmath57 and @xmath58 are the incomplete elliptic integrals of the first and second kind , respectively , while @xmath59 and @xmath60 are the complete elliptic integrals of the first and second kind . mention that @xmath61 gives the total number of dislocations of either sign in the pileups ( cf . ) . up to now @xmath46 is an unknown quantity . it can be found by minimizing energy as function of @xmath47 . substituting the solution into the energy functional and using we obtain @xmath62 the direct consequence of is that for @xmath63 the minimum of energy is achieved at the end - point @xmath64 giving @xmath65 . indeed , for @xmath43 we have @xmath66 , so , @xmath67 in this case and the minimum is achieved at @xmath64 . thus , the threshold stress @xmath68 . note that this threshold stress is inversely proportional to the size of the grain exhibiting clearly the size effect ( cf . with @xcite ) . : a ) bold line : theory with self - energy of dislocations , b ) dashed line : classical theory , width=302 ] in general the integral in can not be found in closed analytical form . however it is easy to find minimum of with respect to @xmath47 numerically . for this purpose let us introduce the following dimensionless variable and quantities @xmath69 then function @xmath70 is given by @xmath71 \quad \text{for $ \zeta \in ( \alpha , 1)$},\ ] ] where @xmath72 formula becomes @xmath73 we evaluate this function numerically using _ mathematica_. the minimum of @xmath74 with respect to @xmath75 is also sought numerically . after finding @xmath76 at which energy reaches the minimum , we find the total number of dislocations of either sign in the pileups by ( cf . ) @xmath77.\ ] ] in fig . [ fig:1 ] the graph of @xmath78 versus the shear strain @xmath79 is plotted , where , for comparison , the dashed straight line @xmath80 obtained by the classical theory is also shown . for the numerical calculation we took @xmath81 , @xmath82 , @xmath83 . it is seen that for @xmath84 no dislocations can be formed . the curve @xmath85 obtained from the energy minimization is a non - linear function of @xmath79 , but for @xmath86 it approaches quickly the straight line @xmath80 . : a ) bold line : theory with self - energy of dislocations , b ) dashed line : classical theory , width=377 ] fig . [ fig:2 ] shows the normalized dislocation density @xmath87 as function of the dimensionless coordinate @xmath88 for @xmath89 which , in contrast to the classical distribution @xmath90 ( presented by the dashed line ) , depends on the stress ( or the shear strain @xmath79 ) through the half - length of the dislocation - free zone @xmath75 . however , this haft - length approaches quickly zero for @xmath86 , so the distribution of dislocations approaches quickly that of the classical theory at large @xmath79 . it is shown in this paper that the account of self - energy of dislocations leads to the existence of the threshold stress at which the dislocations nucleate . the total number of dislocations depends non - linearly on the applied shear stress , however it approaches quickly that of the classical theory for the stress three times larger than the threshold value . 40 berdichevsky , v.l . , le , k.c . dislocation nucleation and work hardening in anti - planed constrained shear . continuum mech . 18 , 455 - 467 . chakravarthy , s.s . , curtin , w.a . effect of source and obstacle strengths on yield stress : a discrete dislocation study . solids 58 , 625 - 635 eshelby , j.d . , frank , f.c . , nabarro , f.r.n . , 1951 . the equilibrium of linear arrays of dislocations . philos . mag . a 42 , 351 - 364 . friedman , l.h . , chrzan , d.c . , 1998 . continuum analysis of dislocation pile - ups : influence of sources . philos . mag . a 77 , 1185 - 1204 . gradshteyn , i.s . , ryzhik , i.m . , 2000 . table of integrals , series , and products . academic press , san diego . gurtin , m. e. , 2002 . a gradient theory of single - crystal viscoplasticity that accounts for geometrically necessary dislocations . j. mech . solids 50(1 ) , 5 - 32 . gurtin , m.e . , anand , l. , lele , s.p . , 2007 . gradient single - crystal plasticity with free energy dependent on dislocation densities . solids 55(9 ) , 1853 - 1878 . hall , c.l . , 2010 . asymptotic expressions for the nearest and furthest dislocations in a pile - up against a grain boundary . philos . mag . a 90 , 3879 - 3890 . hall , e.o . the deformation and ageing of mild steel : iii discussion of results . proc . phys . soc . b 64 , 747 - 753 . hansen , n. , kuhlmann - wilsdorf , d. , 1986 . low energy dislocation structures due to unidirectional deformation at low temperatures . mater . 81 , 141 - 161 . hirth , j.p . , lothe , j. , 1968 . theory of dislocations . mcgraw - hill , new york . kaluza , m. , le , k.c . , 2011 . on torsion of a single crystal rod . j. plasticity 27 , 460 - 469 . kochmann , d.m . , le , k.c . dislocation pile - ups in bicrystals within continuum dislocation theory . j. plasticity 24 , 2125 - 2147 . kochmann , d.m . , le , k.c . plastic deformation of bicrystals within continuum dislocation theory . solids 14 , 540 - 563 . lardner , r.w . , 1974 . mathematical theory of dislocations and fracture . university of toronto press . an inequality for a quadratic functional . vietnam journal of mechanics 31 , 155 - 158 . le , k.c . , 2010 . introduction to micromechanics . nova science . le , k.c . , nguyen , b.d . , 2012 . polygonization : theory and comparison with experiments . 59 , 211 - 218 . le , k.c . , nguyen , b.d . , 2013 . on bending of single crystal beam with continuously distributed dislocations . j. plasticity 48 , 152 - 167 . le , k.c . , sembiring , p. , 2008a . analytical solution of plane constrained shear problem for single crystals within continuum dislocation theory . arch . 78 , 587 - 597 . le , k.c . , sembiring , p. , 2008b . plane - constrained shear of a single crystal strip with two active slip - systems . solids 56 , 2541 - 2554 . , sembiring , p. , 2009 . plane constrained uniaxial extension of a single crystal strip . j. plasticity 25 , 1950 - 1969 . leibfried , g. , 1951 . verteilung von versetzungen i m statischen gleichgewicht . zeitschrift fr physik 130 , 214 - 226 . ohno , n. , okumura , d. , 2007 . higher - order stress and grain size effects due to self - energy of geometrically necessary dislocations . solids 55 , 1879 - 1898 . petch , n.j . , 1953 . the cleavage strength of polycrystals . j. iron steel inst . 174 , 25 - 28 . ockendon , j.r . , voskoboinikov , r. , chapman , s.j . , 2007 . continuum and discrete models of dislocation pile - ups . i. pile - up at a lock . solids 55 , 2007 - 2025 . weertman , j. , 1996 . dislocation based fracture mechanics . singapore , world scientific .	a continuum model of dislocation pileups that takes the self - energy of dislocations into account is proposed . an analytical solution describing the distribution of dislocations in equilibrium is found from the energy minimization . based on this solution we show ( i ) the existence of a critical threshold stress for the equilibrium of dislocations within a double pileup , and ( ii ) the existence of a non - linear regime in which the number of dislocations in a double pileup does not scale linearly with the resolved external shear stress , contrary to the classical double pileup model . dislocations , self - energy , dislocation nucleation , yield stress
a lattice formulation is one of the most powerful tools for a nonperturbative study . we present here a completely new formulation of the super yang - mills theory on lattice with an exact fermionic symmetry . there are several pionieering attempts for contructing super yang - mills theories on lattice . the first approach , the wilson fermion and mass fine - tuning procedure , is a well - simulated method @xcite . it is based on an observation that , in the perturbative expansion , a gaugino mass is the unique relevant operator which breaks supersymmetry around the susy fixed point @xcite . nevertheless , nobody knows whether this observation is nonperturbatively correct and there exists the susy fixed point in the wilson fermion theory . second is an appoach with the staggered fermion @xcite . this formulation may allow us to have a correspondence between fermion and boson . however , the counting of the degrees of freedom is still difficult for yang - mills theory and the full theory has not been constructed until now . third approach is based on the domain wall fermion @xcite . this approach naturally has a vanishing mechanism of the gaugino mass , although , in order to impose a majorana condition , one must introduce a nonlocal condition along the fifth direction . a simulation has also been done @xcite . our new formulation may look like the second approach in some aspects . however , the crucial difference is the presence of an exact fermionic symmetry in our formulation . no fine - tuning is required for the symmetry and we show that it is not brs - like . we start our consideration with a model defined on a fundamental lattice . the model will be called as the one - cell model . then we show that the model may be extended to the whole space while keeping the fermionic symmetry found for the one - cell model . the extended model has a slightly unusual lattice structure , which will be explained later . in this section , we present the lattice action and susy transformations for the one - cell model . each coordinate of a site takes only 0 or 1 in the unit of @xmath0 ( lattice constant ) . our action of gauge and fermion system is a normal one except the facts that it is resricted on a fundamental lattice and fermi fields , @xmath1 , are real . @xmath2 @xmath3 where @xmath4 is a base point for the plaquette @xmath5 and a symbol @xmath6 implies @xmath7 for the @xmath8 coordinate . the lattice has one plaquette in d=2 case and six plaquettes in d=3 . the coefficients @xmath9 of the fermion action appear similarly to the usual staggered fermion . however the fermion here is the real fermion . we also use the symbol @xmath10 to indicate either @xmath11 or @xmath12 , allowed inside the cell when we start from the site @xmath13 . the susy transformation of gauge fields is defined as @xmath14 where @xmath15 is a gauge - covariantly translated fermion , @xmath16 and @xmath17 is a grassmannian parameter . for fermi fields , we write the transformation as @xmath18 where @xmath19 implies @xmath20 and @xmath21 is a grassmannian parameter . now , we derive relations between the introduced parameters , which are required for the susy invariance . they come from four important conditions . this is the invariance of the fermion action under gauge field variation . it is nothing but the vanishing condition of fermion cubed terms : @xmath22 note that this condition is consistent with the staggered majorana fermion ; @xmath23 here @xmath24 is a constant and the constraint on @xmath17 is @xmath25 in particular , we find that any diagonal element vanishes @xmath26 by satisfying the first and second conditions altogether , the total action becomes invariant under the susy transformation . the second condition produces two types of graphs , i.e. those with three different indices and those with two different indices . corresponding to two types of graphs , we find the relations , @xmath27,\ ] ] @xmath28 third one is to confirm that this symmetry is not brs - like . we see that this condition is trivially satisfied because the left hand sides of the relations obtained from the second condition are nonzero . the condition is not so trivial . we consider in two steps . ( a ) in the first step , only fermion variables change . @xmath29 @xmath30 ( b ) in the second step only gauge field variables change . @xmath31 @xmath32 where @xmath33 . it is easy to show that jacobian @xmath34 . but @xmath35 implies @xmath36 . this is consistent with the solution for the first condition . we have completed the check of the susy invariance of our one - cell model . now , we extend one - cell model to the whole space . if we do it naiively , we encounter @xmath37 reduced susy not @xmath38 . so we have to consider a new type lattice theory . let us decompose the whole lattice into three units . ( b ) in a blue unit , a one - cell model lives . ( r ) in a red unit , another one - cell model lives . ( bl ) in blank , there are no cell models . our total action is obtained by summing over blue and red cells . from a point of view in fields , \(a ) ( gauge fields ) are in either a blue or a red cell . \(b ) are put only in a blue or red cell but . \(c ) ( fermi fields ) are on a boundary site of a blue or red cell . so fermi fields have dual property . now we have to check not only the in - cell condition but also inter - cell conditions . there are many constraints but we can find a solution . let us summarize our work and give some discussions . we presented a new model for lattice super yang - mills theory . after finding the susy invariant one - cell model , it is extended to the whole space . note that the representation of a link variable , @xmath39 , must be real as the susy transformation of a fermi field . finally , the embedding pattern of the one - cell model allows the modulo 2 translational invariance . the pattern may imply an embedding of spinor structure to lattice . in our forthcoming paper @xcite , some discussions will be given on this point . 9 g. curci and g. veneziano , nucl.phys . b292 ( 1987 ) 555 . i. montvay , nucl . 63 ( 1998 ) 108 . y. taniguchi , phys . d63 ( 2001 ) 014502 . h. aratyn , m. goto and a.h . zimerman , nuovo cim . a88 ( 1985 ) 225 . h. aratyn and a.h . zimerman , j.phys.a18 ( 1985 ) l487 . d. b. kaplan and m. schmaltz , chin . 38 ( 2000 ) 543 . g. t. fleming , j. b. kogut and p. m. vranas , phys . d64 ( 2001 ) 034510 . k. itoh , m. kato , h. sawanaka , h. so and n. ukita , in preparation .	we present a new lattice super yang - mills theory and its susy transformation . after our formulation of the model in a fundamental lattice , it is extended to the whole lattice with a substructure of modulo 2 .
ic10 x-1 is a dynamically confirmed black hole ( bh ) + wolf - rayet ( wr ) binary , which exhibits @xmath035 hour periodic modulation in the x - ray and optical lightcurves @xcite ; the likely association between the bh and wr was first reported by @xcite . its mass function is [email protected] @xmath6 , giving a bh mass @xmath02338 @xmath6 for the likely range in wr mass and system inclination @xcite . ic10 x-1 was previously observed by xmm - newton in 2003 . @xcite reported large intensity variation over @xmath010 ks : a factor of up to @xmath06 in the 0.52.0 kev band , and a factor 4 in the 2.07.5 kev band ; they suggested that this could be due to an eclipse by the companion star . they found that the emission spectrum was well characterised by a multi - temperature disk blackbody , and rejected a simple power law model ; the observed [email protected] kev inner disk temperature was consistent with a black hole binary in its thermal dominant state @xcite . however , this fit required a metalicity @xmath110.01 solar because solar abundance models failed to provide acceptable fits . they inferred a bh mass of @xmath04 @xmath9 for a non - spinning bh , and a mass @xmath06 times greater for a rapidly spinning bh . @xcite attempted to compare the spectra in and out of the eclipse , but there were too few counts to observe any significant differences . ic10 x-1 was also observed several times with chandra and swift , leading to the discovery of its orbital period @xcite . the 0.310 kev lightcurve of ic10 x-1 exhibits deep , periodic intensity dips that have been interpreted as eclipses by the wr ; however , the full dip profile had never previously been seen . ic10 x-1 was observed with xmm - newton for @xmath0130 ks , equivalent to @xmath01 orbital cycle , in 2012 june ( obsid 0693390101003 , pi t. strohmayer ) . @xcite found the intensity dip to be asymmetric , with ingress and egress times of @xmath03.3 and 4.6 hours , with a maximum eclipse lasting @xmath05.2 hours . additionally , they found @xmath07 mhz quasiperiodic oscillations ( see also * ? ? ? such evolution , with the egress substantially longer than the ingress , is highly reminiscent of the behavior observed in the high inclination galactic neutron star binaries known as the dipping sources ( see e.g. * ? ? ? it is possible that ic10 x-1 experiences similar processes , and we therefore present a brief review of the galactic dipping sources . when discussing these systems , the term `` eclipse '' is reserved for energy - independent occultation by the companion star ; the energy - dependent photo - electric absorption by other material is referred to as `` dipping '' , and individual events are called `` dips '' . however we note that hmxbs such as vela x-1 can also exhibit such asymmetric lightcurves due to asymmetries in the wind . it is of fundamental importance to determine whether this intensity modulation is energy - dependent or not ; energy - dependent dipping can not be due to occultation of the x - ray source by the star itself . around 10 galactic low mass x - ray binaries ( lmxbs ) exhibit periodic intensity dips in their x - ray lightcurves on the orbital period . this behavior is caused by photoelectric absorption of x - rays by cold material in the outer accretion disk , such as the bulge where the accretiion stream collides with the disk @xcite . the spectral evolution during dipping is well understood as the progressive covering of a point - like thermal component ( from the ns ) and an extended comptonized component by an extended absorber ( see e.g * ? ? ? * ; * ? ? ? * ; * ? ? ? the thermal component sees little additional absorption until the covering fraction of the comptonized component reaches @xmath050% ; then the absorption of the thermal component jumps to very large values ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? we note that some dipping sources exhibit shallow `` shoulders '' on either side of the dip ; these are energy - independent and consistent with electron scattering by the ionized outer region of the absorber ( e.g. * ? ? ? * ; * ? ? ? 4u1624@xmath1249 is a galactic dipping source with orbital period @xmath021 hours that does not exhibit eclipses . its @xmath0110 kev lightcurve from the 1985 march 25 exosat observation exhibited flat - bottomed dipping @xcite that strongly resembles the `` eclipses '' observed in the 0.310 kev xmm - newton lightcurve of ic10 x-1 . however , rxte observations of 4u 1624@xmath1249 exhibit no signs of dipping above 15 kev @xcite . it is therefore possible that the periodic intensity dips observed in ic10 x-1 could be caused by material in the outer accretion disk instead of the donor star . in some cases , the duration of dipping can be extremely high : the transient , eclipsing , galactic ns xb exo 0748@xmath12676 suffered dipping for @xmath1370% of the orbital cycle during a 1993 may asca observation @xcite . one common property of the absorption features in dipping sources is that the egress is shallower than the ingress ( and strohmayer et al . 2013 noted this property in the ic10 x-1 lightcurve ) . since the thermal and non - thermal components of the dipping sources suffer increased absorption at different rates during dipping , it is clear that the ingress is caused by a change in covering fraction , rather than a change in the density of the absorber . it is therefore possible to estimate the scale of the comptonized region ( i.e. the corona ) . @xcite estimated the corona sizes for 5 dipping sources with 130 kev luminosities @xmath14@xmath15 erg s@xmath16 using the following approximation : @xmath17 where @xmath18 is the orbital period , @xmath19 is the ingress time , @xmath20 is the radius of the accretion disk , and @xmath3 is the radius of the disk corona . @xcite estimated @xmath20 to be 80% of the roche lobe equivalent radius , @xmath21 , based on values 0.740.84@xmath21 from the simulations of @xcite and 0.9@xmath21 from @xcite . the radius of a sphere with volume equivalent to the roche lobe of the accretor , @xmath21 , was calculated from the following expression derived by @xcite ( accurate to 1% ) : @xmath22},\ ] ] where @xmath23 is the binary separation , and @xmath24 and @xmath25 are the masses of the accretor and donor respectively . @xcite found corona radii @xmath020,000700,000 km ( @xmath01060% of the disk , with 15% typical ) ; they also found that @xmath3 @xmath26 @xmath27 with 99% confidence . they presented evidence that @xmath3 is comparable to the compton radius , i.e. the maximum radius for hydrostatic equilibrium ; this radius is expected to increase with luminosity because the electron temperature is reduced by the higher flux of soft photons . more recently , the extended corona model was used to explain the odd dipping behavior exhibited by cygnus x-2 , where the blackbody component is unaffected @xcite . cygnus x-2 was simultaneously observed in radio , optical and x - ray with xmm - newton , chandra , and the european vlbi network , and a combination of ccd and grating spectroscopy was used to probe the dip evolution . the blackbody emission was unaffected by these dips , while @xmath040% of the comptonized component from the extended corona suffered extra absorption equivalent to 3@xmath28 h atom @xmath29 . the dips occured at phase 0.35 in the orbital cycle , opposite the bulge where the accretion stream impacts the disk ; @xcite demonstrated that such material viewed at inclinations @xmath160@xmath30 could cover large parts of the corona without reaching the ns . @xcite found independent evidence for extended coronae in ns lmxbs from chandra grating spectroscopy of cygnus x-2 . they examined several emission lines between 1.525 with the high energy transmission grating . these emission lines were observed at rest , and had widths between 1120 and 2730 km s@xmath16 ; these corresponded to a corona radius @xmath020,000110,000 km . one model describing such a system is that of @xcite , where the hot corona is coplanar with the cold accretion disk ; photons from the cold disk feed the corona , while high energy photons from the corona heat the disk . indeed xmm - newton observations of several bright extra - galactic x - ray binaries have revealed spectra that strongly prefer models where the comptonized component dominates the low - energy spectrum to models where the thermal component dominates @xcite . we note that @xcite suggested an alternative explanation for dipping after discovering absorption lines in xmm - newton observations of 4u1323@xmath1262 that grow stronger during dipping . they proposed that variations in the continuum and absorption lines could be produced by using a warm , highly - ionized absorber ; their model does not require partial covering , or an extend corona . @xcite were able to get acceptable fits to individual spectra of 4u1323@xmath1262 at different stages of dipping . however , @xcite examined xmm - newton observations of several other galactic dipping lmxbs , fitting different stages of dipping simultaneously ; they quote best fits with @xmath31/dof = 1.151.3 , but the many degrees of freedom ( 7201340 ) meant that the fits were formally unacceptable , and it is unclear whether this is due to the model or uncertainties in calibration etc .. furthermore , the warm absorber model can not account for the dipping in cygnus x-2 where the blackbody is unaffected @xcite . the discovery of a highly - ionized absorbing region in dipping xbs is very interesting ; however , the warm absorber model is not a credible replacement for partial covering . while we are unlikely to see such features in ic10 x-1 from the disk , it is possible that the warm wr wind may act in a similar way . one substantial difference between ic10 x-1 and the dipping lmxbs is that the donor is a wolf - rayet rather than a roche lobe - filling low mass star . @xcite showed that the wr could not be roche lobe filling , and expected only a tenuous accretion disk from the stellar wind . such a tenuous disk may not support the extended corona or the substantial absorber observed in the galactic ns dipping sources . however @xcite proposed that ic10 x-1 contains a substantial accretion disk that is fed by the wr wind rather than roche lobe overflow . the bh hmxbs cygnus x-1 , lmc x-1 , and lmc x-3 have longer orbital periods than ic10 x-1 , yet their x - ray spectra still exhibit substantial disk emission ( see e.g. * ? ? ? * and references within ) ; lmc x-3 is powered by roche lobe overflow , but cygnus x-1 and lmc x-1 are likely powered by a wind - fed disk , and ic10 x-1 is too . we note that the x - ray emission from galactic ns hmxb vela x-1 is dominated by 2 - 3 power laws with the same photon index but with different normalizations and column densities ; this is interpreted as direct emission from the neutron star as well as a scattered component from free electrons in the wind ; vela x-1 exhibits eclipses on a @xmath09 day period , and some direct emission is still seen during eclipse ( see e.g. * ? ? ? * ; * ? ? ? * and references within ) . since ic10 x-1 has a very strong wind , it is possible that some of its emission is scattered also . vela x-1 also exhibits strong fluorescence line emission , as well as recombination lines and continuum emisssion from highly ionzed gas @xcite , which we are unlikely to detect in ic10 x-1 . we consider three scenarios in this work that could produce the observed lightcurve : energy - independent occultation by the donor star , energy - dependent absorption from structure in a substantial disk , and energy - dependent absorption from the stellar wind . the latter two scenarios are indistinguishable from the x - ray data , so our spectral analysis only considers energy - independent vs. energy - dependent evolution . ic10 x-1 was observed by xmm - newton on 2012 august 18 for @xmath0130 ks ( obs i d 0693390101003 , pi t. strohmayer ) . we analyzed these data with the xmm - newton software analysis system ( sas ) version 13.0.0 . substantial background flares occurred near the beginning and end of the observation @xcite ; we identified these events by extracting a 1012 kev lightcurve with pattern==0 and flag==0 , and highlighting intervals with pn intensities @xmath10.4 count s@xmath16 . we extracted source and background spectra from the pn instrument , and analyzed these spectra with xspec version 12.8.1 . these spectra were grouped to a minimum of 20 counts per bin . the background was accumulated from a circular region of empty sky on the same ccd as the source region . since the intensity dips are thought to be caused by absorption , we simultaneously fitted non - dip and dip spectra ; the emission parameters were the same for the non - dip and dip spectra , but free to vary . unfortunately , we were unable to fit the ingress or egress because the timescale of spectral evolution is significantly shorter than the exposure time required for good spectra . we used the abund command in xspec to specify a metalicity 0.15 times solar ( appropriate for ic10 , see * ? ? ? * ; * ? ? ? * ) , assuming the solar abundances reported by @xcite . however , xspec does not allow us to specify a solar - metalicity foreground absorber when the abundance is set this way ; the galactic foreground absorption is translated into some portion of the measured low - metalicity absorption . in figure [ lc ] we present the background - subtracted 0.310 kev pn intensity lightcurve of ic10 x-1 ( black ) , and the background lightcurve in gray for comparison . horizontal lines represent the intervals used when extracting spectra : the solid line represents the non - dip interval ( 90,000 - 110,000 s after t@xmath32 ) , and the dashed line represents the dip interval ( 20,00040,000 s after t@xmath32 ) . the non - dip spectrum yielded 9250 net source counts , while the dip spectrum yielded 1150 net source counts . simultaneous fitting of the non - dip and dip spectra allowed us to test different models for the spectral evolution , which we discuss in turn . these dips could be energy - independent if caused by the donor star itself ; alternatively they could be energy - dependent absorption features due to structure in the outer accretion disk ( as seen in the galactic dipping sources , e.g. * ? ? ? * ; * ? ? ? * ) , or due to the dense stellar wind from the wolf - rayet donor . to test whether the absorption is energy - dependent , we examined the low and high energy lightcurves . to accomplish this , we created source and background lightcurves in the 0.31.0 and 4.010 kev bands . the left panel of figure [ cols ] shows the background - subtracted 0.31.0 kev and 4.010 kev lightcurves from the pn , binned to 1000 s. the right panel shows the hardness ratio defined by 4.010 kev intensity divided by 0.31.0 kev . the best fit value for a constant hardness ratio yielded @xmath31/dof = 62/12 this has a good fit probability @xmath33 , and is rejected at a 5.6 @xmath2 level . therefore , the absorption can not be simple occultation by the wr , and must come from a lower - density region such as the bulge in the outer accretion disk , or from the wr wind . we tested a series of spectral models by simultaneously fitting non - dip and dip data ( as defined above ) , with only the absorption free to vary . all of our spectral models include line - of - sight absorption , @xmath34 , and a partial covering component ( pcf in xspec ) that is described by the amount of absorption , @xmath35 , and the fraction of the emission region covered , @xmath36 . for the non - dip spectrum , @xmath35 and @xmath36 are frozen at 0 . the models also include a disk blackbody component ( diskbb in xspec ) , parameterized by inner disk temperature @xmath37 , and some sort of comptonization component ( simpl , comptt , or po short for power law in xspec ) . we use the cflux convolution model to estimate the 0.310 kev intrinsic luminosity of each emission component ; we adopt a 715 kpc distance to ic10 @xcite , which is subject to @xmath1010 kpc random uncertainties and @xmath1060 kpc systematic uncertainties . we first consider a compact corona scenario ( model i ) , where the emission consists of a disk blackbody and a comptonized component , represented by tbabs(cfluxdiskbb+cfluxcomptt ) in xspec . many authors believe that the corona resides inside the accretion disk , and can only access the hottest photons at the inner edge of the disk ( see e.g. ? ? ? * ; * ? ? ? * ) ; hence , we set the seed photon temperature of the comptt component to the inner disk temperature of the disk blackbody component . the best fit model yielded @xmath37 = [email protected] kev , with the comptt parameters unconstrained ; @xmath38 = [email protected]@xmath39 atom @xmath29 , @xmath35 = 8.2@xmath40 atom @xmath29 , @xmath36 = [email protected] , and @xmath31/dof = 420/427 . statistically , this model is a good fit to the spectrum ; however , the inner disk temperature is very low , indicating a near maximum retrograde spin for a black hole with mass @xmath030 @xmath6 , which is possible but unlikely . the total 0.310 kev luminosity for this fit is [email protected]@xmath41 erg s@xmath16 , 47@xmath1018% eddington for a @xmath030 @xmath6 bh . we next applied the simpl @xcite convolution model for comptonization ( model ii ) , where a fraction of the thermal emission , @xmath42 is comptonized , and described by a power law with photon index @xmath43 . the appropriate spectral model was tbabspcfcfluxsimpldiskbb . the best fit yielded @xmath37 = [email protected] kev , @xmath43 = [email protected] , @xmath42 = [email protected] ; @xmath38 = [email protected]@xmath39 atom @xmath29 , @xmath35 = [email protected]@xmath44 atom @xmath29 , @xmath36 = [email protected] , and @xmath31/dof = 438/428 . the results from this model are very similar to those obtained with the previous model , although the comptonized component is better constrained . again , the low temperature makes this scenario unlikely . the 0.310 kev luminosity is [email protected]@xmath41 erg s@xmath16 , i.e. 100@xmath1040% eddington for a @xmath030 @xmath9 bh . we note that since simpl is a convolution model , we can not treat the thermal and non - thermal components separately . we are therefore unable to apply different absorptions to the thermal and comptonized components . this may explain why the best fit parameters resemble those for model i , even though simpl does not specify a compact corona . finally we considered a compact corona model that includes scattering from free electrons in the wind ( model iii ) . the xspec model used was pcftbabs(cfluxdiskbb+cfluxcomptt)+ + pcftbabs(cfluxdiskbb+cfluxcomptt ) , with the seed temperature of the comptt component linked with the inner disk temperature as with model i. the spectral shapes of the two components ( i.e. inner disk temperature , electron temperature , and opacity ) were the same for each component but free to vary ; the normalizations of each component were unconstrained . this model fitted the spectra very well , which is perhaps unsurprising given that it contained 4 emission components and 11 free parameters : @xmath31/dof = 383/422 . however , the inner disk temperature ( [email protected] kev ) indicated a strong retrograde spin , as with previous models with a compact corona . in table 1 we present two values each for the absorption , normalization and partial covering parameters , corresponding to the direct and scattered emission respectively . we note that for the scattered component , the thermal flux goes to zero , while the comptonized flux is higher than for the direct component ; this is probably an artifact caused by fitting the high energy excess in the dip spectrum . since the extended corona may have access to soft photons from the outer disk in addition to the harder photons at the inner disk ( e.g. * ? * ) , the comptonized component is decoupled from the thermal component , and may be represented as a power law . we explore two models for the extended corona scenario : one where both components suffered the same absorption , and one with different absorption for the thermal and comptonized component . we modeled these scenarios as tbabs(tbabs@xmath45cfluxdiskbb+ + pcfcfluxpo ) . the additional absorption suffered by the thermal component , i.e. tbabs@xmath45 , was frozen at 0 for the non - dip spectrum . forcing the disk blackbody and power law components to experience the same absorption ( model iv ) yielded the following results : @xmath37 = [email protected] kev , @xmath43 = [email protected] ; @xmath38 = [email protected]@xmath39 atom @xmath29 , @xmath35 = 640@xmath10170@xmath39 atom @xmath29 , @xmath36 = [email protected] ; @xmath31/dof = 468/428 . such a spectrum is entirely consistent with a black hole in its steep power law state @xcite . the 0.310 kev luminosity for this model is [email protected]@xmath46 erg s@xmath16 , @xmath07% eddington for a @xmath030 @xmath6 bh . setting the in - dip blackbody absorption to 10@xmath47 atom @xmath29 and freely fitting the power law in - dip absorption ( model v ) yielded @xmath37 = [email protected] kev , and @xmath43 = [email protected] ; @xmath38 = [email protected]@xmath39 atom @xmath29 , @xmath35 = 84@xmath1014@xmath39 atom @xmath48 , @xmath36 = [email protected] , and @xmath31/dof = 412/428 . the 0.310 kev luminosity of this model is [email protected]@xmath46 erg s@xmath16 , or @xmath010% eddington for a @xmath030 @xmath6 bh . this model yielded the best fit out of the constrained models ( good fit probability 0.71 ) and also believable parameters for a bhc in the steep power law state @xcite ; the high disk blackbody temperature corresponds to a high positive spin , as seen in @xcite . the folded and unfolded spectra are presented in the left and right panels of fig . [ fold ] . cccccccccccc model & @xmath38 / 10@xmath49 & @xmath37 / kev & @xmath50 & @xmath43 & @xmath51 & @xmath42 & @xmath52 / 10@xmath49 & @xmath36 & @xmath31/dof + + model i & 6.2(6 ) & 0.133(10 ) & 150(70 ) & ? & 27.5(18 ) & & 820(180 ) & 0.886(4 ) & 420/427 + model ii & 6.9(5 ) & 0.118(11 ) & 370(170 ) & 2.43(6 ) & & 0.020(13 ) & 930(240 ) & 0.886(4 ) & 438/428 + model iii & 4.7(6)/19(5 ) & 0.20(3 ) & 180(90)/0 ( ? ) & ? & 8(2)/28(5 ) & & 100(30)/@xmath53 & 0.873(8)/0.94(3 ) & 383/422 + model iv & 3.38(16 ) & 1.49(15 ) & 4.7(13 ) & 2.50(6 ) & 21.3(16 ) & & 635(166 ) & 0.854(12 ) & 468/428 + model v & 3.91(16 ) & 1.68(9 ) & 8.1(4 ) & 3.12(10 ) & 29(3 ) & & 84(14 ) & 0.847(8 ) & 412/428 + we present a summary of the best fits from models i v in table [ modsum ] . for each model we give the line - of - sight absorption ( @xmath38 ) , the disk blackbody temperature ( @xmath37 ) and 0.310 kev luminosity / 10@xmath54 erg s@xmath16 , the photon index ( @xmath43 ) and 0.310 kev luminosity of the comptonized component , degree of comptonization where applicable ( @xmath42 ) , plus the column density and covering fraction during the dip ( @xmath35 and @xmath36 respectively ) . unconstrained parameters are indicated by ? , and parameters that do not apply to the model are indicated by . numbers in parentheses represent the 1@xmath2 uncertainties in the last digit . for the compact corona models ( i iii ) , the emission is dominated by the disk component , which is an extra problem for model ii where the luminosity was near the eddington limit . for the extended corona models ( iv v ) , the comptonized emission dominates . models i iii yield extremely low inner disk temperatures ( @xmath00.10.2 kev ) , indicating strong retrograde bh spin for a @xmath030 @xmath5 accretor . such negative spin is unlikely , as only recently has there been observational evidence for a small minority of retrograd spins ( see e.g. ? ? ? * ; * ? ? ? * ; * ? ? ? * and references within ) ; we note that @xcite studied a xmm - newton observation of a m31 ulx ( pi r. barnard ) ; analysis of this spectrum revealed that the while the spin is negative for a 10 @xmath6 bh , it is positive for a 17 @xmath6 bh ( their soft limit to the maximum bh mass ) . model iii ( compact corona and disk emission scattered by free electrons in the wind ) yielded the best @xmath31/dof ( 383/422 ) . however , we do not favor this model for two reasons . firstly , the temperature ( @xmath00.2 kev ) is very low for a @xmath030 @xmath6 bh , as discussed above . secondly , the thermal emission is absent from the scattered component , while the comptonized emission is stronger for the scattered component than the direct component ; this is more likely to represent the high energy excess in the dip spectrum than a real scattering component . we conclude that the corona is most likely to be extended . if we set the absorption of the two emission components to be the same ( model iv ) , the best fit absorption = [email protected]@xmath44 atom @xmath29 , with @xmath31/dof = 468/428 . @xmath55 = 56 with respect to model v , a @xmath16@xmath2 significance for 7 degrees of freedom . _ hence it is clear that the two emission components suffer different amounts of absorption , meaning that the dip evolution is most probably due to changes in the covering fraction rather than a simple change in opacity . _ we note that the background contributes @xmath04% of the non - dip spectrum and @xmath020% of the dip spectrum . however , our results are not particularly sensitive to the background ; removing the background completely yields @xmath31/dof = 437/428 for model v , with @xmath111@xmath2 differences in parameter values . we estimated the size of the corona from the ingress , assuming that the absorber was located in the outer accretion disk . to do this , we estimated the disk radius to be 0.8@xmath56 for the bh ( following * ? ? ? * ) , i.e. 0.300.34@xmath23 . from equation 1 , we estimate the corona size to be @xmath00.1@xmath23 . from kepler s third law , we estimate @xmath23 to be @xmath01.31.5@xmath57 cm for the range of bh and wr masses discussed by @xcite , and a corona radius @xmath010@xmath58 cm , a factor @xmath112 times bigger than observed in the galactic dipping source 4u1624@xmath1249 at a 130 kev luminosity @xmath010@xmath59 erg s@xmath16 @xcite . the best fit 0.310 kev luminosity for ic10 x-1 = [email protected]@xmath46 erg s@xmath16 , @xmath00.1@xmath60 ; the best fit 130 kev luminosity is [email protected]@xmath46 erg s@xmath16 , hence this corona estimate is consistent with the empirical relation derived by @xcite . in figure [ adcvl ] we present the estimated corona radius vs. 130 kev luminosity for ic10 x-1 and the five galactic dipping ns xbs studied by @xcite . the luminosities and corona sizes were gleaned from that paper , but the uncertainties and best fit were judged by eye ; @xcite did not cite the uncertainties for each point explicitly , but stated that the uncertainties were in the range 1030% . we set the 1@xmath2 uncertainty in @xmath3 for ic10 x-1 to 30% . we note that the scale of the corona is strongly related to the scale of the absorber , which is proportional to its distance from the x - ray emission regions . we assume that the absorbing material is located at the edge of the disk ; if it were instead located further in , for example near the circularization radius , then the corona sizes of ic10 x-1 and the galactic dipping sources would all be smaller by similar amounts . if the absorption is caused by the wr wind , then the corona could be substantially larger . galactic ns xbs can exhibit dipping due to material in the outer disk for inclinations @xmath1360@xmath30 @xcite . furthermore , the minimum inclination for eclipsing is @xmath17578@xmath30 for wr masses 1735 @xmath6 . if we assume the best fit mass function obtained by @xcite , 7.64 @xmath6 , then the allowed mass range for the bh in ic10 x-1 is @xmath02441 @xmath6 asssuming disk absorption with inclination 6078@xmath8 , slightly higher than the 2338@xmath6 range quoted by @xcite . @xcite estimated the deepest dip to last 5.2 hr , with a 3.3 hr ingress and a 4.6 hr egress , although the ingress was observed at the start of the observation , when the background radiation was high . these times correspond to @xmath015% , 10% , and 14% of the orbital period , with the total dip lasting @xmath040% of the orbital cycle . if the absorber is related to the secondary ( either an eclipse or due to stellar winds ) , then the maximum dip requires an absorber size @xmath61 , and the total dip needs @xmath02.4@xmath23 . @xcite estimated the black hole mass for a range of wr masses ( 17 , 25 , and 35 @xmath6 ) and inclinations ( 65 , 78 , and 90@xmath30 ) , and we calculated the roche lobe radius for the wr ( @xmath62 ) for each scenario . we found that @xmath62 @xmath00.340.38@xmath23 , hence the roche lobe is somewhat smaller than the maximum dip ( 0.96@xmath23 ) . the long egress would require a thick stellar wind out to @xmath13@xmath62 . @xcite found that the wr optical properties were well described by @xmath085000 k emission with a luminosity 10@xmath63 @xmath64 , although other solutions were also possible . if we assume that the wr is emitting as a blackbody , then the optically thick part of the star has a radius @xmath00.60.75 @xmath62 . therefore , the absence of eclipsing rules out inclinations @xmath137580@xmath30 ; a more massive wr allows a higher inclination , because @xmath23 increases with the mass of the system . we note that it is possible for the x - ray photosphere of the wr in ic10 x-1 to be substantially larger than the optical photosphere ( our spectra see absorption equivalent to @xmath132@xmath65 h atom @xmath29 as energy - independent ) ; a larger photosphere would impose tighter constraints on the maximum inclination . the x - ray lightcurve of ic10 x-1 contrasts with that of m33 x-7 , another high inclination bh xb @xcite . in this case , the donor ( 3649 @xmath6 ) is @xmath046 times more massive than the bh for various inclinations . @xcite folded the 0.55.0 kev lightcurves from several chandra observations , revealing a flat - bottomed dip where the minimum lasted @xmath00.16 orbital cycles . using equation 2 we calculated the roche lobe radius for the m33 x-7 donor to be @xmath00.5@xmath23 ; this is entirely consistent with a true eclipse if the donor fills , or nearly fills , its roche lobe . the light curve also shows scrappy dipping before the eclipse that may be due to an accretion stream that flows through the l1 point ; cygnus x-1 exhibits similar behavior @xcite . the wr wind system in ic10 x-1 is expected to be highly complex , and detailed modeling is well beyond the scope of this paper . @xcite found the observed properties of the wr to be well described by a model where the mass loss rate is 4@xmath66 @xmath6 yr@xmath16 . this value is considerably higher than the accretion rate required to power the observed x - ray luminosity of @xmath67 erg s@xmath16 ; if @xmath68 = @xmath69 , @xmath70 is the accretion rate , and @xmath71 is the efficiency ( assumed to be 0.1 ) , then @xmath70 @xmath07@xmath72 @xmath6 yr@xmath16 . only @xmath02% of the wr wind is required to power the x - ray source . furthermore , the intense x - ray source is expected to substantially alter the structure of the wind . a large portion of the wind will be highly ionized ; if the ionization parameter , @xmath73 , is calculated from @xmath74 where @xmath75 is the wind density , and @xmath76 is the distance from the x - ray source @xcite , then @xmath73 @xmath01000 at the location of the wr star for reasonable parameters . however , the observed optical he line used by @xcite and @xcite to calculate the mass function shows that some of the gas is neutral ; perhaps this neutral material is in the x - ray shadow of the wr . the ionization state of the absorbing material will greatly impact its absorbing signature . the x - ray ionization also impacts the wind velocity , because the winds are line - driven ( see e.g. * ? ? ? this complicates matters further , since the wind density ( therefore ionization state ) is affected by its velocity @xcite . the fact that the wr is in a close binary system with a roche lobe radius not much larger than wr radius complicates things yet further ; for instance , it is not even clear whether the galactic wr x - ray binary cygnus x-3 contains a black hole or a neutron star , because much of the line emission is dominated or contaminated by the wind ( see e.g. * ? ? ? one thing we can do is examine the absorption in the non - dip and dip spectra , and compare the difference with variations observed in other windy systems . for our favored model ( model v , assuming metalicity 0.15 solar ) , @xmath38 = [email protected]@xmath39 atom @xmath29 for the non - dip spectrum , a factor @xmath08 higher than the galactic hi column density in the direction of ic10 in both the @xcite survey and the leiden / argentine / bonn ( lab ) survey @xcite . the absorption during deepest dipping is [email protected]@xmath28 atom @xmath29 . hence the absorption varies by a factor @xmath020 over the orbital cycle . we note that this is the equivalent column density for neutral hydrogen ; the wr wind in ic10 x-1 has little hydrogen , and is unlikely to be neutral . therefore the density of the ionized material needs to be substantially higher than the @xmath38 equivalent to produce similar absorption . such an ionized absorber is expected to produce absorption features similar to those seen in galactic high inclination xbs @xcite . however , any such features in ic10 x-1 are beyond the detection powers of current x - ray telescopes . cygnus x-3 is the only galactic wr + compact object binary known @xcite , although @xcite claim that oao 1657@xmath12415 either has a wr donor already , or soon will have . even today the nature of the accretor in cygnus x-3 is unconfirmed due to complications caused by the wind ; the evidence suggests a bh accretor , but is still consistent with a ns @xcite . cygnus x-3 exhibits a near sinusoidal x - ray lightcurve with a 4.8 hr period @xcite that is considerably shorter than that of ic10 x-1 ; it is also shorter than the period of ngc300 x-1 , the only other known bh+wr xb ( @xmath033 hr * ? ? ? the absorption experienced by cygnus x-3 varies between @xmath03@xmath39 and 2@xmath28 h atom @xmath29 , assuming cold material with normal cosmic abundances ; the absorbing material appears to be symmetrical around the donor star , i.e. asymmetrical with respect to the x - ray source @xcite . @xcite proposed that the x - ray variation of cygnus x-3 included compton down - scattering of the emission by a thompson - thick , low temperature plasma cloud . @xcite noted that scenarios such as a wind - fed circumbinary envelope , or outflows from a disk are expected to be symmetrical around the accretor , and would not result in the observed flux variation ; instead they suggest that the absorption may come from an inflated bulge where the stellar wind collides with the accretion disk . they were unable to perform formal fits , due to the complexity of the model , but were able to reproduce observed spectra . with additional information from the power spectra , @xcite were able to estimate the size of the cloud to be @xmath77 cm , with a temperature @xmath78 @xmath03 kev and an optical depth @xmath07 . the behavior exhibited by ic10 x-1 is strikingly different to the proposed behavior of cygnus x-3 , in that the high energy photons contribute a larger portion of the total flux in the dip spectrum than in the non - dip spectrum . at high energies , the spectra are visibly converging in fig . [ fold ] ; also the models with energy - independent absorption systematically underestimated the high energy flux . an obvious difference between the two systems is in the orbital periods : 4.8 hr vs. @xmath034 hr ; accordingly , the accretor in cygnus x-3 is likely to be embedded in a considerably more dense wind than the bh in ic10 x-1 . we note that if we assume cosmic abundances for model v , the non - dip absorption is just 7@xmath79 atom @xmath29 , only 2@xmath79 @xmath29 above the line - of - sight absorption @xcite , and a factor @xmath015 lower than for cygnus x-3 . extra - galactic x - ray sources are often described as `` ultra - luminous '' if their luminosities are thought to be too high for stellar mass black holes , and are unassociated with the galaxy nucleus . some of these ultra - luminous x - ray sources ( ulxs ) could be stellar mass black hole binaries in an special ultra - luminous accretion state @xcite , while others may contain intermediate mass black holes ( hlx1 in eso 243 - 49 is a good candidate , see * ? ? ? @xcite analyzed twice - weekly swift observations of the well - known ulx ngc5408 x-1 over a 485 day interval , and found evidence for orbital modulation on a 115@xmath104 day period . their spectral modeling suggested 0.310 kev luminosities up to @xmath02@xmath80 erg s@xmath16 . the modulation was energy - dependent with an amplitude [email protected] below 1 kev and [email protected] above 1 kev ; they suggest that the system has a 115 day orbital period , with a giant or supergiant donor . @xcite considered the possibility that a super - orbital period was observed , but found it unlikey because it was shorter than any such period observed in bh xbs to date . they calculated the mean density for a roche lobe - filling secondary to be 1.5@xmath81 g @xmath82 , consistent with a giant or supergiant @xmath05 @xmath9 donor and a 1000 @xmath6 accretor . however , this 115 day period was considerably weakened when the swift observations were extended to @xmath01200 days ; instead , sharp intensity dips were observed on a @xmath0245 day period @xcite . @xcite studied the phase distribution of the 27 observed dips assuming a 243 day period , which was consistent with a gaussian with fwhm 0.24 in phase . the galactic bh hmxb cygnus x-1 exhibits short , energy - dependent dips that are concentrated at phase @xmath00.7 that is thought to be due to the stream of material passing through the l1 point @xcite , and @xcite suggest that ngc5408 x-1 could be exhibiting similar behavior , and infer a high inclination ( @xmath070@xmath30 ) , similar to that of ic10 x-1 . however , we note that @xcite found no evidence for energy dependent dips in the swift lightcurve of ngc5408 x-1 . furthermore , @xcite analyzed non - dip and dip spectra , and the differences are consistent with a change in normalization rather than in column density . @xcite favor an orbital period @xmath0245 days , but note that it could be super - orbital . however , @xcite consider the @xmath0245 day period to be super - orbital , as they find that the strongest peak in their power spectrum is at @xmath02.6 days , with other peaks at 115 and 183 days . we note that the m31 globular cluster xb bo 158 ( xb158 ) exhibits deep dipping on a @xmath010 ks period in some observations but not others , demonstrating that the disk is precessing @xcite ; we found that the luminosity of bo 158 varied over @xmath0440@xmath83 erg s@xmath16 during our @xmath0monthly chandra monitoring program , and suggested that this luminosity variation was due to changes in accretion rate over the disk precession cycle @xcite . it is therefore possible that the dips observed in ngc5408 x-1 are due to variations in accretion rate on the superorbital cycle also . a shorter orbital period would suggest a lower bh mass than the 1000 @xmath6 assumed by @xcite and @xcite . if ngc5408 x-1 exhibited a coplanar , extended corona as suggested by our modeling of ic10 x-1 , then the observed luminosity of @xmath84 erg s@xmath16 could be produced by the corona while keeping the emission locally sub - eddington . however , we note that there is some evidence for a 1000 @xmath6 bh from the energy spectrum and quasi - periodic oscillations exhibited by ngc5408 x-1 ( see * ? ? ? * and references within ) . another aspect to consider is whether the wind is spherical or aspherical . knowledge of the wind geometry will constrain the inclination of the system , and therefore the bh mass . when calculating the bh mass , we adopt the mass function obtained by @xcite , and estimate the bh mass for the range of wr masses used in that work . the duration of the observed x - ray dipping suggests that the absorber have a radius @xmath03 @xmath62 , yet our spectral modeling suggests that @xmath01015% of the emission is completely unabsorbed in the deepest dipping . if the absorber were spherical , this would require either a low inclination ( @xmath040@xmath8 ) or electron scattering in the wind . since the non - dip and dip spectra converge at high energies ( see fig . 3 ) , the residual emission is unlikely to be dominated by scattering . such a low inclination would suggest a bh mass @xmath05065 @xmath6 for a wr with mass 1735 @xmath6 . a 65 @xmath6 bh is not impossible , if it was formed via direct collapse of a metal poor , high mass star so that little mass was lost during the main sequence lifetime @xcite . we note that @xcite cite ic10 x-1 as an example of a star born in a galaxy with moderate metalicity ( they assume 0.3 solar ) , with a bh mass that is in good agreement with what they expect for such metalicities . @xcite found the maximum bh mass to be @xmath015 @xmath6 for solar metalicity , @xmath030 @xmath6 for 30% solar , and @xmath080 @xmath6 for 1% solar . we are unable to simply interpolate the maximum mass for 15% solar metalicity , but it seems unlikely to be as high as 5065 @xmath6 . one possible cause for an aspherical wr wind is stellar rotation . @xcite examined a population of single wr stars , and found evidence that 20% of the sample have flattened winds , using linear spectropolarimetry . also , @xcite suggested that the wind of the donor in cygnus x-3 is highly perturbed , and may be consistent with the presence of a disk wind ; this may be in the orbital plane . another possible cause for asymmetrical absorption in the wind would be the gas stream ( as seen at phase 0.6 in cygnus x-1 , * ? ? ? * ) , so long as the inclination is high enough . @xcite performed two dimensional simulations of such streams , finding that density of the stream is enhanced by a factor 2030 ; this is similar to the observed density variation in ic10 x-1 . furthermore , the tidal stream would produce stable absorption features that are fixed in phase , and that last for a substantial portion of the orbital cycle , again similar to ic10 x-1 . while this scenario is appealing , there are two caveats . firstly , @xcite assumed an inclination of 90@xmath30 for their simulations ; the stream would likely have a reduced effect at lower inclinations . secondly , the strength of the gas stream was found to be strongly dependent on the separation , @xmath23 , between the stars ; the simulations covered separations of 1.51.7 stellar radii , while in ic10 x-1 , the separation is 45 wr radii . in order to avoid energy - independent eclipsing by the wr , and yet still see a aubstantial stream , the inclination would likely be @xmath07578@xmath30 . using the @xcite mass function and wr mass estimates yields a bh mass @xmath02433 @xmath6 for this scenario . we have analyzed the spectral and temporal evolution of the bh+wr xb ic10 x-1 , during a 130 ks xmm - newton observation . ic10 x-1 exhibits x - ray intensity dips on a @xmath035 hr orbital period , and we considered three scenarios : energy - independent eclipses , and energy - dependent absorption from either material in the accretion disk or the wr wind . analyzing the 0.31.0 kev and 4.010 kev lightcurves showed the probability for a constant hardness ratio ( @xmath85 ) was @xmath02@xmath86 , rejected at a 5.6@xmath2 level . therefore , the intensity dips can not be due to simple occultation by the wr , and must come from a lower - density region such as the outer accretion disk or the wr wind . our favoured results were obtained from a model consisting of a disk blackbody component that was completely removed during the dip , and a partially - covered comptonized component ; a similar model has been successfully applied to high - inclination ns xbs in our galaxy ( see e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . one model for such a system is the two - phase accretion disk proposed by @xcite ; in this scenario , a coplanar , hot corona embeds the cold disk , and is fed by cool photons from the disk while hot photons from the corona heat the disk . if we assume that the dip is caused by material in the outer disk , then we may estimate the size of the corona from the dip ingress ; the thermal and comptonized emission regions in galactic dipping sources increase in absorption at different rates during dip evolution , meaning that the ingress is a change in absorbed fraction rather than a change in density ( see * ? ? ? * and references within ) . we estimate the corona size to be @xmath010@xmath4 km . remarkably , this result is consistent with the relation found between corona size and 130 kev luminosity in galactic dipping ns xbs @xcite ; the 130 kev luminosity was [email protected]@xmath46 erg s@xmath16 . for this scenario the inclination is likely to be @xmath06080@xmath30 , giving a bh mass @xmath02441 @xmath6 , which is slightly larger than the current estimate . if the absorption is due to the wr wind , then this wind is unlikely to be spherical . a spherical wind would require an inclination @xmath040@xmath30 , and a bh mass @xmath05065 @xmath6 ; however , @xcite found that the maximum mass for a stellar mass bh formed in a moderately metal poor galaxy like ic10 to be @xmath030 @xmath6 ( although they assumed a metalicity @xmath030% solar ) . it is possible that the absorption is caused by the gas stream that results from part of the stellar wind flowing through the l1 point . for this scenario , the inclination would need to be as high as is allowed by the absence of energy - independent eclipsing , @xmath07578@xmath30 . the bh mass would therefore be @xmath02433 @xmath6 , within the current range . we note that an absorbing wind would need to be considerably larger than an absorber in the disk to produce the same ingress and egress times , hence the corona would be similarly larger . while it is unclear whether the absorbing material originates in the accretion disk or wr wind , our results provide hard evidence that ic10 x-1 has a substantial extended corona . it would be capable of producing a total luminosity that exceeds the eddington luminosity while remaining locally sub - eddington . therefore , many ulxs could do this too . this would allow stellar mass bhs to power many of the ulxs observed to date . we thank the referee for their insights on how to improve the paper . we thank paul crowther for useful discussions about the properties of the wind . this research has made use of data from xmm - newton , an esa science mission with instruments and and contributions directly funded by esa member states and the us ( nasa ) . rb was supported by nasa grant go3 - 14095x . jfs was supported by nasa hubble fellowship grant hst - hf-51315.01 .	we have analyzed a @xmath0130 ks xmm - newton observation of the dynamically confirmed black hole + wolf - rayet ( bh+wr ) x - ray binary ( xb ) ic10 x-1 , covering @xmath01 orbital cycle . this system experiences periodic intensity dips every @xmath035 hours . we find that energy - independent evolution is rejected at a @xmath15@xmath2 level . the spectral and timing evolution of ic10 x-1 are best explained by a compact disk blackbody and an extended comptonized component , where the thermal component is completely absorbed and the comptonized component is partially covered during the dip . we consider three possibilities for the absorber : cold material in the outer accretion disk , as is well documented for galactic neutron star ( ns ) xbs at high inclination ; a stream of stellar wind that is enhanced by traveling through the l1 point ; and a spherical wind . we estimated the corona radius ( @xmath3 ) for ic10 x-1 from the dip ingress to be @xmath010@xmath4 km , assuming absorption from the outer disk , and found it to be consistent with the relation between @xmath3 and 130 kev luminosity observed in galactic ns xbs that spans 2 orders of magnitude . for the other two scenarios , the corona would be larger . prior bh mass ( @xmath5 ) estimates range over 2338 @xmath6 , depending on the inclination and wr mass . for disk absorption , the inclination , @xmath7 , is likely to be @xmath06080@xmath8 , with @xmath5 @xmath02441 @xmath9 . alternatively , the l1-enhanced wind requires @xmath7@xmath080@xmath8 , suggesting @xmath02433 @xmath6 . for a spherical absorber , @xmath7@xmath040@xmath8 , and @xmath5 @xmath05065 @xmath6
gelfand and shilov @xcite have first introduced the elementary solution of the @xmath8-dimensional classical diamond operator . later , kananthai @xcite has proved the distribution related to the @xmath8-dimensional ultra - hyperbolic equation , the solutions of @xmath8-dimensional classical diamond operator and fourier transformation of the diamond kernel of marcel riesz . furthermore , kananthai @xcite has showed that the solution of the convolution form @xmath11 is a unique elementary solution of the @xmath12 . in this article , we will define the bessel ultra - hyperbolic type operator iterated @xmath1 times with @xmath13 , @xmath14 we will show that the generalized function @xmath15 as defined by ( 10 ) is the unique elementary solution of the operator @xmath16 , that is @xmath17 where @xmath18 and @xmath19 is the dirac - delta distribution . @xmath20 is the shwartz space of any testing functions and @xmath21 is a space of tempered distribution . the operator @xmath0 can be expressed as the product of the operators @xmath26 and @xmath27 , that is , @xmath28^{k}\nonumber\\[.3pc ] & = \left [ \sum\limits_{i=1}^{p } b_{x_{i } } - \sum\limits_{i = p+1}^{p+q } b_{x_{i}}\right ] ^{k}\left [ \sum\limits_{i=1}^{p}b_{x_{i } } + \sum\limits_{i= p+1}^{p+q } b_{x_{i } } \right]^{k}\nonumber\\[.3pc ] & = \box_{b}^{k } \delta_{b}^{k}.\end{aligned}\ ] ] denoted by @xmath29 the generalized shift operator acting according to the law @xcite @xmath30 & \quad\ , \times \left ( \prod\limits_{i=1}^{n } \sin^{2v_{i } - 1 } \theta_{i } \right ) { \rm d}\theta_{1 } \ldots { \rm d}\theta_{n},\end{aligned}\ ] ] where @xmath31 , @xmath32 . we remark that this shift operator is closely connected with the bessel differential operator @xcite . @xmath33 u(x,0 ) & = f(x),\\[.3pc ] u_{y}(x,0 ) & = 0.\end{aligned}\ ] ] the convolution operator determined by @xmath29 is as follows : @xmath34 convolution ( 3 ) is known as a @xmath35-convolution . we note the following properties for the @xmath35-convolution and the generalized shift operator : 1 . 2 . @xmath37 . 3 . if @xmath38 @xmath39 , @xmath40 is a bounded function , @xmath41 and @xmath42 + then @xmath43 4 . from ( c ) , we have the following equality for @xmath44 . @xmath45 5 . @xmath46 . the fourier bessel transformation and its inverse transformation are defined asfollows @xcite : @xmath47 ( f_{b}^{-1}f ) ( x ) & = \left ( f_{b}f\right ) \left ( -x\right),\;c_{v}=\left ( \prod\limits_{i=1}^{n}2^{v_{i}-\frac{1}{2}}\left ( v_{i } + \frac{1}{2}\right ) \right)^{-1},\end{aligned}\ ] ] where @xmath48 is the normalized bessel function which is the eigenfunction of the bessel differential operator . the following are the equalities for fourier bessel transformation @xcite : @xmath49 & f_{b}(f\ast g)(x ) = f_{b}f(x ) \cdot f_{b } g(x).\end{aligned}\ ] ] for the equation @xmath57 , we have @xmath58 first , we consider the left side of ( 7 ) @xmath59 & = c_{v}\int_{\mathbb{r}_{n}^{+}}\left ( \sum\limits_{i=1}^{n } \frac{\partial ^{2}e(y)}{\partial y_{i}^{2}}\right ) \left ( \prod\limits_{i=1}^{n}j_{v_{i}-\frac{1}{2}}\left ( x_{i},y_{i}\right ) y_{i}^{2v_{i}}\right ) { \rm d}y \\[.3pc ] & \quad\ , + c_{v } \int_{\mathbb{r}_{n}^{+ } } \left ( \sum\limits_{i=1}^{n } \frac{2v_{i}}{y_{i } } \frac{\partial e(y)}{\partial y_{i}}\right ) \left ( \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } \left ( x_{i } , y_{i}\right ) y_{i}^{2v_{i } } \right ) { \rm d}y.\end{aligned}\ ] ] if we apply partial integration twice in the first integral and once in the second integral , then we have @xmath60 & \quad\ , + c_{v } \int_{\mathbb{r}_{n}^{+}}\;e(y)\left ( \sum\limits_{i=1}^{n } \frac{2v_{i}}{y_{i } } \frac{\partial } { \partial y_{i } } \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2}}\left ( x_{i } , y_{i}\right ) \right ) y_{i}^{2v_{i } } { \rm d}y \\[.3pc ] & = c_{v } \int_{\mathbb{r}_{n}^{+ } } e(y ) \left ( \sum\limits_{i=1}^{n } b_{y_{i } } \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2}}\left ( x_{i } , y_{i}\right ) \right ) y_{i}^{2v_{i } } { \rm d}y.\end{aligned}\ ] ] here , if we use the following equality @xcite , @xmath61 then we have @xmath63 & = -\left\| x\right\| ^{2}f_{b}e(x).\end{aligned}\ ] ] the right side of equality ( 7 ) is @xmath64 . then @xmath65 from lemma 1 and inverse fourier bessel transformation we obtain @xmath66 that completes the proof.@xmath67 given the equation @xmath68 for @xmath18 , where @xmath69 is the laplace bessel operator iterated @xmath1 times defined by @xmath70 then @xmath71 is an elementary solution of the operator @xmath69 where @xmath72 then @xmath79 is the unique elementary solution of the operator @xmath77 where @xmath80 & = \frac{(x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2 } - x_{p+1}^{2 } - \cdots - x_{p+q}^{2})^{\left(\frac{2k - n-2 \left\| v\right\| } { 2}\right)}}{k_{n}(2k ) } \label{9}\end{aligned}\ ] ] for @xmath81 now @xmath85 & \quad = \frac{1}{k_{n}(2k ) } \sum\limits_{i=1}^{n } x_{i } \frac{\partial } { \partial x_{i } } ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2 } \!-\ ! \!- \cdots - x_{p+q}^{2})^{\left(\frac{2k - n-2\left\| v\right\| } { 2}\right)}\\[.3pc ] & \quad = \frac{(2k-\!n-\!2\left\| v\right\| ) } { k_{n}(2k ) } ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2 } \!-\ ! x_{p+1}^{2 } \!- \cdots - x_{p+q}^{2})^{\left(\frac{2k - n-2\left\| v\right\| } { 2 } - 1 \right)}\\[.3pc ] & \qquad\ , \times ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2 } - x_{p+1}^{2 } - \cdots - x_{p+q}^{2})\\[.3pc ] & \quad = \frac{(2k - n-2\left\| v\right\| ) } { k_{n}(2k ) } ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2 } \!-\ ! } \!- \cdots - x_{p+q}^{2})^{\left(\frac{2k - n-2\left\| v\right\| } { 2}\right)}\\[.3pc ] & \quad = ( 2k - n-2\left\| v\right\| ) r_{2k}(x).\end{aligned}\ ] ] hence @xmath15 is a homogeneous distribution of order @xmath86 as required and similarly @xmath82 is also a homogeneous distribution of order @xmath87.@xmath67 from lemmas 1 and 2 we have @xmath93 thus from ( 4 ) we obtain @xmath94 & = \left\| x\right\| ^{-2k-2 m } \\[.3pc ] s_{2k}(x)\ast s_{2m}(x ) & = f_{b}^{-1}\left\| x\right\| ^{-2k-2 m } \\[.3pc ] & = c(v , m , k , n)\left\| x\right\| ^{-2k-2m - n-2\left\| v\right\| } \\[.3pc ] & = s_{2k+2m}(x),\end{aligned}\ ] ] where @xmath95 now @xmath99 can be written in the form @xmath103 @xmath35-convolving both sides by the function @xmath22 defined by ( 8) , we obtain @xmath104 and @xmath105 since @xmath106 we have @xmath107 hence @xmath108 by keeping on @xmath35-convolving @xmath22 , @xmath109 times , we obtain @xmath110 it follows that @xmath111 by ( 12 ) as required.@xmath67 before proving the theorems , we need to define the @xmath35-convolution of @xmath101 with @xmath15 defined by ( 10 ) with @xmath112 . now for the case @xmath113 , we obtain @xmath101 and @xmath15 as analytic functions that are ordinary functions . thus the @xmath35-convolution @xmath114 exists . now for the case @xmath115 , by lemma 6 we obtain @xmath101 and @xmath15 as tempered distributions . let @xmath89 be a compact set and @xmath116 where @xmath117 is defined closer to @xmath118 choose the support of @xmath15 equal to @xmath119 then supp@xmath15 is compact ( closed and bounded ) . so the @xmath35-convolution @xmath120 exists and is a tempered distribution from lemma 6 . given the equation @xmath121 for @xmath18 , where @xmath0 is a diamond bessel operator iterated @xmath1 times defined by @xmath122 then @xmath123 , defined by @xmath124 and @xmath125 , is a unique elementary solution of the operator @xmath0 . now @xmath126 can be written as @xmath127 by ( 2 ) . @xmath128 is a unique elementary solution of the operator @xmath77 for @xmath8 odd with @xmath129 odd and @xmath130 even , or for @xmath8 even with @xmath129 odd and @xmath130 odd . by the method of @xmath35-convolution , we have @xmath131 @xmath35-convolving both sides by @xmath101 , we obtain @xmath132 or @xmath133 it follows that @xmath134 by lemma 8 . that completes the proof.@xmath67 from theorem 1 , @xmath139 thus @xmath140 or @xmath141 @xmath35-convolving both sides by @xmath142 , we obtain @xmath143 \ast \diamond _ { b}^{r } [ ( -1)^{k}s_{2k}(x)\ast r_{2k}(x ) ] \\[.3pc ] & \quad\ , = [ ( -1)^{k - r}s_{2k-2r}(x)\ast r_{2k-2r}(x ) ] \ast \delta,\end{aligned}\ ] ] or by theorem 1 @xmath144 = [ ( -1)^{k - r}s_{2k-2r}(x)\ast r_{2k-2r}(x)].\ ] ] it follows that for @xmath135 , @xmath145 = ( -1)^{k - r}s_{2k-2r}(x)\ast r_{2k-2r}(x)\ ] ] as required . for @xmath137 , @xmath146\\[.3pc ] & = \diamond_{b}^{m - k}\delta\end{aligned}\ ] ] by theorem 1 . that completes the proofs.@xmath67 given the linear differential equation @xmath147 where the operator @xmath10 is defined by @xmath122 , @xmath8 is odd with @xmath129 odd and @xmath130 even , or @xmath8 is even with @xmath129 odd and @xmath130 odd , @xmath148 is a constant , @xmath149 is the dirac - delta distribution and @xmath150 . 1 . if @xmath152 and @xmath153 , then ( 15 ) has the solution @xmath154 which is an elementary solution of the operator @xmath10 in theorem 1 and is the ordinary function for @xmath115 . 2 . if @xmath155 @xmath152 , then the solution of ( 15 ) is @xmath156\ ] ] which is an ordinary function for @xmath157 , and a tempered distribution for @xmath158 . 3 . if @xmath159 , and suppose @xmath160 then ( 15 ) has the solution @xmath161 which is only the singular distribution . -.2pc ( the fourier bessel transformation @xmath162 ) . let @xmath163 for @xmath25 . then @xmath164 that is , @xmath165 is bounded and continuous on the space @xmath166 of the tempered distribution . moreover , by the inverse fourier bessel transformation @xmath167^{k}.\ ] ] from the fourier bessel transform we have @xmath168 & = c_{v}\int_{\mathbb{r}_{n}^{+}}\delta _ { b}^{k}\box _ { b}^{k}\delta ( y)\left ( \prod\limits_{i=1}^{n}j_{v_{i}-\frac{1}{2}}\left ( x_{i},y_{i}\right ) y_{i}^{2v_{i}}\right ) { \rm d}y \\[.3pc ] & = c_{v}\int_{\mathbb{r}_{n}^{+}}\delta _ { b}^{k}g(y)\left ( \prod\limits_{i=1}^{n}j_{v_{i}-\frac{1}{2}}\left ( x_{i},y_{i}\right ) y_{i}^{2v_{i}}\right ) { \rm d}y,\end{aligned}\ ] ] where @xmath169 for @xmath170 we have @xcite @xmath171 so we have @xmath172 & = c_{v}(-1)^{k } ( x_{1}^{2 } + \cdots + x_{n}^{2})^{k } f_{b } \box_{b}^{k}\delta ( x).\end{aligned}\ ] ] the same way we have following equality : @xmath173 since @xmath174 , we can write @xmath175 & \quad\ , \times ( x_{1}^{2 } + \cdots + x_{p}^{2 } - x_{p+1}^{2 } - \cdots - x_{p+q}^{2})^{k}\\[.3pc ] & = c_{v } [ ( x_{1}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2}]^{k}.\end{aligned}\ ] ] then there is the following inequality : @xmath176^{k}\\[.3pc ] & \leq c_{v } [ \|x_{1}^{2 } + \cdots + x_{n}^{2}\|]^{2k}\\[.3pc ] & = c_{v } \\| x\\| ^{4k}.\end{aligned}\ ] ] therefore @xmath177 is bounded and continuous on the space @xmath21 of the tempered distribution . and @xmath181\| = c_{\upsilon } m\quad \text{for\ a\ large } \ x_{i}\in \mathbb{r}^{+},\ ] ] where @xmath182 is a constant . that is , @xmath178 is bounded and continuous on the space @xmath166 of the tempered distribution . by lemma 8 , @xmath183 = \delta\ ] ] or @xmath184 = \delta . \label{17}\ ] ] if we applied the fourier bessel transform on both sides of ( 17 ) , then we obtain @xmath185 = f_{b}\delta ( x)\\[.3pc ] & c_{v } \left\langle ( \diamond _ { b}^{k}\delta ) \ast [ ( -1)^{k } s_{2k}(x ) \ast r_{2k}(x ) ] , \ \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } ( x_{i } , y_{i } ) y_{i}^{2v_{i } } \right\rangle = c_{v}.\end{aligned}\ ] ] by the properties of @xmath35-convolution @xmath186 , \phantom{\prod\limits_{i=1}^{n } } \right.\right.\\[.3pc ] & \quad\ , \left.\left . \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } \left ( z_{i } , y_{i } \right ) y_{i}^{2v_{i } } \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } \left ( x_{i } , y_{i } \right ) y_{i}^{2v_{i } } \right\rangle \right\rangle = c_{v},\\[.3pc ] & c_{v } \left\langle [ ( -1)^{k } s_{2k}(x ) \ast r_{2k}(x ) ] , \phantom{\prod\limits_{i=1}^{n } } \right.\\[.3pc ] & \quad\,\left . \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } \left ( z_{i } , y_{i } \right ) y_{i}^{2v_{i } } \right\rangle \left\langle ( \diamond_{b}^{k } \delta ) , \prod\limits_{i=1}^{n } j_{v_{i } - \frac{1}{2 } } \left ( x_{i } , y_{i } \right ) y_{i}^{2v_{i } } \right\rangle = c_{v},\\[.3pc ] & f_{b } [ ( -1)^{k } s_{2k}(x ) \ast r_{2k}(x ) ] \frac{1}{c_{v } } f_{b } ( \diamond_{b}^{k } \delta ) = c_{v}.\end{aligned}\ ] ] by lemma 9 , @xmath187\\[.3pc ] & \quad\ \times [ ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2}]^{k } = c_{v}.\end{aligned}\ ] ] it follows that @xmath180\\[.3pc ] & \quad\ , = \frac{c_{\upsilon}}{[(x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2}]^{k}}.\end{aligned}\ ] ] now @xmath188\|\nonumber\\[.3pc ] & \quad\ , = \frac{c_{v}}{\|x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{n}^{2}\|^{k } \|x_{1}^{2 } + \cdots + x_{p}^{2 } - x_{p+1}^{2 } - \cdots - x_{p+q}^{2}\|^{k}},\label{18}\end{aligned}\ ] ] where @xmath189 with @xmath190 defined by lemma 4 . then @xmath191 @xmath192 and for a large @xmath193 and a large @xmath1 , the right - hand side of ( 18 ) tends to zero . it follows that it is bounded by a positive constant say @xmath182 , that is , we obtain ( 17 ) as required and also by ( 17 ) . @xmath178 is continuous on the space @xmath21 of the tempered distribution . + @xmath62@xmath67 @xmath194 \ast [ ( -1)^{k } s_{2m}(x ) \ast r_{2m}(x ) ] ] \\[.3pc ] & \quad\ , = f_{b } [ ( -1)^{k}s_{2k}(x)\ast r_{2k}(x ) ] f_{b } [ ( -1)^{m}s_{2m}(x)\ast r_{2m}(x)]\\[.3pc ] & \quad\ , = \frac{c_{\upsilon } } { ( ( x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2})^{k+m}},\end{aligned}\ ] ] since @xmath82 and @xmath15 are tempered distribution with compact support , from lemmas 6 and 7 we have @xmath195 * [ ( -1)^{m } s_{2m}(x ) * r_{2m}(x)]\\[.3pc ] & \quad\ , = ( -1)^{k+m } [ s_{2k}(x ) * s_{2m}(x ) ] * [ r_{2k}(x ) * r_{2m}(x)]\\[.3pc ] & \quad\ , = ( -1)^{k+m } s_{2(k+m)}(x ) * r_{2(k+m)}(x).\end{aligned}\ ] ] taking the fourier transform on both sides and using theorem 4 we obtain @xmath196 * [ ( -1)^{m}s_{2m}(x ) * r_{2m}(x)]\\[.3pc ] & \quad\ , = \frac{c_{v}}{[x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2}]^{k+m}}\end{aligned}\ ] ] @xmath197^{k}}\\[.3pc ] & \qquad\,\times \frac{c_{v}}{[(x_{1}^{2 } + x_{2}^{2 } + \cdots + x_{p}^{2})^{2 } - ( x_{p+1}^{2 } + \cdots + x_{p+q}^{2})^{2}]^{m}}\\[.3pc ] & = \frac{1}{c_{v } } f_{b } [ ( -1)^{k } s_{2k}(x ) * r_{2k}(x ) ] f_{b } [ ( - 1)^{m}s_{2m}(x ) * r_{2m}(x)].\end{aligned}\ ] ]	in this article , the operator @xmath0 is introduced and named as the bessel diamond operator iterated @xmath1 times and is defined by @xmath2^{k},\ ] ] where @xmath3 where @xmath4 , @xmath5 [ 8 ] , @xmath6 , @xmath7 is a non - negative integer and @xmath8 is the dimension of @xmath9 . in this work we study the elementary solution of the bessel diamond operator and the elementary solution of the operator @xmath10 is called the bessel diamond kernel of riesz . then , we study the fourier bessel transform of the elementary solution and also the fourier bessel transform of their convolution . = tibi at 13.5pt
smectic elastomers @xcite are rubbery materials with the orientational properties of smectic liquid crystals @xcite . they possess a plane - like , lamellar modulation of density in one direction . in the smectic-@xmath0 ( sm@xmath0 ) phase , the frank director @xmath2 describing the average orientation of constituent mesogens is parallel to the normal @xmath3 of the smectic layers whereas in the smectic-@xmath1 ( sm@xmath1 ) phase , there is a non - zero tilt - angle @xmath4 between @xmath2 and @xmath3 . recently , there has been some controversy about whether shear strain and tilt of the director relative to the layer normal are coupled in smectic elastomers and whether the imposition of one necessarily leads to the development of the other . beautiful experiments by nishikawa and finkelmann @xcite , where a sm@xmath0 elastomer was subjected to extensional strain along the layer normal , found a drastic decrease in young s modulus at a threshold strain of about @xmath5 accompanied by a rotation of @xmath3 through an angle @xmath6 that sets in at the same threshold . interpreting their x - ray data , the authors concluded that there was no sm@xmath1-like order over the range of strains they probed , and that the reduction in young s modulus stems from a partial breakdown of smectic layering . recently , adams and warner ( aw ) @xcite developed a model for sm@xmath0 elastomers which assumes that @xmath2 and @xmath3 are rigidly locked such that @xmath7 . this model produces a stress - strain curve and a curve for the rotation angle @xmath6 of @xmath3 in full agreement with the experimental curves but without needing to invoke a breakdown of smectic layering . evidently , because of the assumption @xmath7 , the aw model predicts no sm@xmath1-like order . more recently , stenull and lubensky @xcite argued that shear strain and sm@xmath1-like order are coupled and that the imposition of one inevitably leads to the development of the other . they developed a model based on lagrangian elasticity that predicts , as does the aw model , a stress - strain curve and a curve for @xmath6 in full agreement with the experiment . in contrast to the interpretation of nishikawa and finkelmann of their data and to the central assumption of the aw model , ref . @xcite found that the tilt angle @xmath8 of @xmath2 is not identical to that of the layer normal , @xmath6 , implying that there is sm@xmath1-like order with non - zero @xmath4 above the threshold strain . however , estimates of @xmath6 and @xmath8 based on reasonable assumptions guided by the available experimental data of the layer normal - director coupling , turn out the same order of magnitude . the upshot is that ref . @xcite predicts sm@xmath1-like tilt above the threshold strain but that the angle @xmath4 is small . it is entirely possible that @xmath4 is smaller than the resolution of the experiments by nishikawa and finkelmann . in this case , there is no disagreement between the predictions of ref . @xcite and the experimental data by nishikawa and finkelmann . the arguments of ref . @xcite imply in particular that a sm@xmath0 elastomer subjected to a shear in the plane containing @xmath3 will develop sm@xmath1-like tilt of the director . very recently , kramer and finkelmann ( kf ) @xcite performed corresponding shear experiments using two different shear geometries . one geometry , which we refer to as tilt geometry , imposes a shear that is accompanied by an effective compression of the sample along the layer normal . in this geometry , the elastomer ruptures at an imposed mechanical shear angle @xmath9 of @xmath10 @xcite or @xmath11 @xcite , and up to these values of @xmath9 no sm@xmath1-like tilt is detected within the accuracy of the experiments of about @xmath12 , as was the case in the earlier stretching experiments of nishikawa and finkelmann . the other shear geometry , which we refer to as slider geometry , imposes simple shear and can be used to probe values of @xmath9 exceeding @xmath13 . for this geometry , the kf x - ray data provides clear evidence for the emergence of sm@xmath1-like tilt . in this communication , we generalize the model introduced by aw @xcite and use it to study the magnitude of sm@xmath1-like tilt under shear for the two geometries investigated by kf . using reasonable estimates of model parameters , we estimate the tilt angle @xmath4 , and we compare our estimates to the experimental results . the outline of the remainder of this communication is as follows . in sec . [ sec : extendingaw ] we develop our model , which generalizes the original aw model . in sec . [ sec : expgeomtiltang ] we describe the two experimental setups that we consider . we discuss the respective deformation tensors for these setups and calculate for both setups the sm@xmath1-tilt angle @xmath4 as a function of the imposed mechanical tilt angle @xmath9 . we address why the tilter apparatus is problematic for experiments where shear induces director tilt . in sec . [ sec : discussion ] we discuss our findings and we make some concluding comments . there are two appendices . in app . [ app : smallangles ] , we comment on analytical calculations of @xmath4 for small @xmath9 . in app . [ app : euler ] , we briefly discuss the euler instability in the context of the experiments by kf . in this section we devise a theory for smectic elastomers based on the _ neo - classical _ approach , developed originally by warner and terentjev and coworkers @xcite for nematic elastomers , and the subsequent extension of the _ neo - classical _ approach by aw to smectics . the _ neo - classical _ approach generalizes the classical theory of rubber elasticity @xcite to include the effects of orientational anisotropy on the random walks of constituent polymer links . it treats large strains with the same ease as they are treated in rubbers in the absence of orientational order , it provides direct estimates of the magnitudes of elastic energies , it is characterized by a small number of parameters , and it accounts easily for incompressibility . in the _ neo - classical _ approach , one formulates elastic energy densities in terms of the cauchy deformation tensor @xmath14 , defined by @xmath15 , where @xmath16 is the target space vector that measures the position in the deformed medium of a mass point that was at position @xmath17 in the undeformed reference medium . in this approach , a generic model elastic energy density for smectic elastomers that allows for a relative tilt between the director and the layer normal can be written in the form @xmath18 here and in the following , incompressibility of the material is assumed , i.e. , the deformation tensor is subject to the constraint @xmath19 . @xmath20 is the usual trace formula of the _ neo - classical _ model with @xmath21 the shear modulus , @xmath22 \ , , \end{aligned}\ ] ] where @xmath23 , with @xmath24 describing the uniaxial direction before deformation , is the so - called shape tensor describing the distribution of conformations of polymeric chains before deformation and @xmath25 is the inverse shape tensor after deformation . @xmath26 denotes the unit matrix , and @xmath27 denotes the anisotropy ratio of the uniaxial sm@xmath0 state . the contribution @xmath28 ^ 2\end{aligned}\ ] ] describes changes in the spacing of smectic layers with layer normal @xmath29 where @xmath30 is the layer normal before deformation and @xmath31 the layer compression modulus @xcite . @xmath32 and @xmath33 are , respectively , the layer spacing before and after deformation which are related via @xmath34 the tilt energy density , @xmath35 incorporates into the model the preference for the director to be parallel to the layer normal in the sm@xmath0 phase . to study the sm@xmath1 phase , we would have to include a term proportional to @xmath36 , which , however , is inconsequential for our current purposes . without the contribution @xmath37 , the model elastic energy density ( [ eq : full_f ] ) is invariant with respect to simultaneous rotations of the smectic layers , the nematic director , and @xmath14 in the target space . to break this unphysical invariance , we include the semisoft term @xcite @xmath38 , \end{aligned}\ ] ] where @xmath39 is a dimensionless parameter . the relation of the model presented here to the aw model is the following : aw assume that the layer normal and the director are rigidly locked such that the angle @xmath4 is constrained to zero . moreover , the semi - soft term @xmath37 is absent in the aw model . essentially , we retrieve the aw model from eq . ( [ eq : full_f ] ) by setting @xmath40 ( or equivalently @xmath41 ) and @xmath42 . as mentioned above , one of the virtues of the _ neo - classical _ approach is that it involves only a few parameters , and for most of these there exist experimental estimates , which we will now review briefly . the shear modulus @xmath21 for rubbery materials is typically of the order of @xmath43 . a typical value for the smectic layer compression modulus in smectic elastomers , as observed in experiments with small strains along the layer normal , is @xmath44 , which is greater than the values in liquid smectics . in previous experiments by the freiburg group , the anisotropy ratio was approximately @xmath45 @xcite , and we adopt this value for our arguments here . the value of @xmath46 can be estimated from experiments by brehmer , zentel , gieselmann , germer and zungenmaier on smectic elastomers and by archer and dierking @xcite on liquid smectics . the former experiment indicates that @xmath46 is of the order of @xmath47 at room - temperature , and the latter experiment produces a room - temperature value of the order of @xmath48 . we are not aware of any experimental data that allows us to estimate @xmath39 for smectic elastomers reliably . for nematic elastomers , it has been estimated from the fredericks effect @xcite and from the magnitude of the threshold to director rotation in response to stretches applied perpendicular to the original director @xcite that @xmath49 or @xmath50 , respectively . for our arguments here , we adopt the latter value acknowledging that @xmath39 may be considerably larger in smectics than in nematics . as we will see in the following , our findings do not depend sensitively on this assumption . for the deformations that we consider , @xmath39 appears only in the combination @xmath51 , that is , semi - softness simply adds to the director - layer normal coupling . uncertainty in estimates for @xmath52 is expected stem mainly from the spread in estimates for @xmath46 . we account for this spread by discussing several values of @xmath52 or , more precisely , for several values of @xmath53 . in this section we apply the model defined in sec . [ sec : extendingaw ] to study the behavior of sm@xmath0 elastomers in shear experiments . we consider two experimental setups . in the first setup , which is perhaps the first that comes to mind from a physicist s viewpoint , a simple shear is applied , i.e. , a shear strain in which the externally imposed displacements all lie in a single direction . figure [ fig : sliderapparatus ] shows a sketch of such an experiment , and we call the apparatus sketched in it the slider apparatus . -direction remains fixed.,width=226 ] the second setup , which is depicted in fig . [ fig : tiltapparatus ] , is one in which opposing surfaces of the sample remain essentially parallel but in which the height of the sample ( its extension in the @xmath54-direction ) decreases upon shearing . hence , the applied shear is , strictly speaking , not just simple shear . in the following we will refer to the this apparatus as a tilt apparatus . to make our arguments to follow as concrete and simple as possible , we now choose a specific coordinate system . that is , we choose our @xmath54 direction along the uniaxial direction @xmath24 of the unsheared samples , @xmath55 , and we choose our @xmath56 direction as the direction along which the sample is clamped , which is perpendicular to @xmath24 , cf . [ fig : sliderapparatus ] . with these coordinates , the deformation tensor for both the slider apparatus and the tilt apparatus is of the form @xmath57 as can be easily checked , for this type of deformation , the layer normal lies along the @xmath54-direction for the unsheared _ and _ for the sheared samples . thus , the tilt angle @xmath4 between @xmath2 and @xmath3 is identical to the tilt angle between @xmath2 and the @xmath54-axis , and we can parametrize the director as @xmath58 it is worth noting that a deformation tensor of the form shown in eq . ( [ formoflambda ] ) leads to a particularly simple expression for the semi - soft contribution to @xmath59 , @xmath60 , which combines with the tilt energy density to @xmath61 . thus , as indicated above , our model depends for the experimental geometries under consideration on @xmath39 and @xmath46 through a single effective parameter , viz.@xmath62 . from the photos of the experimental samples provided in ref . @xcite , see figs . [ fig : sliderapparatus ] and [ fig : tiltapparatus ] , it appears as if @xmath63 does not deviate significantly from 1 . in the following , we set @xmath64 for simplicity @xcite . in this event @xmath65 . in the slider apparatus , the extent of the sample in the @xmath54-direction is fixed , and hence @xmath66 . the incompressibility constraint @xmath19 thus mandates that @xmath67 . the remaining nonzero component of the deformation tensor is entirely determined by the externally imposed shear , @xmath68 , where @xmath9 is the mechanical tilt angle of the sample , see fig . [ fig : sliderapparatus ] . the only remaining degree of freedom in the problem is , therefore , the angle @xmath4 . to calculate @xmath4 as a function of @xmath9 , we insert the just discussed deformation tensor into @xmath59 and then minimize @xmath59 over @xmath4 holding @xmath9 fixed . for @xmath9 small , this can be done analytically by expanding @xmath59 to harmonic order in @xmath4 and by then solving the resulting linear equation of state for equilibrium value of @xmath4 . this type of analysis is presented in the appendix . to provide reliable predictions for larger shears , one has to refrain from expanding in @xmath4 and use numerical methods instead . to this end , we minimize @xmath59 numerically assuming , based on what we discussed at the end of sec . [ sec : extendingaw ] , that @xmath69 and @xmath70 . for this minimization , we use mathematica s findminimum routine . figure [ fig : thetasliderapparatus ] shows the resulting curves for @xmath4 as a function of @xmath9 . the dashed horizontal line in fig . [ fig : thetasliderapparatus ] is a guide to the eye ; it corresponds to the angle resolution in the kf experiments which was about @xmath71 degree @xcite . between the layer normal and the director as a function of the mechanical tilt angle @xmath9 in the slider apparatus for ( i ) @xmath72 , ( ii ) @xmath73 , ( iii ) @xmath74 , and ( iv ) @xmath75 . the dashed line corresponds to the resolution for @xmath4 in the experiments by kf.,width=279 ] for the slider geometry , the kf data produces clear evidence for the development of sm@xmath1-like tilt under simple shear , and our theoretical estimates agree well with the experimental data . however , given that thus far only two data points are available for @xmath76 , it can not be judged reliably whether our theoretical curve could be fitted to an experimental curve with more data points . it is encouraging , though , that our curve for @xmath72 agrees with the available experimental points within their errors . now we turn to the tilt geometry . the essential difference between the slider and the tilt geometry is that in the former the height of the sample @xmath77 remains constant , @xmath78 ( @xmath32 times the number of smectic layers ) whereas in an ideal tilter , @xmath77 is not constant but rather decreases as @xmath9 increases , @xmath79 . this difference has far reaching consequences . in the slider , the sample - height remains larger than the modified natural height of the sample created by the director tilt , @xmath80 . in other words , the sample is under effective tension . in the tilter , on the other hand , the sample - height @xmath81 can be smaller than @xmath82 , i.e. , the sample can be under effective compression . for a sample under effective @xmath83-compression , one has to worry , experimentally and theoretically , about all sorts of complications . most notable are perhaps buckling and wrinkling . the theory of buckling of elastic sheets is well established @xcite , and we now briefly comment on buckling in the context of the tilt geometry . as mentioned above , a sample clamped into the tilt apparatus can develop a buckling instability , if the height of the sample imposed by tilt , @xmath84 , is smaller than the natural height of the sample created by the director tilt , @xmath82 . as we will see below , our results for @xmath4 as a function of @xmath9 imply that @xmath85 , i.e. , buckling is possible . alternatively , this can be seen by calculating the engineering stress @xmath86 , which can be done using the numerical approach outline above . @xmath87 is positive for @xmath88 in the slider apparatus , whereas it turns out being negative for @xmath88 in the tilt apparatus . thus , the sample is effectively under tension in the slider apparatus whereas it is effectively under compression in the tilt apparatus , opening the possibility for buckling in the @xmath89-direction in the latter . the angle @xmath90 at which buckling sets in is expected to be comparable to that for the well known euler strut instability @xcite , @xmath91,\end{aligned}\ ] ] where @xmath92 is the thickness of the sample in the @xmath89-direction , and where clamped ( rather than hinged ) boundary conditions are assumed . a brief derivation of eq . ( [ eulerangle ] ) is given in app . [ app : euler ] . in the experiments of kf , @xmath93 and @xmath94 @xcite , which leads to @xmath95 . the experimental samples buckle immediately before they rupture @xcite at @xmath96 or @xmath97 , which is very close to our estimate for @xmath90 . the observation that buckling occurs immediately before rupturing might suggest that the former actually triggers the latter . a detailed analysis of sample - wrinkling in the tilt geometry is beyond the scope of the present paper . however , a few comments about wrinkling are in order . the wavelength of wrinkling is expected to be much shorter than that of buckling . thus , wrinkling can be harder to detect by visual inspection of an experimental sample than buckling , and there is a risk that it remains unnoticed . the main problem is , however , that wrinkling can act as to effectively reduce the mechanical shear in the sample . when designing a tilt experiment , one thus has be very careful to avoid wrinkling . if not , the effects of wrinkling can bias the data , and there is the risk to underestimate the sm@xmath1-tilt significantly . our calculation of the the sm@xmath1-tilt will be for an ideal tilter . figure [ fig : tiltschematic ] shows the tilter of fig . [ fig : tiltapparatus ] schematically in a non - tilted and a tilted configuration to make clear that shear and compression are complex in the non - ideal case because the frame - axles allowing angle change are off - set from the corners of the sample . using elementary trigonometry , one can deduce for the shears and the angle @xmath9 @xmath98 in particular , the relation between the shear angle @xmath9 and the deformation components @xmath99 and @xmath100 is not simple . returning to an ideal tilter , @xmath101 , @xmath102 , we calculate the sm@xmath1-tilt as a function of the applied mechanical shear , suppressing buckling and wrinkling . then , the deformation tensor is of the same form ( [ formoflambda ] ) as for the slider apparatus , but with the essential difference that here @xmath103 and @xmath104 . as we did for the slider apparatus , we assume @xmath64 . the incompressibility constraint then implies that @xmath105 , leaving the angle @xmath4 as the only degree of freedom . we calculate @xmath4 as a function of @xmath9 in exactly the same way as above , i.e. , we minimize @xmath59 numerically assuming the values of the model parameters discussed in sec . [ sec : extendingaw ] . the resulting curves are shown in fig . [ fig : thetatiltapparatus ] . between the layer normal and the director as a function of the mechanical tilt angle @xmath9 in the tilt apparatus for ( i ) @xmath72 , ( ii ) @xmath73 , ( iii ) @xmath74 , and ( iv ) @xmath75 . the dashed line corresponds to the resolution for @xmath4 in the experiments by kf.,width=283 ] the dip in fig . [ fig : thetatiltapparatus ] occurs because of a competition between @xmath52 which would prefer @xmath40 and @xmath31 which would prefer @xmath106 . if @xmath107 then it is possible to get curves where @xmath4 stays close to @xmath108 . the dashed horizontal line in fig . [ fig : thetatiltapparatus ] indicates the angle resolution of about @xmath71 degree of the kf experiments @xcite . as mentioned in sec . [ sec : sliderapparatus ] , the data points for the slider are compatible with @xmath109 . the elastomers used by kf in the slider and the tilter are identical , and , therefore , curve ( i ) of fig . [ fig : thetatiltapparatus ] should describe the sm@xmath1-tilt if the tilter used by kf were ideal or nearly so . note from fig . [ fig : thetatiltapparatus ] , however , that this implies that the sm@xmath1-tilt for an ideal tilter at @xmath110 or @xmath111 should be significantly larger than the experimental resolution , which is incompatible with the findings of refs . @xcite . since any effective compression of the sample along the layer normal , as in a tilter , promotes rather than hampers director rotation , we obtain a sm@xmath1-tilt at a given @xmath9 in a tilter if it occurs at the same angle in the slider . it is likely that the failure of kf to observe director rotation in their tilter is due to mechanical instability that effectively reduces the mechanical shear of the sample and thus leads to a systematic suppression of the sm@xmath1-tilt in the tilt geometry . in summary , we have developed a _ neo - classical _ model for smectic elastomers , and we used this model to study sm@xmath0 elastomers under shear in the plane containing the director and the layer normal . in particular , we investigated the tilt angle @xmath4 between the layer normal and the director for two different experimental setups as a function of the mechanical tilt angle @xmath9 measuring the imposed shear . our model builds upon the _ neo - classical _ model for smectic elastomers by aw . in their original work , aw chose not to consider the possibility of sm@xmath1 order in their theory : they forced the layer normal and the nematic director to be parallel . in our present theory , the fundamental tenet is different in that the nematic director @xmath2 and the deformation tensor @xmath14 are independent quantities that must be allowed to seek their equilibrium in the presence of imposed strains or stresses . the layer normal @xmath3 is determined entirely by @xmath14 . thus , @xmath2 and @xmath3 are independent variables that rotate to minimize the free energy - they are not locked together . in the present paper , we have for simplicity focused on idealized monodomain samples . boundary conditions in real experiments prevent this simple scenario and produce a microstructure structure of the type that aw discuss in ref . @xcite based on their original model that prohibits sm@xmath1 ordering . our current theory , which admits sm@xmath1 ordering , predicts the same type of microstructure ; stretching will produce a polydomain layer structure just as the aw theory does . in particular , the existence of sm@xmath1 ordering does not contradict the appearance of optical cloudiness . to avoid undue repetition , we refrain discussing this here in more detail and refer the reader to ref . @xcite . the main finding of our present work is that the tilt angle @xmath4 between @xmath2 and @xmath3 is non - zero if a shear in the plane containing @xmath3 is imposed . this angle depends on material parameters , the experimental setup , and the magnitude of the imposed shear . figures [ fig : thetasliderapparatus ] and [ fig : thetatiltapparatus ] depict our results for the slider apparatus and the tilt apparatus used by kf . as expected , in both setups the angle @xmath4 decreases as the value of the parameter @xmath53 increases , at fixed @xmath9 . this is because the director is becoming more strongly anchored to the layer normal . if one keeps @xmath53 fixed and varies @xmath112 instead , then for @xmath113 @xmath4 approaches @xmath9 because in this limit @xmath114 locks @xmath4 to @xmath106 . as @xmath112 is decreased then @xmath4 decreases at fixed @xmath9 , because the @xmath115 term starts to compete with the @xmath116 term . in the opposite limit of very small @xmath112 , @xmath4 is locked to @xmath40 by virtue of the tilt and semi - soft contributions to @xmath59 . we refrain to show the corresponding curves , which are similar to those depicted in figs . [ fig : thetasliderapparatus ] and [ fig : thetatiltapparatus ] , to save space . the main difference between our results for the two setups is that for @xmath117 and large @xmath9 , say @xmath118 degrees or so , @xmath4 is roughly one order of magnitude smaller than @xmath9 in the slider apparatus , whereas it is of the order of @xmath9 in the tilt apparatus . in closing , we would like to stress once more that the experiment using the slider geometry produces clear evidence of the development of sm@xmath1-like tilt under shear as predicted originally in ref . @xcite . our theory developed here allowed us to understand in some detail how the magnitude of this tilt depends on the shear geometry and on model / material parameters , such as @xmath52 , which can be reasonably estimated from nematic analogues . given these estimates , the resulting estimates for the tilt angle vary over a range from being small enough to be essentially unobservable with the x - ray equipment used by the freiburg group to exceeding the experimental resolution . while the latter is consistent with the shear experiments in the slider geometry , the former is consistent with the stretching experiments by nishikawa and finkelmann in which no sm@xmath1-like tilt was detected . we believe that the reason for the discrepancy between the rotations in the slider and tilter apparatus lies in a possible wrinkling of the samples in the tilt apparatus used by kf . mechanical instability leads to an effective reduction of the mechanical shear of the sample causing a systematic underestimation of the sm@xmath1-like tilt in the tilt geometry . another experimental approach that could be used , at least in principle , to investigate @xmath4 is to measure changes in the layer spacing , as was done by nishikawa and finkelmann @xcite . it should be noted , however , that in this approach one measures @xmath119 rather than @xmath4 directly . for small values of sm@xmath1 tilt , say @xmath120 degree , @xmath121 which is practically indistinguishable from the @xmath122 pertaining to sm@xmath0 order . for a refined comparison between experiment and theory it would be useful to critically evaluate and potentially improve the design of the tilt apparatuses regarding sample - wrinkling . also , it would be interesting to perform shear and stretching experiments with an angle resolution better than @xmath123 . we hope that our work stimulates the interest for such experiments . this work was supported in part by the national science foundation under grants no . dmr 0404670 and no . dmr 0520020 ( nsf mrsec ) ( t.c.l . ) . we are grateful to d. kramer and h. finkelmann for discussing with us some of the details of their experiments . for arbitrary sm@xmath1-tilt angle @xmath4 , the total elastic energy density @xmath59 is , for the deformations discussed in sec . [ sec : expgeomtiltang ] , a fairly complicated conglomerate of trigonometric functions ( and powers thereof ) of @xmath4 and @xmath9 . thus , we resorted in sec . [ sec : expgeomtiltang ] to a numerical approach to determine the equilibrium value of @xmath4 over a wide range of the imposed mechanical tilt angle @xmath9 . here , we focus on the regime of @xmath9 where @xmath4 is small , such that it is justified to expand @xmath59 to harmonic order in @xmath4 . in this case , the equations of state are linear in @xmath4 , of course , and are therefore readily solved . for the slider apparatus we obtain @xmath124 } \ , , \end{aligned}\ ] ] and for the tilt apparatus we get @xmath125 } \ , .\end{aligned}\ ] ] equations ( [ thetaanaslider ] ) and ( [ thetaanatilt ] ) imply , that the sm@xmath1-tilt in both setups is identical for small @xmath9 , @xmath126 this is consistent with figs . [ fig : thetasliderapparatus ] and [ fig : thetatiltapparatus ] , where the initial slopes are identical for identical values of @xmath52 ( though this is perhaps somewhat hard to see because the ordinate - scales are different in the two figures ) . an interesting related question , which we have not addressed in the main text because there is currently no experimental data available , is that of the stress that is caused by the imposed shear strains . from the above , it is straightforward to calculate this stress for small @xmath9 . inserting eq . ( [ thetaanaexpanded ] ) into the aforementioned harmonic ( in @xmath4 ) total elastic energy density provides us with an effective @xmath59 in terms of @xmath9 . from this effective @xmath59 , we readily extract the engineering or first piola - kirchhoff stress as @xmath127 equation ( [ engineeringstress ] ) highlights a problem that arises when the tilt and the semi - soft contributions @xmath128 and @xmath37 to the total elastic energy density are missing , and one allows @xmath2 and @xmath14 to be independent quantities . setting @xmath129 in eq . ( [ engineeringstress ] ) leads to zero stress for non - zero @xmath9 , i.e. , this truncated model predicts soft elasticity . this soft elasticity , however , is not compatible with sm@xmath0 elastomers crosslinked in the sm@xmath0 phase , such as the experimental samples of refs . @xcite , where the anisotropy direction is permanently frozen into the system . in this appendix , we give a brief derivation of eq . ( [ eulerangle ] ) . we are interested primarily in a rough estimate for @xmath90 , and therefore , for simplicity , we assume that we can ignore the effects of smectic layering . in the following , we employ , for convenience , the lagrangian formulation of elasticity theory . in the lagrangian formulation , the elastic energy density of a thin elastomeric film with thickness @xmath92 and height @xmath130 that is compressed along the @xmath54-direction can be written as @xmath131 where the choice of coordinates is the same as depicted in fig . [ fig : tiltapparatus ] . @xmath132 is the @xmath89-component of the elastic displacement @xmath133 , and @xmath134 is the @xmath83-component of the cauchy - saint - venant strain tensor , @xmath135 . @xmath136 and @xmath137 are , respectively , the bending modulus and young s modulus of the film , which are given in the incompressible limit by @xmath138 and @xmath139 @xcite . at leading order , @xmath140 is negative when the sample is effectively compressed as in the tilt apparatus . this leads to @xmath141 when we concentrate on the parts of @xmath142 that are most important with respect to buckling in the @xmath89-direction . next , we substitute the strain ( [ uzz ] ) into eq . ( [ flagrangian ] ) and switch to fourier space . to leading order in the elastic displacement , this produces @xmath143 where @xmath144 is the wavevector conjugate to @xmath17 , @xmath145 is the fourier transform of @xmath132 , and so on . equation ( [ ffourier ] ) makes it transparent that buckling occurs for @xmath146 the smallest value of @xmath147 is determined by the specifics of the boundary conditions . from fig . [ fig : tiltapparatus ] it appears as if the sample of kf is clamped such that it prefers to stay parallel to the clamps in their immediate vicinity . in this case , the smallest value of @xmath147 is @xmath148 . for hinged boundary conditions , in comparison , its smallest value would be @xmath149 . using the former value and exploiting that , approximately , @xmath150 , we obtain the estimate ( [ eulerangle ] ) for the onset of the euler instability in the tilt apparatus . for a review on liquid crystal elastomers see m. warner and e.m . terentjev , _ liquid crystal elastomers _ ( clarendon press , oxford , 2003 ) . for a review see p. g. de gennes and j. prost , _ the physics of liquid crystals _ ( clarendon press , oxford , 1993 ) ; s. chandrasekhar , _ liquid crystals _ ( cambridge university press , cambridge , 1992 ) . e. nishikawa and h. finkelmann , macromol . * 200 * , 312 ( 1999 ) . j. m. adams and m. warner , phys . e * 71 * , 021708 ( 2005 ) . o. stenull and t. c. lubensky , phys . e * 76 * , 011706 ( 2007 ) . d. kramer and h. finkelmann , arxiv:0708.2024 . d. kramer and h. finkelmann , phys . e * 78 * , 021704 ( 2008 ) . p. bladon , e. terentjev , and m. warner , j. phys . ii ( france ) * 4 * , 75 ( 1994 ) . l. r. g. treloar , _ the physics of rubber elasticity _ ( clarendon press , oxford , 1975 ) . note that the layer compression term formulated by aw is of the form @xmath151 ^ 2 $ ] whereas setting @xmath152 in eq . ( [ layercompressionen ] ) produces @xmath153 ^ 2 $ ] . this form is chosen for consistency with the chen - lubensky model of smectics . this term can be factorized into @xmath154 ^ 2[(d / d_0 ) + 1]^2\approx 4 [ d / d_0 -1]^2 $ ] for @xmath155 . thus the difference is merely factor of @xmath156 in the definition of the smectic modulus @xmath31 in the two representations . m. brehmer , r. zentel , f. gieselmann , r. germer , and p. zungenmaier , liquid crystals * 21 * , 589 ( 1996 ) . p. archer and i. dierking , phys . e * 72 * , 041713 ( 2005 ) . e. m. terentjev , m. warner , r. b. meyer , and j. yamamoto , phys . e * 60 * , 1872 ( 1999 ) . h. finkelmann , i. kundler , e. m. terentjev , and m. warner , j. phys . ii ( france ) * 7 * , 1059 ( 1997 ) . we checked the validity of this assumption by allowing @xmath63 to relax to its equilibrium value , i.e. , we repeated our analysis using the @xmath63 that minimizes @xmath59 rather than @xmath157 . the results obtained by using this more involved approach do not differ significantly from those obtained our simplified approach with @xmath157 . d. kramer and h. finkelmann , private communication . landau and e.m . lifshitz , _ theory of elasticity _ , 3rd edition ( pergamon press , new york , 1986 ) .	stenull and lubensky [ phys . rev . e * 76 * , 011706 ( 2007 ) ] have argued that shear strain and tilt of the director relative to the layer normal are coupled in smectic elastomers and that the imposition of one necessarily leads to the development of the other . this means , in particular , that a smectic-@xmath0 elastomer subjected to a simple shear will develop smectic-@xmath1-like tilt of the director . recently , kramer and finkelmann [ arxiv:0708.2024 , phys . rev . e * 78 * , 021704 ( 2008 ) ] performed shear experiments on smectic-@xmath0 elastomers using two different shear geometries . one of the experiments , which implements simple shear , produces clear evidence for the development of smectic-@xmath1-like tilt . here , we generalize a model for smectic elastomers introduced by adams and warner [ phys . rev . e * 71 * , 021708 ( 2005 ) ] and use it to study the magnitude of sm@xmath1-like tilt under shear for the two geometries investigated by kramer and finkelmann . using reasonable estimates of model parameters , we estimate the tilt angle for both geometries , and we compare our estimates to the experimental results . the other shear geometry is problematic since it introduces additional in - plane compressions in a sheet - like sample , thus inducing instabilities that we discuss .
the driving force behind the theoretical studies in gravitational radiation is to predict and investigate the sources of gravitational waves that will be observed by interferometric ( geo600 , ligo , virgo , tama , lisa ) and bar ( allegro , auriga , explorer , nautilus ) detectors . the goal of these studies is not limited to facilitating the detection of signals but also to provide tools to extract as much information as possible about the physical nature of the sources . neutron stars have been pointed out as obvious source candidates , since one channel by which an oscillating neutron star looses energy is via the emission of gravitational radiation . this radiation basically consists of a superposition of its characteristic oscillatory modes , which can be grouped into two families : i ) the fluid modes @xcite , which have a newtonian counterpart , but which through their coupling to the spacetime are damped because they now can emit gravitational waves . ii ) the spacetime modes @xcite , which have no newtonian counterpart , and which couple only weakly to the fluid . ( in the odd parity case , they do not couple to the fluid at all ) . these modes usually are strongly damped , however , for ultra - relativistic stars , the spacetime curvature can be so strong that it can trap impinging gravitational waves . those `` trapped '' modes @xcite are quite long lived , since they only leak out slowly from inside the gravitational potential well created by the neutron star . for recent comprehensive reviews on oscillation modes of neutron stars and black holes , see @xcite . however , just knowing the different oscillation modes of a neutron star is not enough . to be able to detect and to interpret a signal in a gravitational wave detector , it is crucial to have accurate templates of the waveforms that result from astrophysical events , such as the formation of the neutron star after the collapse of a progenitor star at the end point of nuclear burning . to that purpose , two ingredients are necessary . firstly , one has to have initial data that represent a realistic stage of the astrophysical system under consideration . secondly , one needs a numerical evolution code that can evolve these initial data in a stable and accurate way . it is only by fully nonlinear simulations that accurate waveforms of oscillating neutron stars after core collapse or binary merger can be obtained . therefore it is a major goal of several groups to push the further development of nonlinear evolution codes . the biggest obstacle , however , that still prevents one from obtaining accurate results , is the enormous computational expenditure to solve the full set of einstein equations in 3d . even modest resolutions easily exceed the capabilities of the largest and fastest present day computers . it is therefore crucial to have some alternative methods which require less computing power but still give valuable physical insight . one of them is perturbation theory . in the non - rotating case , due to the spherical background , one can completely separate off the angular dependence , and the evolution problem boils down to the integration of 1d wave equations . slowly rotating stars can still be treated in 1d as long as the deformation due to rotation is negligible . but for rapidly rotating neutron stars , the spherical symmetry is broken , and one can only separate off the azimuthal angular dependence . hence the problem becomes 2d , which is nevertheless still numerically tractable . the results from those linear evolutions are expected to give accurate waveforms in the range where the neutron star oscillates only weakly . furthermore , they can be used as testbeds for the nonlinear calculations . indeed , recent results of nonlinear evolutions have confirmed the existence of the various modes predicted by linear perturbation theory @xcite . the evolution of polar perturbations of non - rotating stars has been first addressed by allen et al . @xcite . in their formulation they describe the oscillations with three coupled wave equations , two for the metric perturbations in and outside the star , and one for the fluid inside the star . the evolution of various initial data have shown that generic initial data can indeed excite some of the stellar modes such as the @xmath1-mode , some @xmath2-modes and the first @xmath3-mode . in this paper we rederive the perturbation equations using the ( 3 + 1 ) or adm - form @xcite of the einstein equations . these equations are much better suited for the numerical initial value problem , since they immediately yield a hyperbolic evolution system provided that a suitable gauge has been chosen , whereas the original field equations contain mixed time and space derivatives . it is only by the introduction of new variables that those equations can be cast into a well suited form . the adm - formalism has already been addressed by moncrief @xcite , where it was used to derive a hamiltonian gauge invariant formulation of the perturbation equations . albeit preferable from a conceptual point of view , since they are independent of any gauge , those equations do not prove particularly useful for numerical evolutions . our starting point therefore are the field equations written as a set of evolution equations for the metric and the extrinsic curvature of a 3-dimensional spatial hypersurface , together with the constraint equations , which have to be satisfied at every instance of time . as an immediate consequence of using the adm - formalism , it follows that it is always possible to eliminate the fluid variables and thus to write down the evolution equations in terms of metric quantities only . this has already been shown by chandrasekhar and ferrari @xcite for the diagonal gauge and by ipser and price @xcite for the regge - wheeler gauge . however , from our formalism it is clear that this feature is independent of the chosen gauge . after having expanded the equations in spherical harmonics , we will choose the regge - wheeler gauge as the basis of further investigations . due to the spherical harmonics the equations are independent of the azimuthal order @xmath4 and can be divided into two uncoupled sets according to their behavior under parity transformation . the polar or even parity equations transform as @xmath5 , whereas the axial or odd parity equations as @xmath6 . our main focus will be the polar perturbations , nevertheless , due to their simplicity , we also present the axial equations in the adm - form . the derivation of the polar equations is much more involved , because the `` raw '' forms lead to numerical instabilities at the origin due to indefinite expressions , which is a well known consequence of using spherical coordinates . we are therefore forced to recast the polar equations into a form that is well behaved at the origin . having forged the equations into a form that yields stable evolutions for polytropic equations of state , we face a new instability close to the stellar surface when we try to switch to realistic equations of state . in this paper we use a quite recent equation of state , called mpa @xcite . yet , this instability is not a consequence of just this particular equation of state , it will occur for any realistic equation of state . as we shall show , the instability results from the behavior of the sound speed at the neutron drip point , where the equation of state suddenly becomes very soft . this is accompanied by a sharp drop of the sound speed , which can give rise to a numerical instability if the spatial grid resolution is chosen too low . however , by increasing the resolution this instability will eventually vanish . in the non - rotating case , having to resort to higher resolutions does not present too great an obstacle , since the equations are 1d wave equations . however , this instability presumably also occurs for rotating stars , where the equations are 2d , and the required resolution in order to obtain stable evolutions might stretch the computing times to unacceptably large values . therefore we devise a coordinate transformation of the radial coordinate inside the star , which enables stable evolutions for any spatial resolution . the remainder of this paper is organized as follows : in sec . ii we derive the relevant equations . section iii is devoted to a discussion of the boundary conditions . in sec . iv we briefly comment on the construction of initial data , and in sec . v we present results for polytropic equations of state . finally , in sec . vi we discuss the instability associated with the use of realistic equations of state , and we present the coordinate transformation that removes it . conclusions are briefly presented in sec . we adopt the metric signature ( @xmath7+++ ) and work in geometric units with @xmath8 . spacetime and spatial indices are denoted by greek and latin letters , respectively . derivatives with respect to the radial coordinate @xmath0 are sometimes denoted by a prime and derivatives with respect to the time coordinate @xmath9 by an overdot . the background geometry of a non - rotating , spherically symmetric star is given by the line element @xmath10 where @xmath11 and @xmath12 are functions of the radial coordinate @xmath0 . we model the star as a perfect fluid whose energy - momentum tensor has the form @xmath13 with @xmath2 denoting the pressure , @xmath14 the energy density , and @xmath15 the 4-velocity of the fluid . in the fluid rest frame @xmath15 has only one non - vanishing component @xmath16 . einstein s equations @xmath17 and the conservation equations @xmath18 yield the following three independent structure equations for the four unknown @xmath19 and @xmath14 : [ toveqns ] @xmath20 we will refer to those equations as tolman - oppenheimer - volkoff ( tov ) equations , although they are usually written in terms of the gravitational mass function @xmath4 , which is related to @xmath21 through @xmath22 to fully determine this system of equations , an equation of state must be supplemented . in this paper , we restrict ourselves to barotropic equations of state , where the pressure is a function of the energy density alone : @xmath23 in particular , we will use either a polytropic equation of state in form of @xmath24 or a tabulated realistic zero temperature equation of state . to obtain the stellar model , we have to integrate the tov equations together with the equation of state from the center up to the point where the pressure @xmath2 vanishes . this then defines the surface @xmath25 of the star . from birkhoff theorem it follows that the exterior vacuum region of the star is described by the schwarzschild metric with the mass parameter @xmath26 . let us recall that the general line element , when written in the adm - form , reads @xmath27 where @xmath28 , @xmath29 , and @xmath30 are lapse function , shift vector , and spatial metric , respectively . comparing the adm - metric with the background metric reveals that the background shift @xmath31 vanishes and the background lapse function is @xmath32 . in addition , the extrinsic curvature of constant @xmath9 slices vanishes , since the background metric is static . let @xmath28 and @xmath33 now be the perturbations of the lapse function and the ( covariant ) shift vector , and let @xmath34 denote the perturbations of the spatial metric . the line element describing the perturbations to the background metric then reads @xmath35 the perturbations of the energy - momentum tensor can be written in terms of @xmath36 , @xmath37 , and @xmath38 , which are the ( eulerian ) perturbations of the energy density @xmath14 , pressure @xmath2 , and ( covariant ) 4-velocity @xmath39 , respectively . we assume the equation of state to hold for both the unperturbed and perturbed configurations ( isentropic perturbations ) , therefore the perturbations @xmath36 and @xmath37 are related through @xmath40 where @xmath41 is the sound speed inside the fluid . the dynamic equations that govern the metric and extrinsic curvature perturbations @xmath34 and @xmath42 read after linearization @xmath43\\ & \quad { } + e^\nu\left[\,\delta r_{ij } + 4\pi { \left(}{\left(}p - \eps{\right)}h_{ij } + \delta\eps{\left(}c^2_s - 1{\right)}\gamma_{ij}{\right)}\right]\;. \end{split}\end{aligned}\ ] ] therein , the @xmath44 represent the perturbations of the spatial christoffel symbols @xmath45 @xmath46 and @xmath47 the perturbations of the spatial ricci tensor @xmath48 @xmath49 we should note that the indices of perturbations are lowered or raised with the background 3-metric @xmath30 . in constructing initial data , the perturbations must satisfy the linearized version of einstein s constraint equations , namely @xmath50 from the linearized energy - momentum conservation equations @xmath51 , we can deduce equations of motion for the fluid perturbations @xmath52 and @xmath53 , which complete the set of dynamical equations . however , we can as well dispense with the fluid equations , since it is possible to eliminate the energy density perturbation @xmath36 from the evolution equation by virtue of the hamiltonian constraint . in this way we can obtain a consistent system of evolution equations for the metric and extrinsic curvature perturbations alone . the constraints then can serve as a means to compute the matter perturbations @xmath36 and @xmath53 . we should stress that the possibility to completely describe the oscillations of neutron stars with spacetime variables only is a general feature of the perturbation equations and does not depend on any specific gauge choice . due to the spherical symmetry of the background , we can eliminate the angular dependence of the perturbation equations by expanding them into spherical tensor harmonics . those tensor harmonics can be divided into two subsets that behave differently under parity transformation . under space reflection the _ even parity _ or _ polar _ harmonics change sign according to @xmath5 , whereas the _ odd parity _ or _ axial _ harmonics transform like @xmath6 . here , @xmath54 is the number that labels the multipole of the spherical harmonics @xmath55 . because of the spherical symmetry , the axial and polar equations are decoupled and degenerate with respect to the azimuthal number @xmath4 . without picking a specific gauge , the metric perturbations can be expanded using the tensor harmonics given first by regge and wheeler @xcite in the following way ( symmetric components are denoted by an asterisk ) : @xmath56 where @xmath57 and @xmath58 with @xmath59 the notation has been chosen such that the coefficients @xmath60 represent the _ scalar _ parts of @xmath61 , namely @xmath62 , and @xmath63 , whereas the @xmath64 stand for the _ vector _ components @xmath65 and @xmath66 . lastly , the @xmath67 represent the _ tensor _ components @xmath68 and @xmath69 . note that this expansion includes both polar and axial harmonics . the latter are represented by @xmath70 and @xmath71 with respective coefficients @xmath72 and @xmath73 . similarly , the extrinsic curvature tensor @xmath42 will be expanded as @xmath74 here , @xmath75 and @xmath76 are the axial coefficients . last not least , we need the matter variables @xmath77 the sum over the multipoles starts from @xmath78 , since we are only interested in perturbations which are associated with the emission of gravitational radiation . besides , the non - radiative multipoles @xmath79 and @xmath80 would have to be treated in a different way . using the above expansions , we obtain 12 evolution equations for the coefficients of the 3-metric @xmath34 and the extrinsic curvature @xmath42 . however , it is clear that this set can not be used for the evolution , for we have not specified any gauge yet . within perturbation theory , picking a specific gauge usually means to define a gauge vector @xmath81 and transform the metric perturbations according to @xmath82 in our case , however , we follow the spirit of the adm - formalism , that is we pick the gauge by prescribing the coefficients of lapse @xmath28 and shift @xmath33 , for which we do not have any evolution equations . yet , this is not enough as we shall see . to fully fix the gauge , we also have to impose certain constraints on the initial data . the most common gauge is the regge - wheeler gauge @xcite , which can be obtained by setting a certain number of metric perturbation coefficients to zero . translating this into the adm - formalism means to choose shift and lapse in such a way that for initial data which obey the regge - wheeler gauge , the evolution will preserve it . regge and wheeler have chosen their gauge such that the metric coefficients @xmath83 , and @xmath84 vanish . by choosing shift and lapse as ( from now on we omit the indices @xmath54 and @xmath4 ) @xmath85 we obtain the following evolution equations for the coefficients @xmath86 and @xmath73 : @xmath87 in addition the evolution equation for the extrinsic curvature component @xmath88 depends only on @xmath89 and @xmath90 . thus , only for initial data satisfying @xmath91 _ and _ @xmath92 , the evolution equations guarantee the vanishing of those coefficients throughout the whole evolution . hence , the regge - wheeler gauge does not only impose constraints on the initial metric perturbations but also on the extrinsic curvature perturbations . the vanishing of the coefficient @xmath88 is a crucial feature of the regge - wheeler gauge and has to be imposed on the initial data . this is in contradiction with the statement of andersson et al . @xcite , where the authors claim that there are no constraints on the odd or even parity nature of the extrinsic curvature . this is not true , since the vanishing of @xmath88 _ must _ be ensured on the initial slice , otherwise the evolution _ can not _ be performed in the regge - wheeler gauge . for data with non - vanishing @xmath88 , we would have @xmath93 , which is incompatible with the regge - wheeler gauge , because it would lead to a non - vanishing @xmath89 even if @xmath89 was initially set to zero . therefore the initial data presented in @xcite for two colliding neutron stars are not adequate for an evolution code which makes use of the regge - wheeler gauge , since the coefficient @xmath94 in eqs . ( 82 ) of @xcite ( which corresponds to our @xmath88 ) does not vanish . we will discuss the issue of constructing initial data in more detail in section [ sec : init ] . having switched to the regge - wheeler gauge , we are left with evolution equations for the three metric variables @xmath95 , @xmath96 , and @xmath97 , and all the extrinsic curvature variables @xmath98 save @xmath88 . ( in the more common notation of regge and wheeler @xcite , it is @xmath99 , @xmath100 and @xmath101 ) . we now split the equations with respect to their behavior under parity transformation . let us first focus our attention to the axial case . if in the expansions and we use new variables @xmath102 we obtain the following set of evolution equations : [ axial ] @xmath103 and one constraint equation : @xmath104 from the conservation law @xmath105 , it follows that @xmath106 which means that axial gravitational waves do not couple to the stellar fluid . of course , this does not mean that there can not be any axial modes at all , in fact , there exists a whole spectrum of axial spacetime modes , which is quite similar the polar spectrum @xcite . it is only the fluid modes that are missing . the three evolution equations can be combined to yield a single wave equation for the metric perturbation @xmath107 , or better for the variable @xmath108 which completely describes the dynamical evolution of axial perturbations : @xmath109 herein , @xmath110 denotes the tortoise coordinate defined by @xmath111 in the exterior , where @xmath2 and @xmath14 vanish and @xmath112 , eq . reduces to the well known regge - wheeler equation @xcite . for the polar case we have only two dynamical metric perturbations @xmath113 and @xmath114 , since the lapse @xmath115 is proportional to @xmath113 , and the only non - vanishing component @xmath116 of the shift is proportional to @xmath117 . to obtain the relevant equations , we again choose a somewhat different expansion . for the metric we use @xmath118 and for the extrinsic curvature @xmath119 the expansion of the matter variables reads @xmath120 of course , these expansions are to be understood as sums over all @xmath54 and @xmath4 , analogous to the expansions and . it is only through this particular choice of expansion coefficients ( e.g. in @xmath28 , @xmath63 and @xmath121 ) that we obtain a system of equations that can be numerically integrated in a quite straightforward and this is the main point in a stable way . another consequence of this decomposition is that it gives some physical meaning to the metric perturbations @xmath122 and @xmath123 . from eq . , we can see that @xmath123 represents the conformal part of the spatial perturbation @xmath34 , for if we set @xmath124 , the metric is conformally flat . obviously , @xmath122 itself then represents the deviation from conformal flatness . finally , we should note that this decomposition is quite similar to the one used by allen et al . their variables @xmath125 and @xmath126 are related to ours as follows : @xmath127 and @xmath128 . in order to avoid numerical problems at the origin , we have to replace @xmath129 by the following quantity @xmath130 in this way , we obtain a system of five coupled evolution equations , which are of first order in time and second order in space . there are two equations for the metric variables @xmath122 and @xmath123 and three more for the extrinsic curvature variables @xmath131 , @xmath132 and @xmath133 : [ fosys ] @xmath134 \end{split } \end{aligned}\ ] ] @xmath135\\ & \qquad\qquad + 8\pi e^{2\nu}{\left(}1 - c^2_s{\right)}\rho \end{split}\\ \label{k2 } { \frac{\partial { k_2}}{\partial { t } } } & = e^{2\nu-2\lam } { \left(}r { \frac{\partial { s}}{\partial { r } } } + { \left(}2r\nu ' + 1{\right)}s + 2\frac{\nu'}{r}t{\right)}\ , . \end{aligned}\ ] ] we could easily convert this system into a first order system both in time and space by adding another two evolution equations for the first derivatives @xmath136 and @xmath137 . however , the form of the first four equations , which are independent of @xmath132 , suggests to rather convert them into two coupled wave equations for @xmath122 and @xmath123 @xmath138 \end{split}\\ \begin{split } \label{wavet } { \frac{\partial^2 { t}}{\partial { t}^2 } } & = e^{2\nu-2\lam}\bigg [ { \frac{\partial^2 { t}}{\partial { r}^2 } } + { \left(}\nu ' - \lam'{\right ) } { \frac{\partial { t}}{\partial { r } } } + { \left(}\frac{\nu'}{r } + 3\frac{\lam'}{r } + 2\frac{e^{2\lam } - 1}{r^2 } - e^{2\lam}\frac{l(l+1)}{r^2}{\right)}t + 2{\left(}r\nu ' + r\lam ' - 1{\right)}s\,\bigg]\\ & \qquad\qquad+ 8\pi e^{2\nu}{\left(}1 - c^2_s{\right)}\rho\ ; , \end{split}\end{aligned}\ ] ] which are equivalent to eqs . ( 14 ) and ( 15 ) of allen et al . @xcite . as can be seen , the wave equation for @xmath122 is totally decoupled from the fluid variable @xmath139 , which only couples to the metric perturbation @xmath123 in eq . . the equation for @xmath132 is only necessary in the interior region , where it couples to the hydrodynamical equations , which follow from energy - momentum conservation @xmath51 , and are given by [ rho_system ] @xmath140\\ & \qquad { } + { \left(}p + \eps{\right)}{\left(}r { \frac{\partial { k_2}}{\partial { r } } } + { \left(}2 + r\lam'{\right)}k_2 - \frac{r^2}{2}k - \frac{3}{2}k_5{\right)}+ r\eps'k_2 \end{split}\\ \begin{split } \label{u1 } { \frac{\partial { \tilde u_1}}{\partial { t } } } & = c_s^2 { \frac{\partial { \rho}}{\partial { r } } } + { \left(}\nu'{\left(}1 + c_s^2{\right)}+ { \left(}c_s^2{\right)}'{\right)}\rho - \half{\left(}p + \eps{\right)}{\left(}r^2 { \frac{\partial { s}}{\partial { r } } } + { \frac{\partial { t}}{\partial { r } } } + 2rs{\right)}\end{split}\\ \label{u2 } { \frac{\partial { \tilde u_2}}{\partial { t } } } & = c_s^2\rho - \half{\left(}p + \eps{\right)}{\left(}r^2 s + t{\right)}\;. \end{aligned}\ ] ] here , we have defined @xmath141 . by introducing the enthalpy perturbation @xmath142 the fluid equations assume a more convenient form : [ h_system ] @xmath143 - \nu'{\left(}u_1 - \frac{1}{r}u_2{\right)}\\ & \qquad + c_s^2{\left(}r { \frac{\partial { k_2}}{\partial { r } } } + { \left(}2 + r\lam'{\right)}k_2 - \frac{r^2}{2}k - \frac{3}{2}k_5{\right)}- r\nu'k_2 \end{split}\\ \label{uu1 } { \frac{\partial { u_1}}{\partial { t } } } & = { \frac{\partial { h}}{\partial { r } } } - \half{\left(}r^2 { \frac{\partial { s}}{\partial { r } } } + { \frac{\partial { t}}{\partial { r } } } + 2rs{\right)}\\ \label{uu2 } { \frac{\partial { u_2}}{\partial { t } } } & = h - \half{\left(}r^2s + t{\right)}\;. \end{aligned}\ ] ] interestingly , from eqs . and , it follows that the coefficients @xmath144 and @xmath145 are not independent of each other but rather are related via @xmath146 where @xmath125 is a time independent function , which has to be fixed by the initial data . the above system , too , can be cast into a second order wave equation for @xmath147 , which is equivalent to eq . ( 16 ) of allen et al . @xcite ( the different signs in the terms containing @xmath122 and @xmath123 are correct ) : @xmath148\;. \end{split}\end{aligned}\ ] ] as already mentioned in subsection b.1 , we do not necessarily need the fluid equations , for we can eliminate them by means of the hamiltonian constraint , which relates the fluid variable @xmath139 to the two metric variables @xmath122 and @xmath123 : @xmath149 with @xmath139 substituted by the hamiltonian constraint , the wave equation for @xmath123 in the interior now reads @xmath150\\ & \quad { } + e^{2\nu-2\lam}\bigg[\;\nu ' { \frac{\partial { t}}{\partial { r } } } + r { \frac{\partial { s}}{\partial { r } } } + { \left(}2r\nu ' + \half e^{2\lam}l(l+1){\right)}s + { \left(}\frac{\nu'}{r } + \frac{e^{2\lam } - 1}{r^2}{\right)}t\;\bigg]\;. \end{split}\end{aligned}\ ] ] in the exterior , both @xmath122 and @xmath123 propagate with the local speed of light @xmath151 , in the interior , however , @xmath123 changes its character and propagates with the local speed of sound @xmath152 . the last set of equations that is still missing are the momentum constraints , which link the velocity perturbations to the extrinsic curvature variables : [ mc ] @xmath153 those constraint equations have to be solved in order to obtain physically valid initial data . we postpone the discussion on how to construct initial to section [ sec : init ] . with @xmath139 being eliminated , @xmath122 and @xmath123 in the interior now become independent variables , and the hamiltonian constraint serves as a definition for @xmath139 . in the exterior , however , @xmath122 and @xmath123 are not independent but have to satisfy the hamiltonian constraint with @xmath139 set to zero . unfortunately , we can not use the hamiltonian constraint to further eliminate one of those variables , but it is possible to combine @xmath122 and @xmath123 to form a new variable @xmath154 ( we use the definition ( 20 ) of allen et al . @xcite ) @xmath155\;,\end{aligned}\ ] ] with @xmath156 @xmath154 then satisfies a single wave equation , the famous zerilli equation that was first derived in 1970 by f. zerilli @xcite in the context of black hole oscillations : @xmath157 here , we use @xmath158 , and @xmath110 is again the tortoise coordinate defined in eq . . it is also possible to invert eq . and express @xmath122 and @xmath123 in terms of @xmath154 through [ s_t ] @xmath159 lastly , we should note that the radiated energy at infinity can be computed from @xcite @xmath160 since the zerilli function contains the full polar gravitational wave information , it should be possible to recover the original metric and extrinsic curvature variables from it . for @xmath122 and @xmath123 we have already given the relevant formulas in eqs . . since @xmath161 and @xmath162 , they also can computed from with @xmath154 replaced by @xmath163 . in the exterior the momentum constraints can be combined to give an algebraic relation for @xmath132 . finally , the last remaining quantity @xmath129 follows from eq . . in the derivations of all the above equations we have made extensive use of the computer algebra program maple v in order to avoid possible mistakes . furthermore , we have checked that our equations and our numerical results agree with those of allen et al . there are three boundaries we have to take care of : the origin , the surface , and the outer boundary of the numerical grid , which lies somewhere outside the star . at the origin @xmath164 we have to demand all variables to be regular . from taylor expansion around @xmath164 , we can infer the analytic behavior of the various variables . close to the origin , @xmath122 and @xmath123 , for instance , are both proportional to @xmath165 . at the outer boundary , we impose the sommerfeld boundary condition , i.e. we require the waves to be purely outgoing . if computational time does not matter , we can even put this boundary so far away that any contamination that enters the grid from there has not enough time to travel to the region where we extract the signal . the third boundary is the surface of the star at @xmath166 , which is formally defined by the vanishing of the total pressure @xmath167 . since the perturbations will slightly deform the star , the perturbed surface will be displaced by an amount @xmath168 with respect to the unperturbed location at @xmath169 . if the coordinates of the unperturbed surface are denoted by @xmath170 , the vanishing of the total pressure @xmath167 at the displaced surface translates to @xmath171 . from taylor expansion to first order we find that the ( eulerian ) pressure perturbation @xmath37 at the surface has to obey @xmath172 unfortunately , this is not a very convenient boundary condition , since we neither use @xmath37 nor the displacement vector @xmath168 in our set of evolution equations . therefore we must relate this condition to the variables we use . we will try to find a condition that gives us the time evolution of @xmath36 at the stellar surface . the first step is to use the relation @xmath173 , which gives us @xmath174 the time derivative of @xmath175 can then be related to the @xmath0-component of the 4-velocity @xmath176 @xcite : @xmath177 after expansion in spherical harmonics , we finally obtain @xmath178_{r = r}\;.\end{aligned}\ ] ] the equivalent equation for the quantity @xmath147 as defined in reads @xmath179_{r = r}\;.\end{aligned}\ ] ] incidentally , this expression can be obtained directly from the evolution equation just by setting @xmath41 to zero . in the same manner , we can obtain the boundary condition for the wave equation . also , eq . follows immediately from eq . , if one sets @xmath180 . for polytropic equations of state , it is always @xmath181 at the surface of the star , hence the evolution equations and automatically lead to the right boundary conditions . for realistic equations of state , the sound speed at the surface should be that of iron , which is very small compared to the sound speed inside the core , where it might reach almost the speed of light for very relativistic stellar models . for practical purposes , in these cases we just might as well set @xmath182 . of course , this analysis does not hold for constant density models , but those are not considered in this paper , anyway . let us now turn to the junction conditions at the surface of the star . we will always assume that @xmath14 and @xmath41 go to zero when approaching the stellar surface @xmath166 . following the line of reasoning of @xcite , we find that @xmath122 is at least @xmath183 , whereas the differentiability of @xmath123 depends on the value of @xmath139 at the surface . for , if we let the subscripts @xmath184 and @xmath185 represent the values for the interior and the exterior , respectively , we deduce from eq . that for the second derivative of @xmath123 across the surface , the following relation has to hold : @xmath186 from condition , it follows that the values of @xmath139 at the surface depends on the value of @xmath187 at this point . as already discussed above , for polytropic equations of state , we have @xmath188 at the surface . however , @xmath187 does not necessarily vanish , instead we have the following relation @xmath189 which shows that the behavior of @xmath187 critically depends on the value of the polytropic index @xmath190 . we can distinguish three different cases . for @xmath191 , we have @xmath192 , for @xmath193 , we have @xmath194 const . , whereas for @xmath195 , we have @xmath196 ! this is somewhat disturbing , since for the boundary condition this would mean that @xmath197 , unless the expression in brackets vanishes . however , this is not automatically guaranteed ! interestingly , the boundary condition for @xmath147 is harmless for all values of @xmath190 , since @xmath198 is always bounded . but , of course , in eq . , we have to use @xmath139 , which we have to compute from @xmath147 using @xmath199 again , we obtain an infinite value when @xmath200 , unless @xmath147 vanishes at the surface . however , as with eq . , the boundary condition does not guarantee the vanishing of @xmath147 , even if @xmath147 is initially set to zero . the reason for this peculiar behavior of @xmath139 is that in the @xmath201 case , the eulerian description breaks down at the surface , which is a direct consequence of @xmath187 becoming infinite . in this case , a lagrangian description of the fluid perturbations would be much more appropriate , since for any polytropic equation of state , the lagrangian energy density perturbation always vanishes at the surface , because of the vanishing of the lagrangian pressure perturbation . what then happens to the eulerian density perturbation ? by definition , the eulerian density perturbation @xmath52 is the difference between the total energy density and the background density @xmath14 at the same location @xmath0 , whereas the lagrangian density perturbation @xmath202 measures the density change in a fluid element that is displaced by some amount @xmath168 . it is through taylor expansion to linear order that we obtain the connection between the lagrangian and eulerian perturbations : @xmath203 but of course , for the @xmath200 case this expression is ill - defined at the surface , since the second term diverges . it is then clear that the eulerian perturbation @xmath36 has to become infinite , too , in order to compensate for the blow up of @xmath187 and to yield a vanishing lagrangian perturbation @xmath204 . the whole discussion seems somewhat irrelevant , since we have seen that the equations can be written in terms of metric quantities only . however , we have to compute the second derivative of @xmath123 , which , as can be seen from eq . , depends on @xmath139 . in the @xmath201 case , we therefore must have a blow up of @xmath205 at the surface . of course , this is very troublesome for the numerical discretization , and even for @xmath193 , we still have a discontinuity in @xmath205 , which can spoil the second order convergence of the numerical discretization scheme . the numerical evolutions indeed confirm the above analysis . by computing @xmath139 with the aid of the hamiltonian constraint , we find that for polytropic stellar models with @xmath200 , @xmath139 tends to blow up at the stellar surface . for @xmath191 , we have @xmath206 at the surface , whereas for @xmath193 , we obtain a finite value @xcite . to obtain physically valid initial data , we have to solve the constraint equations and . as is well known , there is no unique way to do so , for those equations are underdetermined , and therefore one always has to make some additional assumptions about the geometry of the initial slice . the simplest way is to choose time symmetric initial data , since in this case the momentum constraints can be satisfied trivially by setting all extrinsic curvature variables to zero , and we only have to solve the hamiltonian constraint . of course , time symmetric data are somewhat unphysical , since they represent a stage of a physical system which had an arbitrary amount of incoming radiation in the past . nevertheless , they can give valuable insight into the behavior of the system under consideration . in general , all conceivable initial data for neutron star oscillations fall into two basic categories , or combinations thereof . first , we can assume the star being initially unperturbed . in the exterior vacuum region , we then can arbitrarily prescribe some metric perturbations representing a gravitational wave , which travels towards the star and eventually excites it to oscillations . the most convenient way to do so is to prescribe the initial zerilli function @xmath154 and its time derivative @xmath163 , and to compute @xmath122 , @xmath123 , @xmath131 , and @xmath133 through eqs . . in this way , we have initial data that automatically satisfy the constraints without having to solve any differential equation . the other possibility is to prescribe some fluid perturbations inside the star and then use the constraints to solve for the associated metric perturbations . as is well known , the matter distribution does not uniquely fix the metric , since we can always superpose some additional gravitational waves . it is therefore not clear at all what would represent `` true '' physically realistic initial data . a quite common procedure is to construct conformally flat initial data @xcite . in our case , this amounts to setting @xmath207 in the hamiltonian constraint and solving for @xmath208 for a given initial fluid perturbation @xmath209 . interestingly , the thus obtained initial data do not excite any of the spacetime modes to a significant extent , whereas initial data with the same fluid perturbation @xmath209 , but with @xmath210 and @xmath211 , can show large @xmath3-mode excitations . in fig . 1 , we compare the wave signals of these two cases for time symmetric initial data . in the case where we set @xmath212 , there is no @xmath3-mode signal at all , and the waveform consists of pure fluid oscillations . in the other case , we can see a burst of gravitational radiation , which quickly damps away through the first @xmath3-mode . the final fluid ringing then coincides with the @xmath213 case . this shows that the presence of @xmath3-modes is not a generic feature of neutron star oscillations . they are only excited by a special subset of initial data . the open question now is whether or not they will show up in a real wave signal . andersson and kokkotas @xcite have shown that by extracting the frequencies and damping times of the @xmath1-mode and the first @xmath3-mode in a gravitational wave signal , one can obtain important information such as mass and radius of the neutron star . these data , in turn , can then be used to restrict the possible equations of state . of course , this method stands and falls with the presence of the @xmath3-modes in the wave signal . but as we have seen , it could well be that @xmath3-modes do not play a significant role at all . in a subsequent paper @xcite , we will investigate the excitations of neutron star oscillations by means of particle scattering , which will show that the @xmath3-modes can only be excited by extremely relativistic particles . let us now focus on initial data which are not time symmetric . in this case , we also have to solve the momentum constraints . here , too , we have some additional freedom in specifying the extrinsic curvature . for instance we can choose tr@xmath214 , which is equivalent to setting @xmath215 or , equivalently because of eq . , @xmath216 we then can substitute @xmath131 in eq . and obtain thus two coupled differential equations for the remaining unknowns @xmath132 and @xmath133 . for given fluid perturbations @xmath144 and @xmath145 , the momentum constraints yield a unique solution . before presenting the numerical results , we would like to comment on the applicability of the lichnerowicz - york approach @xcite for the construction of initial data for neutron star perturbations , which was used in @xcite . stated in a somewhat crude way , this method consists in decomposing the six degrees of freedom of the extrinsic curvature into its trace tr@xmath131 ( 1 dof ) , a transverse trace free part @xmath217 ( 2 dofs ) , and a trace free longitudinal part @xmath218(3 dofs ) . since the momentum constraints consist only of three equations with three source functions @xmath219 , which describe the matter distribution , only 3 dofs are fixed , whereas the remaining ones are freely specifiable . the simplest way is to set tr@xmath220 , i.e. the extrinsic curvature is purely longitudinal . it can thus be written in terms of a vector potential with three independent components , for which the momentum constraints provide a unique solution once the sources @xmath219 and appropriate boundary conditions have been specified . if the extrinsic curvature is first decomposed into spherical harmonics , this procedure still works . for the polar part , we then have four components and two constraint equations , i.e. only two components can be freely specified . here , too , by assuming tr@xmath221 , we can write the remaining longitudinal part in terms of a vector potential with two polar components . solving the constraint equations with non - vanishing sources @xmath219 will yield nonzero values for those component , and therefore all four polar components of the extrinsic curvature will in general have nonzero values , either . but this is contradiction with the regge - wheeler gauge , which requires the component @xmath222 to vanish . this could only be accomplished by setting one component of the vector potential to zero . however , this is not possible , since both components are already fixed by the constraint equations , and in general non - vanishing . hence , if we want our initial data to be conform to the regge - wheeler gauge , we have to have @xmath223 , and therefore we can not have both tr@xmath224 and @xmath225 . by choosing tr@xmath224 , we already have used up all degrees of freedom to fix the extrinsic curvature components , since we have to set @xmath223 , too . our actual goal was , by using the adm - formalism , to obtain the evolution equations in a hyperbolic first order form . however , due to numerical problems at the origin , we were forced to modify those equations in such a way that the resulting set is more or less equivalent to the set of wave equations for the variables @xmath122 , @xmath123 and @xmath147 . thus , from the computational point of view , there is no point in sticking to the first order system , since the wave equations can be numerically integrated in a much faster way . also , we will not explicitly integrate the fluid equations , since we can eliminate @xmath139 in eq . by means of the hamiltonian constraint . for the numerical evolution , we therefore use eqs . and in the interior , whereas in the exterior we use eqs . and with @xmath139 set to zero . of course , in the exterior we could try to switch from @xmath122 and @xmath123 to the zerilli function @xmath154 . we thus would have to integrate only one wave equation , and from the resulting waveforms , we could easily read off the emitted gravitational energy , since the radiation power is proportional to the square of @xmath163 . unfortunately , this causes numerical problems , because at the seam , we would have to compute @xmath154 and its derivatives from @xmath122 and @xmath123 and vice versa . however , in order to obtain a correct value of @xmath154 , it is crucial that @xmath122 and @xmath123 satisfy the hamiltonian constraint . on the other hand @xmath122 and @xmath123 satisfy the hamiltonian constraint by construction , at least up to discretization errors , when directly computed from @xmath154 through eqs . . at the seam somewhere outside the star , where we switch from @xmath122 and @xmath123 to @xmath154 , we will always have some violation of the hamiltonian constraint , thus the computed zerilli function @xmath154 will not be quite correct . if we then go back and compute @xmath122 and @xmath123 from the just obtained value of @xmath154 , they will differ significantly from their original values , since now they suddenly satisfy the hamiltonian constraint . this mismatching at the seam gives rise to additional reflections , which rapidly amplify inside the star and cause the numerical code to crash after a few dynamical time scales . there is another point we would like to mention . when evolved with the zerilli equation , the amplitude of an outgoing wave remains constant for large @xmath0 . the same is true for @xmath122 , however , the amplitude of @xmath123 grows linearly with @xmath0 . when computing @xmath154 from @xmath122 and @xmath123 by means of formula , this growth has to cancel , what is indeed the case when @xmath122 and @xmath123 exactly satisfy the hamiltonian constraint . if not , @xmath154 will also start to grow for large @xmath0 . but this is what happens , since numerically @xmath122 and @xmath123 do not satisfy the hamiltonian constraint . it is predominantly the high frequency components that get amplified the most strongly . unfortunately , this can lead to quite rough waveforms for @xmath154 , while those for @xmath122 and @xmath123 look perfectly smooth . by increasing the resolution , this amplification will decrease , but the resolution has to be quite high in order to obtain accurate results . the most practical way to obtain a quite reliable zerilli function , therefore , is not to completely switch to the zerilli function in the exterior region , but to additionally evolve @xmath154 together with @xmath122 and @xmath123 . that is , close to the surface of the star we construct @xmath154 from @xmath122 and @xmath123 by means of formula and then use the zerilli equation to independently evolve @xmath154 parallel to @xmath122 and @xmath123 . of course , this amounts to the additional computational expenditure of evolving an extra wave equation , but we get rewarded by obtaining much more accurate results . to discretize the wave equations , we use the explicit second order leap - frog scheme . if the equations are correctly implemented , the numerical violation of the constraints should converge to zero in second order . by monitoring the hamiltonian constraint in the exterior , this is indeed what we find . of course , as mentioned above , for polytropes with @xmath226 , the stellar surface can reduce the convergence down to first order . we should make one final note concerning the numerical treatment of the origin @xmath164 . it is well known that in radial coordinates , the equations usually show a singular behavior close to the origin . moreover , taylor expansion around the origin shows that the equations admit two kinds of solutions , a regular and a divergent one . on physical grounds , one usually rejects the divergent solution . for the evolution , it is crucial that the numerical scheme preserves the regularity condition and suppresses the divergent solution . we have chosen the dynamical variables such that the only singular terms are the ones proportional to @xmath227 . the regularity condition requires all the perturbation variables to vanish at the origin , however , even for @xmath78 the equations can not be numerically evolved with a time step size @xmath228 close to the maximal allowed time step size @xmath229 , which follows from the cfl - condition @xmath230 where @xmath231 is the grid spacing in the @xmath0-coordinate and @xmath232 the largest propagation speed , because the presence of the divergent @xmath227-terms causes the scheme to become unstable at the origin . only by decreasing @xmath228 can stability be reinforced . however , for large values of @xmath54 , @xmath228 has to be so small that it prevents one from obtaining numerical results within a reasonable time frame . we therefore propose a different way that allows stable evolutions with @xmath228 close to @xmath229 for all values of @xmath54 . instead of having the boundary conditions located exactly at @xmath164 , we move it @xmath233 grid points to the right , i.e. at @xmath234 , we impose @xmath235 . the actual value of @xmath233 depends on @xmath54 in the following simple way : @xmath236 . the numerical experiments and a detailed stability analysis @xcite have shown that with this little trick , we can indeed obtain stable and second order convergent evolutions for arbitrary values of @xmath54 with a courant number of about @xmath237 . results concerning the evolution of various initial data have already been presented in @xcite . since the authors only focused on a single polytropic stellar model with a central density of @xmath238g/@xmath239 , we would like to consider three more models , one being less relativistic and the others being more relativistic . the physical parameters of the models are given in table 1 . all results presented are for @xmath78 . results for more stellar models and for other values of @xmath54 can be found in @xcite . as initial data , we choose a narrow time symmetric gravitational wave pulse , centered at @xmath240 km , where we prescribe @xmath123 to have a gaussian shape , and use the hamiltonian constraint to compute @xmath122 . the numerical resolution is 500 grid points inside the star . the resulting waveforms for the different models are shown in fig . 2 . using a logarithmic scale , we plot the modulus of the zerilli function , extracted at @xmath241 km . the three waveforms clearly have quite different features . for the least relativistic model 1 ( upper panel ) , there is no @xmath3-mode signal at all , and the fall - off , which immediately follows the reflected wave pulse at about @xmath242ms , shows more a tail - like behavior . at @xmath243ms it merges into the fluid ringing , which is dominated by the @xmath1-mode . this is a somewhat unexpected result , since a direct mode calculation reveals that for this model there do exist @xmath3-modes , which should show up in the waveform . however , even the first @xmath3-mode has a quite large imaginary part ( @xmath244 = 0.305 + 0.24i ) , which results from the model being less compact . thus , the @xmath3-modes get buried in the tail - like fall - off . the more compact and therefore more relativistic the stellar model is , the less damped are the @xmath3-modes , and they should eventually dominate over the tail - like fall - off . this is the case for model 2 ( middle panel ) , which is quite close to the stability limit . here , instead of the tail - like fall - off , we now can see the ring - down of the first @xmath3-mode ( @xmath244 = 0.256 + 0.085i ) . still , it is much more strongly damped than the fluid modes , which therefore dominate the late time part . model 3 ( lower panel ) , which is unstable with respect to radial oscillations , is compact enough to allow the existence of trapped modes . because of their comparatively weak damping , we can not have a clear cut discrimination between the two ring - down phases , i.e. at first the @xmath3-modes and then the fluid modes . instead , the signal consists of a mixture of both the trapped and the fluid modes . nevertheless , the trapped modes dominate in early times , which can be seen from the fourier transform in the upper panel of fig . 3 . in the lower panel , we have taken the fourier transform at much later times , where the most strongly damped trapped modes have damped away and the fluid modes prevail . still , the first two trapped modes are present in the spectrum . however , there is still one feature present in the wave signal for model 3 , which was also present in previous evolutions of the axial modes for very compact stars , but which apparently has not been noticed before ( see e.g. fig . 4 of @xcite and fig . 5 of @xcite ) . if , in the wave signal , we determine the frequency and damping time of the very first ring - down ( from @xmath245ms to about @xmath246ms ) , we find that they do not match to any of the quasi - normal modes associated with this particular stellar model . instead , they are almost identical with the complex frequency of the first quasi - normal mode of an equal mass black hole ! how is that ? for black hole spacetimes , the quasinormal modes are derived from the regge - wheeler potential in the axial case and the zerilli potential in the polar case . both potentials exhibit a maximum at about @xmath247 . for ordinary neutron stars the surface lies usually at a radius larger than @xmath248 , therefore these peaks do not exist for those stars . however , for ultra - compact stars these peaks can lie outside . any axial or polar incoming gravitational wave packet has to cross the corresponding peak before penetrating the neutron star . while trying to cross this peak , a large fraction of the wave will be scattered off , which in the black hole spacetime gives rise to the quasinormal ring - down , but since the situation is now almost identical for the neutron star spacetime , this associated ring - down will have black - hole characteristics . since inside the star , the local speed of light @xmath249 is largely reduced , it takes a while for the remaining wave packet to be reflected , and this is gives rise to the strong increase in the signal at about @xmath250ms . here , the trapped wave packet finally finds its way out again , and it is only from thereon , that the wave signal consists of the proper fluid and spacetime modes . ( it is also from this time when we have taken the fourier transformation to obtain the upper spectrum in fig.3 . if we had included the first ring - down , this spectrum would have sat on top of a large broad peak . ) for less compact stars the local light speed inside the star is higher , and therefore the wave packet gets reflected much earlier , making the black - hole - like ring - down phase much shorter . we should mention that this ring - down phenomenon for very compact stars is much more pronounced for larger values of @xmath54 @xcite . of course , we gave a somewhat crude and heuristic explanation of this ring - down phenomenon ; we therefore postpone a more detailed study to a future paper @xcite . ccccc + ' '' '' model & @xmath251g/@xmath252 $ ] & @xmath253 $ ] & @xmath254km@xmath255 $ ] & m / r + ' '' '' 1 & @xmath256 & 0.802 & 10.81 & 0.109 + 2 & @xmath257 & 1.348 & 7.787 & 0.256 + 3 & @xmath258 & 1.031 & 4.992 & 0.305 + so far we have used polytropic equations of state , which are quite decent approximations to realistic equations of state as far as the bulk features of neutron stars like mass and radius are concerned . however , it is in particular the fluid oscillations that are very sensitive to local changes in the equation of state , which are due to the different behavior of the neutron star matter under varying pressure . it is therefore important to use realistic equations of state that take into account the underlying microphysics which determines the state of the matter as a function of pressure and temperature . realistic equations of state can not be given in analytic terms over the whole pressure range inside the neutron star , hence they usually exist in tabulated form only . to solve the tov equations in this case , one has to interpolate between the given values in order to obtain the stellar model with continuous functions of radius @xmath0 . in the following , we will make use of an equation of state called mpa @xcite , which yields a maximal mass model of @xmath259 . the non - radial oscillations modes for various stellar models have been compiled in @xcite . if we try to repeat the evolution of the same initial data , but now for a realistic stellar model with , say , @xmath260 grid points inside the star , we will find , after a few oscillations , an exponentially growing mode that immediately swamps the whole evolution . it is clear that this has to be a numerical instability , since the frequency and @xmath261-folding time of this growing mode strongly depends on the chosen resolution . in fact , by further increasing the resolution , the growing mode starts to weaken and eventually vanishes completely . this happens at about @xmath262 grid points inside the star , depending on the stellar model under consideration . hence , we have this strange fact that for low resolutions the numerical evolution tends to be unstable , whereas for resolutions high enough , the evolution is stable . we should stress that this happens only with the use of realistic equations of state . it does not happen for polytropic stellar models , even if the resolution is extremely low . further investigations show that the origin of this instability comes from the region close to the surface , where the sound speed has a sharp drop . in fig . 4 we show the profile of the sound speed @xmath263 for a stellar model using eos mpa with a central density of @xmath264 g/@xmath239 . it can be clearly seen that at @xmath265 km , there is a local minimum of the sound speed , where it drops down to @xmath266 . for larger @xmath0 we can see a series of much smaller dips , which is an artefact of the numerical spline interpolation between the tabulated points . but the dip at @xmath265 km is physical and is related to the neutron drip point , where the equation of state suddenly becomes very soft . since this occurs still in the low pressure regime , the dip is present for any realistic equation of state . it is indeed this dip in the sound speed that is responsible for the numerical instability , for if we remove it `` by hand '' , we obtain a stable evolution ! moreover , the occurrence of the instability is independent of the actual formulation of the equations , since the terms that are responsible are the same in each case . but what are the `` bad '' terms ? by simply crossing out individual terms , we find that the culprits are those terms in the fluid equations , , or which are not multiplied by @xmath267 . these are the terms which remain when the sound speed goes to zero , i.e. these which constitute the boundary conditions and . without these terms , the evolution would be stable for any resolution . the next question is then whether this instability is just a feature of our finite differencing scheme and does not appear for a different scheme . it turned out that the instability does indeed occur for all discretization schemes we have tried : the central difference scheme for the wave equations , a central difference scheme on staggered grids and the two - step lax - wendroff scheme both for the first order system of equations . of course , these are only explicit schemes , and we do not know whether switching to implicit schemes could be a remedy . however , in all above cases , the instability eventually vanishes when the resolution is increased above a certain threshold . for the central difference schemes , this threshold is at about @xmath268 grid points , depending on the stellar model , whereas for the lax - wendroff scheme , the required minimum resolution in order to overcome the instability can even exceed @xmath269 grid points inside the star , which is clearly unacceptable . a more elaborate analysis of the nature of the instability can be found in @xcite . there it is shown on the basis of a simplified toy equation that the resolution needed in order to overcome the instability strongly depends on the location and depth of the dip in the sound speed . the closer to the surface and the deeper the dip , the higher the required resolution . since this resolution is only needed in a very tiny region close to the surface , it would be enough to just refine the grid in this region and use a coarser grid outside . this could be accomplished , for instance , by fixed mesh refinement , since this region is determined by the profile of the sound speed , which does not change throughout the evolution . however , we then would have to deal with the transition from the coarse grid to the fine grid and vice versa , which might be troublesome . another drawback is that for each different stellar model , we would need a different grid refinement , and it would be a matter of trial and error to find the appropriate refinements for a stable evolution . yet , there is a better way out . we can try to find a new radial coordinate @xmath270 , which is related to the actual radial coordinate @xmath0 in such a way that an equidistant grid in @xmath270 would correspond to a grid in @xmath0 that becomes automatically denser in regions where the sound speed assumes small values . a simple relation between the grid spacings @xmath271 and @xmath231 that has the desired property is @xmath272 an equidistant discretization with a constant grid spacing @xmath271 would correspond to a coarse grid in @xmath0 for large values of @xmath41 , which becomes finer and finer as the sound speed @xmath41 decreases . from the relation , we can immediately deduce the form of our new coordinate @xmath270 as a function of @xmath0 . by replacing the @xmath273 s by differentials we obtain @xmath274 or @xmath275 as a consequence , the derivatives transform as @xmath276 and @xmath277 from the last relation , we see that the thus defined coordinate transformation will transform the wave equation in such a way that the propagation speed with respect to the @xmath270-coordinate will be unity throughout the whole stellar interior . of course , we have to use relation with some caution , for if @xmath278 this transformation becomes singular . and this is what happens at the stellar surface . if , for instance , the profile of the sound speed is given in the form @xmath279 , where we have @xmath280 , we can find an analytic expression for @xmath270 @xmath281 which tells us that at the surface @xmath166 we obtain @xmath282 . in this case , the coordinate transformation seems to be quite useless , since numerically we can not deal with a grid that extends up to infinity . we would have to truncate it somewhere . but from a numerical point of view , this is not that bad , since going to infinity in the @xmath270-coordinate would correspond to an infinitely fine resolution in the @xmath0-coordinate at the stellar surface . but this is numerically impossible as well , so truncating the @xmath270-coordinate at some point means to define a maximal resolution in @xmath0 at the surface . however , this case usually does not happen . in all cases considered , @xmath283 has always a finite value . it should be noted that the above transformation is of the same kind as the definition of the tortoise coordinate @xmath110 in eq . , which also leads to wave equations with unity propagation speed throughout the whole domain . thus , we may call @xmath270 a hydrodynamical tortoise coordinate . figure 5 shows the @xmath0-coordinate as a function of the @xmath270-coordinate for three different stellar models . from the curves , it is clear that an equidistant grid spacing in @xmath270 will result in an equivalent spacing in @xmath0 that gets very dense towards the surface of the star , which means that this part gets highly resolved . and this is exactly what we need in order to overcome the instability and to obtain a decent accuracy . to obtain the background data on the @xmath270-grid , we have to solve the tov equations with respect to @xmath270 . since we also need @xmath0 as a function of @xmath270 , we simultaneously solve eq . the transformed set of equations then reads : [ tovx ] @xmath284 of course , this transformation is only defined in the stellar interior for the fluid equations , for it is only there that the propagation speed is the speed of sound ( apart from the factor @xmath151 ) . if we still wanted to use eq . , where in the interior @xmath123 plays the role of the fluid , we would have to switch from the @xmath270-grid in the interior to the @xmath0-grid in the exterior , which is somewhat inconvenient . it is therefore much more natural to explicitly include the fluid equation , which is defined in the interior only . thence , it is only eq . which will be transformed according to the transformation , whereas we keep the wave equations for @xmath122 and @xmath123 as they are given in and . the transformed fluid part of the fluid equation reads @xmath285\;. \end{split}\end{aligned}\ ] ] here , the subscript @xmath270 denotes a derivative with respect to @xmath270 . the last missing thing is the transformation of the boundary condition . unfortunately , we can not transform eq . in a straightforward way , since at the surface it is @xmath278 , and therefore the transformation of the derivative @xmath286 is not defined . however , the inverse transformation can make sense if we note that if @xmath278 then any derivative with respect to @xmath270 has to vanish . this is in particular true for @xmath147 itself . thus , at the stellar surface @xmath166 , where the sound speed @xmath41 vanishes we impose the following boundary condition for @xmath147 : @xmath287 this corresponds to reflection at a loose end on the @xmath270-grid . since we only transform the fluid equation and not the equations for the metric perturbations @xmath122 and @xmath123 , we have to simultaneously use two different grids : the @xmath270-grid in the interior for the fluid variable @xmath147 , and the @xmath0-grid both in the interior and exterior for the metric variables @xmath122 and @xmath123 . because of the coupling , at each time step we have to interpolate @xmath147 from the @xmath270-grid onto the @xmath0-grid in order to update equation , and , vice versa , both @xmath122 and @xmath123 from the @xmath0-grid onto the @xmath270-grid in order to update equation . this can easily be accomplished by using spline interpolation . in fig . 6 we demonstrate the convergence of this new method using a stellar model based on the realistic eos mpa . for initial data similar to the ones used for fig . 2 , we plot the evolution of a gaussian wave packet together with the evaluation of the hamiltonian constraint for three different spatial resolutions . by doubling the resolution , the error in the hamiltonian constraint should decrease by a factor of 4 , if the scheme is of second order . conversely , the magnitude of the error should remain roughly constant , if for each doubling of the resolution , the error is multiplied by a factor of 4 . this has been done in fig . 6 , and one can clearly see that the magnitudes of the thus rescaled errors remain constant for the different resolutions . furthermore the scheme is stable since the error does not grow with time . the fluid modes have their largest amplitude in the region close to the surface . to obtain the correct mode frequencies in the power spectrum , it is important to have high resolution close to the surface of the star , and this is what is accomplished by our new coordinate @xmath270 . in fig . 7 , we show the power spectra of waveforms obtained from runs with different resolutions . as initial data , we chose a narrow fluid deflection at the center of the star together with @xmath124 . in this case , we suppressed any @xmath3-mode contribution . it is apparent that even for the quite moderate resolution of 50 grid points on the @xmath270-grid inside the star we obtain very accurate frequencies for the first couple of fluid modes . on the @xmath0-grid , however , a resolution of 200 grid points is still not enough to obtain the same accuracy , and the peaks of the higher @xmath2-modes in the spectrum are still quite far off their true values . of course , these results were obtained with a polytropic equation of state , since otherwise we would not have been able to perform the evolution on the @xmath0-grid at all , because of the occurring instability . finally , we want to demonstrate the effectiveness of our coordinate transformation by evolving an initial perturbation for the mpa equation of state . we choose the central energy density to be @xmath288g/@xmath239 , yielding a stellar mass of @xmath289 . the resolution is n = 200 grid points , which , again , would be too low to yield a stable evolution using the @xmath0-coordinate . here we do not have any problems , the evolution is stable for _ any _ chosen resolution . in fig . 8 , we show the power spectrum of the evolution of a sharp initial peak in the fluid perturbation @xmath147 , which leads to the excitation of a multitude of fluid modes . in the interval up to @xmath290khz , we can find 37 modes . the direct calculation of the first couple of modes with an eigenvalue code shows perfect agreement , which is demonstrated in the lower panel of fig . 8 . actually , the direct mode calculation is much more involved and time consuming than the evolution with a subsequent fourier transformation . even with a thousand grid points inside the star , it is a matter of minutes to obtain an accurate fluid mode spectrum for stellar models with realistic equations of state . of course , with this method one can only obtain the real parts of the frequencies , the imaginary parts are too small to be determined . in this paper , we have derived , using the adm - formalism , the perturbation equations for non - rotating , spherically symmetric neutron star models with barotropic equations of state . we have shown how to choose shift and lapse and the initial data on the three dimensional initial slice in order to obtain the regge - wheeler gauge . we pointed out that not only some of the metric components but also one of the extrinsic curvature components has to vanish in order to preserve the regge - wheeler gauge throughout the evolution . unfortunately , the equations that directly come out after the expansion in spherical tensor harmonics are not suited for the numerical treatment due to problems at the origin . we therefore had to introduce different variables that remove the singular structure of the `` raw '' equations . since the final system of equations in those variables is nothing else but a system of coupled wave equations written in first order form , it is therefore much more convenient to use instead the second order wave equations for the numerical evolution . those wave equations have already been previously derived by allen et al . @xcite in a different way . still , our formulation is quite useful because it facilitates the construction of initial data in terms of metric and extrinsic curvature perturbations . moreover , by introducing the adm - formalism for the derivation of the evolution equations of non - rotating neutron stars , we have performed the first step towards the much more involved task of deriving the evolution equations for rotating neutron stars . here , the linearization of the original einstein equations yields quite obscure perturbation equations , which first have to be cast into a suitable hyperbolic form before trying to discretize them on a numerical grid . with the adm - equations this is a much more straightforward procedure . we have presented results for different stellar models with polytropic equations of state . for the least relativistic model , the wave signal did not show any @xmath3-modes , although mode calculations show that they do exist . but they are so strongly damped that the tail - like fall - off wins over the @xmath3-modes . only for more compact stellar models , the @xmath3-modes are less strongly damped , and they can show up in the waveforms . however , even for very relativistic neutron star models , @xmath3-modes are not necessarily part of the waveform . in fact , their occurrence strongly depends on the choice of initial data . for initial data representing a gravitational wave pulse that will hit the star , @xmath3-modes in general will be part of the spectrum , however , for initial fluid perturbations , the choice of conformally flat initial data ( @xmath212 ) can totally suppress the presence of @xmath3-modes , whereas setting @xmath210 and @xmath291 , which can be regarded in some sense as a maximal deviation from conformal flatness , shows strong @xmath3-mode excitation . for very compact stars , we found that the complex frequency of the very first ring - down phase does not correspond to any of the proper quasi - normal modes of the stellar model . instead , the frequency is almost identical with the first quasi - normal frequency of a black hole with the same mass . we explained this phenomenon by observing that those parts of the potential , which are responsible for the first scattering of the impinging wave packet , lie outside the star and therefore are the same for a black hole . only if the part of the wave packet that could penetrate the neutron star is reflected , the proper ring - down with the trapped and fluid modes starts . this phenomenon , although already present in previous evolution results @xcite , apparently has not been recognized before . when switching to realistic equations of state , we are faced with a numerical instability that is due to the sharp drop of the sound speed at the neutron drip point , which is located quite close to the stellar surface . since the neutron drip point is located in the low pressure regime , it is a feature of any realistic equation of state , and therefore the instability should occur for all of them . the quite peculiar fact about this instability is that it disappears when the resolution is increased above a certain threshold . this threshold depends on the depth of the dip of the sound speed and is the higher the deeper the dip is . it seems that this instability is present for any explicit numerical discretization scheme , however , the minimum resolution that is needed in order to suppress it , can vary drastically for different schemes . indeed , when using the two step lax - wendroff scheme for the equations written as a first order system , we were confronted with an almost unsurmountable resolution threshold needed for stable evolution . this instability does not depend on the actual formulation of the equations and is present for any of the given formulations of this paper . in fact , it is not only a peculiar problem of the non - radial perturbation equations , the same instability also occurs in the radial case @xcite ; the reason being that the structure of the fluid equations is the same in both cases . as a remedy , we introduced the coordinate transformation , which leads to a wave equation for the fluid with unity propagation speed throughout the whole star . it is now possible to evolve oscillations of realistic stellar models for any resolution . as a further benefit , the evolution of the fluid on the @xmath270-grid is , for a given number of grid points inside the star , much more accurate than the corresponding evolution on the @xmath0-grid with the same number of grid points . this is because the @xmath270-coordinate resolves the region close to the stellar surface much better than the ordinary @xmath0-coordinate . this coordinate transformation could also prove useful for fast rotating neutron stars , since the evolution equations of the fluid will certainly have the same structure that leads to the numerical instability in the non - rotating case . since the rotating case is a 2d problem , resorting to higher resolutions may become impossible , and therefore our coordinate transformation could do the job . maybe , this transformation could even be of avail in nonlinear evolutions , although it is not that straightforward to implement , since there is no fixed background model with a static profile of the sound speed . a possibility is to view eq . as a dynamic coordinate condition that depends on the actual profile of the sound speed at every instance of time . i would like to thank pablo laguna and hans - peter nollert for helpful discussions and suggestions , and kostas kokkotas for his encouragement in developing this paper . i am also indebted to vincent moncrief , who kindly provided me with some of his notes . this work was supported by the deutsche forschungsgemeinschaft through sfb 382 . # 1#2#3 phys . rev . * # 1 * , # 2 ( # 3 ) # 1#2#3 phys . lett . * # 1 * , # 2 ( # 3 ) # 1#2#3 phys . d * # 1 * , # 2 ( # 3 ) # 1#2#3 phys . b * # 1 * , # 2 ( # 3 ) # 1#2#3 phys . reports * # 1 * , # 2 ( # 3 ) # 1#2#3 physica * # 1 * , # 2 ( # 3 ) # 1#2#3 j. comput . phys . * # 1 * , # 2 ( # 3 ) # 1#2#3 j. math . phys . * # 1 * , # 2 ( # 3 ) # 1#2#3 computer phys . # 1 * , # 2 ( # 3 ) # 1#2#3 class . quantum grav . * # 1 * , # 2 ( # 3 ) # 1#2#3 computers math . applic . * # 1 * , # 2 ( # 3 ) # 1#2#3 math . compt . * # 1 * , # 2 ( # 3 ) # 1#2#3 ann . phys . * # 1 * , # 2 ( # 3 ) # 1#2#3 astrophys . j. * # 1 * , # 2 ( # 3 ) # 1#2#3 astrophys . j. suppl . * # 1 * , # 2 ( # 3 ) # 1#2#3 acta astronomica * # 1 * , # 2 ( # 3 ) # 1#2#3 sov . # 1 * , # 2 ( # 3 ) # 1#2#3 siam j. sci . comput . * # 1 * , # 2 ( # 3 ) #1#2#3 astron . astrophys . * # 1 * , # 2 ( # 3 ) # 1#2#3 mon . not . # 1 * , # 2 ( # 3 ) # 1#2#3 proc . london , ser . a * # 1 * , # 2 ( # 3 ) # 1#2#3 i.j.m.p . c * # 1 * , # 2 ( # 3 ) # 1#2#3 nature * # 1 * , # 2 ( # 3 ) # 1#2#3 commun . phys . * # 1 * , # 2 ( # 3 ) # 1#2#3 prog . . phys . * # 1 * , # 2 ( # 3 ) # 1#2#3 gen . . grav . * # 1 * , # 2 ( # 3 ) # 1#2#3 class . quantum grav . * # 1 * , # 2 ( # 3 ) # 1#2#3 nuovo cimento b * # 1 * , # 2 ( # 3 ) # 1#2#3 j. soc . appl . math . * # 1 * , # 2 ( # 3 ) # 1#2#3 nucl . phys . a * # 1 * , # 2 ( # 3 ) # 1#2#3 rev . phys . * # 1 * , # 2 ( # 3 )	we present a new derivation of the perturbation equations governing the oscillations of relativistic non - rotating neutron star models using the adm - formalism . this formulation has the advantage that it immediately yields the evolution equations in a hyperbolic form , which is not the case for the einstein field equations in their original form . we show that the perturbation equations can always be written in terms of spacetime variables only , regardless of any particular gauge . we demonstrate how to obtain the regge - wheeler gauge , by choosing appropriate lapse and shift . in addition , not only the 3-metric but also the extrinsic curvature of the initial slice have to satisfy certain conditions in order to preserve the regge - wheeler gauge throughout the evolution . we discuss various forms of the equations and show their relation to the formulation of allen et al . new results are presented for polytropic equations of state . an interesting phenomenon occurs in very compact stars , where the first ring - down phase in the wave signal corresponds to the first quasinormal mode of an equal mass black hole , rather than to one of the proper quasinormal modes of the stellar model . a somewhat heuristic explanation to account for this phenomenon is given . for realistic equations of state , the numerical evolutions exhibit an instability , which does not occur for polytropic equations of state . we show that this instability is related to the behavior of the sound speed at the neutron drip point . as a remedy , we devise a transformation of the radial coordinate @xmath0 inside the star , which removes this instability and yields stable evolutions for any chosen numerical resolution .
forbidden line ratios are often employed to determine the hardness of the radiation field and the composition and physical state of irradiated gas within star - forming galaxies . of the quantities revealed by these ratios , the ionization parameter @xmath3 the ratio of ionizing photon density to hydrogen density is of special interest , as it measures the dimensionless intensity of ionizing radiation . because @xmath0 controls the ionization state ( e.g. , * ? ? ? * ) , it is easily disentangled from other parameters @xcite ; and because it indicates the ratio of ionizing radiation pressure to gas pressure ( e.g. , * ? ? ? * ) , it encodes information about the relative importance of these forces in star formation feedback . it is important , therefore , to understand the physical and observational effects which influence our estimates of @xmath0 in distant galaxies . when determined on kiloparsec scales , @xmath4 appears to reach a maximum value of about -2.3 in starburst environments , and indeed holds this value throughout the inner 500 pc of m82 @xcite . these points led @xcite to argue that this maximum @xmath4 is a property of the starburst phenomenon . however , higher values ( up to about -1 ) are sometimes seen in cases where regions are individually resolved @xcite . why do individual regions not exhibit arbitrarily high ionization parameter , and why are starburst determinations limited to @xmath5 ? both limits strongly suggest the operation of a saturation mechanism . we know of four possibilities : _ dynamics . _ the characteristic values of @xmath0 could originate in the dynamics of expanding regions . @xcite ( @xcite , hereafter d05 ) , in particular , explain the saturation of @xmath0 in terms of the ` stalling ' of expanding regions as they decelerate to the ambient sound speed . under a specific set of assumptions about the dynamics of regions and their interaction with ambient gas , this implies a unique value of @xmath0 at the time of stalling , which imprints itself on the line ratios integrated over an entire galaxy . however , there are two problems with the assumptions that enter this explanation . first , d05 follow @xcite in modeling regions as though they were enclosed stellar wind bubbles and therefore use the @xcite model to describe their motions . two pieces of evidence indicate that wind energy escapes regions , contrary to this assumption . trapped wind would produce x - ray luminosities as much as two orders of magnitude higher than what is observed ( e.g. , in the carina nebula : * ? ? ? moreover , d05 find that the wind luminosity must be drastically scaled down to reproduce the observed range of @xmath0 values . ( we revisit this issue below in [ ss : winds ] and [ conclusion ] . ) second , d05 assume that the ambient gas is the warm ( ionized or neutral ) component of the ism , and therefore has a unique sound speed of roughly @xmath6kms@xmath7 . although this is plausible , many of the ionizing photons will be consumed as the dense molecular material or cold neutral hydrogen is being cleared away . as the thermal sound speed of this matter is quite low , the appropriate stalling criterion involves the total ( turbulent , magnetic and thermal ) ambient pressure ( e.g. , * ? ? ? * ; * ? ? ? . _ homogeneous mixtures . _ a separate explanation has been provided by ( * ? ? ? * fs01 ) in the context of the central 500pc starburst core of m82 , where their and later observations @xcite demonstrate that @xmath0 is remarkably uniform . the fs01 model invokes an ensemble of neutral clouds with ionized outer layers , scattered among newly formed stars . the uniformity of @xmath0 must therefore be sought in the properties of this mixture , which are not part of the @xcite model . for this reason we do not investigate mixture models further , except to point out that most physical mechanisms limiting @xmath0 in individual regions would also operate here . _ radiation pressure confinement_. another physical mechanism for the saturation of @xmath0 was introduced by @xcite and elaborated by @xcite ( @xcite , d02 ) in the context of narrow - line regions in active galactic nuclei . when irradiation is sufficiently intense , ionized gas develops a pressure gradient to oppose the force of photons caught by neutral atoms and dust grains , and is squeezed into a thin layer near the ionization front ( if ) . this prevents gas pressure from becoming small compared to the ionizing radiation pressure ( as previously argued by @xcite ) . the corresponding upper limit on @xmath8 depends on the ionizing spectrum and the lines used to infer ionization ; a rough estimate based on pressure equilibrium gives -1.5 . while this is significantly above the value of -2.3 seen in starbursts , the details of line formation , among other effects , can reduce @xmath0 to about its observed maximum . _ dust - limited regions_. finally , the observed saturation could simply be a selection effect , caused by the fact that dust absorbs most of the ionizing starlight in regions with @xmath9 , suppressing their line emission . because extragalactic observations inevitably integrate over a wide range of local conditions , the inferred @xmath0 will be weighted toward regions of lower intensity . such a ` dust - limited ' state has been shown to successfully reproduce the suppression of fine structure lines relative to far - infrared luminosity in ultra - luminous infrared galaxies @xcite . our goal in this paper is to evaluate these possibilities and , in the process , to assess whether @xmath0 can be a useful proxy to measure radiation pressure feedback in regions in external galaxies . we review the physical effects which control the state of ionized gas and the values of @xmath0 inferred from forbidden line ratios , focusing on its saturation in starburst galaxies and the relevance of the last two mechanisms . because radiation pressure and dust opacity are both proportional to @xmath0 , they occur simultaneously . but , since they are truly two separate physical mechanisms , we take care not to conflate them . in this paper we concentrate on the internal structures of regions , only evaluating a very simple model for their dynamics and galactic populations ( [ ss : dust_abs ] ) . a companion paper ( verdolini et al . 2012 , in prep . ) presents a new suite of models for the emission from galactic populations of regions evolved with more sophisticated internal dynamics . we begin in [ s : parameters ] by reviewing the parameters of quasi - static , dusty , spherical regions . we draw here on the models of @xcite ( @xcite , hereafter dr11 ) , examining in detail their planar , radiation - confined limit ( appendix [ appendix : planarregions ] ) , but we also introduce a new parameter to account for an inner , pressurized region of shocked stellar wind . we discuss the scales of dust opacity and radiation pressure in [ ss : basics ] , confinement by radiation pressure and the maximum @xmath0 in [ ss : radconfinement ] , and the influence of an inner bounding pressure in [ ss : pressureparameter ] . in [ s : dynamics ] we focus on the extra effects which can reduce the maximum value of @xmath0 , including stellar winds and neutral clumps in the region . magnetic fields can in principle allow @xmath0 to violate its nonmagnetic upper limit , but this turns out to be difficult to arrange . we discuss the effects of internal and external dust extinctions on limiting @xmath0 in [ ss : dust_abs ] . in [ s : model ] , we provide a survey of @xmath0 values obtained from modeled mir line ratios within static and internally pressurized dusty regions , and demonstrate how stellar winds act to suppress @xmath10 . in [ s : implication ] , we use our results to interpret observations on the scales of individual regions and entire galaxies . in the static or quasi - static state , regions have several well - known limits : classical regions @xcite , which lack significant radiation pressure or dust opacity ; dusty regions @xcite in which radiation pressure is negligible ; and radiation - confined regions @xcite in which both radiation pressure and dust opacity are generally important . recently , dr11 has thoroughly explored the parameter space of dusty regions with radiation pressure , demonstrating a three - parameter family of models which takes all of these limits . ionization and force balance within spherical regions are especially simple under the following set of idealizations ( dr11 ) : the ionized gas temperature @xmath12 , the mean energy of ionizing photons @xmath13 , the effective dust opacity per h atom @xmath14 , and the effective recombination coefficient @xmath15 , are all uniform constants ; pressure from reprocessed radiation , such as the lyman @xmath15 line , is unimportant ; the flow is static , so forces must balance ; the pressure from stellar winds is neglected ( which we will relax below in [ ss : pressureparameter ] and [ ss : winds ] ) ; and there are no additional forces from gravity or magnetic fields . with these idealizations , regions are described by seven dimensional parameters : @xmath16 , the central output of ionizing photons per second ; @xmath17 and @xmath18 , the total force of ionizing and non - ionizing photons refers to the frequency range where the dust opacity is still @xmath19 ; dr11 assumes that the two opacities are equal . ] ; @xmath15 , the effective recombination coefficient ; @xmath20 , the thermal pressure per h density ; @xmath21 , the pressure at the ionization front ; and @xmath14 . other dimensional parameters of interest , determined by the seven above , include : @xmath22 , the h density at the ionization front ; @xmath23^{1/3}$ ] , the radius of a dust - free strmgren sphere of uniform density @xmath24 ; @xmath25 , the recombination - weighted ion density ( eq . [ [ def : n_em ] ] ) , and @xmath26 the radius of a uniform - density , dust - free strmgren sphere whose gas pressure equals the total , unattenuated radiation pressure @xmath27 at its boundary . ( our @xmath28 differs from @xcite s [ hereafter km09 s ] @xmath29 in that it accounts neither for the effects of dust extinction , nor for the additional pressure from stellar winds . ) yet another is @xmath30 , the ionization front radius . we focus on three dimensionless ratios which control the structure and appearance of the region . first of these is the characteristic ratio of ionizing radiation pressure to gas pressure in a classical strmgren sphere , @xmath31 where n@xmath32=10@xmath33 n@xmath34 @xmath35 , @xmath36s@xmath7 , and we have used @xmath37@xmath38s@xmath7 @xcite . here and elsewhere , we use the subscript ` st ' to denote conditions in a classical , uniform strmgren sphere without attenuation , rather than the actual region . an alternative to @xmath39 foregoes @xmath40 as a dimensional scale in favor of the actual ionization front radius @xmath30 ; @xmath41 this is useful if @xmath30 is observed or derived from the dynamics of the region ( km09 ) ; we use @xmath42 to parameterize our simulation output in [ s : model ] . bear in mind that @xmath42 decreases as radiation pressure becomes more important , whereas @xmath39 increases . our second parameter is the dust opacity in a classical region , @xmath43 both @xmath39 and @xmath44 are proportional to @xmath45 ; as a result @xmath46 in uniform regions . the third parameter is the _ dust discriminant _ @xmath47 where @xmath48@xmath49 . this sets the relative importance of dust and neutral h atoms in absorbing ionizing photons when radiation pressure confines the gas : in a situation where gas pressure matches the ionizing radiation pressure , the dust absorbs a fraction @xmath50 of all ionizing photons . fiducial parameters ( @xmath51ev ) give @xmath52 . additional parameters are required to capture the spectral form of the incident starlight and dust opacity , the most relevant of which , @xmath53 , compares the radiation force of non - ionizing starlight to that of ionizing starlight ; see dr11 and appendix [ appendix : planarregions ] . for emission in a given line , and therefore for the determination of line ratios and electron temperatures , the ratio of the characteristic density @xmath24 to the transition s critical density @xmath54 is a significant additional parameter . so long as this ratio is small for all of the lines which are either observed or contribute to the collisional cooling of the region , the density parameter will not be important . we return to this point in [ s : model ] . although the above idealization of regions is reasonable and admits a simple parameter space , we wish to relax its approximations and assumptions when considering observational constraints . microphysical assumptions , such as the constancy of @xmath12 , can be relaxed through the use of a modern photoionization code like cloudy or mappings ; we take this approach in [ s : model ] . macrophysical issues , such as stellar winds and bulk flow , must be addressed separately ( [ s : dynamics ] ) . we are especially interested in observational measures of the ionization parameter @xmath55 . this is directly related to the ionizing radiation - to - gas pressure ratio @xmath56 these are both functions of position @xmath57 , and are each observable only in an integral sense within the region . whereas @xmath58 and the ionizing spectrum set the local ionization structure and recombination spectrum , any recombination line ratio will depend on an average of @xmath0 weighted by the emissivities of the species involved . the value @xmath59 inferred from such line ratios will naturally reflect this weighting . one observationally relevant proxy is therefore the recombination - weighted average @xmath60 where both integrals are over the entire region . another commonly - used proxy is a ` geometrical ' estimate which relies on an observational determination of the characteristic ionized hydrogen density : @xmath61 we also define @xmath62 $ ] = @xmath63 $ ] for consistency . in a classical region ( uniform density , negligible dust absorption , radius @xmath40 ) , @xmath64 and , in such a region , @xmath65 by definition . clcl @xmath66({\mathbf r})$ ] & ratio of [ total , ionizing ] radiation pressure to gas pressure & eq . ( [ eq : xi ] ) & position dependent + @xmath67 & ionization parameter @xmath68 & & position dependent + @xmath69 & geometric estimate @xmath70 & eq . ( [ ugeom ] ) & + @xmath10 & observational estimate of @xmath0 for individual hii region or line of sight & & tracer dependent + @xmath71 & maximum value of @xmath10 in a quasi - static , unmagnetized layer & & tracer dependnet + @xmath72 & obs . estimate of @xmath0 for beams spanning many hii regions & eq . ( [ eq : uobs_of_population ] ) & biased by dust + @xmath73 & recombination - weighted average of @xmath0 & eq . ( [ eq : u_em ] ) & theoretical proxy for @xmath10 + @xmath28 & char . radius for radiation - gas pressure equality & eq . ( [ eq : rch ] ) & similar to @xmath29 in km09 + @xmath42 & radiation pressure parameter , @xmath74 & eq . ( [ psi_definition ] ) & @xmath75 implies @xmath76 + @xmath39 & ionizing radiation pressure parameter , @xmath77 & eq . ( [ eq : gammadefn ] ) & @xmath78 implies @xmath76 + @xmath79 & wind pressure parameter , @xmath80 & eq . ( [ define_omega ] ) & @xmath81 implies @xmath76 + @xmath82 & wind pressure parameter , @xmath83 & km09 & + @xmath84 & wind force parameter @xmath85 & & + @xmath86 & fraction of ionizing photons absorbed by gas rather than dust & dr11 & + @xmath87 & dust optical depth to ion . photons from source to ionization front & eq . ( [ eq : tauddefn ] ) & @xmath88 : dust - limited hii region + @xmath89 & dust discriminant , @xmath90 & eq . ( [ eq : gamma_in_dr11 ] ) , dr11 & @xmath91 implies @xmath92 when rad - confined + @xmath53 & effective ratio of non - ionizing to ionizing radiation force & dr11 & see note in table [ table : spectraparameters ] + @xmath25 & recombination - weighted average density & eq . ( [ def : n_em ] ) & proxy for observed density [ definitions ] these emission - weighted averages are directly related to the dust opacity , as the following argument shows . following dr11 , we define the local ionizing photon density to be @xmath93 , and the integrated recombination rate to be @xmath94 , where @xmath95 is the fraction of ionizing radiation absorbed by hydrogen rather than dust . rewriting both the numerator and denominator of equation ( [ eq : u_em ] ) , we find @xmath96 where @xmath97 is a dust absorption - weighted average of @xmath98 . equation ( [ xi_em - vs - tau_d ] ) is remarkable , because it relates a region s dust optical depth to its ionizing photon - to - gas pressure ratio in terms of the dust discriminant @xmath89 , and the order - unity parameters @xmath86 and @xmath99 take definite forms in each of the limiting regimes of static regions . in the context of radiation - confined regions , which we now discuss , @xmath89 determines the importance of dust absorption . in a classical region ( @xmath100 ) , except for a small region near the ionizing source @xcite , radiation pressure is negligible and has no practical effect on the distribution of ionized gas . in this case the recombination - weighted radiation pressure parameter @xmath101 is much less than unity , as can be seen from equation ( [ eq : xiem_vs_gamma_in_classical ] ) . this ceases to be true as the illumination becomes more intense . all of the momentum in ionizing starlight , and ( thanks to dust ) some of the momentum in non - ionizing starlight , is transferred to the ionized gas . in a quasi - static region without other forces , the radiation force must be balanced by a gas pressure gradient . many have pointed out that when radiation pressure is significant ( @xmath102 , ionized gas can be effectively excluded from the interior @xcite and confined to a thin shell near the if @xcite . within this shell the total pressure is nearly constant , as we show in appendix [ appendix : forcebalance ] , so the rise in gas pressure matches the drop in radiation pressure . although the ratio of radiation and gas pressures varies throughout the region , its characteristic value records a state of pressure balance either @xmath103 , if dust absorption is negligible ( @xmath104 ) , or @xmath105 if it is not ( @xmath106 ) . we explore the radiation - confined limit more quantitatively in appendix [ appendix : planarregions ] . because @xmath107 reaches its maximum value of around unity in radiation - confined layers , the recombination - weighted ionization parameter @xmath73 does as well ( dr11 ) : in a dusty layer with @xmath105 , we expect @xmath108 \simeq 10^{-1.5}$ ] . ( this estimate rises to @xmath109 in low - dust regions with @xmath104 , because of the factor of @xmath110 . ) moreover , the structure of the ionized zone becomes nearly independent of the strength of the illumination , apart from an overall scale , so long as @xmath111 . any ratio of forbidden lines , which are all well above or well below their critical densities , therefore saturates at a finite value in this radiation - confined limit ( e.g. * ? ? ? the ionization parameter @xmath10 which would be inferred from a specific set of line ratios tends to be close , but not equal , to @xmath73 . because the radiation force causes @xmath112 to be strongly stratified , there exists a diffuse inner region where lines preferentially form that require a higher ionization state @xcite . conversely , low - ionization transitions form in outer regions of lower @xmath0 . the observationally - inferred value @xmath59 therefore limits to a maximum value which depends on the lines being observed in addition to all the parameters apart from the strength of illumination : the chemical abundances , dust properties , and incident spectrum . in other words @xmath113 where @xmath114 depends on the lines in question and these other parameters , but not on the radiation intensity ( unless radiation pressure leads the density to cross one of its critical values ) . likewise , if density - dependent tracers are used to derive @xmath115 and @xmath69 , then this will exhibit an upper limit in the vicinity of @xmath116 in the radiation - confined state . in [ s : model ] we demonstrate the saturation of @xmath10 for a selection of mid - infrared emission lines with high critical densities using the cloudy code . we find in figure [ fig : ne3ne2_2myr ] that @xmath10 derived from a particular [ neiii]/[neii ] line ratio exhibits a maximum at @xmath117 for our fiducial parameters . because of the effects just described , the saturation of emission line ratios is a feature of the radiation - confined state in which photon momentum dominates over gas pressure gradients in the structure and dynamics of regions . as we shall see in [ ss : dust_abs ] , however , a selection effect due to dust attenuation ( [ ss : dust_abs ] ) also modulates the apparent value of @xmath10 in large - scale observations of dusty galaxies ( @xmath106 ) . both effects are important in the interpretation of galactic line ratios . in appendix [ appendix : planarregions ] and figures [ taudxifion ] and [ xiestimatesplanar ] , we present the properties of regions in the radiation - confined state , i.e. , the planar limit of dr11 s equations in the absence of a finite pressure at the inner boundary . this limit , which corresponds to @xmath111 ( and @xmath118 in the definitions of [ ss : pressureparameter ] ) , is parameterized only by the dust discriminant @xmath89 and the spectral parameter @xmath53 , which quantifies the non - ionizing radiation pressure . although these solutions are only a subset of dr11 s three - parameter model family , they improve upon the treatment by @xcite by accounting for the gas density gradient induced by radiation pressure . they show that : ( 1 ) @xmath119 saturates at a value which is near unity and only logarithmically dependent on @xmath89 . ( 2 ) dust absorbs most ionizing radiation , so @xmath120 for astrophysically relevant values of @xmath89 . as dr11 discusses , @xmath121 at solar metallicity unless dust is effectively removed . therefore radiation - confined regions are also marginally ` dust - limited ' ; see [ sss : int_abs ] . ( 3 ) for relatively dust - free gas with @xmath104 , the planar solution is not compact in radius . it must either be bounded inward by a bubble of hot gas , or be matched to a spherical solution ( as in dr11 ) . is initially large . we do not account for this effect in our equations and models , as we assume a uniform mixture of gas and dust . ] in the radiation - confined shell , the column density of ionized gas takes a characteristic maximum value : either @xmath122 if dust absorption is negligible , or several times @xmath123 if it is not . what happens if the mean column density of the ambient medium is less than four times this shell column , so there is not enough matter to achieve it ? in this case , it is easy to show that the initial strmgren sphere forms outside @xmath124 , so that @xmath125 and @xmath126 : the radiation - dominated phase never occurs . the ionizing sources of regions ( typically stars , star clusters , or agn ) often produce fast outflows which decelerate or are deflected before they reach the if , thereby creating regions of hot and pressurized gas within the photo - ionized zone . what new parameter should we introduce to capture the effect of such zones on the regions emission ? the ratio of outer to inner pressures is not appropriate , because the two are equal in a classical region and also in a wind - confined shell . neither is the volume fraction of the interior zone , because photo - ionized gas can be confined to a thin shell by radiation pressure just as well . instead we choose @xmath127 where @xmath128 is the difference of the product of pressure and volume between the if and the inner edge of the region ( the outer edge of a pressurized bubble ) . this combination reflects the strength of the interior pressure , both for classical and radiation - dominated regions . in fact @xmath79 directly reflects the contribution of the interior bubble to the total kinetic energy of the region , which , according to the virial theorem @xcite , controls the expansion of it and any shell of neutral matter it pushes outward ( e.g. , km09 ) . in figure [ fig : parameterspace ] we indicate the physical regimes of spherical , quasi - static , dusty regions surrounding a central wind bubble . quasi - static hii regions typically have an inward gradient of @xmath0 . for classical regions this is simply because the ionizing flux is greater at smaller radii , while for radiation - confined regions it is due to the gas density gradient . in either case , replacing a region s inner portion with hot gas has the effect of removing its most highly photo - ionized zone . this effect will be greatest for those lines more which form preferentially in regions of highest @xmath0 . see [ ss : winds ] for a discussion of the effect on @xmath10 . in radiation - dominated ( @xmath111 ) and relatively dust free ( @xmath129 ) regions , low - density ionized gas fills the interior ( appendix [ appendix : planarregions ] ) . because this low - density gas is ideal for the formation of lines sensitive to regions of high @xmath0 , we expect these line ratios to be especially sensitive to interior pressure when @xmath111 and @xmath130 . several physical effects modify @xmath59 relative to its theoretical maximum value @xmath71 within an individual region . in the following subsections we consider stellar winds , neutral clumps , dust absorption of non - ionizing light , and magnetic fields . of these , only magnetic fields act to raise @xmath10 relative to @xmath71 . although highly - magnetized regions can in principle have @xmath131 , we shall see that this is very difficult to arrange in practice . as discussed above in [ ss : pressureparameter ] , any force which clears ionized gas from the interior of an region will have the effect of reducing the characteristic value of @xmath59 , especially for those lines which form in regions of highest @xmath0 . the most relevant force is the pressure of a shocked wind produced by the same source that ionizes the region . most such regions are ionized by hot stars , which inject fast line - driven winds . however the significance of stellar wind pressure depends on the wind strength , e.g. the ratio @xmath132 between the momentum fluxes of winds and starlight , and the trapping of wind energy within the region . km09 combine these into a single parameter @xmath133 which encapsulates the contribution from shocked winds relative to the direct momentum flux of starlight . so long as the combined force of winds and radiation compresses the ionized layer into a thin shell , the enclosed volume is constant and @xmath79 reduces a comparison of pressures . since @xmath134 in this case ( appendix [ appendix : forcebalance ] , ignoring on the last term the factor @xmath135/(1+\beta)$ ] which accounts for incomplete absorption ) , we expect @xmath136 further , if we consider a typical location where half of the incident radiation has been absorbed and its pressure applied to the gas , the same approximations lead to the estimate @xmath137 although , in the case where confinement is accomplished by radiation alone ( @xmath138 ) , the values of @xmath107 plotted in figure [ xiestimatesplanar ] are more accurate . we refer to section 3.1 of km09 for a detailed discussion of the characteristic values of @xmath84 and @xmath139 one might expect around massive star clusters . for our purposes , it suffices to consider three limiting cases : ( 1 ) perfectly trapped , adiabatic bubbles of shocked wind , as envisioned in the models of @xcite and @xcite and employed by @xcite , which have @xmath140 ; ( 2 ) leaky wind bubbles ( @xcite , km09 ) , for which @xmath141 in the case of a very massive cluster of solar metallicity ( km09 ) ; and ( 3 ) very weak wind bubbles with @xmath142 , such as one might expect if the central source consists of low - metallicity or lower - mass main sequence stars . in the first case ( adiabatic wind bubbles ) @xmath143 and @xmath59 will be limited to values well below @xmath144 . indeed , @xmath145 can easily be computed from the self - similar solutions of @xcite : using @xmath146 and evaluating for the case of a steady wind in a uniform medium @xcite , @xmath147 the factor of two in the first equality arises because a pressure - confined ionized layer is nearly uniform and optically thin to dust ; therefore @xmath0 drops linearly in radius from the value @xmath69 at its inner edge . since @xmath148kms@xmath7 from a young cluster which fully samples the imf , equation ( [ ugeominweaver77 ] ) yields @xmath149 . equation ( [ ugeominweaver77 ] ) assumes the ionization front is trapped within the swept - up ambient matter , which requires @xmath150 where @xmath151 is the ambient h density in @xmath35 , @xmath152kms@xmath7 , and @xmath153kms@xmath7 . in the second case ( leaky bubbles ) , the wind force is of order @xmath154 and is therefore significant when @xmath139 exceeds unity . so long as the ionized layer is thin , the suppression of @xmath10 relative to @xmath71 is of order @xmath155 as given by equation ( [ xifromftrapw ] ) . in the last case ( very weak winds ) , @xmath10 will be unaffected by their presence , although high-@xmath0 lines from the innermost regions may be ( especially when @xmath156 ) . in our modeling effort ( [ s : model ] ) , we consider a wide range of possible values for @xmath79 . spherically symmetric , partially - radiative bubbles @xcite , which have @xmath139 lower than the adiabatic case but well above unity , are not especially relevant here because they require slower winds than is typical of ionizing stars . fast - wind bubbles can also become radiative , however , because of mass entrainment from evaporating clumps ( see [ ss : clumps ] ) . note that @xcite argued that this effect prevents the filling factor of shocked wind , and therefore also @xmath139 , from exceeding unity . as an aside , we note that some ultracompact regions are probably formed in quasi - static layers confined by the ram pressure of protostellar winds , which are both highly collimated and much stronger than radiation pressure . combining the outflow force - luminosity correlation reported by @xcite with the collimation model of @xcite , @xmath157 for an angle @xmath158 from the wind axis and a bolometric source luminosity @xmath159 . such regions are typically in the first , wind - dominated regime , even well away from the wind axis . this ram - pressure confined phase can only exist in directions where the wind itself is fully ionized . the alternate case , in which the if forms in the wind , was considered by @xcite . it is inertially - confined in the sense of [ inertialconfinement ] below , and can exhibit @xmath160 . regions are often porous and clumpy instead of spherical , as we invoked above in the context of wind trapping . here we wish to point out that clumpiness reduces the inferred @xmath10 relative to a spherical region of the same radius . the net effect is therefore to reduce @xmath10 relative to @xmath114 , although the upper limit does not necessarily change . in the absence of significant radiation pressure , a clump subjected to ionizing radiation is first crushed by the recoil of its photo - evaporative flow @xcite . if the clump is dense enough to survive the first implosion phase , it comes into a quasi - static equilibrium with the induced pressure @xcite , in which its radius of curvature is @xmath161 ( on the symmetry axis , or the point most directly illuminated by starlight ) . ionization balance in the presence of this accelerating and diverging photoionized flow implies @xmath162 where @xmath163 is the ionized gas density just outside the clump if at this point , @xmath164 is the distance from the clump to the central source , and @xmath165 @xcite the parameter which describes the characteristic ionizing radiation - to - gas pressure ratio ( @xmath166 ) for this outflow is therefore @xmath167 compared to conditions at a spherical if of radius @xmath30 , @xmath168 a clump generally has a smaller value of @xmath169 , and therefore @xmath0 , than its parent region , because its radius of curvature @xmath161 is significantly less than the ionization front radius @xmath30 of the region . a clump s photoevaporation flow ends at a deceleration shock caused by the pressure from ionized gas in the parent region , stellar winds , or radiation . as @xmath170 increases toward unity , radiation pressure requires that the shock approach the sonic point of the photoevaporation flow , and when @xmath171 , radiation impedes the outflow of ionized gas . for @xmath172 , the ionized gas is confined by radiation pressure to a thin layer on the surface of the clump , leading to a saturation of @xmath119 and @xmath10 , exactly as described in [ ss : radconfinement ] and appendix [ appendix : planarregions ] . just as before , dust opacity will be significant so long as @xmath106 . @xcite sketch the flow for @xmath173 . however there is one significant difference between the effects of clumpiness and stellar winds . the ionized layer of a clump can not be hydrostatic : it must either evaporate away , or be dragged along the clump surface by the tangential component of the radiation momentum flux , and also possibly by the stress from a turbulent boundary layer between photoionized clump gas and shocked stellar winds . such wind and photon - driven flows are a plausible source of supersonic line widths in giant regions , for photoevaporative flows only produce motions of up to about twice the ionized sound speed @xcite . it is not at all clear that a dynamical equilibrium exists which would prevent a clump with @xmath174 from being dynamically disrupted , although a magnetic buffer zone may allow this ( e.g. , * ? ? ? magnetic fields supply an additional form of pressure which is available to balance the force of radiation , and therefore have the potential to raise @xmath10 _ above _ however this is somewhat difficult to arrange . consider a planar shock of speed @xmath175 running into a medium of hydrogen density @xmath176 , magnetic field perpendicular to the shock normal @xmath177 and alfvn speed @xmath178 . the post - shocked fluid is in contact with ( and driven forward by the pressure of ) an ionized layer . ignoring any change in pressure across the post - shocked layer as well as streaming of the ionized gas ( both of which can be order - unity effects ) , the density of ionized gas at the contact discontinuity is @xmath179 if it is pressure - supported , and @xmath180 if it is magnetically supported . magnetic pressure support therefore requires @xmath181 where @xmath182kms@xmath7 is the ionized sound speed . magnetic support is therefore only relevant for shock speeds in excess of about @xmath183kms@xmath7 . magnetic fields should be relatively more important within clumpy regions , because of the tendency for streaming and turbulent flows to amplify them toward equipartition with the kinetic energy . our discussion so far has focused on regions which are in force balance because ionized gas resides within them for long enough to be crossed by sound waves . however , some regions will be too young ( or have crossing times too long ) to develop into this state , such as the regions created in cosmological reionization . in others , ionized gas flows through the region at speeds well in excess of the ionized sound speed ; these include ionized winds and jets , ionized ejecta , ionized zones of supersonic turbulence , and regions around runaway stars . indeed , champagne flows @xcite are density - bounded to one side and are only approximately in force balance on the side toward the ionization front . none of these impulsive or inertially - confined flows is limited to @xmath184 and @xmath185 for the reasons we have outlined : therefore , higher values of @xmath10 could be taken as spectral evidence of this state . loosely speaking , an observation of @xmath10 above @xmath114 implies that either a non - thermal form of pressure , such as ram pressure or magnetic fields , balances radiation pressure , that another force ( such as gravity ) is at play , or that there has not been time to establish equilibrium . it is worth noting a couple additional effects involving the interaction among light gas , and dust : _ lyman-@xmath15 trapping and dust absorption_. in the absence of dust , photons from the ly-@xmath15 line scatter many times before escaping . once the scattered line radiation is sufficiently strong its pressure will augment gas pressure , leading to an _ increase _ of @xmath114 ( see * ? ? ? * for an analysis of line pressure in the agn context ) . the effect is analogous to the role of magnetic fields , except that the line pressure is proportional to the ionizing radiation pressure . at solar metallicity , a typical grain population limits the ly-@xmath15 line pressure within the ionized zone to only @xmath186 of the ionized gas pressure @xcite , but line pressure is significant at low metallicity or in regions cleared of dust . even in the absence of dust , the dynamical effect of line pressure limits @xmath187 within quasi - static regions @xcite . _ dust photoelectric effect_. photoelectrons ejected from dust grains represent an additional sources of heat and ionization , while collisions and recombinations with dust grains are ( less significant ) sinks . these processes , considered in detail by @xcite , @xcite , and others , are fully accounted for within the cloudy code . they depend both on the grain abundance ( captured here by @xmath89 ) and the grain size distribution . of these , trapped line radiation tends to make @xmath114 to decrease as @xmath89 increases , whereas photoelectric heating has the opposite effect by raising the ionized gas temperature in dustier gas . cccc @xmath188 $ ] & 6.99 & 15.76 & @xmath189 + @xmath190 $ ] & 8.99 & 27.63 & 3.0@xmath191 + @xmath192 $ ] & 12.81 & 21.56 & 6.5@xmath19310@xmath194 + @xmath195 $ ] & 15.55 & 40.96 & 1.3@xmath19310@xmath194 + @xmath196 $ ] & 18.71 & 23.34 & 2.0@xmath19310@xmath33 + @xmath197 $ ] & 10.51 & 34.79 & 6.0@xmath19310@xmath33 [ mirlines ] emission lines usually arise from regions large enough to encompass many local environments , so their ratios reflect an average of the local conditions . dust absorption has the effect of dimming high-@xmath0 regions , suppressing their contribution to the average and causing a saturation of the apparent ionization level . this bias arises from the dust absorption of ionizing radiation within regions , and from the absorption of the forbidden lines themselves , predominantly outside the regions themselves . if a region is sufficiently dusty , then the dust optical depth will exceed unity before radiation pressure becomes important . the critical value of the dust discriminant @xmath89 is apparent if we combine its definition ( eq . [ eq : gamma_in_dr11 ] ) with the characteristic ionizing radiation - to - gas pressure ratio in a classical strmgren sphere ( eq . [ eq : xiem_vs_gamma_in_classical ] ) : @xmath198 although this refers to the classical case of a filled strmgren sphere , for which both @xmath199 and @xmath200 , it demonstrates that , when @xmath201 , dust extinction of ionizing photons is significant even for some regions which are not radiation dominated . conversely , when @xmath202 , radiation pressure confinement can occur when dust extinction is not strong . the former scenario ( dust - limited regions ) is consistent with our expectations of milky way conditions , so long as dust is not cleared from the gas ; figure [ fig : parameterspace ] maps the physical regimes in this case . the latter applies to low - metallicity environments , or more broadly if dust clearing is effective . ( we will not distinguish the critical @xmath89 from unity in our following discussion . ) because recombination light is always redder than the photons which ionize an region , and because dust opacity increases with frequency , the line photons of interest tend to escape the ionized gas relatively unabsorbed . nevertheless , because dust consumes ionizing photons , the rate of recombinations is reduced in dust - dominated regions . the ionization efficiency @xmath86 drops from unity ( for a classical region ) to an asymptotic value of order @xmath203 ( in the planar limit , when @xmath106 : appendix [ appendix : planarregions ] ) as radiation pressure and dust extinction become significant ; see dr11 s eq . ( 21 ) . for the parameters of interest , the minimum value of @xmath86 is about 0.3 . because of this , recombination light from the regions of highest @xmath10 is suppressed by the factor @xmath86 so that , on a galactic scale , @xmath10 is reduced relative to @xmath114 . specifically , so long as the extinction of line photons ( [ sss : ext_abs ] ) is negligible , the inferred value of @xmath10 on galactic scales will be approximately the emission - weighted average of all the regions encompassed by the observation : we call this @xmath204 . if we label a population of @xmath205 regions by their ionizing luminosities and apparent ionization parameters , then @xmath206 where in the second equality we assume @xmath207 and define @xmath208 to be the average of @xmath209 among regions whose ionization rates lie in a narrow range around @xmath16 . galactic region luminosity functions obey @xmath210 at high @xmath16 , but are significantly shallower at lower luminosities ( @xmath211s@xmath7 , where the imf sampling is incomplete ; * ? ? ? each decade of @xmath16 above this break therefore contributes about equally to @xmath204 , while those below it are increasingly insignificant . in general , a galaxy with more massive star clusters ( a population extending to higher @xmath16 ) and a denser circum - cluster medium ( so that regions are more compact ) will have regions characterized by higher values of @xmath10 and therefore attain a higher @xmath204 . however , the decline in @xmath86 accompanying the increase in @xmath44 suppresses the contribution of those regions with highest @xmath10 . so long as the region population includes a range of intensities and a significant range of @xmath87 and @xmath86 , the net effect will be that @xmath204 is an underrepresentation of the starlight - weighted average of @xmath10 ( the result of taking @xmath212 in eq . [ eq : uobs_of_population ] ) . in particular , @xmath204 will be closer to the value it has when @xmath213 , @xmath214 it is important to remember that this selection effect does not apply to individual regions , or parts thereof , which saturate at a higher value ( @xmath114 , [ ss : radconfinement ] ) . it operates only when starlight photons are distributed rather evenly across environments with a range of @xmath10 and @xmath86 , so that variations of @xmath86 can affect the apparent value of @xmath10 . for this reason , it does not prevent @xmath204 from limiting @xmath114 if _ all _ the regions in a galaxy are radiation - confined . we demonstrate these points in figure [ fig : uobsgalmodel ] , for which we generate a toy model of the region distribution within a galaxy . we start by examining a population of @xmath215 associations generated for a prior work ( @xcite , based on @xcite , in which @xmath216 at high @xmath16 ) , extracting from it the luminosity function @xmath217 and the typical ionizing lifetime @xmath218 as functions of @xmath16 . we then realize a scenario in which all regions expand into density profiles with @xmath219 , taking the mean column density @xmath220 to be constant and the same for all regions . ( for simplicity , we neither vary @xmath221 with @xmath16 , nor alter the upper limit of @xmath16 with @xmath221 although both effects are likely to occur in real galaxies . ) in each luminosity bin we evaluate the time evolution @xmath222 using km09 s equation ( 13 ) , which accounts for the combined force due to radiation , partially trapped winds ( if any ) , and photo - ionized gas . we limit regions growth according to their finite initial size and stalling in the finite hydrostatic pressure of the environment ( km09 s 2.3 ) . we assume for this model that stellar winds are weak : @xmath223 . with this assumption , we can construct @xmath224 and @xmath225 ( from the neiii / neii line ratio ) using our own numerical models ( [ s : model ] ) . taking the time average for @xmath226 to get @xmath227 and @xmath228 , we arrive at @xmath204 . for comparison we also plot the starlight - weighted averages of @xmath10 , @xmath86 , and the radiation - to - gas force ratio @xmath229 ( see km09 eq . 10 ) . although the dust selection effect is not strong in our toy model , amounting to a suppression of @xmath204 by only @xmath230 over a relevant range of @xmath221 , we suspect it might be more significant if some of our toy model assumptions were relaxed . our toy model does not account for heterogeneity of the ambient medium , which permits the existence of a diffuse ionized component ; it is likely to under - estimate @xmath231 and over - estimate @xmath204 for this reason . in the expansion of an region , most of the ambient medium is swept before the ionized gas rather than being ionized itself . therefore , although dust within the region does not severely attenuate recombination line photons , the surrounding matter poses a thicker barrier . this external dust extinction is also an increasing function of @xmath10 , as the following argument shows . suppose that each star cluster , of mass @xmath232 , forms within an overdensity of radial profile @xmath233 , and moreover that , statistically speaking , @xmath234 . the toy model of [ sss : int_abs ] adopted @xmath235 and @xmath236 , but @xmath237 could be as low as 0 or as high as 2 , and @xmath238 is likely to be positive . for the star clusters massive enough to sample the imf , @xmath239 . how then does the dust optical depth of the neutral matter outside an region scale with @xmath0 within it ? we restrict attention to @xmath240 , so that optical depth varies significantly with radius . when the region size is @xmath30 , the external dust optical depth at any frequency is then @xmath241 ; at the same time , the radiation pressure parameter @xmath242 , so that @xmath243 for massive clusters . when @xmath240 , regions of a given @xmath16 are most enshrouded when they are most radiation - dominated ( i.e. , when @xmath42 is lowest ) . the region luminosity increases or decreases @xmath87 at _ fixed _ @xmath42 , depending on whether @xmath244 or not . however , regions with higher @xmath16 tend to also be more radiation dominated ( i.e. , have lower @xmath42 ) . when @xmath244 it is clear that @xmath87 decreases as @xmath42 increases . insofar as this external dust is optically thick at the frequencies of the lines being observed , this will tend to suppress emission from high-@xmath10 regions , reducing this quantity relative to its theoretical maximum value . note that , as we saw in [ ss : radconfinement ] , the radiation dominated state which produces @xmath114 is in fact impossible if the column density of the ambient medium is too low . the external attenuation of line photons is minimized by considering the longest possible transition wavelengths ; we focus on mid - infrared lines in [ s : model ] . it should also be mitigated by holes in the swept - up neutral shell , as well as clumpiness of the ambient dust distribution ( e.g. , @xcite ) . in this section we present numerical results for spherical , dusty , quasi - static regions . our goal is to account explicitly for the possibility of an inner boundary pressure due to shocked wind ( as parameterized by @xmath79 ) and to demonstrate the saturation of line ratios and inferred @xmath10 due to radiation pressure confinement . we consider only a few ionizing spectra and do not vary the composition of gas or the grain population ; a full exploration of these parameters is beyond the scope of this work . we use cloudy version 08.00 , last described by @xcite , to account for many important microphysical effects including the photoelectric effect , collisional cooling , and the pressure due to optically thick recombination lines . however we consider only quasi - static regions in perfect force balance , and do not account for secular effects , such as grain drift , which might lead to inhomogeneities in the composition . using starburst99 @xcite , we generate the ionizing continua from coeval star clusters of different ages , all of which we assume are massive enough to fully sample the stellar initial mass function , which we take to have exponents -1.3 and -2.3 between stellar mass boundaries 0.1 , 0.5 , and 120 m@xmath245 . we employ the geneva high mass - loss evolutionary tracks with solar metallicity . these are optimized for modeling atmospheres of high mass stars and are recommended by @xcite . we adopt pauldrach / hillier atmospheres , as these include non - lte and line - blanketing effects @xcite for o stars @xcite and wolf - rayet stars @xcite . the combination of the geneva high mass - loss tracks and pauldrach / hillier atmosphere is recommended when wolf - rayet stars are important @xcite . starburst99 output spectra are recorded from 0 to 11 myr with 0.5 myr steps . starburst99 output continuum spectra are fed into cloudy as the ionizing continuum of each simulated region . each region is spherical and in perfect force balance ; we allow radiation pressure to exceed gas pressure , in contrast to cloudy s default setting . we adopt cloudy s default ism abundances and dust grain size distributions . each calculation stops where temperature drops to 100 k , and so encapsulates the if . each set of the simulations outputs the integrated luminosity of selected mir emission lines form table [ mirlines ] , including [ ariii]@xmath246 , [ arii]@xmath247 , [ neiii]@xmath248 , [ neii]@xmath249 , [ siii]@xmath250 , and [ siv]@xmath251 . in both observational and theoretical studies it is necessary to translate some observed line ratios into a set of physical parameters , including @xmath0 ; as we have already discussed , we use @xmath10 to designate the apparent @xmath0 which characterizes a single region . a standard definition for @xmath10 is the value of @xmath0 at the inner boundary of an ( often uniform ) slab which reproduces the observed lines ; however this has pitfalls both for low and high @xmath0 ( [ conclusion ] ) , and ignores the possible role of wind pressurization . we choose instead to associate @xmath10 with the value of @xmath0 within a _ homogeneous _ mixture of gas and radiation . to do so , we require a library of the line luminosities emitted by such homogeneous mixtures , in which physical parameters such as @xmath0 , @xmath252 , and input spectrum are varied . we construct this library using the very innermost zone of a sequence of cloudy models . selected results are presented in figures [ fig : ar3ar2_1zone ] to [ fig : s4s3_1zone ] . figure [ fig : onezoneu ] demonstrates how well the [ neiii]/[neii ] ratio reflects @xmath0 . other line ratios , [ ariii]@xmath253/[arii]@xmath254 and [ siv]@xmath255/[siii]@xmath256 , are also most sensitive to the intensity of radiation field , i.e. , to @xmath0 . we work with a fixed dust population and several input spectra corresponding to fully sampled , coeval star clusters of different ages . therefore , while the ratio @xmath53 of non - ionizing to ionizing radiation force varies greatly from young ( 2 myr ) to relatively old ( 6 myr ) cluster spectra , the dust discriminant @xmath89 is almost constant . we list the relevant quantities in table [ table : spectraparameters ] . whenever we are not exploring age dependence , we employ spectra from 2 myr clusters , as this is halfway through their typical ionizing lifetimes . cccccc 0 & 20.6 & 0.61 & 0.61 & -20.9 & 11.2 + 2 & 19.1 & 1.25 & 1.12 & -20.8 & 13.8 + 4 & 19.2 & 2.81 & 2.51 & -20.8 & 13.2 + 6 & 16.7 & 16.7 & 11.3 & -20.7 & 19.3 [ table : spectraparameters ] for each spectrum our goal is to map the line luminosities , line ratios , and resulting @xmath10 as function of @xmath42 and @xmath169 , i.e. , to scan the parameter space displayed in figure [ fig : parameterspace ] . practically , we accomplish this by varying the central luminosity and innermost density of the region , at a fixed inner radius . for this study we restrict ourselves to regions with ionized densities well below the critical densities for the transitions being considered ( table [ mirlines ] ) so that density does not enter as a third parameter . our plots in this section should therefore not be used for regions with pressures exceeding @xmath257@xmath35 for [ siii ] and [ siv ] , or @xmath258@xmath35 for the others . moreover we keep the metallicity of gas and stars fixed , and vary only the age of the stellar population ( rather than its imf or other properties ) . in addition to the line luminosities , we compute @xmath86 from the ratio of total h@xmath53 emission to the amount of h@xmath53 emission expected in the absence of dust . we also compute @xmath259 , a useful observational diagnostic of the strength of ionization , where @xmath260 the recombination - averaged density a proxy for the observationally inferred density @xmath261 . for regions without internal wind bubbles , @xmath259 determines @xmath169 and @xmath87 ( [ s : parameters ] , dr11 ) ; but pressurization breaks this relation . we begin by comparing the physical structures of a radiation - confined shell ( @xmath262 , @xmath263 ) with those of a wind - confined shell ( @xmath264 , @xmath265 ) in figure [ fig : ne_p_twomodels ] . as expected , radiation confinement leads to a dramatic gradient of electron density and pressure , whereas wind pressure causes the region to be nearly uniform . figures [ fig : ne3ne2_2myr ] to [ fig : ne3ne2_6myr ] show @xmath10 derived from [ neiii]/[neii ] ratios for cluster ages 2 , 4 , and 6 myr . immediately apparent is the saturation of @xmath10 in the radiation - confined state , @xmath266 and @xmath267 , as is its suppression in wind - confined shells ( @xmath268 ) . at age 4 myr , the most massive stars have left the main sequence , so the ionizing flux drops and @xmath10 values decrease accordingly . the pattern of saturation in the radiation - dominated quadrant is , however , unchanged . by 6 myr , only stars with mass @xmath269 30 m@xmath245 are still on the main sequence , which are responsible for producing only 7.5% of the ionizing flux at zero age . therefore @xmath10 values decreased gradually comparing to that at younger ages . note that @xmath10 still saturates in the same manner as the 2 myr and 4 myr cases . a second set of contours in the @xmath42@xmath79 plots , labeled with dashed black lines , indicate @xmath270 . this quantity increases as @xmath42 decreases , which implies that dust opacity increases as radiation pressure also becomes more important . this is consistent with the analytical prediction in [ s : parameters ] . @xmath271 values are weakly correlated with @xmath79 in the @xmath272 regimes where stellar winds are negligible . then the contours become a linear function of both @xmath42 and @xmath79 when @xmath273 . as we have discussed in [ ss : dust_abs ] , dust grains absorb a fraction @xmath274 of the ionizing photons in regions . we estimate @xmath86 using the ratio of total h@xmath53 emission to the dust - free value h@xmath275 , which we calculate by assuming that all ionizing photons are used for photoionization , and each ionizing photon is responsible for one recombination . the emitted h@xmath53 luminosity which we tally zone - by - zone to avoid extinction of the line photons reflects the suppression of ionizations due to dust absorption of starlight . the variation of h@xmath53/h@xmath275 with @xmath42 and @xmath79 is shown in figure [ fig : fion ] with solid blue contours , for the ionizing spectrum of a 2 myr - old cluster . again , dashed black contours denote @xmath276 , providing a calibration with respect to previous @xmath10 contour plots at the same age . in the quadrant corresponding to radiation - confined shells , the h@xmath53/h@xmath275 line ratio is 0.35 . this is consistent with our prediction , in appendix [ appendix : planarregions ] , that @xmath86 falls to @xmath277 in the limit @xmath278 , @xmath279 for the values of @xmath89 and @xmath53 which characterize a 2 myr cluster . ccccc m82 central 500 pc & 10 600 & 54.09 & -2.1 -2.6 & a + ngc 253 & 430@xmath280 & 53.00 & -2.2 -2.6 & b , c , d + ngc 3256 & @xmath2691400 & 52.30 53.78 & -2.3 & e + antennae & & & -1.57 & f [ table : physical_quantity ] on the basis of the theory developed above , we expect to find that individual regions ( or regions within them ) display line ratios consistent with ionization parameters ranging up to the maximum value corresponding to radiation confinement . although the same physical process limits the inferred ionization parameter for observations on galactic scales , the value of the upper limit is not the same , because of the different measures used : usually geometric estimates ( our @xmath69 ) in resolved regions , as opposed to pure line ratios ( our @xmath10 , or more precisely @xmath204 ) in unresolved ones . moreover , averaging over many regions tends to blend together those at @xmath114 with those below it , and the selective dimming of high-@xmath10 regions by internal dust absorption favors those below the maximum . we estimate the upper limit of @xmath69 to be roughly @xmath109 using the solutions of appendix [ appendix : planarregions ] , by replacing @xmath115 with the emission - weighted density @xmath25 . this is only approximate , as it does not account for the formation of the lines used to derive @xmath115 . for unresolved regions we apply @xmath114 , in which we do account for line formation : because most of the observations employ the the [ neiii]/[neii ] line ratio of figure [ fig : ne3ne2_2myr ] , we apply the upper limit of @xmath117 we found there . part of the difference between the two upper limits can be seen in the internal structure of radiation - confined zones , in which @xmath281 is about 0.3 dex higher than the recombination - weighted average @xmath119 . the rest of the difference is presumably due to the details of line formation . to test these expectations against observed regions , we plot in figure [ fig : uexamples ] a sample of observationally - derived ionization parameters from the literature , divided according to the scope of the observations involved and according to the method used . as expected , the data range up to values which are approximately our estimated maximum value for each method . for individually - resolved regions , the observed upper limit appears to be somewhat higher than our prediction , but we attribute this to the approximate nature of our @xmath282 ( rather than to magnetic fields , say ) . for galactic - scale observations highest @xmath10 is slightly lower than our @xmath114 , which could be a product of the averaging process . these results in hand , we turn to the importance of wind pressure in regions . for a resolved , spherical region with well - constrained observations of @xmath16 , @xmath283 , and @xmath10 , it is possible to estimate the wind parameter @xmath79 by reference to a plot like our figure [ fig : ne3ne2_2myr ] . with only @xmath10 or @xmath69 , one obtains an upper limit on @xmath79 . by this logic , we can state quite firmly that the wind parameter @xmath79 is often small in resolved regions : otherwise , our example values of @xmath10 in the upper panel of figure [ fig : uexamples ] would be much less than the theoretical maximum . the same argument applies , if somewhat more weakly , to entire galaxies : for them , @xmath284 ranges up to about @xmath285 . this is far higher than what would be possible if they were adiabatic wind bubbles ( eq . [ ugeominweaver77 ] ) unless they are exceptionally weak . this can be seen , for instance , in figure [ fig : ne3ne2_2myr ] , where we use the values of @xmath204 and the product of _ total _ @xmath16 and inferred @xmath286 to place several starburst galaxies within the parameter space of individual regions . because this is an upper limit on the actual values of @xmath271 within each galaxy s regions , and because @xmath204 can not be higher than the typical values of @xmath10 in those regions , we infer that the characteristic values of @xmath287 can not exceed about @xmath288 . while wind - dominated bubbles are likely to exist , we find no evidence for them either in individual regions or on galactic scales . we conclude from this that the leakage of wind energy past the if , either by advection or radiation , is significant . our primary motivation in this study has been to assess whether the ionization parameter can be used to measure the importance of radiation pressure in the inflation of regions in external galaxies , especially starburst galaxies . we have been interested in how well the observational estimate @xmath10 reflects the radiation - to - gas pressure ratio @xmath155 in local conditions and its average over galactic scales . because @xmath289 in each parcel of gas , and because @xmath0 controls the ionization state and therefore the emitted spectrum , it is possible , in principle , to derive the role of photon momentum in the kinematic feedback from young star clusters into the nearby dense neutral gas . ( although supernovae tend to inject more energy , they are thought to couple very poorly to dense , star - forming gas ; @xcite . ) because emission line ratios reflect local conditions , this information would be complementary to the global analysis of @xcite . to this end we explored the properties of strongly radiation - confined layers using the analytical theory of dr11 ( appendix [ appendix : planarregions ] ) , considered a host of physical effects and scenarios which could affect @xmath10 in real regions ( [ s : dynamics ] and [ ss : dust_abs ] ) , and engaged in a suite of numerical models of dusty regions in quasi - static force balance with an interior pressurized wind bubble ( [ s : model ] ) . our models , though not meant to probe the full parameter space of regions , are the first to explore a wide range of both the radiation force parameter ( @xmath169 or @xmath42 ) and the wind pressure parameter ( @xmath79 ) . two effects complicate our plan to measure radiation pressure feedback using the apparent ionization parameter . first is the fact that strongly radiation - dominated regions compress their ionized gas into thin layers whose internal pressure gradient balances the radiation force @xcite . the ionization parameter recorded in any line ratio reflects the ratio of radiation to gas pressure where the lines are produced , and in radiation - confined regions @xmath10 saturates at a maximum value . the resulting values of @xmath290 depend on the lines in question ( [ s : model ] ) and parameters such as the dustiness of the region and the ionizing spectrum , but are roughly @xmath291 , the value for which @xmath292 . for the particular neon line ratio plotted in figure [ fig : ne3ne2_2myr ] , the upper limit is actually @xmath293 . only significant levels of ram pressure or magnetic pressure , or serious departures from force balance , allow for @xmath10 to exceed this putative maximum ( [ ss : magnetic]-[inertialconfinement ] ) , and this point may be useful in diagnosing the physical states of unresolved regions . conversely any significant pressure on its inner boundary , such as the pressure due to shocked stellar winds , reduces @xmath155 and @xmath294 ( as well as the dust optical depth @xmath87 ) below their maximal values ( [ ss : winds ] ) . our limited compilation of ionization parameters in individual regions ( figure [ fig : uexamples ] , top panels ) is consistent with the proposition that @xmath10 varies up to the maximum imposed by radiation confinement ; for instance , @xcite find @xmath295 in individual regions in the antennae galaxies . another complicating effect is the suppression of the apparent ionization parameter on galactic scales by selective dust absorption of ionizing photons in the regions where @xmath0 is highest . gas in the local universe is sufficiently dusty ( @xmath296 ) that the dust optical depth @xmath87 exceeds the radiation - to - gas pressure ratio @xmath155 , both of which vary proportionally to @xmath0 : all radiation - confined regions ( and some others ) are dust - limited . such dust - limited regions convert a large portion of the stellar ultraviolet radiation into far infrared continuum , limiting the efficiency @xmath86 of line production . for observations which span a wide range of local conditions , this biases the inferred estimate of @xmath0 ( which we call @xmath204 in this case ) toward regions of lower @xmath297 and @xmath10 . within the toy model for galactic region populations we present in [ sss : int_abs ] , it appears this selection effect suppresses @xmath204 by a small amount ( @xmath2980.05 dex ) relative to an unbiased ( i.e. , starlight - weighted ) average of @xmath10 , primarily because the line efficiency @xmath86 gradually declines by only 0.5 dex ( to a minimum of about 0.35 ) as radiation confinement becomes strong . although this result depends on the assumptions in our toy model , it strongly suggests that @xmath204 can be corrected for selective absorption and used to infer the physical characteristics of the galactic region population . chief among these would be the ratio of radiation and gas - pressure forces averaged over a galaxy s ionized zones , a quantity we plot ( for the toy model ) in the top panel of figure [ fig : uobsgalmodel ] . it has long been appreciated that the dust - bounded state contributes to the suppression of fine - structure cooling lines relative to the far - infrared continuum in ulirgs @xcite , and that recombination luminosity under - represents the ionizing luminosity in this state ; these effects provide observational checks on the degree of dust saturation one might infer from recombination line ratios . the galaxies collected in the bottom panel of figure [ fig : uexamples ] display an upper limit of @xmath299 , with some uncertainty . this is very close to the upper limit of @xmath293 we found for @xmath10 derived from a [ neiii]/[neii ] line ratio in young regions ( [ s : model ] ) , especially considering the bias induced by dust attenuation . we infer , therefore , that radiation pressure strongly modifies the ionized zones in these galaxies . a result of dust - induced bias is that the upper limit of @xmath204 will be weakly anti - correlated with metallicity . to test this , one must disentangle variations in @xmath0 from variations in the incident spectrum due to changes in stellar metallicity , and possible influences of metallicity on the stellar initial mass function ( and more subtly , on stellar rotation and multiplicity ) . moreover , our analysis of the radiation - confined state ( appendix [ appendix : planarregions ] ) suggests a change in the properties of radiation - confined regions with metallicity above and below about @xmath300 , , @xmath301 , and @xmath302 . ] corresponding to @xmath303 . for higher values of @xmath89 , as we have already mentioned , dust absorption is significant in all radiation - confined zones . further , for @xmath106 radiation - confined zones are compact ( the density scale height is much smaller than the radius ) and the lyman-@xmath15 line pressure is negligible . none of these is true for @xmath304 : regions with significant radiation pressure extend inward toward the central source , are supported by line pressure as well as gas pressure , and are not significantly attenuated by dust . moreover , the wind pressure will typically be lower on account of the lower opacity of stellar atmospheres . we leave a full investigation of this transition to future research , but point out that many works on photo - ionization in active galactic nuclei have been carried out in the limit @xmath305 . we firmly rule out the hypothesis that resolved regions are predominantly the ionized shells of energy - conserving wind bubbles ( as envisioned for giant regions by @xcite ) . the pressure of shocked winds in such models lowers @xmath0 well below observed values . although this discrepancy can be remedied by decimating the wind luminosity @xcite , it is more natural to attribute the reduction in wind pressure to leakage of hot gas beyond the ionization front , either via direct outflow or radiative losses . it is preferable to employ models which explicitly account for such leakage ( * ? ? ? such models form the basis for populations of regions in the toy model of [ sss : int_abs ] and in a future companion paper ( verdolini et al . in prep . ) , in which line emission by star - forming galaxies will be reconsidered in greater detail than is possible here . it is possible to interpret entire galaxies in terms of models of individual regions , as we discussed in [ s : implication ] . this must be done with caution , because observations on galactic scales must necessarily average over a wide range of conditions , and are biassed by dust absorption . acknowledging this , one can nevertheless derive an effective upper limit on the wind pressurization @xmath79 from the fact that a larger value of this parameter would limit @xmath204 below what is observed . for observational and theoretical studies of the ionization parameter , an important point is that the method used to infer @xmath10 affects the result , leading to systematic offsets between methods . traditionally line ratios are compared with suites of simulations in which @xmath10 represents the interior of an ionized layer ( often one of uniform density ) , a definition which corresponds to our @xmath69 so long as the layer is thin . this definition encounters a couple difficulties . first , the incident flux is attenuated by recombination and dust within the layer so that the characteristic value of @xmath0 is lower than the incident value ( e.g. , equation [ [ ugeominweaver77 ] ] ) . second , ionized gas can only be confined to a thin layer by an inner , pressurized region of hot gas , or ( when @xmath106 ) by strong radiation pressure ; however , radiation confinement introduces a pressure and density gradient within the confined layer , and wind confinement limits the maximum values of @xmath155 and @xmath0 well below unity and 0.1 , respectively . for these reasons , uniform - slab calculations are physically inconsistent at high values of @xmath0 . it is preferable to use suites of simulations involving spherical regions in quasi - static force balance , but a single sequence of these excludes the role of a wind pressurization parameter ( our @xmath79 ) . as an alternate standard , we advocate using , for @xmath10 , the value of @xmath0 within a uniform mixture of gas photons for which the emitted line ratios match those observed . we thank the anonymous referee for positive feedback , and c.f . mckee and b.t . draine for insightful comments which helped to clarify the paper . sccy thanks p.g . martin , g.j . ferland , k. blagrave , j. fischera , and p. van hoof for helpful discussions regarding photoionization codes . cdm thanks a. bressan for an illuminating conversation . this work is supported by a national sciences and engineering research council of canada ( nserc ) grant , and a university of toronto fellowship . the momentum equation for fluid flow is @xmath306 where @xmath307 is the unit tensor and @xmath308 and @xmath309 are the magnetic and radiation force per unit volume , respectively . as elsewhere in the paper , @xmath310 ( with no subscript ) is the gas pressure . our analysis is based on a quasi - static idealization of an region in which @xmath311 and @xmath312 , so this simplifies to a statement of force balance : @xmath313 as discussed by @xcite , @xmath314 where @xmath315 is the radiation pressure tensor . in spherical symmetry , @xmath316 for radiation pressure @xmath317 and energy density @xmath318 . splitting the radiation into radially free - streaming and isotropic components @xmath319 with energy densities @xmath320 , we have @xmath321 . our free - streaming radiation is meant to represent attenuated radiation from the central source , whose luminosity at @xmath322 is @xmath323 so that @xmath324 , whereas the isotropic radiation field represents trapped line emission , especially lyman @xmath15 , as well as the diffuse lyman continuum radiation emitted during recombinations to the ground state . inserting this and @xmath325 into the force balance equation and writing only the radial component , @xmath326 writing @xmath327 and integrating across a range @xmath328 , @xmath329 this shows that the change in @xmath330 is bounded across any range of @xmath322 . since @xmath323 is non - increasing , the magnitude of the first integral is maximized if radiation and matter do not interact , so that @xmath323 remains constant : @xmath331 where the arrow represents the limit @xmath332 . moreover , the magnetic tension term @xmath333 is of order @xmath334 if @xmath164 is the curvature radius of the magnetic field . when ionized gas is compressed in a thin shell of width @xmath335 , @xmath164 likely to be intermediate between @xmath335 and @xmath322 . we can therefore say in general that when the shell is thin , @xmath336 which is typically quite small : force balance implies near - constancy of the total pressure . we are interested in the properties of the dr11 solutions in the limit that ionized gas is restricted to a geometrically thin shell , as occurs in radiation - dominated regions when @xmath337 and ( as we see below ) @xmath106 . we restrict equations ( 1)-(3 ) of dr11 to the limit of a thin shell ( constant @xmath322 ) and combine them to find @xmath338 ( we use @xmath339 for the local dust optical depth , reserving @xmath87 for that of the entire region ) and @xmath340 where @xmath341 for a characteristic density @xmath342 which is defined so that , when @xmath343 , dust absorption and ionization consume ionizing photons at equal rates . the solution with @xmath344 and @xmath345 at @xmath346 is @xmath347 - 1 \over \gamma-1}\ ] ] and @xmath348,\ ] ] valid up to @xmath349 , for which @xmath350 , so that @xmath351 for @xmath352 , which shows that when @xmath353 and @xmath354 , @xmath355 $ ] . is @xmath356 , where @xmath357 . ] because of the planar geometry ( appendix [ appendix : forcebalance ] ) these solutions obey pressure equilibrium , @xmath358 , or in dimensionless terms , @xmath359 so that , at the if , @xmath360/\gamma$ ] . within these solutions , @xmath363 and the fraction of ionizing starlight caught by gas is @xmath364\\ & \simeq & 1-\exp[0.7(\ln \tau_d - 1 ) - ( \ln \tau_d)^2/6],\end{aligned}\ ] ] where the final approximation is correct to 2% for the entire range of figure [ taudxifion ] . in the very dusty limit @xmath365 , for which @xmath366 , the recombination luminosity is suppressed by a constant fraction which is roughly @xmath203 . the radial coordinate can be determined by integrating @xmath367 . because of the low density at the inside of the slab , radiation force is transmitted to the gas only through the dust , leading to a constant acceleration and ( since the temperature is assumed to be constant ) an exponential interior density distribution @xmath368 $ ] . note , however , that equation [ nhat_planar ] implies a negative density if @xmath156 ; this simply means that the solution is incompatible with the imposed inner boundary of zero density and optical depth . for such low dust opacity , the solution must extend inward to join the spherical solutions of dr11 , or meet an inner , pressurized region of hot gas . the difference between compact ( @xmath106 ) and extended ( @xmath156 ) radiation - confined regions ( with no inner hot gas ) is apparent in the top panels of dr11 s figure 2 . for @xmath156 , radiation - confined regions with @xmath111 are still concentrated toward the ionization front , but the scaling of radiative acceleration with the neutral fraction allows gas to permeate the interior this effect would only be accentuated by the pressure of lyman @xmath15 line photons , which becomes significant when dust is scarce @xcite . in figure [ xiestimatesplanar ] we plot two characteristic values of the ionizing radiation - to - gas ratio @xmath166 in the radiation - confined state : the recombination - weighted average @xmath119 , and the geometrical @xmath369 . the latter is calculated using the emission - weighted density @xmath370 . estimates of @xmath10 and @xmath69 derived from these limits are only approximate , because line emission is not strictly proportional to the recombination rate . , r. , de messires , g. e. , madden , s. , engelbracht , c. , smith , j. d. , meixner , m. , brandl , b. , smith , l. j. , boulanger , f. , galliano , f. , gordon , k. , hora , j. l. , sewilo , m. , tielens , a. g. g. m. , werner , m. , & wolfire , m. g. 2009 , , 694 , 84	the ionization parameter @xmath0 is potentially useful as a tool to measure radiation pressure feedback from massive star clusters , as it directly reflects the ratio of radiation to gas pressure and is readily derived from mid - infrared line ratios . we consider a number of physical effects which combine to determine the apparent value of @xmath0 in observations encompassing one or many regions . an upper limit is set by the compression of gas by radiation pressure , when this is important . the pressure of shocked stellar winds and the presence of neutral clumps both tend to reduce @xmath0 for a given intesity of irradiation . the most intensely irradiated regions are selectively dimmed by internal dust absorption of ionizing photons , leading to a bias for observations on galactic scales . we explore these effects in analytical and numerical models for dusty regions and use them to interpret previous observational results . we find that radiation pressure confinement sets the upper limit @xmath1 seen in individual regions . unresolved starbursts are known to display a maximum value of @xmath2 . while lower , this is also consistent with a large portion of their regions being radiation pressure dominated , given the different technique used to interpret unresolved regions , and given the bias caused by dust absorption . we infer that many individual , strongly illuminated regions can not be significantly overpressured by stellar winds , and that even when averaged on galactic scales , the shocked wind pressure can not be large compared to radiation pressure . therefore , most regions can not be adiabatic wind bubbles . our models imply a metallicity dependence in the physical structure and dust attenuation of radiation - dominated regions , both of which should vary strongly across a critical metallicity of about one - twentieth solar .
hadronic systems produced in annihilation have their origin in a uniquely simple quark - antiquark state . while the standard model of particle physics provides a well tested description of the reaction , the subsequent production of observable hadrons is less well understood . a parton shower , described by perturbative quantum chromodynamics ( pqcd ) , is normally invoked to describe the initial fragmentation phase . in the subsequent non - perturbative hadronization process , the partons become the hadrons which are experimentally observed . multihadronic annihilation events , with no beam or target fragments to confuse their experimental or theoretical interpretation , provide the most powerful system available for the study of the transition from the partons of perturbative qcd to the hadrons of the laboratory . between 1989 and 1995 , the large electron positron collider ( lep ) at cern delivered an integrated luminosity of about 170 pb@xmath2 per detector at and around the z@xmath0 peak , providing each of its four dedicated experiments , aleph , delphi , l3 and opal , with some 6 million e@xmath3e@xmath4 annihilation events , 70% of which were multihadronic . these events have enabled the experiments to conduct detailed studies of many aspects of parton hadronization . over the same period , the slac linear collider ( slc ) delivered a more modest number of events , initially to the mark ii detector and then to the sld detector , which has now accumulated about 200k z@xmath0 events . although its luminosity is lower than that of lep , the slc is able to provide highly polarized electron beams . while lep has now entered phase 2 , running at higher energies , analysis of the z@xmath0 peak data will continue for some time . the slc , with polarized beams , is scheduled to run at the z@xmath0 for several more years . in this review , we look at the present understanding of the non - perturbative hadronization process in the context of the recent experimental results from lep and slc . a brief introduction to the electroweak theory of annihilation is followed by a heuristic picture of the transition from perturbative partons to final - state hadrons . after this , the role of perturbative qcd in determining the structures and particle content of events is discussed . models of the non - perturbative hadronization phase are then covered independent fragmentation , string fragmentation and cluster fragmentation are discussed in detail and a comparison is given of the main features of the different models . the lep and slc machines and their associated detectors are then described , with emphasis on the features relevant to measurements of individual hadrons from z@xmath0 decay . next the experimental data on single inclusive identified particles are summarised , and the results interpreted in the light of the different models . results on spin phenomena , such as baryon polarization and vector meson spin alignment are then covered , and this is followed by discussion of results on correlations , including bose einstein effects , strange particle rapidity and angular correlations , and intermittency ; where possible the results are interpreted within the context of theory and models . differences between quark and gluon initiated jets are then dealt with , and the review ends with a look forward to what may still be learned from the available lep and slc data , and the forthcoming slc data . the ( initial ) numbers , directions , polarizations and flavours of the quarks produced in annihilation are determined by the electroweak couplings of the exchanged vector bosons . the full , tree - level expression for the differential cross section for in the centre of momentum ( com ) frame is given as a function of quark s polar angle , @xmath5 , measured relative to the electron beam direction , and total com energy squared , @xmath6 , by : @xmath7 \nonumber \\ \fl\hspace{3mm}+\bigl({\cal p}^{(1)}_{\rm e}a^{ij}_{\rm e } + { \cal p}^{(2)}_{\rm e}\bigr ) \bigl({\cal p}^{(1)}_{\rm q}a^{ij}_{\rm q } + { \cal p}^{(2)}_{\rm q}\bigr)2\cos\theta^\star\biggr\}\end{aligned}\ ] ] where , @xmath8 the summation is over all exchanged bosons , @xmath9z@xmath10 , with mass and width @xmath11 and @xmath12 , and vector and axial couplings @xmath13 and @xmath14 . for example @xmath15 , @xmath16 and @xmath17 etc , with @xmath18 the fermion s electric charge , normalized to that of the electron , @xmath19 the fermion s third component of weak , su(2)@xmath20 , isospin and @xmath21 the sine(cosine ) of the weinberg angle . the number of colours is @xmath22 and @xmath23 is the electromagnetic fine structure constant . the dependence on the initial lepton and final quark longitudinal polarizations are via @xmath24 , where @xmath25 for spin along the com direction of travel ( helicity basis ) . in the absence of transverse beam polarization there is no azimuthal angular dependence . at @xmath26 the z@xmath0 exchange term dominates and simplifies considerably . in this limit the partial width for is given by : @xmath27\ ] ] this implies relative hadronic branching ratios of @xmath28 to each up - type quark and @xmath29 to each down - type quark at the z@xmath0 ; the relative fraction of b quarks peaks on resonance . shows the relative fractions of light ( u , d , s ) , charm and bottom quarks as a function of @xmath30 obtained using the full expression . the total hadronic branching fraction is @xmath31 . the beam polarization has little influence on the relative rates of produced quarks , although the total rate is proportional to @xmath32 . however only a single polarized beam can significantly alter the quark polar angle distribution , as can be seen from the forward - backward asymmetry ( for pure z@xmath0 exchange and @xmath33 ) : @xmath34 since @xmath35 is only @xmath36 and @xmath37 ) then , given polarized beams , one has a powerful statistical way to identify separately quark and antiquark jets based on their direction . the quarks are naturally produced highly polarized : @xmath38 this result is for unpolarized lepton beams , pure z@xmath0 exchange and @xmath33 ; the full expression depends little on the lepton beam polarization . for qcd studies at the z@xmath0 @xcite the effect of initial - state electromagnetic radiation ( isr ) is of rather minor significance , unlike in the case of electroweak studies @xcite . the primary effect of isr is to lower the effective @xmath30 , leading to a distortion of the breit - wigner lineshape given by . however the basic quantities of relevance from a qcd perspective , quark flavour mix , polarization etc , are only rather weak functions of @xmath30 . further , because of the z@xmath0 resonance the cross section falls rapidly when isr occurs , mitigating against its effects ( even so the peak cross section falls by @xmath39 compared to ) , in contrast to the situation at lep 2 and higher energies where isr is very important @xcite . the effects of _ final - state _ radiation , particularly of gluons but also of photons , form one theme of this review . the effect of final - state radiation on the total hadronic cross section can be summarized in the multiplicative factor @xcite : @xmath40 where @xmath41 is the strong coupling constant @xmath42 and @xmath43 is a measure of the quark - gluon coupling strength . flavour ( mass ) dependent effects and electroweak radiative corrections are small both occur at the 1% level , and need not concern us here @xcite . the description of a multihadronic event commences by specifying a set of primary partons ( q//g ) distributed according to an exact z@xmath0 decay matrix element , such as . it is customary to identify three basic stages in the transition of these partons into detected hadrons @xcite . first a parton ( or , equivalently , dipole ) cascade , formulated according to pqcd , evolves the primary partons from the hard scattering scale @xmath44 into secondary partons at a ( fixed ) cut - off scale @xmath45 gev . it is during the calculable stages , hard subprocess and shower , that the event s global features are determined : energy dependences , event shapes , multiplicity etc . in a second stage , carried out at the fixed , low virtuality scale @xmath46 , a model is employed to convert the secondary partons into hadrons . the second stage is essentially energy @xmath47 independent , up to power corrections , and assumed to be local in nature . finally unstable primary particles are decayed into stable hadrons and leptons according to decay tables @xcite . schematically the fragmentation function is given by : @xmath48 here , @xmath49 is the probability to find a hadron of type h carrying a fraction @xmath50 of the parton s momentum , in a jet initiated by the parton @xmath51 , whose maximum virtuality is @xmath52 . the h represent possible intermediate hadrons . ( 140,50 ) ( 70,20 ) ( 68.5,20)(-1,0)48.5 ( 71.5,20)(1,0)33.5 ( 105,20)(4,-1)35 ( 111,20)@xmath53 ( 20,20)(4,20 ) ( 18,20)(4,20)[l ] ( 18,30)(1,0)2 ( 18,10)(1,0)2 ( 4,26)(2,8 ) ( 2,26)(2,8)[l ] ( 2,30)(1,0)2 ( 2,22)(1,0)2 ( 4,14)(2,8 ) ( 2,14)(2,8)[l ] ( 2,18)(1,0)2 ( 2,10)(1,0)2 ( 4,5)(-1,0)0 ( 4,5)(1,0)16 ( 0,0)@xmath54 fm ( 30,30)(10,0)3 ( 35,25)_gluers _ ( 27,14)@xmath55 fm ( 25,15)(-1,0)5 ( 64,15)(1,0)5 ( 75,10)@xmath56 ( 71,15)(-1,0)0 ( 71,15)(1,0)34 ( 121,23)@xmath57 fm ( 120,19)(0,-1)0 ( 120,19)(0,1)10 ( 100,0)@xmath58 ( 105,5)(-1,0)0 ( 105,5)(1,0)15 ( 57,35)@xmath59 fm ( 63.5,32)(1,0)5 ( 76.5,32)(-1,0)5 ( 81,21)@xmath60 ( 65,0)@xmath61 ( 68.75,20 ) ( 1000,1000 ) ( 0,400)[2 ] ( , ) ( 0,-800)[4 ] ( , ) ( 63.5,20 ) ( 1000,1000 ) ( 0,0)[2 ] ( , ) ( 73.5,20 ) ( 1000,1000 ) ( 0,0)[2 ] ( , ) ( 103.5,20 ) ( 1000,1000 ) ( 0,0)[5 ] ( , ) by 200 by -500 ( , ) @xmath62 we consider this process in terms of its space - time structure : the discussion is based on @xcite and illustrated in figure [ fxt ] . the hard subprocess , , may be viewed as the production of a highly virtual photon , or a real z@xmath0 , which impulsively kicks a pair out of the vacuum ; the time scale is short , @xmath63 fm . in this non - adiabatic process the quarks shake off most of their cloud of virtual particles , so that any structure they have is on a scale below @xmath64 . that is they behave as bare ( more properly half - dressed ) colour charges until the gluon field has had time to regenerate out to a typical hadron size @xmath65 fm . allowing for the boost , this takes a time @xmath66 fm , where the second approximation is appropriate to light hadrons . the fact that @xmath67 raises the issue of how charges are conserved over the space - like separated distances involved of course the accelerated quarks will radiate gluons and here two new time scales are relevant . first , from the off - shellness of the quark prior to emission , the formation time of a real gluon of 3-momentum ( = energy ) @xmath68 is @xmath69 ( the uncertainty relation gives the proper lifetime of the virtual state as the reciprocal of its off - shellness , during which time it will travel a 4-distance @xmath70 in the laboratory frame , where @xmath71 is the virtual state s 4-momentum ) . second , for the gluon to reach a transverse separation of @xmath72 and become independent of the quark takes a time @xmath73 , whilst the hadronization time may be written @xmath74 . for this quark - gluon picture to make sense we require @xmath75 so that @xmath76 ; this is natural as it implies that @xmath77 , making perturbation theory applicable . if @xmath78 then we can say nothing . on the borderline are quanta with @xmath79 ; these feel the strong interaction and are responsible for holding hadrons together . we distinguish these from the essentially free perturbative gluons by the name _ gluers_. the first _ gluers _ form after only 1 fm , having @xmath80 . thus _ gluers _ immediately form in the wake of the primary partons blanching their colour field and leaving two fast separating charge neutral systems . thus the slowest hadrons form first , close to the interaction point , in what may be called an ` inside - out ' pattern @xcite . on quite general grounds the distribution of _ gluers _ in qcd can be estimated using : @xmath81\times\frac{\d k}{k}\propto\frac{\d k}{k}\ ] ] here the logarithmic @xmath82 dependence reflects the fact that the coupling constant is dimensionless , the logarithmic @xmath68 dependence follows from the gluon being massless and only the kernel function @xmath83 , which is of order unity , depends on details of the quark - gluon vertex . the distribution is thus governed by longitudinal phase space and leads to hadron production on rapidity plateaux along the directions of the initial partons . not all quanta are emitted at low @xmath84 and in a significant range , @xmath85 , a shower of perturbative gluons is possible . a hard gluon emitted at an angle @xmath53 becomes a separate colour source at @xmath86 , again raising the issue of charge conservation . fortunately a _ gluer _ emitted at the same angle @xmath53 would appear just in the right place and at the right time to blanch the tail of the separating gluon s colour field . this gluon then acts as a new source of _ gluers _ , restricted such that @xmath87 . that is equivalent to a boosted ( @xmath88 ) quark jet of a reduced scale @xmath89 and with the substitution @xmath90 for the parton s charge in @xcite . it should be noted that _ gluers _ emitted at larger angles @xmath91 have @xmath92 and should therefore be associated with the parent q+g system . for example at time @xmath72 the first _ gluers _ emitted actually see ensembles of partons equivalent to the original q@xmath93 pair . the coherence of soft emissions simply reflects colour charge conservation and is fundamental to a gauge theory such as qcd . finally , the time scales for the decay of the primary particles are set by the reciprocal of their widths : for the strong resonances typical values of @xmath94 mev give @xmath95 fm . clearly these scales are commensurate with those for primary hadron production so that any distinction between hadronization and resonance decays may be only semantics . this is illustrated in which shows a string - inspired space - time picture ( see ) of two equivalent ways to form a @xmath96 pair differing only in the time sequence of the string breaks . in the first figure the pions are directly produced whilst in the second they appear to come from a @xmath97 ; however observationally they are indistinguishable . the time scale for the weak decays of s , c and b hadrons is of order @xmath98 ( where @xmath99 is the hadron s @xmath100 factor ) ; this is @xmath101 fm for c and b and much longer for s , and so these decays may safely be treated as separate subprocesses . ( 125,60 ) ( 5,5)(1,0)10 ( 0,10)@xmath102 ( 5,5)(0,1)10 ( 10,0)@xmath50 ( 35,5)(-1,1)25 ( 35,5)(1,1)25 ( 25,25)(-1,1)5 ( 25,25)(1,1)20 ( 45,45)(-1,1)6 ( 39,51)(1,1)8 ( 35,41)(1,1)14 ( 49,55)(-1,1)4 ( 45,25)(1,1)5 ( 45,25)(-1,1)20 ( 25,45)(1,1)6 ( 31,51)(-1,1)8 ( 35,41)(-1,1)14 ( 21,55)(1,1)4 ( 95,5)(-1,1)25 ( 95,5)(1,1)25 ( 85,25)(-1,1)5 ( 85,25)(1,1)6 ( 91,31)(-1,1)14 ( 77,45)(1,1)4 ( 95,21)(-1,1)14 ( 81,35)(1,1)6 ( 87,41)(-1,1)8 ( 105,25)(1,1)5 ( 105,25)(-1,1)6 ( 99,31)(1,1)14 ( 113,45)(-1,1)4 ( 95,21)(1,1)14 ( 109,35)(-1,1)6 ( 103,41)(1,1)8 a typical hadronic final state in z@xmath0 decay contains about 20 stable charged particles and 10 neutral hadrons , mainly @xmath103 s . this is a complex , non - perturbative system beyond direct , first principles calculation at the present time . however monte carlo event generator programs @xcite have been developed which provide remarkably accurate , detailed descriptions of complete hadronic events . these are based on a combination of judicious approximations to pqcd , discussed in , and per / in - spirational models for hadronization , to be discussed more fully in . here we list the main features of the major event generators used in z@xmath0 studies and refer the reader to later sections and to the literature for more complete descriptions . ariadne : : @xcite implements a pqcd shower based on the dipole cascade model @xcite , which is equivalent to a coherent parton shower . the evolving chain of dipoles corresponds naturally to a string and indeed the hadronization is performed using the jetset implementation of the lund string model @xcite . cojets : : @xcite is based on a virtuality - ordered parton shower , that takes no account of colour coherence . the resulting jets are hadronized using a refined version of the field - feynman independent fragmenation model @xcite . herwig : : @xcite is based on a highly developed parton shower algorithm @xcite , that automatically takes into account colour coherence , and a relatively simple cluster hadronization scheme @xcite . jetset : : @xcite uses a virtuality - ordered parton shower with an imposed angular ordering constraint to take into account colour coherence . a relatively sophisticated hadronization scheme based on the lund string model is provided @xcite . an option also exists which does not employ a parton shower but tries to use instead higher - order hard - subprocess matrix elements plus string hadronization . as stated in the bulk properties of hadronic events in z@xmath0 decay are established early in the fragmentation when virtualities are large and qcd perturbation theory is valid . it is an important issue to establish to what extent pqcd dominates and what contributions are made by non - perturbative effects . two basic approaches are used to calculate hadronic event properties . the use of fixed - order perturbation theory is justified by the smallness of the running strong coupling constant @xcite at @xmath104 . known results include the total hadronic cross section to three loops , order-@xmath105 @xcite . complete one - loop , order-@xmath106 calculations are available @xcite for planar ( i.e. 3-jet dominated ) event shapes , including contributions due to the orientation of the event plane with respect to the beam direction @xcite and to quark mass effects @xcite . partial order-@xmath107 results are known for 4-jet distributions @xcite . tree - level calculations , up to order-@xmath105 , of 5-jet distributions are also available @xcite . however a complication arises in this approach because tree - level diagrams diverge whenever external partons become soft or collinear and related negative divergences arise in virtual ( loop ) diagrams ( in addition to ultraviolet divergences ) . fortunately , in sufficiently inclusive measurements such as the total hadronic cross section it is guaranteed that the two sets of divergences cancel @xcite . unfortunately in more exclusive quantities such as event shapes the cancellation is no longer complete and large logarithmic terms remain , generically of the form @xmath108 . since @xmath109 is of order unity this can spoil the convergence of naive perturbation theory . in the second approach the perturbation series is rearranged in terms of powers of @xmath109 : @xmath110 . the first infinite set of terms represent the leading logarithm approximation ( lla ) , then the @xmath41-suppressed next - to - lla and so on . many of these terms have been identified @xcite and summed using renormalization group techniques . they are conveniently expressed via @xmath111-dependent fragmentation functions whose evolution is controlled by altarelli - parisi type equations @xcite : @xmath112 here , the so - called splitting functions , @xmath113 , may be thought of as giving the probability of finding parton @xmath114 ( and @xmath115 ) inside @xmath51 and carrying a fraction @xmath116 of its momentum . monte carlo event generators implement solutions to these equations as parton showers @xcite . more recently the resummed results for a number of event shapes have been calculated @xcite and combined with the fixed order approach @xcite . the durham jet clustering algorithm @xcite for example was proposed to allow such a resummation for jet rates . at very low @xmath30 little structure is present in hadronic events , beyond that due to the presence of hadronic resonances , and these can be described essentially by isotropic phase space @xcite . this is because the momentum scales involved in the hadronization , typically 200 mev , are comparable with those of the proto - jets . at higher energy , due to the preferred collinear nature of gluon radiation , the hadrons clearly form collimated pairs of back - to - back jets with angular distributions compatible with those of the primary q@xmath117 pairs @xcite . around @xmath118 gev a significant fraction of events deviate from a simple linear configuration : this was interpreted as first evidence for gluons @xcite . at the z@xmath0 , hard , non - collinear , gluon radiation is manifested in multi - jet event structures @xcite . event shape variables have been developed to be sensitive to the amount of acollinearity and hence the presence of hard gluon radiation @xcite . since theoretical predictions are at parton level @xcite , hadronization corrections , derived from monte carlo models , must be applied before making comparisons to data . away from phase space boundaries the corrections are modest , but not negligible ( of order 10% ) , and order-@xmath106 calculations generally describe the data well @xcite , particularly so when resummed @xcite . in the case of planar variables such as thrust , transverse momentum in the event plane , wide jet broadening etc , this allows @xmath41 to be measured , and good consistency is seen with other measurements @xcite . only a leading - order description of aplanar , four - parton , event shapes such as the @xmath119-parameter , transverse momentum out of the event plane @xmath120 , the four - jet rate etc , are presently available , although partial results are available @xcite in higher order . here agreement between data and monte carlo is less satisfactory , with 30% discrepancies occurring for @xmath120 and the 4-jet rate @xcite . hopefully the situation will be improved by matching the showers to exact fixed - order results . this will prove important because of a reliance on monte carlo models in establishing variables to discriminate at higher energies between continuum qcd events and other physics signals , such as w@xmath3w@xmath4 pairs . this is in addition to their importance in qcd and in increasingly demanding electroweak measurements such as those involving jet charge @xcite . establishing agreement between the predictions of event generator programs and data for event shape variables has proved very important for confirming our ability to model qcd as a whole and has led to a first generation tuning of the monte carlo models @xcite . a second generation of tunings , which also take into account data on identified particle production , are now becoming available @xcite . this tuning is not a trivial exercise as it is often the case that a distribution depends in a complex way on a model s free parameters . however some specific sensitivities have been identified . the 3-jet rate is very sensitive to the qcd scale parameter @xmath121 , indicating the dominance of the shower and pqcd . event shapes and inclusive momentum spectra are sensitive to the shower cut - off @xmath46 , and also to parameters controlling the generation of transverse momentum in the hadronization . in programs that lack a shower stage it is necessary to retune the hadronization model at each @xmath30 since this must describe the whole fragmentation process including the perturbative @xmath30 dependence found in a shower . as noted above , hadronization corrections , which at present are inherently model dependent , need to be applied to partonic predictions before comparison can be made to hadronic event properties . empirically the differences between quantities measured at the hadron level and the partonic predictions are found to be power behaved : @xmath122 . for example , in the case of massless quarks , @xmath123 @xcite , where @xmath124 is the usual ratio of hadronic to @xmath125 pair cross sections ; in fact this result follows from the operator product expansion ( ope ) @xcite and can be related to the value of the gluon condensate : @xmath126 . in the case of average event shape variables , a @xmath127 behaviour is found ( the ope does not apply in this case , due to the presence of multiple scales , and indeed no corresponding dimension-1 local operator exists ) . this behaviour may be established by going to the power - enhanced , low @xmath52 , region where a @xmath128 variation is easily distinguished from the slow logarithmic variation of the perturbative expressions . such a @xmath127 behaviour is significant because @xmath129 ; thus to be able to take advantage of a future @xmath130 prediction will require a better understanding of hadronization . three sources of power corrections are known : instantons , infrared ( ir ) renormalons and ultraviolet renormalons @xcite . only ir renormalons are believed to be relevant to the issue of hadronization corrections ; instanton effects are too highly power - suppressed . the coefficients of pqcd series in @xmath41 generically suffer from a factorial growth , @xmath131 , making the series formally divergent . a standard technique for dealing with this type of behaviour is to employ a borel transformation @xcite . a pole in the transform at @xmath132 would correspond to an @xmath133 term in the borel summed series : these poles are called renormalons . unfortunately the ` residue ' of the pole , equivalent to the coefficient of the @xmath134 term , is not calculable by this purely mathematical technique . recently a relationship has been suggested between the positions of the renormalons and the power behaviour of perturbative series in the presence of an ir ` regulator ' @xcite . the basic idea is that the ( resummed ) perturbative calculation probes regions of phase space involving low - virtuality partons where non - perturbative confinement effects should also be important . if these regions are isolated by introducing a ` cut - off ' into the perturbative calculation then requiring that any cut - off dependence is compensated by the non - perturbative hadronization correction allows the power behaviour to be determined . using this idea , together with a gluon mass @xcite , a @xmath127 correction was derived for the thrust @xmath135 , @xmath136-parameter and longitudinal cross section @xmath137 with all the coefficients proportional to a common scale : @xmath138 and @xmath139 . refinements of this calculation @xcite have attempted to relate the common scale to a fixed @xmath140 , representing an effective measure of long - range confinement forces at an inclusive level . a dispersive approach indicates that the numerator of the @xmath127 term may involve several coefficients , for example proportional to @xmath141 . the subject of power corrections is an active theoretical area and still subject to dispute . however there exists the exciting prospect of a phenomenology of power corrections . this is particularly so if the corrections to various processes can be related and so shown to be universal @xcite or could be developed for full distributions rather than average values . for example , according to the calculations above , @xmath142 would have no leading @xmath127 power correction , enabling a more accurate high-@xmath52 prediction @xcite . it is interesting to compare these speculations with the results of a simple tube model @xcite calculation . the idea is that after hadronization a parton jet is equivalent to a tube of hadrons distributed uniformly in rapidity , @xmath143 , along the jet axis with transverse mass @xmath144unit @xmath143 and length @xmath145 . thus the total energy and momentum of the jet are : @xmath146 and @xmath147 . in a two - jet event the thrust is simply given by @xmath148 , so that @xmath149 ; likewise @xmath150 and @xmath151 . fitting to data , all these estimates are consistent with @xmath152 gev . direct photons , which are unaffected by hadronization , offer an alternative way to probe the early perturbative event structure @xcite . in these events , cross sections are not simply given in terms of the quark charges but reflect the competition in showers between q@xmath153qg and q@xmath153q@xmath100 branchings in a way that is sensitive to the choice of evolution variable @xcite . here agreement with experiment has been less satisfactory @xcite , particularly for rather soft or very isolated photons ; for example @xmath154-jet+@xmath100 cross - section results @xcite disfavour the virtuality ordering employed in jetset s parton shower @xcite . this has led to some development of the monte carlo shower algorithms @xcite . a more critical test will be the rate of soft , wide - angle photons , but here one must be especially wary of non - prompt photons arising from decays of particles produced in the hadronization @xcite . long - wavelength quanta see event structures on larger scales and so are sensitive to the presence of neighbouring charged partons @xcite . this results in the effective radiating units being charge - anticharge dipoles @xcite and leads to a suppression of soft gauge quanta due to the requirement of coherence . this is true of qed and qcd but in the latter , because gluons carry colour charge , one must also allow at leading order for changes in the colour antennae formed by the hard partons . effects due to colour coherence are expected in both the perturbative and non - perturbative stages of an event @xcite . in the parton shower they can be simply incorporated by requiring angular ordering that is successive branchings are nested @xcite ; this is known as the modified leading logarithm approximation ( mlla ) . as noted earlier , in jestet angular ordering is imposed on an initially incoherent shower @xcite , in herwig it is built into the choice of evolution variable @xcite , it is intrinsic to the colour dipole model used by ariadne @xcite and it is not included in cojets @xcite . hadronization models also respect an event s colour structure in accordance with local parton hadron duality ( lphd ) @xcite , discussed below . strings may be regarded as the natural limit of an evolving chain of colour dipoles whilst clusters may be thought of as the final , colour neutral , dipoles . in all monte carlo models the treatment of both stages is based on the large @xmath155 approximation to the colour flow @xcite which gives very good agreement with known full analytic results @xcite . the classic direct test of inter - jet colour coherence is the ` string effect ' a comparison of the particle flows between the jets in g events @xcite . as the historic name suggests , the relative depletion between the and qg jets was expected to have a non - perturbative origin @xcite ; later , assuming lphd , a perturbative explanation was found @xcite . to describe the z@xmath0 data @xcite it is necessary to employ a coherent shower with an approximately equal contribution coming from the non - perturbative hadronization model ( but see @xcite ) . a second classic measurement @xcite , proposed in @xcite , is the relative depletion between the pair in g and @xmath100 events @xcite . the azimuthal angular dependence of particle flow in g events has also been studied @xcite . in order to avoid having to find jets , the energy - multiplicity - multiplicity correlation can be used to measure the inter - jet coherence @xcite , though in this case the pqcd calculation suffers large corrections @xcite . ( a related quantity , the asymmetry in the particle - particle correlation , has even been suggested @xcite as a way to measure intra - jet coherence effects @xcite . ) again all of the above studies at the z@xmath0 confirm that monte carlo models incorporating a coherent shower give the best descriptions of the data . unfortunately a test for the ` negative dipole ' , expected in g events when the large @xmath155 approximation is not used , does not appear practical @xcite . the intra - jet effect of colour coherence is to limit the production of soft particles in the parton shower , leading to a ` hump - backed plateau ' shape for the particle spectrum in the variable @xmath156 @xcite , where @xmath50 is the ratio of the hadron momentum to the beam energy . by contrast , in the incoherent tube model , based on , there is no gluon suppression and no broad peak forms . the actual calculation is of the parton ( essentially gluon ) spectrum in a jet at the cut - off scale : @xmath157 , with @xmath158 , which it is then argued also applies to the hadron spectrum . the result is a rather unwieldy expression so that often lphd is invoked , in the rather technical sense of taking the limit @xmath159 @xcite , to obtain the simpler , more fully evolved , limiting spectrum . finally in the peak region a distorted ( downward skewed , platykurtic ) gaussian function @xcite is applicable . the z@xmath0 data @xcite are qualitatively well described ; quantitatively the limiting spectrum is a little too narrow in the peak region , where the distorted gaussian fits better , although not in the tails . the coherent monte carlo event generators @xcite give the best fits ( however see @xcite ) . a particularly interesting measurement on the @xmath160 spectrum is that of the peak position , @xmath161 . this occurs at low momentum where the occurence of successive parton branchings should lessen any dependence on the primary quark flavour . ( measurements of the width , skewness , etc are statistics limited and sensitive to primary quark flavour effects in the distribution s high momentum tail ) . in pqcd the @xmath52 dependence of @xmath161 is predicted to be linear in @xmath162 whie the width grows as @xmath163 @xcite . specifically @xmath164 ( i.e. @xmath165 ) with @xmath166 in the incoherent double logarithm approximation and @xmath167 in the coherent mlla , corresponding to a harder spectrum . the z@xmath0 data when combined with lower energy results ( see @xcite ) clearly prefer @xmath167 especially when higher - order corrections are included @xcite . similar conclusions have been reached from breit frame analyses of deep inelastic scattering ep events at hera @xcite . a closely related measurement is that of two - particle momentum correlations @xcite . qualitative agreement is seen with the expectations based on the pqcd distribution of gluon pairs in a shower @xcite . however to obtain quantitative agreement large corrections to the prediction should be anticipated @xcite . finally , scaling violations are seen when comparing the fragmentation function , @xmath168 , measured at the z@xmath0 to those from lower energies @xcite , after making allowance for the varying primary quark favour mix ( see ) . as in deep inelastic scattering these variations are controlled @xcite by altarelli - parisi equations which allows @xmath41 to be determined @xcite . the integral of the momentum distribution of all particles gives the event multiplicity . soft particles make a significant contribution to this total so that again colour coherence is important . the @xmath30 dependence of the first few moments of the distribution of event multiplicities is calculable in mlla @xcite . the effect of coherence is to slow the growth of the mean multiplicity with @xmath30 @xcite as compared to the lla result @xcite . the mlla prediction for the mean works well when higher - order corrections are included @xcite , as do several more phenomenological functions of @xmath30 @xcite , including a simple statistical phase space model @xcite . however this is not the case for the incoherent cojets model . the width ( or equivalently the second binomial moment ) is larger than the data @xcite although further relatively large higher - order corrections might be anticipated from the ratio of leading to next - to - leading order predictions @xcite . at high energy the ratios of these moments are expected to become @xmath30 independent , a feature already present in the data . this would imply kno scaling @xcite which is indeed seen to work well . the ratios of moments as given by qcd behave approximately like those of the negative binomial distribution ( nbd ) @xcite ( with parameter @xmath169 ) , though empirically the discretized log normal distribution ( lnd ) @xcite is a better fit to the multiplicity distribution . multiplicity distributions have also been studied in restricted rapidity intervals @xcite where the influence of global conservation laws may be reduced . the shape of the distribution becomes narrower for small rapidity ( @xmath143 ) windows , and some structure , possibly attributable to multi - jet events @xcite , arises at intermediate @xmath143 ranges . these features are described well by jetset , but less well by herwig or the simple nbd and lnd . in addition to studies of the natural flavour mix , light and heavy quark initiated events have been investigated separately @xcite . a large quark mass reduces phase space and shields the the collinear singularity associated with forward gluon emission , causing a relative suppression of forward hadrons @xcite . in practice observation of the reduction in associated multiplicity due to this ` dead cone ' @xcite is hampered by the presence of the heavy hadron s decay tracks ( @xmath170 per b - hadron @xcite ) which actually ensure a higher multiplicity than in u , d , s events . however the difference in multiplicity is predicted @xcite and seen to be @xmath30 independent @xcite . ( a variation of 0.4 tracks is expected from the change in the heavy quark fraction between 12 ( @xmath171 ) and 91 ( z@xmath0 ) gev @xcite ; this is below the level of the experimental errors ) . implicit in the above calculations of event features , including those by monte carlo methods , is the assumption that pqcd provides the dominant contribution . this is especially true for infrared ( soft gluon ) sensitive quantities . in other words hadronization and resonance decays causes little disruption of the features already established by the cascade . in the absence of a well developed theory of hadronization this is in fact a minimum requirement for pqcd calculations to be worthwhile . in terms of the earlier space - time picture , described in subsection [ sxtpic ] , this idea of ` soft confinement ' appears very natural and it is indeed in many instances supported by data @xcite . since in this picture it is the relatively soft _ gluers _ , following in the wake of the hard quarks and gluons , which cause hadronization to occur , then except for a ` collective ' action by the _ gluers _ , large momentum transfers during hadronization are precluded . an important observation which allows this picture to be taken further is that after a coherent shower , hadronization should occur locally . this is seen by considering two separating perturbative partons , with opening angle @xmath53 , and two _ gluers _ each emitted at @xmath172 to their parent partons @xcite . at the time the two _ gluers _ simultaneously hadronize , their transverse and longitudinal separations are given by : @xmath173 here the inequalities follow from the strong angular ordering condition @xmath174 @xcite . this implies that the hadronizing _ gluers _ form only a rather low density system in configuration space . the large inter-_gluer _ separations effectively limit the influence which one hadronizing parton system can have on a neighbour . this locality hypothesis also receives strong support from the pre - confinement property of pqcd @xcite ( see ) . in this scenario the local , and essentially independent , nature of each parton s hadronization leaves little scope for long - range effects . therefore hadronization should not significantly alter an event s angular structure or its energy and multiplicity distributions or their correlations , but only provide a ` correction ' . it is beyond this approach to predict actual production rates of specific hadrons ; however it would support the notion that these rates are essentially constants , dependent on at most a few nearest neighbour partons . to a certain extent present models do respect this idea and it is the quantum numbers of a few nearby partons which determine the properties of a produced hadron . this in turn gives rise to an approximate local conservation of flavour , baryon number etc . this idea has become known as local parton - hadron duality and represents perhaps the simplest working hypothesis for the effects of hadronization . experimental results on ` hard event properties ' clearly indicate that the concept contains a grain of truth but also that it is manifestly untenable at the level of ` one parton one hadron ' . its failure should be interpreted as a need for a less trivial , non - perturbative model of hadronization . a more specific , technical definition of lphd in terms of the limiting momentum spectrum ( @xmath159 ) @xcite was encountered in and will be discussed further in . so far it has been argued that the parton to hadron transition should occur locally in space - time , only involving a few neighbouring partons , and that local parton - hadron duality provides a reasonable summary of its effects on perturbative event structures . however this does not mean that hadronization is simple in detail . in practice many factors influence the production of a specific hadron and a realistic description is presently feasible only within the context of stochastic , non - perturbative models . independent fragmentation is perhaps the earliest framework for hadronization @xcite , later becoming synonymous with the work of field and feynman @xcite . as the name suggests , the hadronization of each individual parton is treated in isolation as a sequence of universal , iterative q@xmath175q@xmath176+h branchings based on the excitation of ( di)quark pairs . unfortunately the scheme has no strong theoretical underpinning and is rather arbitrary in its details , leading to many variants . it is used by cojets @xcite ( and isajet @xcite for hadron - hadron collisions ) and is a available as a set of non - default options in jetset @xcite . ( 140,35 ) ( 0,16)@xmath177 ( 15,17)(1,0)35 ( 34,17)(1,0)0 ( 50,17)(2,1)24 ( 62,23)(2,1)0 ( 50,15)(2,1)24 ( 62,21)(-2,-1)0 ( 78,27)@xmath178 ( 50,12)(8,6)[l ] ( 50,9)(4,-1)16 ( 58,7)(4,-1)0 ( 70,3)@xmath179 illustrates the ` unit cell ' used in the iterative implementation of the scheme . the flavours of the ( di)quarks generated are selected in fixed ratios . empirically it is found necessary to suppress both strange quarks and diquarks , u : d : s = 1:1:@xmath180 with @xmath181 and q : qq@xmath182 = 1:@xmath183 with @xmath184 . further rules are required to choose between the various ( low lying ) hadron states of a given flavour . the light - cone momentum fraction , @xmath185 , of the produced hadron , @xmath186 , is given by a longitudinal fragmentation function , such as @xmath187 in cojets a dependence on the mass of an emitted light hadron is also built in . special forms are employed in the case of heavy quarks where harder momentum spectra are expected ( see ) . the transverse momentum is chosen from a gaussian distribution @xmath188 possibly with a width that narrows as @xmath189 and @xmath190 , as expected from phase space considerations . the iteration continues until a backward moving hadron would be produced ( @xmath191 ) or , in the case of cojets @xcite , the sum of jet masses violates an available energy bound @xcite . diquarks are treated just as quarks , whilst gluons are first split into a light q@xmath192 pair ( the momentum is shared equally in cojets ) and the above algorithm used but with retuned parameters . typically a large cut - off value , @xmath193 gev , is used , resulting in only a few final - state partons . since these partons are treated in strict isolation , essentially ad hoc remedies must be used to ensure global conservation of quantum numbers . conserving four - momentum after the partons have acquired masses proves particularly troublesome because event shape variables , and hence @xmath41 determinations , are sensitive to the nature of the chosen solution @xcite . a fully lorentz - invariant scheme has been proposed in @xcite but it is hard to implement in general and only jetset @xcite contains a simplified version , as an option . in addition to the independent hadronization of the final - state partons , independent fragmentation models also naturally employ incoherent parton showers , the combination of the two features offering , by today s standards , a mediocre description of the exacting z@xmath0 data . however when applied to the hadronization of a back - to - back q@xmath192 pair there is little practical difference between independent and string - based models . ( it is only after gluon jets occur that differences become apparent ) . one can therefore speculate that a possible way to improve the situation is , after the forced gluon splitting , to apply the model to each neighbouring , colour singlet q@xmath194 pair in its own com frame just as to a pair of back - to - back jets . when a pair of oppositely coloured quarks move apart it is thought that the self - interacting colour field between them collapses into a long , narrow flux tube / vortex line , called a string . neglecting a short - range ` coulomb ' term the energy of this system appears to grow linearly with the separation . that is , the string has a uniform ( rest ) energy density or constant tension , estimated to be @xmath195 gev / fm . this is equivalent to a linear confining potential as expected from regge phenomenology ( @xmath196 ) @xcite , bag model calculations @xcite , lattice studies @xcite and quarkonium spectroscopy @xcite . this picture of a collapsed field is analogous to a chain of magnets @xcite and the behaviour of magnetic fields in ( type i / ii ) superconductors @xcite . the transverse size of a string , @xmath197 , is small compared to a typical length . therefore it is reasonable to try to model the string dynamics on those of a ` massless ' , relativistic string possessing no transverse excitations . the classical equations of motion , derived from a covariant area action , imply that in the com frame the two end quarks simply oscillate backwards and forwards along a line in what are known as yo - yo modes @xcite , seen as diamonds in . the equations of motion also admit solutions that include localized energy - momentum carrying ` kinks ' which have successfully been identified with hard gluons @xcite . at the end of a perturbative shower , string segments develop between neighbouring partons , the ends terminating on quarks . the full three - dimensional treatment of such a string system is rather complex , being characterized by the appearance of new string segments as intermediate gluons lose all their momentum to the system . however , robust , covariant evolution algorithms are available @xcite , and whilst ambiguities exist they are largely confined to the matching at gluon ` corners ' . the classical equations of motion are also more complex , becoming non - linear when the end quarks are massive @xcite . it is worth noting at this point that the quantized theory of the idealized string has spawned a rich and still growing subject of its own @xcite , which may still prove of relevance to the hadronization and confinement problems @xcite . an alternative scenario for a string description of gluons arises when the possibility of an octet colour flux tube is admitted . for example , in a q@xmath192 g event the quarks may be attached to triplet strings and the gluon to an octet string , all three of which join at a central node . a bag model calculation @xcite suggests the ratio of octet to singlet string tensions is @xmath198 . if however @xmath199 ( as suggested by lattice calculations ) then it becomes favourable for the octet string to split into two triplet strings and the above picture is recovered . the montevay independent fragmentation model @xcite requires choosing a frame in which the central junction is at rest . the above describes the motion of an idealized classical string due to the exchange of energy between the end quarks and the string ; in reality a second process is also believed to contribute . quantum mechanical effects allow the the creation of q@xmath192 or q@xmath200 pairs in the colour field of a stretched string , causing it to break in two _ la _ the snapping of a magnet . this picture is the basis for the lund group s familiar string hadronization model @xcite . however the lund version is only one of several possible @xcite , as illustrated in . ( 140,50 ) ( 35,40)(75,10)string area law ( 40,40)(0,-1)6 ( 40,34)(0,-1)4 ( 22,34)clusters ( 59,34)_breakup products _ ( 20,20)(40,10)mass regulator ( 25,20)(0,-1)6 ( 25,14)(0,-1)4 ( 55,20)(0,-1)6 ( 55,14)(0,-1)4 ( 0,0)(35,10)artru - mennessier ( 45,0)(25,10)cal - tech ii ( 105,40)(0,-1)6 ( 105,34)(0,-1)4 ( 110,34)hadrons ( 85,20)(40,10)lsff ( 90,20)(0,-1)6 ( 90,14)(0,-1)4 ( 73,14)quarks ( 93.5,14)_flavour rules _ ( 120,20)(0,-1)6 ( 120,14)(0,-1)4 ( 125,14)hadrons ( 80,0)(20,10)lund ( 110,0)(20,10)ucla the starting point for all string - inspired hadronization models is wilson s exponential area decay law , @xmath201 @xcite . this describes the probability of a string break occurring due to the creation of a q@xmath192 or q@xmath202 pair in the colour field at a point containing the space - time area @xmath203 within its backward light cone . as strings are believed to be uniform along their length , the probability of pair creation , @xmath204 , is a constant per unit area . now the momentum of a string fragment is proportional to its spatial extent ( @xmath205 and @xmath206 ) such that the string fragment mass is given by @xmath207 , so that the decay law can be reformulated as @xmath208 with @xmath209 . since @xmath210 this implies that on average the string break - up points , and hence the hadron formation points , lie scattered about a hyperbola : @xmath211 . this in turn implies that the slowest moving fragments form first near the centre of the string ( this is true in any frame as the break - up points are space - like separated ) an inside - out pattern is assured @xcite . before constructing an actual model one must decide what the string fragments are to be identified with : either continuous mass fragments @xcite clusters which then decay into hadrons , or actual discrete - mass hadrons @xcite . the first choice leads to the original artru - mennessier @xcite or caltech - ii @xcite schemes and the latter to the lund @xcite and ucla @xcite schemes . repeatedly applying the area decay law alone will result in an infinite sequence of ever smaller string fragments @xcite . one way to see this is to note that the area law is equivalent to a joint distribution in light - cone momentum fraction @xmath50 and transverse mass @xmath212 , which reduces to a divergent @xmath212 distribution : @xmath213 ( @xmath214 is the first exponential integral function . ) since a physical interpretation of string fragments with very low mass is implausible , in practice a regulator is required . in the more fully developed caltech - ii model @xcite this regulator is supplied by introducing a probability to allow a given string fragment to undergo any further splitting : @xmath215\ ] ] where @xmath204 and @xmath216 are free parameters , the latter related to the threshold mass for a string s decay to two hadrons . the function @xmath217 is the heaviside step function : @xmath218 for @xmath219 and @xmath220 for @xmath221 . if allowed , a break - up point is selected according to the area law with the q@xmath192 or q@xmath202 flavour chosen according to fixed probabilities from those kinematically allowed ; at this point no transverse momentum is introduced . occasionally a final string fragment is below a second cut - off and this is replaced by a single hadron . otherwise the fragments undergo a comparatively complex sequence of cluster decays in which phase space determines the produced flavours and momenta . whilst the caltech - ii model has attempted to combine the desirable features of both string and cluster models , its success in confronting data has been at best mixed . consequently the monte carlo program has fallen out of favour and is no longer being actively developed . the alternative to a continuous mass spectrum for the string fragments is a discrete spectrum , the allowed values of which are identified with the masses of known hadrons . this approach is followed by the lund @xcite and ucla @xcite models . requiring a string to fragment into hadrons of given ( transverse ) mass leaves only the choice of the hadrons light - cone momentum fractions , @xmath50 , free . these @xmath50 values may be iteratively chosen according to several possible distributions @xcite and still remain true to the area decay law . however a set of plausible assumptions greatly reduces the number of allowed fragmentation functions @xcite . the assumptions are : the equations of motion are those of a classical , relativistic , constant tension string with no transverse excitations ; a statistical left - right symmetry ; a central rapidity plateau ; and negligible end effects . the resulting @xmath50-distributions are known as the lund symmetric fragmentation functions ( lssf ) . @xmath222 the coefficient @xmath223 relates to the parent quark flavour and @xmath224 to that of the quark or diquark produced in the colour field : in practice only diquarks are allowed a different value of @xmath51 @xcite . taking every @xmath225 the lsff simplifies ( compare to where @xmath226 ) and is equivalent @xcite to a flat rapidity , @xmath227 $ ] , distribution and the wilson area law modified by a perimeter ( coulombic ) term . an important issue now is to prescribe how the actual hadrons are chosen , and here again the models diverge . the lund approach @xcite is based on an attempt to model , principally through flavour and spin selection rules , the supposed quark dynamics in the strong colour field that is a string . the idea is a development of the concept of fermion pair production in a strong electromagnetic field @xcite . to supply the energy for a pair , each of transverse mass @xmath228 , it is necessary to consume a finite length of string ( @xmath229 ) . the quarks have equal and opposite @xmath230 since no transverse string excitations are permitted in the model . if the pair are produced locally at a point then they must tunnel out to this classically required separation . using the wkb approximation for the matching of the quark wavefunctions at the classical turning points suggests a suppression factor @xcite : @xmath231 this would be a crude approximation to the known full qed expression for the production of a single pair , @xmath232 . unfortunately it is not known what to use for the quark masses in ( see reference @xcite for some discussion of the range spanned by current and constituent quark masses ) and so only qualitative conclusions can be drawn . among these are : * the transverse momentum suppression is the same for all quark flavours * charm and bottom quarks will not be produced from the string * since @xmath233 ( su(3)@xmath234 is broken ) , strange quarks will be suppressed relative to up and down quarks : @xmath235 * production of diquarks will be suppressed : @xmath236 the power 4 arising in the last term because of the quadratic dependence of the tunnelling probability on the diquark mass . also the gaussian @xmath230-distribution in is best regarded as a first approximation , leaving open the possibility of long tails originating from unresolved gluon emission @xcite . this may account for the fact that the prediction for the width of the hadron @xmath230-distribution , @xmath237 gev , proves too narrow compared with measurements of @xmath238 gev @xcite . as yet the model does not supply as much guidance on how to account for any possible spin dynamics . two factors influence the relative production ratios of same flavour mesons @xcite . first are the ( @xmath239 ) spin counting factors . second is the need for the quark produced in the string to match onto the wavefunction of the produced meson . at the classical boundary to the tunnelling region the meson wavefunction is expected to behave as @xmath240 @xcite . the result for the vector to pseudoscalar meson production ratio is therefore : @xmath241 as @xmath242 is @xmath243 for @xmath244 and 0.004 for b / b@xmath245 , the primary @xmath246 ratio is taken to be 1 whilst the b@xmath247b ratio is fixed at 3 . radially and orbitally excited mesons are expected to be suppressed @xcite , although @xmath248 mesons may be included at the expense of a new parameter for each of the four additional families of states . the fact that the transverse mass is involved in the suppression factor proves unimportant except for pions , since @xmath249 . hence it is anticipated that ( directly produced ) pions will be enhanced at low @xmath230 and have a tighter @xmath230 distribution than , say , @xmath250 mesons or kaons . also , because of their small mass , neighbouring - pion correlations are anticipated @xcite . to summarize , the lund prescription for an iterative string fragmentation scheme is : first choose a quark flavour ; then choose a produced hadron species ; select a quark transverse momentum ; and finally select an @xmath50 value . the normalization , @xmath251 , of the lsff in for this case is thus : @xmath252 the resulting model requires the specification of a relatively large number of free parameters @xcite . much of this can be traced back to the unknown properties and dynamics of ( di)quarks . the ucla model @xcite attempts to finesse these problems by formulating an iterative scheme only in terms of known hadron properties , thereby trying to avoid the issue of quark production in a string at the heart of the difference between the lund and ucla models is a reinterpretation of the lsff @xcite . in the ucla model @xcite the lsff is used to choose both @xmath50 and the species of produced hadron . this means that @xmath251 becomes an _ absolute _ normalization : @xmath253 the sum runs over all hadrons containing the parent quark , with cg the appropriate clebsch - gordon coefficients for the hadron wavefunction . since @xmath251 is now a common constant the hadron mass dependence appearing in the exponential term in immediately implies a suppression of heavy hadrons and a stiffening of their fragmentation function @xcite . an instructive way to view the difference between the lund and ucla approaches is to consider the complete weight for the production of a set of @xmath251 hadrons @xmath254 in a string of mass @xmath6 stretched between two quarks q@xmath255 and @xmath256 , as illustrated in . ( 100,80 ) ( 5,5)(0,1)15 ( 0,15)@xmath102 ( 5,5)(1,0)15 ( 15,0)@xmath50 ( 50,0 ) ( 1,1)40 ( 50,0)(-1,1)40 ( 48,10)@xmath203 ( 30,10)@xmath257 ( 65,10)q@xmath255 ( 10,40)(1,1)10 ( 17.5,43)@xmath258 ( 20,50)(-1,1)20 ( 0,70)(1,1)5 ( 25,35 ) ( 1,1)5 ( 25,35)(-1,1)20 ( 21,30)@xmath259 ( 32.5,38)@xmath260 ( 5,55)(1,1)10 ( 15,65)(-1,1)10 ( 11,51)h@xmath261 ( 30,52)(2,0)5 ( 90,40)(-1,1)15 ( 75,55)(1,1)17.5 ( 85,45.5)@xmath262 ( 83,57)h@xmath263 ( 70,35)(1,1)25 ( 95,60)(-1,1)12.5 ( 68,30)@xmath264 ( 60,38)q@xmath263 ( 77.5,38)@xmath265 ( 70,35)(-1,1)20 ( 62.5,43)@xmath266 ( 62,58)h@xmath176 ( 50,55)(1,1)22.5 ( 42.5,27.5)(-1,1)5 ( 42.5,27.5 ) ( 1,1)35 ( 40.5,22.5)@xmath267 ( 50,30.5)@xmath268 ( 77.5,62.5)(-1,1)15 the ` master ' equation for this process s weight is given by : @xmath269 ^ 2 \exp\{-\kappa'(p^2_\perp)_{{\rm h}_i}\}\d^4p_{{\rm h}_i } \delta(p^2_{{\rm h}_i}-m^2_{{\rm h}_i})\end{aligned}\ ] ] the first two terms impose overall four - momentum conservation and the area decay law . the second set of terms is associated with the production of the @xmath270 intervening pairs : @xmath271 is a possible quark flavour suppression factor and @xmath272 controls the quark s gaussian @xmath230-distribution . the third set of terms is associated with the @xmath251 hadrons @xmath254 formed out of the quark - antiquark pairs @xmath273 : cg is the clebsch - gordon coefficient for the hadron s su(6 ) wavefunction , @xmath274 is an additional `` knitting factor '' and @xmath275 controls the hadron s assumed gaussian @xmath230-distribution . the approaches to these various terms in the ucla and lund models are summarized in table 1 . factor & ucla @xcite & lund @xcite & controls + @xmath276 & 1 & @xmath277ud@xmath278ud@xmath279 & quark flavour suppression + @xmath272 & 0 & `` @xmath280 '' & quark @xmath281 suppression + @xmath275 & @xmath282 & 0 & hadron @xmath281 suppression + @xmath283 & @xmath251 ( const . ) & @xmath284 & hadron `` knitting '' factor + a number of points in the iterative implementation of the ucla scheme are noteworthy . first , the original model was designed to look ahead to the next iterate . for example , if the first hadron leaves behind a u or an s quark then the next hadron is most likely to be a pion or a kaon , and the latter choice is ( doubly ) suppressed because of the higher masses of strange hadrons . this was an attempt to mimic the quantum mechanical projection of the string onto a set of hadrons @xmath285 . in fact , using and the ucla ansatz for an event s overall weight it is possible to derive an iterative scheme that automatically generates chains of hadrons according to this overall distribution . that is , the projection from partons to hadrons is automatically taken into account . the required fragmentation function is remarkably similar to the original lsff : @xmath286 compared to a finite mass correction term appears together with a term , @xmath287 , coming from local transverse momentum compensation , ( @xmath288 ) . in the case of hadrons containing a heavy quark , @xmath50 is replaced by @xmath289 which softens the momentum spectrum and favours heavier states @xcite . second , ` multiple ' so - called popcorn ( discussed below ) baryon production b@xmath290bm@xmath290bmm@xmath291 is included ( here b represents a baryon and m a meson ) , but due to its slow ` convergence ' an ad hoc suppression needs to be introduced : @xmath292 with @xmath293 free ( @xmath294 gev@xmath2 ) . third , the mechanism for the local conservation of hadron transverse momentum proves awkward due to an ambiguity between the quark and hadron levels . finally , but perhaps most significantly , the model only contains four ( + two ) free parameters : @xmath295 and @xmath296 and @xmath46 ) . a further interesting aspect of string models is that inferences can be drawn from their associated space - time picture . these include predictions on spin correlations and effects due to quantum statistics . if a pair are produced with some ( equal and opposite ) transverse momentum , @xmath230 , with respect to the string , then because they are separated by a distance @xmath229 a non - zero angular momentum @xmath297 is necessarily introduced . since total angular momentum @xmath298 must be conserved and @xmath299 the pair typically form in a @xmath300 state , particularly so at higher @xmath230 . this is expected to lead to @xmath301transverse - spin correlations , and spin correlations between neighbouring hadrons @xcite . transverse polarizations are only possible because the string introduces a preferred axis . ( such an axis is implicit in the chain - like structure found in cluster models , though it is not presently utilized in them ) . conservation of angular momentum is also expected to lead to a suppression of orbitally excited hadrons . ( 100,65 ) ( 5,5)(1,0)10 ( 7,14)@xmath102 ( 5,5)(0,1)10 ( 17,4)@xmath50 ( 42.5,45)(1,1)5 ( 47.5,50)(1,-1)5 ( 52.5,45)(1,1)7.5 ( 60,52.5)(1,-1)5 ( 65,47.5)(1,1)5 ( 70,52.5)(1,-1)7.5 ( 60,5)(1,1)40 ( 60,5)(-1,1)40 ( 55,32.5)(1,1)7.5 ( 62.5,40)(1,-1)5 ( 67.5,35)(1,1)10 ( 77.5,45)(1,-1)5 ( 82.5,40)(1,1)5 ( 87.5,45)(1,-1)5 ( 92.5,40)(1,1)5 ( 55,32.5)(-1,1)12.5 ( 42.5,45)(-1,-1)5 ( 37.5,40)(-1,1)5 ( 32.5,45)(-1,-1)5 ( 27.5,40)(-1,1)5 ( 57.5,42.5)@xmath302 ( 73,37)1 ( 46,36)2 ( 45.5,45)1 ( 71,46)2 if two identical hadrons are produced from a string there exists an ambiguity in the possible rank ordering of the particles , unless they have exactly equal momentum . these two string drawings , illustrated in , enclose different areas . now if , as is believed , the string area law derives from the modulus squared of a matrix element @xmath303 $ ] then quantum mechanical interference will occur @xcite . dependent on the bose - einstein or fermi - dirac nature of the particles the expected joint production probability becomes @xmath304\ ] ] which is clearly modified from the naive sum of the weights . ( a quark transverse momentum correction , @xmath305 , is not shown in . ) as the momentum difference squared between the identical hadrons vanishes , @xmath306 and any enhancement or cancellation becomes maximal . the area difference , @xmath302 , is related to the size of the emitting volume , likened to the hanbury - brown twiss effect in optical astronomy @xcite , but not related directly to the total string size . this approach therefore has close parallels to a more geometric picture based on the fourier transform of the source distribution @xcite . only identical neutral pions can be produced side by side from a string identical charged pions must have at least one intervening hadron . therefore @xmath302 will be larger for charged pions , leading to smaller correlations than for neutral pions @xcite . the effects of correlations on short lived resonance decays can also be included @xcite . two basic algorithms are currently available for including a bose - einstein event ` weight ' @xcite . the standard scheme available in jetset @xcite involves rearranging identical boson momenta so that they are distributed according to the pairwise correlation function . full multiboson correlations may be included , but at the cost of additional computing time . the effect of including bose - einstein correlations may be likened to adding an attractive inter - boson force , leading to ` lumpier ' distributions . the experimental situation is discussed in . in the lund string model , baryon production poses particular problems of principle . the basic difficulty appears to stem from incomplete knowledge of the internal structure of a baryon @xcite . is it a quark - diquark system or a three quark system ? this ignorance poses less of a problem for herwig and to some extent the ucla string model because these essentially only need to know a baryon s mass and spin . however the lund string works directly with the ( di)quarks themselves and so in the absence of a guiding principle it therefore allows for two options , the diquark @xcite and popcorn @xcite baryon production mechanisms . ( 140,15 ) ( 4,0)(2,0)3 ( 10,0)(1,0)5 ( 14,0)(1,0)0 ( 15,5)(10,10)[br ] ( 20,5)(0,1)5 ( 20,9)(0,1)0 ( 19.5,11)m ( 22,5)(0,1)5 ( 22,6)(0,-1)0 ( 27,5)(10,10)[b ] ( 32,5)(0,1)5 ( 32,9)(0,1)0 ( 34,7)(0,1)3 ( 34,9)(0,1)0 ( 32.5,11)b ( 36,7)(0,1)3 ( 36,9)(0,1)0 ( 41,7)(10,10)[b ] ( 41,7)(14,14)[b ] ( 46,7)(0,1)3 ( 46,6)(0,-1)0 ( 48,7)(0,1)3 ( 48,6)(0,-1)0 ( 46.5,11)@xmath307 ( 50,5)(0,1)5 ( 50,6)(0,-1)0 ( 55,5)(10,10)[bl ] ( 55,0)(1,0)5 ( 58,0)(1,0)0 ( 62,0)(2,0)3 ( 74,0)(2,0)3 ( 80,0)(1,0)5 ( 84,0)(1,0)0 ( 85,5)(10,10)[br ] ( 90,5)(0,1)5 ( 90,9)(0,1)0 ( 92,7)(0,1)3 ( 92,9)(0,1)0 ( 90.5,11)b ( 94,7)(0,1)3 ( 94,9)(0,1)0 ( 99,7)(14,14)[bl ] ( 99,7)(10,10)[b ] ( 104,7)(0,1)3 ( 104,6)(0,-1)0 ( 103.5,11)m ( 106,7)(0,1)3 ( 106,9)(0,1)0 ( 111,7)(10,10)[b ] ( 116,7)(0,1)3 ( 116,6)(0,-1)0 ( 99,0)(1,0)12 ( 107,0)(1,0)0 ( 111,7)(14,14)[br ] ( 118,7)(0,1)3 ( 118,6)(0,-1)0 ( 116.5,11)@xmath307 ( 120,5)(0,1)5 ( 120,6)(0,-1)0 ( 125,5)(10,10)[bl ] ( 125,0)(1,0)5 ( 128,0)(1,0)0 ( 132,0)(2,0)3 the diquark mechanism is a straightforward generalization of the quark meson production model and was the first to be fully developed @xcite . however a stepwise quark model for the production of baryons was the first to be proposed @xcite and implemented @xcite . whilst this only evolved later @xcite ( and is continuing to evolve @xcite ) into the popcorn mechanism it would be a misconception to regard it as especially contrived or unnatural . when a qq@xmath308 ( or ) pair is produced in a string s colour field with the same colour as the end quarks they precipitate a string break : the diquark mechanism @xcite . when a ( virtual ) pair is produced the possibility that they have a different colour to that of the end quarks allows a non - zero colour field to exist between them in which further real pairs could form , leading to the sequence b@xmath309bm@xmath307 etc : popcorn production @xcite . perhaps not unsurprisingly the popcorn mechanism requires a ( modest ) number of new free parameters . the main practical consequence , so far , of introducing popcorn production appears to be a lessening of the phase space correlations between baryon - antibaryon pairs ( see ) . the fragmentation function for baryons is also expected to be softened @xcite ( compared to a meson with the same transverse mass ) due to popcorn production : in particular bm@xmath310 sequences cause a suppression of leading baryons @xcite . ( this would suggest using @xmath311 in the lsff . ) the actual level of popcorn production required to describe data is still the subject of debate but it may be related to the @xmath51 parameter of the lsff @xcite . interestingly a search for the expected chain like structures , such as correlated @xmath312k@xmath313 systems , has failed to see any positive evidence @xcite . allowing three body decays of clusters is expected to have similar effects to introducing popcorn production in string models . whilst clusters initially appeared as intermediate states in the string model of artru and mennessier @xcite the first fully fledged cluster hadronization models originated at caltech @xcite . ( cluster - like structures also naturally arose in the earlier statistical bootstrap model @xcite and multiperipheral models @xcite . ) today the scheme is best known through the herwig implementation of webber @xcite . this is based on the pre - confinement property proved for pqcd @xcite : at the end of the perturbative shower the mass and spatial distributions of colour singlet clusters of partons spanned by quark - antiquark pairs have a universal distribution . in practice gluons remaining at the cut - off scale @xmath46 are forcibly split into light pairs , the wolfram ansatz @xcite , so that in the planar approximation @xcite neighbouring quark - antiquark pairs form colour singlets . illustrates the resulting cluster mass distributions for u - quark initiated jets at four energies , showing clearly the universality of the colour singlet cluster mass distribution , in contrast to the distributions for random clusters . the mean cluster mass is of order @xmath46 , a few gev , and for a colour - coherent parton shower has a spectrum falling faster than any power @xcite . the distribution is independent of the initial parton type ( q or g ) and virtuality : this is in contrast to random quark - antiquark pairs . this universality is suggestive of the formation of intermediate ` super - resonances ' which independently decay into the familiar resonances . illustrates the stages in the fragmentation of a parton into hadrons via cluster hadronization . ( 140,50 ) ( 5,45)perturbative ( 0,20)(15,10)parton ( 15,25)(1,0)4 ( 19,25)(1,0)1 ( 20,20)(15,10)shower ( 35,25)(1,0)4 ( 39,25)(1,0)1 ( 40,20)(15,10)clusters ( 55,0)(0,2)10(0,1)1 ( 55,31)(0,2)10(0,1)1 ( 55,30)(1,1)8.5 ( 63.5,38.5)(1,1)6.5 ( 65,34)heavy ( 70,40)(25,10)splitting ( 82.5,40)(0,-1)6 ( 82.5,34)(0,-1)4 ( 55,25)(1,0)8.5 ( 63.5,25)(1,0)6.5 ( 70,20)(25,10)2-bdy decay ( 55,20)(1,-1)8.5 ( 63.5,11.5)(1,-1)6.5 ( 65,14)light ( 70,0)(25,10)1-bdy decay ( 95,28)(1,0)6 ( 101,28)(1,0)4 ( 95,22)(1,0)6 ( 101,22)(1,0)4 ( 95,5)(1,0)6 ( 101,5)(1,0)4 ( 105,0)(15,30 ) ( 120,15)(1,0)4 ( 124,15)(1,0)1 ( 125,10)(15,10)hadrons ( 105,45)non - perturbative three masses of cluster need to be treated : average : : most clusters have masses , @xmath314 , close to @xmath46 and undergo two - body decays @xcite . a quark or light diquark pair ( here the diquarks serve only as a mnemonic device for quantum number conservation ) , @xmath315q@xmath316q@xmath317q@xmath318 , is selected at random , and for a cluster of flavours q@xmath319 the putative hadrons h@xmath320q@xmath321q@xmath263q@xmath317q@xmath322 and h@xmath323q@xmath324 are formed . the selection is accepted or rejected according to its phase space weight : @xmath325 where @xmath326 and @xmath327 are the hadron spins and masses , and @xmath328 is the com momentum of the kinematically allowed two - body decays . if accepted , the decay momenta are selected isotropically in the cluster rest frame . this prescription is in accord with the ozi rule @xcite . heavy : : a number of clusters in the tail of the distribution are too massive for an isotropic decay to be plausible . the criterion used is : @xmath329 where @xmath330 and constituent quark masses are used . here the label f denotes the flavour type of the cluster and @xmath331 and @xmath154 are variable parameters of the model . by the ( repeated ) device of introducing a light quark pair , heavy clusters are forcibly split into two daughter clusters whose masses are chosen according to a power law spectrum and whose directions are aligned with the q@xmath332 axis in the cluster rest frame @xcite . this appearance of a preferred , colour field , axis is reminiscent of a string splitting . light : : occasionally it is kinematically impossible for a cluster to decay into two hadrons . such a cluster is decayed into the lightest hadron h@xmath333q@xmath334 and four - momentum transferred to a neighbouring cluster to satisfy the mass - shell constraint @xcite . in the absence of a theory of cluster decay the simplest viable approach , pure phase space , is adopted ( okham s razor ) @xcite resulting in very few free parameters @xcite . here two - body dominance of the decays is one such simplification @xcite . however non - trivial matrix elements are not precluded and parameters controlling the relative production rates of ( di)quark pairs in the cluster decays and also weights for the hadron representations ( knitting factors ) are made available to users @xcite . the details of heavy cluster splitting prove important : @xmath331 ( related to the available kinetic energy ) influences the light baryon yield , @xmath154 the heavy hadron yield and both influence the momentum spectra @xcite . the one - body cluster decay mechanism is important for the production of rare , heavy states such as the @xmath335 or b@xmath336 , and also for describing leading particles , to which end a number of extra parameters have been added @xcite . despite the fact that its initial purity has been somewhat corrupted by the need to handle exceptional mass clusters , the model remains a simple , compact and predictive scheme . since clusters are typically light , limited transverse momentum is automatic , hadrons with non - zero strangeness and baryon number are suppressed because they are heavier , and the spin ratios of iso - flavour hadrons follow partly from the ( @xmath239 ) factor and partly from the larger masses of higher spin states . the model represents a well founded attempt to go as far as possible with as little as possible . all the major hadronization models , cluster , independent and string , are local , universal , stochastic models . they are based on a small number of recursively applied branchings , where at each iterative step probabilistic rules are applied to select flavours , spins and momenta . the main features the models are summarized in table 2 . feature & + & + & cluster & independent & + & & & + & & & lund & ucla + principle & very simple & simple & complex & less complex + lorentz invariant & yes & no & + flavour , charge & automatic & ad hoc & + etc , conservation & & & & + mass dep . via & hadrons & quarks & quarks & hadrons + strangeness supp . & predicted & free param . & restricted params . & predicted + baryon supp . & predicted & free param . & restricted params . & predicted + @xmath337 ratios & predicted & free params . & restricted params . & predicted + limited @xmath338 & natural & built in & built in & natural + frag . & n / a & free & + cut - off @xmath339 dep . & significant & very strong & + stability & infrared prob . & collinear prob . & + limitations & massive clusters & requires & + & treated like & large @xmath46 & + & ` strings ' & & & + lorentz invariance and quantum number conservation are not troublesome for strings or clusters but are an issue for independent hadronization models requiring post facto adjustments . these essentially ad hoc remedies are not always implemented ( particularly in application to dirtier hadronic collisions @xcite ) , and , worse , physical observables are known to be sensitive to details of the chosen solution @xcite . only the cluster and ucla models , with their emphasis on hadron properties , provide succinct basic algorithms for flavour , spin and momentum selection . in their initial formulations the lack of free parameters gives these models a laudable predictive power , which is in marked contrast to the lund string and independent hadronization models . of course , poor fits to data could mandate elaborations of the models which in turn may then dilute their predictive power . the lund model ought to have an advantage because it is based on a semi - quantitative picture of an underlying dynamics but this is undermined by the indirect measurability of the basic ( di)quark parameters . thus whilst the model s parameters possess a large degree of internal coherence , their must be disappointment at their large number and seemingly byzantine complexity . to emphasize this point , 13 inputs are needed in the basic lund model @xcite to describe the 13 ( assuming u@xmath340d isospin ) @xmath341 light ( u , d , s ) hadrons ( the number of hadrons increases to 27 if u and d are distinguished ) . thus the flavour and spin aspects of the lund model show little practical difference to the rather similar , but assumed , rules found in an empirical independent hadronization model . however a significant difference is the way in which a string s causal structure restricts the longitudinal fragmentation function to a unique family of left - right symmetric functions @xcite . no such restriction applies to independent models . the issue of stability with respect to collinear and soft gluons arises in cluster and independent hadronization models . specifically , the spectrum of clusters is sensitive to the emission of isolated , soft gluons . this infrared instability may be regarded either as a serious problem or , perhaps , as a warning that it is important to treat the theory correctly . perturbative qcd does not like isolated colour charges , which are strongly sudakov suppressed ; this is responsible for the fast falling tail of the cluster mass spectrum @xcite . ( recall also that many other observables , such as the mean multiplicity , are known to be infrared sensitive . ) in independent fragmentation a similar problem arises when one final - state parton is replaced by two parallel partons of equal net energy , giving a different multiplicity . this collinear instability occurs essentially because the two partons are oblivious to each other s presence . to recap , the motivation for the rules used in these models varies from the qcd - inspired , complex dynamics of strings through the minimalism of clusters to the simple expediency of independent fragmentation . in the subsequent sections we shall see how well these hadronization models compare to the various z@xmath0 data . however since independent hadronization makes no claim to be based on qcd , a fact reflected in the extreme arbitrariness of its parameters , we drop it from further discussion . while it is not considered a viable scheme for describing z@xmath0 physics it does however survive in the isajet @xcite monte carlo program for hadron hadron collisions . recently interest has arisen in the possibility that soft , long wavelength , gluons may cause non - perturbative rearrangements in the colour structures of events developed during the showering stage , particularly in relation to w@xmath3w@xmath4 pair production @xcite . several phenomenological models are available based on the three main event generators , and are summarized in table 3 . feature & + basic mc & & herwig & + author(s ) & j. hkkinen & l. lnnblad & b.r . webber & t. sjstrand & s. todorova + & g gustafson & & & v a khoze & + reference(s ) & @xcite & @xcite & @xcite & @xcite & @xcite + rearrangement : & & & decrease cluster & + - criterion & & spatial size & + - in shower & no & yes & no & no & yes + - in hadron . & yes & yes & yes & yes & yes + - mult . re - arr . & yes & yes & yes & no & yes + - inter - singlet & yes & yes & yes & yes & yes + - intra - singlet & no & yes & yes & no & yes + the rearrangement criterion in the ariadne - based models is a decrease in the @xmath342-measure @xcite which quantifies the momentum - space size of a string system . only the second model @xcite should be considered as realistic for typical events . in herwig at the end of the perturbative shower pairs of @xmath182 singlet clusters may be rearranged , with fixed probability ( @xmath343 ) , if this results in a reduction in the quadratic sum of their space - time sizes . in jetset the rearrangement criterion is based on the space - time evolution of the strings . in model i the spatial overlap of ` wide ' flux tubes is used to assign probabilities for rearrangement , whilst in model ii rearrangement occurs at the first crossing of the ` narrow ' vortex lines . todorova s model @xcite is an elaboration of the original model @xcite , allowing for multiple reconnections , including within a single colour singlet system , and self - interactions ; this leads to string loops glueballs . as yet these models have not been thoroughly investigated , nor their consequences for z@xmath0 decays found . note that when including colour rearrangement the monte carlos must first be retuned : for example reducing the @xmath342-measure also lowers the average multiplicity , which must be compensated . since the physics of colour rearrangement is universal in nature , effects should also be anticipated in all other types of hadronic interactions , including for example b - hadron decays ( b@xmath344 ) and rapidity gap events @xcite , where important constraints may be found . the lep machine , a 27 km circumference storage ring , was conceived and constructed primarily as a z@xmath0 factory . in the lep 1 mode , running at and near the z@xmath0 peak , a luminosity of @xmath345 @xmath346s@xmath2 could be obtained , with beam lifetimes of up to 20 hours . until 1992 , four bunches of electrons and positrons crossed every 22 @xmath125s at the experiments . during 1993 and 1994 lep ran in a mode allowing eight bunches , with a crossing time reduced to 11 @xmath125s . in 1995 , so - called bunch trains were introduced , with four trains , each of three bunches of particles , providing a further increase in luminosity . the beam spot in 1995 , the last year of extensive running at the z@xmath0 , was 250 @xmath125 m in the vertical and 5 @xmath125 m in the horizontal directions . in the years 1990 , 1992 and 1994 , lep ran on the z@xmath0 peak to produce the largest possible number of events . in 1991 , 1993 and 1995 , the centre - of - mass energy was scanned across the z@xmath0 peak to allow studies of the z@xmath0 line shape and other tests of the electroweak theory . unlike lep , the slc is a single pass collider , with the electron and positron bunches lost after each pass . the bunch crossing frequency , 120 hz , is therefore very much lower than at lep . the slc is however capable of providing electron beams with large polarization . in order to increase the luminosity , the beams are tightly squeezed before the collisions to reduce their cross sectional area . this technically difficult procedure results in a beam spot of diameter 0.5 @xmath125 m in the vertical plane and 2.3@xmath125 m horizontally . the experiment , sld , is able to exploit this by placing precision microvertex detectors only 3 cm from the interaction point . since 1989 the slc luminosity has steadily improved to around @xmath347 @xmath348s@xmath2 , and electron beam polarization values of 80% are now routinely achieved . both lep 1 and slc could provide their host experiments with clean , low background , experimental conditions . in addition , the relatively low beam crossing rates allowed the experiments to implement bias - free triggers and to collect data with essentially no dead time . for studies of multihadron production at the z@xmath0 peak , initial - state radiation is negligible . multihadronic z@xmath0 decay events could be triggered with efficiencies greater than 99% and selected offline with backgrounds ( mainly due to @xmath349 lepton pair production ) of less than 1% . the four lep detectors , aleph , delphi , l3 and opal , and the sld detector at the slc are typical large , multipurpose , particle physics detectors designed to allow measurement of events over a large part of the solid angle . for studies of hadronization , certain features of the detectors are particularly relevant . a large , active tracking volume within a strong magnetic field allows reconstruction of jets of charged particles and of secondary vertices from strange particle decays ; good track momentum and direction measurement give accurate reconstruction of systems of two or more particles ; charged particle identification permits the study of inclusive identified hadrons ; electromagnetic calorimetry allows measurement of inclusive @xmath103 and @xmath293 mesons ; and precision secondary vertex reconstruction and inclusive electron and muon identification may be used to identify charm and beauty particles . the coordinate systems used by the experiments define @xmath116 to be along the beam direction , so that the @xmath350 plane is perpendicular to the beams . then @xmath351 and @xmath352 are the usual cylindrical polar coordinates , and the @xmath350 plane is also called the @xmath353 plane . the angle @xmath53 is normally the polar angle to the @xmath116 ( beam ) axis . the aleph detector @xcite was designed to provide high three - dimensional granularity with large solid angle coverage for charged particle tracking and for calorimetry . its silicon vertex detector , drift chamber and large time projection chamber ( tpc ) , in a 1.5 t magnetic field , give a resolution on momentum transverse to the beam directions of @xmath354 ( gev / c)@xmath2 for 45 gev / c tracks . at low momentum , the resolution is dominated by multiple scattering which contributes @xmath355 to @xmath356 . the silicon vertex detector permits the measurement of track impact parameters with an accuracy of @xmath357m@xmath358m@xmath359 ( with @xmath360 in gev / c ) in both the @xmath353 and @xmath361 planes , allowing excellent reconstruction of charm and beauty particles . the tpc also measures ionization energy loss , @xmath362 , giving good electron identification to high momenta , @xmath363/k separation at two standard deviations ( 2@xmath364 ) in the relativistic rise region above 2 gev / c , and k / p separation at 1@xmath364 for momenta over 5 gev / c . the efficiency to measure k@xmath365 and @xmath121 particles is about 50% at maximum , which occurs at about 8 gev/@xmath366 momentum . the lead / wire - chamber electromagnetic calorimeter has an energy resolution of @xmath367 and an angular resolution of @xmath368 mrad ( @xmath369 in gev ) . along with the @xmath362 information , the calorimeter gives an average electron identification efficiency of 65% in hadronic jets . the efficiency for reconstruction of @xmath103 mesons peaks at about 50% at 10 gev , falling to 20% at 2 gev and 10% at 30 gev . the @xmath103 energy resolution is around 7% , independent of energy . muons are identified using the hadron calorimeter and muon chambers ; only muons above 3 gev/@xmath366 momentum penetrate the system , and these are detected with an average efficiency of 86% . a pivotal feature of the delphi @xcite detector is the particle identification capability of its ring imaging cherenkov detectors ( rich ) . these cover both the barrel and end cap regions , and have both liquid and gas radiators . when combined with ionization energy loss information from the tracking detectors , the system gives clear identification of charged particles over the whole momentum range at lep 1 . the tracking detectors of delphi operate in a magnetic field of 1.23 t and consist of a silicon vertex detector , an inner drift chamber , a tpc and an outer detector of drift tubes . in addition there are two forward chambers , containing planes of drift tubes , to improve reconstruction at low polar angles . the system gives a momentum resolution @xmath370 of 0.7% to 1.4% in the barrel region , varying with track momentum . measurement of k@xmath365 and @xmath121 in multihadron events has an efficiency of 3040% over a wide momentum range . the silicon vertex detector allows track impact parameters to be measured with an accuracy which depends on momentum and polar angle , and which varies between 30 and 130 @xmath125 m in @xmath353 and between 40 and 200 @xmath125 m in @xmath116 . delphi s electromagnetic calorimeters consist of high - density projection chambers in the barrel region and lead glass blocks in the endcap regions . they are preceded by @xmath371 radiation lengths of material in the barrel region , and more in the endcap regions , so that resolution is somewhat degraded . when combined with @xmath362 information , the system allows electron identification with efficiency and purity values both around 50% over a wide momentum range . photons are identified both in the calorimeters and by measuring conversion e@xmath3e@xmath4 pairs in the tpc , allowing reconstruction of @xmath103 mesons . muons above 3 gev/@xmath366 are detected using a hadron calorimeter and muon chambers with an efficiency between 75 and 85% . [ [ l3 ] ] l3 ^^ the l3 detector @xcite was designed with the primary aim of reconstructing electrons , muons and photons . it is therefore more limited than the other detectors in its capabilities for studying hadronization . l3 has a low magnetic field of 0.5 t in a large cylindrical volume of diameter 12 m. the central time expansion chamber measures tracks out to a radius of 31.7 cm with a high spatial resolution in the @xmath353 plane . it is supplemented by a @xmath116-chamber to measure the polar angle of tracks . the l3 arrangement gives optimized momentum resolution for penetrating muons , with @xmath372 for 45 gev / c muons . its bismuth germanium oxide ( bgo ) electromagnetic calorimeter is preceded by less than 10% of a radiation length in the barrel region , and has a spatial resolution better 2 mm above 2 gev . the energy resolution is about 5% at 100 mev and 1.4% at 45 gev . the calorimeter permits electron identification , with only 0.1% probability of misidentifying a hadron . it is also well suited for measurements of inclusive @xmath103 , @xmath293 and @xmath373 production . the opal @xcite detector has a warm solenoid providing a magnetic field of 0.435 t. the main central tracking jet chamber lies outside of a silicon microvertex detector and a precision vertex drift chamber ; a set of @xmath116-chambers around the jet chamber give precise measurement of track polar angles . the combination of the tracking chambers gives a momentum resolution of @xmath374 with @xmath375 in gev/@xmath366 . efficiency for reconstruction of k@xmath365 and @xmath121 particles varies with momentum , having a maximum value of 30% at about 5 gev/@xmath366 . the silicon detector , orginally an @xmath353 device but improved in 1993 to also measure @xmath116 , gives an impact parameter resolution of 15 @xmath125 m in @xmath353 for high momentum tracks . the jet chamber measures ionization energy loss , @xmath362 , of tracks in multihadronic events with a resolution of 3.8% , allowing excellent identification efficiency and high purity for electrons , pions , kaons and protons over almost the whole momentum range at lep 1 . opal s lead glass electromagnetic calorimeter , which is preceded by some two radiation lengths of material , has an energy resolution varying with energy from 1 to 5% . the muon chambers , together with the instrumented flux return of the hadron calorimeter , are highly efficient for identification of muons with momentum greater than 3 gev/@xmath366 . the sld @xcite experiment , like delphi , makes use of cherenkov ring imaging to identify charged particles , with a detector which uses both liquid and gas radiators to allow coverage of a wide momentum range . tracking is done by a central drift chamber within a 0.6 t magnetic field which gives a momentum resolution , @xmath376 , with @xmath375 in gev/@xmath366 . for electromagnetic calorimetry , the sld uses a liquid argon device , inside the magnet coil , which gives a resolution of around 15%/@xmath377 ( with @xmath369 in gev ) . the sld has silicon charge - coupled - device pixel detectors for microvertex measurements . the pixels are 22 micron square , and the setup covers radii from 3 to 4 cm from the interaction vertex . resolution on track impact parameter in @xmath353 is in the range 11 to 20 @xmath125 m , depending on track momentum and polar angle . muons are identified by layers of streamer tubes between the slabs of iron which make up sld s warm iron calorimeter . inclusive single identified particles are usually studied in terms of their fractional energy ( @xmath378 ) or momentum ( @xmath379 ) relative to that of the beams , with the fragmentation functions being reported as @xmath380 . here @xmath381 is the total cross section for . its inclusion is experimentally advantageous since it obviates the need to measure absolute cross sections , so reducing systematic errors : @xmath382 is simply calculated as the number of particles , @xmath383 , in bin @xmath384 relative to the total number , @xmath385 , of z@xmath0 hadronic decays . in a real measurement the bin width is finite , @xmath386 , and the differential cross section is taken as @xmath387 . because fragmentation functions vary rapidly with @xmath50 , care has to be taken for large values of @xmath386 in interpreting a measurement as a differential cross section at some particular value of @xmath50 @xcite . total inclusive particle yields , or average multiplicities per hadronic z@xmath0 decay , are obtained by integrating the measured fragmentation functions and extrapolating into any unmeasured regions of @xmath50 with the aid of one or more models or interpolation functions . systematic errors are included to account for uncertainties in this procedure . fragmentation functions and total inclusive yields have been measured for an impressively large number of particle species at lep . the string and cluster models , as implemented in jetset and herwig , are usually confronted with the data . a comprehensive compilation of inclusive particle production data in annihilation at all available com energies above the @xmath335 mass , as of mid-1995 , may be found in @xcite where measured fragmentation functions are plotted along with curves obtained from jetset version 7.4 . a more recent review @xcite contains a good summary of measurements published after @xcite . tables 4 and 5 list of all the measured inclusive yields of mesons and baryons published to date . where an experiment has reported more than one measurement , only the most recent is taken . for each measurement of a particular particle , statistical and systematic errors have been combined in quadrature ; then the weighted mean of the available measurements has been calculated to give the results shown in the tables ( no attempt has been made to take into account systematic errors correlated between experiments ) . yields reported over a restricted @xmath50 range are given separately . in general there is very good agreement among the measurements of the different experiments . in only two cases , where the measurements are listed separately in the tables , is there evidence of disagreement : the delphi and opal measurements of the @xmath388 are possibly inconsistent , and the @xmath389 yield reported by delphi does not agree well with the numbers given by aleph and opal . particle & multiplicity & herwig59 & jetset74 & ucla74 & comments + charged & @xmath390 & 20.40 & 20.95 & 20.88 & adlmo @xcite + @xmath391 & @xmath392 & 16.62 & 16.95 & 17.04 & o @xcite + @xmath103 & @xmath393 & 10.15 & 9.59 & 9.61 & adl @xcite + @xmath293 & @xmath394 & 0.246 & 0.286 & 0.232 & a @xcite @xmath395 + & @xmath396 & 0.92 & 1.00 & 0.78 & l @xcite + @xmath397 & @xmath398 & 1.12 & 1.50 & 1.17 & ad @xcite + @xmath399 & @xmath400 & 1.05 & 1.35 & 1.01 & al @xcite + @xmath401 & @xmath402 & 0.071 & * 0.127 * & 0.061 & a @xcite @xmath395 + & @xmath403 & 0.143 & 0.297 & * 0.121 * & l @xcite + f@xmath404 & @xmath405 & 0.068 & & & d @xcite @xmath406 + @xmath407 & @xmath408 & * 0.181 * & * 0.194 * & * 0.132 * & ado @xcite + f@xmath409 & @xmath410 & 0.137 & & & d @xcite @xmath411 + f@xmath412 & @xmath413 & 0.021 & & & d @xcite + k@xmath3 & @xmath414 & 2.08 & 2.30 & 2.24 & do @xcite + k@xmath0 & @xmath415 & * 1.87 * & 2.07 & 2.06 & adlo @xcite + k@xmath416 & @xmath417 & * 0.524 * & * 1.10 * & 0.779 & ado @xcite + k@xmath418 & @xmath419 & * 0.530 * & * 1.10 * & 0.760 & ado @xcite + k@xmath420 & @xmath421 & 0.067 & & & d @xcite + & @xmath422 & * 0.054 * & & & o @xcite @xmath423 + d@xmath3 & @xmath424 & 0.190 & 0.174 & 0.196 & ado @xcite + d@xmath0 & @xmath425 & 0.406 & 0.490 & 0.497 & ado @xcite + d@xmath426 & @xmath427 & 0.151 & * 0.242 * & * 0.227 * & ado @xcite + d@xmath365 & @xmath428 & 0.087 & 0.129 & 0.130 & o @xcite + b@xmath429 & @xmath430 & * 0.182 * & 0.260 & 0.254 & d @xcite + b@xmath431 & @xmath432 & * 0.032 * & & & d @xcite + j/@xmath433 & @xmath434 & * 0.0018 * & 0.0050 & 0.0050 & adlo @xcite + @xmath435 & @xmath436 & 0.00097 & 0.0019 & 0.0019 & do @xcite + @xmath437 & @xmath438 & 0.00088 & & & dl @xcite + @xmath335 & @xmath439 & @xmath440 & @xmath441 & @xmath442 & o @xcite @xmath443(3 lightest @xmath335 ) + particle & multiplicity & herwig59 & jetset74 & ucla74 & comments + p & @xmath444 & * 1.41 * & 1.19 & 1.09 & do @xcite + @xmath445 & @xmath446 & * 0.278 * & * 0.189 * & * 0.139 * & d @xcite + & @xmath447 & 0.278 & 0.189 & 0.139 & o @xcite + @xmath121 & @xmath448 & * 0.605 * & 0.385 & * 0.332 * & adlo @xcite + @xmath449 & @xmath450 & & & & o @xcite + @xmath451 & @xmath452 & 0.123 & 0.073 & 0.061 & o @xcite + @xmath453 & @xmath454 & 0.102 & 0.068 & 0.056 & o @xcite + @xmath455 & @xmath456 & 0.225 & 0.140 & 0.118 & do @xcite + @xmath457 & @xmath458 & 0.093 & 0.073 & 0.074 & ado @xcite + @xmath459 & @xmath460 & * 0.202 * & * 0.074 * & * 0.074 * & ado @xcite + @xmath461 & @xmath462 & * 0.0746 * & 0.0271 & * 0.0220 * & ado @xcite + @xmath389 & @xmath463 & * 0.0352 * & 0.0053 & 0.0081 & a @xcite + & @xmath464 & * 0.0352 * & 0.0053 & * 0.0081 * & d @xcite + & @xmath465 & * 0.0352 * & 0.0053 & 0.0081 & o @xcite + @xmath466 & @xmath467 & * 0.0093 * & 0.00072 & 0.0011 & ado @xcite + @xmath468 & @xmath469 & 0.0129 & 0.059 & * 0.026 * & o @xcite + in tables 4 and 5 the measured rates are compared with the outcome of the three major monte carlo models which attempt a full simulation of particle production . the numbers in bold font show results which are more than three standard deviations from the experimental measurements . in each case , the default versions of the programs have been used , although for jetset version 7.4 various sets of alternative parameters have been suggested which improve the agreement with the overall rates . the most recent herwig version 5.9 does not fit as well as version 5.8 , but a new default set of parameters will no doubt follow careful comparisons with data . it is clear from the various jetset tunings suggested by the four lep experiments in @xcite that there are strong correlations among the program s parameters , possibly such that there is no unique best set . to this extent jetset may be underconstrained despite the large number of experimental measurements . delphi have published @xcite comprehensive sets of tuned parameters for various monte carlo models ( including ariadne and jetset with matrix elements as well as with parton showers ) in which they take account of event shape variables as well as inclusive identified particle rates . this exercise is useful but probably premature . some of the recent measurements differ significantly from those used in the tuning , and many have much reduced errors . for example the @xmath466 baryon is now known to be produced at a much lower rate than previously measured , and it turns out the delphi tuned rate fits better with this new rate than with the one used as input ; the same is true of the @xmath407 meson rate . all of the models considered in @xcite describe the inclusive rates reasonably well , with the exception of the performance of herwig in the baryon sector . although the available meson measurements are predominantly of the @xmath341 pseudoscalar and vector states , the presence of the @xmath248 mesons shows clearly their importance in the hadronization ; this is confirmed also in the baryon sector with the observation by opal of the @xmath449 . other scalar , axial vector and tensor mesons , together with orbitally excited baryons , must presumably also be produced , although large widths and small branching ratios will make them difficult to measure . many of the lighter particles are therefore decay products of other hadrons , and care has to be taken in any interpretation of the data using only relative light particle rates . so far only herwig 5.9 includes by default the p and some d wave meson states , and as the tables show their inclusion does not mar the agreement with data in the u , d sector . however herwig does poorly with mesons containing s quarks as well as with baryons . the production of the light tensor mesons is discussed by delphi in @xcite where a comparison is made of relative rates of tensor to corresponding vector mesons . while the production ratio f@xmath470 is similar to that for f@xmath471 , at about 20% , there is evidence for a lower k@xmath472/k@xmath429 ratio , in agreement with results from hadroproduction experiments . this suggests an extra suppression of strange tensor mesons . however opal @xcite measures a larger rate for k@xmath472 and so the picture is not yet clear . some evidence has also been reported @xcite for a rise in the ratio f@xmath470 with meson momentum ; in other words the fragmentation function of the tensor meson may be harder than that of the vector meson , as generally expected of heavier hadrons @xcite . however , these are difficult resonances to measure ; they have large widths , large combinatorial backgrounds and uncertainties in the resonance line shapes . as usual , more results would help . only one measurement of a scalar meson , the f@xmath404 , has been reported @xcite , with a ratio f@xmath473 of @xmath474 . thus if the f@xmath255 is indeed a conventional @xmath475 meson ( see @xcite for a mini - review of the scalars ) then the scalar and tensor mesons are produced with similar rates . but again the measurements are difficult , and other studies @xcite of the inclusive @xmath96 mass spectrum with higher statistics have failed to report a measurement of the f@xmath404 because of systematic uncertainties . one should therefore be wary of too much interpretation of one measurement . the f@xmath404 and @xmath476 have aroused interest @xcite as potential probes of the gribov confinement scenario @xcite . in this theory the qcd vacuum is likened to the intense qed fields expected around super charged ions , @xmath477 ( or @xmath478 for point - like charges ) @xcite . this results in the production of spatially compact ` novel vacuum scalars ' , identified with the f@xmath404 , which are expected to be produced in relative isolation . particular signatures include enhanced production at central rapidities ( with respect to the thrust axis ) and in low multiplicity events @xcite . from a semiclassical point of view , the orbital angular momentum @xmath479 of a pair from string fragmentation is given by @xmath480 where @xmath481 is the mean quark momentum transverse to the string and @xmath482 is the size of the resulting hadron @xcite . typical values give @xmath483 so that the rate for mesons with non - zero orbital angular momentum is expected to be less than about 10% of the rate for the corresonding l=0 mesons . the large average quark - antiquark separation in radially excited states also works to prevent production of these mesons in string fragmentation . although the @xmath248 mesons can be simulated in jetset at the correct rates by the adjustment of appropriate parameters , the large experimental rates are nevertheless a problem for the basic model assumptions about production of hadrons from strings . strangeness suppression is immediately apparent from both the meson and baryon measurements given in the tables . there is a large number of ways to determine , from the data , values for @xmath484 , the quark - level strangeness suppression assumed in the string model . results are tabulated for example in @xcite where all measurements agree on a value of 0.3 , a result which accords with reasonable values for the strange quark mass @xcite and the string energy density @xmath272 . since the various methods in @xcite use both light and heavy quark states , this consistency suggests that the suppression occurs at the quark level , in agreement with the string - model assumption . on the other hand the ucla model also reproduces reasonably well the strange particle rates , and previous versions of herwig have been tuned to do so . and hadronization studies in ep collisions at hera @xcite give a lower value of @xmath485 , in apparent disagreement with the z@xmath0 decay measurements . therefore it is fair to say that the data are not yet conclusive . a direct method to measure the strangeness suppression in has been proposed @xcite which makes use of the electroweak forward - backward asymmetry and which could possibly distinguish between quark - level and hadron - level suppression . recently sld @xcite have applied this method to their data on inclusive k@xmath486 and @xmath487 production using 150k events , with the result @xmath488 . in the absence of mass effects , the ratio of direct pseudoscalar to vector meson production may depend simply on spin statistics , in which case the value p/(p+v ) would be expected to be 3/4 . however feed - down from decays is also important and may obscure the interpretation of the experimental results . one approach to determine the underlying p/(p+v ) value is to tune the appropriate parameter in the jetset model , but this can only be done within the limited knowledge available on production of the higher states . alternatively , one can use the measurements in the b and c sectors where vector to pseudoscalar mass differences are much smaller and there are some hopes to measure the orbitally excited states . the average ratio for primary b@xmath429/(b+b@xmath429 ) is found to be @xmath489 @xcite , in excellent agreement with simple spin counting . however the picture is not so clear when charm is considered : here the ratio d@xmath429/(d+d@xmath429 ) is measured @xcite to be @xmath490 . the difficulty here arises from incomplete knowledge of the production rate and decay modes of the orbitally excited d@xmath491 states which are likely to feed down into the d and d@xmath429 production . the question then arises as to why b@xmath491 production does not similarly muddy the waters in the b sector ; it may be that production and decay rates of the four different @xmath337 states of b@xmath491 conspire to leave the value of p/(p+v ) at 3/4 . more measurements are needed before definite conclusions can be reached . in the string model , baryon yields are determined by many parameters . the overall baryon rates relative to mesons depend on the relative probability to produce a diquark pair from the sea . spin-1 diquarks may be suppressed relative to spin-0 diquarks . the strangeness suppression enters in a similar way as for meson production , but there is in addition the possibility of extra suppression of strangeness in a diquark . and the popcorn mechanism may introduce one or more mesons locally in phase space between a baryon and an antibaryon . although jetset does rather well , there are some discrepancies with measured rates . attempts by opal @xcite to tune the parameters which control baryon production have shown that it is not possible to reproduce simultaneously all of the measured rates . the suggestion then is that the mechanisms for baryon production in the string model , and particularly for the strangeness suppression , are deficient . however , as has been said , there is now clear experimental evidence also for orbitally excited states in baryon production . the rate for the @xmath492 @xmath449 , at 0.02 per hadronic z@xmath0 decay , is around 5% of that for the @xmath493 @xmath121 . and there are many similar baryon states which can not be measured experimentally but which , it is fair to assume , must be produced in the hadronization . so since jetset , like all of the other models , does not include production of orbitally excited baryons , no clear conclusions can yet be reached . neither herwig nor ucla , both of which rely only on phase space , mass and spin , are successful in the strange baryon sector , with the former consistently overestimating the rates , and the latter underestimating them . as presently implemented , baryon production in herwig does not take account of the appropriate su(6 ) clebsch - gordon coefficients and this will lead to an overestimation of baryon production rates . as with jetset , neither of these models includes the production of baryons other than the lowest lying @xmath341 states . in principle their inclusion would lower the predicted rates for primary low lying baryons , since some higher states would be produced in their stead . however there would be a compensatory increase in the rates due to feed down from decays . since herwig consistently overproduces and ucla underproduces the baryons , it seems reasonable to deduce that at least one of them is incapable of being fixed up by this mechanism . the conclusion then from the baryon yields is that the string model of jetset , while certainly not perfect , is in reasonable accord with the measurements . recent work on the baryon production within the popcorn model has further improved the agreement @xcite . whether or not this is to be taken as a strong endorsement of the model is however an arguable point , given the large amount of freedom available to tune parameters in order to reproduce the observations . both the herwig cluster model and the ucla string model are clearly in difficulty , and it remains to be seen whether they can be rescued . a number of models have been proposed to treat only the overall yields of identified particles . such models are necessarily of limited physical content , although they turn out to be reasonably successful in describing the inclusive rates . why they do so is not at all clear . in the thermodynamic model @xcite the source of particles is assumed to be a hadron gas in thermal and chemical equilibrium . the model has three parameters , a temperature , a volume and a parameter to allow for incomplete strange chemical equilibrium ( similar to the strangeness suppression of the lund string model ) . the model gives a good fit to the lep data , as well as to lower energy annihilation data , with a temperature of around 170 mev , close to @xmath494 and the temperature found in the earlier statistical model of hagedorn @xcite . the author of the model speculates that the thermal equilibrium could be a feature of the quark - hadron transition , brought about by strong interactions . this argument was also invoked by fermi @xcite to justify his phase space model for hadron production . however it is difficult to reconcile this picture with the conventional view of hadronization in annihilation as occuring locally in the wake of rapidly separating colour sources . a `` striking regularity '' @xcite has been noted in the particle yields , and a simple formula proposed which reproduces well the observations ( apart from the pions and possibly the @xmath466 baryon ) : @xmath495 . here , @xmath251 is the yield for a particle of spin @xmath326 and mass @xmath496 , and @xmath497 is the isospin for baryons and a `` modified '' isospin for k and k * mesons and isosinglet pseudoscalar mesons . the introduction of the `` modified '' isospin appears rather ad hoc , although there are plausible arguments to justify it . the model makes no attempt to explain the yield differences between members of the same isomultiplet which , for example , are significant for kaons . the parameters @xmath51 and @xmath114 are fitted to the data , and the slope parameter @xmath114 , at about 3.9 gev@xmath498 , is found to be the same for lep as for lower energy measurements , implying that the regularity may be universal . it is unclear as to the physical origin of the expression , and it will be interesting to see how the formula copes with future measurements . this observed regularity has a less successful predecessor of the form @xmath499 @xcite , which was applied to lower energy data . another model @xcite of the total yields is based on string fragmentation , and proposes a simple formula with only three parameters : a strangeness suppression , an effective temperature and a relative normalization factor between mesons and baryons . following the string model , the rate @xmath251 of light meson and baryon production is taken as @xmath500 . the normalization @xmath136 depends on the centre - of - mass energy , @xmath501 is a relative suppression of baryons , @xmath326 is the particle spin , @xmath502 gives the suppression for a hadron containing @xmath503 strange quarks , @xmath504 is the hadron binding energy and @xmath135 is the effective temperature . the model gives a good simultaneous fit to lep and lower energy data , with a temperature of @xmath505 mev and a strangeness suppression @xmath484 of @xmath506 . the model also gives a good description of heavy flavour production . its predictions for production rates of excited charm states have recently been shown to agree with opal measurements @xcite . due to their narrow widths and the availability of clean leptonic decay channels the principle heavy quarkonium states measured are the @xmath507 and @xmath335(1s , 2s,3s ) , based upon which further excitations can be reconstructed @xcite . the production of these heavy q@xmath508 bound states is thought to be rather atypical of hadron production in general , especially for the @xmath335 , due to the significant part played by perturbative physics . in the case of charmonium the dominant production mechanism is expected to be weak b hadron decays : b@xmath153c@xmath509s ) plus subsequent colour rearrangement . in the case of b@xmath510 mesons the j@xmath511 branching ratios ( @xmath5121.15% ) have been previously measured at the @xmath171 @xcite , so allowing reliable predictions for charmonium rates at the z@xmath0 . ( the presence of b@xmath513 and b baryons at the z@xmath0 makes little difference to the inclusive b hadron branching ratio @xcite ) . perturbative fragmentation contributes only at the few percent level to charmonium production @xcite , as is indeed observed @xcite , but 100% to bottomonium production . three basic pqcd production mechanisms , illustrated in , are considered @xcite : * heavy quark fragmentation @xcite , z@xmath514b@xmath515 . here the production of a primary b@xmath515 pair is followed by the radiation of a gluon which splits into a second b@xmath515 ; a b and @xmath516 from these two pairs then bind in a colour singlet system . it is noteworthy that in addition to the quarkonium two other heavy hadrons occur in this process . * gluon fragmentation @xcite , z@xmath517q@xmath192g@xmath518g@xmath519b@xmath520gg . the need to remove the colour from the b@xmath515 pair , whilst forming a positive @xmath136 parity @xmath335 ( or @xmath507 in the case of c@xmath521 ) , requires the emission of two perturbative gluons . thus this is an order-@xmath522 process and typically the quarkonium state has a relatively large transverse momentum . * gluon radiation @xcite , z@xmath523b@xmath520gg . here two gluons are emitted from a primary heavy ( anti)quark allowing it to recoil and form a colour singlet with the other ( anti)quark . this results in a very hard , isolated , quarkonium state . all of the above prompt production mechanisms give comparatively isolated quarkonia originating from the interaction point , in contrast to charmonium from b decays . here the b@xmath515 system is produced in a colour singlet state by the emission of perturbative gluons . the dominant process is quark fragmentation @xcite . however the theoretical predictions are significantly low compared to the z@xmath0 data . also no evidence for displaced vertices , associated with additional heavy quarks , is observed in quarkonium events . a similar situation has occurred at the tevatron where colour singlet fragmentation mechanisms fail to account for the number of observed high-@xmath375 quarkonium states @xcite . fortunately recent theoretical developments suggest that colour octet quarkonium production may play an important role @xcite . here the q@xmath508 forms a colour octet system from which the colour is leached away ( by exchange of soft gluons ) in an , as yet unspecified , non - perturbative mechanism . two new contributions , also shown in , arise : gluon fragmentation z@xmath517q@xmath192g@xmath524g@xmath525b@xmath520 ; and gluon radiation z@xmath517b@xmath526 g . these new diagrams are lower order in @xmath41 but suppressed by larger powers of @xmath527 , the relative velocity of the b and @xmath515 . to calculate these processes requires knowledge of the octet matrix elements ; until recently these have been taken from fits to the cdf data @xcite though now a potential model based calculation is available @xcite . ( _ ab initio _ lattice calculations have so far only been performed for octet decay matrix elements @xcite and not the technically demanding production matrix elements ) . in the the octet case the dominant fragmentation contribution to @xmath335 production becomes gluon fragmentation @xcite . taking into account all contributions , agreement with the z@xmath0 data is possible . although the data are insufficient to be able to isolate components due to the individual processes , the lack of an observed hard @xmath335 spectrum does rule out a large contribution from octet gluon radiation @xcite . it is also noteworthy that the octet mechanism predicts a large transverse polarization for the vector quarkonium states @xcite particularly so at high momentum . ( 130,180 ) ( 20,175)colour singlet ( 90,175)colour octet ( 0,120 ) ( 16000,16000 ) ( 0,8000)[3 ] ( , ) [ 8000 ] ( , ) by 300 by -500 ( , ) @xmath528 ( , ) [ 4000 ] ( , ) ( , ) [ 4000 ] ( , ) by 300 by -500 ( , ) @xmath529 ( , ) [ 1 ] ( , ) [ 5000 ] ( , ) ( , ) [ 5000 ] ( , ) by 300 by -500 ( , ) @xmath52 ( 40,115)heavy quark fragmentation ( 70,60 ) ( 16000,16000 ) ( 0,8000)[3 ] ( , ) [ 8000 ] ( , ) by 300 by -400 ( , ) @xmath530 ( , ) [ 3000 ] ( , ) ( , ) [ 5000 ] ( , ) by 300 by -200 ( , ) @xmath531 ( , ) [ 2 ] ( , ) [ 1500 ] ( , ) ( , ) [ 5000 ] ( , ) by 500 by -1300 ( , ) @xmath529 ( , ) [ 1500 ] ( , ) ( , ) [ 5000 ] ( , ) ( 0,60 ) ( 16000,16000 ) ( 0,8000)[3 ] ( , ) [ 8000 ] ( , ) by 300 by -400 ( , ) @xmath530 ( , ) [ 3000 ] ( , ) ( , ) [ 5000 ] ( , ) by 300 by -200 ( , ) @xmath531 ( , ) [ 2 ] ( , ) [ 7000 ] ( , ) by 500 by -1000 ( , ) @xmath529 ( , ) [ 3000 ] ( , ) ( , ) [ 3 ] ( , ) [ 3000 ] ( , ) ( , ) [ 3 ] ( , ) [ 4000 ] ( , ) ( 45,55)gluon fragmentation ( 70,0 ) ( 16000,16000 ) ( 0,8000)[3 ] ( , ) [ 8000 ] ( , ) by 500 by -1000 ( , ) @xmath529 ( , ) [ 3000 ] ( , ) ( , ) [ 3 ] ( , ) [ 8000 ] ( , ) ( 0,0 ) ( 16000,16000 ) ( 0,8000)[3 ] ( , ) [ 8000 ] ( , ) by 500 by -1000 ( , ) @xmath529 ( , ) [ 3000 ] ( , ) ( , ) [ 3 ] ( , ) [ 3000 ] ( , ) ( , ) [ 3 ] ( , ) [ 5000 ] ( , ) ( 50,0)gluon radiation looking at the j@xmath511 data in table 4 , one can conclude that since b hadron decays dominate charmonium production , herwig is deficient in this aspect of its b decay description . possible remedies include adding more explicit decay channels to the decay tables or refining the partonic b decay model , particularly in its treatment of light clusters . next , because none of the models considered include the dominant octet production mechanism the large underestimates of @xmath335 @xcite production are not too surprising . only ariadne , version 4.09 onwards , been extended to include an approximation to the perturbative colour octet production mechanism @xcite . this program is therefore the only one containing the necessary physics to attempt to describe the @xmath335 data . having discussed the total production rates of identified particles , we now turn to their momentum distributions : @xmath168 . all particle spectra vanish as @xmath532 and are expected to vanish as @xmath533 , though measurements are rarely available at sufficiently low momentum to see indications of this behaviour . however changing to the commonly used variable @xmath534 , a clear ` hump - backed plateau ' shape ( see ) is seen . a compilation of many of the measurements of the @xmath50 and @xmath160 spectra of identified hadrons at the z@xmath0 can be found in @xcite and @xcite . references to the more recent measurements are listed in table 6 . in addition , a recent measurement of the charged particle momentum spectrum is available in @xcite . + particle & & particle & + @xmath103 & a & @xcite & f@xmath535 & d & @xcite + @xmath293 & a & @xcite & k@xmath0 & a & @xcite + @xmath399 & l & @xcite & k@xmath536 & a & @xcite + @xmath401 & al & @xcite & k@xmath537 & d & @xcite + @xmath407 & d & @xcite & k@xmath538 & d & @xcite + + particle & & particle & + @xmath121 & a & @xcite & @xmath461 & a & @xcite + @xmath539 & a & @xcite & @xmath389 & a & @xcite + in the calculation of the charged particle momentum spectrum was discussed ; these results can also be applied to identified hadron spectra @xcite . especially in the case of identified hadrons a dependence on the primary quark flavour is anticipated : monte carlo studies and recent sld results @xcite suggest that the main effects are at high momentum but that residual effects could also occur at low momentum . consequently investigations have again concentrated on the peak position , @xmath161 , which occurs at low momentum and for which the complete prediction is of the form @xcite : @xmath540 + \frac{1}{2}\ln\left(\frac{q}{\lambda}\right)+\cdots\ ] ] observe that the cut - off , @xmath46 , dependence only occurs through the first term which has the property that @xmath541 whilst the scale , @xmath52 , dependence only arises in the second term : the constant @xmath121 is the usual ( effective ) qcd scale . in order to create a particle of a given mass it is reasonable to expect that the final - state partons will require virtualities of the same magnitude , so that @xmath46 is a simple function of the hadron mass @xmath542 @xcite . invoking the lphd concept one can hope that these parton level calculations are then sufficient to predict an identified hadron s energy spectrum . application of has caused some confusion . first , there are small differences between experiments in the practical definition of @xmath161 . second , it is common to set @xmath543 so that the first term vanishes , the limiting spectrum case , but then somewhat inconsistently to introduce a new variable @xmath544 in the second term . that is is replaced by : @xmath545 given the availability of low energy data it is possible to investigate the @xmath546 dependence of @xmath161 for the @xmath547 , k@xmath3 , k@xmath365 , p and @xmath121 hadrons separately @xcite . the data for each hadron species appear to lie on straight lines each of slope @xmath548 but with differing offsets , indicating that @xmath549 in decreases with increasing hadron mass . aleph also finds that the linearity of these fits can be improved by using jetset to remove the effects of secondary hadron decays @xcite . for example kaons produced in the weak decays of hard b mesons stiffen the kaon spectrum causing a decrease in their @xmath161 . however this correction procedure is somewhat at odds with the original idea of lphd which was thought to account for such decays . at the z@xmath0 it is also possible to study the hadron mass dependence of @xmath161 with @xmath550 fixed @xcite , which probes the relationship between @xmath551 and @xmath542 . adopting opal @xcite claim two linear relationships between @xmath161 and @xmath542 ( an exponential dependence of @xmath552 on @xmath542 ) one describing the mesons , except pions , and the other the baryons . this pattern is well reproduced by jetset allowing a correction to remove secondary decays to be applied after which all the points , including pions , now fall on a single curve . however delphi @xcite report that mesons and baryons lie on two curves corresponding to a negative logarithmic dependence of @xmath161 on @xmath542 ( linear relationship between @xmath552 and @xmath542 ) which again reduce to a single curve after correcting back to primary hadrons using jetset . whilst the full shape of the momentum spectrum can be calculated within pqcd @xcite comparison is more commonly made to monte carlo predictions , particularly so in the case of identified particle momentum spectra , reflecting the expected interplay between perturbative and non - perturbative contributions . from a practical point of view the primary concern is with a correct description of the pion momentum spectra @xcite which dominate the total event multiplicity . these fall by five orders of magnitude over the measured momentum range . here , and in general , the predictions of ariadne and jetset are rather similar and typically agree with the data a little better than herwig , though all suffer problems at very low and very high @xmath50 . meson spectra are described reasonably well on the whole , though the k@xmath553 spectrum is notably too soft @xcite which might be attributable to an inadequate description of heavy hadron decays . however the description for baryons is less satisfactory . in particular the number of fast , predominantly primary , protons is consistently overestimated @xcite . in the case of ariadne and jetset this may be remedied @xcite by invoking a suppression of leading uds - baryons @xcite , which may have an underlying physics motivation @xcite . an intriguing possibility is that of an inequality in the fragmentation function of strange quarks into protons and antiprotons , @xmath554 @xcite . this prediction follows from a possible asymmetry , of non - perturbative origin , in the sea quark structure functions of protons @xcite and application of the ` reciprocity rule ' @xcite . a test of this idea may be possible by sld using a strangeness tag @xcite and exploiting their polarization asymmetry to distinguish the quark from the antiquark jet . hadrons containing heavy ( c , b ) quarks are special since the heavy quark is expected to be principally ( if not exclusively ) of perturbative origin . in practice non - perturbative physics also plays a role and the delineation of the two contributions is not clear cut . if the fragmentation function can be written as a convolution of perturbative ( @xmath555 ) and non - perturbative ( @xmath556 ) parts , @xmath557 , then it immediately follows that @xmath558 . in leading order pqcd predicts @xcite @xmath559 ; the nlo calculation of @xmath560 @xcite is sharply forward - peaked , vanishing at @xmath561 . the perturbative result alone is too hard ( @xmath562 ) and a non - perturbative component is required , especially in those rare cases ( of order 1% ) where no gluon radiation occurs at all . a simple argument @xcite shows that the non - perturbative hadronization should also be very hard for heavy quarks due to their inertia , @xmath563 , where @xmath72 is a typical hadron size , with the remaining energy uniformly distributed in rapidity between @xmath564 . a full expression for @xmath565 depends on the details of the hadronization mechanism assumed , and several are available . the most commonly adopted standard is due to peterson et al @xcite . in an independent fragmentation approach , the amplitude for the transition q@xmath566q@xmath567q@xmath182 in perturbation theory is proportional to the inverse of the energy transfer , assuming a constant matrix element . the fragmentation function is then given by the amplitude squared and the appropriate flux factor as : @xmath568 theoretically one predicts @xmath569 ; experimental fits yield small @xmath570 values with @xmath571 , consistent with @xmath572 @xcite . on average , @xmath573 as anticipated above . a perceived problem with the peterson form is its @xmath574 behaviour ( it is sometimes argued that including gluon radiation is equivalent to an additional @xmath575 factor ) as @xmath190 which is in conflict with the ` reciprocity ' rule @xcite . this posits that the @xmath532 behaviour of the q@xmath576h fragmentation function should equal the @xmath532 behaviour of the h@xmath576q structure function @xmath577 . furthermore it is expected from dimensional counting arguments @xcite that as @xmath532 , @xmath578 where @xmath579 is the number of spectators in the hadron h. in a meson @xmath580 and therefore the @xmath532 behaviour of the fragmentation function should be @xmath575 . two alternatives are available . by adopting an explicit meson wavefunction @xcite and thereby introducing a non - trivial matrix element , a ` refined ' peterson form may be derived : @xmath581 an earlier approach @xcite , based directly on the structure function analogy , leads to : @xmath582 here @xmath583 so that @xmath584 in accord with expectation . in string models two approaches have been elaborated according to whether the momentum or space - time aspects are emphasized . in the presence of massless quarks , which move along linear light - cone trajectories , these pictures are simply related , @xmath585 ; however massive quarks move along displaced hyperbolae . in the lund momentum - space approach the relevant string area is given , as for light quarks , by @xmath586 and the same reasoning then leads to the lsff . however this gives @xmath587 which is too hard compared to both theoretical expectation and experiment @xcite . adopting a space - time based approach @xcite , _ la _ artru mennessier , the string fragments randomly into clusters and the relevant string area becomes : @xmath588\ ] ] where @xmath589 is the mass of the string segment containing the heavy quark . in the @xmath590 limit this results in a generalization of the lsff but with no @xmath51-term present . elaborating this scheme @xcite to multiple string breaks along a finite size string leads to the following effective fragmentation function : @xmath591 now @xmath592 should be identified with the mass of the lightest q hadron , where in fact the cluster mass spectrum peaks . this has a softer spectrum , @xmath593 , and is very similar to the peterson form in practice . it should be noted that the non - perturbative fragmentation functions can be expected to describe effects due to the perturbative emission of soft gluons , as found in a parton shower . however effects due to the emission of hard gluons can not be accounted for . in particular when defining @xmath50 as a light - cone momentum fraction this may lead to a difference between the reconstructed value @xmath594 , measured along a jet axis , and the primary value @xmath595 , generated with respect to the ` string ' axis : the lund model indicates @xcite that @xmath596 . the experimental measurement of the b fragmentation function is difficult ; to date it has been based on reconstructed b@xmath576d@xmath597 decays or the rapidity method @xcite . also , the interpretation of the results is delicate , and ambiguous conclusions have been drawn . two sources of confusion are the delineation of the perturbative and non - perturbative contributions and whether a distinction is made between primary and secondary b hadrons cascading down from excited states . at the z@xmath0 there is substantial production of b@xmath245 @xcite and the four b@xmath598 @xcite mesons : the primary rate is approximately 1:3:2 for b : b@xmath245:b@xmath599 . opal @xcite finds reasonable agreement with the peterson , collins and spiller , kartvelishvili and lund - bowler fragmentation functions . on the other hand , the aleph data @xcite would rule out the collins and spiller form whilst favouring the kartvelishvili form ; the data also disfavour an untuned version of herwig , which is too soft . preliminary studies from delphi @xcite indicate that jetset with parton showers ( ps ) and the peterson form , , gives the best description of the data but with a peak that is appreciably too wide . using jetset ps with lund - bowler functions , , gives a very poor fit as does using jetset with matrix elements and the peterson function . this last combination gives too narrow a peak , indicating the need for a contribution from gluon radiation . herwig gives a reasonable description of the data but must be tuned to avoid too many heavy b clusters . delphi @xcite also give measurements of the primary b@xmath245 and b@xmath599 fragmenation functions . l3 @xcite find agreement within errors with the peterson fragmentation function . as discussed in [ selwk ] , the primary quarks produced via are highly polarized . whether , and in what circumstances , this polarization survives the hadronization or whether spin - spin forces wash out any memory of the initial polarization is an open question . the primary quarks may become constituents of unstable baryons or of vector or tensor mesons , the angular distribution of whose decay products may be used to extract information about spin states . in the particular case of leading ( large @xmath50 ) spin-1/2 @xmath121-type baryons , considerable polarization is expected @xcite , though also see @xcite . in the constituent quark model , the spin of such baryons is carried by the heavy quark , with the light diquark system in a spin zero , isospin zero state ( though see the discussion in ) . thus fast particles could carry a substantial fraction of the polarization of the initial b quark , with the light diquark system carried along as a spectator . similarly , high-@xmath50 , and @xmath121 baryons formed in fragmentation of s quarks , are expected to be polarized . since @xmath121 particles can also arise from hadronization of initial u and d quarks , the polarization in this case is considerably reduced . aleph have measured the polarization of baryons @xcite using semileptonic decays , @xmath600+charmed hadrons . the method @xcite is based on measurement of the ratio of the average lepton to average neutrino energy . the measured longitudinal polarization , @xmath601 , is well below the theoretical expectation of @xmath602 although the error is large . this is a surprising result and , if confirmed , would indicate the likely existence of depolarizing mechanisms in the b quark hadronization . because of the parity - violating nature of the decay , @xmath603 , the distribution of the polar angle , @xmath604 , of the proton direction in the @xmath121 rest frame , relative to the @xmath121 direction in the laboratory , is proportional to @xmath605 , where @xmath606 is the degree of longitudinal polarization . the value of the weak decay parameter @xmath607 is well measured @xcite . aleph @xcite have exploited this distribution to measure a value @xmath608 for leading ( @xmath609 ) @xmath121 baryons . when all sources of @xmath121 baryons are taken into account , a result of @xmath610 is expected if @xmath121 s containing a primary s quark carry all of its initial polarization . thus the aleph measurement is in agreement with standard electroweak theory together with the assumption that the initial strange quark polarization survives hadronization to become a leading @xmath121 baryon . this result is clearly at odds with the conclusion from the study which indicated significant depolarization in the hadronization . indeed , the polarization of the heavier b quark may be expected to survive more easily than that of the s quark . study of spin alignment of vector mesons , particularly at large @xmath50 , where the meson may be expected to contain one of the primary quarks , may provide information on the nature of the quark to meson transition . such analyses are normally done in terms of the vector meson helicity density matrix , @xmath611 , some of whose elements can be determined by measuring the distribution of the vector meson decay products @xcite . the element 00 is the fraction of mesons which are in the helicity zero state . in strong vector meson decays it is not possible to infer separately the values of elements @xmath612 and @xmath613 , since the decay angular distributions are the same for helicity @xmath614 and helicity @xmath615 vector mesons . in statistical models @xcite , the fragmentation is assumed to produce extra quarks with both helicities equally likely . parallel alignment of primary and secondary quark spins will produce a vector meson with helicity @xmath616 . if the spins are initially antiparallel , the value of 00 will depend on the relative probability to produce a vector or a pseudoscalar meson ; in this case 00@xmath617 , with a maximum value of 1/2 when @xmath618 and a value of 1/3 if there is no suppression of vector mesons . in the model of @xcite vector mesons are produced via vector currents q@xmath619qv which conserve the quark helicity . the vector meson then has helicity zero . another model @xcite , for production of leading mesons , assumes multiple emission of soft gluons by the fragmenting quark , a process which conserves the quark helicity . the leading vector meson may be formed when the leading quark combines with a soft antiquark in a process which results in meson helicity @xmath620 . the basic string and cluster models have little to say about vector meson spin alignment , although @xcite does point out that no alignment may be expected in the simplest string picture . aleph @xcite , delphi @xcite and opal @xcite have measured the alignment of b@xmath429 mesons using the angular distribution of @xmath100 rays from the decay b@xmath621b@xmath100 . the measurements agree , and produce a weighted average of 00@xmath622 ( although they all express the result in terms of a relative contribution of longitudinal polarization states ) . these results , which imply no spin alignment , are consistent with simple spin counting and with heavy quark effective theory ( hqet ) . they are also in accord with the measurements of the ratio b@xmath429 to b meson production @xcite which imply no suppression of the vector state ( see [ sec - vtop ] ) . results from opal @xcite on lighter vector mesons show deviations from 00@xmath623 . for d@xmath624 mesons , a value 00 @xmath625 has been measured , consistent with lower energy results @xcite . and for primary @xmath407 mesons at @xmath626 , opal report an even larger value , 00@xmath627 . these mesons therefore appear to be preferentially in the helicity zero state . measurements of some off - diagonal elements of the helicity density matrix show small deviations from zero for both mesons . such non - zero off - diagonal elements are a natural consequence of coherence in the hadronization , and are a firm prediction @xcite of any general model other than independent fragmentation . there is clearly more to be learned about hadronization from such measurements . correlations in hadronic systems may be defined as departures from phase space in distributions for groups of two or more particles . such correlations may be associated with * quantum mechanics bose - einstein and fermi - dirac effects * hadron dynamics resonances , reflections and final - state interactions * local baryon number conservation baryon - antibaryon phase space correlations * local strangeness conservation strange particle rapidity correlations * soft gluon coherence in qcd showers two - particle momentum correlations and possibly intermittency * fragmentation dynamics transverse - momentum limitation of phase space in so far as one wishes to understand the underlying dynamics of the hadronization phase , it is important to take into account all of these effects . bose - einstein correlations ( becs ) have been extensively studied in hadronic systems from z@xmath0 decay . a review may be found in @xcite . the correlations are interesting in their own right as a quantum mechanical phenomenon whose experimental details can give information on the space - time structure of the source of hadrons . the correlations arise from the necessity to symmetrize the wavefunction for systems of two or more identical bosons . in most experimental analyses , a simple model is assumed where the source of particles is spherical with a gaussian density . then the two - particle phase space is enhanced by a factor @xmath628 relative to its density in the absence of the correlations . here @xmath52 , the square of the 4-momentum difference between two bosons , is the measure of the separation in phase space , @xmath342 measures the degree of coherence in the particle emission ( @xmath629 corresponds to full coherence ) and @xmath72 is the radius of the gaussian source . particle system & r ( fm ) & @xmath342 & references & comments + @xmath630 & @xmath631 & @xmath632 & a @xcite & + & @xmath633 & @xmath634 & d @xcite & direct @xmath363 + & @xmath635 & @xmath636 & o @xcite & + @xmath637 & @xmath638 & @xmath639 & d @xcite & + k@xmath553k@xmath553 & @xmath640 & @xmath641 & d @xcite & + k@xmath642k@xmath642 & @xmath643 & @xmath644 & a @xcite & not corrected for f@xmath255 + & @xmath645 & @xmath646 & d @xcite & + & @xmath647 & @xmath648 & o @xcite & + becs have been studied at lep for systems of @xmath630 , @xmath637 , @xmath649 , k@xmath553k@xmath553 and k@xmath650k@xmath650 . a summary of measured values of @xmath342 and @xmath72 is given in table 7 . it is not straightforward to compare the various results , nor to interpret the findings . for example , in order to measure the enhancement @xmath651 due to becs it is necessary to know the phase space density in their absence , and there are several ways to tackle this . in addition particles which arise from resonance or weak decays may be removed ( if the detector has the capability ) . the bec effect will be diluted in data samples containing mixtures of different particle types , and the purity of pion or kaon samples may be increased , depending again on the capabilities of the detector . in the lep analyses , various different approaches to these problems have been used . nevertheless , in all cases it is clear that the model based on a spherical source with gaussian density gives a reasonable fit to the observations , although there is no real evidence that other models would not fit equally well . the coherence parameter @xmath342 varies from close to zero to greater than one ; the latter is unphysical and possibly indicates deviations from the model . the source has a typical size of 1 fm , which at first sight seems inconsistent with a picture of hadrons arising from a rapidly expanding linear colour string . however the length scale measured by the becs is not the longitudinal size of the string , but the distance in production points for which particles are close together in momentum space @xcite . recently bec studies at lep have been extended to three - particle systems @xcite , and the multiplicity dependence of the correlations has been investigated @xcite . even if all of the measurements had been made in a consistent way , the interpretation of the results would not be straightforward : the correlations depend in a poorly understood way on final - state interactions , resonance production and rescattering of resonance decay products . according to @xcite the situation may be `` impossibly complicated '' . and @xcite says `` it seems very difficult to make progress in studying the bose - einstein effect in the context of physics , and it is not clear to what extent it can be considered a useful and interesting activity . '' it is nevertheless important to understand at least the phenomenology of the bose einstein correlations since they impact on studies of other features of hadronic systems . for example , although the effect primarily influences systems of identical bosons , in the relatively high - multiplicity , jet - like environment of z@xmath0 decay , `` residual '' correlations @xcite arise between pairs of unlike particles . the becs produce a general collimation of the jets and a tendency to reduce the mean transverse momentum ; this brings all particles closer together in momentum space . the effect is to produce a distortion of @xmath96 mass spectra , especially at low momentum where the multiplicity is highest . this means for example that the use of opposite charge particle pairs to determine the phase space in the absence of becs can result in biased values of @xmath342 and @xmath72 . an important practical result of the residual correlations is the considerable difficulty in inclusive measurements of @xmath96 resonances @xcite such as the @xmath397 and the f@xmath652 whose line shapes are distorted by the correlations and possibly also by other mechanisms @xcite . in the process at lep 2 , correlations may arise between the hadrons from one w and those from the other , an effect recently investigated by delphi @xcite . such an effect could result in a shift of the reconstructed w mass in multihadronic w decays @xcite . a good understanding of the role of becs in hadronization at the z@xmath0 may help to reduce uncertainties introduced by this effect . while becs between identical bosons are firmly established experimentally , anticorrelations between identical fermions are more difficult to observe . in z@xmath0 decay , baryons are produced at much lower rates than mesons , and local conservation of baryon number suppresses production of identical baryons close together in phase space . the only evidence for anticorrelations comes from a study by opal @xcite of @xmath653 , @xmath654 @xmath654 and @xmath655 pair production . close to threshold it is found the the spin states of the di - hyperon systems agree with expectations of a simple statistical mixture , with no indication of any resonance in the @xmath655 threshold enhancement . in the case of the identical baryon pairs , there is a tentative indication of suppression close to threshold , as would be expected by the pauli exclusion principle . however , further measurements are needed to confirm this result . since baryons , and strange hadrons , are both heavier and less frequently of secondary origin than ordinary hadrons investigating their production and pairwise correlations appears to offer a more direct probe of the momentum and quantum number flow during hadronization . however , allowance must always be made for secondary hadrons coming from decays , such as @xmath656 ( here further data @xcite on @xmath657 correlations would be welcome ) . baryon number conservation implies that a baryon is always accompanied by an antibaryon ; flavour conservation , via diquark pairs , suggests that the baryons may be preferentially particle - antiparticle pairs ; and lphd argues that these baryons occur close by in phase space . so far , studies have concentrated on measuring the proximity in phase space of @xcite , @xcite and @xmath657 @xcite baryon pairs . by introducing an event axis , typically the thrust or sphericity axis , correlations can be studied in rapidity , azimuthal angle or polar angle with respect to the axis . several models of baryon production are available @xcite . discusses the independent fragmentation @xcite , the diquark @xcite and popcorn @xcite ( triplet ) string options and the cluster @xcite models . in addition a possible contibution from direct , @xmath658 , diquark production has been proposed @xcite . also there is the recombination model of hadron production @xcite . in this approach a spectrum of partons occuring after a shower at the fixed scale @xmath659 is convoluted with an explicit wavefunction for a particular hadron . in the case of a baryon @xcite this gives ( c.f . ): @xmath660 here a typical wavefunction is @xmath661 , consistent with reciprocity @xcite and quark counting rules @xcite , and on the assumption of uncorrelated partons @xmath662 . it may be mentioned in passing that the recombination model makes interesting predictions for baryon polarizations @xcite . at the z@xmath0 strong , short - range correlations in rapidity are seen between baryon - antibaryon b@xmath663 pairs . ( unlike - baryon distributions have the like - baryon ( bb ) distributions subtracted to remove secondary correlations due to more than one b@xmath307 pair in an event . ) for example , the distribution of baryon rapidity ( @xmath143 ) in such events , @xmath664 , is measured to be compact and centred on @xmath665 @xcite . in a b@xmath307 pair , given the rapidity of the @xmath666 there is @xmath667% probability that the p ( @xmath121 ) will be found within @xmath668 and vice versa @xcite ; a marginally weaker short - range @xmath669 correlation is found @xcite . a much weaker long - range anticorrelation is also seen for far forward / backward b@xmath307 pairs , as anticipated from a leading particle effect . in contrast , when bb pairs occur in an event , the @xmath670 distribution is nearly flat with just a weak short - range anticorrelation , particularly so away from the central region @xcite where phase space constraints become important . similar , though statistically limited , results have been seen at lower energy @xcite . qualitatively these features are reproduced by both the herwig and jetset models , though quantitatively the strength of the correlations is overestimated by herwig and by jetset without popcorn @xcite . the recombination model predicts rapidity correlations which are much too weak whilst direct diquark production predicts long - range correlations that are too strong ; both are disfavoured by the data @xcite . somewhat weaker correlations are seen between the azimuthal angles in baryon pairs at the z@xmath0 @xcite . b@xmath307 pairs show a tendency towards @xmath671 , though in those pairs which lie out of the event plane there is a tendency towards @xmath672 . bb pairs show a weaker tendency towards @xmath672 . two competing mechanisms might be envisaged to explain the azimuthal angle correlations . first is a local compensation of transverse momentum which leads to an enhancement for @xmath672 . second is a tendency for any off - axis boost to be shared by neighbouring baryons , leading to an enhancement at @xmath671 . the latter effect might be expected to become more important at higher energies where three - jet effects come into play ( these by definition involve higher transverse momentum scales than hadronization ) , more so for baryons lying in the event plane . at low com energy ( 10 gev ) the first , back - to - back , effect appears to dominate @xcite but this is seen to weaken at higher energy ( 30 gev ) @xcite and at the z@xmath0 the second , side - by - side , effect is more important . historically a very powerful way to discriminate between hadronization models is the orientation of the b@xmath307 pair with respect to the event axis @xcite . if @xmath5 is the angle between the axis of the b@xmath307 pair and the event axis as seen in the b@xmath307 pair rest frame then @xmath673 is measured to be highly forward - backward peaked @xcite . by utilizing the lepton beam polarization information ( see section [ selwk ] ) sld are capable of determining the primary quark direction and have recently been able to demonstrate that the baryon preferentially follows the quark direction @xcite . in string models the colour field typically aligns along the event axis , so that provided that the transverse momenta acquired by the baryons are small compared to the longitudinal momenta transferred from the string , the baryons will retain a strong memory of the string / event axis direction . in contrast , because clusters are deemed to be structureless , they have no means of retaining any information about an original event axis ( they are unpolarized ) and so decay isotropically in their rest frame . not surprisingly then the highly peaked @xmath673 strongly disfavours the essentially flat prediction from herwig whilst jetset offers a good description of the data . in herwig version 5.7 an option was introduced to allow non - isotropic decays of clusters containing primary quarks ; this was implemented to stiffen the momentum spectrum of c and b hadrons and has little influence on b@xmath307 pair polar angle distributions . in the previous section , [ sbcor ] , phase space correlations were discussed for baryons ; we now turn to similar measurements made on strange hadrons . the conservation of strangeness in strong interactions implies that strange hadrons are pair - produced during hadronization . the probability that hadron @xmath674 is accompanied by @xmath675 is defined as @xmath676 where the 2 is included because of double counting . measurements give @xmath677% whilst @xmath678% @xcite , @xmath679% and @xmath680% @xcite , @xmath681% @xcite and @xmath682% @xcite . two sources of strange hadrons can be anticipated : leading hadrons associated with initial s@xmath683 quarks and those pair - produced locally in the event s colour field . to help distinguish these possibilities requires information on the phase - space correlations between pairs of strange hadrons . a short - range rapidity correlation occurs between @xmath684 pairs and a slightly weaker one between @xmath685 pairs @xcite indicating a local mechanism for strangeness compensation . also visible are : weaker long - range correlations , as expected from leading quarks ; and evidence of phase - space suppression , particularly when both hadrons are leading . weak correlations are also seen at @xmath686 for centrally produced @xmath684 and @xmath685 pairs . the possible influence of introducing the popcorn baryon production mechanism , see , is of particular interest for string models . the presence of an intermediate meson , bm@xmath307 tends to soften all correlations . at present , taking account of systematic errors , measurements of baryon and strangeness correlations are insufficient to place any significant constraint on the level of popcorn production required . the b@xmath307 pair rapidity difference is mainly sensitive to the amount of popcorn production , with data favouring a substantial component @xcite . the substantial rate of @xmath687 pair production also favours a high level of popcorn production in order to supply an intermediate kaon @xcite . interestingly the number of b@xmath688 pairs decreases linearly with the amount of popcorn introduced so that their measured multiplicity can be ( rather simplistically ) used to constrain the amount of popcorn to around 50% @xcite . a more direct test of the popcorn mechanism is to look at rapidity ordered bm@xmath307 triples @xcite . using nearby baryon pairs , @xmath689 , a probability of 7% ( 25% ) is obtained for finding an intervening kaon ( pion ) in a or @xmath669 pair . in addition to there being no enhancement of popcorn - favoured @xmath121k@xmath690 triples , unfavoured @xmath121k@xmath691 triples are found equally likely . however , in events containing a kaon and a close @xmath121p pair , the kaon is found very close in rapidity to the baryon pair . intermittency @xcite , the non - random clustering of particles in phase space , is a somewhat obscure phenomenon of uncertain dynamical origin . in essence , intermittency corresponds with large , non - statistical , fluctuations in the numbers of particles in particular events which are found in narrow rapidity bins . it is normally studied by measuring factorial moments of multiplicity distributions in rapidity bins . intermittent behaviour has been observed in hadronic systems from z@xmath0 decay at both lep @xcite and slc @xcite as well as in lower energy collisions . the intermittency observed in hadronic z@xmath0 decays is in fact reproduced by the lund parton shower monte carlo model with string fragmentation . it is a possibility that the self - similarity inherent in the qcd parton shower , with its successive q@xmath619qg and g@xmath619gg branchings , is the source of the intermittency seen in the distribution of the final - state hadrons . certainly the jetset model contains no feature explicitly introduced to simulate the dynamics of intermittency . the appearance of intermittency in the z@xmath0 decay data would seem to be an ideal opportunity to gain a good understanding of the mechanisms which lie behind it . previously its interpretation in hadroproduction experiments has been obscured by the complicated nature of the final states and the effects of beam and target fragments . the determining property of quark and gluon jets is the colour charge of the initiator partons , @xmath692 and @xmath693 respectively . due to the gluon s larger charge it should radiate more subsequent gluons in a parton shower . this leads to the anticipation that gluon jets , as compared to quark jets , will have : a higher multiplicity , softer momentum spectrum and wider angular distribution @xcite . these global features are largely borne out by experimental results @xcite . however whilst clear differences are now established between gluon and light - quark jets , b - jets appear to be rather like gluon jets at z@xmath0 energies @xcite ; the ratios of measured properties typically fall short of the naive asymptotic predictions . for example the multiplicity ratio is , in leading order , predicted to be @xmath694 @xcite ; at nnlo this becomes @xmath695 @xcite and after imposing energy ( but not momentum ) conservation on the shower this drops to @xmath696 @xcite ( the exact predictions depend on jet energy and the scale used for @xmath41 ) . the measured ratio , which is seen to be sensitive to the precise jet definition , is typically in the range @xmath697 , though opal has obtained 1.55 in an event hemisphere - based analysis @xcite . it should be borne in mind that the these basic predictions are made for back - to - back pairs of quark or gluon jets whilst growing evidence suggests that the relative topology of a jet is important in determining the appropriate scale @xcite . the above discussion relates to the perturbative properties of quark and gluon jets . current cluster and string models of hadronization make no distinction between whether a set of final - state partons arose from a fragmenting primary quark or gluon ; they are treated the same . this does allow the possibility of a ` leading particle ' effect ( for example one should expect more leading kaons in an s - quark jet than a gluon jet ) but no other ` anomalous ' effects @xcite . an interesting possibility is that some isoscalar states , @xmath698 , may contain a significant gg component and hence might appear more frequently as leading particles in gluon jets . these particles are very often primary hadrons coming directly from a cluster or string . at present the possibility of hadrons containing gluons is not allowed for and only quark constituents are considered . indeed in the cluster framework gluons are split into light pairs whilst in the string approach they represent energy - momentum ` kinks ' on a string . however modifying the models to accommodate gluonium would not , _ a priori _ , appear to pose significant problems of principle or practice . a more radical scenario is offered by the independent fragmentation model of peterson and walsh @xcite . this is based on the suggestion that a gluon is attached to an octet colour flux tube and quarks to triplet flux tubes ( see ) . a gluon now fragments into a sequence of isoscalar gg or gq@xmath192 g clusters leading to a prediction of greatly enhanced @xmath699 etc production and harder momentum spectra as compared to quark jets . a third alternative scenario for gluon jet hadronization is provided within the recombination model @xcite discussed in . calculations predict : a softer pion spectrum @xcite , an enhanced multiplicity ratio @xmath700 @xcite and enhanced baryon production @xcite in gluon jets compared to quark jets . the l3 collaboration have reported tentative indications that @xmath293 production is enhanced in gluon jets @xcite . studying the lowest energy gluon jet in three - jet events they see a a harder @xmath293 momentum spectrum than predicted by both the herwig and jetset monte carlos with an enhancement in the ratio @xmath701 . the use of a ratio takes into account the established increase in multiplicity found in gluon compared to quark jets and is designed to make the measurement sensitive to any additional enhancement or suppression . the monte carlos provide a satisfactory description of the spectrum and @xmath702 ratio in quark jets . delphi have studied the production rates of kaons , @xmath703 and @xmath704 in multi - jet events @xcite . they find that the relative yields of strange hadrons in multi - jet , normalized to two - jet , events is constant in events with widely separated jets but favours increased production in multi - jet events at small resolutions , particularly so for kaons . in a more direct study delphi have looked at identified particle production rates in actual quark ( natural flavour mix ) and gluon jets , normalized to the charged multiplicity in the jet : the double ratios @xmath705 @xcite . the evidence suggests values for the double ratios of approximately @xmath706 for the k@xmath0 , @xmath707 for k@xmath3 , @xmath708 for p and @xmath709 for @xmath121 . in general these results are in qualitative agreement with the monte carlos but quantitatively the deviations from unity are larger than typically predicted , apart for the k@xmath0 result where a small suppression was expected . however the errors are relatively large . opal have also reported preliminary studies of the double ratios @xcite . for k@xmath365 and @xmath352 mesons they report a slight , @xmath710% , increase in the relative production rates in gluon jets , whilst for p and @xmath121 baryons they measure a significant 3040% increase . these enhanced production rates are presumably at the expense of pion production . at present more measurements are required before hard conclusions can be drawn . the possibility of enhanced identified hadron production in gluon as compared to quark jets has received previous attention . the results for mesons , in particular hard @xmath293 relative to @xmath103 , production are not conclusive . at @xmath711 gev argus @xcite saw no evidence for enhanced @xmath293 or @xmath352 production in the continuum , @xmath712q@xmath192 . also comparison has been made between continuum and @xmath713ggg events ; here crystal ball @xcite see no enhancement whilst dasp - ii , cleo and argus @xcite see a slight enhancement for a number of mesons . at @xmath714 gev jade @xcite reported very weak ( statistically insignificant ) evidence for a small enhancement in the @xmath702 ratio in acollinear , gluon rich , events . the situation is clearer for baryons . the production rate of baryons in @xmath715 decay ( gluon dominated ) to continuum ( quark dominated ) events shows an excess of 200300% @xcite . in the context of conventional monte carlos a number of partial explanations have been offered . a study using jetset indicates that the relative production rates of mesons in quark and gluon jets is energy independent and just less than one , whilst for baryons it is 2025% larger and shows a slight increase with jet energy @xcite . this latter effect may be attributed to an edge effect associated with the suppression of leading baryons . in cluster models it has been argued that in gluon rich environments , the topology , rather than any intrinsic properties of the jets , leads to heavier clusters and hence larger baryon production rates @xcite . at the @xmath335 it has also been emphasized @xcite that secondary decays are important , and that 40% of the events are in the continuum to b and c quarks ( see ) which , being heavy , ` eat up ' the available phase space for baryon production . that monte carlo event generators , solidly based on sound physics , are essential in modern high energy experiments , from detector conception to data analysis , ought not to be forgotten . it therefore almost goes without saying that the reliability of monte carlo predictions should be a prime concern if only for mere practical reasons . to this end the large event rates and pristine conditions available in hadronic z@xmath0 decays at colliders play a particularly important role . here , as nowhere else , the physics assumption built into the monte carlo models can be confronted with ever more exacting tests . the relatively complex conditions associated with initial - state hadrons have largely precluded this activity using ep and p@xmath716 data , thereby making physicists reliant on the quality of the z@xmath0 data . two caveats to the wider application of the models are the questions of the reliability of the factorization theorem in perturbative qcd and the presumption of universality in the hadronization process . focusing on the models for the hadronization processes one may ask what is left that should be done with the z@xmath0 data and what might be learned from it . we list a number of topics , perhaps not all of which can be done with available data ( but one can always hope for more z@xmath0 data ) : * a study of the transverse momentum distributions of identified particles , especially pions where low mass effects and correlations may occur @xcite in string models the fact that the predicted width of the gaussian @xmath230-distribution , @xmath717 , proves too narrow has raised questions about the adequacy of a tunnelling mechanism explanation : is unresolved ( non - perturbative ) gluon emission @xcite the real explanation ? * an attempt to establish properties of the relatively rare , directly produced pions pions are special particles by virtue of their nature as goldstone bosons ( of the chiral symmetry ) ; this mandates them to have small masses which in a string model implies a very small size , in fact uniquely less than a string s width . * further measurements of orbitally excited mesons and baryons are the production rates for these states too high to be compatible with the string model ? * a search for d - wave states , such as the k@xmath718 these could help elucidate the role of the wavefunction in determining a hadron s production rate . * a study of f@xmath404 production as a function of rapidity and of total event multiplicity this would test the gribov confinement scenario @xcite . * a search for deuterium production this has been reported previously in studies at the @xmath335 @xcite and is expected , on the basis of a string model calculation @xcite , at the level of @xmath719 per hadronic z@xmath0 event . * further measurements of strangeness suppression ( @xmath484 ) which ideally can be directly compared to those available in ep collisions , particularly those associated with the current region of the breit frame @xcite if it turns out that @xmath720 then effort should go to establishing any other differences . * a measurement of the s - quark to p and @xmath721 fragmentation functions is there a measurable difference as suggested @xcite by the possible asymmetry in sea quark structure functions ? * measurements of identified particle production rates in quark and gluon jets are there measurable differences and , if so , what is the mechanism ? * a search for glueball candidates , particularly in the gluon - rich environment provided by gluon jets . * further measurements of leading baryon polarization is there a depolarizing mechanism for the b baryons which does not apply to those arising from s quark fragmentation ? * measurements of the helicity density matrix for the light and heavy vector mesons does hadronization produce spin - aligned vector mesons , and does any alignment depend on @xmath50 ? are some off - diagonal elements non - zero as predicted by coherent hadronization ? spin physics has a long history of producing surprises . * more detailed studies of intermittency can it be firmly established that the phenomenon is due to the self - similar evolution of the initial state via a qcd parton shower ? * a search for direct evidence of popcorn - type baryon production via , for example , further study of @xmath121k@xmath313 type correlations three - body cluster decays would also induce such correlations but perhaps with different intensity . * a search for @xmath722 type correlations these are possible in a generalized popcorn mechanism although they will be hard to find experimentally because of low rates . * an attempt to establish directly the existence of the ` dead cone ' @xcite in non - leading particles , by removing the leading particles using fully reconstructed b - hadron decays . * direct measurement of the rates and momentum spectra of hadrons produced in b - quark events to ensure adequate descriptions of the weak decays in the monte carlos b hadrons contribute a tenth of all particles in hadronic z@xmath0 decays , and significantly more at large @xmath50 . * a tuning of the monte carlo models to z@xmath0 data , incorporating colour rearrangement this is important in order to determine the effects of colour rearrangement and to provide constraints which may prove significant for later w mass measurements . * a simultaneous tuning of the monte carlo programs to the z@xmath0 data and lower energy data , particularly from pep and petra at lower energies the relative contribution to event properties made by hadronization is more important , whilst the influence of the perturbative shower can be tested using the com energy dependence of observables . * an ` ultimate ' tuning of the monte carlo models to the final z@xmath0 data this will be an invaluable service to experimentalists and theorists allowing the experience gained from lep1/slc to be applied at future machines . * a continuation of the search for tests which discriminate between the competing models of the non - perturbative physics underlying the hadronization , and continuing development of these models . in summary , the data have already provided a wealth of information on the phenomena of parton hadronization . of the available models of the non - perturbative physics involved , the lund string model , as implemented in jetset , has met with most success , particularly in the baryon sector , and most notably in its prediction of the angular distribution of correlated baryon - antibaryon pairs in their rest frame . however , the model has many free parameters and consequently has little predictive power . but the parameters are not arbitrary most are based on incomplete knowledge of physics . therefore it could be argued that their values , when fully tuned to reproduce observations , provide important information about hadronic physics . on the other hand , while the weight of evidence tends to favour the string picture , the other models , particularly the cluster model of herwig , are not dead , and more analyses of the existing and future data are essential to provide further discrimination between the models and to help elucidate the physics of hadronization . igk wishes to thank n. alguacil conde for encouragement . 999 dokshitser yu l , khoze v a and troyan s i 1989 _ perturbative qcd _ ed a h mueller ( world scientific : singapore ) p 241 + dokshitser yu l , khoze v a , mueller a h and troyan s i 1991 _ basics of perturbative qcd _ ( gif - sur - yvette : editions frontires ) gross d j and wilczek f 1973 1343 and 1973 3633 + politzer h d 1973 1346 + tarasov o v , vladimirov a a and zharkov a yu 1980 429 + van ritbergen t , vermaseren j a m and larin s a 1997 preprint archive hep - ph/9701390 sding p and wolf g 1981 _ ann . nucl . part . sci . _ * 31 * 231 + duinker p 1982 _ rev . phys . _ * 54 * 325 + wu s l 1984 _ phys . rep _ * 107 * 59 + mark - j collaboration , adeva b 1984 _ phys . rep . _ * 109 * 131 + narosoka b 1987 _ phys . _ * 148 * 67 aleph collaboration , decamp d 1991 623 and 1991 479 + delphi collaboration , abreu p 1992 55 + l3 collaboration , adriani o 1992 471 + opal collaboration , acton p d 1992 1 + sld collaboration , abe k 1994 5580 and 1995 4240 krzywicki a and petersson b 1972 924 + finkelstein j and peccei r d 1972 2606 + niedermayer f 1974 355 + casher a , kogut j and suskind l 1974 732 + hoyer p 1979 349 + ali a , pietarinen e , kramer g and willrodt j 1980 155	the confinement transition from the quark and gluon degrees of freedom appropriate in perturbation theory to the hadrons observed by real world experiments is poorly understood . in this strongly interacting transition regime we presently rely on models , which to varying degrees reflect possible scenarios for the qcd dynamics . because of the absence of beam and target remnants , and the clean experimental conditions and high event rates , annihilation to hadrons at the z@xmath0 provides a unique laboratory , both experimentally and theoretically , for the study of parton hadronization . this review discusses current theoretical understanding of the hadronization of partons , with particular emphasis on models of the non - perturbative phase , as implemented in monte carlo simulation programs . experimental results at lep and slc are summarised and considered in the light of the models . suggestions are given for further measurements which could help to produce more progress in understanding hadronization . feynman 00@xmath1 european laboratory for particle physics cern - ppe/97 - 040 + edinburgh - ppe/97 - 01 + man - hep/97 - 02 + 16th april 1997 topical review to be published in journal of physics g : nuclear and particle physics
the value of compilation , evaluation and computer storage of neutron cross section data was first recognized in the early 50 s @xcite prompted by the urgent needs of nuclear industry . these cross sections were summarized in bnl-325 report and evaluated nuclear data file ( endf ) library @xcite . over the years , the nuclear data activities were extended to all low- and intermediate - energy nuclear physics topics . computer storage and worldwide dissemination of nuclear data strongly depend on the presently available computer technologies @xcite and data format developments . in present days , size and representation of nuclear data files are no longer limited by the computer hardware and software . however , for historic and consistency reasons , nuclear data are still stored in 80-character long formats . for years , nuclear data research and developments were driven by nuclear science and technology applications . these developments evolved and lead to the creation of complementary products such as experimental nuclear reaction data ( exfor ) database @xcite . the overall nuclear data improvements , maturation and modern computer technologies created many new opportunities for nuclear reaction calculations , data mining and analysis . for the capture @xmath0-ray spectroscopy purposes , we will concentrate on the selected nuclear astrophysics and reactor operation integral quantities that can be extracted from the beta version of endf / b - vii.1@xmath13 library . current results will be finalized by the end of this year during the public release of endf / b - vii.1 library @xcite . 3 thermal neutron capture cross sections . ] neutron capture cross sections govern the production of chemical elements in the agb and red giant stars , safe operation of nuclear reactors , serve in nuclear structure measurements , etc . the low - energy neutron cross section values are often influenced by the contributions from resolved and unresolved resonance regions . to estimate these contributions across the whole endf / b - vii.1 library range of nuclei @xcite and provide additional insights on the data quality for nuclear reactor and astrophysics applications , we have selected thermal and maxwellian - averaged cross sections , resonance integrals and westcott factors @xcite for calculation and further analysis . endf / b - vii.0 @xcite and endf / b - vii.1@xmath13 evaluated neutron cross sections were doppler broadened using the code prepro @xcite with the precision of 0.1% . these reconstructed and linearized data were used to calculate the selected quantities using the definite java integration method @xcite . the ratio of thermal @xmath2(n,@xmath0 ) is shown in fig . [ fig1 ] . using the method of visual inspection we notice the deviations for light and medium nuclei and minor actinides evaluations . these differences , in the low- and medium - z region , are attributed to the lack of or insufficient experimental data for @xmath3b , @xmath4o , @xmath5ca , @xmath6kr , @xmath7pd , deficiencies for @xmath8co,@xmath9mo and recent re - evaluation of @xmath10zr . while , in the actinide region , deviations are due to new evaluations from the actinoid file @xcite . neutron capture resonance integrals were calculated for 0.5 ev - 20 mev incident neutron energy range and shown in fig . [ iq.3 ] . 3 thermal neutron capture resonance integrals . ] several data outliers in this case could be traced to the lack of measurements and incomplete overlap of experimental and theoretical data for @xmath4o , @xmath11ru , @xmath12ho and @xmath13ca , @xmath8co , @xmath14cs , @xmath15hg , respectively . however , there are neutron capture cross section deficiencies in the kev region of energies for @xmath16o,@xmath17si and @xmath18pb evaluations . maxwellian - averaged cross sections play an important role in power reactor developments and slow - neutron capture ( _ s_-process ) nucleosynthesis calculations @xcite . the @xmath19-process is mostly responsible for element formation in stars from @xmath20fe to @xmath21bi . the detailed analysis of the fig . [ iq.5 ] data @xcite demonstrates the nuclear astrophysics potential of endf libraries as a complimentary source of evaluated cross sections and reaction rates . 3 maxwellian - averaged cross sections at _ kt_=30 kev . ] there are noticeable differences between kadonis @xcite and endf / b - vii.1@xmath13 libraries for light and medium nuclei . @xmath22h deviation is due to differences between center of mass and lab system cross section values . for @xmath23si , @xmath24p , @xmath25ni and @xmath26hg kadonis values are based on a single recent measurement . due to lack of experimental data theoretical values were adopted for @xmath27ar , @xmath28se , @xmath29cd , @xmath30ce , @xmath31pr and @xmath32pm . deficiencies in @xmath16o , @xmath33ca and @xmath34s originate from the old or insignificant for integral tests endf evaluations and coverage problems in exfor database @xcite , respectively . shown in fig . [ iq.6 ] ratio of capture westcott factors indicate large deviations for @xmath35u and @xmath36lu . 3 libraries . ] these deviations reflect the changes in endf / b - vii.1@xmath13 library where westcott factors evolved from 3.997 to 0.989 and from 1.002 to 1.711 for @xmath35u and @xmath36lu , respectively . the last number agrees well with the recommended value of 1.75 @xcite . smaller deviations as in @xmath37xe are due to adoption of new evaluations in endf / b - vii.1@xmath13 library and lack of experimental data for this nucleus . the present work demonstrates large potential of endf libraries for capture @xmath0-ray spectroscopy applications . maxwellian - averaged and thermal capture cross sections , resonance integrals and westcott factors have been extracted from the beta version of endf / b - vii.1@xmath13 library . these results are important in fundamental science , nuclear technologies and for nuclear data validation . several @xmath1-version @xmath2(n,@xmath0 ) deficiencies ( @xmath16o , @xmath8co , ... ) will be resolved during the official library release . the new endf / b - vii.1 library @xcite will be publicly available in december of 2011 at the national nuclear data center website _ http://www.nndc.bnl.gov _ future nuclear reaction data mining work will involve extensive analysis of neutron cross section covariance files and extended coverage for neutron capture , fission and elastic scattering and all major neutron libraries . the author is grateful to m. herman , s.f . mughabghab and m. blennau ( bnl ) for support of this work , productive discussions and careful reading of the manuscript and useful suggestions , respectively . this work is supported by the office of nuclear physics , office of science of the u.s . department of energy , under contract no . de - ac02 - 98ch10886 with brookhaven science associates , llc . 20 s. pearlstein , nuclear news , 73 no . 11 november ( 1970 ) . mughabghab , atlas of neutron resonances , resonance parameters and neutron cross sections z = 1 - 100 , elsevier ( 2006 ) . m.b . chadwick , p. obloinsk ' y , m.w . _ , nuclear data sheets * 107 * , 2931 ( 2006 ) . b. pritychenko , a.a . sonzogni , d.f . et al . _ , energy * 33 * , 390 ( 2006 ) . international network of nuclear reaction data centres ( nrdc ) , compilation of experimental nuclear reaction data ( exfor / csisrs ) , ( available from _ _ h__ttp://www - nds.iaea.org / exfor/ , http://www.nndc.bnl.gov/exfor/ ) . chadwick , m.w . herman , p. obloinsk ' y _ et al . _ , nuclear data sheets * 112 * , no.12 ( 2011 ) . b. pritychenko , s.f . mughabghab , and a.a . sonzogni , atomic data and nuclear data tables * 96 * , 645 ( 2010 ) . cullen , the endf / b pre - processing codes ( prepro ) , " available from @xmath38http://www - nds.iaea.org/ ndspub / endf / prepro/@xmath39 . b. pritychenko , proc . nuclei in cosmos , pos(nic - xi)197 , heidelberg , germany , july 19 - 23 ( 2010 ) . o. iwamoto , t. nakagawa , n. otuka _ _ , nuclear data sheets * 109 * , 2885 ( 2008 ) . bao , h. beer , f. kppeler _ et al . _ , atomic data and nuclear data tables * 76 * , 70 ( 2000 ) . i. dillmann , m. heil , f. kppeler _ et al . _ , aip conf . proc . * 819 * , 123 ( 2006 ) ; data downloaded from @xmath38http://www.kadonis.org@xmath39 on april 14 , 2011 . mughabghab , m. divadeenam , and n. holden , _ neutron cross sections _ , vol . * 1 * , part a , academic press , new york ( 1981 ) . k. shibata , o. iwamoto , t. nakagawa _ et al . _ , * 48 * , 1 ( 2011 ) .	nuclear reaction data play an important role in nuclear reactor , medical , and fundamental science and national security applications . the wealth of information is stored in internally adopted endf-6 and exfor formats . we present a complete calculation of resonance integrals , westcott factors , thermal and maxwellian - averaged cross sections for z=1 - 100 using evaluated nuclear reaction data . the addition of newly - evaluated neutron reaction libraries , and improvements in data processing techniques allows us to calculate nuclear industry and astrophysics parameters , and provide additional insights on all currently available neutron - induced reaction data . nuclear reaction calculations will be discussed and an overview of the latest reaction data developments will be given .
when we estimate the present star formation rate ( sfr ) in galaxies , we use the luminosity of radiation from young massive stars as its indicator . various observational quantities are adopted in order to count the number of photons originating from these stars . indeed , we use the data of hydrogen recombination lines , ultraviolet ( uv ) , infrared ( ir ) , radio , etc . , as the indicators of the photon flux and the sfr ( e.g. , @xcite ) . observational evidence suggests , however , that star - forming regions are often associated with dust ( e.g. , @xcite , p. 125 ) . thus , we can not obtain the true sfr of galaxies unless we correct observational data for dust extinction . in other words , understanding the extinction property is important in estimating the sfr ( e.g. , @xcite ) . therefore , we should examine the properties of dust extinction in star - forming regions in detail . the extinction by the interstellar dust from uv ( @xmath6 ) to near - ir ( nir ) has been studied well to date ( e.g. , interstellar extinction curve : @xcite ) . according to those researches , it is widely accepted that when we estimate the sfr from the h@xmath7 luminosity , for example , we should correct decrement of the line luminosity due to the uv nir extinction . it can be performed by using a proper extinction curve and the observation of the balmer decrement ( e.g. , @xcite ) . on the other hand , how about the effect of the extinction in the lyman continuum band ? because of the decrement of the stellar radiation owing to the lyman continuum extinction ( lce ) , we may underestimate the sfr of galaxies . in this paper , this point is extensively examined . indeed , many observations of h ii regions have revealed that dust grains are associated with these regions ( e.g. , @xcite ) . this means a very important fact that the lyman continuum photons from the central young massive stars suffer extinction by dust grains within h ii regions . as a result , the number of lyman continuum photons contributing to hydrogen ionization is reduced . from the intensity of recombination lines or thermal radio radiation , we can only estimate the number of the photons really contributing to the ionization . therefore , we inevitably underestimate the number of lyman continuum photons and the sfr unless we correct the observational data for the lce by dust . in studies on each individual h ii region , it has been reported that a part of lyman continuum photons are absorbed by dust if h ii regions contain dust grains ( e.g. , petrosian , silk , & field 1972 ; @xcite ) . according to @xcite , for instance , the fraction of lyman continuum photons used by hydrogen ionization is estimated to be 0.26 for the orion nebula . their result has shown that the chance which the lyman continuum photons are extinguished by dust is significantly large . on a galactic scale , smith , biermann , & mezger ( 1978 ) have examined lce by dust for a large number of radio h ii regions in the galaxy , and then , have shown about half lyman continuum photons from exciting stars in many h ii regions are absorbed by dust within these regions . however , they did not discuss the effect of the lce on estimating the sfr of galaxies . unfortunately , it seems that we often estimate the sfr of galaxies from lyman continuum flux ( e.g. , @xcite ) without any correction of the lce effect . hence , it is indispensable for us to re - examine lce by dust within h ii regions and to discuss the effect of lce on determining the sfr of galaxies . first , we formulate the way of estimating the fraction of lyman continuum photons used by hydrogen ionization , and determine the fraction in each individual h ii region of the galaxy in section 2 . we also present another way for estimating lce by dust in section 3 . then , we discuss the effect of lce by dust on estimating the galaxy - scale sfr in section 4 . finally , our conclusions are summarized in the last section . the lyman continuum photons from young massive stars in an h ii region are absorbed by dust grains as well as by neutral hydrogen atoms . then , the number of lyman continuum photons contributing to hydrogen ionization , @xmath8 , is smaller than the intrinsic number of lyman continuum photons , @xmath9 . we parameterize this effect by @xmath1 , which is defined as the fraction of lyman continuum photons contributing to the hydrogen ionization ; @xmath10 . unfortunately , we are able to estimate only @xmath8 from observations of recombination lines or thermal radio continuum . this means that the `` real '' sfr , which should be estimated from @xmath9 , is @xmath11 times larger than the `` apparent '' sfr estimated from @xmath8 . hence , if the parameter , @xmath1 , is much smaller than unity , the effect of lce by dust becomes very important to the estimation of the sfr . we call @xmath11 _ correction factor _ or _ increment factor _ for the sfr . in this section , we examine the value of @xmath1 in each individual h ii region in the galaxy to find the correction factor quantitatively . first , when no dust grains exist in an h ii region , the radius of the region is estimated to be the strmgren radius , @xmath12 . according to @xcite , we express the number of lyman continuum photons emitted per unit time , @xmath9 , by using the strmgren radius , @xmath12 : @xmath13 where @xmath14 and @xmath15 are the number densities of electrons and protons , respectively , and @xmath16 is the recombination coefficient excluding captures to the @xmath17 level . here , we employ the case b approximation , which is the assumption that h ii regions are optically thick for the photons of all lyman - emission lines ( e.g. , @xcite ) . we also assume spherical symmetry and spatial uniformity of h ii regions in this paper for simplicity . next , we consider the effect of lce by dust on the size of h ii regions . the actual radius of the ionized region becomes smaller than the strmgren radius in equation ( [ eq1 ] ) by the lce effect . when we express the actual ionized radius as @xmath18 , we obtain @xmath19 moreover , if the ratio of @xmath18 to @xmath12 is expressed by @xmath20 , we obtain the following relation from equations ( [ eq1 ] ) and ( [ eq2 ] ) , @xmath21 to estimate @xmath1 , we need to find the quantitative relation between @xmath1 and the amount of dust . we determine @xmath1 in terms of the optical depth of dust for lyman continuum photons . according to @xcite , then , we define @xmath22 as the optical depth of dust for lyman continuum photons over the strmgren radius in order to formulate the dependence of @xmath1 on the dust - to - gas ratio . here , @xmath22 is approximated to be the optical depth at the lyman limit ( 912 ) . when we adopt the galactic extinction curve , the dust extinction at 912 is about 13@xmath23 mag . thus , @xmath24 . if we assume that the dust - to - gas mass ratio , @xmath25 , is proportional to @xmath26 , where @xmath27 denotes the column number density of hydrogen , then , according to @xcite and @xcite , @xmath28\ , , \ ] ] where @xmath25 is estimated with respect to a typical galactic value , @xmath29 . thus , @xmath30 here , we define @xmath27 as the column density over the strmgren radius . that is , @xmath31 , where @xmath32 denotes the volume number density of hydrogen . therefore , equation ( [ eq19 ] ) is reduced to @xmath33 where @xmath12 is eliminated by equation ( [ eq1 ] ) , and we suppose that @xmath34 ( fully ionized in the ionized region ) and that the temperature of h ii regions is @xmath35 k. since we can not observe the real number of lyman continuum photons , @xmath9 , we rewrite equation ( [ eq18 ] ) in terms of the apparent one , @xmath36 ; @xmath37 in addition , @xcite derived the following relation between @xmath20 and @xmath22 : @xmath38 where @xmath39 is the radius normalized by @xmath12 . integrating equation ( [ eq22 ] ) , we obtain @xmath40 where @xmath41 is defined as @xmath42 . it is worthwhile to note that @xmath41 is identified with the optical depth of dust over the actual ionized radius , @xmath18 . according to equation ( [ eq3 ] ) , @xmath43 . thus , we finally obtain @xmath44 this is the same as equation ( 8) in @xcite . now , once @xmath32 ( @xmath45 ) , @xmath8 , and @xmath25 are determined , we derive @xmath46 via equation ( [ eq20 ] ) , and then , we calculate @xmath1 by using equation ( [ eq31 ] ) . in figure 1 , we show the @xmath1 vs. @xmath41 relation obtained from equation ( [ eq31 ] ) . which quantity determines @xmath1 the most effectively ? according to figure 1 , decrement of @xmath1 is due to increment of the dust optical depth . thus , to find the most effective parameter for @xmath1 , let us examine the parameter dependence of the optical depth . we find in equation ( 7 ) that the optical depth increases according to the dust - to - gas ratio , lyman continuum flux , and density of hii regions . then , when any of those three parametes increase , @xmath1 decreases . by the way , among the power indices in equation ( 7 ) , that of the dust - to - gas ratio is the largest . then , we find that @xmath1 depends most largely on the dust - to - gas ratio . one might think that the effect of helium on the ionization structure of h ii regions is important ( e.g. , @xcite ) . in h ii regions , there are extremely energetic photons whose wavelengths are shorter than 504 . such photons can ionize neutral helium . hence , the number of lyman continuum photons absorbed by hydrogen seems to decrease because of the absorption by helium . however , fortunately , almost all ( 96% according to @xcite ) photons produced by helium recombination can ionize neutral hydrogen . therefore , we can assume safely that the number of lyman continuum photons does not significantly decrease owing to the presence of helium . moreover , the fraction of the photons with shorter wavelengths than 504 is small ( 0.14 in our calculation adopting the stellar spectra assumed to be the plank function and salpeter s imf ) . thus , the effect of helium on the photon count is small in estimating the sfr . another effect of helium appears on the electron number density . for example , in the derivation of equation ( [ eq18 ] ) , we must replace @xmath47 with @xmath48 . since helium abundance is typically 0.1 by number , @xmath49 . hence , we consider this effect to be less significant than that caused by some other uncertainties , e.g. , the change of the imf . therefore , we neglect these effects of helium in this paper . let us estimate @xmath1 for the galactic h ii regions in order to estimate the effect of lce by dust on determining the sfr . to do this , as formulated in the previous subsection , we need the data set of the number of lyman continuum photons , the electron ( or hydrogen ) number density , and the dust - to - gas ratio . the amount of dust is considered to be related to the metallicity . in this section , we determine the dust - to - gas mass ratio , @xmath50 , from the observational metallicity ( abundance of oxygen ) for each h ii region , using a global relation between @xmath25 and ( o / h ) proposed and modeled by hirashita ( 1999a , b ) . and @xmath51 here , @xmath52 is the dust mass fraction in the material injected from stars , and @xmath53 and @xmath54 are defined as gas consumption timescale ( gas mass divided by star formation rate ) normalized by dust growth timescale and by dust destruction timescale , respectively ( see @xcite for details ) . ] we display in figure 2 the global relation between @xmath25 and @xmath55 . this is the same as figure 4 of @xcite . here , we adopt @xmath56 as the solar abundance ( @xcite ) . we should note here that this @xmath25(o / h ) relation is the average relation for the interstellar medium ( ism ) in nearby spirals and dwarfs . that is , we assume that @xmath25(o / h ) relation of h ii regions is not different significantly from the averaged @xmath25(o / h ) of ism . this assumption may not be very good , but the dust properties in h ii regions are still uncertain . hence , as a first step , we apply the @xmath25(o / h ) relation of the ism to each h ii region . we make two sets of the suitable samples to estimate @xmath1 of each individual h ii region . one is the sample of seven representative giant h ii regions in the galaxy . their properties are summarized in table 1 . these h ii regions are famous and have been studied well . thus , the uncertainties of their data , especially their distance ( i.e. , their luminosity ) , can be regarded as small . their apparent number flux of lyman continuum photons ( @xmath57 ) are estimated from radio observations and more than about @xmath58 s@xmath59 . hereafter , we call this sample _ luminous _ sample . the determined @xmath1 values are also shown in column ( 7 ) of table 1 . we find a good agreement between petrosian et al.s @xmath1 and ours for the orion nebula ( @xmath60 ) . the mean @xmath1 of this sample is 0.35 . the other sample is a part of h ii regions observed and investigated by @xcite and @xcite . we select the h ii regions whose kinematic distance and photometric distance are consistent within a factor of 1.5 from their sample . this is because the uncertainty in distance affects the estimation of the number of lyman continuum photons and then prevents us from determining @xmath1 reasonably . these h ii regions tend to be less luminous than the sample of table 1 and their apparent number flux of lyman continuum photons ( @xmath57 ) are estimated from their h@xmath7 luminosities . hereafter , the sample is called to _ less luminous _ sample . their properties are tabulated in table 2 . @xcite also observed the orion nebula and m16 , both of which are also included in the sample in table 1 , but their h@xmath7 luminosities are smaller than that expected from the radio observations adopted in table 1 . this is because @xcite did not observe whole parts of these nebulae . thus , we exclude the orion nebula and m16 from the sample of table 2 . the finaly determined @xmath1 of the sample distributes over 0.51 , and the mean is 0.8 . in figures 3 ( a)(f ) , we present several relations among various parameters of our sample h ii regions in tables 1 and 2 . open symbols are h ii regions in table 1 , and filled symbols are those in table 2 . in the panels of ( a)(c ) , we show some relations among the basic parameters to determine @xmath1 . from these panels , we can check our standpoint for the subsequent considerations . firstly , we may find , in panel ( a ) , the sample regions with higher metallicity have larger observed lyman continuum flux , @xmath8 . indeed , the sample correlation coefficient is 0.41 , so that the correlation exists with a confidence level of 95% for the sample . however , we find no correlation for the sample of table 2 alone . we also can not find the correlation for the extragalactic h ii regions . thus , we consider that the correlation in panel ( a ) may be caused by our sample selection . hence , we assume it conservatively in this paper that @xmath8 dose not depend on the metallicity . in panel ( b ) , the correlation between the metallicity and the electron number density , @xmath14 , is not found . thus , we dose not consider that @xmath14 depends on the metallicity . in panel ( c ) , interestingly , we can find a correlation between @xmath14 and @xmath8 for only samples in table 2 . next , in the panels of ( d)(f ) , we present the dependece of @xmath1 on the parameters of h ii regions . the estimated @xmath1 is regarded as a function of the metallicity in panel ( d ) , i.e. , as the metallicity in an h ii region increases , @xmath1 becomes smaller . this is because , in our formulation , the dust - to - gas ratio ( i.e. , metallicity ) has a positive dependence on the optical depth of dust as shown in equation ( [ eq20 ] ) , and then , a larger dust content leads to a larger optical depth of dust and smaller @xmath1 . although the correlation in ( d ) is not the result directly derived via the analysis of the observational quantities , we confirm a trend that @xmath1 depends on the metalicity ( see also hirashita et al . 2001 ) . in panel ( e ) , we also find @xmath1 is related to @xmath8 which has a positive dependence on the dust optical depth ( eq . [ [ eq20 ] ] ) . this issue will be discussed again in section 3 . we can not find a significant correlation in panel ( f ) , probably because @xmath14 has a weak dependence in the dust optical depth ( index 1/3 in eq . [ [ eq20 ] ] ) and its dynamic range is small relative to @xmath8 which has the same dependence in equation ( [ eq20 ] ) . now , we concentrate on the dependence of the dust - to - gas ratio ( i.e. , metallicity ) on @xmath1 , because the variation of dust - to - gas ratio has the largest influence on determining the dust optical depth as seen in equation ( [ eq20 ] ) . that is , we assume that the optical depth of dust is a function of the dust - to - gas ratio only . if we consider @xmath8 and @xmath14 dose not depend on @xmath61 ( or metallicity ) , we can re - express equation ( [ eq20 ] ) as @xmath62 where @xmath41 is the dust optical depth over the actual ionized radius ( see section 2.1 ) , and @xmath63 is a factor containing the dependence of @xmath8 and @xmath64 . we find that the best fit value of @xmath63 for the adopted galactic h ii regions is 1.8 , although the dispersion is large . in figure 4 , we display the estimated @xmath1 of the individual h ii region in our sample as a function of its dust - to - gas mass ratio , @xmath25 . the open and filled symbols represent the sample of tables 1 and 2 , respectively . this figure is basically identical to figure 3 ( d ) . the solid line is the best fit line ( @xmath65 ) . we also show the lines of @xmath66 , and 3.0 for comparisons . here , let us discuss the increment factor for the sfr , @xmath11 , which is introduced at the beginning of section [ sec : f_value ] . it is useful for us and readers to check quantitatively how @xmath1 is affected by @xmath25 . as a representative case , we consider the model of @xmath67 , which corresponds with @xmath68 and @xmath69 ( nearly equal to the lyman continuum number flux and electron number density of the orion nebula ) . numerically , from the set of equations ( 7 ) and ( 10 ) , we find @xmath1 to be about 0.2 for @xmath70 ( the typical value of the ism in the galaxy ; @xcite ) . if @xmath25 is 1/3 , 1/10 , or 1/100 of the galactic value , @xmath1 becomes 0.6 , 0.9 , or nearly unity , respectively . thus , it is found that the increment factor for the sfr , @xmath11 , increases from almost unity to about 5 as @xmath25 increases from @xmath71 ( 1/10 of a typical galactic value ) to @xmath29 . clearly , the correction factor for the sfr is very sensitive to the dust - to - gas ratio ( or metallicity ) , especially around @xmath72 . hence , we should determine @xmath25 precisely to find an actual @xmath1 and sfr . in any case , we must take account of the effect of lce by dust at least for the objects whose metallicity is as high as that of the orion nebula . @xcite have also determined @xmath1 for a numerous radio sample of giant h ii regions in the galaxy . although our equation ( [ eq20 ] ) is equivalent to their equation ( a.2 ) , there is a difference in the method of determining the dust - to - gas ratio of h ii regions . indeed , they used an empirical relation of the absorption cross section as a function of the galactocentric radius ( @xcite ) , whereas we determine @xmath25 for each h ii region individually as described above . the determined mean @xmath1 for the giant h ii regions by @xcite is 0.56 . since @xmath8 of their sample is larger than @xmath73 , their sample h ii regions correspond to our sample in table 1 , which has mean @xmath74 . the estimated @xmath1 values by us may be systematically small . this is due to the differences of some adopted parameters in calculation , for example , @xmath75 . also , @xcite took account of the effect of helium which we neglect as mentioned in the previous subsection . the parameter of @xmath1 in smith et al . is thus determined as the sum of the fractions of lyman continuum photons absorbed by both hydrogen and helium . this increment by helium in @xmath1 is about several % ( up to 10% ) in their framework . if the increment is removed , smith s mean @xmath1 will approach our value . throughout this section , we have found that the determination of @xmath25 is very important to estimate the `` real '' sfr ( or @xmath1 ) from an observational count of lyman continuum photons . to obtain @xmath25 accurately , it is a recommendable way that we use the luminosity of ir thermal emission of dust in h ii regions . since dust grains absorb the radiative energy from the young massive stars and then re - emit the energy in the ir range , the amount of dust is determined from the ir luminosity . in the next section , we discuss lce by dust from observation of ir luminosity in the previous section , we estimate the amount of dust ( the dust - to - gas ratio ) in h ii regions from their metallicity . since the amount of dust is also reflected in the ir emission , @xmath1 is also determined from ir luminosity independently . in this section , then , we determine @xmath1 by using the ir luminosity of h ii regions in order to check consistency between the results of the previous and the current section . based on the theory of the ir emission from dust in h ii regions by @xcite , we formulate the relation between the observed ir luminosity , @xmath76 , and the observed number of lyman continuum photons , @xmath8 ( inoue , hirashita , & kamaya 2000 , 2001 ) . the derived formula is @xmath77 where @xmath78 is the average efficiency of the dust absorption for uv optical photons from ob stars . here , we have also adopted the salpeter s imf ( the stellar mass range is 0.1100 @xmath79 ) and proper stellar properties . moreover , the intrinsic number of lyman continuum photons , @xmath9 , is @xmath80 per unit solar luminosity of the bolometric luminosity of ob stars , in our calculation . as defined in the previous section , @xmath81 . instead of determining the sed of the ir radiation , we consider the total ir luminosity of dust , @xmath76 , in the range of 8 1000 @xmath82 m which covers almost the whole wavelength range of dust emission . first , let us consider the correlation between the luminosities of ir and h@xmath7 for h ii regions . we make a cross - reference between the sample of @xcite and the _ iras _ point source catalog ( _ iras _ psc ; @xcite ) . we select the sample h ii regions by the following criteria ; ( a ) the separation between the positions of the h@xmath7 observations and sources in _ iras _ psc is within @xmath83 . ( b ) @xmath84 . ( c ) @xmath85 . ( d ) @xmath86 jy . here , @xmath87 , @xmath88 , @xmath89 , and @xmath90 are the _ iras _ fluxes at 12 , 25 , 60 , and 100 @xmath82 m , respectively . the criterion ( a ) is based on the typical radius of optical h ii regions ( @xmath91 110 pc ) and _ iras _ positional error ( @xmath92 , @xcite ) . the criteria ( b ) ( d ) are proposed by @xcite . the confidence level of the set of criteria for true association between _ iras _ sources and h ii regions is more than 80% ( @xcite ) . as a result , we obtain 18 h ii regions ( 10 samples are also in table 2 ) and their properties are found in table 3 . when we suppose that the dust emission can be fitted by the modified black - body radiation of 30 k ( spectral index is 1 ) , _ iras _ 40 120 @xmath82 m luminosity , @xmath93 , is @xmath94 . this correction factor is consistent with that of @xcite . on the other hand , @xmath8 is estimated in terms of the h@xmath7 luminosity corrected for interstellar extinction : @xmath95 , where we assume the case b and the electron temperature of @xmath35 k ( @xcite ) . thus , equation ( [ eq40 ] ) is reduced to @xmath96 where @xmath97 and @xmath98 are fluxes of _ iras _ and h@xmath7 , respectively . this is a theoretical relation between @xmath97 and @xmath99 . since a flux ratio does not depend on the distance to the objects , equation ( [ eq41 ] ) is free from the uncertainty in the distance . the average efficiency of dust absorption for uv optical photons from ob stars , @xmath78 , is defined as @xmath100 , where @xmath101 is an average dust absorption over the wavelength range of @xmath102 4000 in units of magnitude . this range is suitable for our purpose because the number of photons of @xmath103 4000 emitted by ob stars are negligible . if we adopt the galactic extinction curve ( @xcite ) , then we have @xmath104 . thus , we obtain the following equation ; @xmath105 using this equation , we find @xmath106 for the h ii regions in table 3 . once we obtain the value of @xmath78 for each h ii region from equation ( [ eq42 ] ) , we can determine each @xmath1 from ir and h@xmath7 fluxes by using equation ( [ eq41 ] ) . we show the estimated @xmath1 in column ( 7 ) of table 3 . by definition , however , for some sample h ii regions , @xmath1 appears to be greater than unity . the reason for the discrepancy will be discussed in the last paragraph of this section . by the way , we compare the theoretical @xmath97@xmath98 relation with observational data on the @xmath97 vs.@xmath98 plane in figure 5 . adopting @xmath108 , we make three lines for three values of @xmath1 ; the solid , dotted , and dashed lines for @xmath109 , 0.5 , and 0.1 , respectively . moreover , we make the dash - dotted line for the set of @xmath109 , @xmath110 . this corresponds to the case that only the lyman @xmath7 photons are absorbed by dust grains . as long as all lyman @xmath7 photons are absorbed by dust within h ii regions and re - emitted in the ir range ( @xcite ) , the area below the dash - dotted line on the @xmath111 vs. @xmath98 plane is the forbbiden area within the framework of our theory . that is , the dash - dotted line stands for the lower boundary of _ iras _ flux corresponding to h@xmath7 ( lyman continuum photon ) flux in our model . while the correlation between @xmath97 and @xmath99 is expected theoretically from equation ( [ eq41 ] ) , we do not find a significant correlation in observational data points in figure 5 ( the sample correlation coefficient is 0.35 ) . this point is also examined in the final part of this section . on the other hand , it is well known that there is a tight observational correlation between the ir luminosity and the number of lyman continuum photons estimated from the radio luminosity ( @xcite ) . using equation ( [ eq40 ] ) , we compare this observational correlation with the theoretical one in figure 6 . here , we read the data directly from the figure in @xcite and re - plotted them . the sample correlation coefficient of their data is 0.96 , showing a stong correlation . the parameters of lines are the same as those in figure 5 . comparing the data and the model lines , we consider the ir luminosity of @xcite ( 40 350 @xmath82 m ) is nearly equal to the whole dust - ir luminosity in equation ( [ eq40 ] ) because the dust luminosity of the wavelength range of @xmath112 or @xmath113 is negligible ( @xcite ) . from figure 6 , we find that the theoretical lines agree with the data of @xcite excellently . this shows that our model is supported by the observation . if we assume @xmath108 for the sample of @xcite , we can determine their @xmath1 from equation ( [ eq40 ] ) . we find @xmath114 as an average in the current sample . now , we compare it with another mean @xmath1 value estimated by the method of the previous section . since h ii regions in figure 6 have @xmath115 s@xmath59 , we shall choose 18 regions whose @xmath8 is greater than @xmath116 s@xmath59 from tables 1 and 2 . thus , we obtain a mean @xmath1 as being 0.56 by the method of section 2 . fortunately , we find a good agreement between the two @xmath1 s . by the way , @xmath1 values estimated by the both methods are not fully consistent with each other . for example , the sample of @xcite includes the orion nebula , m8 , and m17 as the sample of table 1 does . their @xmath1 values via equation ( [ eq40 ] ) are 0.44 , 0.59 , and 0.59 , respectively . obviously , these values are larger than those of the orion nebula , m8 , and m17 in table 1 . this may be caused by the aparture difference between two photometories , ir and radio ( see also the last paragraph in this section ) . this discrepancy may also result from our assumption in equation ( [ eq40 ] ) that all of the energy of photons not used by hydrogen ionization is converted into ir radiation . generally , the efficiency of conversion to ir radiation is never unity because of the escape of lyman continuum photons from h ii regions or the scattering of these photons by dust . as a result , we expect rather large ir luminosities for the h ii regions by using equation ( [ eq40 ] ) . moreover , we can not find in figure 6 the trend that the more luminous h ii regions have the smaller @xmath1 . this trend was seen in section 2.1 , where we determined @xmath1 from the metallicity . this may be the effect of ir cirrus . the observed ir flux originates from dust associated with the star forming regions and from diffuse dust heated by the interstellar radiation field . if we consider that the ir surface brightness of the cirrus component is almost constant in a galaxy , the effect of cirrus contamination in the observed ir flux of an h ii region is larger as its ir flux is lower . in the case , we may overestimate the ir luminosity for low luminosity regions , so that we may underestimate @xmath1 of such regions . figure 6 has other interesting implications : first , the dash - dotted line ( @xmath109 and @xmath117 ) means that only lyman @xmath7 photons heat dust grains and only this energy is re - emitted in ir . obviously the predicted ir luminosity in this case falls short of the observed one . next , the solid line ( @xmath109 and @xmath118 ) represents the case that only the energy of photons whose wavelength is longer than the lyman limit is absorbed by dust and re - emitted in ir . even in this case , the predicted ir luminosity is not enough to explain the observed one . hence , we conclude that a part of lyman continuum photons ought to be aborbed and reprocessed by dust . the fraction of lyman continuum photons absorbed by dust is about 0.5 or more . in terms of the fraction of the photons used by hydrogen ionization , @xmath2 in many h ii regions of our galaxy this is one of the main conclusions in the current paper . to close this section , we must discuss a few remaining questions in the previous paragraphs : why is @xmath1 determined by our theory greater than unity for some h ii regions in table 3 ? why can not we find the observational correlation between @xmath97 and @xmath98 in figure 5 ? these may be because we use the _ iras _ psc . since the galactic h ii regions lie close to us and extend over large areas , they can never be regarded as point - like sources . hence , the ir flux for h ii regions in _ iras _ psc may be underestimated since the aperture effect of h ii regions is not neglected . therefore , _ iras _ psc may not be suitable when we measure the ir flux of galactic h ii regions ( see also @xcite ) . conversely , the correlation between the ir luminosity and the number of lyman continuum photons inferred from radio observations is tight . this observational correlation is explained very well by our theory . this is because the ir luminosities of sample in @xcite are observed in detail individually . seeing the observational tight correlation in figure 6 , the absence of the correlation in figure 5 is caused by flux underestimation of _ iras _ psc due to the finite aperture effect . thus , if we use the _ image _ data of the h ii regions observed by _ iras _ or recent / future facilities such as _ iso _ , sofia , _ sirtf _ , or astro - f ( see @xcite for a summary of the facilities ) , the observational correlation between @xmath119 and @xmath99 will appear . in this section , we estimate @xmath1 averaged on galactic scale for the nearby spiral galaxies . then we discuss the effect of lce by dust on determining their galaxy - wide sfr . in the privious sections , we have discussed lce by dust in h ii regions . as shown in section 2 , the parameter @xmath1 strongly depends on the dust - to - gas mass ratio . for the galactic h ii regions , we show that @xmath120 is appropriate in equation ( [ eq30 ] ) . that is , the orion nebula is likely to be a typical galactic h ii region in gas density and lyman continuum flux . here , we consider the orion nebula is also a typical h ii region in the other galaxies . in short , we assume @xmath67 in equation ( [ eq30 ] ) to be universal for the galaxies . then , we estimate a typical @xmath1 for the nearby spiral galaxies from their dust - to - gas ratio . by the way , we can determine @xmath1 of the galaxies from equation ( [ eq41 ] ) as perfomed for the galactic h ii regions in section 3 . this alternative way is discussed in the last part of this section . in order to determine their dust - to - gas ratio , we estimate the amount of dust mass in the nearby spiral galaxies from their ir emission re - processed by dust . as considered in the final paragraph of section 3 , we do not use _ iras _ psc since we want to estimate the entire ir luminosity of the sample galaxies safely . we adopt only the sample galaxies whose entire images by _ iras _ are available . we note another issue here . if we convert the ir emission calculated only from _ iras _ data to the dust mass , the total amount of dust will be underestimated . this is because there is large amount of cold dust ( 1020 k ) in the spiral galaxies , even in those with starburst regions ( @xcite and references therein ) . such cold dust can not be traced by _ iras _ bands . on the other hand , _ iso _ has better detectability for the cold dust than _ iras _ because _ iso _ covers longer wavelengths than _ iras_. thus , we expect that _ iso _ data improve the estimation of dust mass . @xcite present the imaging properties of seven nearby spiral galaxies obtained by both _ iras _ and _ iso_. their sample galaxies are ngc 134 , ngc 628 ( m74 ) , ngc 660 , ngc 5194 ( m51 ) , ngc 5236 ( m83 ) , ngc 6946 , and ngc 7331 . some basic properties of these galaxies are tabulated in tables 1 , 2 and 3 of @xcite . we estimate the dust - to - gas mass ratio for alton s sample galaxies , and then , determine their @xmath1 averaged over galactic scale . to avoid the confusion from @xmath1 of the h ii regions , in this section , we express the galactic @xmath1 as being @xmath3 . of course , each sample galaxy has unique @xmath3 . we , first , calculate the dust temperature of the sample galaxies from the ratio of flux densities at 100 and 200 @xmath82 m . in this calculation , we assume the spectrum of dust emission to be the modified black - body radiation with the spectral index of 1 . here , we note that the determination of the dust temperature strongly affects the successive estimation of the dust mass . the obtained dust temperatures are shown in column ( 5 ) of table 4 . next , we estimate the dust mass , @xmath121 , in the sample galaxies from the observed flux density at 100 @xmath82 m via the following equation : @xmath122 where @xmath123 , @xmath124 , and @xmath125 are the grain radius , grain density , and galaxy distance , respectively . @xmath126 , @xmath127 , and @xmath128 are the observed flux density , grain emissivity , and the plank function of the temperature @xmath129 at the frequency @xmath130 , respectively . according to @xcite , we set @xmath131 g @xmath132 at the frequency of 100 @xmath82 m ( @xmath133 @xmath82 m and @xmath134 g @xmath135 ) . the estimated dust masses are tabulated in column ( 6 ) of table 4 . then , we determine the dust - to - gas mass ratio , @xmath25 , of the sample galaxies by using their total ( h i + h@xmath136 ) gas masses reported in literatures ( column 7 ) . the values of @xmath25 are in column ( 8) , and the average is about @xmath137 . this is quite consistent with @xmath25 estimated by @xcite . @xcite also claim that the median value of the dust - to - gas mass ratios of sample galaxies of isophot 170 @xmath82 m serendipity survey ( @xcite ) is about @xmath138 . thus , our estimation of @xmath25 is reasonable . it still remains open whether @xmath25 in h ii regions can be approximated to an averaged @xmath25 of the ism in the galaxies . as seen in subsection 2.2 , there is a large variation of @xmath25 ( i.e. observed metallicity ) of the galactic h ii regions . moreover , most of the sample h ii regions in tables 1 and 2 have smaller @xmath25 than the typical value for the galactic ism ( @xmath29 ; spitzer 1978 ) . it may suggest that @xmath25 in h ii regions is systematically smaller than that of the ism . however , it is reported that the mean @xmath25 in the ism is not so different from that in h ii gas . @xcite have determined the dust - to - gas mass ratio of the h i , h@xmath136 , and low - density ( @xmath139 ) h ii gases in the galaxy from the data of @xmath140 . according to them , the mean value of @xmath25 in the galactic ism ( h i + h@xmath136 + h ii ) is 4.66.6@xmath141 , and that in h ii gas is 5.110@xmath141 . with the uncertainty of a factor of 2 , therefore , we assume that @xmath25 in h ii regions is approximated to the mean @xmath25 of the ism in the galaxies . here , let us calculate a characteristic value of the @xmath3-parameters for the sample seven spiral galaxies , adopting their dust - to - gas mass ratio ( column 8 in table 4 ) , equations ( [ eq31 ] ) and ( [ eq30 ] ) with @xmath67 . the derived values of @xmath3 are tabulated in column ( 9 ) of table 4 . the averaged @xmath3 is 0.3 . this is nearly equal to the mean @xmath1 of the sample h ii regions in table 1 which are more luminous than the orion nebula . this small @xmath3 means that the correction factor of the sfr ( @xmath4 ; see also section [ sec : f_value ] ) for these spiral galaxies is large ( typically @xmath5 ) . although @xmath1 values of h ii regions distribute over a wide range ( 0.021 ) as shown in tables 1 and 2 , @xmath3 is approximately equal to the averaged value of the luminous h ii regions ( i.e. , the sample in table 1 ) . this may be caused by the effect of luminosity function of h ii regions . indeed , it is implied that the total luminosity of h ii regions is dominated by the regions more luminous than the orion nebula ( @xcite ) . thus , @xmath3 may be shifted toward smaller values because we expect smaller @xmath1 in more luminous h ii regions ( see section 2.1 ) . therefore , we think it reasonable that a characteristic @xmath3 for nearby spiral galaxies is about 0.3 . now we discuss the importance of the increment of a factor of @xmath142 for the estimated sfr . there are two other major uncertainties for the estimation of the sfr . one is related to the initial mass function ( imf ) . the choices of a specific imf and its upper / lower limit masses affect the coefficient of the conversion formula of the sfr . the uncertainties are within a factor of about two ( e.g. , @xcite ) . the other uncertainty is related to stellar properties . the choices of the stellar properties adopted affect the conversion formula of the sfr . @xcite derived the formula , @xmath143 , adopting their stellar population synthesis model and salpeter imf ( 0.1100@xmath79 ) . we also derive the same type formula by using the same imf but by using very simplified stellar properties ( @xcite ) . in comparison with them , we find that the uncertainty due to the adopted stellar properties is at most a factor of two . therefore , we conclude that the effect of lce by dust may be more significant than the uncertainties due to the specific imf or the stellar properties for nearby spiral galaxies . finally , we discuss the alternative way for determining galaxy - wide @xmath3 . we can use equation ( [ eq41 ] ) to do it . however , there are three difficulties to determine @xmath3 reasonably . ( i ) one is the uncertainty of the cirrus fraction of the ir luminosity of the galaxies . it is widely believed that the ir luminosity of the galaxies consists of two components : the warm component originating from the dust nearby the star forming regions and the cirrus component originating from the diffuse cold dust heated by the interstellar radiation field from the older stars ( e.g. , @xcite ) . it is a very difficult task to subtract the cirrus component from the observed ir luminosity of the galaxies in order to obtain the `` pure '' ir luminosity originating from the star forming regions . since the cirrus fraction differs from galaxy to galaxy ( e.g. , 0.30.8 ; @xcite ) , the subtraction of cirrus causes a significant uncertainty in the estimation of @xmath3 . the other difficulties are the correction of the h@xmath7 luminosity for ( ii ) the interstellar extinction and ( iii ) the [ n ii ] contamination . indeed , @xcite noted that the interstellar extinction at the wavelength of h@xmath7 is 1.1 mag on average for his sample galaxies . that is , the increment correction factor for h@xmath7 luminosity is 2.8 . for that case , we do not neglect such large extinction in order to determine @xmath3 reasonably . at that time , we may also have to correct the inclination of the galaxies along the line of sight to find the correct @xmath3 . the [ n ii ] contamination is more complicated if we do not have sufficient resolution of the spectrograph . fortunately , the h@xmath7 flux of @xcite is not contaminated by [ n ii ] because of their high dispersion of spectrograph . then , our main conclusions on @xmath1 of the galactic h ii regions are not altered . however , we are sure of that the [ n ii ] line contaminates the galactic h@xmath7 emission ( e.g. @xcite ) . when we use equation ( [ eq41 ] ) to find @xmath3 , therefore , we always need the h@xmath7 luminosity corrected for the interstellar reddening and [ n ii ] contamination to find the true @xmath3 . unfortunately , we face the deficiency of the data to correct the h@xmath7 luminosity of each individual galaxy . because of several large uncertainties , thus , we can not safely determine @xmath3 of the galaxies via equation ( [ eq41 ] ) in this paper . this problem is examined by us in the near future . for each individual h ii region in the galaxy , we examine lce by dust within such regions . then , we discuss the lce effect on determining the sfr of galaxies . we reach the following conclusions . [ 1 ] we estimate the fraction of lyman continuum photons contributing to hydrogen ionization within the galactic h ii regions , @xmath1 . the estimated @xmath1 values distribute over a wide range ( 0.021 ) . in our framework , an h ii region with higher metallicity and larger lyman photon flux tends to have a smaller @xmath1 . [ 2 ] from the correlation tests in figure 3 , we find that @xmath1 is approximately treated as a function of only the dust - to - gas mass ratio , @xmath25 . for our sample of the galactic h ii regions , the orion nebula is regarded as a typical region in gas density and lyman continuum flux . [ 3 ] we explain theoretically the observed correlation between ir luminosity and radio luminosity of the galactic h ii regions . then , we manifest that a part of lyman continuum photons ( about a half or more ) ought to be absorbed by dust directly in many regions . [ 4 ] we do not find the observational correlation between fluxes of _ iras _ psc and h@xmath7 for our sample h ii regions , which is expected by our model theoretically . this may indicate that _ iras _ psc is not suitable to determine the ir flux of nearby extended sources like galactic h ii regions as pointed out in literatures . [ 5 ] if we assume that @xmath25 in h ii regions is not significantly different from a mean @xmath25 of the ism , we estimate @xmath1 averaged over galactic scale to be about 0.3 for the nearby spiral galaxies ( i.e. @xmath144 ) . consequently , the increment factor for the sfr due to the lce by dust ( @xmath4 ) is about 3 . the correction factor is larger than the uncertainties of determining the sfr due to the choice of a specific imf or stellar properties ( within a factor of 2 ) . therefore , we can not neglect the effect of lce by dust when we estimate the sfr in galaxies . we thank the anonymous referee for very helpful comments , which improved quality of the paper significantly . we are also grateful to tsutomu t. takeuchi for careful reading of this paper and giving us very useful comments . one of us ( h.h . ) acknowledges the research fellowship of the japan society for the promotion of science for young scientists . we have made extensive use of nasa s astrophysics data system abstract service ( ads ) . orion & 0.5 & 7.59 & 150 & 8.76 & 5.0 & 0.26 & 1 + m8 & 1.1 & 10.6 & 60 & 8.74 & 4.5 & 0.37 & 1 + m16 & 2.5 & 29.1 & ( 100 ) & 8.76 & 5.0 & 0.16 & 1 + rosette & 1.4 & 43.2 & 13 & 8.20 & 0.79 & 0.85 & 2 + m17 & 1.8 & 143 & ( 100 ) & 8.81 & 6.3 & 0.016 & 1 + carina & 1.4 & 167 & 13 & 8.49 & 1.8 & 0.56 & 2 + ngc3603 & 8.5 & 897 & 25 & 8.51 & 2.0 & 0.23 & 2 + s54 & 42.0 & 3.08 & 245 & 8.60 & 2.7 & 0.54 + s88 & 2.17 & 0.16 & 28 & 8.43 & 1.5 & 0.94 + s93 & 8.65 & 0.63 & 269 & 8.57 & 2.5 & 0.71 + s101 & 3.61 & 0.26 & 7 & 8.34 & 1.1 & 0.97 + s104 & 76.7 & 5.62 & 56 & 8.51 & 2.0 & 0.71 + s127 & 38.4 & 2.81 & 545 & 8.23 & 0.79 & 0.80 + s131 & 0.20 & 0.01 & 18 & 8.39 & 1.3 & 0.98 + s142 & 16.6 & 1.22 & 21 & 8.27 & 0.89 & 0.94 + s148 & 58.2 & 4.27 & 235 & 8.30 & 1.0 & 0.78 + s153 & 8.90 & 0.65 & 106 & 8.12 & 0.59 & 0.94 + s156 & 59.4 & 4.35 & 907 & 8.42 & 1.5 & 0.55 + s168 & 18.4 & 1.35 & 137 & 8.30 & 1.0 & 0.87 + s184 & 4.56 & 0.33 & 90 & 8.39 & 1.4 & 0.90 + s206 & 31.8 & 2.33 & 313 & 8.37 & 1.3 & 0.75 + s209 & 201 & 14.7 & 645 & 8.18 & 0.71 & 0.69 + s211 & 26.0 & 1.91 & 135 & 8.03 & 0.48 & 0.93 + s219 & 4.26 & 0.31 & 166 & 8.27 & 0.89 & 0.92 + s252 & 5.22 & 0.38 & 144 & 8.35 & 1.2 & 0.89 + s257 & 2.55 & 0.19 & 160 & 8.17 & 0.69 & 0.95 + s83 & 1.75 & 6.88 & 19223 + 2041 & 0.98 & 1 & 1.35 + s93 & 1.76 & 12.6 & 19529 + 2704 & 6.91 & 1 & 0.49 + s100 & 1.31 & 19.4 & 19598 + 3324 & 50.8 & 1 & 0.11 + s101 & 0.56 & 6.07 & 19579 + 3509 & 0.22 & 0.98 & 2.49 + s127 & 1.65 & 2.77 & 21270 + 5423 & 1.29 & 1 & 0.56 + s128 & 1.83 & 6.31 & 21306 + 5540 & 3.53 & 1 & 0.48 + s138 & 1.38 & 2.80 & 22308 + 5812 & 4.15 & 1 & 0.20 + s146 & 1.57 & 6.16 & 22475 + 5939 & 4.05 & 1 & 0.42 + s148 & 0.99 & 4.09 & 22542 + 5815 & 2.73 & 1 & 0.41 + s152 & 1.10 & 9.28 & 22566 + 5830 & 6.68 & 1 & 0.38 + s153 & 0.68 & 2.45 & 22571 + 5828 & 2.84 & 0.99 & 0.25 + s156 & 1.12 & 16.5 & 23030 + 5958 & 7.34 & 1 & 0.58 + s168 & 0.99 & 8.95 & 23504 + 6012 & 1.01 & 1 & 1.55 + s206 & 1.36 & 52.1 & 03595 + 5110 & 2.44 & 1 & 2.32 + s211 & 1.61 & 5.08 & 04329 + 5045 & 1.29 & 1 & 0.91 + s212 & 0.92 & 13.8 & 04368 + 5021 & 0.73 & 1 & 2.22 + s217 & 0.73 & 2.72 & 04551 + 4755 & 0.50 & 1 & 1.14 + s252 & 0.33 & 7.90 & 06066 + 2029 & 3.82 & 0.90 & 0.51 + ngc 134 & 21.1 & 67.0 & 127 & 21 & 8.18 & 10.40 & 5.9 & 0.21 + ngc 628 & 8.8 & 67.5 & 209 & 19 & 7.72 & 10.04 & 4.8 & 0.29 + ngc 660 & 11.3 & 120 & 158 & 24 & 7.51 & 10.04 & 2.9 & 0.48 + ngc 5194 & 6.2 & 303 & 533 & 22 & 7.62 & 9.88 & 5.5 & 0.24 + ngc 5236 & 6.9 & 624 & 889 & 23 & 7.90 & 10.54 & 2.3 & 0.56 + ngc 6946 & 10.1 & 338 & 743 & 20 & 8.38 & 10.48 & 8.0 & 0.12 + ngc 7331 & 10.9 & 120 & 243 & 21 & 7.85 & 10.00 & 7.0 & 0.16 +	we re - examine the effect of lyman continuum ( @xmath0 ) extinction ( lce ) by dust in h ii regions in detail and discuss how it affects the estimation of the global star formation rate ( sfr ) of galaxies . to clarify the first issue , we establish two independent methods for estimating a parameter of lce ( @xmath1 ) , which is defined as the fraction of lyman continuum photons contributing to hydrogen ionization in an h ii region . one of those methods determines @xmath1 from the set of lyman continuum flux , electron density and metallicity . in the framework of this method , as the metallicity and/or the lyman photon flux increase , @xmath1 is found to decrease . the other method determines @xmath1 from the ratio of infrared flux to lyman continuum flux . importantly , we show that @xmath2 via both methods in many h ii regions of the galaxy . thus , it establishes that dust in such h ii regions absorbs significant amount of lyman continuum photons directly . to examine the second issue , we approximate @xmath1 to a function of only the dust - to - gas mass ratio ( i.e. , metallicity ) , assuming a parameter fit for the galactic h ii regions . we find that a characteristic @xmath3 , which is defined as @xmath1 averaged over a galaxy - wide scale , is 0.3 for the nearby spiral galaxies . this relatively small @xmath3 indicates that a typical increment factor due to lce for estimating the global sfr ( @xmath4 ) is large ( @xmath5 ) for the nearby spiral galaxies . therefore , we conclude that the effect of lce is not negligible relative to other uncertainties of estimating the sfr of galaxies .
the authors acknowledge b. w. eichhorn for experimental assistance , and n.p.b . acknowledges support from a cnam glover fellowship . this work was supported by afosr - muri grant fa9550 - 09 - 1 - 0603 .	the effect of alkaline earth substitution on structural parameters was studied in high - quality single crystals of ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2grown by the self - flux method . the results of single - crystal and powder x - ray diffraction measurements suggest a continuous monotonic decrease of both @xmath3- and @xmath4-axis lattice parameters , the @xmath5 tetragonal ratio , and the unit cell volume with decreasing alkaline earth atomic radius as expected by vegard s law . as a result , the system experiences a continuously increasing chemical pressure effect in traversing the phase diagram from @xmath6 in ba@xmath0sr@xmath1fe@xmath2as@xmath2 to @xmath7 in sr@xmath0ca@xmath1fe@xmath2as@xmath2 . the recent discovery of high - temperature superconductivity in iron - based compounds has attracted much interest . the parent phases of these compounds generally show antiferromagnetic order that onsets between 130 k and 200 k , with superconductivity emerging when the antiferromagnetic order of the parent compounds is suppressed @xcite . this proximity of magnetic and superconducting order parameters is widely thought to be a key argument for an unconventional pairing mechanism , likely mediated by spin fluctuations @xcite similar to the cuprates @xcite . but in strong contrast to the copper oxides , superconductivity in iron arsenides can be induced without changing the carrier concentration , either by applying external pressure @xcite or by isovalent chemical substitution . the highest @xmath8 achieved so far in these materials is @xmath9 k in smo@xmath0f@xmath1feas @xcite and ( sr , ca)feasf @xcite . oxygen - free feas - based compounds with the thcr@xmath2si@xmath2-type ( 122 ) structure also exhibit superconductivity induced by chemical substitution of alkali or transition metal ions @xcite , the application of large pressures @xcite , or lattice strain @xcite , with transition temperatures as high as @xmath10 k. for the 122 phase , superconductivity has been induced by substituting fe with not only 3@xmath11-transition metals such as co and ni , but also some of the 4@xmath11- and 5@xmath11-transition metals . recently , ru , ir , and pt substitution for fe were also shown to induce superconductivity in srfe@xmath2as@xmath2 and bafe@xmath2as@xmath2 @xcite . superconductivity with @xmath8@xmath12 31 k has also been shown to occur by isovalent substitution of p for as @xcite . this gives the opportunity to tune magnetic character without nominally changing charge carrier concentrations , for instance making the interpretation of transport coefficients much simpler than in the case of charge doping . in order to investigate the possibility of applying uniform chemical pressure in a continuous manner , we have synthesized the series of solid solutions ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2by substituting isovalent alkaline earth atoms , and investigated the evolution of the crystal structure by high - resolution powder and single - crystal x - ray diffraction . here we present our preliminary results that suggest the unit cell of the ba - sr - ca substitution series experiences a monotonic uniform chemical pressure as a function of alkaline earth atomic radius . single - crystal samples of ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2 were grown using the feas self - flux method @xcite . fe was first separately pre - reacted with as via solid - state reaction of fe ( 99.999% ) powder with as ( 99.99% ) powders in a quartz tube of partial atmospheric pressure of ar . the precursor materials were mixed with elemental sr ( 99.95% ) with either ba ( 99.95% ) or ca ( 99.95% ) in the ratio 4:1@xmath13 , placed in an alumina crucible and sealed in a quartz tube under partial ar pressure . the mixture was heated to 1150@xmath14c , slow - cooled to a lower temperature and then quenched to room temperature . typical dimensions of as - grown single crystal specimen are @xmath12100 @xmath15 m thickness and up to 5 mm width . chemical analysis was performed using both energy- and wavelength - dispersive x - ray spectroscopy ( eds and wds ) . both eds and wds analysis of all ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2 samples showed the proper 1:2:2 stoichiometry in all specimens reported herein , with no indication of impurity phases . figure [ fig1 ] compares the nominal alkaline earth concentration @xmath16 in sr@xmath0ca@xmath1fe@xmath2as@xmath2 crystals with that measured by wds and eds analysis , using an average value determined from 8 different spots on each specimen . as shown by the dotted line guide , the actual concentrations found by wds are equal to the nominal values of @xmath16 to within experimental error , indicating homogeneous substitution in this series of solid solutions . diffraction patterns were obtained by both powder and single - crystal x - ray diffraction and rietfeld refinement ( shelxs-97 ) to @xmath17 structure . powder x - ray diffraction was performed at 250 k using a smart apex2 diffractometer with mo - k@xmath18 radiation and a graphite monochromator . figure [ fig2 ] shows a typical x - ray diffraction pattern obtained from a single - crystal sample of sr@xmath19ca@xmath20fe@xmath2as@xmath2 . all of the main peaks can be indexed to the thcr@xmath2si@xmath2 structure , with no impurity phases detected . .[tabl1]crystallographic data for srfe@xmath2as@xmath2 and sr@xmath21ca@xmath22fe@xmath2as@xmath2determined by single - crystal x - ray diffraction at 250 k. the structure was solved and refined using the shelxs-97 software , yielding lattice constants with residual factor @xmath23= 1.36% and 1.95% for srfe@xmath2as@xmath2 and sr@xmath21ca@xmath22fe@xmath2as@xmath2 , respectively . [ cols="<,<,<",options="header " , ] table 1 shows the crystallographic parameters determined by single - crystal x - ray - diffraction at 250 k in sr@xmath19ca@xmath20fe@xmath2as@xmath2 . a bruker smart apex2 diffractometer with mo - k@xmath18 radiation , a graphite monochromator with monocarp collimator , and a ccd area detector were used for this experiment . the structure was refined with shelxl-97 software using 1033 measured reflections of which 115 were unique and 108 observed . the final residuals were @xmath24= 1.36% and 1.96% for the observed data and @xmath25= 3.31% and 4.52% for all data for srfe@xmath2as@xmath2 and sr@xmath19ca@xmath20fe@xmath2as@xmath2 respectively . sr and ca atoms were found to reside in the same site with a refined ca : sr ratio of 0.33(1):0.67(1 ) , giving the exact formula sr@xmath21ca@xmath22fe@xmath2as@xmath2 from x - ray analysis . figure [ fig3 ] shows the unit cell of sr@xmath21ca@xmath22fe@xmath2as@xmath2 determined by single crystal x - ray diffraction at 250 k. the size of the ellipsoids map the thermal agitation of the particular ion at 250 k with a 50% probability factor , which means the probability of finding the center of the atom inside the ellipsoid . figure [ fig4]presents the variation of the @xmath3- and @xmath4-axis lattice constants ( upper panel ) , the unit cell volume and the tetragonal @xmath5 ratio ( lower panel ) with ba - sr and sr - ca concentrations determined from refinements of the single crystal x - ray diffraction data for ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2 crystals taken at 250 k. within experimental accuracy , the @xmath3- and @xmath4-axis lattice constants , the @xmath5 ratio , and the unit cell volume all show a monotonic linear decrease with alkaline earth substitution in the continuous series from bafe@xmath2as@xmath2 to srfe@xmath2as@xmath2 to cafe@xmath2as@xmath2 . this fact indicates that the whole ( ba , sr , ca)fe@xmath2as@xmath2 series progression experiences a uniform chemical pressure effect due to the reduction of the cation size that follows vegard s law , as expected for the decreasing ionic radii of ba , sr and ca , respectively . the lattice parameters of ba@xmath0sr@xmath1fe@xmath2as@xmath2 obtained in our experiments are consistent with the data reported in a recent study @xcite , which found a systematic increase of @xmath26 with increasing sr content and no superconductivity . on the other hand , substitution of arsenic for the smaller phosphorus atoms , also instituting a chemical pressure effect , induces superconductivity in bafe@xmath2as@xmath27p@xmath1 @xcite . thus , a pressure - volume effect is clearly an oversimplified explanation for superconductivity in bafe@xmath2as@xmath27p@xmath1 . in the future , it will be interesting to investigate the evolution of superconductivity combining both the chemical pressure effect of alkaline earth substitution studied here and another tuning parameter that induces superconductivity in order to investigate the role of lattice density in these phenomena . in summary , we have systematically studied the crystallographic properties in the solid solution series ba@xmath0sr@xmath1fe@xmath2as@xmath2 and sr@xmath0ca@xmath1fe@xmath2as@xmath2 by growing high - quality single crystals by the self - flux method . x - ray diffraction and data refinement reveal that the entire ( ba , sr , ca)fe@xmath2as@xmath2series progresses according to vegard s law , experiencing a uniform chemical pressure effect due to substitution with reduced ionic size from ba to sr to ca , respectively .
the emission of gamma - ray bursts ( grbs ) and radio loud active galactic nuclei ( agns ) is generally believed to be produced in relativistic jets powered by their central black holes ( bhs ) . grbs are the most luminous transients in the universe . their progenitors are thought to be either core collapses of massive stars ( woosley 1993 ; paczyski 1998 ) or mergers of two compact objects ( eichler et al . such a catastrophic event likely gives birth to a new - born stellar - mass bh , which accretes mass from a torus . an ultra - relativistic jet is launched from the central engine , which dissipates its kinetic or magnetic energy at a large radius to power the gamma - rays observed . radio - loud agns are believed to be powered by a super - massive rotating bh , which launches a mildly relativistic jet ( urry & padovani 1995 ) . most confirmed extragalactic gev - tev sources are blazars , a sub - sample of radio - loud agns . they are divided into flat spectrum radio quasars ( fsrqs ) and bl lac objects ( bl lacs ) , based to whether or not strong emission line features are observed . the broad band spectral energy distributions ( seds ) of blazars usually show two bumps , which are generally explained as synchrotron radiation and inverse compton ( ic ) scattering of a same population of relativistic electrons , respectively the seed photons for the ic process can be from the synchrotron radiation itself ( i.e. , the synchrotron self - compton [ ssc ] model , maraschi et al . 1992 ; ghisellini et al . 1996 ; zhang et al . 2012b ) , or from an external radiation field ( e.g. , the external compton [ ec ] model , dermer et al . 1992 ) . narrow - line seyfert 1 galaxies ( nls1s ) were also identified as a new class of gev gamma - ray agns by the fermi / lat ( abdo et al . 2009a ) . their broad band seds are similar to that of fsrqs , which are also well explained by this synchrotron+ic leptonic jet model ( abdo et al . 2009a ) . in contrast to blazars whose radiation mechanisms ( especially for the low - frequency component ) are well understood , the radiation mechanism of grb prompt gamma - ray emission is still not identified . this prompt emission spectrum is usually well described by a smooth broken power - law function called the `` band function '' ( band et al . 1993 ) , with a spectral peak typically in the range of 100s of kev . the origin of this spectrum is under intense debate . although synchrotron emission has been proposed as the leading model ( e.g. , mszros et al . 1994 ; wang et al . 2009 ; daigne et al . 2011 ; zhang & yan 2011 ) , other mechanisms including comptonization of quasi - thermal photons from the fireball photosphere ( e.g. , rees & mszros 2005 ; peer et al . 2006 ; giannios 2008 ; beloborodov 2010 ) and synchrotron self - compton ( e.g. , racusin et al . 2008 ; kumar & panaitescu 2008 ) have been also suggested . even though photosphere emission is believed to dominate the spectrum in a small fraction of grbs such as grb 090902b ( abdo et al . 2009b ; ryde et al . 2010 ; zhang et al . 2011 ) and probably grb 090510 ( ackermann et al . 2010 ) , the band component in most grbs may not be the modified photosphere emission . in particular , zhang et al . ( 2012a ) show that the peak energy of the band component of grb 110721a is beyond the `` death line '' in the @xmath10 diagram for the photosphere models , where @xmath11 is the isotropic prompt gamma - ray luminosity and @xmath12 is the peak energy of the @xmath13 spectrum , suggesting that at least some grbs have a dominant emission component above the photosphere in the optically thin region . since the ssc mechanism suffers some other criticisms ( e.g. , piran et al . 2009 ; resmi & zhang 2012 ) , synchrotron radiation is a natural candidate ( see also veres et al . 2012 ) . because both a quasi - thermal photosphere emission component and a non - thermal component are predicted to co - exist in the standard fireball shock model ( mszros & rees 2000 ; peer et al . 2006 ) , the non - detection of the photosphere component ( or in some cases a weak component ) would point towards a magnetically dominated jet in grbs ( zhang & peer 2009 ; daigne & mochkovitch 2002 ; zhang et al . 2011 ) . such a scenario is favored in the fermi era , although further independent support is needed to make a stronger case . it has long been speculated that the physics in different bh jet systems may be essentially the same ( mirabel 2004 ; zhang 2007 ) . a comparative study between the properties of agns and grbs may shed light on the nature of these jets in different scales . by comparing the radio - to - optical spectral properties of optically bright grb afterglows and blazars , wang & wei ( 2011 ) suggested that grb afterglows share a similar emission process with high frequency - peaked bl lac objects . wu et al . ( 2011 ) found that grbs and blazars may have a similar intrinsic synchrotron luminosity in the co - moving frame , which is doppler - boosted by different degrees in different systems , with grbs having a higher doopler boosting factor ( @xmath14 ) than blazars . lately , nemmen et al . ( 2012 ) found that the energy dissipation efficiency of bh powered jets is similar over ten orders of magnitude in jet power , establishing a physical connection between agns and grbs . this letter compares the synchrotron radiation of agns with the prompt gamma - rays of grbs and study their radiation efficiency ( @xmath3 ) and jet composition of agns and grbs . we make use of our detailed modeling results for blazars and nls1s to infer more physical parameters . the data and sample are presented in 2 . a comparison of grbs and agns in the jet power - luminosity plane is presented in 3 . jet radiation efficiency and magnetization degree ( @xmath9 ) are presented in 4 . conclusions are presented in 5 . our agn sample includes 19 bl lacs , 23 fsrqs , and four nls1s . they are gev - tev sources and 51 well - sampled broad - band seds are available for these sources . the seds are taken from zhang et al . ( 2012b ) significance level . we include only these seds in this analysis . ] and abdo et al . ( 2009a , 2010 ) . we fit the seds with a synchrotron+ssc model for bl lacs . the details of our sed fits for the bl lacs can be found in zhang et al . ( 2012b ) . for nls1s and fsrqs , a synchrotron+ssc+ec model is used to fit the seds . the external photon field of the ec process is taken as emission from the broad - line regions ( blrs ) of fsrqs and nls1s . the radiation from the blrs is taken as a black body spectrum , with energy density measured in the comoving frame as @xmath15 erg/@xmath16 ( ghisellini et al . 2010 ) , where @xmath17 is the bulk lorentz factor of the radiating region . most agns in our sample are blazars . since for blazars we are likely looking at the jet within the @xmath18 cone , and that the probability is the highest at the rim of the cone , we take @xmath19 in all the calculations , where @xmath14 is the beaming factor . the radiation region is taken as a homogeneous sphere with radius @xmath20 , which is given by @xmath21 . the minimum variability timescale is taken as @xmath22 hr for the fsrqs and nls1s . the electron distribution is taken as a broken power law , which is characterized by a normalization term ( @xmath23 ) , a break energy @xmath24 and indices ( @xmath25 and @xmath26 ) in the range @xmath27 $ ] . these parameters are preliminarily derived from the observed seds and are further refined in our sed modeling ( zhang et al . @xmath28 and @xmath24 depend on the synchrotron peak frequency ( @xmath29 ) and peak flux ( @xmath30 ) as well as other model parameters . we let @xmath29 and @xmath30 as free parameters in stead of @xmath23 and @xmath24 . @xmath31 is poorly constrained , but it does not significantly affect our results . we fix it at a large value . therefore , the free parameter set of our sed modeling is @xmath32 , @xmath33 , where @xmath34 is the magnetic field strength of the radiating region . we randomly generate a parameter set in broad spaces and measure the consistency between the model result and data with a probability @xmath35 , where @xmath36 is the reduced @xmath37 . the center values and @xmath38 uncertainties of these parameters are derived from gaussian fits to the profiles of the @xmath39 distributions . since we set @xmath40 , the @xmath39 distributions of @xmath41 for some sources have a cutoff at @xmath42 . the @xmath41 of nls1s are poorly constrained in @xmath38 confidence level . taking pks 1454 - 354 as an example , figure [ fig : sed_fit ] shows our best sed fits and the probability distributions of the parameters . our results for the fsrqs and nls1s are reported in table 1 . the observed seds with our best model fitting parameters are presented online . we adopt the conventional assumption that the jet power is carried by electrons and protons with one - to - one ratio , magnetic fields , and radiation , i.e. , @xmath43 , where @xmath44 and @xmath45 is the energy density associated with each ingredient measured in the co - moving frame ( ghisellini et al . note that the radiation power @xmath46 should be a part of @xmath0 before radiation , since @xmath47 is only the power carried by the electrons after radiation . it is estimated with the bolometric luminosity @xmath48 , i.e. , @xmath49 . the radiation efficiency and the magnetization parameter of the jets can be calculated by @xmath50 @xmath51 respectively . the jet luminosities ( bolometric @xmath7 , synchrotron peak @xmath2 , and ic peak @xmath52 ) are derived with @xmath53 , where `` w '' stands for `` bol '' , `` s '' , or `` c '' , and @xmath54 is the relativistic beaming correction factor . our results are reported in table 2 . the grb sample includes 52 typical grbs reported by nemmen et al . ( 2012 ) , excluding two low - luminosity grbs ( 980425 and 060218 ) since they may have a different origin from most high - luminosity long grbs ( liang et al . their isotropic gamma - ray luminosity , jet luminosity , and jet power are calculated with @xmath55 , @xmath56 , and @xmath57 , respectively , where @xmath58 is the duration and @xmath59 is the equivalent isotropic gamma - ray energy of grbs , @xmath60 is the beaming factor estimated with the jet opening angles derived from the observed break in their afterglow lightcurves , and @xmath61 is the isotropic kinetic energy of grb fireballs estimated from the x - ray luminosity during the afterglow phase using the standard afterglow model . the grb radiation efficiency is defined as @xmath62 ( zhang et al . we first plot @xmath0 as a function of @xmath63 , @xmath2 , @xmath7 , and @xmath52 for agns in figure 2 . we make the best linear fits to the data with the minimum @xmath37 technique by considering the errors in both x and y axes , if available , for the fsrqs and bl lacs , respectively . the derived @xmath64 relations are marked in figure 2 . tight correlations between @xmath0 and luminosities are observed for fsrqs , but not for bl lacs . we do not make correlation analysis for nls1s because we have only four nls1s in our sample . since the prompt emission mechanism of grbs is not identified , we add the grbs to these @xmath65 planes by plotting their @xmath0 against @xmath11 or @xmath1 . our best linear fit results are shown in figure 2 . interestingly , the @xmath66 relation of fsrqs is well consistent with the @xmath67 relation of grbs . we have @xmath68 with @xmath69 and @xmath70 for the fsrqs , and @xmath71 with @xmath72 and @xmath73 for the grbs . we make best linear fits to the combined fsrq and grb sample and obtain @xmath74 , with a pearson correlation coefficient of @xmath75 ( chance probability @xmath76 ) and a dispersion of 0.44 dex . nls1s are roughly in the @xmath77 confidence band of this relation , but most bl lacs are out of this band . this relation spans ten orders of magnitude in luminosity with a small dispersion , indicating that both fsrqs and grbs form a well sequence . although tight @xmath78 , @xmath79 and @xmath80 relations are also found for the fsrqs , their intercepts are significantly different from the @xmath67 relation of the grbs . these results suggest that the physical properties between fsrqs and grbs are likely similar and the dominant radiation mechanism of grbs may be the synchrotron radiation of relativistic electrons in the jets . the standard internal shock model of grbs predicts @xmath81 ( e.g. , kumar 1999 ; panaitescu et al . 1999 ) . some grbs satisfy such a constraint , but most of them do not(zhang et al . 2007 ; see also fan & piran 2006 ) . dissipative photosphere emission ( lazzati et al . 2011 ) and internal - collision - induced magnetic reconnection and turbulence ( icmart ) in a poynting - flux - dominated wind ( zhang & yan 2011 ) have been suggested to achieve high radiation powers of grbs . these two scenarios invoke distinctly different jet composition . while the photosphere model invokes a hot , matter - dominated fireball , the icmart model invokes a magnetically dominated emitter in the emission region . figure 3(a ) shows @xmath3 as a function of @xmath7 for agns and @xmath3 as a function of @xmath1 for grbs . the @xmath3 of both grb and fsrq jets are comparable , with a large fraction being greater than 10% , and they increase with @xmath7 or @xmath1 with similar power - indices . lloyd - ronning & zhang ( 2004 ) reported a similar correlation between @xmath3 and @xmath59 for grbs . our best linear fits yield @xmath82 with @xmath83 and @xmath84 for the grbs and @xmath85 with @xmath86 and @xmath87 for the fsrqs . the slopes of both fsrqs and grbs are consistent within error bars . the bl lacs have a lower efficiency ( normally @xmath88 ) than fsrqs and grbs . a large fraction of the bl lacs are out of the 3@xmath9 confidence band of the @xmath89 relation for the fsrqs . a weak @xmath8 correlation is also found for the bl lacs alone , which is @xmath90 with @xmath91 and @xmath92 . the index is different from that of the fsrqs and grbs . we further study the jet composition of agns with our modeling results . even though current grb modeling does not allow us to constrain the magnetization parameter @xmath9 , this can be done for the agn sample . in figure 3(b ) , we plot @xmath3 against @xmath9 for all agns in our sample . the fsrqs tend to have higher @xmath3 and @xmath9 values than the bl lacs . their @xmath9 values are close to or exceeding unity . an anti - correlation between @xmath3 and @xmath9 is found for the fsrqs , i.e. , @xmath93 with @xmath94 and @xmath95 . this is due to @xmath96 dominates @xmath0 for fsrqs , and an anti - correlation between @xmath3 and @xmath9 may be expected from eqs ( 1 ) and ( 2 ) . nls1s show similar trend , but systematically have a lower @xmath3 than fsrqs . bl lacs have lower @xmath3 and @xmath9 values , and a weak correlation is found , i.e. , @xmath97 with @xmath98 and @xmath99 . the dramatic difference of the @xmath100 correlations between fsrqs and bl lacs may further signal the different jet properties of two kinds of sources . the fsrq jets are likely highly magnetized and the bl lac jets are less radiation efficiency and matter dominated . since grbs have a similar radiation efficiency and efficiency - luminosity dependence as fsrqs , one may suggest that the jet properties of grbs are analogous to fsrqs . this supports the idea that grb emission is due to magnetic dissipation in a highly magnetized jet ( e.g. zhang & yan 2011 ) . we have presented a comparative study of grbs and radio loud agns , making use of our systematical sed modeling results for a gev - tev selected sample of agns . we show that the @xmath101 relation of fsrqs is consistent with the @xmath102 relation of grbs . the radiation efficiencies of both fsrqs and grbs are comparable and even increase with @xmath7 with a similar power - law index . bl lacs typically have a lower @xmath3 and @xmath7 than fsrqs , and a tentative @xmath8 relation is found with a different slope from that of the fsrqs . an anti - correlation between @xmath3 and @xmath9 is found for fsrqs , but this correlation is positive for the bl lacs . based on the analogy between grbs and fsrqs , we suggest that grbs are likely produced by synchrotron process in a magnetized jet with high radiation efficiency . the jets of nls1s potentially share similar properties with fsrqs . the jets of bl lacs , on the other hand , are low radiation efficiency and likely matter dominated . compared with agns , grbs are less understood . there is a list of open questions in grb physics , including jet composition , energy dissipation and radiation mechanisms ( e.g. , zhang 2011 ) . our comparative study between grbs and agns shed new light on some of these open questions of grbs . for example , the clear jet power - luminosity correlation suggests that the dominant radiation mechanism of grb prompt emission is similar to the low energy peak of seds for blazars , namely , synchrotron radiation ( see also an independent study of uhm & zhang 2013 ) . the close analogy between grbs and fsrqs in their radiation efficiency - luminosity dependence and the fact that fsrqs have a moderate to high magnetization parameter @xmath9 suggest that grb emission is likely from energy dissipation in a highly magnetized jet ( e.g. , zhang & yan 2011 ) . the high radiation efficiency argument alone may not disfavor the photosphere model of grbs . however , when combining the luminosity - jet power correlation as presented in figure 2 , the magnetic dissipation model is further favored since it invokes synchrotron radiation as the dominant radiation mechanism . all these are also consistent with roming et al . ( 2006 ) , who discovered that some grbs with high radiation efficiency tend to have tight early uvot upper limits , which could be caused by suppression of the reverse shock emission in a magnetized jet ( e.g. , zhang & kobayashi 2005 ; mimica et al . 2009 ) . above conclusions are based on the leptonic model of radiation for the agns and comparative analysis between the agns and grbs . the derivation of @xmath0 of agns is essential for our analysis . the estimate of @xmath47 and @xmath103 , especially @xmath103 are significantly affected by the @xmath104 values , which are simply taken as unity in previous work ( e.g. , ghisellini et al . as shown in table 1 , we find that the typical @xmath41 value is 45 , which lowers both @xmath47 and @xmath103 , with more drastic decrease of @xmath103 , as compared with ghisellini et al . ( 2010 ) . as a result , both @xmath3 and @xmath9 derived in our paper are systematically higher than those derived in ghisellini et al . note that we assumed one cool proton for one relativistic electron in jet power calculations . in case of that the jet power is carried by positron - electron pairs , magnetic field , and radiation , but no protons , the main results for fsrqs hold and the conclusions of our analysis are still valid , but no significant luminosity - radiation efficiency and magnetization parameter - radiation efficiency correlations are found for bl lacs . we thank helpful discussion with g. ghisellini , jian - yan wei , xue - feng wu , and zi - gao dai . this work is supported by the national basic research program ( 973 programme ) of china ( grant 2009cb824800 ) , the national natural science foundation of china ( grants 11078008 , 11025313 , 11133002 , 10725313 ) , guangxi science foundation ( 2013gxnsffa019001 , 2011gxnsfb018063 , 2010gxnsfc013011 ) . bz acknowledges support from nsf ( ast-0908362 ) . 99 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2009a , , 707 , l142 abdo , a. a. , ackermann , m. , ajello , m. , et al . 2009b , , 706 , l138 abdo , a. a. , ackermann , m. , agudo , i. , et al . 2010 , , 716 , 30 ackermann , m. , asano , k. , atwood , w. b. , et al . 2010 , , 716 , 1178 band , d. , matteson , j. , ford , l. , et al . 1993 , , 413 , 281 beloborodov , a. m. 2010 , , 407 , 1033 daigne , f. , bonjak , . , & dubus , g. 2011 , , 526 , a110 daigne , f. , & mochkovitch , r. 2002 , , 388 , 189 dermer , c. d. , schlickeiser , r. , & mastichiadis , a. 1992 , , 256 , l27 eichler , d. , livio , m. , piran , t. , & schramm , d. n. 1989 , , 340 , 126 fan , y. , & piran , t. 2006 , , 369 , 197 ghisellini , g. , maraschi , l. , & dondi , l. 1996 , , 120 , 503 ghisellini , g. , tavecchio , f. , foschini , l. , et al . 2010 , , 402 , 497 giannios , d. 2008 , , 480 , 305 kumar , p. , & panaitescu , a. 2008 , , 391 , l19 kumar , p. 1999 , , 523 , l113 lazzati , d. , morsony , b. j. , & begelman , m. c. 2011 , , 732 , 34 liang , e. , zhang , b. , virgili , f. , & dai , z. g. 2007 , , 662 , 1111 lloyd - ronning , n. m. , & zhang , b. 2004 , , 613 , 477 maraschi , l. , ghisellini , g. , & celotti , a. 1992 , , 397 , l5 mszros , p. , & rees , m. j. 2000 , , 530 , 292 mszros , p. , rees , m. j. , & papathanassiou , h. 1994 , , 432 , 181 mimica , p. , giannios , d. , & aloy , m. a. 2009 , , 494 , 879 mirabel , i. f. 2004 , 5th integral workshop on the integral universe , 552 , 175 nemmen , r. s. , georganopoulos , m. , guiriec , s. , et al . 2012 , science , 338 , 1445 paczynski , b. 1998 , , 494 , l45 panaitescu , a. , spada , m. , & mszros , p. 1999 , , 522 , l105 peer , a. , mszros , p. , & rees , m. j. 2006 , , 642 , 995 piran , t. , sari , r. , & zou , y .- c . 2009 , , 393 , 1107 racusin , j. l. , karpov , s. v. , sokolowski , m. , et al . 2008 , , 455 , 183 rees , m. j. , & mszros , p. 2005 , , 628 , 847 resmi , l. , & zhang , b. 2012 , , 426 , 1385 roming , p. w. a. , schady , p. , fox , d. b. , et al . 2006 , , 652 , 1416 ryde , f. , axelsson , m. , zhang , b. b. , et al . 2010 , , 709 , l172 uhm , z. l. , zhang , b. 2013 , arxiv:1303.2704 urry , c. m. , & padovani , p. 1995 , , 107 , 803 veres , p. , zhang , b .- b . , & mszros , p. 2012 , , 761 , l18 wang , j. , & wei , j. y. 2011 , , 726 , l4 wang , x .- y . , li , z. , dai , z .- g . , & mszros , p. 2009 , , 698 , l98 woosley , s. e. 1993 , , 405 , 273 wu , q. , zou , y .- c . , cao , x. , wang , d .- x . , & chen , l. 2011 , , 740 , l21 zhang , b .- b . , zhang , b. , liang , e .- w . , et al . 2011 , , 730 , 141 zhang , b. 2011 , comptes rendus physique , 12 , 206 zhang , b. , & kobayashi , s. 2005 , , 628 , 315 zhang , b. , liang , e. , page , k. l. , et al . 2007 , , 655 , 989 zhang , b. , lu , r .- j . , liang , e .- w . , & wu , x .- f . 2012a , , 758 , l34 zhang , b. , & peer , a. 2009 , , 700 , l65 zhang , b. , & yan , h. 2011 , , 726 , 90 zhang , j. , liang , e .- w . , zhang , s .- n . , & bai , j. m. 2012b , , 752 , 157 zhang , s. n. 2007 , highlights of astronomy , 14 , 41 lccccccccc fsrqs & & & & & & & & & + 3c 279&0.536&1.6&3.6&2@xmath105&5000&219@xmath10637&([email protected])e3&[email protected]&[email protected] + 3c 273&0.158&1.4&4.0&6@xmath107&2000&328@xmath10679&([email protected])e3&[email protected]&[email protected] + 3c 454.3&0.859&1.4&3.7&33@xmath10614&1700&203@xmath10625&([email protected])e3&[email protected]&[email protected] + pks 1454 - 354&1.424&2.4&3.6&49@xmath10614&2000&276@xmath10699&([email protected])e5&[email protected]&[email protected] + pks 0208 - 512&1.003&1.3&3.6&48@xmath10610&3000&227@xmath10682&([email protected])e3&[email protected]&[email protected] + pks 0454 - 234&1.003&1.4&3.6&17@xmath108&2500&184@xmath10656&([email protected])e2&[email protected]&[email protected] + pks 0727 - 11&1.589&1.8&3.9&50@xmath10612&5000&254@xmath10661&([email protected])e4&[email protected]&[email protected] + pks 0528 + 134&2.07&1.04&4.2&80@xmath1068&7000&222@xmath10654&([email protected])e3&[email protected]&[email protected] + 4c 66.20 & 0.657&1.8&3.2&48@xmath10611&3000&240@xmath10695&([email protected])e4&[email protected]&[email protected] + 4c 29.45&0.729&2.0&4.3&49@xmath1068&5000&400@xmath106112&([email protected])e4&[email protected]&[email protected] + b2 1520 + 31&1.487&2.2&4.2&48@xmath10616&2500&283@xmath10693&([email protected])e4&[email protected]&[email protected] + pks 0420 - 01&0.916&2.64&4.06&51@xmath10611&5000&385@xmath106118&([email protected])e6&[email protected]&[email protected] + 1jy 1308 + 326&0.997&2.2&3.4&44@xmath10610&8000&353@xmath106129&([email protected])e5&[email protected]&[email protected] + pks 1510 - 089&0.36&1.12&3.72&20@xmath10612&1500&305@xmath10632&([email protected])e2&[email protected]&[email protected] + 4c 28.07&1.213&1.4&3.96&41@xmath10610&10000&223@xmath10651&([email protected])e3&[email protected]&[email protected] + pmn 2345 - 1555&0.621&2.0&4.0&46@xmath10611&8000&202@xmath10658&([email protected])e4&[email protected]&[email protected] + s3 2141 + 17 & 0.213&1.8&3.9&53@xmath10610&13000&504@xmath106154&([email protected])e3&[email protected]&[email protected] + s4 0133 + 47&0.859&1.4&3.2&49@xmath1068&4000&191@xmath10665&([email protected])e3&[email protected]&[email protected] + s4 0917 + 44&2.19&1.3&3.54&10@xmath109&5000&213@xmath10666&([email protected])e3&[email protected]&[email protected] + pks 0227 - 369&2.115&1.88&4.1&86@xmath1068&2800&331@xmath106102&([email protected])e5&[email protected]&[email protected] + pks 0347 - 211&2.944&1.2&3.9&59@xmath10636&4000&222@xmath10668&([email protected])e2&[email protected]&[email protected] + pks 2325 + 093&1.843&1.4&4.6&72@xmath1066&900&354@xmath106107&([email protected])e3&[email protected]&[email protected] + pks 1502 + 106&1.839&1.2&3.6&14@xmath110&3000&192@xmath10666&([email protected])e2&[email protected]&[email protected] + nls1s & & & & & & & & & + pks 1502 + 036&0.409&1.2&4.5&55&10000&435@xmath106129&([email protected])e2&[email protected]&[email protected] + pks 2004 - 447&0.24&1.2&4.3&30&3000&311@xmath10642&([email protected])e3&[email protected]&[email protected] + pmn j0948 + 0022&0.5846&1.2&4.1&18&35000&297@xmath10640&([email protected])e3&[email protected]&[email protected] + 1h 0323 + 342&0.061&1.2&4.0&11&25000&637@xmath106128&([email protected])e3&[email protected]&[email protected] + bl lacs & & & & & & + mkn 421@xmath111&44.87&44.31&[email protected]&44.39@xmath112&(7.7@xmath113)e-3&(2.2@xmath114)e-2 + mkn 501@xmath111&44.29&43.77&[email protected]&43.94@xmath115&(2.6@xmath116)e-2&(7.2@xmath117e-2 + w com@xmath111&44.91&44.86&[email protected]&44.65@xmath118&(2.5@xmath119)e-2&(1.9@xmath120)e-2 + bl lacertae@xmath111&45.06&44.79&[email protected]&45.21@xmath121&(4.5@xmath122)e-3&(7.0@xmath123)e-3 + 1es 1959 + 650@xmath111&44.95&43.92&[email protected]&45.75@xmath124&(2.4@xmath125)e-3&([email protected])e-2 + 1es 2344 + 514@xmath111&44.10&43.71&[email protected]&44.55@xmath126&(3.9@xmath127)e-3&([email protected])e-3 + pks 2155 - 304@xmath111&45.87&45.35&[email protected]&44.61@xmath128&(1.5@xmath129)e-2&(5.9@xmath130)e-2 + 1es 1101 - 232@xmath111&45.54&44.60&[email protected]&44.61@xmath131&(9.7@xmath132)e-2&0.34@xmath133 + mkn 501@xmath134&45.46&44.96&[email protected]&44.95@xmath135&(2.5@xmath136)e-2&(3.7@xmath137)e-4 + w com@xmath134&45.00&45.37&[email protected]&44.88@xmath138&(4.6@xmath139)e-2&(7.6@xmath140)e-3 + bl lacertae@xmath134&44.69&44.81&[email protected]&45.79@xmath141&(9.3@xmath142)e-4&(3.6@xmath143)e-4 + pks 2005 - 489@xmath134&45.34&44.11&[email protected]&44.99@xmath144&(3.5@xmath145)e-3&0.11@xmath146 + 1es 1959 + 650@xmath134&45.26&44.83&[email protected]&46.89@xmath147&(4.8@xmath148)e-4&(6.7@xmath149)e-5 + pks 2155 - 304@xmath134&46.12&46.83&[email protected]&45.44@xmath150&(8.9@xmath151)e-2&(3.8@xmath152)e-3 + 3c 66a&46.60&47.14&[email protected]&45.77@xmath153&(9.1@xmath154)e-2&(6.8@xmath155)e-3 + pg1553 + 113&46.61&46.11&[email protected]&44.91@xmath156&(9.5@xmath157)e-2&0.49@xmath158 + 1es 1218 + 30.4&45.39&45.19&[email protected]&44.14@xmath159&0.10@xmath160&0.11@xmath161 + 1es 1011 + 496&46.20&45.64&[email protected]&46.07@xmath162&(1.4@xmath163)e-2&(2.1@xmath164)e-2 + pks 1424 + 240&47.05&46.39&[email protected]&45.31@xmath165&(7.5@xmath166)e-2&0.53@xmath167 + 1es 0806 + 524&44.87&44.35&[email protected]&46.23@xmath168&(7.0@xmath169)e-4&(2.4@xmath170)e-3 + mkn 180&43.88&43.63&[email protected]&44.33@xmath171&(2.2@xmath172)e-2&([email protected])e-2 + rgb j0152 + 017&43.89&43.93&[email protected]&45.45@xmath173&(4.3@xmath174)e-3&(3.8@xmath175)e-4 + h1426 + 428&44.74&45.27&[email protected]&44.91@xmath176&(8.4@xmath177)e-2&(1.3@xmath178)e-3 + pks 0548 - 322&44.31&43.54&[email protected]&44.98@xmath179&(1.2@xmath180)e-2&(8.2@xmath181)e-3 + fsrqs & & & & & & + 3c 279&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 3c 273&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 3c 454.3&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 1454 - 354&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0208 - 512&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0454 - 234&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0727 - 11&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0528 + 134&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 4c 66.20 & [email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 4c 29.45&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + b2 1520 + 31&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0420 - 01&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 1jy 1308 + 326&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 1510 - 089&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 4c 28.07&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pmn 2345 - 1555&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + s3 2141 + 17 & [email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + s4 0133 + 47&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + s4 0917 + 44&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0227 - 369&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 0347 - 211&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 2325 + 093&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 1502 + 106&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + nls1s & & & & & & + pks 1502 + 036&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pks 2004 - 447&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + pmn j0948 + 0022&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + 1h 0323 + 342&[email protected]&[email protected]&[email protected]&[email protected]&[email protected]&[email protected] + ) as a function of the jet luminosity ( geometrically - corrected bolometric luminosity of agns and prompt gamma - ray luminosity of grbs ) and the magnetization parameter ( @xmath9 ) of agns in our sample . color lines are the best linear fits along with 3@xmath9 confidence bands to the data of the grbs ( _ cyan _ ) , fsrqs ( _ magenta _ ) , and bl lacs ( _ green _ ) . ]	gamma - ray bursts ( grbs ) and gev - tev selected radio loud active galactic nuclei ( agns ) are compared based on our systematic modeling of the observed spectral energy distributions of a sample of agns with a single - zone leptonic model . we show that the correlation between the jet power ( @xmath0 ) and the prompt gamma - ray luminosity ( @xmath1 ) of grbs is consistent , within the uncertainties , with the correlation between jet power and the synchrotron peak luminosity ( @xmath2 ) of flat spectrum radio quasars ( fsrqs ) . their radiation efficiencies ( @xmath3 ) are also comparable ( @xmath4 for most sources ) , which increase with the bolometric jet luminosity ( @xmath5 ) for fsrqs and with the @xmath1 for grbs with similar power - law indices . bl lacs do not follow the @xmath6 relation of fsrqs . they have lower @xmath3 and @xmath7 values than fsrqs , and a tentative @xmath8 relation is also found , with a power - law index being different from that of the fsrqs . the magnetization parameters ( @xmath9 ) of fsrqs are averagely larger than that of bl lacs . they are anti - correlated with @xmath3 for the fsrqs , but positive correlated with @xmath3 for the bl lacs . gev narrow - line seyfert 1 galaxies potentially share similar properties with fsrqs . based on the analogy between grbs and fsrqs , we suggest that the prompt gamma - ray emission of grbs is likely produced by synchrotron process in a magnetized jet with high radiation efficiency , similar to fsrqs . the jets of bl lacs , on the other hand , are less efficient and are likely more matter dominated .
nowadays , the evidence of neutron flux enhancement in the lightning discharges in the atmosphere have been obtained in a number of experiments @xcite . in the period of thunderstorm activity a significant excess of the neutron flux over the cosmic background is observed , and the registered neutron energy is in the range from the thermal ( 0.01 ev ) to the fast one ( tens of mev ) . the nuclear fusion processes @xmath6h(@xmath6h , n)@xmath7he in the lightning channel @xcite , the reaction @xmath8c(@xmath6h , n)@xmath9n , @xmath10n(@xmath6h , n)@xmath11o @xcite and the photonuclear ( @xmath12,n ) reactions have been considered @xcite as possible mechanisms , leading to the formation of a statistically significant gain of a neutron flux in a thunderstorm atmosphere . however , a satisfactory explanation of the aggregate of the existing experimental results is still lacking . in this connection , the experiments on studying the neutron emission in the laboratory conditions , in high - voltage discharges in the air , which are similar to conditions observed in a natural storm , seem to be important . the experiments have been carried out with a high current electron beam accelerator erg , reconstructed for studying the high - voltage discharge in the air @xcite . the scheme of the experiment is shown in fig . [ fig:1]a . the voltage amplitude up to 1 mw from the marx - generator with the store energy of 60 kj is fed through the entranced oil - air isolator at to duralumin flange with replaceable electrodes ( cathode ) , and provides discharge current of @xmath13 ka in the air . the diameter of the external grounded cylindrical conductor limits the allowable distance between the cathode and the anode at the level of 1 m. the anode flange can be moved along the axis of the system for a smooth change of the gap . the anodes and cathodes used the same set of replaceable electrodes . the scheme of laboratory experiment . ( a ) layout of diagnostics : 1 , 2 - capacitive and active dividers , 3 - high voltage input from marx - generator , 4 , 13 - magnetic probes , 5 - track detectors , 6 - anode shunt , 7 - rogowski coil , 8,9 - integral cameras , 10 - scintillation detectors , 11 - uv radiation detector , 12 - pmt to visible light , 14 - cathode , 15 - anode . ( b ) layout of cr-39 track detectors : 1 - inside the anode , 2 - inside the cathode , 3 , 4 , 5 - axially placed in water , 6 , 7 , 8 - radially placed at different distances from the discharge . ] electrophysical diagnostics included the recording of currents waveforms ( anode shunt , rogowski coils ) , the voltage ( active and capacitive divider ) , and azimuthal magnetic field probes with a bandwidth of 200 mhz . the channel formation of the discharge is monitored using integral shooting in the optical range from two angles . to register the integral neutron flux the track detectors cr-39 , produced by fukuvi chemical industry company , which are insensitive to the electromagnetic radiation , have been used . the track detectors cr-39 are located near the zone of discharge as shown in fig . [ fig:1]b . the calibration of the cr-39 detector by charged particles has been carried out on the beam of protons from an electrostatic accelerator ( e@xmath14 mev ) , with standard @xmath15-sources ( e@xmath16 mev ) , and on the beam of a cyclotron ( e@xmath17 mev ) at sinp msu . after the irradiation the detectors were etched in the solution 6 m naoh in h@xmath18o at the temperature of 70c during 7 hours . a detailed procedure of the track detector calibration is considered in @xcite . the neutron calibration of the cr-39 detector was performed using a @xmath19cf source with the activity of @xmath20 n@xmath1s@xmath2 in the solid angle of 4@xmath21 . a detector with the radiator of 120 @xmath22 m polyethylene ( detectors pe in fig . [ fig:1 ] ) was used . recoil protons produced by fast neutrons are registered by track detector . the calibration measurements have shown that the diameters of the proton tracks are in the range of @xmath23 @xmath22 m . the average efficiency of fast neutron registration by cr-39 detector was @xmath24 . for the registration of thermal neutrons by the reaction @xmath25 the detectors were placed into the 20@xmath26 solution of na@xmath18b@xmath27o@xmath28 in glycerol ( detectors with b in fig . [ fig:1 ] ) . the neutrons were registered by counting the tracks of @xmath15-particles with e@xmath29 2 mev , which , according to the calibration , have the diameters of @xmath30 @xmath22 m . the average efficiency of thermal neutron registration by cr-39 detector was @xmath31 . fast neutrons of the energy e@xmath32 mev was detected by the reaction of @xmath33 with the energy threshold of about 10 mev . a characteristic signature of the desintegration of the nucleus @xmath8c presents three @xmath15-particles , and their tracks come from a single point . the average efficiency of registration of the fast neutrons by reaction ( [ eq2 ] ) using the cr-39 detector with a 120 @xmath22 m pe radiator was @xmath34 @xcite . the registration of neutron emission in the real - time mode was performed by plastic scintillation detectors . for the detection of fast neutrons and x - rays popop doped polystyrene scintillators were used . the scintillators have an active area of @xmath35 cm@xmath6 and the thickness of 5.5 cm . signals were registered by 4-channel digital storage osciloscope with 1 ghz bandwidth . the intrinsic efficiency of the detector to fast neutrons , measured with the @xmath19cf source , was equal to @xmath37 . in a series of 180 shots the cr-39 track detectors , placed inside spherical anode and cathode , showed the excess of neutron flux over background . the track diameter distributions for the charged particles in the detectors are shown in fig . [ fig:2 ] . the round - shaped tracks with an angle of incidence close to the normal have been chosen . the cr-39 detector with pe radiator of 120 @xmath22 m , placed inside the anode , revealed @xmath385 time excess of tracks above the background within the range of recoil proton diameters of @xmath39 @xmath22 m ( fig . [ fig:2]a ) . this excess should correspond to the average fast neutron flux in the location of the detector @xmath40 n@xmath1@xmath41 per shot . a similar detector , located inside the cathode , also showed the excess above the background in the same range ( fig . [ fig:2]b ) . the average flux of fast neutrons , estimated by the recoil protons , in the detector area was @xmath42 n@xmath1 @xmath41 per shot . the detector inside the anode placed into a 20@xmath26 solution of na@xmath18b@xmath27o@xmath28 in glycerol , has also showed the excess above the background ( fig . [ fig:2]c ) in @xmath30 @xmath22 m range of diameters of @xmath15-particle tracks ( e@xmath43 mev ) . the average thermal neutron flux in the detector area , estimated by the reaction ( [ eq1 ] ) , was @xmath44 n@xmath1@xmath41 per shot . significant excess above the background in the @xmath39 @xmath22 m range of the diameters of recoil proton tracks was also observed . comparison of data from the detectors 1 and 3 has shown that the thermal neutron flux in the far detector ( see . [ fig:2]d ) is greater . probably , the fast neutrons are primary in the discharge , and then they slow down in the water moderator up to thermal energies . very important result is connected with direct observation of desintegration of @xmath8c nucleus into 3 @xmath15-particles . clear signature for such type events was observed on detectors located near discharge area . for example , at the detector 1 ( 2 cm@xmath6 ) , placed near the anode , there were observed 10 events of the desintegration of @xmath8c nucleus into 3 @xmath15-particles ( fig . [ fig:3 ] ) . therefore the number of 3@xmath15-events was n@xmath45 @xmath41 at the background n@xmath46 @xmath41 , i.e. the confidence level of the effect of @xmath8c nucleus desintegration is more than 10 @xmath47 . the average neutron flux of the energy e@xmath32 mev in the detector area , estimated by the reaction of ( [ eq2 ] ) , was @xmath48 n@xmath1@xmath41 per shot . on the different detectors , located near the discharge , n@xmath49 @xmath41 . the observation of neutron signals with plastic scintillation detectors is shown in fig . [ fig:4]a . the detectors are located at a distance of @xmath50 cm from the discharge zone . one of detectors was shielded by 50 @xmath22 m al foil and could detect the pulses from the x - rays ( e@xmath51 10 kev ) and neutrons ( e@xmath52 1 mev ) . the other one had an additional 10 cm thick pb shield . during a discharge both detectors indicated the time structure of signals . the start signal ( presumably , from fast neutrons ) in the second detector with the protection of 10 cm of pb was delayed relative to the first signal ( x - rays ) by 35 ns . using time - of - flight estimation it was obtained the energy of neutron e@xmath53 mev . taking into account the geometric factor ( @xmath54 ) and detection efficiency , fast fraction of neutron flux was estimated as @xmath55 into 4@xmath21 sr per shot . an additional neutron moderator ( 20 cm of polyethylene ) leads to the absence of a signal from fast neutrons at the second detector . this is in agreement with the assumption of fast neutrons emission . as seen from fig . [ fig:4]b , the neutron bursts occur on the pre - pulse current prior to the formation of the discharge . at the moment of neutron generation the pre - pulse current amplitude is @xmath56 times smaller then the amplitude of the main current pulse . two independent methods ( track detector cr-39 and the detector on the basis of plastic scintillators with pmt ) have revealed that in the process of a high - voltage discharge in the air the neutrons are emitted in a wide energy range ( from thermal up to the energies greater than 10 mev ) with the intensity of @xmath57 neutrons per shot into 4@xmath21 solid angle . the obtained data allows one to assume that during the discharge the fast neutrons are produced , and their generation occurs at the initial phase of the discharge and is correlated with the generation of x - ray radiation . to explain the mechanism of the observed emission of neutrons and clarify the location of their source one needs additional experiments . the authors express their gratitude to s.m . zakharov , v.a . bogachenkov and e.i . saunin for assistance in conducting the experiments , a.v . gurevich , s.s . gershtein and g.a . mesyats for helpful discussions . the work was supported in part by the rfbr grant no . 13 - 08 - 01379 . g. n. shah et al . nature ( london ) 313 ( 1985 ) 773 . gurevich et al . prl 108 ( 2012 ) 125001 . fleisher r.l . j. gephys . 80 ( 1975 ) 5005 . m. kuzhevskiy . moscow univ . phys . bull . 5 ( 2004 ) 14 , in russian . fleisher et al . j. gephys 79 ( 1974 ) 5013 . geomagnetism and aeronomy 47 ( 2007 ) 702 . a. v. agafonov , a. v. oginov , and k. v. shpakov , physics of particles and nuclei letters 9 ( 2012)380 . belyaev et al . physics of atomic nuclei 72 ( 2009 ) 1123 . a.m. marenniy , n.a . nefedov , a.s . kratkie soobshenia po fizike fian , n 6 ( 1998 ) 42 , in russian .	data on the observation of neutron bursts in the process of high - voltage discharge in the air at an average electric field strength @xmath0 mv@xmath1m@xmath2 and discharge current @xmath3 ka are presented . two independent methods ( cr-39 track detectors and plastic scintillation detectors ) registered neutrons within the range from thermal energies up to the energies above 10 mev with the flux of @xmath4 neutrons per shot into @xmath5 solid angle .
the ginzburg - landau ( gl ) equation is one of the most - studied nonlinear partial differential equations ( pde ) @xcite . spiral waves are important patterns in many systems , such as physics , chemistry , materials , biology , etc . exploring exact solutions of nonlinear pde is a hot and difficult topic in mathematical physics . many methods were developed for studying exact solutions of nonlinear pdes . recently , a new method named extended f - expansion was proposed @xcite@xcite . it s useful to obtain more jacobi elliptic function solutions . in this paper it is used to study the one - dimensional gl equation . we obtain a spiral wave solution , which is exactly same with hagan s solution @xcite . the ginzburg - landau equation is given by @xmath0 where @xmath1 is a complex function of time @xmath2 and space @xmath3 ; @xmath4 and @xmath5 are real parameters describing linear and nonlinear dispersion , respectively . it s useful to represent the complex function @xmath1 in the following way @xmath6 then the eq . ( [ glequ ] ) becomes @xmath7r - b(2\bigtriangledown\theta\cdot\bigtriangledown r + r\bigtriangleup\theta ) + ( 1 - r^2)r,\\ & r\partial_{t}\theta = b[\bigtriangleup - ( \bigtriangledown\theta)^2]r + 2\bigtriangledown\theta\cdot\bigtriangledown r + r\bigtriangleup\theta -cr^3 . \end{split}\ ] ] if @xmath8 , then it becomes a class of reaction - diffusion equations called @xmath9 systems which have the following general form @xmath7r + r\lambda(r),\\ & r\partial_{t}\theta= 2\bigtriangledown\theta\cdot\bigtriangledown r + r\bigtriangleup\theta + qr\omega(r).\label{rw } \end{split}\ ] ] for one - dimensional systems , let us assume the analogues of spiral wave solutions have the form @xmath10 where @xmath11 is an arbitrary constant and @xmath12 is a long - time scale @xcite . substitution into eq . ( [ rw ] ) yields @xcite @xmath13 let us assume , @xmath14 where @xmath15 and @xmath16 ( @xmath17 ) are constants to be determined later . @xmath18 is a solution of the first - oder nonlinear ode @xmath19 according to the homogeneous balance method @xcite , we know @xmath20 , hence @xmath21 so we get the equations , @xmath22 by requiring the coefficients of each term @xmath23 in the third equation of eqs . ( [ gl1 ] ) are zero , we obtain , @xmath24 from the above equations we obtain @xmath25 , @xmath26 , @xmath27 . then from the first two equations of eqs . ( [ gl1 ] ) we have @xmath28 finally , we obtain @xmath29 .[tab : paraf ] relations between the parameter@xmath30 and @xmath31 , where @xmath31 is satisfied with ode @xmath32 . [ cols="^,^,^,^,^",options="header " , ] according to the table [ tab : paraf ] , the only possible jacobi elliptic function which is satisfied with the eqs . ( [ s0s2s4 ] ) is @xmath33 with @xmath34 , i.e. @xmath35 @xcite . so we obtain , @xmath36 so the spiral wave solution for the one - dimensional gl equation is @xmath37 so we finally get the one - dimensional gl equation s solution @xmath38}.\ ] ] this is the same as hagan s solution @xcite . based on balance mechanism in nonlinear pdes , the extended f - expansion method is widely used to obtain single and combined non - degenerative jacobi elliptic function solutions , as well as their corresponding degenerative solutions , for many kinds of pdes . as example , we give a spiral wave solution for the one - dimensional gl equation in this paper . from this example we can see that the extended f - expansion method is a powerful tool in the nonlinear pde field . we can expect that solutions of the higher dimensional gl equations also will be obtained in this way . 99 i. aranson , l. kramer , rev . phys . , * 74 * ( 2002 ) , 99 j. liu , k. yang , chaos , solitons and fractals * 22 * ( 2004 ) 111 j. zhang , m. wang , y. wang , z. fang , phys . lett . a * 350 * ( 2006 ) 103 hagan , p. s. , siam ( soc . appl . math . ) 42 ( 1982 ) , 762	the one - dimensional ginzburg - landau ( gl ) equation is considered . we use the recently developed extended f - expansion method to obtain spiral wave solution of one - dimensional gl equation .

Subsets and Splits

No saved queries yet

Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.